Skip to main content

57 th EASD Annual Meeting of the European Association for the Study of Diabetes

27 September – 1 October 2021

Abstracts

Index of Oral Presentations

OP 01 Macrovascular disease: large cohorts and large trials

OP 02 Deep and shallow looks at human beta cell gene expression

OP 03 Many faces of diabetic pregnancy

OP 04 GLP-1 receptor agonism: higher dose, combination therapy, or both?

OP 05 Epidemiology of diabetes complications

OP 06 Insights into diabetic retinopathy

OP 07 When men are mice: the study of human physiology in humans

OP 08 Bugs on fire

OP 09 SGLT2 inhibitor trials

OP 10 Peripheral neuropathy: pathophysiology and intervention

OP 11 Cardiac complications: mechanisms and possible treatments

OP 12 The long and winding road to prevention and treatment of diabetes

OP 13 Diverse landscape of type 1 diabetes risk

OP 14 "Humanomics" in diabetes

OP 15 Fat in the liver: where it comes from and how it can be prevented

OP 16 CKD in diabetes - a costly complication

OP 17 Don't stop moving: beneficial effects of exercise on diabetes and beyond

OP 18 Stressed out beta cell organelles

OP 19 Diet and nutrition

OP 20 Keeping the balance in islet secretion

OP 21 The adipocentric angle

OP 22 Understanding kidney disease in diabetes

OP 23 Advances in insulin therapy

OP 24 Epidemiology of diabetes complications

OP 25 Disparities and diversity in diabetes epidemiology

OP 26 Beta cells: sensing, signalling and secreting

OP 27 Prediction tools for outcomes in diabetes

OP 28 Pathogenic mechanisms of complications

OP 29 Understanding muscle and liver metabolism

OP 30 GLP-1 receptor agonism: putative mechanisms of benefit

OP 31 Modelling diabetes long term complications

OP 32 Benefits of GLP-1: from traditional to non-traditional complications

OP 33 Diabetic foot problems: from prediction to treatment

OP 34 SGLT2 inhibition: putative mechanisms of benefit

OP 35 Omics and more for type 2 diabetes and complications

OP 36 Optimising insulin therapy

OP 37 Cytokine storm in type 1 diabetes: from signalling to interventions

OP 38 Novel agents

OP 39 Glucagon metabolism in humans

OP 40 Protecting the kidney in diabetes

OP 41 Building the evidence base for new devices

OP 42 Cardiovascular disease: predictors and outcomes

OP 43 Genes, epigenes and telomeres in type 1 diabetes

OP 44 Diabetes around the clock!

OP 45 What surgery can do for you

OP 46 (Path)ways to develop human beta cells

OP 47 Hypoglycaemia consequences at system level

Index of Short Oral Discussions

SO 01 Diabetes epidemiology at scale: registries and large databases

SO 02 Diabetes across generations

SO 03 Diet, lifestyle and behaviour

SO 04 Prediction models and precision medicine

SO 05 Therapeutic advances

SO 06 Genes and genomic engineering

SO 07 Risk factors and consequences of poor glycaemic control

SO 08 Preservation versus destruction of beta cell mass

SO 09 The ins and outs of insulin secretion

SO 10 Beta cells to the grave in type 1 diabetes

SO 11 In vivo and ex vivo beta cell function in diabetes

SO 12 Sugar Moms

SO 13 From pregnant women and mice

SO 14 Exercise effects beyond blood glucose

SO 15 Novel methods to study metabolism in diabetes

SO 16 The many faces of insulin sensitivity

SO 17 Novel aspects of beta cell and insulin secretion

SO 18 It must be my hormones

SO 19 Gastro-entero pancreatic factors

SO 20 Carbohydrate and protein metabolism

SO 21 Metabolic control during and after pregnancy

SO 22 News from the drug pipeline

SO 23 Fatty matters

SO 24 Glucose-lowering drugs and the liver

SO 25 Interaction of digestive system and glucose metabolism

SO 26 Cardiorenal consequences of SGLT2 inhibition

SO 27 Glucose-lowering agents: Real World Evidence

SO 28 GLP-1 receptor agonists and weight loss

SO 29 Novel glucose-lowering agents

SO 30 The advantage of dual agonists

SO 31 Clinical aspects of semaglutide

SO 32 Different aspects of SGLT2 inhibitors

SO 33 Non-insulin treatment in type 1 and type 2 diabetes

SO 34 News from new insulins

SO 35 More on insulins

SO 36 Determinants and consequences of hypoglycaemia

SO 37 New approaches to health care delivery

SO 38 Psychological aspects of diabetes care

SO 39 Apps, devices and tools and their impact on diabetes care

SO 40 CGM

SO 41 Closed loop systems

SO 42 Other aspects of managing blood glucose levels

SO 43 Insulin pumps

SO 44 Glycaemic management in special settings and populations

SO 45 Seeing the full picture of diabetic retinopathy

SO 46 Diabetic foot: from cost to COVID

SO 47 Autonomic neuropathy

SO 48 Peripheral neuropathy - predictors of disease and prognosis

SO 49 Neuropathy: from mechanisms to memory

SO 50 Nephropathy interventions: from blueberries to SGLT2

SO 51 Burdens and bones in CKD and diabetes

SO 52 Predictors of diabetic kidney disease

SO 53 New insights from animal models of complications

SO 54 Pathogenic mechanisms of complications

SO 55 Not so sweet: cancer and diabetes

SO 56 Looking at the brain and its function

SO 57 All you need to know about atherosclerosis and diabetes

SO 58 Brain, kidney and vascular complications

SO 59 Vascular complications: mechanisms and risk factors

SO 60 Hepatic fibrosis: from screening to treatment

SO 61 Fatty liver always hides some complications

SO 62 Cardiovascular complications and drugs

SO 63 COVID-19: from the known to unknown

SO 64 Cardiac complications in diabetes and prediabetes

OP 01 Macrovascular disease: large cohorts and large trials

1

Association of increased intima media thickness and arteriosclerosis with elevated fasting insulin levels in middle-aged persons

M. Röhling1, K. Kempf1, H. Kolb2, M. Schneider3, S. Martin1;

1West-German Center for Diabetes and Health, Dusseldorf Catholic Hospital Group, Dusseldorf, 2Faculty of Medicine, Heinrich-Heine-University Dusseldorf, Dusseldorf, 3Occupational Health and Medical Services, Boehringer Ingelheim Pharma GmbH & Co. KG, Ingelheim upon Rhine, Germany.

Background and aims: Recently published genetic studies could show for the first time a causal link between elevated insulin levels and cardiovascular disease (CVD) risk. We, therefore, hypothesized that increased fasting insulin levels are also associated with precursors of CVD such as increased intima media thickness (IMT > 1 mm) or arteriosclerosis (AS, defined by the occurrence of plaques).

Materials and methods: In an ongoing study, middle-aged (≥40 years) employees of Boehringer Ingelheim were regularly followed up as part of an occupational health care program. Clinical and laboratory parameters were examined. Multivariable logistic regression analyses were performed to determine odds ratios (OR) for the incidence of increased IMT or AS in relation to fasting insulin levels at baseline and its change until follow-up in comparison to basal BMI and its change in comprehensive models (Model 1: baseline laboratory and clinical parameters; Model 2: Model 1 + BMI change; Model 3: Model 1 + insulin change). Adjusted relative risk for increased IMT and AS was calculated by using Mantel-Haenszel analysis. For this purpose, fasting insulin and BMI were recoded into dichotomous variables defined as hyperinsulinemia (fasting insulin: >15 μU/ml) or overweight (BMI: ≥25 kg/m²).

Results: From n=6825 persons who entered the healthcare program, n=3332 had their first follow-up after 5.0±1.0 years. After excluding all participants with incomplete data (n=1327) or already existent cardiovascular impairments (n=366), data sets of 1639 participants were included in the analysis. Increased IMT during follow-up was diagnosed in 238 participants (15 %) and 328 (20 %) developed AS. Logistic regression analysis identified fasting insulin, BMI and smoking as risk factors for both cardiovascular endpoints (all p<0.05), whereas age and diastolic blood pressure were risk factors for increased IMT only, and male sex was associated with incident arteriosclerosis only (all p<0.01). Additional adjustment for BMI change during follow-up did not modify these associations (including fasting insulin), but adjustment for fasting insulin change during follow-up removed BMI as risk factor for both cardiovascular endpoints. Fasting insulin change during follow-up but not BMI change was found to be associated with increased IMT (OR [95% CI]: 1.055 [1.030; 1.082]) and AS (OR [95% CI]: 1.057 [1.030; 1.085]) (both p<0.001). Subgroup analyses showed that high baseline values of fasting insulin or BMI in combination with greater change during follow-up, respectively, yielded highest risks for both cardiovascular endpoints. The analysis of adjusted relative risks indicated that both, fasting insulin and BMI added to age and sex as risk factors. Interestingly, including the parameter smoking did not lead to a higher risk for increased IMT or arteriosclerosis than conferred by high fasting insulin levels.

Conclusion: Higher basal fasting insulin levels and increases in fasting insulin over time are associated with atherogenic progression and supersede BMI as risk factor.

Supported by: KK and SM received a research grant from Boehringer Ingelheim (grant no. 43034856)

Disclosure: M. Röhling: None.

2

Kidney function measures and cardiovascular outcomes in people with diabetes: the Hoorn Diabetes Care System cohort

E. Dal Canto1,2, A.A. van der Heijden2, A.J. van Ballegooijen1, B.I. Lissenberg-Witte1, F. Rutters1, P. Elders2, J.J.W. Beulens1,3;

1Epidemiology and Data Science, Amsterdam University Medical Center, Amsterdam, 2General Practice, Amsterdam University Medical Center, Amsterdam, 3Julius Center for Health Sciences and Primary Care, Utrecht University Medical Center, Utrecht, Netherlands.

Background and aims: Kidney function measures - estimated glomerular filtration rate (eGFR) and albuminuria - have been consistently associated to cardiovascular disease (CVD) risk in people with T2D. Since multiple mechanisms underlie the link between kidney dysfunction and diabetes macrovascular complications, each of the manifestations of kidney disease in T2D could contribute to increase the risk of specific CVD subtypes. Accordingly, we aimed to assess the prospective association between longitudinally repeated measures of eGFR and albuminuria and the occurrence of different cardiovascular events - including myocardial infarction (MI), coronary heart disease (CHD), stroke and heart failure (HF) - as well as cardiovascular mortality in people with diabetes.

Materials and methods: 13,507 people with T2D from The Hoorn Diabetes Care System cohort, a cohort study consisting of nearly all people with diabetes in primary care from the West-Friesland region in the Netherlands, were followed-up annually since 1998. Multivariate time-dependent Cox regression models adjusted for common cardiovascular risk factors as well as cardiovascular drug use, were used to assess the association of repeated measures of eGFR and urinary albumin/creatinine ratio (UACR) as clinical categories with CVD outcomes (time to first MI, CHD, HF and stroke events) and cardiovascular mortality. We tested for effect modification by sex, that was observed only for the association between UACR categories and HF (interaction dummies for UACR category 3.0-30.0 and >30.0 mg/mmol: both P<0.001).

Results: Mean age at baseline was 62.4±11.8 years and 53.7% were males. During a median follow-up time of 7.0 years (interquartile range 3-12) event rates per 1000 person-years were 3.08 for MI, 3.72 for CHD, 1.12 for HF, 0.84 for stroke and 6.25 for cardiovascular mortality. Categories of moderately (60-90) and severely reduced eGFR (<60 mL/min/1.73m2) were prospectively and independently associated with a higher risk of MI, CHD and stroke but not of HF nor cardiovascular mortality, compared to eGFR>90 mL/min/1.73m2. Conversely, categories of moderately (3-30 mg/mmol) and severely increased (>30.0 mg/mmol) UACR were prospectively associated with a higher risk of cardiovascular mortality in men and women and with a higher risk of HF in women only, compared to normal or mildly increased UACR (<3.0 mg/mmol). No significant association were observed between increased UACR and the risk of MI, CHD or stroke.

Conclusion: This study indicated differential prospective associations between each manifestation of kidney disease in T2D and cardiovascular events and mortality. In particular a longitudinal decline of eGFR associated to a higher risk of atherosclerotic CVD, while increased albuminuria associated to increased risk of HF and cardiovascular mortality. These findings suggest a regular monitoring of kidney function over time could be of potential utility to identify diabetes patients at higher cardiovascular risk.

Disclosure: E. Dal Canto: None.

3

Comparative evaluation of GLP-1 receptor agonists and SGLT-2 inhibitors neuroprotective properties in transient brain ischaemia

A. Simanenkova1,2, N. Timkina1, O. Fuks1, A. Khalzova2, A. Shimshilashvili1, V. Timofeeva1, T. Karonova1,2, T. Vlasov2;

1Almazov National Medical Research Centre, Saint-Petersburg, 2Pavlov First Saint-Petersburg State medical university, Saint-Petersburg, Russian Federation.

Background and aims: Nowadays guidelines for type 2 diabetes mellitus (DM) treatment focus not only at glucose-lowering properties but also at cardiovascular effects of the drugs. Recently, according to LEADER ang EMPA-REG OUTCOME studies, GLP-1 receptor agonist liraglutide (LIRA) and SGLT-2 inhibitor empagliflozin (EMPA) have demonstrated outstanding cardioprotective potential. Importantly, ischemic stroke has high incidence in type 2 diabetes mellitus (DM) comprising the second leading cause of death in these patients. Nevertheless, there are lack of data concerning potential neuroprotective effect of above-mentioned drugs. The aim of this study was to investigate neuroprotective actions of LIRA and EMPA, in comparison with metformin (MET) in acute rat brain ischemia. We have chosen MET as it is the first-line therapy for type 2 DM.

Materials and methods: male Wistar rats 200-255 g were treated with LIRA 1 mg/kg s.c. once daily for 7 days (group “LIRA”, n=12), EMPA 2 mg/kg per os once daily for 7 days (group “EMPA”, n=9), MET 200 mg/kg kg per os once daily for 7 days (group “MET”, n=8) or 0.9% NaCl s.c. once daily for 7 days (“Control” group, n=12). Five hours after last drug administration all the animals were subjected to 30-min filament middle cerebral artery occlusion (MCAO). 48 hours after MCAO neurological deficit was evaluated by means of Garcia score - healthy animals have 18 points while maximal neurological deficit is characterized by 3 points. After that rats were euthanized and brain slices were prepared and incubated with 1% 2,3,5-triphenyltetrazolium chloride for necrosis measurement. Blood glucose level (BGL) was studied 3 times every second day during the treatment.

Results: Brain infarct volume was significantly smaller in “LIRA” group in comparison with “Control” (5.50 (3.97; 5.50) and 16.56 (13.33; 24.65) % of total brain volume, respectively). Brain infarct volume in “EMPA” group (4.91 (2.67; 14.49) % was also significantly smaller than in “Control” group. Importantly, there was no difference between “LIRA” and “EMPA” groups. Treatment with MET also led to brain damage volume decrease (8.67 (5.39; 30.07) %, comparing with control, but still it was significantly larger than both in “LIRA” and “EMPA” groups. Rats in group “LIRA” had less prominent neurological deficit and more points according to Garcia score (14.0 (11.5; 15.5) comparing with “Control” group (12.0 (9.0; 14.0). Neither EMPA, nor MET diminished neurological deficit, comparing with “Control” group (12.0 (9.5; 14.0) and 12.0 (6.5; 12.5) points vs 12.0 (9.0;14.0)). BGL was normal in all groups, with no hypoglycemic episodes during treatment.

Conclusion: LIRA, EMPA and MET are neuroprotective in rat transient brain ischemia and this effect is not connected with influence on glucose metabolism. Infart-limiting effect of LIRA and EMPA is similar and is more prominent than that of MET. Only LIRA administration prior to ischemia modelling causes neurological deficit diminishing.

Disclosure: A. Simanenkova: None.

4

Insulin resistance and risk of first stroke in type 2 diabetes: a nationwide cohort study

A. Zabala1, V. Darsalia1, M. Lind2, A.-M. Svensson3, S. Franzén3, B. Eliasson4, C. Patrone1, M. Jonsson5, T. Nyström1;

1Department of Clinical Science and Education, Södersjukhuset, Karolinska Institute, Stockholm, 2Department of Molecular and Clinical Medicine, University of Gothenburg, Gothenburg, 3Centre of Registers in Region Västra Götaland, Gothenburg, 4Department of Molecular and Clinical Medicine, Gothenburg University, Gothenburg, 5Department of Molecular Medicine and Surgery, Karolinska Institute, Stockholm, Sweden.

Background and aims: Insulin resistance contributes to the development of type 2 diabetes (T2D) and is a cardiovascular risk factor. The aim of this study was to investigate the potential association between insulin resistance measured by estimated glucose disposal rate (eGDR) and risk of first stroke and death in people with T2D.

Materials and methods: This was a nationwide population based observational cohort study that included all T2D patients from the Swedish national diabetes registry between 2005 to 2016 with full data on eGDR which was calculated by the formula (based upon the euglycemic hyperinsulinemic clamp technique): eGDR (mg/kg/min) = 21.158 - (0.09 * WC) - (3.407 * HT) - (0.551 * HbA1c) [WC = waist circumference (cm), HT = hypertension (yes=1/no=0), and HbA1c = HbA1c (DCCT %)]. eGDR was categorized as following: <4 (highest grade of insulin resistance), 4-5.99, 6-7.99, and ≥8 mg/kg/min). We calculated the crude incidence rates and 95% confidence intervals (CIs) and used multiple Cox regression to estimate hazard ratios (HRs) to assess the association between the risk of stroke and death and the eGDR categories in which the lowest category served as a reference. The relative importance of each factor in the eGDR formula was measured by the R2 (±SE) values by calculating the explainable log-likelyhood that was attributable to each risk factor.

Results: A total of 104 697 T2D patients (woman 44.5%) with a mean age of 63 years was included in this study. During a median follow up-time of 5.6 years, 4201 strokes occurred (4.0%). Crude survival curves for freedom of stroke are shown in the figure. After multivariate adjustment the HR (95% CI) for stroke in patients with eGDR categories between 4 to 5.99, 6 to 7.99 and >8 were: 0.77 (0.69-0.87), 0.68 (0.58-0.80) and 0.60 (0.48-0.76), compared to the reference, i.e. eGDR <4). The corresponding numbers for risk of death were: 0.83 (0.76-0.89), 0.77 (0.69-0.77) and 0.72 (0.59-0.88). The estimated explained relative risk (R2) for each factor in the eGDR formula was for the risk of stroke: hypertension (0.045±0.0024), HbA1c (0.013±0.0014), and waist (0.006±0.0009), respectively.

Conclusion: Insulin resistance, measured as eGDR, is associated with an increased risk of stroke and death in people with T2D. The relative importance of the predictors in the eGDR formula for the risk of stroke was highest for hypertension followed by glycemia.

figurea

Disclosure: A. Zabala: None.

5

Gender differences in cardiovascular risk, treatment, and outcomes: a post-hoc analysis from the REWIND trial

G. Ferrannini1, J.M. Maldonado2, S. Raha2, P. Rao-Melacini3, R. Khatun3, C. Atisso2, L. Shurzinske2, H. Gerstein3, L. Rydén1, M.A. Bethel2;

1Karolinska Institutet, Stockholm, Sweden, 2Eli Lilly and Company, Indianapolis, USA, 3Population Health Research Institute, Hamilton, Canada.

Background and aims: Cardiovascular disease (CVD) is the leading cause of death in women and more common in those with type 2 diabetes (T2D). Evidence suggests that the development of T2D adversely affects metabolic and CVD risk factor profiles more in women than men. The aim of this analysis was to investigate gender differences in risk factor management and outcomes in the REWIND trial.

Materials and methods: Analysis was performed on a subset of the REWIND participants by excluding patients with missing data at baseline (BL) or 2 years for HbA1c, systolic BP, LDL-cholesterol, and concomitant medications, or BL history of CVD either missing or unknown. Gender differences in BL characteristics, cardioprotective therapies use at BL and after two years, achievement of relevant treatment targets, and observed cardiovascular (CV) outcomes were analysed. The risk for CV outcomes including fatal/nonfatal stroke, fatal/nonfatal myocardial infarction, CV death, all-cause mortality, and heart failure hospitalisation in women versus men were analysed using Cox proportional hazards models adjusted for randomised treatment and key baseline characteristics identified using stepwise variable selection. Time-to-event analysis was performed in the subgroups with and without CVD history using Cox proportional hazards regression models, including gender, subgroup, randomised treatment, and the gender-by-subgroup interaction.

Results: Of 9901 study participants with either high CV risk or established CVD, 4589 (46.3%) were women. Significantly fewer women than men had a history of CVD (20.0% vs 41.4%; P<0.001). Although the majority of women met clinically relevant treatment targets for blood pressure (96.7%) and lipids (72.8%) at BL, fewer women than men were at target for relevant clinical targets of ACE inhibitor/ARB use, lipid control/statin use, or aspirin use (P<0.001 for all). Overall, women had a lower risk than men for all CV outcomes except fatal/nonfatal stroke, a pattern echoed among the subgroup without a history of CVD at BL. Compared to men (Figure), women with a history of CVD had a similar risk for stroke, heart failure hospitalisation, all-cause mortality, and CV death.

Conclusion: In this international trial cohort of patients with T2D and high CV risk or established CVD, women, overall, were less likely than men to reach treatment targets for CV risk management. Nonetheless, they remained at lower risk for all adverse CV outcomes except stroke. These findings warrant further investigation in women with T2D.

figureb

Clinical Trial Registration Number: NCT01394952

Supported by: Eli Lilly and Company

Disclosure: G. Ferrannini: None.

6

The importance of addressing multiple risk markers in type 2 diabetes: results from LEADER and SUSTAIN 6

E. Hein Zobel1, B.J. von Scholten1,2, T.W. Hansen1, F. Persson1, S. Rasmussen2, B. Wolthers2, P. Rossing1,3;

1Steno Diabetes Center Copenhagen, Gentofte, 2Novo Nordisk A/S, Søborg, 3University of Copenhagen, Copenhagen, Denmark.

Background and aims: The extent to which improvements in multiple risk markers affect outcomes in type 2 diabetes (T2D) is unclear. Our aim was to investigate whether improvement in multiple risk markers confers lower risk of vascular disease in patients with T2D and cardiovascular disease (CVD)/high risk for CVD, independent of specific treatments.

Materials and methods: This was a post-hoc analysis of the LEADER (n=8638; median follow-up 3.8 years) and SUSTAIN 6 (n=3040; median follow-up 2.1 years) cardiovascular outcome trials in patients with T2D. Patients had baseline and year-1 assessment of at least one of the parameters of interest. We pooled the liraglutide/semaglutide- and placebo-treated groups and categorised them by number of risk markers with clinically relevant improvements after 1 year. We investigated risk of major adverse cardiovascular events (MACE: cardiovascular death, nonfatal myocardial infarction and nonfatal stroke); expanded MACE (MACE + coronary revascularisation and hospitalisation for heart failure or unstable angina pectoris); cardiovascular death; or nephropathy (new onset of macroalbuminuria or doubling of the serum creatinine level and eGFR ≤45 mL/min/1.73 m2, or the need for continuous renal-replacement therapy, or death from renal disease). Clinically relevant change: body weight loss ≥5%, HbA1c reduction ≥1%, systolic blood pressure reduction ≥5 mmHg, low-density lipoprotein cholesterol reduction ≥0.5 mmol/L, eGFR reduction ≥0 mL/min/1.73m2 and urinary albumin-to-creatinine ratio reduction ≥30% of baseline value. Numbers of risk markers with change were classified as: none (G0) to ≥4 (G4). Cox regression was used; models were adjusted for continuous baseline levels of risk markers and treatment group (liraglutide/semaglutide and placebo) and stratified by trial.

Results: Baseline characteristics were similar in each subgroup. Compared with patients with 0 risk-marker improvements, patients with 2, 3 or ≥4 improved risk markers had reduced risk of expanded MACE (HR 0.80; 0.80; 0.82), cardiovascular death (HR 0.66; 0.67; 0.60) and nephropathy (HR 0.71; 0.48; 0.43) (Figure). One improved risk marker conferred no risk reduction. The trend of decreased risk with each additional risk marker improvement was observed for expanded MACE (p=0.004), cardiovascular death (p=0.005) and nephropathy (p<0.0001). We observed an increasingly higher number of patients on liraglutide/semaglutide treatment in groups with 0, 1, 2, 3 or ≥4 risk marker improvements as follows: 30.5% in G0, 38.0% in G1, 48.8% in G2, 61.6% in G3 and 75.3% in G4.

Conclusion: In patients with T2D, improvements in ≥2 risk markers confers reduced risk of CVD and nephropathy as compared with 0 or 1 improved risk marker. The results stress the importance of multifactorial intervention targeting all risk markers and the benefit of pleiotropic antidiabetic treatments.

figurec

Clinical Trial Registration Number: NCT01179048; NCT01720446

Supported by: Novo Nordisk A/S

Disclosure: E. Hein Zobel: None.

OP 02 Deep and shallow looks at human beta cell gene expression

7

About time: functional and molecular effects of prolonged ex vivo human islet culture

M. Suleiman1, E. Bosi1, A. Piron2, C. De Luca1, M. Tesi1, S. Del Guerra1, D.L. Eizirik2,3, M. Cnop2, P. Marchetti1, L. Marselli1;

1University of Pisa, Pisa, Italy, 2ULB Center for Diabetes Research, Universitè Libre de Bruxelles, Bruxelles, Belgium, 3Indiana Biosciences Research Institute, Indianapolis, USA.

Background and aims: Several studies have addressed human islet features during ex vivo culture under different conditions. It is, however, unclear if and to which extent culture time affects human islet cell functional and molecular traits. Here we studied whether glucose-stimulated insulin secretion (GSIS), insulin content and transcriptome signatures of human islets change upon 8 days of culture after isolation.

Materials and methods: Islets were prepared by enzymatic digestion and density gradient purification from 17 non-diabetic organ donors (age: 70±4 years, mean±SEM; BMI: 24±1 Kg/m2; sex: 14F/3M) and cultured in control M199 medium. GSIS, insulin content and RNA-sequencing were studied 2 (D2), 4 (D4) and 8 (D8) days from isolation. Transcriptomes were compared to identify gene expression changes over time.

Results: GSIS (n=17), expressed as insulin stimulation index, was similar at D2 (3.5±0.4), D4 (3.1±0.4) and D8 (3.1±0.3). Insulin content was 301.2±28.6 μU/islet at D2, 258.1±35.6 μU/islet at D4 (p=0.13 vs D2) and 205.5±27.4 μU/islet at D8 (p=0.20 vs D4, p<0.01 vs D2 after Tukey’s correction). Gene expression trajectories (n=11) revealed that, at D8 vs D2, 1125 genes were differentially expressed (FDR<0.05, absolute fold-change>2), of which 425 and 770 were respectively up- and down-regulated. The top 5 up- and down-regulated genes were: AC004017.1, PTH2R, NKIRAS2, CTD-3199J23.6, KLHL22; GJC1, ZNF469, WNT5A, COL4A1, CAV1. Gene Set Enrichment Analysis retrieved 40 significant (FDR<0.05) KEGG pathways. Twenty of them were positively enriched (including Fatty Acid Metabolism, Peroxisome and Glycolysis/Gluconeogenesis), and 20 were negatively regulated (including Cell Cycle, ECM receptor interaction, Apoptosis and DNA Replication).

Conclusion: In the present study prolonged ex vivo culture of human islets did not significantly affect beta cell insulin responsiveness to glucose stimulation, but reduced insulin content. The respective, associated transcriptomic signatures could unveil some of the molecular mechanisms regulating beta cell function and insulin reservoir, to be possibly surveyed in the study of beta cell resilience or subsidence.

Supported by: RHAPSODY, INNODIA HARVEST

Disclosure: M. Suleiman: None.

8

Identification of mRNA and microRNA transcripts that differ in expression across islet donor sex, age and BMI

W.K.M. Wong1, M.V. Joglekar1, A.E. Sørensen2, Y. Chew3, F. Cheng4, T. Loudovaris5, H.E. Thomas5, R.C.W. Ma4, W.J. Hawthorne3, L.T. Dalgaard2, A.A. Hardikar1,2;

1School of Medicine, Western Sydney University, Sydney, Australia, 2Roskilde University, Roskilde, Denmark, 3Westmead Institute for Medical Research, University of Sydney, Sydney, Australia, 4The Chinese University of Hong Kong, Hong Kong, Hong Kong, 5St. Vincent’s Institute for Medical Research, Melbourne, Australia.

Background and aims: Donor age, sex and body mass index (BMI) have been recognized to influence pancreatic islet gene expression. These traits have also been implicated with metabolic health and/or associated with diabetes risk. Since human islets are widely used as a research tool in understanding metabolic pathways in disease, it is important to identify transcriptomic features that are significantly different merely due to sex, age, or BMI of islet donors. Such data obtained from several human islets is an important resource to understand true effects of metabolic stressors on human islet gene expression profiles.

Materials and methods: We analyzed mRNA (58,190 transcripts) and microRNA (miRNA; 754 known/validated) profiles from 131 non-diabetic donor islet preparations (males n=38, female n=25 for mRNA dataset; male n=53, female n=48 for miRNA dataset; of which 33 (male n=18 and female n=15) from the 131 samples, both mRNA and miRNA data were available) generated via bulk RNA-sequencing and TaqMan real-time quantitative PCR respectively. Islet transcriptome profiles were then assessed along with islet metadata on sex, age, BMI and other donor characteristics. Univariate analyses, adjustments to other co-variates and machine-learning penalized regression with bootstrap (at 1000 iterations) analyses were applied. Correlation matrix and predictive target tools were used to identify potential miRNA-mRNA transcript interaction within islets.

Results: Analyses on entire human islet transcriptome identified 1628 significantly different gene transcripts across sex (excluding sex-chromosome linked transcripts), 2208 transcripts correlated with age (range 16 to 69 years) and 1774 transcripts correlated significantly with BMI (range 17.9 to 47.3 kg/m2). Of these, 85 transcripts were common between the sex and age comparison; and 56 transcripts were common between the sex and BMI comparison, while three transcripts were common between each comparison. Interestingly, there were 142 (sex-), 292 (age-) and 267 (BMI-) associated transcripts which were also present in their respective regression bootstrap model. Analyses across the miRNA profiles in human islet samples identified 40 miRNAs associated with sex differences, 23 and 82 miRNAs correlated with age (range 16 to 66 years) and BMI (range 17.9 to 43.6 kg/m2) respectively. Of these, miR-147b was common between the sex and age comparison; and 18 miRNAs were common between the sex and BMI comparison, while miR-99c-3p was common between all three comparisons. Regression bootstrap analysis identified 17, 18 and 37 miRNAs as significantly associated with sex, age and BMI respectively that are common to univariate analysis. Correlative analysis in our dataset and predictive target gene tools identified potential miRNA-mRNA interactions in human islets.

Conclusion: Our analyses reveal multiple genes and miRNAs that are associated with different islet phenotype characteristics. These findings serve as an important resource to understand the underlying mechanisms between donor variables and gene expression profiles in human islet biology.

Supported by: AAH acknowledges JDRF Australia CDA.

Disclosure: W.K.M. Wong: None.

9

A single cell atlas of de novo beta cell regeneration in adult zebrafish identifies hybrid cell states that facilitates diabetes reversal

P. Chawla1, S.P. Singh2, L.D. Silva1, A. Hnatiuk1, M. Kamel1, B. Spanjaard3, J.P. Junker3, N. Ninov1,4;

1Center for Regenerative Therapies Dresden, Dresden, Germany, 2IRIBHM, Université Libre de Bruxelles (ULB), Brussels, Belgium, 3Max Delbrück Center for Molecular Medicine, Berlin, Germany, 4Paul Langerhans Institute Dresden of the Helmholtz Zentrum München at the University Hospital and Faculty of Medicine Carl Gustav Carus of Technische Universität Dresden, Dresden, Germany.

Background and aims: To achieve restoration of β-cell mass after near-compete β-cell destruction, it will be imperative to stimulate endogenous β-cell formation from progenitor-differentiation or via trans-differentiation of non β-cells. However, we lack a systematic understanding of the cellular sources underlying β-cell regeneration. In our study, using the regeneration competent model zebrafish (Danio rerio), we focused on mapping the plasticity of different pancreatic cells during β-cell regeneration. To this end, we generated a transcriptomic profile of the adult zebrafish pancreas during four stages encompassing the period from β-cell destruction to the emergence of new insulin-expressing cells and the resolution of diabetes.

Materials and methods: We employed single-cell RNA sequencing (both 10X-Genomics and SmartSeq2), time-course live-imaging and in vivo calcium imaging to study β-cell regeneration following near-complete destruction. Single-cell sequencing (10X-Genomics) was performed on dissected pancreata from adult zebrafish at 0 (β-cell destruction), 2, 7 (hyperglycemia) and 14 (glucose normalization) days post β-cell destruction.

Results: Using single-cell RNA sequencing, we classified the cell populations in the adult zebrafish pancreas. Among the pancreas population, we discovered stable populations of endocrine cells with hybrid identity, sharing the hormones and fate-determinants of both δ- and β-cells. Moreover, we revealed that upon β-cell destruction, a novel population of plastic δ-cells gives rise to such intermediate lineage cells. The hybrid cells serve as the chief-insulin expressers during diabetes reversal. These plastic δ-cell population also have a distinct expression of progenitor markers such as pdx1, ppdpfa, ppdpfb. These δ-cells also express the Wnt regulator dkk3b whose role has not been previously implicated in endocrine differentiation. We found that overexpression of dkk3b led to an increase in the pool of hybrid cells in the absence of injury. Finally, using in vivo calcium imaging, we show that the hybrid cells acquire glucose-responsiveness during the course of regeneration.

Conclusion: Our study has identified a novel population of δ cells in the zebrafish pancreas, which exhibit higher phenotypic plasticity compared to other pancreatic cells and rapidly acquire insulin expression and glucose-stimulated calcium influx post β-cell destruction.

Supported by: CRTD, DZD, DFG, IRTG, ERC

Disclosure: P. Chawla: None.

10

An integrated analysis of human pancreatic islet single cells reveals autocrine and paracrine

interactions

E. Bosi1, L. Marselli1, M. Suleiman1, M. Tesi1, C. De Luca1, S. Del Guerra1, M. Cnop2, D.L. Eizirik2,3, P. Marchetti1;

1Department of Clinical and Experimental Medicine, University of Pisa, Pisa, Italy, 2Center for Diabetes Research, Universite Libre de Bruxelles, Brussels, Belgium, 3Diabetes Center, Indiana Biosciences Research Institute, Indianapolis, USA.

Background and aims: The function of the endocrine pancreas requires a concerted communication between islet cells to achieve a closely regulated hormone secretion. However, with the exception of a few well-characterized paracrine interactions, a comprehensive picture of the human islet interactome is still missing. New developments in the analysis of single-cell transcriptomics may allow to overcome this. Here we used an integrated single-cell dataset of islets from diabetic (T2D) and non-diabetic (ND) donors to leverage islet cell interactions in health and disease.

Materials and methods: We integrated three single-cell transcriptome datasets of human islets using our recently published pipeline. The collection contains 3,046 single cells classified in 7 different cell types (alpha, beta, delta, PP, ductal, stellate and acinar) based on expression of marker genes. The cells were divided according to donor clinical state. Next, each group was analysed with CellPhoneDB to identify significant ligand-receptor interactions occurring across cell types. The interactions were modelled using a multi-edge, multi-partite directed network with 3 distinct layers corresponding to cell types, ligands and receptors. With this data structure, each interaction corresponds to a path going from a cell type to another, through a ligand and its receptor.

Results: The most abundant cell types were alpha and beta cells (731 and 386 in ND, 807 and 414 in T2D), but, surprisingly, the least abundant stellate cell type showed the highest number of interactions. Considering the topological features of the networks, the T2D interactome displayed, as compared to ND, an increased number of interactions, with 11% more nodes (333 vs 301) and edges (2528 vs 2272). Of such interactions, the most prominent change in T2D was the establishment of 20% novel ligand-receptor pairs (346 vs 289), whereas the addition of new cell types to existing interactions was less relevant. The analysis of beta cell interactions in T2D (2709) revealed an extensive interactome rewiring, with 591 newly emerging interactions and 404 disrupted ones. Examining the most markedly altered connections in T2D, we identified ligand-receptor pairs of potential pathophysiological relevance, such as the emergence of EphA-Efrin A communications between beta and exocrine cells, or the loss of C5AR1-RPS19 between delta and beta cells.

Conclusion: This analysis represents the first reconstruction of the human islet interactome at the single-cell level. The comparison between ND and T2D islet cells allowed us to identify signatures potentially relevant for T2D pathophysiology and to generate hypotheses that merit further investigation.

Supported by: RHAPSODY, INNODIA-HARVEST

Disclosure: E. Bosi: None.

11

The transcription factor CEBPG regulates beta cell function

A. Lopez-Pascual1, A. Lindqvist1, J.A. Martínez-López2, J. Hjerling-Leffler2, N. Wierup1;

1Clinical Research Centre, Neuroendocrine Cell Biology, Lund University, Malmö, 2Medical Biochemistry and Biophysics, Karolinska Institute, Stockholm, Sweden.

Background and aims: Impaired beta cell function is central for the development of type 2 diabetes (T2D). However, the underlying molecular mechanisms have been hard to determine due to the complexity of the pancreatic islets. We previously identified T2D-affected Gene Regulatory Networks (GRNs) in human beta cells by a comprehensive analysis of data from single-cell RNA sequencing (sc-RNA-seq). This analysis pinpointed CCAAT Enhancer Binding Protein Gamma (CEBPG) as a node gene in a GRN of unfolded protein response (UPR) in the beta cells. The aim of our study was to investigate the regulatory and biological functions of CEBPG in pancreatic beta cells.

Materials and methods: We performed siRNA-mediated Cebpg knock-down (KD) and RNA-seq in INS-1 832/13 cells to assess affected pathways and genes. Moreover, we used cell stressors and glucose-stimulated insulin secretion after KD in INS-1 832/13 cells to examine the role of Cebpg in beta cell function. Also, CRISPR/Cas9 gene editing was used to generate Cebpg KO INS-1 832/13 cells. CEBPG-mediated KD was performed in human islets to investigate if our findings in cell lines can be translated to humans. Further, we performed a chromatin immunoprecipitation assay followed by sequencing (ChIP-Seq) in INS-1 832/13 cells to identify genes regulated by CEBPG in beta cells.

Results: Cebpg KD in INS-1 832/13 cells stimulated insulin mRNA expression (Ins1 1.3-fold, p<0.001; Ins2 1.4-fold, p<0.001). Glucose-stimulated insulin secretion (GSIS) was increased 72 h after Cebpg KD at 16.7 mM glucose with IBMX (2-fold, p<0.001), L-arginine (2.0-fold, p<0.001) and palmitate (1.7-fold, p<0.001), and at 2.8 mM glucose with α-KIC (2.1-fold, p<0.001). Moreover, the ER stressor thapsigargin, increased Cebpg (3.7-fold, p<0.001), Atf6 (1.8-fold, p<0.05), Atf4 (3.0-fold, p<0.001) and Xbp1s (2.9-fold, p<0.001) mRNA expression, and decreased insulin mRNA expression by 85% (Ins1 and Ins2p<0.01). Glucotoxicity (GTX, 20 mM glucose) decreased Ins1 (51%, p<0.001), Ins2 (55%, p<0.001), and Cebpg (29%, p<0.001) expression. Cebpg KD suppressed the stress response produced by thapsigargin and GTX treatments on Ins1 and Ins2 mRNA expression and blocked thapsigargin-induced Atf6 and Atf4 mRNA expression. CRISPR/Cas9 Cebpg KO increased GSIS at 16.7 mM glucose with IBMX (1.5-fold, p<0.001). Moreover, CEBPG silencing in human islets led to decreased ATF6 (80%, p<0.01), ATF4 (79%, p<0.01), DDIT3 (70%, p<0. 01) and XBP1 (78%, p<0.05) mRNA expression. Our RNA-seq studies provides the first genome-wide identification of CEBPG target genes in rat beta cells. CEBPG-bound genes in the promoter region in the control group analyzed by KEGG pathway clearly point to insulin secretion, while in GTX conditions an enrichment in genes related to maturity onset diabetes of the young (MODY) was seen. A comprehensive bioinformatic KEGG analysis using a combined approach of CEBPG-bound and Cebpg KD differentially expressed genes in INS-1 832/13 cells pointed to a dysregulation of insulin secretion, insulin resistance and insulin signaling pathways.

Conclusion: Our findings show that CEBPG is a novel regulator of insulin secretion and transcription, as well as UPR in beta cells, indicating an important role for CEBPG in beta cell function.

Supported by: EFSD/AstraZeneca Cellular Plasticity Programme 2015, Novo Nordisk, SSR, DW Sverige, Fisiograf. Sällsk. Lund, Påhlsson

Disclosure: A. Lopez-Pascual: None.

12

A long non-coding RNA that harbors a SNP associated with insulin levels regulates TGM2 gene expression in pancreatic beta cells

I. González-Moro1,2, M. Sebastián-delaCruz1,2, H. Rojas-Márquez1, A. Olazagoitia-Garmendia1,2, J. Mentxaka1, L.M. Mendoza1, A. Lluch3, J.M. Fernández-Real3,4, F.J. Ortega3,4, A. Castellanos-Rubio1,2, I. Santin1,2;

1University of the Basque Country (UPV/EHU), Leioa, 2Biocruces Bizkaia Health Research Institute, Barakaldo, 3Institut d’Investigació Biomèdica de Girona, Girona, 4Physiopathology of Obesity and Nutrition Networking Biomedical Research Centre, Madrid, Spain.

Background and aims: The Transglutaminase 2 (TGM2) is a multi-functional enzyme which catalyzes transamidation reactions that acts as a G-protein in intracellular signaling. TGM2 is expressed in human pancreatic islets and has been implicated in insulin secretion in rodent pancreatic beta cells. Recently, it has been shown that the expression of a lncRNA (named LOC107987281) that is transcribed from an intron of the TGM2 coding gene, correlates with the expression of TGM2 in several human cell lines. Against this background, the aim of the present study was to characterize the potential role of LOC107987281 in pancreatic beta cell function and in the development of diabetes.

Materials and methods: The SNP rs2076380, located within LOC107987281 sequence but intronic for TGM2, was genotyped in 557 individuals (46±10 y; 30% men) using a Taqman Genotyping assay. LOC107987281 and TGM2 expression were determined in a set of human tissues and in the human pancreatic beta cell line EndoC-βH1. Cellular localization of LOC107987281 was assessed by analyzing its expression in nuclear and whole cell RNA fractions. LOC107987281 overexpression and knockdown experiments in beta cells were performed by transfection of an overexpression vector or specific siRNAs. Promoter activation assays were performed using a luciferase plasmid under the control of the TGM2 gene promoter (promTGM2).

Results: Genotyping of rs2076380 revealed that this SNP was associated with basal and glucose-stimulated (120 min of an oral glucose tolerance test) insulin levels in women (p=0.0016), as well as with the Homeostatic Model Assessment for Insulin Resistance (HOMA-IR) (p=0.001) and fasting glucose levels (p=0.005). The expression of LOC107987281 and TGM2 was correlated across human tissues (Spearman’s R= 0.87 (0.59-0.9); p<0.001)). Regarding the localization of LOC10798728 in pancreatic beta cells, our results demonstrated that this lncRNA was preferentially nuclear, suggesting a potential role in transcriptional regulation. Indeed, inhibition of LOC107987281 in EndoC-βH1 cells using siRNAs led to a significant decrease in TGM2 expression. In contrast, overexpression of LOC107987281 induced a nearly 2-fold increase in TGM2 mRNA expression. Finally, cells co-transfected with the LOC107987281 overexpressing vector and the promTGM2 plasmid presented higher luciferase activity than control cells, suggesting that LOC107987281 binds to the promoter of TGM2 to induce its expression.

Conclusion: Our study reveals that LOC10798728 harbors a SNP that is associated with several parameters related to pancreatic beta cell function and type 2 diabetes (T2D). In addition, our results show that LOC10798728 regulates TGM2 expression in pancreatic beta cells. Taking into account the role of TGM2 in insulin secretion, it is plausible to think that LOC10798728 might be implicated in the regulation of insulin production and release.

Supported by: EFSD/JDRF/Lilly Programme on Type 1 Diabetes Research 2019 and Spanish Ministry of Science, Innovation and Universities (PID2019-104475GA-I00)

Disclosure: I. González-Moro: None.

OP 03 Many faces of diabetic pregnancy

13

Sweet pregnancy: digital solutions for an effective management of diabetic pregnancies

M. Löhnert, S. Stichling, C. Eberle;

Hochschule Fulda - University of Applied Sciences, Fulda, Germany.

Background and aims: Hyperglycemia in Pregnancy (HIP) is a global concern and closely associated with adverse outcomes in mothers and offspring, such as pregnancy and birth complications respectively comorbidities, as well as metabolic and cardiovascular diseases. Influences during the pre- and perinatal period play a decisive role for health and illness in the course of later life - for mothers and offspring. In short and in view of the offspring, mechanisms of transgenerational programming (“fetal programming”) may lead to "mal-programming" in organ functions and metabolic regulations. In particular HIP may contribute to such “mal-programming” respectively short- as well as long-term consequences. To optimize the management of diabetic pregnancies, sustainable strategies are needed. Technological solutions present innovative opportunities to improve clinical care for women with diabetes mellitus (DM) in pregnancy. The aim of this review was to examine the clinical effectiveness of current diabetes technologies in diabetic pregnancies. We considered maternal glycemic outcomes as well as pregnancy and birth-related outcomes, and neonatal outcomes.

Materials and methods: Empirical analysis were performed by investigating clinical studies in relevant databases including MEDLINE (PubMed), Cochrane Library, EMBASE, CINAHL and Web of Science Core Collection (2008 - September 2020). Study quality was assessed using “Effective Public Health Practice Project” (EPHPP) tool. Of m=974 records, we analyzed m=15 randomized controlled trials (RCT), m=3 randomized crossover trials (RcrT), m=2 cohort studies, and m=2 controlled clinical trials (CCT) [excluded: duplicates, wrong topic/population/intervention]. We assessed m=9 studies with strong, m=11 with moderate, and m=2 with weak quality.

Results: Overall, the various diabetes technologies seem to have a particularly positive effect on maternal glycemic control in all types of diabetes, as shown by studies of strong and moderate quality. In detail, we found:

T1D (m=9 studies; n=1,142 patients): Improvements towards HbA1c (P<0.05), maternal and neonatal hypoglycemia, insulin dose, fewer caesarean sections, birth weight, large-for-gestational age (LGA) using continuous glucose monitoring (CGM; n=435). Continuous subcutaneous insulin infusion (CSII; n=643) significantly improved HbA1c values (P=0.002) and insulin dose (P=0.02). GDM (m=10 studies; n=1,164 patients): Positive effects regarding HbA1c (P=0.006), fasting blood glucose (FBG), birth weight, macrosomia, LGA, neonatal hypoglycemia using CGM (n=301). Significant Improvements (P<0.05) by mHealth-Apps (n=813) in HbA1c (P<0.001), FBG (P<0.001), off-target blood glucose measurement (P<0.001), compliance (P<0.001) as well as positive trends regarding hypertension, preeclampsia, preterm birth. T1D/T2D (m=3 studies; n=267 patients): Improvements regarding HbA1c (P=0.007), caesarean section, neonatal hypoglycemia by CGM (n=225). Improving trends in terms of HbA1c, birth weight, estimated fetal weight through CSII (n=42).

Conclusion: Digital solutions in the management of diabetic pregnancies showed clearly potential for remotely improving diabetic metabolic conditions in pregnant women and therefore in their offspring. Against this background short- and long-term outcomes in mothers and offspring may be improved, subsequently. Further research is urgently required.

Supported by: DFG Grant EB 440/4-1

Disclosure: M. Löhnert: None.

14

Decreased insulin sensitivity during pregnancy after assisted reproductive therapy in women in the Stork cohort

E. Qvigstad1,2, S.D. Steintorsdottir1, K. Godang1, M.-C.P. Roland3, T. Lekva4;

1Dept of Endocrinology, Morbid Obesity and Preventive Medicine, Oslo University Hospital, Oslo, 2Dept of Endocrinology, Morbid Obesity and Preventive Medicine, University of Oslo, Oslo, 3Dept of Obstetrics and Gynecology, Oslo University Hospital, Oslo, 4Research Institute of Internal Medicine, University of Oslo, Oslo, Norway.

Background and aims: Assisted reproductive therapy (ART) seems to increase the risk of gestational diabetes (GDM) during gestation and glucose intolerance after pregnancy, although the nature of the association remains unclear. Older age and the treatment per se could be primary determinants of altered glucose metabolism. To gain more insight in this association we have investigated levels of glucose and insulin, beta-cell function (BCF) and insulin sensitivity (IS) in singleton pregnancies, comparing women with normal glucose tolerance (NGT), ART and GDM from the STORK cohort.

Materials and methods: 1031 healthy pregnant Norwegian women attended OGTTs at gestational week (GW) 14-16 and 30-32. In the women that had available blood samples and OGTT results, 75 women received ART, 104 women were diagnosed with GDM (WHO 2013 criteria), and 847 women served as controls (NGT). Eleven subjects that were diagnosed with GDM in the ART group were removed for data analysis (n= 64 women in the ART group). None of the participating women required antidiabetic medication during the study period. Glucose was measured using the hexokinase method at an accredited clinical chemistry laboratory. Insulin levels were measured using radioimmunoassay. We calculated indices of BCF (ISSI-2) and IS (Matsuda) from the 5-point OGTTs, and performed linear regression analysis to adjust for BMI and age.

Results: The ART women were 32.8 years (SD 3.5) at inclusion, ie. 1.8 years older than the NGT women (p<0,001), but had similar age to the GDM women. The reasons for ART were: 30% male factors, 21% endometriosis, primary/unspecified 21%, 11% PCOS, and 15% combined etiologies. Although fasting glucose and fasting insulin levels in the ART women were at an intermediate level compared to NGT and GDM women, only fasting insulin levels at GW 14-16 differed significantly. Both BCF and IS were significantly lower than in NGT women in GW 14-16, but at GW 30-32 the difference was significant only for IS. BCF levels in ART women were higher at both time points compared to the GDM women. The influence of ART persisted for IS after adjustment for age and BMI.

Conclusion: The women with ART demonstrated reduced IS in pregnancy compared with the NGT women, and a similar trend for BCF in early pregnancy. In late pregnancy insulin measurements and indices of BCF or IS in the ART subjects were more similar to the NGT women than to the GDM women. Our data suggest that ART treatment exerts a negative influence on insulin sensitivity already during gestation, which persists after adjustment for age and BMI. However, ART did not result in high rates of glucose intolerance, indicating compensatory effects such as increased insulin secretion.

figured

Supported by: JS Kvanes fund for Diabetes Research, Norway

Disclosure: E. Qvigstad: Grants; 4200 euros for biomarker analysis from J.S.Kvanes' fund for diabetes research, Norway.

15

Maternal C-peptide reappearance in type 1 diabetes pregnancy: Evidence of beta cell regeneration or fetal hyperinsulinism?

C.L. Meek1, R. Oram2, T. McDonald2, D.S. Feig3, A.T. Hattersley2, H.R. Murphy4;

1University of Cambridge, Cambridge, UK, 2University of Exeter, Exeter, UK, 3University of Toronto, Toronto, Canada, 4University of East Anglia, Norwich, UK.

Background and aims: Reports of increasing endogenous insulin secretion have occurred in pregnant women with established type 1 diabetes. However, the cause of this phenomenon, and its relevance to maternal-fetal pregnancy outcomes are unclear. The aim of this study was to assess longitudinal patterns of maternal C-peptide concentration using a highly sensitive electrochemiluminescent assay to examine the hypothesis of pregnancy-induced beta cell regeneration in women with type 1 diabetes.

Materials and methods: The Continuous Glucose Monitoring in Women with Type 1 Diabetes in Pregnancy Study (CONCEPTT) was a multinational randomised controlled trial to assess the effects of real-time CGM in comparison to standard care (capillary blood glucose monitoring) in pregnant women with type 1 diabetes. Women with type 1 diabetes who delivered a liveborn singleton infant and gave samples for the biorepository were included in this analysis. Highly-sensitive direct and solid-phase competitive electrochemiluminescent immunoassays were used to measure C-peptide in maternal serum (12, 24, 34 wks; n=127) and cord blood (n=85). Associations were made with adjudicated pregnancy outcomes using adjusted and unadjusted logistic regression.

Results: Three discrete patterns of maternal C-peptide trajectory were identified (see figure 1): Pattern 1 undetectable throughout pregnancy (n=74; 58%); Pattern 2 detectable at baseline (n=22; 17%); Pattern 3 undetectable C-peptide (12 & 24 weeks), became detectable at 34 weeks (n=31; 24%). Women in patterns 1 and 3 had comparable baseline characteristics. Women in pattern 2 had shorter duration of diabetes (median 10.6 years vs 16.9 years overall; p<0.001). Offspring of women in pattern 3 had higher rates of neonatal hypoglycemia (42% vs 14%; p=0.001), large-for-gestational-age (90% vs 60%; p=0.002) and elevated cord blood C-peptide (see figure 1; geometric mean 1319 vs 718 pmol/l; p=0.007) compared to offspring from women in pattern 1, despite comparable 34-week glycemia.

Conclusion: Increased C-peptide in maternal serum at 34 weeks suggests fetal hyperinsulinism with fetal-to-maternal transfer not improved maternal beta cell function. First appearance of C peptide in late pregnancy could identify pregnancies at highest risk of neonatal complications.

figuree

Supported by: EFSD /Novo Nordisk Foundation Future Leaders Award 2019, JDRF 17/2011/533; JDRF 80/2010/585; DUK-HKF 17/0005712; EASD- NNF19SA058974; DUK-HKF 16/0005529

Disclosure: C.L. Meek: None.

16

Maternal efficacy, safety, and pregnancy outcomes with degludec vs detemir in the treatment of pregnant women with type 1 diabetes: an international, multicentre, randomised trial

E.R. Mathiesen1,2, R. Corcoy3, F. Dunne4, M. Ekelund5, D.S. Feig6, M. Hod7, T. Jia5, B. Kalyanam8, S. Kar8, A. Kautzky-Willer9, P. Damm1,2, on behalf of the EXPECT study group;

1Departments of Endocrinology and Obstetrics, Center for Pregnant Women with Diabetes, Copenhagen, Denmark, 2Department of Clinical Medicine, University of Copenhagen, Copenhagen, Denmark, 3Department of Medicine, Universitat Autònoma de Barcelona, Barcelona, Spain, 4College of Medicine, Nursing and Health Sciences, National University of Ireland Galway, Galway, Ireland, 5Novo Nordisk A/S, Søborg, Denmark, 6Department of Medicine, University of Toronto, Toronto, Canada, 7Faculty of Medicine, Tel Aviv University, Tel Aviv, Israel, 8Novo Nordisk Service Centre India Private Ltd., Bangalore, India, 9Division of Endocrinology & Metabolism, Medical University of Vienna, Vienna, Austria.

Background and aims: To compare the efficacy and safety of insulin degludec (degludec) versus insulin detemir (detemir) in pregnant women with type 1 diabetes (T1D).

Materials and methods: EXPECT was an open-label, randomised trial of women aged ≥18 years with T1D and previously treated with insulin, who were at 8-13 weeks’ gestation or planned to become pregnant within 52 weeks. Women were randomised to degludec once daily or detemir 1-2 times daily, both with insulin aspart 2-4 times daily. The primary analysis aimed to demonstrate the non-inferiority (margin of 0.4%) of degludec to detemir with respect to the last planned glycated haemoglobin (HbA1c) measurement prior to delivery (>16 weeks’ gestation) using ANCOVA. Secondary endpoints were maternal efficacy, safety, and pregnancy outcomes.

Results: In total, 225 women (degludec: 111; detemir: 114) were randomised including 144 who were, and 81 who planned to become, pregnant. Mean (±SD) HbA1c at treatment baseline was 6.8% (±0.7%) and 6.7% (±0.8%) with degludec and detemir, respectively. Over the trial, 188 women (degludec: 92; detemir: 96) with singleton pregnancies were included and there were 171 live-born infants (degludec: 86; detemir: 85), with no perinatal deaths (between ≥20 week’s gestation and <1 week after delivery). Estimated mean HbA1c prior to delivery was 6.23% with degludec and 6.34% with detemir (estimated treatment difference: −0.11% [−0.31; 0.08]95% CI, confirming non-inferiority). See Table for supportive maternal safety and pregnancy outcomes.

Conclusion: In pregnant women with T1D, degludec was non-inferior to detemir with respect to HbA1c prior to delivery. Hypoglycaemia rates and pregnancy outcomes were comparable between insulins.

figuref

Clinical Trial Registration Number: NCT03377699

Supported by: Novo Nordisk A/S

Disclosure: E.R. Mathiesen: Grants; Novo Nordisk. Lecture/other fees; Novo Nordisk, Sanofi, Lilly.

17

Epigenetic alterations in offspring born to mothers with type 1 diabetes (the EPICOM study)

S. Knorr1,2, A. Skakkebæk2,3, J. Just2, C. Trolle2,4, S. Vang2, Z. Lohse5, B. Bytoft6, P. Damm6, K. Højlund5, D.M. Jensen5, C.H. Gravholt2;

1Steno Diabetes Center Aarhus, Aarhus, 2Department of Molecular Medicine, Aarhus, 3Department of Clinical Genetics, Aarhus, 4Department of Endocrinology and Internal Medicine, Aarhus, 5Steno Diabetes Center Odense, Odense, 6Center for Pregnant Women with Diabetes, Department of Obstetrics, Rigshospitalet, Copenhagen, Denmark.

Background and aims: Offspring born to women with pregestational type 1 diabetes (T1D) are exposed to an intrauterine hyperglycaemic milieu and has later in life an increased risk of metabolic disease. We hypothesize that in utero exposure to T1D induces epigenetic and transcriptome changes that may alter the offspring’s risk of metabolic disorders later in life.

Materials and methods: EPICOM (Epigenetic, Genetic and Environmental Effects on Growth, Cognitive Functions and Metabolism) is a Danish prospective nationwide follow-up study in 278 offspring of women with T1D and a control group of 303 offspring of mothers without diabetes. In the present study, we performed whole-blood DNA methylation profiling using the 450K-Illumina Infinium assay in addition to RNA-Seq transcriptome profiling among the oldest participants of the EPICOM cohort (n=40, mean age 19.8 (SD 0.9)). Besides fulfilling the matching criteria for inclusion in the EPICOM study, we made sure that they did not differ in respect to body mass index, glucose tolerance, HbA1c, cholesterol or triglyceride levels.

Results: We identified 1,043 differentially methylated positions (DMPs, p<0.005) between controls and offspring born to women with T1D. Of these DMPs, 14 showed a substantial change in methylation (9 hypermethylated and 5 hypomethylated, |deltaM|>1). In the transcriptome analysis, we found 38 up-regulated genes and 1 down-regulated gene (p<0.005 and absolute logFC≥0.3). One of the up-regulated genes, ARHGDIA, was found close to, and possibly under epigenetic control of one of the hypomethylated DMPs (p<0.005). However, no association between the level of methylation and ARHGDIA expression was found in the correlation analysis (rho=-0.13, p=0.42). This gene has previously been described to be associated with ß-cell dysfunction in metabolic disease. Furthermore, using Gene Set Enrichment analysis on gene expression data, we observed enrichment in ontologies or pathways relating to diabetes, carbohydrate metabolism, glucose metabolism disease and pathways including MAPK1/MAPK3 and MAPK family signalling and genes relating to type 1 diabetes, obesity and atherosclerosis.

Conclusion: Our result could indicate methylation changes as a possible pathway between the intrauterine environment, later life methylation status and changes in RNA expression.

Supported by: EFSD/Lilly European Diabetes Research Programme 2015, Fam. Hede Nielsen

Disclosure: S. Knorr: None.

18

Maternal exercise in gestational diabetes has sex-specific effects on offspring adiposity and beta cell function

N. Boonpattrawong, S. Schaffner, A. Jang, M.S. Kobor, I. Laher, A.M. Devlin;

University of British Columbia, Vancouver, Canada.

Background and aims: Adult offspring from dams with gestational obesity and diabetes have greater adiposity and insulin resistance. The goal of this study was to determine if maternal exercise mitigates the adverse effects of maternal gestational diabetes with obesity on offspring metabolic health.

Materials and methods: Female (C57BL/6N) mice were fed from weaning a control (MC; 10%kcal fat) or western (MW; 45% kcal fat) diet to induce excess adiposity and glucose intolerance (obese dams). At 13 weeks, mice were bred and maintained on the diets, with or without access to a running wheel (exercise), throughout pregnancy and lactation. Male and female offspring were studied at two timepoints; i) postnatal day 7 (P7); and ii) adulthood (age 4-5months, fed the control or western diet from weaning). At P7, pancreatic islets were isolated for ex vivo assessment of insulin secretion and β cell mass was quantified by immunohistochemistry. In adult offspring, physiological assessments of glucose tolerance, β cell function and insulin sensitivity were conducted. DNA methylation was assessed by reduced representation bisulfite sequencing of mature adipocytes from male offspring.

Results: At P7, islets from male and female pups from exercised dams fed the control diet (MCE) had greater insulin secretion when stimulated by low glucose (2.6mM; p<0.01) and KCl (40mM; p<0.05) compared to pups from sedentary dams (MCS). Islets from female MCE offspring had greater insulin secretion when stimulated with high glucose (16.7mM); no differences were observed in male offspring. This was accompanied by increased (p<0.01) β cell mass in both male and female P7 offspring from exercised control and obese dams compared to those from sedentary dams. Adult male offspring fed the control diet from obese dams that exercised (MWE) had greater (p<0.05) retroperitoneal, mesenteric, and subcutaneous adipose tissue than those from obese sedentary dams (MWS). Adult female offspring from MWE dams had greater retroperitoneal adipose tissue (p<0.05).

Despite an increase in adiposity in offspring from exercised dams, the offspring did not have impaired glucose tolerance, insulin sensitivity or β cell function. We observed 371 differentially methylated CpG sites (FDR<0.05 and delta beta>10%); 95 CpG sites had increased methylation and 276 CpG sites had decreased methylation in MWE male offspring compared to MWS male offspring. This included sites involved in adipocyte differentiation (Wnt10b, Sirt4), fatty acid transporter (Slc25a29), adipocyte cell proliferation (Mybl2), cell adhesion (Cdc181), and lipid synthesis (Sirt4). We also challenged some of the adult offspring with a western diet from weaning. This diet exacerbated the effect of maternal obesity on male offspring glucose intolerance. However, maternal exercise improved glucose tolerance (p<0.05) from MWE offspring compared to offspring from MWS dams.

Conclusion: Our findings suggest that the effect of maternal exercise on offspring metabolic health is sex-specific. Islets from female, but not male, offspring have improved β cell function in response to high-glucose, which appear to protect the female offspring against obesogenic diet-induced glucose intolerance later in adulthood.

Supported by: CIHR

Disclosure: N. Boonpattrawong: None.

OP 04 GLP-1 receptor agonism: higher dose, combination therapy, or both?

19

Effect of semaglutide 2.4 mg on glucose metabolism and body weight in adults with overweight or obesity and type 2 diabetes in the STEP 2 trial

S.D. Pedersen1, M. Davies2, L. Færch3, O.K. Jeppesen3, A. Pakseresht3, L. Perreault4, J. Rosenstock5, I. Shimomura6, A. Viljoen7, T. Wadden8, I. Lingvay9;

1C-ENDO Diabetes & Endocrinology Clinic Calgary, Calgary, Canada, 2Diabetes Research Centre, University of Leicester, Leicester, UK; NIHR Leicester Biomedical Research Centre, Leicester, UK, 3Novo Nordisk A/S, Søborg, Denmark, 4Department of Medicine, Division of Endocrinology, Metabolism and Diabetes, University of Colorado Anschutz Medical Campus, Aurora, USA, 5Dallas Diabetes Research Center at Medical City, Dallas, USA, 6Department of Metabolic Medicine, Graduate School of Medicine, Osaka University, Osaka, Japan, 7Borthwick Diabetes Research Centre, Lister Hospital, Stevenage, UK, 8Department of Psychiatry, Perelman School of Medicine, University of Pennsylvania, Philadelphia, USA, 9Departments of Internal Medicine/Endocrinology and Population and Data Sciences, University of Texas Southwestern Medical Center, Dallas, USA.

Background and aims: The STEP 2 trial showed mean weight loss of 9.6% with semaglutide 2.4 mg vs 7.0% with semaglutide 1.0 mg and 3.4% with placebo (both p<0.0001). We present detailed glucose metabolism outcomes.

Materials and methods: Adults with BMI ≥27 kg/m2, type 2 diabetes and HbA1c 7-10% (53-86 mmol/mol) were randomised to 68 weeks’ once-weekly s.c. semaglutide 2.4 mg (N=404), 1.0 mg (N=403) or placebo (N=403). Glucose metabolism outcomes were secondary or exploratory endpoints. Treatment comparisons were done using an ANCOVA model with treatment and stratification factors as fixed effects and baseline endpoint values as covariate (all patients in the full analysis set).

Results: At baseline, patients had a mean age of 55 years, body weight 99.8 kg, HbA1c 8.1% (65.3 mmol/mol) and diabetes duration 8.0 years; 88% were on 1-2 oral antihyperglycaemic drugs. From weeks 0 to 68, semaglutide 2.4 mg and 1.0 mg reduced HbA1c to a greater extent vs placebo (estimated treatment difference -1.2% [-13.5 mmol/mol] and -1.1% [-11.8 mmol/mol]; both p<0.0001) and improved fasting plasma glucose, HOMA-IR and -β, and reduced oral antihyperglycaemic drug use (Table). With semaglutide 2.4 mg, more patients achieved HbA1c <7.0% and ≤6.5%, and composite endpoints of HbA1c <7.0% with weight loss ≥10% or ≥15% vs semaglutide 1.0 mg or placebo (Table). Exploratory analyses showed that a greater proportion of patients receiving semaglutide at either dose decreased their oral antihyperglycaemic medication intensity vs placebo by week 68.

Conclusion: Weight loss with semaglutide 2.4 mg was accomplished with improvements in insulin resistance and β-cell function, two key mechanistic drivers and pathophysiologic abnormalities that cause type 2 diabetes and fuel diabetes progression. With semaglutide 2.4 mg, more patients achieved HbA1c <7.0% and weight loss ≥10% (composite endpoint) vs placebo or semaglutide 1.0 mg, and reduced oral antihyperglycaemic drug use vs placebo.

figureg

Clinical Trial Registration Number: NCT03552757

Supported by: Funded by Novo Nordisk A/S

Disclosure: S.D. Pedersen: Employment/Consultancy; Novo Nordisk, Janssen, AstraZeneca, Abbott, HLS therapeutics,Bayer, and Dexcom (all consultancy only). Grants; Lilly, AstraZeneca, Abbott, Boehringer Ingelheim, and Sanofi. Honorarium; Novo Nordisk, Janssen, Lilly, Merck, Bausch Health, AstraZeneca, Abbott, Boehringer Ingelheim, Sanofi, HLS Therapeutics, Bayer, and Dexcom. Lecture/other fees; Novo Nordisk, Janssen, Lilly, Merck, Bausch Health, AstraZeneca, Abbott, Boehringer Ingelheim, Sanofi, HLS Therapeutics, Bayer, and Dexcom. Non-financial support; (travel to meetings) Novo Nordisk, Janssen, Lilly, Bausch Health, AstraZeneca, Boehringer Ingelheim, and Sanofi. Other; (fees for clinical trials) Novo Nordisk, Lilly, AstraZeneca, Sanofi, Prometic, and Pfizer.

20

Tirzepatide, a dual GIP/GLP-1 receptor agonist, is effective and safe when added to basal insulin for treatment of type 2 diabetes (SURPASS-5)

D. Dahl1, Y. Onishi2, P. Norwood3, R. Huh4, H. Patel4, A. Rodriguez4;

1Gemeinschaftspraxis für Innere Medizin und Diabetologie, Hamburg, Germany, 2The Institute of Medical Science, Asahi Life Foundation, Tokyo, Japan, 3Valley Endocrine and Research, Fresno, USA, 4Eli Lilly and Company, Indianapolis, USA.

Background and aims: Tirzepatide (TZP) is a novel dual glucose-dependent insulinotropic polypeptide/glucagon-like peptide-1 (GIP/GLP-1) receptor agonist in development for treatment of type 2 diabetes (T2D). Efficacy and safety of TZP vs placebo was assessed in people with T2D as an add-on to titrated insulin glargine with or without metformin.

Materials and methods: In this double-blind, placebo-controlled, 40-week Phase 3 study, 475 people with T2D (mean baseline age 60.6 y; T2D duration 13.3 y; HbA1c 67.4 mmol/mol [8.31%]; BMI 33.4 kg/m2) were randomised (1:1:1:1) to TZP (5 mg, 10 mg, 15 mg) or placebo, as an add-on to their existing therapy. Primary efficacy measure was mean change in HbA1c from baseline at 40 weeks. Secondary measures included change in fasting serum glucose (FSG) and body weight (BW), and proportion of people achieving HbA1c and BW loss goals.

Results: All three TZP doses were superior to placebo in mean change from baseline in HbA1c, FSG, BW, and percentage patients achieving all HbA1c and BW loss targets at week 40 (Table). LSM treatment difference vs placebo (95% CI) in HbA1c for all TZP groups were statistically significant (p<0.001 all doses): for TZP 5 mg (-14.2 [-16.6,-11.7] mmol/mol or -1.30 [-1.52, -1.07] %); TZP 10 mg (-18.1 [-20.6,-15.7] mmol/mol or -1.66 [-1.88, -1.43] %); and TZP 15 mg (-18.1 [-20.5,-15.6] mmol/mol or -1.65 [-1.88, -1.43]%). The mean insulin glargine dose change from baseline was 4.3, 2.3, -3.9, and 25.6 U/day for TZP 5, 10, 15 mg, and placebo groups, respectively. TZP was generally well tolerated and the most common adverse events were gastrointestinal and a vast majority were mild-to-moderate in severity. Diarrhoea, nausea, and vomiting were reported in 12-21%, 13-18%, and 7-13% of patients treated with TZP vs 10%, 3%, and 3% in the placebo arm, respectively. Incidence of hypoglycaemia (blood glucose <3.0 mmol/L) or severe hypoglycaemia did not differ between TZP and placebo groups (14-19% vs 13%). Three episodes of severe (level 3) hypoglycaemia were observed in 2 episodes in TZP 10 mg group and 1 in TZP 15 mg group.

Conclusion: In conclusion, TZP demonstrated superior and clinically meaningful improvements in glycemic control and BW loss without increasing hypoglycaemia vs placebo in patients with T2D when added to titrated basal insulin.

figureh

Clinical Trial Registration Number: NCT04039503

Supported by: Eli Lilly and Company

Disclosure: D. Dahl: Grants; Novo Nordisk, Afimmune, Novartis, Eli Lilly and Company. Lecture/other fees; Eli Lilly and Company.

21

Semaglutide reduced cardiovascular events regardless of metformin use: a post hoc exploratory subgroup analysis of SUSTAIN 6 and PIONEER 6

M. Husain1, A. Consoli2, A. De Remigis3, A.S.P. Meyer3, S. Rasmussen3, S. Bain4;

1Ted Rogers Centre for Heart Research, University of Toronto, Toronto, Canada, 2D’Annunzio University of Chieti-Pescara, Pescara, Italy, 3Novo Nordisk A/S, Søborg, Denmark, 4Swansea University Medical School, Swansea, UK.

Background and aims: Post hoc analyses of liraglutide, dulaglutide and empagliflozin cardiovascular outcomes trials (CVOTs) showed reductions in major adverse cardiovascular events (MACE) vs placebo regardless of baseline metformin use. The aim of this post hoc analysis of the SUSTAIN 6 and PIONEER 6 CVOTs was to determine if semaglutide provided cardiovascular (CV) benefits independently of baseline metformin use in adults with type 2 diabetes at high risk of a CV event.

Materials and methods: Pooled data from the SUSTAIN 6 and PIONEER 6 trials were used to assess the following CV and metabolic outcomes in the two MET subgroups (baseline metformin, yes/no): time to first CV outcomes (3-point MACE [a composite of CV death, nonfatal myocardial infarction or nonfatal stroke], CV death, all-cause mortality and hospitalisation for heart failure), metabolic outcomes (change in glycated haemoglobin and body weight from baseline to week 80 for SUSTAIN 6 and week 83 for PIONEER 6) and safety outcomes (serious adverse events and severe hypoglycaemic episodes). CV analyses were adjusted for baseline characteristics in a Cox proportional hazards model. Consistency of treatment effects across subgroups was assessed using interaction p-values, with pinteraction <0.05 indicating a statistically significant difference.

Results: Compared with subjects with baseline metformin (n=4,881; 75%), those without (n=1,599; 25%) were at a higher CV risk, were older (64.8 vs 67.1 years), had lower eGFR (79.4 vs 61.9 mL/min/1.73 m2) and higher body weight (90.8 vs 93.6 kg) at baseline. Fewer subjects with than without baseline metformin received concomitant insulin (48.2% vs 70.7%), and more received sulphonylureas (40.5% vs 28.7%). Semaglutide reduced the risk of 3-point MACE vs placebo in both subgroups (HRs with/without metformin: 0.70/0.86; 1-year absolute risk reduction with/without metformin: 1.05%/0.78%). There was no significant interaction between treatment effect (semaglutide vs placebo) on MACE (pinteraction: 0.36) or other CV outcomes and subgroup (metformin, yes/no; Figure A). Glycated haemoglobin and body weight reductions from baseline with semaglutide vs placebo were similar in both metformin subgroups (Figure B).

Conclusion: No significant differences in outcomes could be detected between those subjects receiving metformin and those who were not. These findings indicate that the CV and metabolic benefits of semaglutide can be observed regardless of metformin use at baseline.

figurei

Clinical Trial Registration Number: NCT01720446 and NCT02692716

Supported by: Novo Nordisk research support

Disclosure: M. Husain: Employment/Consultancy; Dr. Husain has served as a paid consultant for Novo Nordisk on advisory boards. Grants; Dr. Husain has previously received investigator-initiated research grants from Novo Nordisk. Honorarium; Dr. Husain has received honoraria from Novo Nordisk for crafting and delivering educational content related to diabetes, obesity, cardiovascular disease and GLP-1RA.

22

Efficacy and safety of tirzepatide versus semaglutide once weekly as add-on therapy to metformin in people with type 2 diabetes (SURPASS-2)

M.J. Davies1, J.P. Frias2, J. Rosenstock3, F. Pérez Manghi4, L. Fernández Landó5, B.K. Bergman5, B. Liu5, X. Cui5, K. Brown5;

1Diabetes Research Centre, Leicester Diabetes Centre – Bloom, University of Leicester, Leicester, UK, 2National Research Institute, Los Angeles, USA, 3Dallas Diabetes Research Center at Medical City, Dallas, USA, 4CINME S.A., Buenos Aires, Argentina, 5Eli Lilly and Company, Indianapolis, USA.

Background and aims: The novel dual glucose-dependent insulinotropic polypeptide and glucagon-like peptide-1 receptor agonist tirzepatide (TZP) is in development for type 2 diabetes (T2D). Efficacy and safety of once weekly TZP vs semaglutide (SEMA) was assessed in people with T2D on background metformin.

Materials and methods: In this open-label, 40-week Phase 3 study, people with T2D (N=1879; mean baseline HbA1c: 67 mmol/mol [8.28%], age: 56.6 years, T2D duration: 8.6 years, body mass index: 34.2 kg/m2) were randomised (1:1:1:1) to once weekly TZP (5, 10, 15 mg) or SEMA (1 mg). Primary efficacy objective was noninferiority of TZP 10 and/or 15 mg vs SEMA for mean change in HbA1c from baseline at 40 weeks. Secondary objectives included noninferiority (TZP 5 mg) for HbA1c change and superiority (all TZP doses) for change in HbA1c, body weight and fasting serum glucose, and proportion of patients with HbA1c <53 mmol/mol (<7%), ≤48 mmol/mol (≤6.5%) and <39 mmol/mol (<5.7%; except TZP 5 mg) and body weight loss ≥5%, ≥10%, ≥15% at 40 weeks.

Results: At 40 weeks, all TZP doses were superior to SEMA in mean HbA1c change from baseline (Table). LSM treatment differences vs SEMA (95% CI) were -2.5 mmol/mol (-3.9, -1.1) (-0.23% [‑0.36, ‑0.10]) for TZP 5 mg, -5.6 mmol/mol (-7.0, -4.1) (-0.51% [‑0.64, ‑0.38]) for TZP 10 mg, and -6.6 mmol/mol (-8.0, -5.1) (-0.60% [-0.73, -0.47]) for TZP 15 mg (p<0.001, all TZP doses). TZP 5, 10 and 15 mg were also superior to SEMA in achieving all HbA1c targets. Fasting serum glucose significantly decreased from baseline with TZP 5, 10 and 15 mg vs SEMA (-0.41 mmol/L [-7 mg/dL], -0.72 mmol/L [-13 mg/dL], and -0.82 mmol/L [-15 mg/dL]; p≤0.001, all TZP doses). Significant body weight loss was achieved with TZP 5, 10 and 15 mg vs SEMA (-1.7 kg [-2.6, -0.7], -4.1 kg [-5.0, -3.2], -6.2 kg [-7.1, -5.3]; p<0.001, all TZP doses). A greater proportion of patients achieved body weight loss ≥5%, ≥10% and ≥15% with TZP vs SEMA (p≤0.002, all TZP doses). TZP was well tolerated. The most common adverse events were gastrointestinal and mostly mild to moderate in severity (nausea: 17-22% vs 18%, diarrhoea: 13-16% vs 12%, vomiting: 6-10% vs 8%). Hypoglycaemia <54 mg/dL or severe hypoglycaemia events were reported in 4 (0.9%), 1 (0.2%) and 8 (1.7%) patients treated with TZP vs 2 (0.4%) with SEMA. Two patients (TZP 5 and 15 mg) had one episode of severe hypoglycaemia.

Conclusion: In conclusion, all TZP doses demonstrated superior and clinically meaningful improvement in glycaemic control and substantial weight loss vs once weekly SEMA 1 mg in people with T2D treated with metformin.

figurej

Clinical Trial Registration Number: NCT03987919

Supported by: Eli Lilly and Company

Disclosure: M.J. Davies: Employment/Consultancy; Novo Nordisk, Sanofi-Aventis, Eli Lilly and Company, Merck Sharp & Dohme, Boehringer Ingelheim, AstraZeneca, Janssen. Grants; Novo Nordisk, Sanofi-Aventis, Eli Lilly and Company, Boehringer Ingelheim, AstraZeneca, Janssen. Lecture/other fees; Novo Nordisk, Sanofi-Aventis, Eli Lilly and Company, Merck Sharp & Dohme, Boehringer Ingelheim, AstraZeneca, Janssen, Napp Pharmaceuticals, Mitsubishi Tanabe Pharma Corporation, Takeda Pharmaceuticals International. Other; Novo Nordisk, Sanofi-Aventis, Eli Lilly and Company, Merck Sharp & Dohme, Boehringer Ingelheim, AstraZeneca, Janssen, Servier, Gilead Sciences.

23

Efficacy and safety of GLP-1RAs with or without baseline SGLT-2i: post hoc analysis of the SUSTAIN 10 trial

M. Capehorn1, A.-M. Catarig2, O. Frenkel2, M. Marre3,4, H. Price5, A.L. Søndergaard2, R. Pratley6;

1Rotherham Institute for Obesity, Rotherham, UK, 2Novo Nordisk A/S, Søborg, Denmark, 3Clinique Ambroise Paré, Neuilly-sur-Seine, France, 4Centre de Recherche des Cordeliers, INSERM, Université de Paris, Sorbonne Université, Paris, France, 5West Hampshire Community Diabetes Service, Lyndhurst, UK, 6AdventHealth Translational Research Institute, Orlando, USA.

Background and aims: As glucagon-like peptide-1 receptor agonists (GLP-1RAs) and sodium-glucose cotransporter-2 inhibitors (SGLT-2is) are often used together in clinical practice, data on the safety and efficacy of these two classes in combination are valuable to clinical decision-making, yet such data are sparse. The SUSTAIN 10 trial analysed the efficacy and safety of the GLP-1RAs semaglutide and liraglutide in subjects with type 2 diabetes. This post hoc analysis of SUSTAIN 10 examined clinical outcomes with these GLP-1RAs in subjects not receiving an SGLT-2i and in those on background SGLT-2i treatment.

Materials and methods: Randomisation in the SUSTAIN 10 trial was stratified based on background medication. In this post hoc analysis, treatment effects with once-weekly semaglutide 1.0 mg or once-daily liraglutide 1.2 mg on HbA1c and body weight (BW) without the confounding effect of rescue medication use (‘on-treatment without rescue medication’) were assessed in subjects without or with SGLT-2i use at screening. Effects on systolic blood pressure (SBP) were also assessed in these subgroups, as was safety (premature treatment discontinuation due to adverse events [AEs]) (‘on treatment’ data).

Results: Regardless of SGLT-2i use (without SGLT-2i: n=435; with SGLT-2i: n=142), in the semaglutide 1.0 mg and liraglutide 1.2 mg treatment arms, baseline use of metformin (without SGLT-2i: 97.2% and 94.0%; with SGLT-2i: 93.2% and 91.3%, respectively) and sulphonylureas (without SGLT-2i: 50.7% and 51.4%; with SGLT-2i: 35.6% and 31.9%) was similar. Regardless of SGLT-2i use, there were reductions from baseline to week 30 with semaglutide 1.0 mg and liraglutide 1.2 mg in HbA1c (without SGLT-2i: 1.8 and 1.1 %-points, respectively; with SGLT-2i: 1.6 and 0.7 %-points, respectively; Figure 1A), BW (without SGLT-2i: 5.9 and 2.2 kg; with SGLT-2i: 5.3 and 1.1 kg; Figure 1B) and SBP (without SGLT-2i: 5.2 and 3.6 mmHg; with SGLT-2i: 2.4 and 3.2 mmHg). There were no unexpected safety concerns in subjects receiving both a GLP-1RA and SGLT-2i; premature treatment discontinuations due to AEs remained relatively low with semaglutide 1.0 mg and liraglutide 1.2 mg in both SGLT-2i subgroups (without SGLT-2i: 13.4% and 6.4%; with SGLT-2i: 5.5% and 7.2%).

Conclusion: Our results suggest that addition of a GLP-1RA to SGLT-2i treatment was associated with further reductions from baseline in HbA1c, BW and SBP without additional safety concerns. As SGLT-2is and GLP-1RAs are often used together in clinical practice, our data on the safety and efficacy of combining these two classes are meaningful to clinical decision-making.

figurek

Clinical Trial Registration Number: NCT03191396

Supported by: Trial sponsored by Novo Nordisk

Disclosure: M. Capehorn: Employment/Consultancy; RIO Weight Management Ltd and Lighterlife and McDonalds UK. Honorarium; Novo Nordisk, Boehringer Ingelheim, Lilly and Abbot. Lecture/other fees; Novo Nordisk, Lilly. Other; Research support: Novo Nordisk, Boehringer Ingelheim, Lilly.

24

Adherence and persistence in patients with type 2 diabetes initiating once-weekly versus once-daily injectable GLP-1 RAs in US clinical practice (STAY study)

W.H. Polonsky1,2, R. Arora3, M. Faurby4, J. Fernandes4, A. Liebl5;

1Behavioral Diabetes Institute, San Diego, CA, USA, 2University of California, San Diego, USA, 3Novo Nordisk Service Centre Pvt. Ltd, Bangalore, India, 4Novo Nordisk A/S, Søborg, Denmark, 5Center for Diabetes and Metabolism, m&i-Fachklinik Bad Heilbrunn, Bad Heilbrunn, Germany.

Background and aims: Drugs with prolonged action allow for a reduced dosing frequency, which may decrease treatment burden and improve adherence and persistence. We assessed persistence and adherence in patients with type 2 diabetes (T2D) initiating once-weekly or once-daily injectable glucagon-like peptide-1 receptor agonists (GLP-1 RAs) in US clinical practice.

Materials and methods: Adults (≥ 18 years) with T2D in the MarketScan Explorys Claims-EMR Data Set who were GLP-1 RA- and insulin-naïve at first claim for once-weekly or once-daily injectable GLP-1 RA were included. Patients required ≥ 180 days’ enrolment pre-index and ≥ 365 days’ enrolment post-index; the index period was 1 July 2012-31 January 2019, and follow-up was index date + 365 days. Individuals with type 1 diabetes, gestational diabetes or secondary diabetes were excluded from the analysis. Patients were propensity score matched 1:1 by baseline age, sex, baseline Charlson Comorbidity Index score, baseline HbA1c and weight (90 days pre-index), and use of sulfonylureas, metformin, dipeptidyl peptidase-4 inhibitors, and sodium–glucose co-transporter-2 inhibitors (180 days pre-index). Discontinuation was defined as ≥ 60 days not covered by medication. Persistence was defined as stay time (patients who had not discontinued within 12 months were censored at 360 days) and assessed using Kaplan–Meier analysis and Cox proportional hazard models. Adherence was defined as proportion of days covered (PDC) ≥ 0.8.

Results: At baseline, the matched cohorts (n = 784 each) had similar mean age (once-weekly vs once-daily, 54.6 years vs 54.4 years), percentage of women (50% vs 51%), mean body mass index (36.5 kg/m2 vs 36.7 kg/m2), mean HbA1c (8.5% vs 8.4%) and mean Charlson Comorbidity Index score (1.0 each), and received similar antidiabetic medications (biguanides, 46% vs 48%; dipeptidyl peptidase-4 inhibitors, 18% each; sodium–glucose co-transporter-2 inhibitors, 12% vs 5%; sulfonylureas, 23% each). Once-weekly injectable regimens were associated with significantly higher persistence compared with once-daily injectable treatments (median stay time: 333 days vs 269 days; HR 0.80 [95% CI, 0.71-0.90], p < 0.01). Once-weekly GLP-1 RAs were also associated with higher adherence, relative to once-daily regimens, at 6 months (+23%) and 12 months (+35%), irrespective of glycaemic control and weight change. Improvement in glycaemic control was greater with once-weekly than once-daily treatments at 6 months (mean HbA1c change, -1.1% vs -0.9%) and 12 months (-0.9% vs -0.7%). Over 12 months, adherent patients experienced greater mean changes in HbA1c with both once-weekly (-1.1%) and once-daily (-1.0%) regimens than patients with PDC < 0.8 (-0.6% each).

Conclusion: Our results suggest that, in a real-world setting, once-weekly injectable treatments are associated with better persistence and adherence than once-daily regimens, irrespective of glycaemic or weight control.

Supported by: Novo Nordisk A/S

Disclosure: W.H. Polonsky: Employment/Consultancy; Consultant for Eli Lilly, Novo Nordisk and Sanofi.

OP 05 Epidemiology of diabetes complications

25

The natural history of 786 episodes of diabetic ketoacidosis in adults with type 1 and type 2 diabetes

K. Nash1, E. Ooi2, L. Rengarajan3, E. Melson3,4, L. Thomas1, A. Johnson1, D. Zhou1, L. Wallett3, S. Ghosh3, P. Narendran3,4, P. Kempegowda3,5;

1College of Medical and Dental Sciences, Birmingham, UK, 2RCSI & UCD Malaysia Campus, Penang, Malaysia, 3University Hospitals Birmingham NHS Foundation Trust, Birmingham, UK, 4Institute of Immunology and Immunotherapy, Birmingham, UK, 5Institute of Metabolism and Systems Research, Birmingham, UK.

Background and aims: Although diabetic ketoacidosis is traditionally associated with type 1 diabetes, it is increasingly recognised that people with type 2 diabetes can also develop this condition. Currently, there is limited evidence studying the differences between these groups. We aimed to explore the differences in the demographics, presentation and management of diabetic ketoacidosis in adults with type 1 diabetes and type 2 diabetes.

Materials and methods: This cohort study included all episodes of diabetic ketoacidosis from April 2014 to September 2020 in a tertiary centre in the United Kingdom. Diabetes was classified into type 1 diabetes and type 2 diabetes in accordance with previously established diagnoses, autoantibody status, and/or phenotypic features. Data for diabetes type, demographics, biochemical and clinical features at presentation and management of diabetic ketoacidosis were collected. As data were skewed, Wilcoxon sum rank test was applied to compare the two groups on various parameters.

Results: From 786 consecutive diabetic ketoacidosis episodes, 583 (75.9%) type 1 diabetes and 185 (24.1%) type 2 diabetes episodes were included in the final analysis. The median (IQR) age for overall cohort was 38.2 years (23.8-56.8) with male: female ratio of 1:1.04. People with type 2 diabetes were older (type 1 diabetes: 28.97 years [21.9 - 48.7] vs type 2 diabetes 61.54 years [52.0-75.1]; p<0.0001) and had more ethnic minority representations than those with type 1 diabetes (non-white: type 1 diabetes 19.7% vs type 2 diabetes 26.5%; p=0.030). Intercurrent illness (35.4%) was the most common precipitating causes for diabetic ketoacidosis in both cohorts. Severity of diabetic ketoacidosis as assessed by pH (type 1 diabetes 7.22 [7.09-7.29] vs type 2 diabetes 7.24 [7.11-7.30]; p=0.3266), glucose (type 1 diabetes 28.0 mmol/l [20.5-34.8] vs type 2 diabetes 13.2 mmol/l [7.8-17.8]; p=0.4496) and lactate (type 1 diabetes 2.6 mmol/l [1.8-4.3] vs type 2 diabetes 2.6 mmol/l [2.0-4.2]; p=0.6532) at presentation was similar in both groups. However, urea was higher in those with type 2 diabetes (type 1 diabetes 7.1 mmol/l [5.1-10.6] vs type 2 diabetes 8.9 mmol/l [6.4- 16.8]; p=0.0001). Insulin requirements (type 1 diabetes 91.6 units [43.5 - 143.9] vs type 2 diabetes 90.2 units [53.6 - 157.4]; p=0.5551) and total duration of diabetic ketoacidosis (type 1 diabetes 13.9 hours [9.1-21.9] vs type 2 diabetes 13.9 hours [7.7-21.1]; p=0.4638) were similar between the two groups. People with type 2 diabetes had longer hospital stay (type 1 diabetes 3.0 days [1.7-6.1] vs type 2 diabetes 11.0 days [5.0-23.1]; p<0.0001).

Conclusion: A quarter of diabetic ketoacidosis episodes occur in people with type 2 diabetes who were older and with greater ethnic minorities proportion compared to type 1 diabetes. However, the severity of presentation and management is similar in both groups suggesting that the same protocol is equally effective in either type of diabetes. People with type 2 diabetes admitted with diabetic ketoacidosis have a longer hospital admission, perhaps reflecting a more complex need of care.

Disclosure: K. Nash: None.

26

Burden of established cardiovascular disease in people with type 2 diabetes and matched controls: hospital-based care, days absent from work, costs, and mortality

S. Persson1,2, K. Nilsson2, K. Karlsdotter3, J. Skogsberg3, S. Gustavsson3, J. Jendle4, K. Steen Carlsson1,2;

1Lund University, Lund, 2The Swedish Institute for Health Economics, IHE, Lund, 3Boehringer Ingelheim AB, Stockholm, 4Örebro University, Örebro, Sweden.

Background and aims: Established cardiovascular disease (eCVD) is associated with need for healthcare interventions, reduced work capacity and excess mortality. People with type 2 diabetes have increased risk and earlier onset of eCVD compared to people in general. The objective was to assess the burden of hospital-based care, days absent from work, associated costs, and excess mortality for people with type 2 diabetes with and without eCVD with comparison to matched controls.

Materials and methods: The study used a Swedish retrospective database cross-linking longitudinal individual-level data (2007-2016) from national population-based health, social insurance, and socio-economic registers for 454,983 people with type 2 diabetes and their matched controls (5:1 on year of birth, sex, and region of residence). Status eCVD (coronary artery disease, stroke, amputation, periphery vascular disease, non-fatal cardiac arrest, or related interventions) was derived from retrospective data 1997-2006 and updated at change of status 2007-2016. Use of hospital-based care, days absent from work (calendar days) and mortality used data 2007-2016. Regression analysis accounting for individual-level clustering was used for comparison to controls and to attribute costs of hospital-based care and days absent from work to eCVD and to individual complications considering multimorbidity. Excess mortality adjusted for age, sex, and educational level was attributed to eCVD using Cox proportional hazards.

Results: Thirty percent (n=136,135) of people with type 2 diabetes up to age 70 years were observed with eCVD ≥1 observation year in 2007-2016 (women 24% n=43,847; men 34% n=92,288). The mean annual costs of hospital-based care for diabetes complications were EUR 2,758 (95% CI 2,729 to 2,787) of which EUR 2,461 (95% CI 2,432 to 2,490) were attributed to people with eCVD (89%). Main drivers of costs of hospital-based care for people with eCVD were acute myocardial infarction, angina pectoris, and stroke; but also end-stage renal disease (ESRD) and eye disease confirming that eCVD is associated with an increased burden from other complications. eCVD was a leading cause behind work absence for both diabetes and controls. People with type 2 diabetes <66 years had on average 146 days absent (95% CI 145-147) of which 68 days (47%) were attributed to eCVD. Controls had 106 days absent of which 63 days (59%) were attributed to eCVD. The annual cost of eCVD work absence was EUR 9,337 (95% CI 9,150 to 9,523) per individual. The highest work absence was attributed to ESRD, stroke, and heart failure. Type 2 diabetes without eCVD did not differ from controls regarding mortality risk, but type 2 diabetes with eCVD had a four-fold risk of death hazard rate 4.13 (95% CI 4.10 to 4.18) adjusting for age, sex, and educational level.

Conclusion: This study assessed the size of the burden of eCVD-status in people with type 2 diabetes in three measures: 1) eCVD was associated with excess mortality; 2) 9 out of 10 EUR spent on hospital-based care for diabetes complications; and 3) even higher costs of days absent from work in the long-run. Reducing the risks of living with eCVD and postponing the onset of eCVD remain central goals to reduce the burden of type 2 diabetes on the individual and on society.

Supported by: Boehringer Ingelheim AB, Stockholm, Sweden, to the Swedish Institute for Health Economics, Lund, Sweden

Disclosure: S. Persson: Grants; Research grant Boehringer Ingelheim.

27

One-year trajectories of impaired hypoglycaemia awareness in young people with type 1 diabetes

A. Messaaoui1, S. Tenoutasse1, L. Hajselova1, L. Crenier2;

1Diabetology Clinic, Hôpital Universitaire des Enfants Reine Fabiola, Brussels, 2Department of Endocrinology, Hôpital Erasme, Brussels, Belgium.

Background and aims: Impaired hypoglycemia awareness has mainly been studied in the era of capillary blood glucose monitoring and in adults with diabetes. It is less known in youth. The aim of this study is to describe impaired hypoglycemia awareness and its evolution during a one-year follow-up in children with type 1 diabetes using a flash glucose monitoring system.

Materials and methods: This 1-year follow-up study included people with type 1 diabetes using flash glucose monitoring, aged 4–20 years. Impaired hypoglycemia awareness was diagnosed using the Gold method, which consists of a 7-point Likert scale that we adapted for pediatric usage, with a cut-off level of ≥ 3 to diagnose impaired hypoglycemia awareness.

Results: Of the 283 participants included, 48% were diagnosed with impaired hypoglycemia awareness. Participants diagnosed with impaired hypoglycemia awareness were younger at diagnosis (p = 0.045), more often C-peptide negative (p = 0.026) and performed more scans (p = 0.045) than participants with normal awareness. They also reported worse quality of life in usual activity (p = 0.035). They had an increased time below 70 mg/dl (p = 0.003), an increased time below 54 mg/dl (p = 0.003) and an increased Low Blood Glucose Index (p = 0.012) compared to participants with normal hypoglycemia awareness and were 3.2 times more likely to experience severe hypoglycemia (p = 0.002). When defining impaired hypoglycemia awareness using the alternative cut-off of 4 on the Gold scale (as validated for adults), 90 participants (30.9%) remained with an impaired hypoglycemia awareness diagnosis, but this threshold was less discriminating in terms of the occurrence of severe hypoglycemia. After 1-year follow-up, 41% of participants diagnosed with impaired hypoglycemia awareness had reverted to normal awareness, while 22% of participants classified as normal awareness at baseline were newly diagnosed with impaired hypoglycemia awareness at follow-up.

Conclusion: Impaired hypoglycemia awareness is highly prevalent in youth using Flash Glucose Monitoring but affects people differently over time and is associated with an increased and consistent risk of (severe) hypoglycemia . Using the Gold method, in a pediatric population, a cut-off of 3 is more discriminating than the historical threshold of 4 (validated for adults) in terms of hypoglycemia risk. Frequent assessments of impaired hypoglycemia awareness in clinical follow-up may help to improve management of hypoglycemia risk in young people with type 1 diabetes.

Supported by: Grant of The Belgian Kids’ Fund for Paediatric Research

Disclosure: A. Messaaoui: None.

28

Association and familial co-aggregation of type 1 diabetes with depression, anxiety and stress-related disorders: a population-based cohort study

S. Liu1, M. Leone1, H. Larsson1,2, P. Lichtenstein1, B.M. D'Onofrio1,3, S.E. Bergen1, R. Kuja-Halkola1, A. Butwicka1;

1Karolinska Institute, Solna, Sweden, 2Örebro University, Örebro, Sweden, 3Indiana University, Bloomington, USA.

Background and aims: People with type 1 diabetes are known to be at heightened risk of common mental health problems. However, it remains unknown whether genetic liability contributes to the elevated risk. This study aimed to investigate the association and familial co-aggregation of type 1 diabetes with depression, anxiety and stress-related disorders.

Materials and methods: Using multiple Swedish nationwide registers, we obtained a population sample of individuals born 1973-2007 and still residing in Sweden at age 5. Individuals were linked to their biological parents, full-siblings, half-siblings, full-cousins and half-cousins. We obtained information from the National Patient Register (since 1973) on the diagnoses of type 1 diabetes, depression, anxiety and stress-related disorders using ICD codes and from the Prescribed Drug Register (since 2005) on the prescribed antidepressants and anxiolytics using ATC codes. Primary outcomes were any or specific diagnosis of 1) depression, 2) anxiety and 3) stress-related disorders. We examined a secondary outcome of using antidepressants or anxiolytics in those who resided in Sweden after 2005.

Results: In this cohort study of about 3.5 million individuals, 20005 (53.9% male) were diagnosed with childhood-onset type 1 diabetes (< 18 years of age, the median age at onset: 9.7). We used Cox models to estimate the association between type 1 diabetes and each outcome. Individuals were regarded as unexposed before diagnosis and exposed after. During a median follow-up of 22.2 years, individuals with type 1 diabetes were at a higher risk of all outcomes after adjusting for sex and birth year: any diagnosis (HR [95%CI]: 1.73 [1.67-1.80]), depression (1.93 [1.84-2.02]), anxiety (1.41[1.33-1.50]), stress-related disorders (1.75 [1.62-1.89]) and using antidepressants or anxiolytics (1.30 [1.26-1.34]). Familial co-aggregation was evaluated using Cox models, where an individual’s relative was regarded unexposed before the individual’s diabetes diagnosis and exposed after. Overall, higher risks of all outcomes were observed in relatives of individuals with diabetes and declined proportionally with decreasing genetic relatedness. Highest HRs were found in parents: any diagnosis (1.21 [1.16, 1.26]), depression (1.20 [1.13-1.26]), anxiety (1.22 [1.15, 1.30]), stress-related disorders (1.25 [1.17-1.34]) and using antidepressants or anxiolytics (1.18 [1.16, 1.21]). HRs decreased but remained significant in full-siblings after adjusting for sex and birth year of the sibling: any diagnosis (1.11 [1.05, 1.17]), depression (1.11 [1.03-1.19]), anxiety (1.10 [1.02, 1.1]), stress-related disorders (1.20 [1.08-1.32]) and using antidepressants or anxiolytics (1.05 [1.01, 1.09]). HRs decreased and were not significant in maternal and paternal half-siblings (HRs 0.90-1.10; 1.00-1.11), full-cousins (HRs 0.98-1.05) and half-cousins (HRs 0.80-1.02).

Conclusion: Our findings support existing evidence that individuals with childhood-onset type 1 diabetes were at higher risks of depression, anxiety, stress-related disorders and using antidepressants and anxiolytics and suggest that familial liability may contribute to these associations. The results highlight the importance of family support integrated with pediatric diabetes care.

Disclosure: S. Liu: None.

29

High adherence to recommended diabetes follow-up procedures by general practitioners is associated with lower estimated cardiovascular risk

K. Nøkleby1, T.J. Berg1, I. Mdala1, E.S. Buhl1, T. Claudi2, J.G. Cooper3, K.F. Løvaas3, S. Sandberg3, A.K. Jenum1;

1University of Oslo, Oslo, 2Nordland Hospital, Bodø, 3Norwegian Quality Improvement of Laboratory Examinations, Bergen, Norway.

Background and aims: The relation between diabetes quality indicators and diabetes outcomes is unclear. Our aims were to explore whether general practitioners’ (GPs’) performance of recommended processes of care were associated with estimated risk of cardiovascular disease (CVD) and poor glycaemic control in their type 2 diabetes patients.

Materials and methods: A cross-sectional study from Norwegian general practices including 6015 people with type 2 diabetes <75 years old, without CVD and their 275 GPs. The GPs were split into quintiles based on each GP’s average performance of six recommended processes of care (measurements of HbA1c, LDL-cholesterol, albuminuria, blood pressure and foot examination the past 15 months; and recorded eye examination the past 30 months). We estimated the 10-year risk of cardiovascular events using the risk calculator NORRISK 2. To avoid excess confounding of strong non-modifiable risk factors, our main outcome was the modifiable fraction of estimated CVD risk. The exposure variable were the quintiles of GPs in multilevel regression models with estimated CVD risk (total and modifiable fraction) and poor glycaemic control; HbA1c >69 mmol/mol (>8.5%) as outcome variables.

Results: The mean total and modifiable estimated 10-year CVD risk was 12.3% and 3.3% respectively. We found a difference in the mean estimated 10-year risk of CVD of 2.2 percent points between persons registered with GPs in the lowest (13.5%) and the highest (11.3%) quintile for the performance of procedures. Adjusted for confounders, those belonging to GPs in the lowest quintile had a 1.88 (1.17 - 2.60) percent point higher total CVD risk, representing a relative difference of 16.6% higher average risk among patients in quintile 1 compared with quintile 5. The mean modifiable CVD risk was almost 2 percent points higher in patients registered with GPs in the lowest (4.22%) compared with the highest quintile (2.30%). After adjusting for confounders, being registered with a GP in the lowest quintile was independently associated with a 1.78 (1.14 - 2.41, p<0.001) percent point higher modifiable CVD risk compared with those of GPs in the highest quintile. This represents a relative difference of 74.8% (59.0 - 85.8) higher average modifiable risk fraction among those in quintile 1 compared with quintile 5. In the lowest quintile, 14.1% of the people had HbA1c above 69mmol/mol (8.5%), compared with 7.8% in the highest quintile. For patients with GPs in the lowest-performing quintile, the adjusted odds of poor glycaemic control was 1.77 (1.27 - 2.46) times higher than for patients with a GP in the highest quintile.

Conclusion: We found a pattern of lower CVD risk and better glycemic control in patients of GPs performing more recommended diabetes processes of care.

figurel

Supported by: The Norwegian Research Fund for General practice funded the doctoral program of K.N.

Disclosure: K. Nøkleby: Grants; The Norwegian Research Fund for General practice funded the doctoral program of K.N., The Lillian and Werner Næss Fund and the Norwegian Diabetes Association granted this subproject funds.

30

Assessment of the effect of the COVID-19 pandemic on HbA1c testing: implications for diabetes management and diagnosis

D. Holland1, A.H. Heald2, C.J. Duff3, A.A. Fryer4;

1The Benchmarking Partnership, Alsager, 2Salford Royal Hospital, Salford, 3Keele University, Keele, 4School of Medicine, Keele University, Keele, UK.

Background and aims: Since its arrival in 2020, the COVID-19 pandemic has challenged all healthcare systems across the world. The focus on mitigating the effects of the virus led to many routine healthcare services being disrupted. In the UK, services for the diagnosis and management of people with diabetes was one such area. Routine blood testing, a mainstay of diabetes diagnosis and management, became challenging, not least because of the potential risk it posted to facilitating transmission of the virus and associated concerns members of the public had regarding attending for tests.We aimed to quantify this effect of the COVID-19 pandemic on diabetes diagnosis and management, using HbA1c as a surrogate, across regions covered by six UK testing laboratories.

Materials and methods: We estimated the number of missed HbA1c tests, both for diagnostic and management purposes during the Covid Impact Period (CIP) between 23 Mar 2020 and 30 Sep 2020. We then examined the potential impact of these missed tests in terms of (i) effect on diabetes monitoring in people with or at risk of diabetes and, (ii) detection of new diabetes cases. The 6 sites represented mixed rural and urban populations with a range of deprivation scores. We categorised tests into three groups - ‘monitoring’, those already diagnosed with diabetes; ‘screening’ as part of general health and ‘diagnostic’ on those at risk.

Results: The 6 laboratories supported 3.7 million population or 6% of the UK population over the 3 years studied, with 3.6 million HbA1c tests on 1.7 million people. The change between 12 months pre-CIP and the 6 month CIP tests/month were 32k vs 19k monitoring; 46k vs 32k screening and 31k vs 12k diagnostic. 79,000 monitoring tests were missed. 28,500 of these for people with suboptimal control, this delay in monitoring was linked to 2-3 mmol/mol increase in HbA1c. Around 149,000 screening tests in high-risk groups were missed, including nearly 27,000 with HbA1c values within the pre-diabetes range. We also identified that 142,000 diagnostic tests were missed during the period, of which ~12,000 would be expected to be in the pre-diabetes range and 3,800 in the diabetes range. Across the whole UK, these data would equate to 2.54m missed/delayed diagnostic tests during the CIP (on average, 0.40m per month), including ~213,000 in the pre-diabetes range in whom lifestyle advice might be delayed, and 68,500 delayed new diagnoses of diabetes (~24,000 and ~11,000 per month, respectively).

Conclusion: Our findings illustrate the widespread collateral impact of implementing measures to mitigate the impact of COVID-19 in people with, or being investigated for diabetes Ironically, failure to focus of the wider implications for people with diabetes and other groups with long-term conditions, may place them at increased risk of poor outcomes from COVID-19 infection itself, irrespective of the implications for their longer-term health prospects.

figurem

Disclosure: D. Holland: None.

OP 06 Insights into diabetic retinopathy

31

Evaluation of maculopathy deep learning prediction models using the SDRNT1BIO cohort

J. Mellor1, A. Storkey2, P.M. McKeigue1, H.M. Colhoun3, SDRNT1BIO Investigators, Diabetic Retinopathy Clinical Research Network;

1Usher Institute, University of Edinburgh, Edinburgh, 2School of Informatics, University of Edinburgh, Edinburgh, 3Institute of Genetics and Cancer, University of Edinburgh, Edinburgh, UK.

Background and aims: Diabetic Retinopathy and Maculopathy are the leading cause of blindness for people with diabetes. Many countries including Scotland have screening programmes to enable early detection which currently screen people based on fundus image photography. In Scotland, it is found that many people are referred from the screening programme to a specialist with suspected maculopathy but later return to the screening programme - suggesting that maculopathy was not confirmed by the specialist. Accurate prediction of maculopathy from fundus images would reduce unnecessary visits to specialists for people. The objective of this study was to establish whether deep learning prediction models using retinal images could predict those people in the Scottish Diabetic Retinopathy screening programme who were referred to a specialist with suspected maculopathy and did not later return to the screening programme.

Materials and methods: We trained a RegNetY016 network to predict Diabetic Macula Edema using data from the Diabetic Retinopathy Clinical Research Network as well as the Kaggle-EyePACS and Messidor datasets using a combined 25557 fundus images. We evaluated our model and also the publicly-available DeepSeeNet and ARIANNA age-related macular degeneration prediction models. The evaluation data contained 68488 fundus images (886 referred and did not return) from 5539 people in the Scottish Type 1 Diabetes Bioresource who attended the Scottish National Screening programme between the 2005 and 2017. The evaluation outcome was defined as an image being graded as referable and the person having no subsequent screenings. Images within 2 years of the latest image in the dataset and images taken within 2 years of a patient’s death were excluded from evaluation.

Results: The RegNetY016 model discriminated images graded as having suspected maculopathy who did not return to the screening programme with an AUROC of 0.85. The DeepSeeNet Drusen prediction model discriminated the same outcome with an AUROC 0.75 whereas the ARIANNA model only attained an AUROC 0.60.

Conclusion: Deep learning prediction models were able to discriminate between fundus images from patients referred for maculopathy in the Scottish screening programme and did not return to the programme subsequently. This suggests deep learning models may be an effective tool to help reduce unnecessary specialist appointments for maculopathy.

Supported by: JDRF Grant 2-SRA-2019-857-S-B

Disclosure: J. Mellor: Grants; 2-SRA-2019-857-S-B.

32

Whole exome- and whole genome sequencing reveals novel rare genetic variants for severe diabetic retinopathy in type 1 diabetes

N. Vuori1,2, J. Haukka1,2, A. Antikainen1,2, V. Harjutsalo1,3, C. Forsblom1,2, N. Sandholm1,2, P.-H. Groop1,4, FinnDiane Study Group;

1Folkhalsan Institute of Genetics, Folkhälsan Research Center, Helsinki, 2Research Program for Clinical and Molecular Metabolism, Faculty of Medicine, University of Helsinki, Helsinki, 3National Institute for Health and Welfare, The Chronic Disease Prevention Unit, Helsinki, 4Department of Nephrology, University of Helsinki and Helsinki University Hospital, Helsinki, Finland.

Background and aims: Genetic factors play a role in the susceptibility to diabetic retinopathy but studies aiming to identify genetic factors associated with diabetic retinopathy remain scarce. The aim of the study is to discover rare variants associated with severe diabetic retinopathy (SDR) in type 1 diabetes (T1D) with whole exome- (WES) and whole genome sequencing (WGS) studies in the Finnish Diabetic Nephropathy (FinnDiane) Study.

Materials and methods: We performed WGS and WES on ~600 and ~500 individuals with T1D, respectively. WES was done on the Illumina HiSeq2000, and WGS on the Illumina HiSeq X platform with at least 30x coverage, and both were aligned to the GRCh38 reference panel. The SDR phenotype was based on data given by the attending physician, medical records, and the Finnish Hospital Discharge Registry until the end of 2017, and SDR was defined as the first laser treatment (photocoagulation) event, or a diagnosis code for proliferative diabetic retinopathy. Controls had at least a 20-year duration without SDR. After phenotypic ascertainment and quality control, the study population consisted of 490 individuals with WES (390 cases) and 583 individuals with WGS (405 cases). For exonic regions in WES and WGS, association was tested with the score test and meta-analysed with METAL; for other regions, the association was tested with the Firth regression for WGS. The SKAT-O gene aggregate test was performed with WES and WGS for protein altering or truncating variants with two minor allele frequency (MAF) thresholds, 5% and 1%, to improve the power to detect low frequency and rare variant associations. Covariates in the analysis included sex, age at onset, and sequencing data set principal components 1 and 2.

Results: Single marker meta-analysis revealed six variants with a p-value < 10-5 for SDR, with the strongest association for a synonymous variant rs372789789 on LDB3 (MAF=0.2%; p=4.6×10−7) and a synonymous variant rs56043607 in PASK (MAF=0.6%; p=8.2×10−7). Outside exons, variants in 11 loci reached a p-value < 10−5 for SDR, with the strongest associations obtained for an intergenic variant rs9940767 (MAF=22%; p=5.7×10−7). The strongest burden of protein truncating variants in WGS (both in MAF < 5% and MAF < 1%) was found for UACA (SKAT-O p=9.6×10−4), and replicated in WES (p=0.02). UACA is an uveal autoantigen with coiled-coil domains and ankyrin repeats and is associated with panuveitis. In CACNB2 - a gene previously associated with SDR - we found an intron variant associated with SDR (rs4747341, WGS p=8×10-4). No signal for association was found with gene-aggregate tests for CACNB2, but protein altering variants on CACNA1C, which is part of the same calcium channel as the CACNB2, were associated with SDR in WGS (MAF < 1%, SKAT-O p=0.001).

Conclusion: Data from WES and WGS suggest that rare protein truncating variants in UACA contribute to the risk of SDR in individuals with T1D. Look-up of CACNB2 revealed an intron variant association on CACNB2 and protein altering variants on CACNA1C, which is part of the same calcium channel.

Supported by: Silmä- ja kudospankkisäätiö, Mary och Georg C. Ehrnrooths stiftelse

Disclosure: N. Vuori: None.

33

Mapping the daily rhythmic transcriptome in the diabetic retina

H.R. Winter1, R. Silk2, V. Tiwari1,2, D. Simpson1,2, A. Stitt1, E. Beli1;

1Wellcome-Wolfson Institute For Experimental Medicine, Belfast, 2Bioinformatics & Computational Genomics, Belfast, UK.

Background and aims: The eye is a rhythmic organ specifically evolved to function around the light cycle via functional circadian clocks. Diseases such as diabetes have been reported to disrupt circadian rhythms and circadian disruption emerges as an important factor in the prognosis of disease outcomes and treatment success. Herein, we mapped the rhythmic transcriptome in the mouse retina to understand the extent of circadian disruption due to diabetes.

Materials and methods: Healthy control and Ins2Akita/J diabetic mice were kept under physiological 12h:12h light dark cycle until 4 months of age. Deep mRNA sequencing was conducted in retinas collected every 4 hours around the day/night cycle. Computational approaches were used for the detection of rhythmicity (empirical JTK_CYCLE, p<0.05, fc>1.25), acrophase prediction (harmonic regression analysis), differential rhythmic patterns (DORD, p<0.05), phase set enrichment analysis (p<0.05) and upstream regulator predictions (Ingenuity Pathway analysis, Qiagen). Validation was done with in vitro experiments on human retinal endothelial cells.

Results: Almost 10% of the retinal transcriptome was identified as rhythmic with a clear 12hr axis of transcriptional activity, peaking at midday and midnight. Although the 12 hour transcriptional axis is retained in the diabetic retina, it is phase-advanced by approximately 1-3 hours. A total of 478 genes were identified as differentially rhythmic in diabetes compared to control. A higher number of genes identified as phase shifted (405 transcripts) compared to genes with altered amplitude (154 transcripts). Diabetes did not alter the phase of the circadian clock, but phase set enrichment analysis revealed that oxidative phosphorylation was phase advanced by almost 4 hours in the diabetic retina. Diabetic mice had a peak of glucose levels in the blood at that time, possibly driving the metabolic shifts in the retina. Downstream analysis identified oxygen sensing mechanisms and HIF1A as major predicted upstream regulators for the phase shifts.

Conclusion: To our knowledge, this is the first study mapping the effect of diabetes on the rhythmic output in the retina. Importantly, we have started to dissect the source of this rhythmic disruption in the diabetic retina and identified that in an entrained retina, the circadian clock is preserved but metabolic perturbations shift a set a genes within the retina that may contribute to an internal jet-lag within the tissue.

Supported by: JDRF 1-FAC-2019-878-A-N

Disclosure: H.R. Winter: None.

34

Transcriptional and post-transcriptional impact of methylglyoxal in human retinal endothelial cells: New players in diabetic retinopathy?

M. Aprile1, A. Leone2, S. Cataldi1, F. Scognamiglio1, C. Perfetto1, A. Nicolò2, C. Nigro2, V. Costa1, C. Miele2, A. Ciccodicola1;

1National Research Council (CNR), IGB-ABT, Naples, 2National Research Council (CNR), URT GDD IEOS, Naples, Italy.

Background and aims: The accumulation of methylglyoxal (MGO) - a highly reactive byproduct of glycolysis - in diabetic patients has harmful effects on vascular function. MGO mediates the hyperglycemia-induced alterations in macrovascular endothelial cells, but the impact on human retinal endothelial cells (hRECs) has not yet been elucidated. Thus, we assessed the functional and molecular effects of MGO in hRECs, defining regulatory networks of endothelial damage in diabetic retinopathy (DR).

Materials and methods: Functional assays (i.e. MTT, transwell and tube formation) were performed on commercially available primary hRECs exposed to 500μM MGO for 72h. Transcriptome and miRNome analysis were performed by RNA-Seq (~60M paired-end reads/sample) and smallRNA-Seq (10M single-end reads/sample). TopHat, RNA-SeqGUI and iSMART were used for mapping and data normalization. Biological processes/pathways of differentially expressed (DE) genes (edgeR tool; FDR≤0.05) and miRNAs (NOISeq tool; Prob≥0.9) were identified by open-source databases/tools (KEGG, DAVID, Panther, miRPathDB, Reactome). ENCODE ChIP-Seq tool and miRPathDB were used for selecting transcription regulators (TRs) and miRNAs with enriched binding sites in DE genes. Protein-protein and miRNA-genes interactions were analyzed by STRING and miRnet.

Results: MGO significantly impairs cell migration and tube formation capacity of hRECs (p≤0.05), also inducing a wide transcriptional perturbation of genes involved in gene expression regulation (i.e. 454 genes), cell cycle (i.e. 108), proliferation (i.e. 104), cell death (i.e. 95), adhesion (i.e. 79) and vasculature development (i.e. 44). The neuron-restrictive silencer factor (NRSF/REST) is one of TRs with enriched binding sites (Q≤0.05) in ~60% of DE genes (i.e. 1009). According to ENCODE ChiP-Seq data, NRSF/REST is predicted to regulate most of DE genes involved in gene expression regulation (i.e. ~70%), cell cycle (i.e. ~78%) and cell death (i.e. ~74%), revealing to be a reliable mediator of MGO-induced deregulation in hRECs. Of note, a restricted interaction network indicates NRSF/REST as hub connecting multiple deregulated factors with known roles in retinogenesis (e.g. Hipk, Ngfr), retinal neuronal cells (e.g. Jun, FosL2, Hes1) and/or in DR (e.g. p53, Vegfa, Sirt1). Moreover, our analysis revealed 83 miRNAs deregulated by MGO and targeting ~75% of DE genes (i.e. 1354). Notably, the upregulated factors involved in NRSF/REST restricted network are targets of 23 downregulated miRNAs. Thus, we reconstructed a deregulated mRNA-miRNA interaction network potentially mediating the MGO-induced transcriptional and post-transcriptional perturbation in hRECs.

Conclusion: Our work reveals that MGO induces functional and molecular effects in hRECs, representing a putative mediator of primary events in DR. MGO directly induces transcriptional and post-transcriptional alterations in retinal endothelial cells. The interaction network between NRSF/REST, other key altered protein-coding genes and deregulated miRNAs could sustain these effects. Experimental assays are in progress to address whether targeting the candidate mRNAs/miRNAs restrain MGO-induced detrimental effects in hRECs, revealing molecular mediators of glucotoxicity in microvascular retinal cells.

Supported by: EFSD/Boehringer Ingelheim European Research Programme in Microvascular Complications of Diabetes 2018

Disclosure: M. Aprile: None.

35

Two-year progression of retinal neurodegeneration in paediatric patients with type 1 diabetes: the role of glycaemic variability

M. Menduni1, F. Picconi1, M.C. Parravano2, B. Russo1, L. Chioma3, S. Cianfarani3, D. Ylli4, P.I. Patera3, S. Frontoni1;

1Endocrinology, Diabetes and Metabolism, S. Giovanni Calibita Fatebenefratelli Hospital, Department of Systems Medicine, University of Rome Tor Vergata, Rome, Italy, 2Ophthalmology, IRCCS-G.B. Bietti Foundation, Rome, Italy, 3Diabetes Unit, Bambino Gesù Children’s Hospital, IRCCS, Rome, Italy, 4Endocrinology, MedStar Washington Hospital Center, MedStar Health Research Institute, Washington DC, USA.

Background and aims: Retinal neurodegeneration (RN) is considered an early marker of diabetic retinopathy (DR), which precedes vascular damage. Few data are available on the impact and predictive role of glycemic control and daily glycemic variability (GV) on early RN signs in the pediatric population with type 1 diabetes mellitus (T1DM). The aim of our study is to evaluate for two years the structural alteration of neuroretina and the predictive role of GV on RN in pediatric T1DM subjects without any complications.

Materials and methods: 25 patients with T1DM (ages 10-20 years), using Continuous Glucose Monitoring (CGM) and treated with Continuous subcutaneous insulin infusion, without any complication, and 18 healthy control subjects (C), comparable in age and gender, were enrolled and followed for 2 years. All subjects underwent an Optical Coherence Tomography Heidelberg Spectralis, with analysis of all macular neuroretinal layers measuring mean of subfoveal, inner and outer quadrants. In T1DM patients, glycated hemoglobin (HbA1c), GV indexes and standardized CGM metrics were calculated. All the data were collected at baseline (V0) and after 12 (V1) e 24 months (V2).

Results: At V0, Retinal Nerve Fiber Layer (RNFL) thickness was significantly thinner in inner sectors, (91.8 ± 6.8 μm vs 96.5 ± 5.5 μm; p = 0.02) in T1DM versus C. At V1 and V2, the Outer Plexiform Layer (OPL) was significantly thinner in the inner quadrants (152.8 ± 9.4 vs. 163.9 ± 12.8, p < 0.01) (150.3 ± 9.5 vs 163.5 ± 12.8, p < 0.01) and in the whole quadrants (257.1 ± 12.6 μm vs. 286.4 ± 66.5 μm, p = 0.05), (254.3 ± 10.2 vs. 289.6 ± 67,6, p = 0.05) in T1DM versus C. At V2, the Inner Retinal Thickness (IRT) was significantly thinner (1201.3 ± 40.5 vs. 1244.1 ± 61.6, p = 0.04) in T1DM versus C. In the T1DM, a progressive reduction in OPL and IRT was observed after the two-year follow-up (p<0.05). In T1DM patients, a negative correlation was observed between Continuous Overlapping Net Glycemic Action (CONGA-1h) and inner RNFL at V0 (r=-0.42, p=0.05). A negative correlation between Mean Absolute Glucose (MAG) and inner OPL (r=-0.53, p=0.04) was observed at V2. A negative correlation between the IRT delta thickness (V2-V1) and Lability Index (r=-0.64, p=0.01) and MAG (r=-0.61, p=0.02) were found in T1DM patients. No significant correlation between HbA1c and macular layers thickness was observed (p>0.05). Among metabolic parameters, a negative correlation between triglycerides levels and the IRT delta thickness (V2-V1) (r=-0.67, p<0.01) was found.

Conclusion: Very early morphological alterations of neuroretina are already present in pediatric T1DM patients without both vascular retinopathy and neuropathy, supporting the hypothesis that RN occurs early in the course of diabetes. GV seems to play a predictive role in the morphological abnormalities of neurosensory retina in T1DM pediatric population. Next step is a longitudinal evaluation to identify if neuroretinal nerve damage could be a predictive marker of diabetic neuropathy.

Disclosure: M. Menduni: None.

36

Effect of calcium dobesilate in patients with subclinical diabetic macular oedema: the CADODIAME study

O. Simo-Servat1, E. Cordero-Vázquez1, H. Brosa2, A. Simó-Servat3, M. Barraso4, X. Valldeperas5, M. Barahona3, J. García-Arumí2, M. Rivas1, C. Hernández1, L. Lehr6, D. Zingg6, R. Simó1;

1Endocrinology Department, Hospital Vall Hebron, Barcelona, Spain, 2Ophtalmology Department, Hospital Vall Hebron, Barcelona, Spain, 3Endocrinology Department, Hospital Mutua de Terrassa, Terrassa, Spain, 4Ophtalmology Department, Hospital Mutua de Terrassa, Terrassa, Spain, 5Ophtalmology Department, Hospital Germans Trias i Pujol, Badalona, Spain, 6OM-Pharma, Geneve, Switzerland.

Background and aims: Calcium dobesilate (CaD) is indicated and approved for the treatment of diabetic retinopathy (DR) in several countries. Several randomised placebo-controlled clinical trials demonstrated CaD efficacy on the progression of early DR. However, in the CALDIRET study, CaD did not reduce the risk of developing clinical significant macular edema (CSME). Overall the data of those studies suggest that CaD is beneficial in the very early stages of DR but more supportive clinical evidence is needed. Retinal thickness seems a good predictor of diabetic macular edema (DME) and, therefore, patients with subclinical DME represent a good target population to evaluate therapeutic strategies for preventing CSME. On this basis, we aimed to test the effect of CaD in early stages of DR, including the progression from subclinical DME to CSME.

Materials and methods: In this randomized, multicenter placebo-controlled study of parallel groups, 60 eligible patients with type 2 diabetes with subclinical DME were randomized to receive CaD (Calcium dobesilate 1000mg twice per day) or placebo for 1 year in a 1:1 ratio. The primary endpoint was to determine whether CaD is able to prevent the increase of retinal thickness (RT). The secondary endpoints were to assess whether CaD is able to: 1) arrest the progression from subclinical to clinically significant DME; 2) arrest the progression of the ETDRS level (≥ 1 step); 3) arrest the progression of neurodegenerative changes, 4) arrest the reduction of visual acuity, and 5) arrest the progression of foveal avascular zone (FAV).

Results: A total of 50 patients completed the 1 year follow-up. RT decreased by a mean of 2.5 μm (SD14) in patients treated with CaD, whereas in patients treated with placebo RT increased by 3.8 μm (SD24), but this difference did not reach statistical significance (p= 0.26). Regarding secondary endpoints the only difference was the effect of CaD on the progression of ETDRS. We found an improvement of ETDRS in 9 of patients under CaD in comparison with 3 in the placebo group (37.5% vs. 12%; p=0.05). In the logistic regression analyses adjusting by age, hypertension and the mean of HbA1C (four measurements), treatment with CaD was independently associated with an improvement of ETDRS level at the end of 1 year follow-up (P=0.03). We did not find any difference in adverse effects when comparing placebo and CaD groups.

Conclusion: In our study, the treatment with CaD did not have any significant impact on the progression of retinal thickness, at least after one year of follow up. However, it was a safe and useful treatment for early microvascular impairment in the setting of DR as shown by the improvement of ETDRS level. Further studies with a larger number of patients and longer follow-up seem warranted.

Clinical Trial Registration Number: 2017-000250-19

Supported by: OM-Pharma

Disclosure: O. Simo-Servat: Employment/Consultancy; Consultancy fees from OM-Pharma.

OP 07 When men are mice: the study of human physiology in humans

37

Exogenous LEAP2 improves postprandial glucose tolerance and reduces ad libitum food intake in men

C.A. Hagemann1,2, M.S. Jensen1, S. Holm3, L.S. Gasbjerg1,3, M.B. Christensen1,4, S.M.N. Heimbürger1,5, F. Dela3, T. Vilsbøll1,5, B. Holst3, F.K. Knop1,6;

1Gentofte Hospital, University of Copenhagen, Hellerup, 2Gubra Aps, Hørsholm, 3Department of Biomedical Sciences, Faculty of Health and Medical Sciences, University of Copenhagen, Copenhagen, 4Bispebjerg Hospital, University of Copenhagen, Copenhagen, 5Steno Diabetes Center Copenhagen, Gentofte, 6Department of Clinical Medicine, Faculty of Health and Medical Sciences, University of Copenhagen, Copenhagen, Denmark.

Background and aims: The gastric hormone ghrelin stimulates food intake and increases blood glucose through activation of the growth hormone secretagogue receptor (GHSR). Liver-expressed antimicrobial peptide 2 (LEAP2) has been proposed to inhibit actions of ghrelin through inverse effects on GHSR activity. In the present study, we investigated the effects of LEAP2 on postprandial glucose metabolism and ad libitum food intake in healthy volunteers.

Materials and methods: On two separate experimental days performed in randomised order, twenty healthy men (age 18-25 years, BMI 20-30 kg/m2) underwent a liquid mixed meal test (3.0 ml/kg body weight; 1,010 kJ, 29.7 g carbohydrate, 9.6 g protein and 9.3 g fat per 100 ml) with double-blind i.v. infusions of LEAP2 (30 pmol/kg/min) and saline (placebo), respectively. After 260 minutes of infusion, an ad libitum meal of pasta Bolognese (565 kJ, 15.0 g carbohydrate, 5.3 g protein and 5.6 g fat per 100 ml) was served. Each experimental day was preceded by 12 hours of fasting including any liquids and 48 hours of abstinence from alcohol, strenuous physical activity, medicine and intermittent fasting and/or excessive eating.

Results: Compared to placebo, LEAP2 reduced postprandial peak plasma glucose (mean differences 0.36 (95% CI 0.13-0.60) mmol/l, p = 0.0048) and baseline-subtracted AUC for plasma glucose (mean differences 40.8 (95% CI 0.2-81.3) mmol/l × min, p = 0.049). LEAP2 significantly decreased ad libitum food intake assessed by total intake and intake per kg body weight, respectively, compared to placebo (mean differences 432 (95% CI 155-709) kJ, p = 0.0041; 5.4 (95% CI 2.1-8.7) kJ/kg, p = 0.0029). There was no statistically significant difference in visual analogue scale ratings of hunger, satiety, prospective food consumption, fullness, nausea, comfort or thirst.

Conclusion: Exogenous LEAP2 improves postprandial glucose tolerance during a standardised liquid mixed meal test and reduces food intake during an ad libitum meal test in healthy young men.

figuren

Clinical Trial Registration Number: NCT04621409

Supported by: Gubra Aps

Disclosure: C.A. Hagemann: None.

38

Effect of a hypercaloric diet on brain insulin sensitivity and liver fat in normal-weight men

S. Kullmann, L. Wagner, R. Veit, J. Machann, A. Fritsche, H. Preissl, A.L. Birkenfeld, M. Heni;

Institute for Diabetes Research and Metabolic Diseases of the Helmholtz Center Munich at the University of Tübingen, Tübingen, Germany.

Background and aims: Obesity and prediabetes are associated with brain insulin resistance with unfavorable effects on long-term weight and body fat composition. Whether short-term overfeeding can trigger brain insulin resistance in healthy humans is currently not known. Hence, we investigated the effect of a five-day hypercaloric diet on brain insulin responsiveness and body fat composition in healthy normal weight adults.

Materials and methods: Twenty-nine male volunteers (age range 19 - 27 years, BMI range 19 - 25 kg/m2) were enrolled to participate either in a five-day hypercaloric diet (n= 18; increasing their daily caloric intake by 1500 kcal with high caloric snacks) or a control diet (n=11; no additional calories). Brain insulin sensitivity was assessed by functional magnetic resonance imaging (MRI) combined with intranasal administration of insulin to the brain before, directly after the hypercaloric diet and 2 weeks afterwards. Food intake and physical activity was recorded during the course of the study. Moreover, participants underwent two oral glucose tolerance tests, whole-body MRI for body fat distribution/ intrahepatic fat content and different behavioral and cognitive assessments.

Results: Paired t-test in the five-day hypercaloric diet group revealed a significant change in brain insulin action in the insular cortex from before to directly after the intervention (p<0.05, corrected for multiple comparisons). Two weeks after the diet, brain insulin action was restored. No change in brain insulin action was observed in the control group (n=11). Moreover, intrahepatic fat significantly increased in the intervention group directly after the five-day hypercaloric intervention (mean change of 53%; p=0.001), while no change was observed in the control group (mean change of 3%; p=0.673). High-caloric diet induced alterations in brain insulin responsiveness significantly correlated with increased liver fat (r= 0.606, p=0.001). No significant effects were observed on glucose metabolism, peripheral insulin sensitivity and body weight. To compare our results to a normal weight and obese control group without an intervention, we extracted the insular cortex response to intranasal insulin in twenty-nine age-matched participants from a previous study cohort. Interestingly, brain insulin action in the insula of the obese control group showed a similar response to intranasal insulin, as we observed in the current study after the high caloric diet. The normal weight control group, on the other hand, showed a response just as in our current study before the intervention.

Conclusion: In normal weight men, short-term overfeeding with high caloric snacks can trigger brain insulin resistance and liver fat accumulation. Brain insulin action after overfeeding was similar as observed in age-matched persons with obesity.

Clinical Trial Registration Number: NCT03590561

Supported by: DZD Grant 2017-2018

Disclosure: S. Kullmann: None.

39

Effect of lowering branched-chain amino acid levels in patients with type 2 diabetes using sodium-phenylbutyrate

F. Vanweert1, E. Erazo Tapia1, T. van de Weijer1,2, J. Hoeks1, V.B. Schrauwen-Hinderling1,2, P. Schrauwen1, M.K.C. Hesselink1, M. Blair3, M. Neinast3, Z. Arany3, E. Phielix1;

1Nutrition and Movement Sciences, Maastricht University, Maastricht, Netherlands, 2Department of Radiology and Nuclear Medicine, Maastricht University Medical Center, Maastricht, Netherlands, 3Cardiovascular Institute, University of Pennsylvania, Philadelphia, USA.

Background and aims: Patients with type 2 diabetes (T2D) are characterized by elevated branched-chain amino acid (BCAA) levels in plasma, which associate with insulin resistance. Recently, we showed that patients with T2D had reduced whole-body BCAA oxidation rates compared to controls, which may at least partly be explained by low mitochondrial BCAA oxidation. In the present study we aimed to pharmacologically accelerate BCAA oxidation in patients with T2D with use of sodium-phenylbutyrate (NaPB) and evaluated the effect on patients’ metabolic health.

Materials and methods: Sixteen men and women with T2D (BMI: 29.6 ± 3.3 kg/m2, age: 66 ± 6 y, 13m/3f) underwent a 2-week NaPB (4.8 g/m²/day) treatment in a randomized, placebo-controlled, double-blind cross-over design with a wash-out period of 6-8 weeks. The primary outcome was peripheral insulin sensitivity measured with a 2-step hyperinsulinemic-euglycemic clamp, expressed as the change in insulin-stimulated glucose disposal rate minus baseline (ΔRd). Secondary outcomes were metabolic flexibility, defined as the change in insulin-stimulated respiratory exchange ratio minus baseline (ΔRER), and ex vivo mitochondrial oxidative capacity in skeletal muscle measured with high-resolution respirometry expressed as O2-flux. Fasting blood samples were collected to determine levels of BCAA and glucose. In addition, phenylbutyrate levels were determined by LCMS to check compliance to the intervention.

Results: Plasma phenylbutyrate levels increased upon NaPB treatment compared to placebo (5.5-fold, p=2.08*10^-10) indicating compliance to the intervention. Fasting plasma BCAA levels (in μmol/l) decreased (valine 275.3 ± 10.3 vs. 304.7 ± 10.4, p=0.01; leucine 168.7 ± 5.1 vs. 181.5 ± 6.5, p=0.04; isoleucine 105.2 ± 3.5 vs. 112.1 ± 4.1, p=0.08) and fasting glucose levels tended to be lower (7.8 ± 0.4 vs. 8.2 ± 0.5 mmol/l, p=0.06) after 2 weeks of treatment with NaPB compared to placebo. Furthermore, peripheral insulin-sensitivity was 27% higher (ΔRd: 13.2 ± 1.8 vs. 9.6 ± 1.8 to μmol/kg/min, p=0.02) which appeared to be primarily accounted for by a higher peripheral insulin-mediated glucose oxidation (11.9 ± 0.7 vs. 10.8 ± μmol/kg/min, p=0.03) upon NaPB compared to placebo. In addition, metabolic flexibility tended to be higher after NaPB compared to placebo (ΔRER: 0.09 ± 0.01 vs. 0.08 ± 0.01, p=0.10). Also, maximally coupled mitochondrial respiration, fueled by the carbohydrate-derived substrate pyruvate was 10% higher after treatment with NaPB compared to placebo (O2-flux: 74.0 ± 4.1 vs. 67.1 ± 4.3 pmol/(s*mg), p=0.05).

Conclusion: NaPB treatment for two weeks effectively reduced plasma BCAA levels in patients with T2D. This was paralleled by a higher peripheral insulin sensitivity and glucose oxidation. Our findings encourage future research to investigate the underlying mechanisms and whether stimulating BCAA oxidation could be a potential strategy to promote metabolic health in T2D.

Clinical Trial Registration Number: NTR7426

Supported by: Diabetes Fonds Nederland

Disclosure: F. Vanweert: None.

40

Comparative effects of acute hypertriglyceridaemia with or without elevation of non-esterified fatty acids on glucose tolerance, insulin kinetics and beta cell function

D. Trico1, B. Astiarraga2, M. Seghieri1, A. Mengozzi1, L. Nesti1, S. Baldi1, A. Mari3, A. Natali1;

1University of Pisa, Pisa, Italy, 2Pere I Virgili Research Institute, Tarragona, Spain, 3National Research Council, Padua, Italy.

Background and aims: Chronic hypertriglyceridaemia is associated with glucose intolerance and type 2 diabetes risk in large population-based studies. Whether this association is directly mediated by triglycerides or their lipid metabolites (non-esterified fatty acids, or NEFAs), and which mechanisms are involved is unclear. Therefore, the aim of this study was to compare the glucometabolic effects of mild acute hypertriglyceridaemia alone or combined with elevation of circulating NEFAs in non-diabetic subjects.

Materials and methods: Twenty-two healthy lean volunteers underwent two 5-h intravenous infusions of either normal saline or Intralipid, without (n=12) or with heparin (I+H; n=10) to promote the release of NEFAs. Oral glucose tolerance tests (OGTTs) were performed during the last 3 h of infusion. Insulin sensitivity, insulin secretion, β-cell function parameters, and insulin clearance were analyzed by mathematical modeling of glucose, insulin, and C-peptide data at baseline (i.e. after 2 h of lipid infusion) and during the OGTTs.

Results: As for study design, plasma triglycerides reached similar levels during both lipid infusions (~2.5 mmol/L, p=0.40), while NEFAs were markedly higher during I+H (AUC 249 [130] vs 77 [20] mmol/L, p=0.0001). In fasting conditions, Intralipid alone reduced insulin clearance, while I+H increased insulin secretion and reduced insulin sensitivity and clearance, without significant changes in glucose levels. During the OGTT, the lipid infusions impaired glucose tolerance to a similar extent but via different mechanisms. Intralipid alone reduced insulin sensitivity and clearance, inhibited glucose-stimulated insulin secretion (i.e. β-cell glucose sensitivity), and stimulated β-cell potentiation. I+H induced a greater impairment in insulin sensitivity, enhanced the dynamic component of insulin secretion (i.e. β-cell rate sensitivity), and neutralized both the negative and positive effects of triglycerides on the β-cell.

Conclusion: This study dissected the pathogenetic mechanisms of glucose intolerance induced by mild acute hypertriglyceridaemia with or without a parallel increase in plasma NEFAs in non-diabetic individuals, providing new insight into plausible biological signals that generate and sustain insulin resistance and chronic hyperinsulinemia during diabetes progression.

figureo

Supported by: EFSD Future Leaders Mentorship Programme for Clinical Diabetologists 2018 supported by an unrestricted educational grant from AstraZeneca

Disclosure: D. Trico: None.

41

Pathophysiological changes during weight-loss induced remission of type 2 diabetes in non-obese people

A. Al-Mrabeh1, A. Barnes2, K. Irvine1, T. Kelly2, K.G. Hollingsworth1, D. Romeres3, C. Cobelli4, R. Taylor1;

1Magnetic Resonance Centre, Newcastle University, Newcastle upon Tyne, UK, 2Human Nutrition Research Centre, Newcastle University, Newcastle upon Tyne, UK, 3Department of Information Engineering, University of Padova, Padua, Italy, 4Department of Woman and Child’s Health, University of Padova, Padua, Italy.

Background and aims: Classification of people according to BMI leads to the frequent assumption that weight loss is not indicated in non-obese individuals with type 2 diabetes. We have examined the pathophysiological changes during weight loss induced remission of type 2 diabetes in people with BMI <27kg/m2.

Materials and methods: 25 people were studied to date, 2/25 were found to have MODY/ type 1 diabetes and were excluded [type 2 diabetes: n=12, age 58.3±7.2 years, BMI 24.5±2.0kg/m2, diabetes duration 2.5±2.2years; non-diabetic controls: n=11, age 58.8±9.5 years, BMI 21.6±2.5kg/m2]. Controls were age, sex and BMI-matched (after weight loss) to the type 2 diabetes group, and were studied on a single occasion. Dietary weight loss was achieved using a low-calorie diet (~800 kcal/day) followed by 4-6 weeks of weight stability. The cycle of weight loss plus maintenance was repeated 2-3 times (10-15% weight loss) until the clinical end point of remission was reached (HbA1c < 48mmol/mol). Intraorgan fat was quantified by 3-point Dixon magnetic resonance. Indices of insulin sensitivity (SI) and beta cell responsivity (Φtot) after a standard meal were estimated by mathematical modelling.

Results: After weight loss (Table 1), fasting plasma glucose and HbA1c decreased (7.3±0.3 to 6.3±0.3 mmol/l, and 53.8±2.1 to 48.0±1.5 mmol/mol, respectively, p<0.05 for both). Fasting plasma insulin normalised (52.2±9.5 to 23.7±3.6 pmol/L, p=0.007). Total body fat decreased (33.1±1.9 to 27.4±2.0%, p<0.001) as did plasma triglyceride (1.6±0.2 to 1.0±0.1mmol/L, p=0.002) with no changes in NEFA (0.67±0.04 to 0.68±0.07mmol/L, p=0.98). Liver fat at baseline was over twice that of controls (4.4±0.8 vs. 1.9±0.3%, p=0.02), and correlated negatively with HDL (r=-0.78, p=0.003). It decreased after intervention (1.4±0.1%, p=0.004 vs. baseline) becoming similar to the control levels. Pancreas fat fell (5.1± 0.6 to 4.5±0.6%, p=0.026) towards control levels (3.4±0.3%, p=0.13). Post meal SI, Φtot, and disposition index (DI) improved after weight loss. Weight loss induced remission (HbA1c <48mmol/mol off medication) was achieved in 67% (8/12) intervention participants.

Conclusion: 2/14 people diagnosed with type 2 diabetes in this BMI range were found to have other diagnoses. Remission and return of beta cell function with fall in liver and intrapancreatic fat content were observed in a similar proportion to previous studies in overweight or obese individuals with type 2 diabetes. Non-obese people with type 2 diabetes exhibit similar pathophysiological changes during remission of diabetes.

figurep

Supported by: Diabetes UK

Disclosure: A. Al-Mrabeh: None.

42

The Single-Point Insulin Sensitivity Estimator (SPISE) in the prediction of abnormal glucose metabolism in obese children: a long term follow-up study

S. Dule1, I. Barchetta1, L. Bertoccini1, F. Cimini1, F. Sentinelli1, G. Marini2, S. Loche2, E. Cossu2, M. Baroni3, M. Cavallo1;

1Department of Experimental Medicine, Sapienza University of Rome, Rome, 2Department of Medical Sciences and Public Health, University of Cagliari, Cagliari, 3Department of Clinical Medicine, Public Health, Life and Environmental Sciences (MeSVA), University of L’Aquila, L'Aquila, Italy.

Background and aims: Overweightand obesity in childhood are conditions epidemically spread worldwide; early onset obesity is an independent risk factor for the development of insulin resistance and type 2 diabetes, more than body mass index alone. The identification of accurate tools for risk stratification in obese/overweight children is a crucial step for designing strategies to prevent metabolic disease later in life. The value of insulin measurement and insulin-derived indicators for metabolic risk stratification in this young population is still debated, as their interpretation is not univocal. The Single-Point Insulin Sensitivity Estimator (SPISE) is a lipids and BMI-based index of insulin sensitivity recently validated in adolescents and adults. Aims of this study were to investigate the relationship between the SPISE index, glyco-metabolic profile and insulin sensitivity in a large population of children with and without obesity and to evaluate whether basal SPISE is predictive of the development of impaired glucose metabolism later in life.

Materials and methods: We analysed data from 909 overweight/obese children undergoing metabolic evaluations at University of Cagliari, Italy; SPISE (=600*HDL0.185/Triglycerides0.2*BMI1.338), static and dynamic insulin-derived indicators of insulin sensitivity/resistance(i.e. ISI, Disposition Index (DI), HOMA-IR, HOMA-β%) were also assessed in 99 normal-weight, age-, sex-comparable children, selected as a reference group. Two-hundred out of 909 overweight/obese children were followed up between 2013 to 2016 (median [IQR] follow-up duration= 6.5[3.5-10] years), and data were used for longitudinal retrospective investigations.

Results: At baseline, 96 out of 909 (11%) overweight/obese children had impaired glucose regulation; in this subgroup, mean SPISE was significantly lower than in youths with normal glucose metabolism (mean±SD SPISE: 6.3±1.7 vs. 7±1.6, p<0.001). At the bivariate analysis, SPISE correlated positively with ISI and DI,and negatively with age, blood pressure, HOMA-IR, basal and 120 mins blood glucose and insulin (all p values <0.001). The presence of a significant association between SPISE and insulin-derived indexes of insulin-sensitivity, -resistance and -secretion,was also reported in normal-weight children. In this cohort, SPISE negatively correlated with age (r= -0.43, p<0.001) and FSI (r= -0.47, p<0.001) whereas no association with FBG was observed. At the 6.5-year follow-up, lower basal SPISE predicted the development of abnormal glucose metabolism with AUROC curve: 0.83(0.72-0.94), p<0.001, regardless of age, sex, fasting/120 mins glucose and insulin at the baseline.

Conclusion: Low SPISE in children is significantly associated with metabolic abnormalities and predicts the development of impaired glucose regulation later in life. In conclusion, our study shows that SPISE represents an easy, reliable and unexpensive surrogate of insulin sensitivity in overweight/obese children to be used as a screening tool for metabolic risk assessment on a large scale.

Disclosure: S. Dule: None.

OP 08 Bugs on fire

43

Leukocyte counts and T cell frequencies differ between novel subgroups of diabetes

J.M. Ratter1,2, H. Maalmi1,2, S. Trenkamp1,2, O.-P. Zaharia1,3, W. Rathmann4,2, N.C. Schloot1, K. Straßburger4,2, J. Szendroedi1,2, C. Herder1,2, M. Roden1,3, GDS Group;

1Institute for Clinical Diabetology, German Diabetes Center, Leibniz Center for Diabetes Research at Heinrich-Heine-University Düsseldorf, Düsseldorf, 2German Center for Diabetes Research (DZD), Partner Düsseldorf, München-Neuherberg, 3Department of Endocrinology and Diabetology, Medical Faculty and University Hospital Düsseldorf, Heinrich-Heine-University Düsseldorf, Düsseldorf, 4Institute for Biometrics and Epidemiology, German Diabetes Center, Leibniz Center for Diabetes Research at Heinrich-Heine-University Düsseldorf, Düsseldorf, Germany.

Background and aims: Distributions of circulating immune cells are altered in type 1 and type 2 diabetes compared to healthy individuals and associate with insulin sensitivity, glycemic control and circulating lipid concentrations. This study aimed to determine whether total leukocyte counts and immune cell frequencies differ among the novel diabetes subgroups.

Materials and methods: We analyzed automated white blood cell counts (n=669) and flow cytometry data (n=201) of participants with recent-onset diabetes (known diabetes duration <1 year) from the German Diabetes Study, who were allocated to the five novel diabetes subgroups (severe autoimmune diabetes (SAID), severe insulin-deficient diabetes (SIDD), severe insulin-resistant diabetes (SIRD), mild obesity-related diabetes (MOD), mild age-related diabetes (MARD)) according to clustering analysis based on clinical variables (age, body mass index, HbA1c, homeostasis model assessment-2 estimates of β-cell function and insulin resistance (HOMA2-B, HOMA2-IR), and glutamic acid decarboxylase antibodies). Pairwise differences between subgroups were analyzed with linear regression models after log-transformation of variables, followed by Tukey-Kramer correction for multiple testing.

Results: Leukocyte numbers were higher in SIRD (median [IQR]; 7100 [6425; 8300] /μL), MOD (7200 [6225; 8400] /μL) and MARD (6500 [5500; 7675] /μL), but not SIDD compared to SAID (5700 [4800; 6900] /μL) and higher in SIRD and MOD vs. MARD (all P< 0.05). The percentage contribution of CD4+ T cells to the total T cell count was higher in SIRD (68.4 [64.0; 76.0] %) than in SAID (58.0 [52.7; 66.2] %), MOD (60.0 [55.5; 68.2] %) and MARD (63.8 [54.8,72.1]). The percentage of CCR4+ regulatory T cells of total regulatory T cells was higher in SIRD (74.8 [72.2; 78.6] %) vs. SAID (63.9 [55.0; 74.5] %) and MOD (67.2 [62.8; 77.3] %) and in MARD (77.1 [68.7, 83.0] %) vs. SAID (all P<0.05). After adjusting for the variables used to define the subgroups, none of these differences remained significant (P>0.05). Frequencies of CD4+ T cells were positively associated with age, BMI, HOMA2-B und HOMA2-IR, and frequencies of CCR4+ regulatory T cells with age, HOMA2-B and HOMA2-IR.

Conclusion: These results show differences in leukocyte profiles between novel subgroups of diabetes. Higher percentages of CD4+ T cells of total T cells in SIRD vs. MOD and MARD and higher percentages of CCR4+ regulatory T cells of total regulatory T cells in SIRD vs. MOD suggest the presence of distinct inflammatory processes in SIRD, MOD and MARD subgroups.

Clinical Trial Registration Number: NCT01055093

Disclosure: J.M. Ratter: None.

44

Visceral adipose tissue-derived PD1-1+ Tconv from obese patients with type 2 diabetes are pro-inflammatory cells with the potential to recirculate to the blood stream

A. Giovenzana1, E. Bezzecchi1,2, C. Socci3, M. Bissolati3, C. Corsini3, A. Terulla3, G.M. Scotti2, S. Cardellini1, A. Saibene4, M. Morelli2, E. Ruggiero5, A. Petrelli1;

1San Raffaele Diabetes Research Institute, IRCCS Ospedale San Raffaele, Milano, 2Center for Omics Sciences, IRCCS San Raffaele Scientific Institute, Milano, 3Transplant and Metabolic/Bariatric Surgery Unit, IRCCS Ospedale San Raffaele, Milano, 4Department of General Medicine, Diabetes and Endocrinology, IRCCS Ospedale San Raffaele, Milano, 5Experimental Hematology Unit, Division of Immunology, Transplantation and Infectious Diseases, IRCCS San Raffaele Scientific Institute, Milano, Italy.

Background and aims: Visceral adipose tissue (VAT) in obesity is featured by chronic low-grade inflammation leading to resistance to insulin action (i.e. insulin resistance). It is yet to be explored, however, why only a fraction of obese patients fails to control glucose metabolism resulting in the development of type 2 diabetes (T2D). Growing data from preclinical models of obesity indicate that, in the cascade of events leading to T2D, T cells are engaged early and precede macrophage recruitment in the VAT. Furthermore, we have shown that CD4 T cells from the VAT of obese patients with T2D display a pro-inflammatory phenotype and are resistant to TGF-β-mediated suppression. Here we aimed to assess profile and spatial dynamics of T cells localized in the VAT of obese patients with T2D.

Materials and methods: Characterization of T cell subsets in the stromal vascular fraction (SVF) obtained from the VAT of obese patients with prediabetes or T2D (OB-PreD&D) (n=21) and without T2D (OB-ND) (n=27) undergoing bariatric surgery was performed using unsupervised and supervised analysis of flow cytometry data. Whole transcriptome sequencing and T cell receptor (TCR) sequencing of VAT-derived PD-1-expressing CD4 conventional T cells (Tconv) was performed in comparison with the PD-1- counterpart (OB-PreD&D n=4, OB-ND n=4). For TCR sequencing a control group of lean donors (n=4) and peripheral blood (PB) samples were also included.

Results: Unsupervised and supervised analysis of flow cytometry data showed that a subset of resident CD4 Tconv cells expressing PD-1 is reduced in the VAT of OB-PreD&D compared to OB-ND (p-value=0.03). Increased frequency of TNF-α-producing PD-1+, but not PD-1-, CD4 Tconv were evident in OB-PreD&D (p-value=0.04). In line with this, transcriptomic analysis showed selective upregulation of pathways involved in inflammatory processes in PD-1+ CD4 Tconv from OB-PreD&D (adj p-value: 5 x 10-09). Analysis of TCR sequencing indicates that clonotypes of PD1+ CD4 T cells from OB-PreD&D tend to be enriched in the peripheral blood compared to the VAT and that their clonality with the level of insulin resistance (p-value=0.04, R2=0,5243).

Conclusion: These data indicate that impairment of glucose metabolism fosters the generation of pro-inflammatory PD1+ CD4 Tconv in the VAT with high potential to recirculate between blood and tissue(s). The PD1+ CD4 Tconv cell subset has the potential to be the mediator of “infectious” inflammation driving insulin resistance in insulin-sensitive tissues.

Supported by: Fit-The-Fat Marie Sklodowska-Curie grant agreement no 704779

Disclosure: A. Giovenzana: None.

45

Plasma lipopolysaccharide levels and its relationship with glycaemic status and gut microbiota changes associated with H. pylori infection

M. Clemente-Postigo1,2, G.M. Martin-Nuñez3,2, M. Roca-Rodriguez4, L. Coin-Araguez3, I. Moreno-Indias3,2, I.M. Cornejo-Pareja3,2, F.J. Tinahones3,2;

1Laboratorio de Investigación (IBIMA). 1° Planta, IBIMA / Universidad de Córdoba, Malaga /Cordoba, 2CIBER Fisiopatología de la Obesidad y Nutrición, Malaga, 3Laboratorio de Investigación (IBIMA). 1° Planta, IBIMA, Malaga, 4Departamento de Endocrinología y Nutrición, Hospital Universitario Puerta del Mar, Cadiz, Spain.

Background and aims: Apart from causing gastric disturbances, Helicobacter pylori (H. pylori) infection has been related to a number of extragastric diseases including diabetes and obesity. Notably, gut microbiota composition is affected by both H. pylori infection and H. pylori antibiotic-based eradication therapy (ET). Considering that gut microbiota dysbiosis has been widely related to metabolic dysregulation, modulation of the gut bacterial community emerges as a linking factor between H. pylori infection, antibiotic treatment and metabolic diseases. Indeed, our group has previously shown that changes in glucose homeostasis after H. pylori ET are related to gut microbiota. However, mechanisms underlying this relationship remains unexplored. Plasma gut-derived lipopolysaccharides (LPS) promote low-grade inflammation and metabolic diseases. The aim of this study was to analyse plasma LPS as the possible mediating factor between gut microbiota and glycemic changes related to H. pylori and its ET.

Materials and methods: 40 H. pylori-infected adult patients and 20 controls without infection were recruited. Infected subjects were evaluated before and two months after ET (omeprazole, clarithromycin, amoxicillin). Faecal microbiota composition, determined by 16S rRNA gene (V3-V4) sequencing using Illumina Miseq, glycemic status [glucose, glycosylated haemoglobin (HbA1c), HOMA-IR, GLP-1] and plasma LPS levels were analysed.

Results: Plasma LPS levels correlated positively and significantly with HbA1c patients (r=0.457; p=0.004) and GLP-1 (r=0.502; p=0.002) in H. pylori-infected patients. However, these correlations were not found after ET treatment or in controls. In addition, plasma LPS positively correlated with the genera Roseburia (r=0.339; p=0.037), Megamonas (r=-0.323; p=0.048), Megasphaera (r=0.446; p=0.05) and Sutterella (r=0.323; p=0.048) in H. pylori-infected patients. However, after ET correlations between LPS and these bacterial groups were abolished, except for Megamonas (r=-0.409; p=0.013).

Conclusion: Plasma LPS levels are related to glycemic status during H. pylori infection, but not after its eradication. Specific gut bacterial groups could be influencing plasma LPS levels in H. pylori-infected patients.

Supported by: Junta de Andalucia-CECEU (DOC_00448); MICINN (FJCI-2017-32194; FJCI-2017-34349); ISCIII (CP16/00163; JR19/00054; PI14/00082, PI15/01114, PI18/01160; CB06/03/0018). Co-founded by FEDER.

Disclosure: M. Clemente-Postigo: None.

46

TNFA mediates inflammation-induced effects on PPARG splicing in adipose tissue and mesenchymal precursor cells

S. Cataldi1, M. Aprile1, C. Perfetto1, D. Melillo1, S. Giorgetti-Peraldi2, M. Cormont2, P. Italiani1, M. Blüher3, J.-F. Tanti2, A. Ciccodicola1, V. Costa1;

1National Research Council, Naples, Italy, 2Université Côte d'Azur, Nice, France, 3University of Leipzig, Leipzig, Germany.

Background and aims: Low-grade chronic inflammation and reduced adipocyte differentiation capacity are hallmarks of hypertrophic adipose tissue (AT). These alterations correlate with low activity of PPARγ, the master adipogenic regulator. We recently identified the dominant-negative isoform PPARγΔ5 overexpressed in adipose tissue (AT) of obese/diabetic patients, able to impair PPARγ activity in hypertrophic adipocytes and differentiation capacity of adipocyte precursors. Our study investigates how inflammatory stimuli - abundant in hypertrophic AT - act on PPARG splicing, whose alteration could contribute to impaired neoadipogenesis and insulin resistance in obese individuals with increased risk to develop type 2 diabetes (T2D).

Materials and methods: Epidydimal AT was isolated from LPS-injected C57BL/6J mice. Murine J774A.1 and human THP 1-derived macrophages (MΦ) were stimulated with LPS. Primary human monocytes were differentiated in LPS/IFN or IL-10 induced MΦ. Murine 3T3-L1 cells and human hTERT-immortalized adipose tissue-derived MSCs were treated with conditioned media of MΦ and recombinant cytokines. RNA from subcutaneous AT (SAT) biopsies were obtained from lean (BMI≤25 kg/m2; n= 14) and obese (BMI≥30 kg/m2; n= 24) individuals. Gene expression was analyzed by qPCR and P val by paired Student’s t test.

Results: The expression analysis of canonical (cPparg) and dominant (PpargΔ5) Pparg transcripts in inflamed AT of LPS-treated mice revealed a marked repression of cPparg and 3.5-fold increase of PpargΔ5/cPparg ratio (n=6; p val<0.01), paralleled by reduced expression of Pparg-regulated genes (AdipoqSlc2a4 and Cd36). As high PPARγΔ5 levels impair adipogenic capacity of precursor cells and it is hallmark of hypertrophic obesity, we evaluated the effects of inflammatory factors on differentiation capacity of 3T3-L1 cells and on the expression and splicing pattern of Pparg. Interestingly, LPS-activated MΦ secretome impairs adipocyte differentiation markedly increasing PpargΔ5/cPparg ratio and impairing AdipoqSlc2a4 and Cd36 expression. Similar effects on PPARG expression and splicing alteration were observed in human MSCs exposed to conditioned media of pro-inflammatory MΦ. We identified TNFα as one of the pro-inflammatory factor able to increase PPARGΔ5/cPPARG ratio (2.5-fold; p val<0.01). This mechanism involves SRp40 phosphorylation through the PI3K/Akt signaling activation. According to in vitro data, TNFA is over-expressed in the SAT of obese (vs lean) patients and positively correlates with PPARGΔ5 levels (r=0.45, p val=0.027) only in obese patients and not in lean individuals.

Conclusion: Our study reveals for the first time the modulation of PPARG splicing in adipose precursor cells by pro-inflammatory stimuli. The alteration of the balance between canonical and dominant negative PPARG isoforms may contribute to impair PPARγ activity in hypertrophic AT. This process exacerbates the defective adipogenic capacity of precursor cells, hallmark of insulin-resistant adipose tissue driving T2D onset and progression.

Supported by: PON Ricerca e Innovazione, PON Ars01_01270 “Innovative Device For SHAping the Risk of Diabetes”

Disclosure: S. Cataldi: None.

47

Empagliflozin-induced gut microbiota alternation reduces obesity in high-fat diet-fed mice

J. Shi1, H. Qiu1, N. Hou1, Y. Liu1, F. Han2, C. Kan1, X. Sun1;

1Department of Endocrinology, Affiliated Hospital of Weifang Medical University, Weifang, 2Department of Pathology, Affiliated Hospital of Weifang Medical University, Weifang, China.

Background and aims: Obesity has become a global epidemic, which can lead to many chronic complications. Empagliflozin,known as Sodium-glucose co-transporter-2 (SGLT2) inhibitors with putative anti-obesity effects, but its anti-obesity mechanisms remain unclear. Gut microbiota plays a crucial role in obesity; however, whether empagliflozin is involved in gut microbiota alternation and improves obesity remains elusive. Here, we sought to determine the anti-obesity effects of Empagliflozin are related to modulations in the gut microbiota and metabolic functions.

Materials and methods: Six-week-old C57BL/6 mice were fed with a normal chow diet or an HFD for 12 weeks and then were treated with or without empagliflozin (10mg/kg) for another 8 weeks. Intestinal substance was taken as samples to estimate alternation of gut microbiota. DNA of intestinal substance was isolated, and whole-genome shotgun libraries were prepared and sequenced. The composition of the gut microbiota and related pathways were assessed by analyzing metagenomic sequences.

Results: HFD mice showed significantly increased body weight, fat mass, decreased glucose tolerance and insulin sensitivity (body weight, 45.4 ± 1.3 g vs. 29.0 ± 1.0 g; fat, 12.6 ± 1.9 g vs. 2.6 ± 1.3 g; P < 0.05), and with lower species diversity of the gut microbiota and down-regulated lipid metabolism. Treatment of HFD mice with empagliflozin reduced body weight, fat mass, improved glucose tolerance and insulin sensitivity (body weight, 40 ± 0.9 g vs. 45.4 ± 1.3 g; fat, 8.5 ± 1.5 g vs. 12.6 ± 1.9 g; P < 0.05). Compared with high-fat diet mice, gut microbiota of mice treated with empagliflozin has a higher species diversity and a lower evenness degree. Mice treated with empagliflozin show a remarkable alteration in microbiota composition compared with that of high-fat diet mice. This alteration is characterized by an enrichment of genus TuricimonasCronobacterMethanobrevibacter_ALawsonella and Parasutterella. We further explored whether altered-microbiota has anti-obesity functions. We found these microbes could reduce sphingolipid metabolism, glycosphingolipid biosynthesis-globo and isoglobo series, galactose metabolism, glycerolipid metabolism, flavone and flavonol biosynthesis. These functions contributed to modulate lipid metabolism, reduce inflammation and improve intestinal barrier function.

Conclusion: Our study demonstrates that empagliflozin-induced microbiota plays a crucial role in controlling obesity development and brings new insights to a potential therapy based on host-microbe interactions.

Supported by: National Natural Science Foundation of China (81870593)

Disclosure: J. Shi: None.

48

Heat shock 70 kDa protein 4 declines after bariatric surgery in association with markers of inflammation and glycaemia

H. Fachim1, Z. Iqbal1, M. Gibson1, I. Baricevic-Jones2, A. Campbell3, B. Geary3, R. Donn4, A. Syed1, A. Whetton3, H. Soran4, A. Heald1;

1Endocrinology & Diabetes, Salford Royal NHS Foundation Trust, Salford, 2Stoller Group, The University of Manchester, Manchester, 3Stoller Group, The University of Manchester, Manchester, 4Faculty of Medicine and Heath, The University of Manchester, Manchester, UK.

Background and aims: Bariatric surgery (BS) can have significant effects on multiple body systems. Heat shock 70 kDa protein 4 (HSPA4) functions as a cytosolic chaperone involved in facilitating protein folding, degradation, complex assembly, and translocation. Elevation of HSPA4 is associated with length of disease in type 2 diabetes mellitus (T2DM). HSPA4 has also been associated with inhibition of nitric oxide production in individuals with T2DM, correlating with CRP. Blood levels of HSPA4 correlate with the degree of microalbuminuria in T2DM.Aim: Identification of plasma proteomic biomarkers of inflammatory change in individuals achieving remission from T2DM following BS.

Materials and methods: Longitudinal plasma samples from 10 individuals who achieved remission of T2DM following Roux-en-Y gastric bypass (n=7) or Sleeve gastrectomy (n=3) were analysed. SWATH MS proteomics was performed on plasma samples taken at 4 months before, and 6 and 12 months after BS and concurrent analyses of a panel of inflammatory/metabolic parameters carried out. Change in absolute abundances of those proteins showing significant change at both 6 and 12 months was correlated with change in anthropometric and other parameters including body mass index (BMI), CRP, IL6, glycosylated haemoglobin (HbA1c) and fasting insulin.

Results: BMI (p<0.001) and HbA1C (p<0.001) declined significantly at 6 and 12 months. IL6, CRP HOMA -IR, HOMA-B also declined. Four proteins related to inflammation were identified that showed consistent change at 6 and 12 months post-BS; three showed negative fold changes (FC): HSPA4 (Figure) (6M FC -0.5, p = 0.026 and 12M FC = -0.39, p = 0.022), Proteoglycan 4 (6M FC -0.89, p = 0.011 and 12M FC = -0.78, p = 0.014), Serotransferrin (6M FC -1.06, p < 0.001 and 12M FC = -0.77, p = 0.002) and one showed positive FC: Sex hormone binding globulin (SHBG) (6M FC 1.48, p = 0.012 and 12M FC = 1.95, p = 0.003). Analysis at 12 months revealed HSPA4 to correlate inversely with IL6 (r =-0.87, p = 0.0045) (Figure), SHBG (r = -0.73, p = 0.04) and HBA1C (r = -0.73, p = 0.02)(Figure).

Conclusion: We have demonstrated a significant reduction in HSPA4 which correlated with change in IL6 levels between baseline and 12 months, and change in HbA1C. This is in keeping with previous work demonstrating reduction in both serum and liver HSPA4 in a murine model after BS, and accords with the narrative that HSPA4 appears to be induced by states of chronic inflammation such as T2DM. Using SWATH MS we have identified HSPA4 as a key protein implicated in inflammatory pathway changes, potentially involved in remission from T2DM after BS.

figureq

Disclosure: H. Fachim: None.

OP 09 SGLT2 inhibitor trials

49

Safety and efficacy of SGLT2-inhibitors in over 70 years old type 2 diabetic patients: 1 year of follow up

M.E. Lunati1, M. Trevisan2, A. Gandolfi1, C. Pace3, N. Betella4, G. Favacchio4, M. Bulgheroni5, L. Bucciarelli5, G. Massari6, C. Mascardi6, A. Girelli6, P.S. Morpurgo1, M. Mirani4, C. Berra5, P. Fiorina1,7;

1ASST FatebeneFratelli-Sacco, Milano, 2Department of Pathophysiology and Transplantation, Università degli studi di Milano, Milano, 3Università degli studi di Milano, Milano, 4IRCCS Humanitas Research Hospital, Milano, 5IRCCS MultiMedica Sesto San Giovanni, Milano, 6ASST Spedali Civili Brescia, Milano, 7Department of Clinical and Biomedical Sciences, Università di Milano, “L. Sacco”, Milano, Italy.

Background and aims: Sodium-glucose co-transporter-2 inhibitors (SGLT2i) may have important benefits for older people, however some safety concerns still limit the use of this treatment in older patients. This is a multicenter study, aimed to analyze efficacy and safety of SGLT2i in elderly in a real-life setting.

Materials and methods: We analyzed 450 adults (mean age 74,7±3,9 years, F/M 183/267) with T2D, that started SGLT2i-based treatment after the age of 70, with at least one year of follow-up. Patients were evaluated at baseline and after 6 and 12 month of SGLT2i treatment. Glycometabolic and antropometric parameters, adverse events and causes of drug discontinuation were recorded.

Results: 450 T2D aged≥70 years started SGLT2i (47.6% Empagliflozin, 41.6% Dapagliflozin, 9.1% Canagliflozin, 1.7% Ertugliflozin) as add-on therapy to Metformin in 79.6% of subjects, insulin treatment in 33.1% and other antidiabetic drugs in 16% (i.e. 38.9% sulfonylureas, 33.3% DPP-IV inhibitors, 16.7% pioglitazone, 8.3% GLP-1RA and 6.9% acarbose). At baseline arterial hypertension affected 88.1% of patients, while cardiovascular events occurred in 38.5% of subjects. The analysis of glucose metabolic parameters during follow-up, showed a statistically significant reduction of HbA1c and FPG during follow-up (HbA1c: 7.7±1.15 vs 7.4±1.0 vs 7.3±0.9%, p=0.001; FPG: 158±48 vs 146±41.3 vs 140±36.5%, p=0.001), while BMI reduced in non statistically significant manner (29,9±5,3 vs 29,2±5,2 vs 29,5±5,1 Kg/m2, p=0.17). Overall, eGFR values remains stable over time (74,5±15 vs 72.2±16 vs 72.4±16,1 ml/min/1,73m2, p=0.314). We then analyzed the subgroup of patients with eGFR decline over 20% (12,4%). At baseline they were characterized by higher percentage of cardiovascular events history (58,6% vs 32,5%, p=0,006), higher U-Albumin values (110,7±153,3 vs 43,9±78,7 mg/L, p=0,036) and higher serum creatinine values (0,95±0,15 vs 0,9±0,2, p=0,009) while age, BMI and HbA1c values were comparable. SGLT2i discontinuation occured in 7,4% of subjects after 6 months and in 7% after 12 months. Main reasons of discontinuation were intolerance in the 60,7% and UTI in the 25.5%. We recorded no cases of diabetic ketoacidosis or amputation. Cardiovascular adverse events during follow-up occured in a small proportion of patients: in the 1,4% of patients during first 6 months and 2,2% between 6 and 12 months of follow-up.

Conclusion: The SGLT2-inhibitor class is well-tolerated and safe in elderly though some caution is also suggested, expecially in frailer subjects.

Disclosure: M.E. Lunati: None.

50

Cardiorenal outcomes with ertugliflozin by baseline cardiorenal medications: an analysis from VERTIS CV

C.P. Cannon1, F. Cosentino2, D.K. McGuire3,4, B. Charbonnel5, S. Dagogo-Jack6, R.E. Pratley7, W.J. Shih8,9, M. Maldonado10, A. Pong11, I. Gantz11, R. Frederich12, J.P. Mancuso13, U. Masiukiewicz13, D.Z.I. Cherney14;

1Cardiovascular Division, Brigham and Women's Hospital, Harvard Medical School, Boston, USA, 2Unit of Cardiology, Karolinska Institute & Karolinska University Hospital, Stockholm, Sweden, 3University of Texas Southwestern Medical Center, Dallas, USA, 4Parkland Health and Hospital System, Dallas, USA, 5University of Nantes, Nantes, France, 6University of Tennessee Health Science Center, Memphis, USA, 7AdventHealth Translational Research Institute, Orlando, USA, 8Rutgers School of Public Health, Piscataway, USA, 9Rutgers Cancer Institute of New Jersey, Piscataway, USA, 10MSD Limited, London, UK, 11Merck & Co., Inc., Kenilworth, USA, 12Pfizer Inc., Collegeville, USA, 13Pfizer Inc., Groton, USA, 14University of Toronto, Toronto, Canada.

Background and aims: VERTIS CV was a placebo-controlled cardiovascular (CV) outcome trial evaluating the sodium-glucose cotransporter 2 inhibitor, ertugliflozin. The aim of the current analyses was to evaluate cardiorenal endpoints by baseline use of CV medications.

Materials and methods: A Cox proportional hazard assessment was used to analyse pre-specified CV and kidney outcomes according to use of renin-angiotensin-aldosterone system inhibitors (RAASi), diuretics and mineralocorticoid receptor antagonists (MRA) at baseline.

Results: Among 8246 randomised patients, at baseline 6686 (81%) were treated with RAASi, 3542 (43%) with diuretics, including 1252 (15%) with loop diuretics, and 674 (8%) with MRA. No significant interactions were observed for cardiorenal outcomes by baseline use of RAASi or MRA (Pinteraction >0.05 for all). Nominally significant interactions for first event of hospitalisation for heart failure (HHF) and HHF/CV death were observed with baseline use of diuretics, including loop diuretics, with an increased benefit of ertugliflozin vs placebo (Figure).

Conclusion: In VERTIS CV, baseline use of diuretics, including loop diuretics, appeared to be associated with greater benefit of ertugliflozin on first HHF and HHF/CV death, with no modification of effect based on baseline use of RAASi or MRA.

figurer

Clinical Trial Registration Number: NCT01986881

Supported by: Merck Sharp & Dohme Corp., a subsidiary of Merck & Co., Inc., Kenilworth, NJ, USA, and Pfizer Inc., New York, NY, USA

Disclosure: C.P. Cannon: Grants; Amgen, Boehringer Ingelheim, Bristol-Myers Squibb, Daiichi Sankyo, Janssen, Merck, and Pfizer. Lecture/other fees; Aegerion, Alnylam, Amarin, Amgen, Applied Therapeutics, Ascendia, Boehringer Ingelheim, Bristol-Myers Squibb, Corvidia, Eli Lilly, HLS Therapeutics, Innovent, Janssen, Kowa, Merck, Pfizer, Rhoshan and Sanofi.

51

Efficacy and safety of dapagliflozin on kidney and cardiovascular outcomes by baseline albuminuria: a secondary analysis of the DAPA-CKD trial

H.J. Lambers Heerspink1, S.W. Waijer1, P. Vart1, D.Z.I. Cherney2, G. Chertow3, A.M. Langkilde4, J.J.V. McMurray5, P. Rossing6, R. Correa-Rotter7, B.V. Stefansson4, R. Toto8, D. Wheeler9;

1University Medical Centre Groningen, Groningen, Netherlands, 2University of Toronto, Toronto, Canada, 3Stanford University School of Medicine, Stanford, USA, 4AstraZeneca, Gothenburg, Sweden, 5University of Glasgow, Glasgow, UK, 6Steno Diabetes Center, Copenhagen, Denmark, 7National Medical Science and Nutrition Institute Salvador Zubirán, Ciudad de México, Mexico, 8Department of Internal Medicine, University College London, London, UK, 9Department of Renal Medicine, University College London, London, UK.

Background and aims: Dapagliflozin reduced the risk of kidney failure and heart failure hospitalisations (HHF) in chronic kidney disease (CKD) patients, with or without type 2 diabetes (T2D), in the DAPA-CKD trial. Patients with CKD with higher levels of albuminuria are at higher risk of kidney failure and HHF. We assessed dapagliflozin according to baseline albuminuria.

Materials and methods: DAPA-CKD was a randomised, placebo-controlled trial. Adult patients with CKD (urinary albumin-to-creatinine ratio [UACR] 200-5000 mg/g; estimated glomerular filtration rate [eGFR] 25-75 mL/min/1.73m2) were randomised to dapagliflozin 10 mg/d or placebo. Primary outcome was a composite of ≥50% reduction in eGFR, end-stage kidney disease (ESKD) and kidney or cardiovascular death. Secondary outcomes were the kidney outcome (≥50% reduction in eGFR, ESKD and kidney death), HHF and cardiovascular death and death from all causes. We used Cox proportional hazards regression analyses to assess relative and absolute effects of dapagliflozin across subgroups of UACR (≤1000, >1000 to ≤3500, >3500 mg/g).

Results: At randomisation, 2225 (51.7%), 1764 (41.0%) and 315 (7.3%) patients had an UACR of ≤1000, >1000 to ≤3500, >3500 mg/g, respectively. Dapagliflozin reduced relative risk of the primary and secondary outcomes consistently across subgroups of UACR (all P-interaction >0.59; primary outcome shown in Table). Absolute risk reductions for the primary outcome were greater in subgroups with higher UACR: the absolute risk reductions across ascending UACR categories were 3.5% (95%CI 1.6 to 5.4), 6.9% (95%CI 3.6, 10.1) and 13.8% (95%CI 3.0, 24.5; P-interaction 0.02). When patients with and without T2D were analysed separately, we observed that relative risk reduction with dapagliflozin on the primary outcome was consistent across UACR subgroups (P-interaction >0.43, Table). Proportion of patients with adverse events leading to study drug discontinuation and proportion of patients with serious adverse events were similar between the groups; this was consistent across UACR subgroups.

Conclusion: Efficacy and safety of dapagliflozin on kidney and cardiovascular outcomes was consistent across subgroups of baseline UACR. These results indicate dapagliflozin is effective and safe across a wide range of baseline UACR levels in patients with CKD with and without T2D.

figures

Clinical Trial Registration Number: NCT03036150

Supported by: Support received from AstraZeneca

Disclosure: H.J. Lambers Heerspink: Employment/Consultancy; AstraZeneca. Grants; AstraZeneca. Non-financial support; AstraZeneca.

52

Ertugliflozin in older patients with type 2 diabetes: an analysis from VERTIS CV

R.E. Pratley1, C.P. Cannon2, B. Charbonnel3, D.Z.I. Cherney4, F. Cosentino5, D.K. McGuire6,7, W.J. Shih8,9, M.N. Essex10, D. Lawrence10, P.L.S. Jones11, J. Liu12, I.A. Adamsons12, S. Dagogo-Jack13;

1AdventHealth Translational Research Institute, Orlando, USA, 2Cardiovascular Division, Brigham and Women's Hospital, Harvard Medical School, Boston, USA, 3University of Nantes, Nantes, France, 4University of Toronto, Toronto, Canada, 5Unit of Cardiology, Karolinska Institute & Karolinska University Hospital, Stockholm, Sweden, 6University of Texas Southwestern Medical Center, Dallas, USA, 7Parkland Health and Hospital System, Dallas, USA, 8Rutgers School of Public Health, Piscataway, USA, 9Rutgers Cancer Institute of New Jersey, Piscataway, USA, 10Pfizer Inc., New York, USA, 11Pfizer Ltd., Sandwich, UK, 12Merck & Co., Inc., Kenilworth, USA, 13University of Tennessee Health Science Center, Memphis, USA.

Background and aims: VERTIS CV was a cardiovascular (CV) outcome trial of ertugliflozin vs. placebo in patients (pts) with T2D and atherosclerotic CV disease.

Materials and methods: This analysis assessed the cardiorenal outcomes by Cox proportional hazard assessment in older vs. younger patients (<65 yrs and ≥65 yrs). The safety of ertugliflozin in older vs. younger patients was summarised descriptively.

Results: Among 8238 pts randomised and treated with ertugliflozin 5 mg, 15 mg, or placebo, 4093 (49.7%) were <65 yrs, 4145 (50.3%) were ≥65 yrs, with 903 (11.0%) ≥75 yrs. Baseline characteristics were similar across treatment groups within each age group; mean eGFR was lower in older pts (82.2, 69.9, and 63.6 mL/min/1.73m2 for pts <65, ≥65, and ≥75 yrs, respectively). Cardiorenal outcomes for ertugliflozin vs. placebo did not differ between pts ≥65 and <65 yrs, with interaction P values >0.05 for all outcomes assessed (Figure). The incidences of adverse events (AEs), serious AEs (including deaths), and discontinuations due to an AE were higher in older vs. younger pts, but generally similar across treatment groups in pts ≥65 and ≥75 yrs. The incidence of renal-related AEs was similar across treatment groups in younger and older pts. The incidences of genital mycotic infection and urinary tract infection were higher with ertugliflozin vs. placebo in older pts. The incidences of hypovolemia AEs with ertugliflozin vs. placebo in pts ≥65 was 5.3% vs. 4.5% and in pts ≥75 yrs was 6.9% vs. 5.8%.

Conclusion: Cardiorenal outcomes were similar in pts ≥65 and <65 yrs. Ertugliflozin was generally well tolerated in pts ≥65 and ≥75 yrs.

figuret

Clinical Trial Registration Number: NCT01986881

Supported by: Merck Sharp & Dohme Corp., a subsidiary of Merck & Co., Inc., Kenilworth, NJ, USA, and Pfizer Inc., New York, NY, USA.

Disclosure: R.E. Pratley: Employment/Consultancy; AstraZeneca, Glytec, Janssen, Merck & Co. Inc., Mundipharma, Novo Nordisk, Pfizer, Sanofi, Sanofi US Services, Inc. Scohia Pharma Inc., and Sun Pharmaceutical Industries. Grants; Hanmi Pharmaceutical Co., Ltd, Janssen, Lexicon Pharmaceuticals, Inc., Novo Nordisk, Poxel SA, Sanofi. Lecture/other fees; Novo Nordisk.

53

Empagliflozin versus dapagliflozin for type 2 diabetes in combination with metformin, dipeptidyl peptidase-4 inhibitor and sulfonylurea: 3-year prospective study

E. Ku, H. Jeon, D.-H. Lee, T. Oh;

Internal Medicine, Chungbuk National University Hospital, Cheongju, Republic of Korea.

Background and aims: Sodium glucose cotransporter 2 (SGLT2) inhibitor is a relatively new type of antidiabetic medication that has beneficial effects on glycemic control, cardiometabolic aspects, and has been increasingly used by proving not only cardiovascular outcome but also renal protection effect in recent years. However, there was no long-term efficacy and safety study for quadruple combination therapy with empagliflozin and dapagliflozin. Therefore, we aimed to evaluate the efficacy and safety of 3-year follow-up of empagliflozin and dapagliflozin as add-on in patients with type 2 diabetes mellitus (T2DM) inadequately controlled despite triple oral antidiabetic agents (OADs).

Materials and methods: A single center, 3-year, open-label, prospective observational study was started between March 2015 and March 2018. Eligible patients were T2DM aged 19years and older, had suboptimal glycemic control (HbA1c 7.5-12.0%) despite maximum tolerated doses of metformin, dipeptidyl-peptidase 4 inhibitor, and glimepiride with stable dosage for at least 3 months. Patients were treated adding empagliflozin (25mg/day) or dapagliflozin (10mg/day) as a fourth OAD to the existing triple combination therapy at the physician’s discretion. Efficacy outcomes included the change in HbA1c, fasting plasma glucose (FPG), and other cardiometabolic variables at 3 years. Safety outcomes included adverse events (AEs), hypoglycemia, volume depletion, nocturia, genitourinary tract infections (GUTIs) and laboratory tests. All procedures were conducted in accordance with the Helsinki Declaration and the International Conference of Harmonization/Good Clinical Practice guidelines.

Results: A total of 362 patients were enrolled with empagliflozin (25 mg/day, n=185) and dapagliflozin (10 mg/day, n=177), respectively. At 3 years, the mean HbA1c reduction was -1.7% (SD 1.1) in the empagliflozin group versus -1.1% (1.3) in the dapagliflozin group (P=0.001). FPG also showed a similar trend (-59.3 ± 53.3 mg/dL vs -47.4 ± 61.7 mg/dL, empagliflozin and dapagliflozin, respectively, P=0.055). In terms of body weight, empagliflozin showed significantly greater reduction (-4.5 ± 2.5 kg vs. -1.0 ± 3.3 kg, P=0.024). Although the proportion of patients reporting total and serious adverse events was similar, volume depletion was reported higher in the dapagliflozin group and asymptomatic pyuria or bacteriuria was reported higher in the empagliflozin group.

Conclusion: The efficacy of addition of SGLT2 inhibitors as a quadruple combination therapy has been sustained for 3 years. In the respects of blood glucose control, empagliflozin (25 mg/day) was stronger than dapagliflozin (10 mg/day).

Clinical Trial Registration Number: NCT03748810

Supported by: EJK/2021 Research grant from the Chungcheong Society of the Korean Society of Endocrinology

Disclosure: E. Ku: None.

54

Effects of empagliflozin on uric acid levels and gout: observations from the EMPA-REG OUTCOME trial

J.P. Ferreira1, S.E. Inzucchi2, B. Zinman3, M. Mattheus4, T. Meinicke5, D. Steubl5, C. Wanner6;

1Université de Lorraine, Nancy, France, 2Yale University School of Medicine, New Haven, USA, 3Lunenfeld-Tanenbaum Research Institute, Mount Sinai Hospital, University of Toronto, Toronto, Canada, 4Boehringer Ingelheim Pharma GmbH & Co KG, Ingelheim, Germany, 5Boehringer Ingelheim International GmbH, Ingelheim, Germany, 6Würzburg University Clinic, Würzburg, Germany.

Background and aims: Higher serum uric acid (UA) levels are associated with gout and other adverse health outcomes in patients with type 2 diabetes (T2D). Sodium glucose co-transporter 2 inhibitors (SGLT2i) lower UA. Moreover, in the CANVAS program, the SGLT2i canagliflozin reduced gout flares in patients with T2D. Whether empagliflozin also positively impacts new-onset of gout or anti-gout medication in T2D patients has not been fully evaluated. We studied the effect of empagliflozin on UA levels, incident prescription of an anti-gout/UA-lowering medication or an episode of gout as an adverse event in the EMPA-REG OUTCOME trial.

Materials and methods: In EMPA-REG OUTCOME, 7,020 patients with T2D at high CV risk were treated with either empagliflozin (pooled n=4,687; 10 mg/day n=2,345; 25 mg/day n=2,342) or placebo (n=2,333). We assessed the effects of empagliflozin vs. placebo on serum UA levels using MMRM analyses and Cox proportional hazards models for the time-to-first occurrence of either an adverse event attributed to gout flare or initiation of a drug for the treatment of hyperuricemia or gout while patients were on-treatment.

Results: Of the 7,020 patients included at baseline, 413 (6%) were taking UA-lowering medications and 6,607 (94%) were not. Patients on UA-lowering medications were older, more frequently male, had a higher BMI, longer T2D duration, poorer kidney function, higher albuminuria, and more often treated with diuretics. At 12 weeks, mean serum UA level was lower in empagliflozin-treated patients vs. placebo, and this difference was maintained over all subsequent timepoints; by week 52 the adjusted mean (95% CI) treatment difference was −0.37 (−0.42, −0.31) mg/dL. The adjusted mean treatment difference (95% CI) in serum UA at week 52 was more pronounced with baseline UA levels ≥6 mg/dL (−0.46 (−0.55, −0.38) mg/dL, panel A). Among the 6,607 patients not taking UA lowering medications at baseline, 5.2% experienced a gout flare or initiated a drug for the treatment of gout in the placebo group vs. 3.6% in the pooled empagliflozin group, corresponding to incidence rates of 21.6 vs.14.1 events per 1,000 patient-years, and a hazard ratio (95% CI) of 0.67 (0.53, 0.85), P=0.001 (panel B). The event rate reduction was similar with either empagliflozin 10 or 25 mg and in patients with baseline UA levels below or above 6.0 mg/dL as well as with vs. without a medical history of gout at baseline (interaction P=0.53 and 0.48, respectively).

Conclusion: Empagliflozin reduced mean UA, the need for UA lowering medications or gout flares vs. placebo. These data support further investigations into SGLT2i in the prevention of gout and reduction of anti-gout medication.

figureu

Clinical Trial Registration Number: NCT01131676

Supported by: Boehringer Ingelheim & Eli Lilly and Company Diabetes Alliance

Disclosure: J.P. Ferreira: Honorarium; Boehringer Ingelheim.

OP 10 Peripheral neuropathy: pathophysiology and intervention

55

Advanced glycation end-products are associated with diabetic neuropathy in young adults with type 1 diabetes

E.F. Al-Saoudi1, M.M.B. Christensen1, P. Nawroth2, T. Fleming2, E.E. Hommel1, M.E. Jørgensen1,3, J. Fleischer4, C.S. Hansen1;

1Steno Diabetes Center Copenhagen, Gentofte, Denmark, 2Department of Medicine I and Clinical Chemistry, University Hospital Heidelberg, Heidelberg, Germany, 3University of Southern Denmark, Odense, Denmark, 4Steno Diabetes Center Aarhus, Aarhus, Denmark.

Background and aims: Cardiovascular autonomic neuropathy (CAN) and sensorimotor polyneuropathy (DSPN) are common diabetic complications. The formation of advanced glycation end-products (AGEs) as a consequence of hyperglycemia and lipid toxicity has been suggested to play a causal role in the pathogeneses of diabetic neuropathy. The aim was to investigate the association between AGEs and measures of CAN and DSPN in young adults with type 1 diabetes.

Materials and methods: In a cross-sectional study design of young adults with type 1 diabetes, CAN was assessed by three cardiovascular autonomic reflex tests (CARTs); lying-to-standing test, the deep breathing test (E/I) and Valsalva manoeuvre, and by 5-min resting heart rate variability (HRV) indices; the mean square of the sum of the squares of differences between consecutive R-R intervals and standard deviation of normal-to-normal intervals (SDNN), high (HF) and low frequency (LF) power, total frequency power and the LF/HF-ratio. DSPN was assessed by light touch and pain perception, vibration perception threshold (VPT), neuropathy questionnaires, electrochemical skin conductance and measures of sural nerve function. Associations between serum levels of AGEs and neuropathy outcomes were analysed by regression analyses by one unit increases of standardized z-scores of groups of AGEs categorized by origin; i. “Glycolytic dysfunction”; methylglyoxal-derived hydroimidazolone 1, Nɛ-(carboxyethyl)-lysine, methylglyoxal-lysine dimers and argpyrimidine. ii. “Lipid peroxidation”; glyoxal-derived hydroimidazolone 1 and Nɛ-(carboxymethyl)-lysine. iii. “Oxidative stress”; methionine sulphoxide. iv. “Glucotoxicity”; fructoselysine, glucosepane and 3-deoxyclucosone. The analyses were adjusted for relevant confounders; age, gender, HbA1c, diabetes duration, current smoking status, total cholesterol, triglycerides, systolic blood pressure and use of beta-blockers.

Results: One hundred and fifty-one patients were examined aged 22 years (SD 1.6) with a mean diabetes duration of 11 years (SD 5.1) and HbA1c was 67.0 mmol/mol (IQR 58.0;77.0). Higher z-scores in several AGE groups were associated with several outcomes: “Glycolytic dysfunction” was associated with a higher VPT (4.14% (95%CI 1.31;7.04), p = 0.004) and E/I (0.03% (95%CI 0.01;0.05), p = 0.005) with full adjustments. “Lipid peroxidation” was associated with higher VPT only in unadjusted models (12.90% (95%CI -2.30;30.45), p = 0.102), and a higher LF/HF ratio (37.72% (95%CI 1.12;87.57), p = 0.044) with full adjustments. “Glucotoxicity” was associated to lower values of several HRV indices and in adjusted models SDNN retained significance (-4.20% (95%CI -8.1416; -0.0896), p = 0.047).

Conclusion: In young adults with type 1 diabetes, increased levels of AGEs derived from glycolytic dysfunction and lipid peroxidation were associated with worse vibration perception threshold. Increased levels of AGEs from glucotoxicity were associated with autonomic dysfunction. We conclude that even in early stages of diabetes AGEs may play a diverse role in the pathogeneses of different types of diabetic neuropathy.

Disclosure: E.F. Al-Saoudi: None.

56

Progression and regression of small and large nerve fiber pathology and dysfunction in recent-onset type 1 and type 2 diabetes: a 5-year prospective study

G.J. Bönhof1,2, A. Strom2,3, K. Straßburger4,3, Y. Karusheva2,3, J. Szendroedi1,2, M. Roden1,2, D. Ziegler1,2, GDS group;

1Department of Endocrinology and Diabetology, Medical Faculty and University Hospital Düsseldorf, Heinrich-Heine-University, Düsseldorf, 2Institute for Clinical Diabetology, German Diabetes Center, Leibniz Center for Diabetes Research at Heinrich-Heine-University, Düsseldorf, 3German Center for Diabetes Research, Partner Düsseldorf, München-Neuherberg, 4Institute for Biometrics and Epidemiology, German Diabetes Center, Leibniz Center for Diabetes Research at Heinrich Heine University, Düsseldorf, Germany.

Background and aims: It has been traditionally suggested that the early development of diabetic sensorimotor polyneuropathy (DSPN) is characterized by a predominant and progressive injury to small nerve fibres followed by large fibre impairment. We tested an alternative hypothesis that small and large fibre damage due to DSPN in type 1 and type 2 diabetes could develop in parallel and may not only be progressive but also reversible.

Materials and methods: Participants from the German Diabetes Study baseline cohort with recent-onset type 1/type 2 diabetes (n=350/570) and age-matched glucose-tolerant control individuals (control 1/control 2: n=114/190) were assessed by nerve conduction studies (NCS), thermal detection thresholds (TDT), vibration perception threshold (VPT), Neuropathy Symptom Score (NSS), Neuropathy Disability Score (NDS), and intraepidermal nerve fibre density (IENFD) in skin biopsies (type 1/type 2 diabetes: n=102/225; control 1/control 2: n=109/208). Subsets of participants with type 1/type 2 diabetes were followed for 5 years (n=184/307; IENFD subset: n=18/69). DSPN was defined by the Toronto Consensus criteria.

Results: At baseline, DSPN was present in 8.1% and 13.3% of the type 1 and type 2 diabetes groups, respectively. The most frequently abnormal tests in the lower limbs below or above the 2.5th and 97.5th centile of the controls were IENFD (13.7%) and individual NCS (up to 9.4%) in type 1 diabetes participants and IENFD (21.8%), malleolar VPT (17.5%), and individual NCS (up to 11.8%) in those with type 2 diabetes, whereas TDT abnormalities did not differ between the control and diabetes groups. The risk factors most consistently associated with impaired peripheral nerve tests across the groups studied were higher age, height, and weight. IENFD correlated variably with both small and large fibre function tests in the control and diabetes groups. After 5 years in the type 2 diabetes group, the highest progression rates from the normal to the abnormal range were found for IENFD (18.6%), malleolar VPT (18.4%), and NDS (15.0%), while vice versa the highest regression rates were observed for NDS (11.2%), sural nerve amplitude (9.1%), IENFD (8.7%), and NSS (8.2%). In type 1 diabetes participants, no major progression was seen after 5 years, but subclinical DSPN regressed in 10.3%.

Conclusion: These findings point to an early parallel damage to both small and large nerve fibres in well-controlled recent-onset type 2 and, to a lesser extent, type 1 diabetes. After 5 years peripheral nerve pathology and dysfunction progresses in type 2 diabetes, but initial nerve alterations are also reversible to a meaningful degree.

Disclosure: G.J. Bönhof: None.

57

Dys- but not demyelination is the hallmark of diabetic neuropathic lesions

M. Le Marois1, R. Longuespee2, D. Schwarz3, J. Szendrödi1, P. Nawroth1, T. Fleming1;

1Department of Medicine I and Clinical Chemistry, Universitätklinikum Heidelberg, Heidelberg, 2Department of Clinical Pharmacology and Pharmacoepidemiology, Universitätklinikum Heidelberg, Heidelberg, 3Department of Neuroradiology, Universitätklinikum Heidelberg, Heidelberg, Germany.

Background and aims: Diabetic neuropathy (DN) remains the most prevalent complication of individuals with diabetes, affecting up to 50% of patients with either type 1 and type 2 diabetes. Magnetic resonance neurography (MRN) has shown that patients with DN have lesion on the proximal region of the sciatic nerve, which is associated with the clinical scoring. The nature of these lesions, which are visualized in MRN as bright focus spots within the fascicles of the nerve, remains unknown. The aim of this study was to understand the nature of these lesions by studying the differences at the protein level using a microproteomic approach.

Materials and methods: Formalin-Fixed and Paraffin-Embedded (FFPE) nerve material from type 2 diabetic patients were obtained following below the knee-amputation. The location of the lesion(s) was identified by ex vivo MRI. Fascicles, with and without damage were isolated by Laser Micro Dissection (Leica LMD 7000) microscopy. Following tryptic digestion, samples were analyzed by liquid-chromatography, tandem mass spectroscopy. MaxQuant software was used to analyze raw mass spectrometric files and Perseus software was used to perform further downstream statistical and bioinformatic analyses. A fold change cut-off of either -2 or +2 were used to identified proteins that were down- or up-regulated, respectively. Proteins that were found to be significantly changed (P<0.05) were subsequently validated by immunofluorescence.

Results: Using a microproteomic approach, an average of 1141 proteins could be identified in lesioned fascicles as compared to 1228 in healthy fascicles (P > 0.05). Bioinformatics analysis showed that in lesioned fascicles, a total of 12 proteins were significantly down-regulated with 4 proteins up regulated. The most notable of the down regulated proteins were associated with myelin sheath integrity such as Myelin Basic Protein (MBP) (log2 FC = -3.1; -log10 p-value = 5.4), Myelin Protein Zero (MPZ) (log2 FC = -2.6; -log10 p-value= 5.7) and Peripheral Myelin Protein 2 (PMP2) (log2 FC = -4.2; -log10 p-value= 5.3). Proteins associated with carbohydrate metabolism were also found to be downregulated such as Aldo-Keto Reductase Family 1 Member B (AKR1B1) (log2 FC = -3.1; -log10 p-value= 4.6). Pathway analysis showed that myelination deficiency was the most prevalent condition associated with the lesions observed in diabetic patients. This deficiency was associated with an increase in proteins involved in myelin clearance proteins as well as promoters of axon growth. Interestingly, this deficiency was associated with dysmyelination, the process of malformed or defective myelin sheath rather than the classical process of demyelination.

Conclusion: The preliminary results obtained from type 2 diabetic patients suggest that incorrect formation of the myelin sheath, rather than its loss is associated with the development of lesions in DN. The finding that AKR1B1, an enzyme which catalyzes the reduction of glucose to sorbitol as part of the polyol pathway, could suggest that this pathway may have a cause role in malformation of the myelin sheath

Disclosure: M. Le Marois: None.

58

Morphometric abnormalities of the brain in diabetic peripheral neuropathy

G. Sloan1, D. Selvarajah2, K. Teh3, M. Greig1, P. Shillo4, I.D. Wilkinson3, S. Tesfaye1;

1Sheffield Teaching Hospitals, Sheffield, 2Department of Oncology and Human Metabolism, Sheffield, 3Academic Unit of Radiology, Sheffield, 4Department of Diabetes, Chesterfield, UK.

Background and aims: Advanced magnetic resonance imaging (MRI) techniques have enabled detailed interrogation of alterations in the brain in diabetic peripheral neuropathy (DPN). Small studies have demonstrated structural alterations in brain regions involved with the somatosensory and motor system in DPN and painful-DPN (pDPN). However, these preliminary findings have not been confirmed in a large cohort with detailed patient characterization. Moreover, such studies have not considered structural alterations in different pDPN phenotypes. Here we present the largest DPN neuroimaging study to date, which aims to examine morphological differences in brain structure in DPN and different phenotypes of painful-DPN.

Materials and methods: A total of 229 participants were enrolled, 177 with diabetes (46 no-DPN, 56 painless-DPN and 75 pDPN; 24 irritable [IR] and 50 non-irritable [NIR] phenotype), and 52 healthy volunteers underwent detailed clinical and neurophysiological assessments. All subjects underwent 3-dimensional T1-weighted brain MRI (3.0T, Phillips). Brain volume analysis was performed using Freesurfer (http://surfer.nmr.mgh.harvard.edu/). We used the German Research Network on Neuropathic Pain (DFNS) quantitative sensory testing (QST) profile to phenotype patients with pDPN.

Results: There was a significant group effect in the cortical thickness at the pre-central gyrus (ANOVA P=0.001), anterior cingulate cortex (ACC, p=0.036), insula cortex (p=0.021) and post-central gyrus (p=0.015), where there was also a group effect in grey matter volume (p=0.037). At the post-central gyrus, in comparison with HV (1.92mm±0.11) and NoDPN (1.90mm±0.14), there was a significantly reduced cortical thickness in DPN (1.85mm±0.14), in pDPN (1.87mm±0.12) cortical thickness was reduced compared with HV. At the precentral gyrus, both DPN (2.29mm±0.16) and pDPN (2.28mm±0.16) had significantly reduced cortical thickness compared with HV (2.37mm±0.14) and NoDPN (2.36mm±0.13). At the insula, compared with HV (2.93mm±0.15) the cortical thickness was reduced in DPN (2.85mm±0.18) and pDPN (2.84mm±0.17). The cortical thickness at the pre and postcentral gyrus, and insula correlated with measures of nerve conduction. At the ACC mean cortical thickness was reduced in the pDPN group (2.55mm±0.23) compared to DPN (2.62mm±0.25). Moreover, at the ACC the mean cortical thickness was significantly reduced in the IR- (2.40mm±0.23) compared to NIR-pDPN (2.58mm±0.20; p=0.003).

Conclusion: We present the largest study to examine cerebral morphometric alterations in patients with DPN and pDPN. We demonstrate that key somatomotor brain regions have a reduced cortical thickness in patients with DPN and pDPN, which correlate with neurophysiological measures suggestive of an ascending axonopathy. Moreover, we found that the cortical thickness at the ACC differentiated patients with the IR and NIR, suggesting neuroplasticity in this region may play a role determining different clinical phenotypes in pDPN.

Supported by: EFSD/Novo Nordisk A/S Programme for Diabetes Research in Europe 2012, DUK, NIHR, JDRF

Disclosure: G. Sloan: None.

59

Pain processing areas of the brain demonstrate altered microvascular perfusion during spontaneous neuropathic pain

M. Greig1, G. Sloan1, P. Shillo2, D. Selvarajah3, I.D. Wilkinson4, S. Tesfaye1;

1Sheffield Teaching Hospitals, Sheffield, 2Department of Diabetes, Chesterfield, 3Department of Oncology and Human Metabolism, Sheffield, 4Academic Unit of Radiology, Sheffield, UK.

Background and aims: Painful diabetic peripheral neuropathy (painful-DPN) causes distressing neuropathic pain that is only partially responsive to treatment. A better understanding of CNS correlates of painful-DPN is vital to develop more effective therapeutics. We have previously reported increased thalamic vascularity in painful diabetic peripheral neuropathy (PDPN) but the microvascular perfusion of the other pain processing areas of the brain (Pain Matrix - PM), with or without pain at the time of scanning, is unknown. Thus, the aim of this study was to measure the cerebral perfusion characteristics of the PM areas using MR-Dynamic Susceptibility Contrast (MR-DSC).

Materials and methods: 55 T1DM subjects (20 PDPN, 23 painless-DPN, 13 no-DPN) and 19 healthy volunteers (HV) underwent clinical, neurophysiological and MR-DSC of the passage of IV-gadolinium-chelate through the cerebral vascular bed (3-Tesla, Philips, Netherlands) at rest. The PDPN group was further divided into those that had neuropathic pain during scan (P+) and no pain (P-) on a visual analogue scale (VAS). The intensity of the pain at the time of scanning was recorded. The Mean Transit Time -MTT and the time-to-peak -TTP concentrations of gadolinium in PM areas (Thalamus, Insular Cortex - IC, Anterior Cingulate Cortex - ACC) were measured.

Results: Participants experiencing spontaneous neuropathic pain (P+, n=10, VAS score>4) during scanning had shorter mean TTP at the Right-Thalamus: P+ vs. painless-DPN p=0.017, P+ vs. HV p=0.033; Right-IC: P+ vs. painless-DPN p=0.048; and POWM: P+ vs. P- p=0.009, P+ vs. painless-DPN p=0.034, P+ vs. HV p=0.011. The MTT at the Right-Thalamus: P+ vs. HV p=0.043 and at the Left-IC: HV vs. No-DPN p=036, were also significant.

Conclusion: We demonstrate for the first time in PDPN that there are altered cerebral perfusion characteristics in the pain processing areas of the brain. We found shorter TTP during spontaneous pain at the time of scanning. This seems to be related to a bolus dispersal factor as MTT, and CBF do not match the TTP and the control area is also affected. However, whether this altered perfusion is primary or secondary to the experience of pain is yet to be determined. This novel finding may serve as objective marker of spontaneous neuropathic pain, and a target for the development of novel treatments.

Supported by: EFSD/Novo Nordisk A/S Programme for Diabetes Research in Europe 2012

Disclosure: M. Greig: None.

60

Long-term high frequency (10 kHz) spinal cord stimulation in painful diabetic neuropathy: a randomised controlled trial

E. Petersen, SENZA-PDN Investigators;

Neurosurgery, University of Arkansas for Medical Sciences, Little Rock, USA.

Background and aims: The World Health Organization estimates there are 422 million people living with diabetes globally and approximately 20% will develop painful diabetic neuropathy (PDN). Current treatment options are ineffective for many patients; however, previous results suggest high frequency (10 kHz) spinal cord stimulation (SCS) relieves pain and may improve sensation in patients with refractory symptoms.

Materials and methods: Prospective, randomized controlled trial with 216 patients assigned 1:1 to 10 kHz SCS (Nevro Corp.) combined with conventional medical management (CMM) or CMM alone. Key inclusion criteria: PDN symptoms ≥12 months, lower limb pain ≥5cm (on 0-10cm visual analog scale [VAS]), and medically appropriate candidate for SCS procedure. Outcomes include pain (VAS), Brief Pain Inventory for Diabetic Peripheral Neuropathy (BPI-DPN), Pain and Sleep Questionnaire (PSQ-3), Patient Global Impression of Change (PGIC), and neurological examination. Patients could opt to crossover to the other treatment arm at 6 months if they had insufficient pain relief (<50%), were dissatisfied, and their physician deemed it was medically appropriate.

Results: Both the 10 kHz SCS and the CMM treatment arms were similar in baseline characteristics. The mean age of participants was 60.8 years (SD 10.7) with mean hemoglobin A1c of 7.4% (SD 1.2%). Median duration of diabetes was 10.9 years (IQR 6.3-16.4) with median 5.6 years (IQR 3.0-10.1) of peripheral neuropathy symptoms. Patients in the 10 kHz SCS group reported a mean 76% reduction in pain at 6 months as well as reduced pain interference with daily activities, and improvements in sleep (Table). On PGIC, 99% of patients reported improvement at 6 months compared to baseline. None of the patients in the 10 kHz SCS group chose to crossover to CMM at 6 months. Treatment benefits were sustained through 12 months. A majority of patients treated with 10 kHz SCS were noted to have improvement on neurological examination by investigators. Participants in the CMM group experienced no change in pain levels, pain interference with daily activities, or sleep quality over 6 months. Eighty-two percent opted to crossover to 10 kHz SCS at 6 months. Results post-crossover were similar to those in patients originally randomized to 10 kHz SCS: mean 70% reduction in pain as well as reduced pain interference with daily activities, improvements in sleep, and a majority with observed improvements on neurological examination. In total, 154 patients received permanent SCS device implants, including those originally randomized to SCS and those who crossed over at the 6-month timepoint. There were 32 study-related adverse events in 26 (17%) patients. Five (3%) patients developed infections that required device removal.

Conclusion: This study is the largest RCT to-date of SCS management in PDN. Substantial improvements with high frequency (10 kHz) SCS were sustained over 12 months and support this treatment for PDN patients with symptoms refractory to conventional care.

figurev

Clinical Trial Registration Number: NCT03228420

Supported by: Nevro

Disclosure: E. Petersen: Grants; Nevro.

OP 11 Cardiac complications: mechanisms and possible treatments

61

Identifying myocardial insulin resistance by positron emission tomography combined with hyperinsulinaemic euglycaemic clamp: a new strategy for phenotyping type 2 diabetes

R. Simo1, M. Velasquez1, O. Simó-Servat1, A. Rojano1, B. Paun2, R. Marés2, C. Hernández1, S. Aguadé2, J.R. Herance3;

1Diabetes and Metabolism Research Unit. CIBERDEM, Vall d'Hebron Research Institute, Barcelona, 2Medical Molecular Imaging Research Group, Vall d'Hebron Research Institute, Barcelona, 3Medical Molecular Imaging Research Group. CIBBIM-Nanomedicine. CIBERBBN, Vall d'Hebron Research Institute, Barcelona, Spain.

Background and aims: The identification and clinical meaning of myocardial insulin resistance (mIR) in type 2 diabetic population are challenges to be met. In order to shed light to this issue we have conducted a pilot study using a methodology which might open up a new avenue in the phenotyping of type 2 diabetic population.

Materials and methods: The study comprised forty-seven patients with type 2 diabetes without history of previous cardiovascular event. Patients were prospectively recruited from February 2018 to July 2019 at the Outpatient’s Diabetes Unit of tertiary university hospital. The study was conducted according to the tenets of the Helsinki Declaration. The Ethic Committee of our hospital approved all procedures. The assessment of mIR was performed by means of two [18F]FDG PET studies per patient within 2 days: one with at least 8 h of fasting conditions and after 24 h of draw out medication (baseline scan) and the other with the same conditions but after performing hyperinsulinemic euglycemic clamp (HEC scan). Myocardial [18F] FDG uptake was measured in SUVbw normalized by the max SUVbw voxel value. A cardiac CT scan was performed during the baseline PET to determine the Hounsfield units (HU) density of the myocardium and to calculate the coronary artery calcium score (CACs) in Agatstone units (AU). Statistics: Mann-Whitney U test and Spearman’s correlation analysis.

Results: All patients showed a low left ventricle myocardial [18F]FDG uptake at the baseline scan closed to the background-to-noise signal. However, after performing HEC, 17 (39.5%) patients exhibited insulin sensitive myocardium (mIS) showing a strikingly enhancement of myocardial [18F]FDG uptake (1.59 SUV [1.42:1.92] vs.7.67 SUV [7.10-9.49]), whereas 26 (60.4%) showed a marginal increase of [18F]FDG uptake (1.18 SUV [1.06:1.59] vs. 2.27 SUV [1.62:2.93]), thus revealing the presence of mIR. Myocardial radiodensity (mRD) was higher in the mIR group than in mIS group (38.95 HU [33.81-44.06] vs. 30.82 HU [21.48-38.02]; p=0.03). A statistically significant correlation was found between mRD and either SUVHEC (rho= -0.4649; p= 0.0024) or with ΔSUV (rho= -0.5238; p= 0.0005). Regarding the relationship between mIR with cardiovascular risk factors we found that patients with mIR did not presented a significantly higher levels of HbA1c, cholesterol or blood pressure but exhibited a significantly higher CACs than patients with mIS (CACs>400 AU: 52% vs. 29%; p=0.002). This result means that patients with mIR present a very high risk of cardiovascular events.

Conclusion: [18F]FDG PET combined with HEC is a reliable method that permits a clear identification of patients with mIR. This subset of patients with type 2 diabetes present structural changes in the myocardium and a high proportion of CACs>400 AU. Further prospective research is required, not only to confirm these findings but also to determine the global impact of mIR on myocardium remodelling, functionality and cardiovascular outcomes.

Clinical Trial Registration Number: NCT02248311

Supported by: Instituto de Salud Carlos III (PI16/02064,​ PI20/01588)

Disclosure: R. Simo: None.

62

Do cardiovascular risk prediction models developed in primary care patients with type 2 diabetes perform better than the general population models? PREDICT cohort study

R. Pylypchuk;

Epidemiology and biostatistics, University of Auckland, Auckland, New Zealand.

Background and aims: There is only limited evidence that cardiovascular risk scores developed specifically in patients with diabetes estimate CVD risk more accurately. Studies comparing the predictive ability of CVD risk scores developed in diabetic populations, with those developed in general populations, showed that their performance varied considerably. However, comparing models developed from different studies is problematic because of between-study variation in methods and differences in populations’ background risk. Large number of patients with diabetes in the PREDICT cohort provides an opportunity to compare models derived from the same patient population, using a standard set of performance measures, therefore resulting in a more valid comparison than has been possible previously.

Materials and methods: Using data from the PREDICT primary care cohort of 401,752 CVD-free patients aged 30 to 74 years, we developed general population and diabetes-specific prediction models for estimating the risk of first CVD event (coronary heart disease, ischaemic and haemorrhagic stroke, transient ischaemic attack, peripheral vascular disease, congestive heart failure). Identical definitions of the outcome and predictors, and identical approaches to model development were applied in both sets of models. Pre-specified clinically-relevant risk predictors were used to develop sex-specific models with Cox regression methods. For the PREDICT-1o models developed in the total cohort of patients with and without diabetes, these were age, ethnicity, deprivation, smoking, diabetes status, family history of premature CVD, history of atrial fibrillation, systolic blood pressure, TC:HDL, and treatment with antithrombotic, blood pressure lowering and lipid lowering medications. For the PREDICT T2D models, developed in patients with type-2 diabetes, several relevant predictors were added to the above: duration of diabetes, glomerular filtration rate, albuminuria, haemoglobin A1c, body mass index, and treatment with oral hypoglycaemic medications and insulin. We compared performance of the diabetes-specific models and the general population models, in T2D population, using a standard set of performance indicators.

Results: The diabetes-specific models were developed from 46,652 patients with type-2 diabetes. The median follow-up was 5.2 years (244,840 person-years), a total of 4,114 patents experienced an event (9% fatal). The general population models were developed from 401,752 patients, of whom 15,386 experienced an event (10% fatal) during 1,685,521 person-years follow-up. The T2D-specific models performed better than PREDICT-1o, in terms of discrimination, calibration, explained variation and net benefit. Discrimination was good, with C statistic of 0.7 in both men and women, and Royson’s D of 1.082 and 1.346, respectively. Calibration plots also indicated T2D models’ excellent calibration. The net benefit analysis showed that diabetes-specific models were superior to the general population models at all clinically meaningful thresholds of risk.

Conclusion: Comparative assessment of the general population and T2D-specific models demonstrated that addition of clinically relevant predictors improves models’ performance in patients with diabetes. Treatment with insulin, measures of renal function, and body composition contribute significantly to cardiovascular risk prediction in type-2 diabetes.

Supported by: Health Research Council of New Zealand

Disclosure: R. Pylypchuk: None.

63

Autonomic dysfunction is associated with the development of arterial stiffness: the Whitehall II cohort

J.F.R. Schaarup1, M.S. Christensen1, A. Hulman2, L. Bjerg2, C.S. Hansen3, D. Vistisen3, A.G. Tabák4,5, D.R. Witte2,1;

1Aarhus University, Aarhus, Denmark, 2Steno Diabetes Center Aarhus, Aarhus, Denmark, 3Steno Diabetes Center Copenhagen, Copenhagen, Denmark, 4University College London, London, UK, 5Semmelweis University, Budapest, Hungary.

Background and aims: The association between autonomic dysfunction and increased arterial stiffness is well documented among persons with diabetes. However, emerging evidence suggests that this association can be detected without the presence of diabetes. This study examined the association between level and change of autonomic nervous function and subsequent development of arterial stiffness in the general population.

Materials and methods: Data was obtained from 5,067 participants of the Whitehall II cohort in the period between 1997 and 2013. Autonomic nervous function was assessed by a heart rate variability index i.e. standard deviation of normal-to-normal heartbeat intervals (SDNN) measured three times between 1997-2009, while arterial stiffness was assessed by carotid-femoral pulse wave velocity (PWV) measured twice between 2007-2013 First, individual SDNN levels and their annual change were estimated. Then we modelled the development of PWV using the previous SDNN estimates as exposure using linear mixed effect models. We included age in the model and its interaction with SDNN. First, we adjusted the model for sex and ethnicity (model 1), and then for socioeconomic status, BMI, smoking status, alcohol use, physical activity, levels of cholesterol (total), triglycerides, HbA1c, systolic blood pressure, anti-hypertensive (e.g. beta-blockers) and glucose lowering medications (model 2). Similar analyses were performed in the subgroup without diabetes.

Results: The median (25th & 75th percentile) age at the first PWV measurement was 64.5 years (60.6; 70.0), while median PWV was 8.05 m/s (7.03; 9.45). SDNN level and annual change was 31.1 ms (26.1; 36.5) and -1.5% (-1.9; -1.2), respectively. A decrease in SDNN was associated with subsequent higher levels of PWV, but the effect of change in SDNN was less pronounced at higher age. A typical individual aged 65 years with a SDNN level of 30 ms and a 2% annual decrease in SDNN had 1.21 m/s (CI: 0.86; 1.56) higher PWV level and 0.09 m/s per year (CI: 0.05; 0.12) slower annual PWV increase, compared to one with the same age and SDNN level but with a 1% annual decrease in SDNN. The differences in the models were attenuated after further adjustments and exclusion of persons with diabetes.

Conclusion: Steeper decline of autonomic nervous function is associated with higher levels of arterial stiffness. However, the effect of a steeper decline in autonomic nervous function was less pronounced at higher age. Associations were diminished after exclusion of persons with diabetes.

Disclosure: J.F.R. Schaarup: None.

64

Metabolically healthy obese and cardiovascular events in a nationwide cohort study

G. Fauchier1, A. Bisson2, C. Semaan2, J. Herbert2, A. Bodin2, D. Angoulvant2, G. Lip3, P. Ducluzeau1, L. Fauchier2;

1Endocrinologie Diabetologie Nutrition, Centre Hospitalier Universitaire de Tours, Tours, France, 2Centre Hospitalier Universitaire de Tours, Tours, France, 3Liverpool Centre for Cardiovascular Science, Liverpool, UK.

Background and aims: Obesity is a risk factor for cardiovascular disease (CVD) and has been increasing globally over the past 40 years in many countries worldwide. Metabolic abnormalities such as hypertension, dyslipidemia and diabetes mellitus are commonly associated and may mediate some of the deleterious effects of obesity. A subset of obese individuals without obesity-related metabolic abnormalities may be classified as being "metabolically healthy obese" (MHO). We aimed to evaluate the associations among MHO individuals and different types of incident cardiovascular events in a contemporary population at a nationwide level.

Materials and methods: From the national hospitalization discharge database, all patients discharged from French hospitals in 2013 with at least 5 years or follow-up and without a history of major adverse cardiovascular event (myocardial infarction, heart failure [HF], ischemic stroke or cardiovascular death, MACE-HF) or underweight/ malnutrition were identified. They were categorized by phenotypes defined by obesity and 3 metabolic abnormalities (diabetes mellitus, hypertension, and hyperlipidemia). In total, 2,953,816 individuals were included in the analysis, among whom 272,838 (9.5%) were obese. We evaluated incidence rates and hazard ratios for MACE-HF, cardiovascular death, myocardial infarction, ischemic stroke, new-onset HF and new-onset atrial fibrillation (AF). Adjustments were made on age, sex and smoking status at baseline.

Results: During a mean follow-up of 4.9 years, obese individuals with no metabolic abnormalities had a higher risk of MACE-HF (multivariate-adjusted hazard ratio [HR] 1.22, 95% confidence interval [CI]: 1.19-1.24), new-onset HF (HR 1.34, 95%CI 1.31-1.37), and AF (HR 1.33, 95%CI 1.30-1.37) compared with non-obese individuals with 0 metabolic abnormalities. By contrast, risks were not higher for myocardial infarction (HR 0.92, 95%CI 0.87-0.98), ischemic stroke (HR 0.93, 95%CI 0.88-0.98) and cardiovascular death (HR 0.99, 95%CI 0.93-1.04). In the models fully adjusted on all baseline characteristics, obesity was independently associated with a higher risk of MACE-HF events (HR 1.13, 95%CI 1.12-1.14), of new-onset HF (HR 1.19, 95%CI 1.18-1.20) and new-onset AF (HR 1.29, 95%CI 1.28-1.31). This was not the case for the association of obesity with cardiovascular death (HR 0.96, 95%CI 0.94-0.98), myocardial infarction (HR 0.93, 95%CI 0.91-0.95) and ischemic stroke (HR 0.93, 95%CI 0.91-0.96). Hypertension and diabetes mellitus were independent predictors of all types of outcomes.

Conclusion: Metabolically healthy obese individuals do not have a higher risk of myocardial infarction, ischemic stroke or cardiovascular death than metabolically healthy non-obese individuals. By contrast they have a higher risk of new-onset HF and new onset AF. Even individuals who are non-obese can have metabolic abnormalities (including diabetes) and be at high risk of cardiovascular disease events. Our observations suggest that specific studies investigating different aggressive preventive measures in specific subgroups of patients are warranted.

Disclosure: G. Fauchier: None.

65

Distribution of cardiovascular risk in type 2 diabetes: results of an analysis using data from CAPTURE study

J. Westerink1, H. Bleken Oestergaard1, E. Margo Hengeveld2, J. Broe Honore2, V. Humphreys3, F. Mach4, G. Yadav5, O. Mosenzon6;

1Department of Vascular Medicine, University Medical Center Utrecht, Utrecht, Netherlands, 2Novo Nordisk A/S, Søborg, Denmark, 3Diabetes Ireland Advocacy Group, Dublin, Ireland, 4Cardiology Division, Faculty of Medicine, University of Geneva, Geneva, Switzerland, 5Novo Nordisk Global Business Services, Bengaluru, India, 6Department of Endocrinology and Metabolism, Hadassah Medical Center, Hebrew University of Jerusalem, Jerusalem, Israel.

Background and aims: Cardiovascular disease (CVD) is the leading cause of mortality in type 2 diabetes (T2D). CAPTURE, a non-interventional, cross-sectional study across 13 countries in 2019, collected demographic and clinical data in almost 10,000 adults with T2D. Less than 25% of patients with established CVD treated with a glucose-lowering agent received an agent with demonstrated benefit in CV risk reduction. We used the Diabetes Lifetime-perspective prediction (DIAL) competing risk adjusted model to estimate CV risk distribution in the CAPTURE population, and to assess treatment patterns by CV risk. The model calculates absolute 10-year and lifetime risk of myocardial infarction, stroke or cardiovascular death, and life-expectancy free of a CVD event.

Materials and methods: Patient-level data from CAPTURE (age, sex, body mass index, smoking status, HbA1c, CVD history, T2D duration, clinical parameters and treatment history) were used in the DIAL model; missing data were imputed by region. High risk was defined as 10-year risk > 10%, and lifetime risk > 50%.

Results: Data from 9457 patients with T2D aged 30-85 years were included. There was a wide distribution of 10-year and lifetime risk, with higher risk in patients with a history of CVD (n = 2914) than in those without (n = 6543). Among patients with a history of CVD, 96% had a 10-year risk of CVD > 10% and 81% had a lifetime risk of CVD > 50% (Figure). In patients with CVD and a high 10-year risk of recurrent CVD, 81% had a lifetime risk of recurrent CVD > 50%. In patients with no history of CVD, 14% had a 10-year risk > 10% and only 1% had a lifetime risk > 50% (Figure). Among patients with no previous CVD but a high 10-year risk of CVD, only 4% had a lifetime risk > 50%. Of the patients with CVD, 10% received a glucagon-like peptide-1 receptor agonist (GLP-1 RA) and 18% received a sodium-glucose co-transporter-2 inhibitor (SGLT-2i); of patients with CVD and a high 10-year risk of recurrent CVD, 10% received a GLP-1 RA and 17% received an SGLT-2i. Among patients with no CVD, 11% received a GLP-1 RA and 16% received an SGLT-2i; of patients without current CVD but a high 10-year risk of CVD, 12% received a GLP-1 RA and 16% received an SGLT-2i.

Conclusion: There is a wide distribution of CVD risk in the CAPTURE population, and only a minority of patients at high risk of CVD received a glucose-lowering agent with demonstrated benefit in CV risk reduction. Discussing these risks and the CV benefit to be gained from interventions with patients can enhance shared decision-making.

figurew

Supported by: Novo Nordisk A/S

Disclosure: J. Westerink: None.

66

Association of circulating metabolomic biomarkers with incident cardiovascular disease in type 2 diabetes: analysis from the Hong Kong Diabetes Biobank

Q. Jin1, A. Luk1, C. Lim1, E. Lau1, R. Ozaki1, H. Lee1, W. So1, Y. Huang2, J. Chan1, Hong Kong Diabetes Biobank Study Group, R. Ma1;

1Department of Medicine and Therapeutics, The Chinese University of Hong Kong, Hong Kong, 2Li Ka Shing Institute of Health Sciences, The Chinese University of Hong Kong, Hong Kong, China.

Background and aims: Metabolomics are emerging as promising biomarkers for cardiovascular disease (CVD). The current study aimed to investigate the metabolomic signature of CVD in participants with type 2 diabetes (T2D).

Materials and methods: Baseline serum samples from 1,991 participants with T2D were quantified for 250 metabolic measures using proton nuclear magnetic resonance spectroscopy. After excluding 544 subjects with prevalent CVD, 1,447 were included in the analyses. Cox regression was used to estimate hazard ratios (HRs) for a 1-SD increment in each measure, adjusting for established risk factors, and use of lipid-lowering drugs. False discovery rate (FDR) <0.05 was set as the significance threshold.

Results: Mean (SD) age was 59.3 (10.9) years, 815 (56.3%) were male, and mean (SD) diabetes duration was 10.7 (8.4) years. During a median (IQR) 5.2 (5.0-5.4) years of follow-up, 125 (8.6%) events were recorded. Cholesterol in triglyceride-rich, non-high-density, and very low-density lipoproteins (VLDL) were positively associated with CVD (HR: 1.28 to 1.33). Cholesterol in small high-density lipoprotein (HDL) was inversely associated with CVD (HR 0.78), while cholesterol in other HDL particles was not. Triglycerides in all lipoproteins, apolipoprotein B (apoB) and its ratio to apolipoprotein A1 were positively associated with CVD (HR: 1.27 to 1.35). Phenylalanine, acetate, and glycoprotein acetyls were positively associated with CVD (HR: 1.29 to 1.39); while leucine and tyrosine were inversely associated with CVD (HR: 0.78 and 0.77, respectively). A total of 66 metabolites were associated with CVD, including 9 which remained significant after Bonferroni adjustment. A metabolite score composed of concentration of very small VLDL, acetate, and albumin (selected based on backward Akaike information criterion [AIC] elimination) significantly improved the C-statistic (95% CI) (0.754 to 0.790 [0.013, 0.066]), integrated discrimination improvement (0.040 [0.015, 0.076]), and continuous net reclassification improvement (0.386 [0.124, 0.634]) over the RECODe (Risk Equations for Complications Of type 2 Diabetes) model.

Conclusion: Cholesterol and triglycerides in non-HDL and apoB were positively associated with CVD. Cholesterol in small HDL was inversely associated, whereas triglycerides in HDL were positively associated with CVD. Amino acids, ketone bodies, and glycoprotein acetyls were associated with CVD. Incorporating metabolomics could improve the prediction of CVD in T2D beyond an established prediction model.

figurex

Supported by: Research Grants Council Research Impact Fund (R4012-18)

Disclosure: Q. Jin: None.

OP 12 The long and winding road to prevention and treatment of diabetes

67

Predicting sensitivity and resilience to modifiable risk factors for cardiometabolic disease

H. Pomares-Millan1, N. Atabaki-Pasdar1, A. Poveda1, I. Johansson2, P.W. Franks1,3;

1Lund University, Malmo, Sweden, 2Department of Public Health and Clinical Medicine, Umeå University, Umeå, Sweden, 3Nutrition, Harvard University, Boston, USA.

Background and aims: Lifestyle exposures play a major role in the development of disease, yet people vary in their susceptibility. Statistical prediction methods typically focus on population averages, with heterogeneity in effects treated as error. However, effects observed at the extremes of a population’s distribution can be informative of underlying features that can help improve statistical inferences. Thus, we used a novel approach to identify population subgroups that are either especially sensitive or resilient to a range of lifestyle risk exposures and assessed the extent to which allocation to these groups predicted incident type 2 diabetes (T2D) or cardiovascular disease (CVD).

Materials and methods: From participants without T2D or CVD at baseline, in the Västerbotten Intervention Program (VIP, n= 35,440), blood samples were obtained after fasting and a 75-gram oral glucose load. Environmental exposure data (e.g. socio-economy, quality of life, tobacco use, alcohol consumption, diet, physical activity) were used to develop scores that predict levels of 9 intermediate cardiometabolic traits: obesity (BMI), dyslipidemia (triglycerides, LDL, HDL-C, total cholesterol), dysglycaemia (fasting and 2-hr glucose), and high blood pressure (SBP, DBP). 90% prediction intervals were determined using quantile regression forests, from which three exposure susceptibility groups were derived per cardiometabolic score (‘sensitive’ [>90% PI], ‘resilient’ [<10% PI], and ‘normal’ [11-89% PI]). Half the dataset was used for training and the remainder for validation. Incidence rates and Cox proportional hazards models were used to determine risk of developing T2D and CVD in the sensitive and resilient groups separately, using the normal group as reference. All analyses were undertaken in R software (v3.6.1).

Results: During a median of 9.7 y follow-up in the VIP cohort, the rate of developing T2D was ~4x higher in the sensitive compared with normal group for obesity risk factors (IRR=4.0; 95%CI 2.2, 7.2; p=0.02). For high blood pressure risk factors, the risk of CVD was >2x higher in the compared with the normal group (SBP: HR=2.4; 95%CI 1.12, 5.14; p= 0.02; DBP: HR=2.25; 95%CI 1.49, 3.4; p= 1.05E-04); conversely, the risk of CVD was more than halved in the resilient compared with normal group (HR=0.39; 95%CI 0.17, 0.87; p= 2.1E-02). For lipid risk factors, the resilient group had higher hazards for T2D, (HR=2.4; 95%CI 1.17, 4.99; p= 0.02) compared to the reference group. Similar results were seen in UKB (see Figure).

Conclusion: Individuals who are predicted to be sensitive to lifestyle risk exposures are generally at higher risk of developing T2D or CVD. This population subgroup may benefit from intensive preventive interventions focused on reducing the burden of adverse lifestyle exposures.

figurey

Supported by: Swedish Research Council, Swedish Heart-Lung Foundation, NASCENT

Disclosure: H. Pomares-Millan: None.

68

Epidemiology of hypoglycaemic episodes leading to hospitalisations in Denmark over the last two decades

M.H. Jensen1,2, O. Hejlesen3, P. Vestergaard4;

1Aalborg University Hospital, Aalborg, 2Department of Health Science and Technology, Aalborg University, Aalborg, 3Department of Health Science and Technology, Aalborg University, Aalborg University, Aalborg, 4Steno Diabetes Center North Denmark, Aalborg University Hospital, Aalborg, Denmark.

Background and aims: Severe hypoglycaemia is a very unpleasant condition for diabetes patients and has been associated with increased mortality. Especially, episodes leading to hospitalisations, places a significant burden on patients and society. The aim of this study was to investigate the nationwide trends in incidence and associated risk factors with focus on blood glucose lowering medication for hypoglycaemic episodes leading to hospitalisations in Denmark among people with type 1 and 2 diabetes mellitus.

Materials and methods: A cohort study of all people with diabetes from 1977-2018 experiencing hypoglycaemic episodes leading to hospitalisation in 1998-2018 was established. Data were extracted from the Danish National Patient Registry. Trends in incidence rates were investigated descriptively and with linear regressions, and risk factors were investigated with Cox Proportional Hazards models using time-varying covariates.

Results: 66,453 hypoglycaemic episodes leading to hospitalisation in 1998-2018 were investigated among 644,009 people with type 1 (age 37 [SD:22] years) and type 2 diabetes (age 61 [SD:17] years). The incidence rate for hypoglycaemic episodes has declined since 2003 with 0.4 and 0.05 episodes per 100 person years per calendar year (p<0.0001) for type 1 and type 2 diabetes, respectively. Insulin glargine (HR: 1.20, 95% CI: 1.05-1.36, p=0.0059), insulin detemir (HR: 1.17, 95% CI: 1.04-1.32, p=0.0078) and insulin degludec (HR: 1.04, 95% CI: 0.81-1.33, p=0.7716) compared to human insulin (intermediate-acting HR: 1.38, 95% CI: 1.25-1.52, p<0.0001, combination: HR: 1.84, 95% CI: 1.65-2.05, p<0.0001) and especially SGLT2i (HR: 0.43, 95% CI: 0.33-0.57, p<0.0001), GLP-1-RA (HR: 0.50, 95% CI: 0.44-0.57, p<0.0001) and DPP-4i (HR: 0.44, 95% CI: 0.38-0.49, p<0.0001) compared to sulfonylureas (HR: 2.27, 95% CI: 2.18-2.37, p<0.0001), see figure 1 below, seemed safer with respect to hypoglycaemic episodes.

Conclusion: Incidence rates of hypoglycaemic episodes leading to hospitalisation are declining in Denmark, and the advent of new treatment alternatives may play a significant role in this decline. From a safety perspective, these findings are important and could be considered by clinicians when assessing treatment options for patients.

figurez

Disclosure: M.H. Jensen: None.

69

The association of hypoglycaemia exposure with subsequent adverse events and severe hypoglycaemia: preliminary results from Hypo-RESOLVE

J.E. O'Reilly, A. Jeyam, T.M. Caparrotta, P. McKeigue, H. Colhoun, the Hypo-RESOLVE consortium;

University of Edinburgh, Edinburgh, UK.

Background and aims: Accurate characterisation of the association of hypoglycaemia with complications of diabetes is required for the prevention of their eventual occurrence. The Hypo-RESOLVE (Hypoglycaemia REdefining SOLutions for better liVEs) project brings together partners from different sectors to alleviate the burden that hypoglycaemia places on people with diabetes. A key aim for this project is an improved understanding of the clinical consequences of hypoglycaemia through the analysis of data pooled from dozens of clinical trials. We leveraged the quantity and quality of data collected during these clinical trials to estimate the association of hypoglycaemia with clinical outcomes.

Materials and methods: We used a mixed GLMs to estimate the association between hypoglycaemic events (HE) and subsequent severe hypoglycaemia (SH). We used proportional hazard models with time-updated HE exposure to estimate the association of several adverse events with HE exposure.Three definitions of HE severity were used: level 1 (events with an associated recorded blood glucose level < 3.9 mmol/L and ≥3.0 mmol/L), level 2 (< 3.0 mmol/L) and severe (need for third-party assistance). The adverse events of interest were CVD, subsequent SH, nephropathy, and neuropathy. Separate analyses were performed by type of diabetes and for each distinct definition of HE severity. All models were adjusted for age, sex and diabetes duration.

Results: Our study population consisted of 2,116 adults with available age, sex, diabetes type, diabetes duration, adverse event, and HE data. Among those with type 1 diabetes, the number of HE in the previous 45 days was associated with an increased probability of subsequent SH. The size of this association increased with the severity of HE exposure: odds-ratio = 1.02 [95% CI = 1.01, 1.04] for level 1 HE, 1.06 [1.03, 1.09] for level 2 HE, and 1.58 [1.25, 2.00] for SH (all p < 0.01). Similar patterns were observed for those with type 2 diabetes: OR=1.09 [1.05, 1.15], 1.20 [1.12, 1.29] and 4.30 [2.18, 8.49], respectively (p < 0.01 for all). An increase in cumulative SH events was associated with an increased hazard of CVD among those with type 1 diabetes (HR=1.28 [1.13, 1.44], p < 0.01) as well as an increased hazard of nephropathy (1.16 [1.06, 1.28], p = 0.01). We found no evidence for an association between HE and neuropathy in those with type 1 diabetes. An association between CVD and level 1 HE was observed in type 2 diabetes (1.01 [1.00, 1.02], p = 0.03). Estimates were 1.02 [0.99, 1.04] (p = 0.17) for level 2 HE, 1.37 [0.99, 1.89] (p = 0.06) for SH. Analyses suggested a potential association between level 1 HE and neuropathy (1.02 [1.00, 1.04], p = 0.06). There was no evidence of association between HE and nephropathy, although these outcomes were relatively rare in the analysed data.

Conclusion: In these preliminary analyses, increased HE exposure in the previous 45 days was associated with increased odds of SH, regardless of HE severity. Our results support an association between SH and subsequent CVD and provide some support for associations with subsequent neuropathy and nephropathy. These preliminary findings suggest that the consequences of HE can be profound, supporting the key role that prevention of HE plays in diabetes care. As the Hypo-RESOLVE project continues to accrue data, the statistical power, accuracy and precision of estimates will improve, allowing greater insight into the consequences of HE.

Supported by: This study was performed and funded as part of hypo-RESOLVE

Disclosure: J.E. O'Reilly: None.

70

Heterogeniety in the effects of two anti-diabetic drugs evaluated using machine learning applied to registry data

S. Franzén1,2, A.-M. Svensson3,2, M. Miftaraj2, K. Eeg-Olofsson3, B. Eliasson3;

1Institute of Medicine, Sahlgrenska Academy, School of Public Health and Community Medicine, Gothenburg, 2Centre of Registries in Region Västra Götaland, Gothenburg, 3Institute of Medicine, Sahlgrenska Academy, Gothenburg, Sweden.

Background and aims: The vast majority of both randomized and epidemiological studies of the effect of anti-diabetic drugs are aimed at estimating the average treatment effect in some defined and often selected population. However, this approach will probably not be able to provide a well-informed basis for offering the right medicines to different patients, and thus calling for precision medicine. Precision medicine is commonly described as a means of exploiting genetic markers to predicate optimal effects, but such are rarely available in clinical practice. Therefore, we set out to assess the existence of, and if so, to characterize the heterogeneity of the treatment effect of two anti-diabetic drugs compared to SU, using state of the art machine learning combined with causal inference, utilizing data routinely available in a clinical setting.

Materials and methods: Based on data from the national diabetes register, the patient registry, the prescribed drug registry and the longitudinal integrated database for health insurance and labour market studies in Sweden we formed two cohorts. One cohort included 18587 persons who started treatment with liraglutide (n=2971) or sulphonyl urea (SU) (n=15616) added to metformin during 2010-2017, and one cohort included 10226 persons starting treatment with dapagliflozin (n=1051) or SU (n=9175) during 2013-2017. The average conditional treatment effect with respect to change in HbA1c, weight, systolic blood pressure (SBP) and kidney function after 6 months treatment was estimated using causal random forest based on 10.000 trees. The algorithm uses a propensity score adjusted doubly robust estimator in the evaluation of each split and mitigates overfitting by using half of the data to derive the split points and half of the data to estimate the treatment effects. The conditional average treatment effect was estimated conditional on age, gender, weight, smoking, HbA1c, SBP, DBP, LDL, HDL, triglyceride, eGFR, retinopathy, microalbuminuria, macroalbuminuria, physical activity, anti-hypertensive treatment, lipid-lowering treatment, education, income, marital status, geographic origin, and previous CHD, AMI, stroke, CVD, HF, AF, kidney failure, hyperglycaemia, psychiatric disease, alcohol or drug abuse, cancer or dementia.

Results: The average treatment effect on change in HbA1c was -2.57 [-3.20, -1.94] mmol/mol for liraglutide compared to SU, and -1.10 [-1.92, -0.27] mmol/mol for dapagliflozin vs SU. We found evidence of treatment effect heterogeneity in the effect of liraglutide on change in HbA1c (p<0.001) and SBP (p=0.01), and in the effect of dapagliflozin on change in HbA1c (p=0.003). Key drivers for the heterogeneity in the effect on HbA1c were baseline HbA1c, and to a lesser extent baseline weight, eGFR and age. There was also a marked difference in the effect of liraglutide with respect to change in HbA1c between the genders.

Conclusion: The effect of both liraglutide and dapagliflozin with respect to change in HbA1c after 6 months is heterogeneous and varies primarily by baseline HbA1c, and to a lesser extent also by baseline weight, eGFR and age. Heterogeneity of the effects of treatments should be examined in greater detail in order to better tailor diabetes care.

Disclosure: S. Franzén: None.

71

Alendronate use and risk of type 2 diabetes: a Danish population-based case-control study

R. Viggers, P. Vestergaard;

Endocrinology, Steno Diabetes Center North Jutland, Aalborg University Hospital, Aalborg, Denmark.

Background and aims: There has been proposed a link between glucose homeostasis and bone metabolism. Bisphosphonates are first line treatment of osteoporosis and we aimed to investigate if the risk of developing type 2 diabetes was altered by previous use of alendronate.

Materials and methods: We conducted a population-based case-control study through access to all discharge diagnoses (ICD-10 system) from the National Danish Patient Registry and all redeemed drug prescriptions (ATC classification system) from the Health Service Prescription Registry. All cases with a diagnosis of type 2 diabetes between 2008 and 2018 were matched on sex and age with 3 randomly selected controls by incidence-density sampling. Exposure was defined as ever use of alendronate and further grouped as effective and compliant use. ORs were calculated by conditional logistic regression analysis with adjustment for several confounders and test for trend for dose-response relationship.

Results: A total of 163,588 patients with type 2 diabetes and 490,764 matched control subjects were included with a mean age of 67 years and 55% male subjects. The crude OR of developing type 2 diabetes after alendronate use was 0.93 (95% CI 0.90-0.96) and decreased further after adjustment (multiple adjusted OR: 0.64 [95% CI 0.62-0.66]). The adjusted OR decreased to 0.47 (95% CI 0.40-0.56) among those with more than 8 years of alendronate use. A test for trend suggested a dose-response relationship between longer effective use of alendronate and lower risk of type 2 diabetes (p=0.002).

Conclusion: These results suggest a possible protective effect of alendronate in a dose-dependent manner against development of type 2 diabetes with a potential 50% risk reduction after 8 years of alendronate use. We propose future clinical research to investigate if alendronate impacts on glucose homeostasis, e.g. insulin sensitivity and glycemic control, and if it differs among people with and without pre- or type 2 diabetes.

Supported by: This work was supported by Steno Collaborative grant, Novo Nordisk Foundation, Denmark (Grant no. NNF18OC0052064).

Disclosure: R. Viggers: Grants; Novo Nordisk Foundation, Denmark (Grant no. NNF18OC0052064).

72

Sex-specific differences in patients deceased after bariatric surgery: a retrospective, registry analysis

H. Beiglböck1, E. Mörth2, B. Reichardt3, T. Stamm4, B. Itariu1, J. Harreiter1, M. Hufgard-Leitner1, P. Fellinger1, J. Eichelter5, G. Prager5, A. Kautzky6, A. Kautzky-Willer1, P. Wolf1, M. Krebs1;

1Medical University of Vienna, Division of Endocrinology and Metabolism, Vienna, Austria, 2University of Bergen, Department of Informatics, Bergen, Norway, 3Austrian Health Insurance Fund, Vienna, Austria, 4Medical University of Vienna, Center for Medical Statistics, Informatics and Intelligent Systems, Institute for Outcomes Research, Vienna, Austria, 5Medical University of Vienna, Division of General Surgery, Department of Bariatric Surgery, Vienna, Austria, 6Medical University of Vienna, Department of Psychiatry and Psychotherapy, Division of Social Psychiatry, Vienna, Austria.

Background and aims: Bariatric surgery is an effective treatment strategy for patients with obesity. The majority of patients undergoing bariatric surgery are female. Motivation and reasons for bariatric surgery are different in men and women. Thus, the aim of this study was to analyze sex-specific differences with emphasis on patients deceased with a history of bariatric surgery in a large, registry analysis.

Materials and methods: The Austrian health insurance provides service for about 99% of all Austrian inhabitants. Data from inpatient and outpatient services comprising reimbursed drug prescriptions based on Anatomical Therapeutic Chemical (ATC) codes, medical diagnoses based on International Classification of Diseases (ICD) and medical procedures as MEL (medical single procedure) were available. Overall, 19 901 patients with a history of bariatric surgery ((HF220 (Sleeve Gastrectomy - open), HF230 (Sleeve Gastrectomy - laparoscopic), HF240 (Gastric Bypass - open), HF250 (Gastric Bypass - laparoscopic), HF254 (Biliopancreatic Diversion - open), HF255 (Biliopancreatic diversion - laparoscopic), HF260 (Gastric banding - open) and HF270 (Gastric banding - laparoscopic)) from January 2010 to December 2018 with 107 806 patient years of observation were included. In deceased patients, comorbidities of patients were analyzed based on ICD-codes and ATC-codes. Comorbidities associated with obesity were categorized in 4 groups: Diabetes mellitus (DM), cardiovascular diseases (CV), psychiatric disorders (PSY) and malignancies (M).

Results: The mean age at operation of all was 40.6 ± 12.5 years (men: 41.8±12.6, women: 40.1±12.4; p=0.000). Within the observation period from January 2010 to April 2020, 367 (1.8%) patients deceased. The mean follow-up of the total cohort was 5.4 ± 2.6 years. The total rate of mortality per year of observation was 0.34%. The sex-specific rate of mortality was 2.7-fold higher in men compared to women (0.64% vs. 0.24%). The 30-days mortality was 0.2% and five-fold higher in men compared to women (0.5% vs. 0.1%, p < 0.001). In both men and women, cardiovascular comorbidities (48%, men: 53%, women: 44%, p = 0.09) and psychiatric disorders (47%, men: 44%, women: 50%, p = 0.212) were the most common comorbidities. Diabetes and malignant diseases were identified in 36% (men: 41%, women: 32%, p = 0.07) and 33% (men: 29%, women: 36%, p = 0.145), respectively.

Conclusion: This analysis demonstrates a five-fold higher 30-days mortality in men compared to women after bariatric surgery. The long-term mortality was also higher in men. Cardiovascular comorbidities and psychiatric disorders are independently of sex, frequently observed comorbidities in this special cohort.

Disclosure: H. Beiglböck: None.

OP 13 Diverse landscape of type 1 diabetes risk

73

Dietary factors and risk of islet autoimmunity and type 1 diabetes: a systematic review and meta-analysis

A.-M. Lampousi, S. Carlsson, J.E. Löfvenborg;

Institute of Environmental Medicine, Karolinska Institutet, Stockholm, Sweden.

Background and aims: Numerous dietary components have been linked to the development of islet autoimmunity (IA) and type 1 diabetes (T1D); however, no associations have been firmly established. This systematic review and meta-analysis aims to synthesize current knowledge on the association between diet and incidence of IA and T1D.

Materials and methods: The literature search was performed in Medline, Embase, and Cochrane Library, from inception until October 2020. Eligible studies had IA or T1D as outcome, any dietary exposure, case-control, cohort, or randomized controlled trial design, and hazard ratios, risk ratios, or odds ratios as measures of association. Summary relative risks (RR) and 95% confidence intervals (CI) of IA and T1D were estimated with random-effects models. Heterogeneity was quantified with the I2 statistic, risk of bias in individual studies with the ROBINS-I and RoB 2 tools and certainty of evidence with the GRADE tool.

Results: Among 5935 identified records, 152 were eligible and pooled estimates could be produced for 27 dietary components. The risk of T1D, but not IA, decreased with later introduction to cow’s milk (≥2-3 vs <2-3 months, RR: 0.69, CI: 0.59-0.81, I2= 0%, moderate certainty) and gluten (3-6 vs <3-5 months, RR: 0.46, CI: 0.25-0.84, I2=11%, high certainty), while it increased with later introduction to fruits (4-6 vs <4-5 months, RR: 2.14, CI: 1.16-3.94, I2=0%, moderate certainty). Higher consumption of cow’s milk products during childhood was associated with increased risk of both IA (per 2-3 portions/day, RR: 1.25, CI: 1.06-1.47, I2= 0%, moderate certainty) and T1D (≥2-3 vs <2-3 glasses of milk/day, RR: 1.78, CI: 1.36-2.33, I2= 0%, moderate certainty). Lower risk of T1D was observed in relation to longer total (≥6-12 vs <6-12 months, RR: 0.39, CI: 0.26-0.58, I2=43%, high certainty) and exclusive (≥2-3 vs <2-3 months, RR: 0.69, CI: 0.58-0.81, I2=0%, moderate certainty) breast-feeding. Age at introduction to infant formula, cereal, meat and vegetables and maternal intake of gluten, iron and vitamin D were not associated with the risk of T1D.

Conclusion: Early dietary factors, including cow´s milk, gluten, fruit and breast-feeding might play a role in the development of T1D. Further well-designed studies are needed to better understand these associations.

Supported by: Swedish Research Council and FORTE

Disclosure: A. Lampousi: None.

74

Alterations in biomarkers of carbohydrate and lipid metabolism up to 20 years before the diagnosis of type 1 diabetes: findings from the AMORIS cohort

K. Herzog1, V.E.R. Grill2,3, N. Hammar1, H. Malmström1,4, M. Talbäck1, G. Walldius1, S. Carlsson1;

1Institute of Environmental Medicine, Karolinska Institutet, Stockholm, Sweden, 2Norwegian University of Science and Technology, Trondheim, Norway, 3Department of Endocrinology, Trondheim University Hospital, Trondheim, Norway, 4R&d, Swedish Orphan Biovitrum AB, Stockholm, Sweden.

Background and aims: Type 1 diabetes (T1D) is described to have an acute onset, but autoantibodies can appear several years prior to diagnosis, suggesting a longer preclinical phase. We assessed whether elevations in metabolic and inflammatory biomarkers were associated with future T1D risk and the temporal relationships.

Materials and methods: We studied 591,239 individuals free of diabetes from the Swedish AMORIS cohort followed from 1985-1996 (baseline period) to 2012. Through linkage to national patient, diabetes, and prescription registers, we identified incident T1D cases based on ICD-code and insulin or age at first recording ≤30 years. We used Cox regression models to estimate hazard ratios for biomarkers at baseline and incident T1D. We additionally assessed trajectories of biomarkers during the 25 years before T1D diagnosis in a nested case-control design.

Results: We identified 1,122 T1D cases during follow-up. The biomarkers glucose, fructosamine, triglycerides, the apolipoprotein B/A-I ratio, uric acid, and alkaline phosphatase were positively associated with T1D risk. A higher apolipoprotein A-I was associated with a lower T1D incidence. These associations were independent of age and sex of the individuals. We obtained similar results for glucose and fructosamine in individuals with age-of-onset ≤ 30 years. Already 15 years before diagnosis, T1D cases had higher mean glucose, fructosamine, triglycerides, and uric acid levels compared to controls.

Conclusion: Alterations in carbohydrate and lipid metabolism are associated with T1D risk, and biomarkers may be elevated several years preceding diagnosis. These data suggest that disease processes leading to T1D may occur decades prior to disease onset.

Supported by: Jungner Foundation for Laboratory Medicine, Swedish Research Council, FORTE, Novo Nordisk Foundation

Disclosure: K. Herzog: None.

75

Decreasing age-at-onset of type 1 diabetes in a unique multigenerational cohort

P. Leete1, R.J. Aitken2, I.V. Wilson2, A.E. Long2, K.M. Gillespie2;

1Institute of Biomedical and Clinical Science, University of Exeter, Exeter, 2Diabetes and Metabolism, University of Bristol, Bristol, UK.

Background and aims: Type 1 diabetes (T1D) is an autoimmune disease in which both genetic and environmental components play important roles. Heterogeneity is increasingly recognised as a key feature of the condition, and the importance of age-at-onset of T1D is well documented. Diagnosis in early childhood is associated with a distinct immune infiltration of islets and extensive beta-cell loss, a higher genetic risk score and decreased residual serum C peptide, all linked to poorer outcomes in later life. Understanding the drivers of early-onset disease is therefore of critical importance. In recent decades, the increasing incidence of T1D at a population level has been associated with more children developing the condition in early childhood but few family data have been reported. This study aimed to examine age-at-onset in a UK population-based cohort of families of children with type 1 diabetes recruited since 1985.

Materials and methods: The Bart’s Oxford (BOX) family study has been registering cases of T1D diagnosed under the age of 21 years and their relatives since 1985. Families with a grandparent or parent and child diagnosed with T1D were selected from a total of 534 families with 2 or more effected family members. Families where both parents were affected were excluded. In total, 264 offspring/parent pairs and 44 grandchild/grandparent pairs were identified. In a cohort of offspring (incl. probands, siblings and proband children) aged between 0.9 to 28.1 years (y), differences in generational ages at onset of diabetes were examined.

Results: Of 264 parent/offspring pairs, 223(84.4%) of offspring had parents who were diagnosed older than them, with a mean difference in ages of -13.5y(SEM+/-0.8; paired T-test (p)<0.0001, Table 1). This trend was maintained irrespective of the sex of the affected parent [mother (-15.17(1.5)y; p<0.0001) or a father (-12.17(0.96)y; p<0.0001)]. This trend was even more striking in grandparent/grandchild pairs, where the mean of differences was -28.37(2.3)y (p<0.0001). More stringent analyses (Table 1) revealed that the only group of parents not significantly older than their children were those diagnosed under the age of 7y.

Conclusion: The vast majority of T1D is idiopathic, with only 3-5% of subjects having an affected first degree relative. In this observational study of generational T1D, we show that age-at-diagnosis is decreased significantly across successive generations. As islet autoimmunity is increased in children diagnosed at the earliest ages, these data suggest that disease severity in subsequent generations of the same families will be exacerbated. The mechanisms underlying this decrease in age-at-onset between generations are unknown, but it is unlikely to be driven solely by genetic causes, suggesting increasing environmental effects. Further epigenetic and genetic studies may reveal hitherto unidentified age-associated interactions between genes and environmental triggers.

figureaa

Supported by: UKRI E3-EXCEED DUK

Disclosure: P. Leete: None.

76

Follow up of a French cohort of children with a family history of type 1 diabetes

A. Vambergue1, T. Lemoine2, M. Kyheng3, I. Fajardy2, C. Leroy4, P. Pigny5, P. Fontaine1;

1University of Lille, Lille, 2CHRU, Lille, 3Department of Biostatistics, University of Lille ULR 2694 METRICS, Lille, 4Department of Pediatrics, Lille, 5University of Lille INSERM UMR-S1172, Lille, France.

Background and aims: The underlying pathological processes of type 1 diabetes (T1D) commence several years before the appearance of clinical symptoms. The annual incidence of T1D is increasing, particularly in children younger than 15 years. Though T1D is sporadic in 90% of cases, individuals with a family history of T1D have a 20 times greater risk of developing diabetes than the general population and may thus benefit from screening.

Materials and methods: Children of parents with T1D were screened annually at the T1D screening centre in Lille (France) from 1996. Genetic markers (HLA class II) were studied and annual tests for autoantibodies (GAD, AAI, ICA, IA-2) were performed. The principal objective was to identify the proportion of children who developed autoantibodies and those who developed T1D. The secondary objective was to characterise the serological profile of those children who developed autoantibodies and evaluate their risk of developing T1D.

Results: Of the 587 children studied, 16.4% showed at least one persistant autoantibody and 2.7% subsequently developed diabetes. Children with an HLA DR3/4 genotype showed an increased likelihood of developing an autoantibody during the study period. Children who developed their first autoantibody at younger than 8 years of age more frequently showed multiple autoantibodies during the course of the study (61.5% vs 28.1%; p=0.001). 32.3% of children who seroconverted were no longer autoantibody positive at their last visit, and none of these children subsequently developed diabetes. The risk of presenting with diabetes within 5 years increased with the number of autoantibodies detected during the study (4.2%, 5.3%, 59.6% and 45.1% respectively for 1,2,3 and 4 antibodies). The presence of multiple autoantibodies at the time of seroconversion was associated with an approximately 14.5 times greater risk of developing type 1 diabetes (HR 14.46 [4.97-42.07].

Conclusion: Our study evaluated the risk of developing type 1 diabetes in a French cohort of children of parents with type 1 diabetes. Detection of multiple autoantibodies was associated with an increased risk of developing T1D. The presence of a single autoantibody and/or transient seroconversion was associated with a lower risk. Our study, on a French cohort, currently represents the only data of this type, and may be useful in establishing new diabetes prevention protocols.

Disclosure: A. Vambergue: None.

77

Large socioeconomic differences in life expectancy and years spent with complications of diabetes: a cohort study in the Scottish population with type 1 diabetes, 2013-2018

A. Höhn1, S.J. McGurnaghan1, T.M. Caparrotta1, A. Jeyam1, J.E. O'Reilly1, L.A.K. Blackbourn1, S. Hatam1, C. Dudel2, P.M. McKeigue3, H.M. Colhoun, on behalf of SDRN-Epi1;

1Institute of Genetics and Molecular Medicine, Diabetes Medical Informatics & Epidemiology Group, Edinburgh, UK, 2Max Planck Institute for Demographic Research, Rostock, Germany, 3Usher Institute of Population Health Sciences and Informatics, Edinburgh, UK.

Background and aims: We aimed to examine current period life expectancy and life years spent with complications of diabetes among a national cohort of individuals with type 1 diabetes and to quantify socioeconomic differences in these measures.

Materials and methods: We studied 22370 individuals aged 30 and older with type 1 diabetes alive at any point in time between 1/1/2013 and 31/12/2018 in the national diabetes register. We used the clinical records in the register to capture prevalent and incident retinopathy/maculopathy, cardiovascular disease, nephropathy, and neuropathy, and also linked to mortality records. We captured socioeconomic status using the Scottish Index of Multiple Deprivation (SIMD) 2016. For each individual, we constructed a history of transitions to one or more of the above complications or death. We used parametric multistate survival models to estimate remaining life expectancy at an attained age of 30 and the expected number of years spent in particular health states.

Results: At an attained age of 30, remaining life expectancy was 42.18 years ((95% Confidence Intervals) 41.78-42.59) among females and 38.94 years (38.17-39.70) among males with type 1 diabetes, compared with 51.72 years and 48.00 years among the general population. Remaining life expectancy at an attained age of 30 was around 10 years lower among the most deprived quintile when compared with the least deprived quintile: 37.35 years (36.01-38.68) vs. 47.64 years (46.08-49.20) among females and 34.34 years (33.18-35.50) vs. 44.35 years (42.86-45.84) among males. Differences in the expected average number of years without complications were around 4 years lower among the most deprived quintile when compared with the least deprived quintile: 7.10 years (6.00-8.19) vs. 10.66 years (9.14-12.18) among females and 7.30 years (6.37-8.23) vs. 10.82 years (9.49-12.16) among males.

Conclusion: Differences in life expectancy between the population with type 1 diabetes and the general population have decreased in Scotland. However, our results indicate large socioeconomic differences in life expectancy and years of life spent without complications of diabetes. New initiatives to reduce these gaps are required.

figureab

Supported by: Diabetes UK (17/0005627)

Disclosure: A. Höhn: None.

78

Risk of incident obstructive sleep apnoea in patients with type 1 diabetes: a population-based matched controlled cohort study

A.A. Tahrani1,2, Z. Alshehri3,4, A. Subramanian5, N.J. Adderley5, K.M. Gokhale5, M. Karamat1, C.J. Ray3, P. Kumar3, K. Nirantharakumar5;

1Department of Diabetes and Endocrinology, University Hospitals Birmingham NHS Foundation Trust, Birmingham, UK, 2Institute of Metabolism and Systems, University of Birmingham, Birmingham, UK, 3Institute of Clinical Sciences, University of Birmingham, Birmingham, UK, 4Respiratory therapy department, Taibah University, Medina, Saudi Arabia, 5Institute of Applied Health Research, University of Birmingham, Birmingham, UK.

Background and aims: Type 2 diabetes (T2D) is associated with increased risk of incident obstructive sleep apnoea (OSA). In T2D, OSA is associated with a higher risk of diabetes-related vascular complications, poor quality of life, and mortality. Several cross-sectional studies showed high prevalence of OSA in type 1 diabetes (T1D). However, it is unknown whether patients with T1D are at increased risk of incident OSA. Therefore, we aimed to assess the incidence of developing OSA in patients with T1D compared to an appropriately matched control population, and identify the predictors of incident OSA in T1D. This is important in T1D considering the adverse outcomes associated with OSA and the impact on driving.

Materials and methods: We utilised the Health Improvement Network (THIN), which is a large UK primary care database, to conduct a retrospective cohort study. Each patient with T1D (exposed) was matched for age, sex, body mass index (BMI), and general practice to up to four persons without diabetes (control). Patients with OSA at baseline or a diagnosis of T2D at any time were excluded. The study period was 1st January 1990 to 31st August 2019. Cox regression was used to calculate hazard ratios using Stata IC version 15.

Results: 163,647 patients (34,147 exposed and 129,500 matched controls) were included. For the whole cohort, mean (SD) age was 38 (18) years, 93,082 (56.9%) were men, and BMI was 25.8 (4.3) kg/m2. The median (IQR) follow-up was 5.43 (2.19-10.11) years. Compared to patients without T1D, the adjusted hazard ratio (aHR) of incident OSA in patients with T1D was 1.67 (95% CI 1.42-1.96; P < 0.001) after adjusting for age, sex, BMI categories, Townsend quintiles (social deprivation), smoking status, ethnicity, alcohol, cardiovascular disease, hypertension, and atrial fibrillation. In T1D, predictors of incident OSA were people in their forties, male sex, obesity, atrial fibrillation, depression, and being prescribed antihypertensive or lipid-lowering drugs (table 1).

Conclusion: Patients with T1D are at increased risk of incident OSA compared to the general population. Clinicians need to consider OSA in patients with T1D despite having lower prevalence of obesity than T2D, especially in those who have one or more of the incident OSA predictors identified in this study.

figureac

Disclosure: A.A. Tahrani: None.

OP 14 "Humanomics" in diabetes

79

Fasting lipidomic analysis: a tool to unveil type 2 diabetes heterogeneity

A.F. Pina1, M.J. Meneses1,2, R. Patarrão1,2, R.T. Ribeiro2,3, J.F. Raposo2,4, L. Gardete-Correia2, R. Duarte5, J.M. Boavida2, J.L. Medina4, F. Carli6, B. Patrício6, A. Gastaldelli6, M.P. Macedo1,2;

1MEDIR, CEDOC, Lisbon, Portugal, 2APDP - Associação Protectora dos Diabéticos de Portugal, Lisbon, Portugal, 3Ciências Médicas, Universidade de Aveiro, Aveiro, Portugal, 4SPD - Sociedade Portuguesa de Diabetologia, Lisbon, Portugal, 5APDP, Lisbon, Portugal, 6Institute of Clinical Physiology - CNR, Pisa, Italy.

Background and aims: Type 2 diabetes (T2D) is a heterogeneous condition whose greatest impact comes from its complications. However, just the gradient of glycemia does not explain them. Lipid and glucose metabolism are closely linked. Thus, plasma lipidomics will be an important tool to unveil diabetes phenotypes. We hypothesize that groups stratified by insulin resistance (HOMA-IR), secretion rate (ISR) and clearance (IC) are associated with distinct glucose and lipid profiles. Considering both glycemia and lipidomics will help understand T2D’s heterogeneity and identify risk biomarkers associated with distinct mechanisms and new therapeutic targets, towards precision medicine.

Materials and methods: PREVADIAB2 includes 974 subjects profiled for glucose, insulin and C-peptide (OGTT) and surrogate indexes of dysmetabolism including a fatty liver score (NAFLD-FLS). For a sample of 145 women (68 NGT/NFG, 63 IGT and/or IFG, and 14 T2D), mean age 62y, lipidomic profile was performed by LC/MS on fasting serum samples with the quantification of triacylglycerols (TAG), ceramides (CER), sphingomyelin (SM) and phospholipids (PC, LPC). These subjects were then stratified using a hierarchical clustering algorithm informed by HOMA-IR, fasting ISR (fISR) and IC (fIC). The resulting clusters (Cl) were profiled for several parameters and it was assessed whether there was a lipidomic pattern in tandem with the one from other dysmetabolic parameters. Statistical significance of mean differences was assessed with Bonferroni correction.

Results: The identified groups and their lipidomic profile are shown in Fig. Cl1 had the lowest HOMA-IR and fISR and highest fIC (p<0.001). Also, had low glycemia (OGTT), the lowest NAFLD-FLS (p<0.01), TAG (p<0.01) and low CER. Consistently, it had high LPC18:2, that has been inversely correlated with IR. TAG48:0, a marker of de novo lipogenesis, was increased in Cl2, 3 and 4 (p<=0.02). Cl2 and 3 had lower fIC and higher NAFLD-FLS, HOMAIR and fISR than Cl1 (p<0.01), however fISR was lower on Cl3 (p<0.01). Additionally, Cl3 had higher glycemia at 30’ and 120’(OGTT) than Cl1 and 2 (p<0.01). Cl2 and 3 had higher TAG than Cl1 (p<0.01). Interestingly, Cl2 also had higher CER than Cl1(p<0.01) whereas Cl3 had lower SM (p=0.02). Cl4 had a high proportion of dysglycemia, similar to Cl3. However, it had the highest HOMAIR, fISR and NAFLD-FLS (p<0.01), whereas fIC was similar to Cl3 (p=1). Cl4 showed high levels of TAG and low levels of most SM species, as Cl3. Surprisingly, Cl4 CER levels were lower than Cl2 (p=0.01).

Conclusion: Clustering individuals by distinct metabolic defects identifies different degrees of glucose tolerance and lipidomic patterns.

figuread

Supported by: PDTC, FoieGras

Disclosure: A.F. Pina: None.

80

Metabolite differences between type 2 diabetes and type 3c diabetes secondary to chronic pancreatitis based on an untargeted metabolomics approach

L. Qi1, L. Li1,2;

1Department of Endocrinology, Zhongda Hospital, School of Medicine, Southeast University, Nanjing, Jiangsu, 2Pancreatic Research Institute, Southeast University, Nanjing, Jiangsu, China.

Background and aims: Type 3c diabetes mellitus (T3cDM) referred to diabetes secondary to exocrine pancreatic diseases, and chronic pancreatitis (CP) had proven to be one of the leading causes. In this study, a nontargeted metabolomics approach was established to characterize serum metabolic profile in T3cDM and compare with type 2 diabetes mellitus (T2DM).

Materials and methods: There were 16 patients with T3cDM, twelve gender and age-matched T2DM and 12 healthy controls recruited for metabolite analysis basing on liquid chromatography-mass spectrometry (LC-MS). Principal component analysis (PCA), partial least squares method-discriminant analysis (PLS-DA), differential metabolite volcanic plot, cluster heatmap, and KEGG metabolic pathway enrichment analysis were used to analyze the specific and differential metabolites.

Results: Fifty-one metabolites including lipids, carnitine, bile acid and hippuric acid were found to be different between T2DM and T3cDM, mainly enriched in fatty acids biosynthesis, primary bile acid biosynthesis, cholesterol metabolism, bile secretion and phenylalanine metabolism.

Conclusion: By LC-MS approach, this study identified the potential differential metabolites between T2DM and T3cDM. T3cDM is characterized by increased carnitine, bile acid and most of lipids, providing novel biomarkers for clinical diagnosis and future direction in pathophysiological mechanisms.

figureae

Clinical Trial Registration Number: ChiCTR1800018247

Disclosure: L. Qi: None.

81

Regional adipose tissue differences of the proteome reveals an enhanced antioxidative and chaperone activity as a feature of lower body fat in humans

M. Ahmed1, M. Todorcevic2, A. van Dam2, F. Karpe2,3;

1Department of Endocrinology and Diabetes, Gävle Hospital, Gävle, Sweden, 2The Oxford Centre for Diabetes, Endocrinology and Metabolism, University of Oxford, Oxford, UK, 3NIHR Oxford Biomedical Research Centre, OUHFT, Oxford, UK.

Background and aims: The special functions of gluteofemoral adipose tissue appear to convey protection against type 2 diabetes. We searched for specific features in the proteome of gluteofemoral fat tissue to further understand the specialised function.

Materials and methods: Adipose tissue biopsies were taken from abdominal subcutaneous (ASAT) and from gluteal adipose tissue (GSAT) of six healthy women. The proteome was analysed by two‑dimensional gel (2‑DG) electrophoresis followed by tandem mass spectrometry. Top candidates were followed up by transcriptomic profiles, scRNASeq and immuno-histochemistry.

Results: A total of 131 proteins showed ≥2‑fold between ASAT and GSAT samples. The top candidate was Haptoglobin (HP) which was 18-fold higher expressed (p=0.01) in GSAT compared to ASAT. Single cell RNASeq of the human mature adipose and stromovascular fractions (SVF) revealed HP expression in both adipocytes and macrophages, and most highly in a macrophage M2 subpopulation. The pattern of expression was the same in ASAT and GSAT but significantly lower in ASAT (p=0.01). In vitro differentiation of ASAT and GSAT preadipocytes showed maintained higher HP expression in differentiating GSAT cells (p=0.01). Of note, the macrophage expression of HP had a cellular co-expression with its receptor, CD163. Further analysis of GSAT proteome revealed a number of additional enriched proteins protecting against redox and oxidative attack (eg, superoxide dismutase, hemopexin, catalase, peroxiredoxin-1, peroxiredoxin-2, ferritin light chain) and certain chaperones (heat-shock protein beta-5, heat-shock protein 27).

Conclusion: The present study reveals a co-ordinated defence against oxidative stress in GSAT. We propose that local production of haptoglobin provides a first line defence for preadipocytes in the adipose tissue to support the safe long-term storage of fat in the tissue.

Disclosure: M. Ahmed: None.

82

Distinct associations of plasma methionine and cysteine with regional fat distribution: the CODAM and the Maastricht studies

E.C. Tore1, M.C.G. Brouwers2, P.C. Dagnelie2, A. Elshorbagy3, S.J.P. Eussen4,2, J.F.A. Jansen5,6, C.J.H. van der Kallen2, E. Kooi6,2, S.J.R. Meex7, T. Olsen8, H. Refsum8, C.G. Schalkwijk2, C.D.A. Stehouwer2, K. Vinknes8, M.M.J. van Greevenbroek2;

1Department of Internal Medicine, CARIM School for Cardiovascular Disease, Maastricht University, Maastricht, Netherlands, 2CARIM School for Cardiovascular Disease, Maastricht University, Maastricht, Netherlands, 3Department of Pharmacology, University of Oxford, Oxford, UK, 4CAPHRI Care and Public Health Research Institute, Maastricht University, Maastricht, Netherlands, 5MHENS School for Mental Health and Neuroscience, Maastricht University, Maastricht, Netherlands, 6Department of Radiology & Nuclear Medicine, Maastricht University Medical Center, Maastricht, Netherlands, 7Central Diagnostic Laboratory, Maastricht University Medical Center, Maastricht, Netherlands, 8Department of Nutrition, Institute of Basic Medical Sciences, University of Oslo, Oslo, Norway.

Background and aims: Sulfur amino acids (SAAs), including the essential amino acid methionine and its main derivative cysteine, have been associated with obesity and related metabolic diseases, including type 2 diabetes (T2D). Yet, it is not known if SAAs are related to specific patterns of regional fat distribution.

Materials and methods: We examined the relationships of fasting methionine and total cysteine, measured in EDTA plasma using LC-MS/MS, with measures of obesity, body fat distribution and liver fat in two cross-sectional studies enriched with (pre)diabetic individuals: the CODAM cohort (n=478, 61.5% men, 67±7 yrs) and The Maastricht Study (DMS; n=424, 54.7% men, 61.6±8.7 yrs). Main outcomes included BMI, waist circumference, trunk subcutaneous or visceral fat (CODAM and DMS), body composition on dual-energy X-ray absorptiometry (DMS), a liver enzyme score calculated from the standardized plasma levels of the three liver enzymes AST, ALT and GGT (CODAM), fatty liver on ultrasound (CODAM) and liver fat % on MRI (DMS). Associations were examined with linear or logistic regressions, as appropriate, with z-standardized primary exposures and outcomes. Adjustment was made for relevant confounders related to cohort structure, demographics and lifestyle (including protein and alcohol intake, and physical activity), and in DMS also for lean mass (not available in CODAM).

Results: In CODAM, methionine was associated with liver enzyme score (β=0.24; 95% CI: 0.14, 0.36) and fatty liver (OR=1.49; 1.19, 1.88), but not with any measure of obesity. Cysteine was associated with BMI (β=0.19; 0.09, 0.28), waist (β=0.17; 0.08, 0.25) and subcutaneous fat (β=0.15; 0.05, 0.24), but not with visceral fat (β=0.05; -0.04, 0.14) or fatty liver (OR=1.15; 0. 29, 1.43). In DMS, methionine, but not cysteine, was associated with liver fat % (β=0.13; 0.01, 0.26). Cysteine was associated with BMI (β=0.11; 0.02, 0.19) and gynoid fat distribution (β=0.14; 0.06, 0.22), but not with waist (β=0.04; -0.04, 0.11), subcutaneous fat (β=0.07; -0.03, 0.17), visceral fat (β=0.04; -0.05, 0.12) or android fat distribution (β=0.07; -0.02, 0.15).

Conclusion: Methionine and cysteine showed distinct associations with different fat depots, which were similar between studies. Methionine was associated with liver fat, implicated in the development of insulin resistance and T2D. By contrast, cysteine was associated with BMI and peripheral fat depot. These results call for further research on the role of SAAs in regional fat distribution, and their relationship with T2D.

Supported by: Zon-MW and JPI-HDHL

Disclosure: E.C. Tore: Grants; Zon-MW, JPI-HDHL.

83

Thrifty energy phenotype predicts weight regain: results of a randomised controlled trial

K. Mai, L. Spranger, J. Bredow, U. Zeitz, U. Grittner, M. Boschmann, S. Dickmann, N. Stobäus, R. Jumpertz-von Schwartzenberg, J. Spranger;

Charité-Universitätsmedizin Berlin, Berlin, Germany.

Background and aims: Weight loss is associated with an improvement of insulin sensitivity. Both, a negative energy balance and changes of body composition are integrative components of weight loss interventions. However, the individual impact of these two components on insulin sensitivity and energy metabolism is unclear.

Materials and methods: We performed a randomized controlled trial including 80 overweight or obesepost-menopausal women. Participants randomly assigned to the intervention group underwent an 800 kcal/d liquid diet for 2 months followed by four weeks in which the formula diet was substituted by a calorie reduced healthy diet to facilitate further weight loss. This weight loss phase was followed by a 4-week weight maintenance phase, where weight stability was achieved by individualized daily caloric intake without negative energy balance. Volunteers of the control group were instructed to keep their weight stable during the entire period of 4 months. Metabolic phenotyping was performed in both groups at baseline (M0), after weight loss (M3) and after the maintenance period (M4). Additional phenotyping was performed during follow-up at 12 (M12) and 24 months (M24). Primary outcomes were changes of lean body mass (LBM) and changes of insulin sensitivity (ISIClamp) between baseline and M3 and M4. Estimates of energy metabolism were secondary endpoints.

Results: No significant changes of body weight or LBM were found in the control group between any time points. A significant reduction of body weight, fat mass (FM) and LBM was found in the intervention group between M0 and M3, while no further change was seen between M3 and M4. Only subjects of the intervention group were characterized by an improvement of the second primary outcome ISIClamp at M3, which was preserved until M4. Notably, a lower resting energy expenditure per LBM (REELBM) at M3 as well as the individual difference of REELBM between M3 and M4 significantly predicted a stronger regain of fat mass during follow-up.

Conclusion: In summary, our data demonstrate that modulation of LBM and insulinsensitivity during weight loss is predominantly driven by changes in body weight and body composition, rather than an individual effect of negative energy balance. However, the variance in energy expenditure during negative and steady energy balance indicates a thrifty phenotype, which is highly susceptible to future regain of fat mass.

Clinical Trial Registration Number: NCT01105143

Supported by: BMBF

Disclosure: K. Mai: None.

84

Cardiovascular risk of former obesity in healthy-weight Americans

M.P. Smith, W. Mansour, B. De Gale;

St. George's University, St George's, Grenada.

Background and aims: Little research investigates whether effects of obesity persist in those who subsequently achieve and maintain healthy weight. Here we profile formerly-obese American adults and compare their cardiovascular risk factors to those who are currently obese, and to those who were always healthy weight.

Materials and methods: Using data from NHANES 1999-2013 (n=20,271), we define three groups: currently obese; always healthy-weight; and formerly obese. These groups were compared in their prevalence of hypertension, dyslipidemia, and diabetes, in models first crude and then corrected for age, gender, smoking and ethnicity.

Results: Formerly-obese subjects (n=326) were older than those who were never (n=6235) or currently-obese (n=13,710), and likelier to smoke cigarettes (36%, vs. 24 and 19%.) However, after correction their risks of hypertension and dyslipidemia were comparable to those who were never obese (ORs 1.08 and 1.13, both p>0.10.) Their risk of diabetes was intermediate between those who were never- and currently-obese (OR 2.93 for former obesity, OR 7.53 for current) Currently-obese subjects were also at elevated risk of hypertension (OR 3.14) and dyslipidemia (OR 3.11.) All p-values were <0.01 unless stated.

Conclusion: Major weight loss appeared to reverse most of the cardiovascular risks associated with obesity, even in those who continued to smoke and especially in those who quit. Risk of diabetes, but not hypertension or dyslipidemia, remained elevated over those who were never obese: however, even diabetes risk dropped with weight loss.

Disclosure: M.P. Smith: None.

OP 15 Fat in the liver: where it comes from and how it can be prevented

85

Long-term effect of a diet high in unsaturated fat and dietary protein on intrahepatic lipids and circulating FGF21 levels: results of a randomised controlled trial

C. Wernicke1, A. Pohrt2, L. Pletsch-Borba1, N. Meyer1, J. Machann3, C. Gerbracht4, A.F.H. Pfeiffer1, J. Spranger1, K. Mai1;

1Charité – Universitätsmedizin Berlin, corporate member of Freie Universität Berlin and Humboldt-Universität zu Berlin, Department of Endocrinology and Metabolism, Berlin, 2Charité – Universitätsmedizin Berlin, corporate member of Freie Universität Berlin and Humboldt-Universität zu Berlin, Institute of Biometry and Clinical Epidemiology, Berlin, 3Institute for Diabetes Research and Metabolic Diseases (IDM) of the Helmholtz Center Munich at the University of Tübingen, Tübingen, 4Department of Clinical Nutrition, German Institute of Human Nutrition, Potsdam-Rehbruecke, Nuthetal, Germany.

Background and aims: Short-term trials indicate reduction of intrahepatic lipids (IHL) by either high protein or increased unsaturated fat intake beyond energy restriction. Human Fibroblast Growth Factor 21 (FGF21), a hepatokine closely linked to hepatic lipid accumulation, was shown to be strongly affected by dietary protein content. However, human data on dietary induced changes of FGF21 levels are scarce. We investigated the long-term effect of a diet focusing on high protein and increased unsaturated fat intake (NutriAct) on IHL as well as circulating FGF21 levels to analyze potential underlying mechanisms.

Materials and methods: Within a 36-months multi-center trial, 502 subjects (50-80 years) were randomized to either NutriAct intervention group (IG) high in mono-/polyunsaturated fat (15-20%E/10-15%E), plant protein (15-25%E) and fibre (≥30g/day) or control group (CG, usual care) based on recommendations of the German Nutrition Society. The intervention included nutritional sessions and supplementation of newly designed foods mirroring NutriAct pattern. IHL were analyzed by proton magnetic resonance spectroscopy in 346 subjects before and in 258 subjects after 12 months. Subjects with significant alcohol consumption were excluded from the present analysis. Plasma FGF21 levels were measured before and after 12 months using ELISA.

Results: A stronger IHL reduction was seen in the IG after 12 months by trend (-33.3% [95%-confidence interval (CI) -24.5,-41.1] vs. -21.8% [95%-CI -11.0,-31.3] in CG, p=0.071). Analysis of IG subjects adherent to NutriAct intervention revealed significantly higher IHL reduction compared to CG subjects with lower protein and unsaturated fat intake (-40.7% [95%-CI -31.9,-48.4] vs. -15.7% [95%-CI -1.1,-28.2], p=0.001), which persists after adjustment for weight loss, sex and age (p=0.014). At baseline, plasma FGF21 levels correlated with IHL (r=0.326, p<0.001, n = 345) and inversely with dietary protein content (r= - 0.191, p<0.001, n = 429). FGF21 levels decreased in both groups with comparable magnitude (IG (n = 187): - 27.1 pg/ml (SD 136.0), p = 0.007 vs. CG (n = 187): - 12.4 pg/ml (SD 133.9), p = 0.208; p for difference between groups = 0.291). Reduction of FGF21 was associated with increase of dietary protein intake (r = -0.128, p = 0.015, n = 358) and with IHL decrease (r = 0.279, p < 0.001, n= 258) among all participants.

Conclusion: The NutriAct pattern induced a more pronounced long-term improvement of liver fat. This was associated with FGF21 reduction, which could be a potential mediator in the context of diet-induced decline in liver fat. The NutriAct pattern might constitute a beneficial strategy in middle-aged and elderly people in Germany.

Clinical Trial Registration Number: DRKS00010049

Supported by: Federal Ministry of Education and Research (BMBF funding code 01EA1408)

Disclosure: C. Wernicke: None.

86

Fructose intake from fruit juice and sugar sweetened beverages, but not from fruit, is associated with higher intrahepatic lipid content: The Maastricht Study

A.M. Buziau1, S.J.P. Eussen2, M. Kooi3, C.J.H. van der Kallen4, M.C.J. van Dongen2, N.C. Schaper4, R.M.A. Henry4, M.T. Schram4, D.C. Pieter4, M.M.J. van Greevenbroek4, A. Wesselius5, C.G. Schalkwijk1, C.D.A. Stehouwer4, M.C.G. Brouwers4;

1Internal Medicine, Maastricht University, Maastricht, 2Epidemiology, Maastricht University, Maastricht, 3Radiology and Nuclear Medicine, Maastricht University Medical Center+, Maastricht, 4Internal Medicine, Maastricht University Medical Center+, Maastricht, 5Complex Genetics and Epidemiology, Maastricht University, Maastricht, Netherlands.

Background and aims: There has been an ongoing debate on whether dietary fructose is a modifiable risk factor of non-alcoholic fatty liver disease. The aim of the present study was, therefore, to assess the relationship between different sources of dietary fructose and intrahepatic lipid content (IHL) at the population level.

Materials and methods: We used cross-sectional data from The Maastricht Study, a population-based cohort (n=3,981; 60±9 years; 50% women). We assessed the relationship between habitual fructose intake (assessed by a FFQ) - total and derived from fruit, fruit juice and sugar sweetened beverages (SSB) - and log-transformed IHL (quantified by 3T Dixon MRI) with adjustment for age, sex, type 2 diabetes (T2D), educational level, smoking status, physical activity, and intakes of total energy, alcohol, saturated fat, protein, vitamin E, and dietary fiber.

Results: Energy-adjusted total fructose intake was not associated with IHL in the fully adjusted model (p=0.647). Energy-adjusted intake of fructose from fruit was associated with lower IHL after adjustment for dietary factors (p=0.044), but this association was no longer statistically significant after additional adjustment for dietary fiber (p=0.767). In contrast, energy-adjusted intake of fructose from fruit juice and SSB was associated with higher IHL in the fully adjusted model (p=0.019 and p=0.006, respectively). Individuals in the highest quartile of energy-adjusted intake of fructose from fruit juice and SSB had a 1.05-fold (95% CI: 0.98; 1.12) and 1.08-fold (95% CI: 1.00; 1.16) higher IHL, respectively, when compared to the lowest quartile in the fully adjusted model. Finally, the association for fructose from fruit juice were stronger in individuals with T2D (p for interaction=0.071).

Conclusion: Fructose from fruit juice and SSB, but not from fruit, is independently associated with higher IHL. These findings support the recommendation to reduce intake of fructose-containing beverages as a means to prevent hepatic steatosis and cardiometabolic disease at the population level.

Supported by: The Maastricht Study was supported by the European Regional Development Fund via OP-Zuid, the Province of Limburg, the Dutch Ministry of Economic Affairs (grant 31O.041), Stichting De Weijerhorst (Maastricht, the Netherlands), the Pearl String Initiative Diabetes (Amsterdam, the Netherlands), School for Cardiovascular Diseases (CARIM, Maastricht, the Netherlands), School for Public Health and Primary Care (CAPHRI, Maastricht, the Netherlands), School for Nutrition and Translational Research in Metabolism (NUTRIM, Maastricht, the Netherlands), Stichting Annadal (Maastricht, the Netherlands), Health Foundation Limburg (Maastricht, the Netherlands), and by unrestricted grants from Janssen-Cilag B.V. (Tilburg, the Netherlands), Novo Nordisk Farma B.V. (Alphen aan den Rijn, the Netherlands), Sanofi-Aventis Netherlands B.V. (Gouda, the Netherlands), and Medtronic (Tolochenaz, Switzerland).

Disclosure: A.M. Buziau: None.

87

Hepatic glycogen and whole-body fat oxidation are not modulated by one night of prolonged fasting in people with non-alcoholic fatty liver

K.H.M. Roumans1, P. Veeraiah1,2, J. Mevenkamp1,2, B. Havekes3, H.P.F. Peters4, L. Lindeboom1,2, P. Schrauwen1, V.B. Schrauwen-Hinderling1,2;

1Nutrition and Movement Sciences, Maastricht University, Maastricht, 2Department of Radiology and Nuclear Medicine, Maastricht University Medical Center, Maastricht, 3Department of Internal Medicine, Division of Endocrinology and Metabolic Disease, Maastricht University Medical Center, Maastricht, 4Unilever Food Innovation Center, Wageningen, Netherlands.

Background and aims: Time-restricted eating has been shown to improve metabolic health, even in energy balance. These beneficial effects may at least partially be due to creating a more pronounced fasting state, leading to larger fluctuations in hepatic glycogen. Here, we investigated in individuals with non-alcoholic fatty liver (NAFL) whether acutely prolonging an overnight fast from 9.5 hours to 16 hours reduces overnight hepatic glycogen and improves substrate metabolism and whether intrahepatic lipid content and lipid composition is improved after 5 days of such a regime.

Materials and methods: Ten overweight/obese volunteers (BMI 30.1±2.2 kg/m2; age 62±8.7 year; 8 men) with NAFL (intrahepatic lipid content: 13.3±7.7 %weight/weight) participated in a randomized cross-over trial, in which hepatic glycogen was measured by 13C-MRS after a standardized lunch at 2 pm and in the morning at 6.30 am after a 9.5h or 16h fast. Substrate oxidation was determined overnight as well as in the morning upon a meal using whole-body respirometry, and hepatic lipid content and composition were measured by 1H-MRS before and after continuing the prolonged overnight fasting protocol for 5 days.

Results: Remarkably, hepatic glycogen did not decline between afternoon and morning after a 9.5h fast (+4.1±3.3%) and prolonging the fast to 16h did not significantly improve this decline in glycogen (-2.4±4.7%) (figure 1A). In addition, prolonging the overnight fast did not have acute effects on whole-body fat oxidation during the night (2.2±0.2 vs. 2.1±0.1 kJ/min, figure 1B) or upon a breakfast meal. Continuing the prolonged overnight fasting protocol for 5 days did not significantly influence hepatic lipid saturation (SFA +0.8 vs. +2.7%) nor hepatic lipid content (-0.6 vs. -0.6% weight/weight).

Conclusion: These results suggest that hepatic substrate metabolism is rather inert in individuals with NAFL and not sensitive to extending acute fasting periods.

figureaf

Clinical Trial Registration Number: NCT03593343

Supported by: This research was in part financed by the Ministry of Economic Affairs and Climate Policy by means of the PPP Allowance made available by the Top Sector Life Sciences & Health to stimulate public-private partnerships and by Unilever R&D Wageningen. HP was an employee of Unilever at the time this research was designed and conducted and has since changed his professional affiliation to Superfoods, Landsmeer.

Disclosure: K.H.M. Roumans: None.

88

Visceral adipose tissue mitochondrial function is reduced in humans with non-alcoholic fatty liver disease and correlates with insulin resistance

K. Pafili1,2, S. Kahl1,2, D. Pesta1,3, K. Strassburger2,4, L. Mastrototaro1,2, J. Pützer1,2, B. Dewidar1,5, T. Sarabhai1,6, I. Esposito7, M. Schlensak8, M. Roden1,6;

1Institute for Clinical Diabetology, German Diabetes Center, Leibniz Center for Diabetes Research at Heinrich-Heine-University Düsseldorf, Düsseldorf, Germany, 2German Center for Diabetes Research, Partner Düsseldorf, München-Neuherberg, Germany, 3Institute of Aerospace Medicine, German Aerospace Center (DLR), Cologne, Germany, 4Institute for Biometrics and Epidemiology, German Diabetes Center, Leibniz Center for Diabetes Research, Heinrich Heine University, Düsseldorf, Germany, 5Department of Pharmacology and Toxicology, Faculty of Pharmacy, Tanta University, Tanta, Egypt, 6Department of Endocrinology and Diabetology, Medical Faculty, University Hospital Düsseldorf, Heinrich-Heine-University Düsseldorf, Düsseldorf, Germany, 7Institute of Pathology, Heinrich Heine University, Düsseldorf, Germany, 8Department of General and Visceral Surgery, Neuwerk Hospital, Mönchengladbach, Germany.

Background and aims: Adipose tissue mitochondrial dysfunction may drive the development of type 2 diabetes (T2D) and non-alcoholic (NA) fatty liver disease (NAFLD), comprising NA fatty liver and steatohepatitis (NAFL and NASH). Potential underlying mechanism comprise impaired adipose tissue mitochondrial mass and respiration favoring increased fatty acid flux to ectopic tissues. Thus, we examined subcutaneous (SAT) and visceral adipose tissue (VAT) mitochondrial oxidation in humans with different degrees of insulin resistance and varying liver histology.

Materials and methods: Obese people without NAFL (OBE-CON, n=20, 38±8 years, body mass index 53±6 kg/m2, 10% T2D), with NAFL (OBE-NAFL, n=20, 40±8 years, 51±5 kg/m2, 25% T2D) or NASH (OBE-NASH, n=20, 42±10 years, 51±6 kg/m2, 40% T2D) underwent intensive baseline metabolic characterisation and tissue biopsies. O2 flux rates from different substrates were measured with high resolution respirometry in SAT and VAT.

Results: In VAT, maximal uncoupled respiration was lower in OBE-NAFL (least square means (LSM): 0.46 pmol*mg wet weight-1*s-1 [95% confidence interval 0.21;0.71], p<0.05) and OBE-NASH (LSM: 0.51 [0.26;0.76], p<0.001) compared to OBE-CON. Similar differences were seen for maximal ADP-stimulated respiration in VAT (following the administration of succinate). In Pearson correlation analysis, maximal ADP stimulated O2 flux in VAT positively correlated with whole-body insulin sensitivity (M value; r= 0.29, p<0.05). Mitochondrial content in VAT, as assessed from mitochondrial DNA copy number, was not different between the groups. Protein expression of electron transport chain complexes was similar in VAT and SAT of all groups, except for complex IV that tended to be lower in VAT in OBE-NASH than in OBE-CON (p=0.065). Of note, a comparable oxidative capacity was revealed in all three groups in the SAT, despite a higher mitochondrial content in OBE-NASH than in OBE-CON (p=0.045).

Conclusion: VAT mitochondrial respiration was reduced in obese humans with NAFLD compared to those without and correlated positively with whole-body insulin sensitivity, demonstrating an adipose tissue-specific effect of mitochondrial function on insulin resistance and hepatic lipid accumulation in the context of obesity.

Clinical Trial Registration Number: NCT01477957

Supported by: BMG, MKW NRW, BMBF, EFRE-0400, DDG, DFG; CRC/SFB 1116/2 B12, E! 113230 DIA-PEP191,

Disclosure: K. Pafili: Grants; German Federal Ministry of Health (BMG), Ministry of Culture and Science of the State North Rhine-Westphalia (MKW NRW), German Federal Ministry of Education and Research (BMBF).

89

Multi-organ multiparametric magnetic resonance imaging reveals distinct ectopic fat distribution in type 2 diabetics with and without co-existing obesity

T. Waddell1,2, A. Bagur1,2, D. Cunha1, H. Thomaides-Brears1, R. Banerjee1, M. Brady1;

1Perspectum, Oxford, 2Department of Engineering Science, The University of Oxford, Oxford, UK.

Background and aims: Type 2 diabetes (T2D) is associated with multi-organ dysfunction and ectopic fat accumulation within the liver, pancreas and as visceral adipose tissue (VAT). Multiparametric magnetic resonance (mpMR) can provide a non-invasive assessment of multi-organ health, including measures of steatosis (PDFF) and fibro inflammation (cT1). Here we collect mpMR markers of the liver, pancreas, VAT, subcutaneous adipose tissue (SAT) and skeletal-muscle index (SMI), comparing T2D and non-T2D (NT2D) participants matched for age, gender and BMI.

Materials and methods: MR images and patient demographics were collected from the UK Biobank online resource. Liver and pancreas images were analysed by placing three regions of interest (ROIs) within the organ parenchyma only on PDFF and T1 maps. Images extracted from the 3rd lumbar vertebrae were manually segmented for SAT, VAT and SMI. Measures of high density lipoprotein (HDL), glycated haemoglobin (HbA1c), aspartate aminotransferase (AST) and alanine transaminase (ALT) were collected. Data from 297 participants (131 T2D, 136 NT2D) were included for analysis. Wilcoxon tests were performed to compare differences in biomarkers between the two groups. Data is median [IQR].

Results: Participants with T2D had significantly higher liver fat (7.4% [4.1-13.8] vs 5.3% [2.7-10.6]; p=0.011) and cT1 (730ms [685-786] vs 709ms [671-753]; p=0.019), despite no differences in AST (p=0.35) or ALT (p=0.11). Significantly lower measures of SMI (45.2cm2/m2 [38.1-52.9] vs 50.6cm2/m2 [42.6-55.9]; p=0.003) and HDL (1.1mmol/L [1-1.3] vs 1.3mmol/L [1.1-1.5]; p<0.0001) were observed within the T2D group. We found no significant differences in VAT (p=0.35), SAT (p=0.43) or pancreas PDFF (p=0.22). However, all mpMR metrics were significantly greater (p<0.001) in obese vs normal weight participants, irrespective of T2D status.

Conclusion: In BMI, age and gender matched participants from the UK Biobank, mpMR reveals significantly greater liver fat and fibroinflammation in those with T2D, despite no differences in AST or ALT. Pancreatic fat and VAT were significantly greater in obese participants but were similar in both T2D and NT2D groups.

figureag

Supported by: Tom Waddell is support by the Royal Commission for the Exhibition of 1851.

Disclosure: T. Waddell: Employment/Consultancy; Employee at Perspectum Ltd.

90

Inferring causal pathways between metabolic processes and liver fat accumulations: an IMI DIRECT study

N. Atabaki Pasdar1, H. Pomares-Millan1, R.W. Koivula2, L.Z. Agudelo3, G.N. Giordano1, A. Viñuela4, E. Thomas5, A. Mari6, R. Hartmut7, M. Ohlsson8, J.D. Bell5, I. Pavo9, E. Pearson10, P.W. Franks1, IMI DIRECT Consortium;

1Lund University, Malmö, Sweden, 2University of Oxford, Oxford, UK, 3Massachusetts Institute of Technology, Cambridge, USA, 4Newcastle University, Newcastle, UK, 5University of Westminster, London, UK, 6CNR Institute of Neuroscience, Padova, Italy, 7Boehringer Ingelheim International GmbH, Mainz, Germany, 8Lund University, Lund, Sweden, 9Eli Lilly Regional Operations GmbH, Vienna, Austria, 10University of Dundee, Dundee, UK.

Background and aims: Many observational studies show that type 2 diabetes (T2D) and non-alcoholic fatty liver disease (NAFLD) coincide. However, the causal nature of these relationships is unclear. Here, we aimed to assess the strength and magnitude of the putative causal pathways linking dysglycemia and fatty liver using a combination of causal inference methods.

Materials and methods: Measures of glycemia, insulin dynamics, MRI abdominal and liver fat content, serological biomarkers, and anthropometry were obtained in participants from the IMI DIRECT cohorts (n=795 with new-onset T2D and n=2234 without T2D). UK Biobank was also used for modeling and replication purposes (n=4617 who had liver fat MRI measurement). Bayesian networks (BNs) were employed to build graphical models from joint probability distributions of variables to infer causal pathways. To attain a stable structure, model averaging was performed on the bootstrapped BNs. To validate these causal pathways, a series of bidirectional two-sample Mendelian randomization (MR) analyses was undertaken among all possible combinations of the selected variables.

Results: BNs in the IMI DIRECT combined cohorts identified higher basal insulin secretion rate (BasalISR) and excess visceral fat (VAT) accumulation most likely to cause liver fat accumulation (shown in the figure). Conditioning on high levels of BasalISR and VAT significantly increased the unconditional probability of high liver fat level on the random observations generated by the BNs; 23% increase after conditioning on VAT, 32% increase after conditioning on BasalISR, and 40% after conditioning on both. Similarly, VAT was the primary causal variable for liver fat accumulation in the UK Biobank BN analysis (BasalISR was not measured in the UK Biobank). Through the MR analyses, several nominal directional associations between hepatic biomarkers, glycemic, and adiposity measures were suggested after Bonferroni correction and pleiotropy sensitivity analysis. In most cases, the effects predicted by the BNs were confirmed in the MR analyses.

Conclusion: Through the conducted analyses BasalISR had the highest upstream causal effect on the liver fat content, highlighting the strong link between NAFLD and T2D. Understanding more about the BasalISR mechanism of action on liver fat accumulation and considering it as a biomarker measurement in clinical practice may help in the diagnosis, prevention, and treatment of these highly prevalent and linked conditions.

figureah

Supported by: ‘IMI DIRECT was supported by the Innovative Medicines Initiative Joint Undertaking under grant agreement n°115317 (DIRECT), resources of which are composed of financial contribution from the European Union's Seventh Framework Programme (FP7/2007-2013) and EFPIA companies’ in kind contribution.’

Disclosure: N. Atabaki Pasdar: None.

OP 16 CKD in diabetes - a costly complication

91

Inside CKD: modelling the clinical and economic impact of routine screening for albuminuria in people with type 2 diabetes

S. Nolan1, J. Ärnlöv2, M.C. Batista3, S. Chadban4, G.M. Chertow5, L. De Nicola6, J.-M. Halimi7, E. Kanda8, G. Li9, F.S. Mennini10, J.F. Navarro-González11, A. Power12, L. Retat13, N. Tangri14, J. Wish15;

1Global Medical Affairs, BioPharmaceuticals Medical, AstraZeneca, Cambridge, UK, 2Karolinska Institutet, Stockholm, Sweden, 3Hospital Israelita Albert Einstein, São Paulo, Brazil, 4Royal Prince Alfred Hospital, Camperdown, Australia, 5Stanford University School of Medicine, Palo Alto, USA, 6University Luigi Vanvitelli, Naples, Italy, 7University Hospital of Tours, Tours, France, 8Kawasaki Medical University, Okayama, China, 9Sichuan Academy of Medical Science, Sichuan Provincial People’s Hospital, Chengdu, China, 10EEHTA CEIS, Faculty of Economics, University of Rome “Tor Vergata", Rome, Italy, 11University Hospital Nuestra Señora de Candelaria, Santa Cruz de Tenerife, Spain, 12North Bristol NHS Trust, Bristol, UK, 13HealthLumen Limited, London, UK, 14University of Manitoba, Winnepeg, Canada, 15Indiana University School of Medicine, Indianapolis, USA.

Background and aims: Early diagnosis of chronic kidney disease (CKD) in patients with type 2 diabetes (T2D) followed by guideline-recommended interventions is key to slowing CKD progression, but adherence to screening recommendations is suboptimal. Inside CKD models the global burden of CKD using country-specific, patient-level microsimulation models. We modelled the effects of targeted implementation of urine albumin:creatinine ratio (UACR) measurement and intervention in patients with T2D.

Materials and methods: We used the Inside CKD microsimulation to model the impact of measuring UACR during routine primary care visits with subsequent intervention in patients with T2D aged ≥ 45 years with a range of kidney functions, versus current practice. Virtual populations were constructed using published country-specific data on demographics, CKD (albuminuria and estimated glomerular filtration rate status), T2D, comorbidities and complications.

Results: Preliminary data for three countries show that from 2020 to 2026, the measurement of UACR with subsequent intervention in patients with T2D would prevent CKD progression to stages 3b to 5 in 164 739 patients in the UK, 964 121 in the US (Figure) and 156 482 in Canada. Associated cost savings would be £0.14B, US$13.81B and C$2.34B. Additional countries will be analysed.

Conclusion: Routine UACR measurement with subsequent intervention could potentially reduce the global burden of CKD and healthcare costs in patients with T2D and improve patient outcomes.

figureai

Disclosure: S. Nolan: Employment/Consultancy; Stephen Nolan is an employee of AstraZeneca. Stock/Shareholding; Stephen Nolan holds stock options at AstraZeneca.

92

Hyper-filtration most strongly predicts decline in estimated glomerular filtration rate: results from analysis of 73,583 person-years

S. Katoh1, K. Yokoyama2, M. Zeniya3, Y. Sakamoto4, K. Utsunomiya5, R. Nishimura1;

1Div. of Diabetes, Metabolism & Endocrinology, The Jikei University School of Medicine, Tokyo, 2Harumi Triton Clinic, The Jikei University School of Medicine, Tokyo, 3Gastroenterology, Akasaka Sanno Medical Center, Tokyo, 4The Jikei University School of Medicine, Tokyo, 5Department of Health-Care Center, The Jikei University School of Medicine, Tokyo, Japan.

Background and aims: We examined predictors of change in estimated glomerular filtration rate (ΔeGFR) per year calculated using serum creatinine.

Materials and methods: This retrospective study comprised from Periods I from April 2005 to September 2006, Period II from April 2009 to September 2010, Period III from April 2013 to September 2014, Period IV from April 2017 to September 2018. Participants were 10,698 people (7137 men, 3561 women; mean age 48 years) who underwent annual medical checkups during the above periods, and we detected the longest observation period for each person in order to evaluate ΔeGFR (ml/min/1.73 m2/year) during the observation period. We performed stepwise regression analysis with ΔeGFR as the dependent variable to examine its associations with baseline variables and use of lipid-lowering drugs.

Results: There were 73,583 person-years of observation, and the mean observation period was 7 ± 3 years. Stepwise regression (SR) in people with diabetes (n = 1037) revealed that baseline eGFR (standardized β [Sβ] = -0.512, p < 0.001) was the strongest independent determinant of ΔeGFR (-1.1 ± 2.6 ml/min/1.73 m2/year), and age (Sβ = -0.259, p < 0.001), fasting plasma glucose (FPG) (mg/dL) (Sβ = -0.107, p < 0.001), uric acid (UA) (mg/dL) (Sβ = -0.096, p = 0.004), female sex (Sβ = -0.094, p = 0.006), exercise habit (EH) (min/week) (Sβ = 0.124, p < 0.001), aspartate aminotransferase (AST) (U/L) (Sβ = 0.100, p = 0.001), and ethanol intake (EI) (g/week) (Sβ = 0.064, p = 0.038) were also significantly associated with ΔeGFR, but BMI (kg/m2), waist circumference (WC) (cm), systolic BP (mmHg), diastolic BP (mmHg), total cholesterol (mg/dL), HDL-cholesterol (mg/dL), triglycerides (mg/dL), LDL-cholesterol (mg/dL), alanine aminotransferase (U/L), gamma-glutamyl transpeptidase (GGTP) (U/L), cholinesterase (U/L), hepatitis B antigen or hepatitis C antibody positivity, C-reactive protein (CRP) (mg/dL), smoking index (SI) (cigarettes per day × years of smoking), vegetable intake (VI) (at least one plate/every day), and use of lipid-lowering drugs were not significantly associated. SR in people without diabetes (n = 9661) revealed that baseline eGFR (Sβ = -0.625, p < 0.001) was the strongest independent determinant of ΔeGFR (-1.6 ± 2.9 ml/min/1.73 m2/year), and age (Sβ = -0.189, p < 0.001), female sex (Sβ = -0.128, p < 0.001), BMI (Sβ = -0.092, p < 0.001), UA (Sβ = -0.067, p < 0.001), diastolic BP (Sβ = -0.023, p = 0.025), WC (Sβ = 0.112, p < 0.001), EH (Sβ = 0.093, p < 0.001), HDL-cholesterol (Sβ = 0.061, p < 0.001), EI (Sβ = 0.044, p < 0.001), FPG (Sβ = 0.034, p = 0.001), GGTP (Sβ = 0.026, p = 0.011), VI (Sβ = 0.024, p = 0.010), and SI (Sβ = 0.020, p = 0.033) were also significantly associated with ΔeGFR, but other variables were not significantly associated. ΔeGFR was significantly higher in people with diabetes than in those without diabetes (p < 0.001).

Conclusion: Among the variables investigated, baseline eGFR indicating hyper-filtration was significantly and most strongly associated with decline in eGFR during the observation period in people with diabetes as well as those without diabetes.

Clinical Trial Registration Number: 20-130 5420

Disclosure: S. Katoh: None.

93

Metabolic syndrome, and not obesity, is associated with chronic kidney disease in the general US population

E. Muraca1, C. Ballabeni2, R. Trevisan3,4, G. Perseghin1,4, S. Ciardullo1,4;

1Medicine and Rehabilitation, Policlinico di Monza, Monza, 2Nephrology, Policlinico di Monza, Monza, 3Endocrinology and Diabetes Unit, ASST Papa Giovanni XXIII, Bergamo, 4Medicine and Surgery, University of Milano Bicocca, Milano, Italy.

Background and aims: Obese patients are at increased risk of chronic kidney disease, but it is still unclear whether this can be attributed to obesity per se or to the associated metabolic derangements. The aim of this study is to evaluate the relative impact of obesity and metabolic syndrome (MS) on kidney disease.

Materials and methods: This is a cross-sectional study based on data obtained in the 2005-2016 cycles of the National Health and Nutritional Examination Survey (NHANES). Adult participants were considered eligible if data were available on body mass index (BMI), estimated glomerular filtration rate (eGFR), urine-albumin to creatinine ratio (UACR) and each of the MS components. Based on the presence of obesity and MS, participants were categorized in four groups: metabolically healthy non-obese (MHNO), metabolically unhealthy non-obese (MUNO), metabolically healthy obese (MHO) and metabolically unhealthy obese (MUO). Primary outcomes were eGFR<60 ml/min and UACR≥30 mg/g. Logistic regression analysis was performed to evaluate the association between obesity, MS and kidney outcomes after adjustment for age, sex and race/ethnicity.

Results: The studied population comprised 12174 participants (MHNO: 5937; MUNO: 1844; MHO: 1574; MUO: 2819). MHO patients were younger and more commonly female. Compared with MHNO participants, an increased prevalence of albuminuria and reduced eGFR were present in both the MUNO group (OR 1.57, 95% CI 1.28-1.93, p<0.001 and OR 1.56, 95% CI 1.17-2.07, p=0.003, respectively) and the MUO group (OR 2.25, 95% CI 1.88-2.70, p<0.001 and OR 1.76, 95% CI 1.37-2.26, p<0.001, respectively), but not in the MHO group (OR 0.95, 95% CI 0.73-1.27, p=0.774 and OR 0.94, 95% CI 0.61-1.46, p=0.789, respectively). When each of the MS components was evaluated separately, elevated blood pressure and HDL cholesterol were associated with both albuminuria and reduced eGFR, while elevated blood glucose and triglycerides were only associated with albuminuria.

Conclusion: This large cross-sectional study suggests that MS and not obesity per se is associated with kidney damage. In particular, the MHO patients does not seem to carry an increased risk of kidney disease.

Disclosure: E. Muraca: None.

94

Non-dipping of nocturnal blood pressure is associated with increased risk of mortality and kidney disease in type 1 diabetes

H. Hjortkjær1, F. Persson1, S. Theilade1,2, S. Winther1, N. Tofte1,3, T. Ahluwalia1,4, P. Rossing1,4;

1Steno Diabetes Center Copenhagen, Gentofte, 2Department of Medicine, Herlev-Gentofte Hospital, Herlev, 3Novo Nordisk A/S, Søborg, 4University of Copenhagen, Copenhagen, Denmark.

Background and aims: Persons with type 1 diabetes (T1D) have increased risk of cardiovascular disease (CVD) and mortality. A 24-hour ambulatory BP measurement (ABPM) evaluates diurnal variations in BP and has previously been shown to be superiorly correlated to future CVD compared to office and home BP measurements; however, its role in T1D remains to be determined. This study aims to determine the prognostic significance of 24-hour ABPM and non-dipping of nocturnal BP for CVD, mortality and kidney disease in persons with T1D.

Materials and methods: A cohort of 654 participants with T1D was examined from 2009 through 2011 at Steno Diabetes Center Copenhagen, including measurements of 24-hour ABPM using a tonometric wrist-watch device (BPro, HealthStats, Singapore). In 2016, outcomes (CV events, all-cause mortality, decline in eGFR ≥30%, end-stage kidney disease (ESKD) and a composite kidney event of decline in eGFR ≥30%, ESKD and mortality) were registered and hazard ratios (HR) were calculated using Cox regression analyses. Non-dipping was defined as a nocturnal decline in systolic BP of less than 10%.

Results: Participants were mean ± SD 55 ± 12 years old, had median (IQR) 35 (24-44) years duration of diabetes, 55.4% were men, BMI 25.5 ± 5.8 kg/m2, eGFR 88 (65-110) ml/min/1.73m2, HbA1c 64.4 ± 12.6 mmol/mol and office systolic and diastolic BP of 132 ± 17 mmHg and 74 ± 9 mmHg. After a median of 6.2 years for all-cause mortality and 5.2 years for the remaining outcomes, we registered 90 new CV events, 54 mortality events, 92 events of decline in eGFR ≥30%, 21 events of ESKD and 120 composite kidney events. The participants had a mean daytime systolic BP of 133 ± 16 mmHg and nighttime systolic BP of 121 ± 16 mmHg with 337 (51.5%) participants being classified as having non-dipping of nocturnal BP. In unadjusted analyses, non-dipping was associated with all-cause mortality (HR 2.59 (95% CI 1.43-4.71), p=0.002, see Figure 1), decline in eGFR ≥30% (HR 1.85 (1.21-2.83), p=0.004), ESKD (HR 4.24 (1.43-12.60), p=0.009) and the composite kidney event (HR 2.45 (1.66-3.62), p<0.001), but not with new CV event (HR 1.25 (0.82-1.89), p=0.30). After adjustments for age, sex, diabetes duration, HbA1c, BMI, HDL, smoking, previous CVD, use of antihypertensive medication/statins/diuretics, office systolic BP, urine AER and eGFR, non-dipping was associated with all-cause mortality (HR 2.01 (1.08-3.91), p=0.03) and the composite kidney event (HR 1.85 (1.21-2.84), p=0.005).

Conclusion: We find that non-dipping of nocturnal BP in persons with T1D is associated with an increased risk of mortality and kidney events. Although the BP measurements were from a wrist-watch device, this is the first time such results are demonstrated in persons with T1D.

figureaj

Disclosure: H. Hjortkjær: None.

95

High plasma concentrations of folic acid are associated with increased risk of graft failure in renal transplant recipients with type 2 diabetes

J.L. Flores-Guerrero, M.C.J. Osté, A. Gomes Neto, M.F. Eisenga, G. Navis, R.P.F. Dullaart, S.J.L. Bakker;

Internal Medicine, University Medical Center Groningen, Groningen, Netherlands.

Background and aims: The role of folic acid in the progression of kidney disease has gained attention in recent years. Furthermore, alterations in one carbon metabolism, and deleterious effects of circulating folate in T2D patients have been described. Nevertheless, the association of circulating folic acid and graft failure remains undetermined. Hence, we aimed to investigate the prospective association of folic acid with graft failure in renal transplant recipients (RTRs) with and without Type 2 Diabetes.

Materials and methods: We included 340 RTRs from the of the TransplantLines prospective cohort, conducted in the Groningen region, in the north of the Netherlands. Out of the 340 RTRs, 170 RTRs with T2D were matched with RTRs without diabetes using a Propensity Score, with an Optimal Matching method. Cox proportional-hazards regression analyses were performed to study the association of plasma folic acid with graft failure. The main outcome was graft failure, which was defined as re-transplantation or return to dialysis and was censored for death.

Results: Among RTRs (age 57.2 ± 10.3 years; 55% males), baseline median folic acid was 11.0 (6.0, 20.0) μmol/L in RTRs with T2D and 10.0 (5.3, 18.0) μmol/L in RTRs without T2D (p = 0.68). After a median follow-up of 6.1 (5.6-6.7) years, graft failure was observed in 24 RTRs with T2D and in 18 RTRs without T2D. In RTRs with T2D folic acid was associated with increased risk of graft failure, independent of age, sex, systolic blood pressure, HDL-cholesterol, triglycerides, immunosuppressive therapy, albuminuria and eGFR (adjHR per 1-SD increase, 1.69 (95% confidence interval (CI): 1.10; 2.61, p = 0.01). Importantly, folic acid was not associated with increased risk of graft failure, neither in the crude model, nor in the full adjusted model in RTRs without T2D (adjHR per 1-SD increase, 0.75 (95% confidence interval (CI): 0.37; 1.38, p = 0.32).

Conclusion: This study suggests that circulating concentrations of folic acid are associated with an increased risk of graft failure only in RTRs with T2D, but not in RTRs without T2D. Further investigation, in relation to folic acid activating mechanisms in T2D patients with kidney disease is needed.

Disclosure: J.L. Flores-Guerrero: None.

96

Fully automated closed-loop versus standard insulin therapy in adults with type 2 diabetes requiring dialysis: a randomised controlled trial

L. Bally1, C.K. Boughton2,3, A. Tripyla1, S. Hartnell4, A. Daly2, D. Herzig1, M.E. Wilinska2, C. Czerlau1, A. Fry5, R. Hovorka2;

1Bern University Hospital, University of Bern, Bern, Switzerland, 2Wellcome Trust-MRC Institute of Metabolic Science, University of Cambridge, Cambridge, UK, 3Wolfson Diabetes and Endocrine Clinic, Cambridge University Hospitals NHS Foundation Trust, Cambridge, UK, 4Cambridge University Hospitals NHS Foundation Trust, Cambridge, UK, 5School of Clinical Medicine, Cambridge, UK.

Background and aims: Diabetes management in patients with end-stage renal disease on dialysis is challenging due to complex interactions with glucose metabolism. We evaluated safety and efficacy of fully closed-loop compared with standard insulin therapy in adults with type 2 diabetes (T2D) requiring maintenance dialysis.

Materials and methods: In an open-label, multinational, two centre, randomised, crossover trial, 26 adults with T2D requiring dialysis (17 males, mean±SD: age 68±11 years, diabetes duration 20±10 years) underwent two 20-day periods of unrestricted living, comparing the CamAPS HX fully closed-loop system using faster insulin aspart, with standard insulin therapy (control) in random order.

Results: The proportion of time in the target glucose range (5.6 to 10.0mmol/L; primary endpoint; Table1) was 52.8±12.5% with closed-loop vs. 37.7±20.5% with control; mean-adjusted difference 15.1 percentage points (95% CI 8.1 to 22.1; P<0.001). Mean glucose was lower with closed-loop than during the control (10.1±1.3 vs. 11.6±2.8mmol/L; P=0.002). Time in hypoglycaemia (<3.9mmol/L) was reduced with closed-loop vs. control (median [IQR] 0.1 [0.0-0.4%] vs. 0.2 [0.0-0.9%]; P=0.040). Total daily insulin requirements were similar between interventions. No episodes of severe hypoglycaemia occurred during control period and one occurred during closed-loop period but not during closed-loop operation.

Conclusion: Fully closed-loop versus standard insulin therapy in adults with T2D requiring dialysis improved glucose control whilst reducing hypoglycaemia, demonstrating safety and efficacy in this vulnerable population.

figureak

Clinical Trial Registration Number: NCT04025775

Supported by: Swiss Kidney Foundation, Swiss Society of Diabetes and Endocrinology, National Institute for Health Research Cambridge Biomedical ResearchCentre and Wellcome Trust Strategic Award (100574/Z/12/Z), Dexcom, Novo Nordisk UK Research Foundation

Disclosure: L. Bally: None.

OP 17 Don't stop moving: beneficial effects of exercise on diabetes and beyond

97

Regular exercise training improves skeletal muscle glucose uptake in monozygotic twin pairs discordant for body weight

J. Hentilä1, R. Ojala1, M.S. Lietzén1, M.A. Heiskanen1, S.M. Honkala1, R. Lautamäki2, E. Löyttyniemi3, T. Malm4, L. Lahti5, J.O. Rinne1,6, K.H. Pietiläinen7,8, J. Kaprio9, J.C. Hannukainen1;

1Turku PET Center, University of Turku, Turku, 2Heart Centre, Turku University Hospital, Turku, 3Department of Biostatistics, University of Turku, Turku, 4A.I. Virtanen Institute for Molecular Sciences, University of Eastern Finland, Kuopio, 5Department of Future Technologies, University of Turku, Turku, 6Turku PET Centre, Turku University Hospital, Turku, 7Obesity Research Unit, Research Program for Clinical and Molecular Metabolism, Faculty of Medicine, University of Helsinki, Helsinki, 8Abdominal Center, Obesity Center, Endocrinology, University of Helsinki and Helsinki University Central Hospital, Helsinki, 9Institute for Molecular Medicine Finland FIMM, HiLIFE, University of Helsinki, Helsinki, Finland.

Background and aims: Skeletal muscle plays an important role for whole-body insulin sensitivity. Body adiposity is associated with decreased skeletal muscle insulin sensitivity, while exercise counteracts this effect. However, it is unclear whether the response to exercise is affected by body adiposity per se, because training-induced adaptations may be confounded by genetic factors. Thus, we investigated the effects of body adiposity and regular exercise training on insulin-stimulated skeletal muscle glucose uptake (GU) in various muscles while controlling genetic variability by studying monozygotic (MZ) twin pairs discordant for body weight.

Materials and methods: We recruited MZ twin pairs discordant for BMI (difference at least 2 kg·m-2) from three twin cohorts (FinnTwin12, FinnTwin16 and the Finnish twin Cohort study). Of the contacted 47 MZ twin pairs, 12 MZ pairs participated in the study (8 female, 4 male pairs; mean age 40.4 (SD) 4.5 years; leaner twins mean BMI 29.1 (SD) 6.3, heavier twins mean BMI 36.7 (SD) 7.0, p<0.001). Of these participants 8 pairs have completed the intervention. Participants performed a 6-month-long exercise-intervention (two endurance, one high-intensity interval training and one resistance exercise session per week). Insulin-stimulated GU was studied by positron emission tomography (PET) using [18F]FDG during euglycaemic hyperinsulinemia in erector spinae (ES), biceps brachii (BB), latissimus dorsi (LD), quadriceps femoris (QF) and psoas major (PM) muscles before and after the intervention. Data was analysed by hierarchical linear mixed model.

Results: At baseline, leaner twins had smaller fat percentage as well as better cardiorespiratory fitness (VO2max) and whole-body insulin sensitivity (M-value; p<0.01 in all), than their heavier co-twins. Compared with heavier twins, the leaner co-twins had higher GU in ES, BB, LD and QF muscles (p<0.05, in all) but not in PM (p=0.68). Exercise intervention increased M-value and VO2max (time p<0.05 for both) similarly in both leaner and heavier twins (time x group: p>0.23), but had no effect on BMI or whole-body fat percentage (time: p>0.30). Exercise intervention increased GU in ES, LD and PM (time: p<0.05) and tended to increase in QF (p=0.068) similarly in both leaner and heavier twins (time x group: p>0.39).

Conclusion: When genetic variability is controlled, body adiposity is associated with decreased insulin-stimulated skeletal muscle glucose uptake. Moreover, regular exercise training improves skeletal muscle insulin sensitivity, independent of body adiposity.

Clinical Trial Registration Number: NCT03730610

Supported by: The Academy of Finland (JCH decision 317332, KHP decisions 272376, 314383, 335443, 266286, JK decision 336823), the Finnish Cultural Foundation (JCH, MAH), the Diabetes Research Foundation of Finland (JCH, MAH, KHP), Novo Nordisk Foundation (KHP, NNF20OC0060547, NNF17OC0027232, NNF10OC1013354), Helsinki University Hospital (KHP), Government Research Funds (KHP), Finnish Medical Foundation (KHP), Gyllenberg Foundation (KHP), Sigrid Juselius Foundation (KHP, JK), and University of Helsinki (KHP, JK), State Research Funding/Hospital District of Southwest Finland (JCH).

Disclosure: J. Hentilä: None.

98

Dynamic profiling of the metabolic response to endurance exercise

M. Hoene1, X. Zhao2, C. Hu2, A. Moller3,4, P. Schneeweiss5,6, M. Heni4,7, A. Birkenfeld4,7, A.M. Niess5,6, A. Peter1,4, R. Lehmann1,4, G. Xu2, C. Weigert1,4;

1Institute for Clinical Chemistry and Pathobiochemistry, Department for Diagnostic Laboratory Medicine, University Hospital Tübingen, Tübingen, Germany, 2CAS Key Laboratory of Separation Science for Analytical Chemistry, Dalian Institute of Chemical Physics, Dalian, China, 3German Center for Diabetes Research (DZD), Tübingen, Germany, 4Institute for Diabetes Research and Metabolic Diseases of the Helmholtz Zentrum München at the University of Tübingen, Tübingen, Germany, 5Department of Sports Medicine, University Hospital Tübingen, Tübingen, Germany, 6Interfaculty Research Institute for Sports and Physical Activity, University of Tübingen, Tübingen, Germany, 7Department of Internal Medicine IV, University Hospital Tübingen, Tübingen, Germany.

Background and aims: Physical exercise is a potent remedy to prevent obesity-associated metabolic diseases. However, the mechanisms underlying its beneficial effects are still incompletely understood. Assessing the dynamic changes in the plasma metabolome during an acute exercise challenge could reveal novel aspects of the metabolic adaptation to a training intervention that are not captured by assessing the resting metabolome only.

Materials and methods: Sedentary obese, non-diabetic subjects (14 female, 8 male) performed 8 weeks of supervised endurance training (3x1 hour/week, 80% VO2 peak). As previously published, this intervention improved VO2 peak and aerobic threshold, with minor reduction of the BMI and a slight, but non-significant increase in insulin sensitivity (ISIMats). The first and last training session were designed as acute endurance exercise challenges. Here, subjects arrived in the fasting state and resting blood samples were obtained before the participants had a standardized breakfast. After another 45 minutes, subjects performed 30 minutes of bicycle ergometer exercise at the end of which another blood sample was drawn. Plasma samples were analyzed using liquid chromatography- mass spectrometry (LC-MS) and capillary electrophoresis (CE)- MS which together covered a broad range of metabolites (>300) with mixed polarity. Statistical analysis was performed using the Mixed Model platform in JMP 14.2.0 (SAS).

Results: The plasma metabolomics pattern was pronouncedly altered by the acute exercise challenges. Metabolites acutely affected by exercise (p<0.05, fold change >10%) were mostly increased. This increase was particularly pronounced (>100%) for hypoxanthine and lactate, both before and after the training intervention. The increase of several metabolites, including most amino acids, during the pre-intervention challenge correlated with the improvement of insulin sensitivity during the intervention. Compared to the acute exercise challenges, training only caused minor alterations in the resting metabolome. However, several metabolites exhibited a different regulation between the initial and the final challenge, and most of these, e.g. different steroid sulfates, were attenuated after training. Thus, the challenged metabolome exhibited a stronger response to the training intervention than the resting metabolome.

Conclusion: Our results indicate that metabolomic profiling of an acute exercise challenge is useful in assessing and, potentially, predicting metabolic adaptations to an endurance training intervention in sedentary, diabetes-prone subjects.

Clinical Trial Registration Number: NCT03151590

Supported by: German Federal Ministry of Education and Research (BMBF)

Disclosure: M. Hoene: None.

99

Effects of regular exercise and carnosine on muscle energy metabolism and cognitive performance in the overweight elderly population

J. Ukropec1, M. Nemec1, M. Schon1, L. Slobodova1, V. Tirpakova2, P. Krumpolec1, J. Cvecka3, P. Turčáni4, M. Sedliak3, P. Valkovič5, B. Ukropcová1,6;

1Dept. of Metabolic Disease, Biomedical Research Center, Bratislava, 2Institute of Physical Education, Slovak Health University, Bratislava, 3Faculty of Physical Education and Sports, Bratislava, 41st Department of Neurology, University Hospital Bratislava, Bratislava, 52nd Department of Neurology, University Hospital Bratislava, Bratislava, 6Inst of Pathophysiology, Medical Faculty, Comenius University, Bratislava, Slovakia.

Background and aims: Sedentary ageing accelerates decline of metabolic health and cognitive reserve, while regular exercise effectively supports metabolic and cognitive health in ageing population. Here we tested the hypothesis that carnosine, a dipeptide with the physiological function in skeletal muscle and brain, enhances the adaptive response to regular exercise in the elderly at the level of whole-body and muscle energy metabolism, physical fitness and cognitive performance.

Materials and methods: Glucose tolerance (oral glucose tolerance test), metabolic flexibility (ΔRQ) & insulin sensitivity (euglycemic hyperinsulinemic clamp/EHC & indirect calorimetry), resting energy expenditure & metabolic substrate preference (indirect calorimetry), body composition (bioelectrical impedance), physical fitness (Rockport 1 mile walk test, VO2max; sit-to-stand test) and cognitive functions (ACE-R, computerized CogState) were assessed. Bergström needle biopsy of m. vastus lateralis was performed. Oxygen consumption rate (pmol/s/mg tissue wet weight) was evaluated in saponin-permeabilized muscle fibers by O2k high-resolution respirometry (Oroboros). Sixty sedentary (BMI 27.1±3.9kg/m2) elderly volunteers (age 66.9±1.2yrs, MMSE score 28.2±1.2) were subjected to either 3-month supervised aerobic-strength training (3x1h/week) (n=36) or stretching exercise training (n=24, active controls). Half of the individuals in each intervention group were randomized to take oral carnosine (2g/day), or placebo. RM-2-way ANOVA and paired T-test was used.

Results: Training intervention combined with carnosine improved performance in sit-to-stand test (9.2%, p<0.05), metabolic flexibility (ΔRQ 7.3%, p<0.05) and aerobic fitness (VO2max 14.8%, p<0.05). Combination of exercise with carnosine has been more effective in stimulating coupled mitochondrial respiration rate in permeabilized muscle fibers, as compared to exercise alone (36.5%, p<0.01). Addenbrooke's cognitive examination (ACE-R) score correlated positively with lean body mass and muscle strength and the training-induced change in ACE-R memory sub-score correlated with concurrent change of lean body mass. Metabolic flexibility (capacity to increase RQ during EHC) correlated positively with short term memory and executive functions (CogState score).

Conclusion: We provide the evidence pointing towards synergistic effects of regular exercise and carnosine on systemic and muscle metabolism which has a potential to improve cognitive functions in overweight sedentary elderly individuals.

Clinical Trial Registration Number: NCT03330470

Supported by: VEGA 2/0107/18; SAS-NSC JRC 2018/10; ITMS 374 313011V344

Disclosure: J. Ukropec: None.

100

Resistance exercise training protects beta cell from apoptosis in an in vitro model of type 1 diabetes

G.A. Bronczek, G.M. Soares, E.M. Carneiro, A.C. Boschero, J.M. Costa-Júnior;

University of Campinas, Campinas - SP, Brazil.

Background and aims: Resistance exercise training exerts beneficial effects on glycemic control in type 1 diabetic (T1D) patients, which could be mediated by exercise-induced humoral factors released in the bloodstream. Here, we aimed to verify if resistance training would be able to protect beta-cell from apoptosis in an in vitro model of T1D by using the serum from resistance-trained mice.

Materials and methods: C57BL/6 mice were randomly assigned into two groups: control (CON) and resistance exercise training (RET). RET mice performed 1 training session/day, 5 days/week for 10 weeks. Each training session consisted of 8 climbs, carrying progressively heavier loads (50, 75, 90 and 100% of the animal’s maximal voluntary carrying capacity “MVCC”). We analyzed strength, body weight, perigonadal fat content and gastrocnemius weight (n=8). For in vitro experiments, we used a rat pancreatic beta-cell line (INS-1E) incubated with medium containing 10% of serum from control or trained mice for 24h; followed by exposure to 10 U/ml recombinant human Interleukin-1β (IL-1β) and 100 U/ml recombinant rat Interferon-γ (IFN-γ) (in vitro T1D), for 24h. The cells were used for Western Blotting analysis and apoptosis measurement by Hoechst-Propidium iodide (HO-PI) fluorescence quantification (n=2-4). Data were analyzed by Student´s t-test or One-Way ANOVA with an unpaired Tukey’s post-hoc test. Data are mean ± SEM, and the difference between the groups were considered statistically significant if P ≤ 0.05.

Results: As expected, after 10 weeks of training, RET mice presented higher MVCC, compared with CON mice (CON 42.83±1.75 g x RET 67.85±1.47 g). RET mice also presented reduced body weight when compared to CON (CON 30.88±1.19 g x RET 26.95±1.0 g), in addition to lower perigonadal fat pad and higher gastrocnemius weight (CON 890.2±95.98, 935.3±30.99 g x RET 521.7±33.53, 1028±23.55 g % body weight, respectively). The in vitro experiments demonstrated that in normal conditions there were no differences between treatment with control or trained serum in neither of the parameters analyzed. However, when cells were exposed to IL-1β plus IFN-γ, the increased iNOS and Cleaved caspase-3 protein content in INS-1E cells, pretreated with control serum, were partially blocked in INS-1E cells pretreated with trained serum (Serum CON 3.66±0.86 AU, 2.55±0.13 AU x Serum RET 1.98±0.45 AU, 1.34±0.16 AU, iNOS/α-Tubulin and Cleaved casp-3/α-Tubulin, respectively). We also measured the pro- and antiapoptotic proteins BAX and bcl-2, both related to the mitochondrial pathway of apoptosis, and there were no differences between groups regarding protein content (Serum CON 3.32±0.88 AU, 1.28±0.17 AU x Serum RET 2.18±0.69 AU, 1.35±0.17 AU, Bax/α-Tubulin and bcl-2/α-Tubulin, respectively) and BAX/bcl-2 ratio (BAX/bcl-2 ratio: Serum CON 1.98±0.45 AU x Serum RET 1.48±0.39 AU). In addition, we observed that the apoptosis rate in INS-1E cells, cultured in a medium containing trained serum was lower than in INS-1E cells, cultured in a medium containing control serum (% of dead cells: Serum CON 21.14±0.55 x Serum RET 11.18±0.02).

Conclusion: These results indicate that the serum from resistance-trained mice is able to protect beta-cell from cytokine-induced apoptosis. Suggesting that resistance exercise training could be an important strategy to protect beta-cell from apoptosis in the diabetes context.

Supported by: FAPESP 2018/15032-9 and 2015/12611-0

Disclosure: G.A. Bronczek: None.

101

High intensity interval training improves insulin sensitivity and affects mitochondria dynamics in skeletal muscle of type 2 diabetes humans

L. Mastrototaro1,2, M. Apostolopoulu2,3, D. Pesta1,2, K. Strassburger4,2, Y. Karusheva1,2, S. Gancheva2,3, J. Szendroedi2,3, M. Roden2,3;

1Institute for Clinical Diabetology, German Diabetes Center, Leibniz Center for Diabetes Research at Heinrich-Heine University Düsseldorf, Düsseldorf, 2German Center for Diabetes Research, Partner Düsseldorf, München-Neuherberg, 3Department of Endocrinology and Diabetology, Medical Faculty and University Hospital Düsseldorf, Heinrich-Heine University Düsseldorf, Düsseldorf, 4Institute for Biometrics and Epidemiology, German Diabetes Center, Leibniz Center for Diabetes Research at Heinrich-Heine University Düsseldorf, Düsseldorf, Germany.

Background and aims: High intensity interval training (HIIT) improves insulin sensitivity and oxidative capacity in humans with type 2 diabetes (T2D). Furthermore exercising induces mitochondrial biogenesis in healthy humans, but its effect on mitochondrial remodeling in insulin resistant states is not known. This study aims to examine the effect of HIIT on mitochondrial fusion and fission as well as on mitochondria respiration in the skeletal muscle of type 2 diabetes (T2D), insulin resistant (IR) and insulin sensitive (IS) glucose tolerant humans in order to elucidate the role of mitochondrial turnover in muscle insulin sensitivity.

Materials and methods: Twenty T2D, 11 glucose-tolerant IR and 12 IS humans, age- and BMI-matched, performed HIIT 3 days per week on a cycle ergometer (4 x 4 min intervals at 90% and 3 min of recovery at 70% of maximal heart rate for a total of 35 min) for 12 weeks. Before the intervention (baseline) and 72 h after the last exercise bout, whole-body insulin sensitivity was measured by hyperinsulinemic-euglycemic clamps, mitochondrial respiration and mitochondrial dynamics were assessed in skeletal muscle biopsies by high resolution respirometry and western blot, respectively. Finally skeletal muscle citrate synthase activity was assayed spectrophotometrically by a commercial kit.

Results: After 12 weeks of HIIT, all participants uniformly improved their cardiorespiratory fitness (p<0.001 vs baseline), whereas insulin sensitivity increased only in T2D and IR individuals (p<0.01 vs baseline). Moreover, muscle maximal oxygen uptake (p<0.001 vs baseline) as well as muscle citrate synthase activity (p<0.01 vs baseline) rose in all groups after HIIT, while biomarkers of mitochondrial fission (p<0.01), fusion (p<0.001) and mitophagy (p<0.001 for phospho-Parkin(Ser65) and p<0.05 for phospho-Pink(Thr257)) increased only in T2D.

Conclusion: In conclusion mitochondrial fusion and fission as well as mitophagy are critical to maintain a functional mitochondrial pool and likely contribute to the exercise training response of insulin sensitivity in overt T2D. These findings help to better understand the metabolic adaptation induced by lifestyle intervention and to identify novel targets for the tailored prevention and treatment of insulin resistance.

Clinical Trial Registration Number: NCT02039934

Disclosure: L. Mastrototaro: None.

102

TGFβ/mir143/145 associated mis-differentiation affects exercise response

S.I. Dreher1, S. Höckele2,3, C. Hoffmann1, P. Huypens2, A. Moller3,4, A.L. Birkenfeld3,4, A. Peter1,4, J. Beckers2,3, M. Hrabe de Angelis2,3, C. Weigert1,4;

1Institute for Clinical Chemistry and Pathobiochemistry, Department for Diagnostic Laboratory Medicine, University Hospital Tübingen, Tuebingen, 2Institute of Experimental Genetics, Helmholtz Zentrum München, Neuherberg, 3German Center for Diabetes Research (DZD), Neuherberg, 4Institute for Diabetes Research and Metabolic Diseases of the Helmholtz Zentrum München, University of Tübingen, Tuebingen, Germany.

Background and aims: Physical training is a promising strategy to prevent type 2 diabetes (T2D), improving insulin sensitivity. However, a substantial number of individuals lack a beneficial outcome in glycemic control. TGFβ was identified as a possible upstream regulator involved in this low-response. TGFβ is known to be a potent regulator of microRNA (miRs) expression. Aim of this study was therefore to elucidate the role of TGFβ-driven miRs on skeletal muscle differentiation and individual exercise response.

Materials and methods: By small and long RNA sequencing, we identified TGFβ-regulated transcripts in primary human myoblasts, afterwards fused to myotubes, utilizing TGFβ and SB431542. Identified miRs were investigated in human skeletal muscle biopsies obtained before and after a supervised 8 weeks-endurance training intervention, in relation to the change in insulin sensitivity (ISIMats) (n=40 consenting donors). Reported results were significant (p ≤ 0.05) according to appropriate statistical tests e.g. ANOVA Tukey.

Results: Enrichment analyses of lcRNAseq data corroborated the negative impact of TGFβ on myotube differentiation. Non-targeted miRNAseq analyses identified several miRs including miR143/708/181/31/145 upregulated and miR499/208/146/139/502 downregulated by TGFβ in myotubes (top 5). miR expression pattern during myotube differentiation revealed that TGFβ effects on miR499/208/206/146/139 could not be distinguished from its negative effect on differentiation. Specific regulation of miR143/145/181/31 by TGFβ was validated in myoblasts and differentiated myotubes. Human skeletal muscle biopsy donors were categorized as responder (RE) or low-responder (LRE) based on their fold change in ISIMats (±1.2). TGFβ signaling as well as miR143/145 expression were stronger induced by training in skeletal muscle of LRE donors compared to RE. Over all donors, changes in miR143/145 cluster expression correlated inversely with changes in ISIMats and clustered together with changes in MYH11ACTA2 and MYOCD. Global transcriptome, correlation and pathway enrichment analyses hinted towards a vascular smooth muscle cell association. Changes in miR143/145 correlated positively with changes in transcripts like MYH11ACTA2 and PPP1R14A. Target mining over all samples revealed several possible miR143/145 targets regulated in our system like HDACs and INSR.

Conclusion: Our results clearly associate TGFβ signaling and its regulation of miR143/145 with a lack of beneficial outcome in glycemic control as a response to training. Correlation analyses revealed a connection to smooth muscle or differentiating muscle cell markers suggesting incomplete or misguided muscle cell differentiation as a response to elevated TGFβ/miR143/145 in low-responders. Identification of the TGFβ/miR143/145 axis in skeletal muscle, and elucidation of involved targets of these miRs, has the potential to identify novel strategies to improve glycemic control in patients with risk for T2D.

Clinical Trial Registration Number: NCT03151590

Supported by: DZD e.V.;No. 01GI0925

Disclosure: S.I. Dreher: None.

OP 18 Stressed out beta cell organelles

103

Boosting the pancreatic beta cell function: the influence of nanotopographical cues on cell clustering and organelles crosstalk

A. Galli1, A. Marku1, N. Dule1, P. Marciani1, M. Castagna1, P. Milani2, C. Lenardi2, G. Tedeschi3, C. Perego1;

1Dept. of Pharmacological and Biomolecular Sciences, Università degli Studi di Milano, Milan, 2Dept. of Physics, Università degli Studi di Milano, Milan, 3Dept. of Veterinary Medicine, Università degli Studi di Milano, Lodi, Italy.

Background and aims: increasing efforts are focusing on the development of engineering microenvironments mirroring the biophysical properties of the islet niche to improve the β-cell differentiation and function in vitro. We have previously demonstrated that mouse and human β-cells sense and respond to the extracellular nanotopography by activating a mechanostransductive pathway, which mainly involved a reorganization of the actin cytoskeleton. Given that the cytoskeleton acts as a hard-wired structure that regulates the intercellular connectivity and the distribution of the organelles within the cytoplasm, aim of the proposed research was to evaluate whether the nanotopography could impact on the β-cell clustering, polarity and organelles organization, thus driving the β-cell fate.

Materials and methods: mouse β-cells and isolated human islets were cultured on metal oxide scaffolds with controlled roughness that mimic the extracellular nanotopography. Changes in the cellular proteome were evaluated by a shot-gun label-free proteomic approach and morphological studies were performed by means of super-resolution fluorescence microscopy (STED, TIRFM).

Results: immunofluorescence staining showed a reorganization of the cell-cell and the cell-substrate adhesion sites on the nanostructure, resulting in an increased intercellular connectivity that promotes the cell clustering. Interestingly, we also found a modification of cilia distribution and dimension in the cells grown on the nanostructure. Quantitative immunofluorescence studies revealed a more elaborated and complex mitochondrial network in the cells grown on the nanostructure, which supports the increased mitochondrial membrane potential and potentiates the glucose-stimulated insulin secretion. Furthermore, the proteomic analysis suggested that the nanotopography modulates the expression of proteins shared by the endoplasmic reticulum (ER) and mitochondria (CARL, VDAC1, HSPs) and by mitochondria and lysosomes (RAB7A and XSPA8). Accordingly, morphological studies confirmed a profound reorganization of the mitochondrial-ER and mithocondrial-lysosomes contact sites.

Conclusion: our results indicate that the extracellular nanotopography promotes the β-cell clustering and regulates the interplay of organelles within the cytoplasm, which are crucial for ensuring a rapid and coordinated cellular response to the extracellular stimuli. Understanding how to target mechanotransductive processes is of particular importance to successfully design scaffolds that can improve β-cell function and viability in vitro.

Disclosure: A. Galli: None.

104

Bcl-xl limits transcriptional and functional decompensation of beta cell mitochondria during chronic exposure to high glucose

D.J. Pasula1, R. Shi2, A.Z.L. Shih2, A. Chaudry2, B. Vanderkruk1, B.G. Hoffman1, D.S. Luciani1;

1Department of Surgery, Faculty of Medicine, University of British Columbia, BC Children's Hospital Research Institute, Vancouver, 2BC Children's Hospital Research Institute, Vancouver, Canada.

Background and aims: In the development of type 2 diabetes, β-cells respond to increasing metabolic demand with an early phase of functional compensation, followed by pathogenic decompensation and failure. Evidence suggests mitochondrial metabolism is enhanced in the adaptive response, while later failure involves changes to β-cell identity and mitochondrial dysfunction. It is not clear what mechanisms drive, and control, the progression from β-cell compensation to decompensation. We previously reported that anti-apoptotic Bcl-xL dampens β-cell mitochondrial metabolism. Here, we aimed to determine if the non-apoptotic functions of Bcl-xL limit β-cell decompensation under chronic glucose excess.

Materials and methods: Islets from Bcl-xflox/flox:Pdx1-CreERTM mice with β-cell knock-out of Bcl-xL (BclxβKO), and Bclxflox/flox wild-type controls (BclxβWT), were cultured 6-days in 11 mmol/l glucose (NG) or 25 mmol/l glucose (HG). Transcriptional profiling was done by RNA-Seq and mRNA expression was quantified by qPCR. We imaged cytosolic Ca2+ using Fura-2, and insulin secretion was assayed in static incubation. Oxygen consumption rate (OCR) was measured in an XFe24 Analyzer. Mitochondrial ROS (mitoROS) was imaged using MitoSOX and scavenged using MitoTEMPO. β-Cell mitochondria were labeled with TMRE and MitoTracker Green and imaged by 2D and 3D confocal microscopy. We quantified mitochondrial morphology and function using our “Mitochondria Analyzer” ImageJ plug-in.

Results: In both BclxβWT and BclxβKO islets, HG culture caused transcriptional re-wiring of metabolism toward glycolysis, a beginning loss of β-cell identity, and robust upregulation of adaptive ER stress responses. There was no significant activation of pro-apoptotic pathways or β-cell death. In HG, 233 genes were differentially expressed between BclxβWT and BclxβKO and these were dominated by mitochondria-related transcripts. This reflected a loss of key transcripts in BclxβKO islets, including Tfam, mitochondrial ribosomal genes, and numerous components of the electron transport chain such as Ndufb8, Sdhb, Cox4l1 and Atp5e. Total OCR, was significantly higher in HG-cultured BclxβKO islets than in BclxβWT, but ATP-coupled OCR was lower and glucose-stimulated mitochondrial hyperpolarization was impaired. Despite this, overall Ca2+ signaling and insulin secretion were amplified in HG-cultured BclxβKO islets. Confocal analysis indicated this was likely because of compensatory increases in mitochondrial fusion and total mitochondrial volume. HG culture increased mitoROS significantly more in BclxβKO cells than BclxβWT. Remarkably, normalizing mitoROS levels with MitoTempo rescued the glucose-induced defects in BclxβKO mitochondrial gene expression and the transcriptional changes to β-cell identity/function (Ins2GckMafaLdhaPdk1) in islets of both genotypes.

Conclusion: We show that mitoROS is a primary driver of transcriptional re-wiring in β-cells exposed to excess glucose, and identify Bcl-xL as a safeguard of the transcriptional and physiological integrity of β-cell mitochondria. Our findings suggest Bcl-xL is essential to prevent early β-cell decompensation during developing hyperglycemia.

Supported by: CIHR

Disclosure: D.J. Pasula: None.

105

The fate of intracellular sphingosine-1 phosphate regulates beta cell response to fatty acids

Y. Tang, E. Gurgul-Convey;

Institute of Clinical Biochemistry, Hannover Medical School, Hannover, Germany.

Background and aims: During type 2 diabetes (T2DM) development pancreatic beta-cells are exposed to lipotoxic stress, which results in beta-cell failure. The bioactive sphingolipid S1P (sphingosine-1 phosphate) has been linked to the development of T2DM. S1P is generated by two isoforms of sphingosine kinases with specific intracellular localizations. S1P can be catabolized either by S1P phosphatase (SPP) in the de-phosphorylation reaction yielding sphingosine, or can undergo an irreversible degradation to hexadecenal and phosphoethanolamine by the action of S1P lyase (SPL). The intracellular site of S1P generation was shown to differentially regulate the mode of lipotoxicity in beta-cells. Little is known, however, about the role of S1P catabolism in beta-cell sensitivity to fatty acids. Therefore the aim of this study was to compare the role of SPP and SPL in beta-cell sensitivity to lipotoxicity.

Materials and methods: Insulin-secreting INS1E cells were stably transfected either with an empty vector (INS1E-ctr), with human SPL (INS1E-SPL) or human SPP1 (INS1E-SPP) vectors. Cells were treated with 500 μM palmitate (PA), 500 μM oleate (OA) or a combination of both fatty acids (FFA) for 24 h. Thereafter cell viability was estimated by a MTT assay, oxidative stress by DCFDA oxidation, protein expression by Western blot and gene expression analyses were performed by qRT-PCR.

Results: The overexpression resulted in a 20-fold increase of the expression of SPL or SPP1. Exposure of INS1E-ctr cells to PA led to a significant decrease of cell viability, an effect that was potentiated by SPL overexpression (~25% stronger vs.INS1E-ctr, p<0.05). In contrast, SPP overexpression significantly protected against PA-mediated viability loss (only 20 % cell viability loss). OA was not toxic to INS1E-ctr cells and protected against PA-mediated cell viability loss (OA 91%, PA+OA 90% viability). Interestingly, in INS1E-SPL and in INS1E-SPP cells the incubation with OA led to a significant decrease of cell viability, and failed to prevent PA toxicity when co-incubated. These effects on cell viability correlated with the observed changes in the FFA-mediated oxidative stress induction and the expression of ER and mitochondrial stress markers.

Conclusion: Our results showed that the fate of intracellular S1P significantly participates in the regulation of beta-cell sensitivity to various FFA. SPL overexpression sensitized INS1E cells towards PA toxicity, while an increased capacity for S1P dephosphorylation by SPP overexpression provided protection against PA-mediated viability loss. An increased rate of intracellular S1P turnover induced toxicity of unsaturated fatty acid OA, which is the major toxic FFA in human beta-cells. Therefore the dynamic re-arrangements of beta-cell sphingolipid composition may impact beta-cell sensitivity to various FFA.

Disclosure: Y. Tang: None.

106

Manf overexpression in pancreatic beta cells protects from streptozotocin-induced beta cell death and diabetes in mice

H. Li1, T. Danilova1, E. Palm1, E. Hakonen2, T. Otonkoski2, M. Lindahl1;

1Institute of Biotechnology, HiLIFE, University of Helsinki, Helsinki, 2Research Programs Unit, Molecular Neurology, Biomedicum Stem Cell Center, University of Helsinki, Helsinki, Finland.

Background and aims: Type 1 Diabetes (T1D) is characterized by a progressive autoimmune destruction of pancreatic beta cells. Unresolved endoplasmic reticulum (ER) stress induced by local pro-inflammatory cytokines and chemokines contribute to the pancreatic beta cell death and insulin deficiency in T1D. Mesencephalic astrocyte-derived neurotrophic factor (MANF) is a small ER stress-regulating factor with cell protective and regenerative roles in various rodent disease models. We previously showed that conventional and pancreas-specific MANF knockout mice develop insulin-deficient diabetes due to postnatal loss of beta cell mass caused by reduced beta-cell proliferation and increased beta cell death preceded by sustained ER stress. Conditional removal of MANF specifically from adult beta cells revealed that endogenous MANF expression is mandatory for postnatal beta cell expansion and maintenance in mice. In human, loss of MANF results in insulin-deficient diabetes due to increased ER stress and defect in proinsulin processing and secretion. Exogenous MANF protein induces proliferation of human and mouse beta cells and protects them from ER stress-induced cell death triggered by chemical ER stressor and cytokines mimicking T1D in vitroIn vivo, AAV-MANF induced overexpression in pancreas protected beta cells from streptozotocin (STZ)-induced beta cell death. Thus, we wanted to test the therapeutic effect of induced transgenic MANF overexpression in two diabetes mouse models induced by STZ.

Materials and methods: We generated doxycycline inducible bi-transgenic mice that overexpressed MANF under the insulin promoter (INS-MANF). To study the effect of MANF overexpression in beta cells, mice were given doxycycline at different ages and different time periods and islet expression of beta cell markers, proliferation and beta cell mass were analyzed. We induced diabetes in the INS-MANF mice by injection of a high single dose of STZ and by injections of multiple low doses of STZ (MLDS). Beta cell mass, proliferation, apoptosis and insulitis scores were analyzed from the pancreases. Primary beta cells isolated from INS-MANF mice were challenged with STZ, and ER stressors to elucidate the mechanism of MANF protective action in beta cells in vitro.

Results: Increased MANF expression in beta cells conferred protection against beta cell apoptosis and hyperglycemia in both models of STZ-induced diabetes (at the end point p<0.01, one-way ANOVA followed by post hoc test). In addition, MANF-overexpression efficiently protected against islet lymphocyte infiltration and insulitis in the MLDS model of T1D. Furthermore, prolonged MANF overexpression for 3 months in pancreatic beta cells showed no change in pancreatic beta cell mass, serum insulin level and ER stress in beta cells.

Conclusion: We conclude that increased endogenous levels of MANF in beta cells is protective against diabetogenic insult and inflammation and induces beta cell proliferation in adult mice, thus constituting a promising therapeutic candidate for T1D.

Supported by: JDRF; Tekes; Finnish Diabetes Research Foundation; Academy of Finland

Disclosure: H. Li: None.

107

Loss of autophagy, Transcription Factor EB (TFEB) and lysosomal homeostasis limit beta cell function and survival under ER stress in vitro and in vivo

Y. Zou, D.J. Pasula, M. Tang, D.L. Dai, G. Soukhacheva, B.C. Verchere, D.S. Luciani;

Department of Surgery, University of British Columbia, Vancouver, Canada.

Background and aims: Autophagy can help protect stressed β-cells by delivering damaged organelles and protein aggregates to lysosomes for degradation. However, autophagic flux is impaired by severe lipotoxicity and inflammation. To examine if this may be a common feature in β-cells under ER stress, we investigated lysosomal homeostasis and the importance of autophagy in β-cells under various ER stress-inducing conditions, including hypoxia and islet transplantation.

Materials and methods: β-Cell autophagy-deficiency was induced by deletion of Autophagy Related 5 (Atg5), in vitro by adenovirus-Cre transduction of Atg5flox/flox islet cells, and in vivo by pancreatic duct injection of Atg5flox/flox with dsAAV6-RIP-Cre virus. Control mice were injected with dsAAV6-Empty vector. A marginal mass of 300 wild-type or Atg5 knock-out islets were transplanted under the kidney capsule of Atg5flox/flox mice that were made diabetic by 175 mg/kg streptozotocin (STZ). Glucose tolerance was assessed by i.p. glucose tolerance tests (IPGTTs). Autophagic flux was assessed in cultured islet cells from CAG-RFP-EGFP-LC3 reporter mice by confocal microscopy. Cell death was quantified by propidium iodide staining. Lysosomal function was determined using the Magic Red Cathepsin B activity assay. Tfeb protein levels were determined by immunofluorescence and western blot. Islet mRNA expression was quantified by qPCR. Human islets were obtained from the University of Alberta IsletCore.

Results: Atg5 knockout reduced β-cell survival under hypoxic stress in vitro and impaired the ability of syngeneic marginal mass islet grafts to maintain glucose homeostasis in STZ-diabetic mice, >80% of recipients of Atg5-deficient islets returned to diabetes within 10 days post-transplant (vs 30% of controls, p=0.05). Mouse islets cultured in 1% O2 activated pro-apoptotic ER stress pathways, and quantification of autophagosomes and autolysosomes in CAG-RFP-EGFP-LC3 islet cells revealed that hypoxia significantly disrupted autophagic flux. The impairment of autophagy was associated with reduced lysosomal cathepsin B activity and a significant loss of transcription factor EB (Tfeb), a master regulator of lysosomal physiology and biogenesis. A similar loss was seen in response to CoCl2-induced hypoxic stress. Notably, hypoxia and CoCl2 also reduced Tfeb in human islet cells. Treating hypoxic mouse islet cells with the mTOR-inhibitor Torin 1 restored Tfeb protein and improved lysosomal function. Lipotoxicity and thapsigargin-induced ER stress reduced Tfeb protein and viability in MIN6 cells, suggesting ER stress-specific mechanisms may drive dysregulation of Tfeb and lysosomes. Conversely, overexpression of Tfeb:GFP partially protected MIN6 cells from thapsigargin-induced death.

Conclusion: Autophagy protects β-cells under hypoxia and following islet transplantation but lysosomal clearance is impaired under prolonged hypoxic stress, possibly through Tfeb down-regulation. Our data further suggest that loss of Tfeb is a common feature of severe ER stress that may exacerbate β-cell failure and death.

Supported by: JDRF

Disclosure: Y. Zou: None.

108

Cell biology of stress granules in pancreatic beta cells

E. Quezada, J. Vasiljevic, M. Solimena;

Paul Langerhans Institute Dresden (PLID), Dresden, Germany.

Background and aims: Pancreatic islet β-cells maintain glucose homeostasis by secreting insulin. β-cells quickly enhance the translation of Insulin mRNA and related transcripts upon glucose stimulation (hyperglycemia). As the levels of these transcripts do not immediately increase after glucose stimulation, their rapid translational increment is mainly regulated by shared post-transcriptional mechanisms. Having a half-life of > 24 hours, the insulin mRNA is likely stored within the cytoplasm. We found that in mouse MIN6-K8 insulinoma cells kept at rest with 2.8 mM of glucose, Insulin1 mRNA is colocalized with the stress granule marker G3BP1. Stress granules are membraneless compartments associated with the storage of mRNAs and part of the translation machinery to halt protein synthesis. Hence, β-cells store Ins1 mRNAs and conceivably other related mRNAs in stress granules until glucose stimulation prompts the recruitment of these mRNAs to the ER to upregulate the biosynthesis of preproinsulin and other insulin secretory granule proteins. Accordingly, glucose stimulation of MIN6 cells correlates with the disappearance of stress granules and the redistribution of G3BP1 and insulin mRNA throughout the cytosol. However, how critical are these insulin mRNA stores, and more in general, the dynamic physiology of stress granules for β-cell function remains unclear. To address these questions, we are investigating the biology of stress granules in wild type and G3BP1-/- MIN6-K8 insulinoma cells.

Materials and methods: In vitro cultures of insulinoma MIN6-K8 cells treated with 2.8 mM glucose (resting) or 25 mM glucose (stimulated) were used as a model of glucose stimulation to β-cells. CRISPR-Cas9 editing was used to generate the G3BP1-/- MIN6-K8 cells. Western blots and immunostainings were performed for detecting G3BP1, phosphor-G3BP1 and insulin levels in wild type and G3BP1-/- MIN6-K8 cells. Transcripts levels were measured by qPCR.

Results: We found that G3BP1-/- MIN6-K8 cells lack stress granules, while their insulin content is increased. Moreover, in G3BP1-/- MIN6-K8 cells Ins1 mRNA levels are reduced compared to control cells. This data suggest that G3BP1 stabilizes Ins1 mRNA while suppressing its translation. In other cell types phosphorylation of G3BP1 on S149 and S232 regulates stress granule assembly and disassembly. However, the levels of phospho-G3BP1S149/S232 in resting and glucose stimulated MIN6-K8 were comparable, pointing to the possible existence of other post-translational modifications. Importantly, we detected G3BP1+ granular structures also in the cytosol of human pancreatic β-cells of a non-diabetic living donor, supporting the physiological roles of these cytosolic organelles. Furthermore, RNAseq analysis revealed that in laser-captured microdissected islets of metabolically phenotyped living donors with type 2 diabetes (T2D) the levels of G3BP1 mRNA are reduced compared to equivalent islets from normoglycemic donors. Hence, our data point to the first time to a link between G3BP1, stress granules and T2D.

Conclusion: In conclusion, this project aims to establish whether stress granules are physiological structures for mRNA storage in β-cells and whether their function is affected in T2D.

Supported by: DZD, IMI-RHAPSODY, IMI-INNODIA, DFG-Beta Stress

Disclosure: E. Quezada: None.

OP 19 Diet and nutrition

109

Reduced carbohydrate and increased protein and fat during weight loss improve the atherogenic lipid profile in type 2 diabetes

M.N. Thomsen1, M.J. Skytte1, A. Samkani1, A. Astrup2, J. Frystyk3, E. Chabanova4, B. Hartmann5, J.J. Holst5, T.M. Larsen2, S. Madsbad6, F. Magkos2, H.S. Thomsen4, R.L. Walzem7, T. Krarup1,2, S.B. Haugaard1;

1Department of Endocrinology, Copenhagen University Hospital Bispebjerg, Copenhagen, Denmark, 2Department of Nutrition, Exercise and Sports, University of Copenhagen, Copenhagen, Denmark, 3Department of Endocrinology, Odense University Hospital, Odense, Denmark, 4Department of Radiology, Copenhagen University Hospital Herlev, Copenhagen, Denmark, 5Department of Biomedical Sciences and NNF Center for Basic Metabolic Research, University of Copenhagen, Copenhagen, Denmark, 6Department of Endocrinology, Copenhagen University Hospital Hvidovre, Copenhagen, Denmark, 7Graduate Faculty of Nutrition, Texas A&M University, College Station, USA.

Background and aims: Elevated triglyceride-rich lipoproteins (TRL), excess small dense LDL particles (LDL5) and decreased HDL2/HDL3 ratio promote atherogenesis in type 2 diabetes (T2D). Carbohydrate restriction reduced intrahepatic triglyceride (IHTG) content beyond the positive effect of weight loss in a group of T2D patients, the present study sought to determine whether parallel improvements in lipoprotein density profiles occurred in these same patients.

Materials and methods: Seventy-two adult T2D patients with a mean±SD BMI of 33±5 kg/m2 were randomised 1:1 to 6 weeks of fully-provided hypocaloric dietary treatment aimed at ~6% weight loss, either with a carbohydrate-reduced high-protein (CRHP, C30E%/P30E%/F40E%) diet or a conventional diabetes (CD, C50E%/P17E%/F33E%) diet. Density profiles of lipoproteins were determined by ultracentrifugation of fluorescently labelled plasma. Magnetic resonance spectroscopy was used to assess IHTG. Treatment effects were evaluated using a constrained linear mixed model with inherent baseline adjustment.

Results: Body weight decreased by 5.8 kg (~6%) in both groups. Compared with the CD diet, the CRHP diet reduced TRL (mean [95% CI]) by -16 [-30;1]% (p=0.07) and LDL5 by -13 [-22;-3]% (p=0.01), and increased HDL2/HDL3 by 11 [1;22]% (p=0.04). The CRHP diet reduced IHTG more than the CD diet (-26 [-45;0]%, p=0.05), and changes in IHTG including both groups correlated significantly with changes in TRL and LDL5 (Spearman’s ρ 0.39 and 0.38, p<0.01).

Conclusion: Carbohydrate restriction adds to the positive effect of weight loss in T2D patients by inducing greater improvements in atherogenic lipid profile, maybe facilitated by a reduction in intrahepatic fat.

figureal

Clinical Trial Registration Number: NCT03814694

Supported by: Arla Foods amba and the Danish Dairy Research Foundation

Disclosure: M.N. Thomsen: None.

110

Intra-organ fat content during weight loss-induced remission of type 2 diabetes in people with normal or raised BMI

R. Taylor1, A.C. Barnes1, K.M. Irvine1, T.L. Kelly1, K.G. Hollingsworth1, N. Sattar2, M.E.J. Lean3, A. Al-Mrabeh1;

1Institute of Translational & Clinical Research, Newcastle University, Newcastle upon Tyne, 2Institute of Health and Wellbeing, Glasgow University, Glasgow, 3Human Nutrition, Glasgow University, Glasgow, UK.

Background and aims: The aetiology of type 2 diabetes is often assumed to be different in people who are overweight or obese compared with those with BMI in the ‘normal’ range. However, clinical observation suggested that losing body fat from above a ‘personal fat threshold’ at any BMI level can bring about remission of diabetes. We therefore compared baseline liver and pancreas fat content in people with type 2 diabetes and BMI above or below 27kg/m2 and examined changes in intra-organ fat after weight loss in relation to remission of diabetes in each group.

Materials and methods: We examined data from two cohorts (DiRECT: BMI 35.1±4.5 kg/m2; n=56, age 53.3±7.6 years, diabetes duration 3.0±1.7years; ReTUNE: BMI 24.3±2.0kg/m2, n=17, age 58.2±7.3 years, diabetes duration 2.6±2.2 years). Age, sex and BMI-matched control groups without diabetes were also studied (BMI>27kg/m2: n=18, age 55.4±6.0 years; BMI<27kg/m2: n=11, age 58.8±9.5 years). Weight loss was achieved using a low calorie diet (~800 kcal/day) followed by food reintroduction and weight maintenance. In those with BMI<27kg/m2, other types of diabetes identified by antibody and genetic tests (n=2: 1 T1DM, 1 MODY) were excluded from analysis. Intra-organ fat was quantified by 3-point Dixon magnetic resonance.

Results: In each BMI category, liver and intrapancreatic fat content were higher in diabetes compared with non-diabetic controls. For BMI>27kg/m2: liver fat was 16±1.3 vs. 5.5±1.4% p<0.0001 for type 2 diabetes vs. control respectively, and intrapancreatic fat 8.5±0.3 vs. 6.8±0.5% p<0.010, respectively. For BMI<27kg/m2: liver fat was 4.7±0.8 vs. 1.9±0.3% p=0.016 for type 2 diabetes vs. control, and intrapancreatic fat 5.0±0.3 vs. 3.4±1.1%, p=0.029, respectively. Both diabetic and control groups with BMI >27kg/m2 had significantly higher liver and intrapancreatic fat than those with <27kg/m2 (p<0.0001). After weight loss, liver fat content fell significantly in both groups (BMI>27kg/m2: to 3.0 ±0.5%, p<0.0001; BMI<27: to 1.4±0.1%, p=0.004). Similarly, intrapancreatic fat decreased (BMI>27- to 7.6±0.3%, p<0.0001; BMI <27 to 4.5±0.6%, p=0.026). Diabetes remissions were similar between BMI groups (60% and 67%).

Conclusion: Intra-organ fat content is elevated in people with type 2 diabetes at any level of BMI. Irrespective of BMI, 10-15% weight loss decreased liver and intrapancreatic fat with notably similar rates of remission of type 2 diabetes, consistent with the twin cycle hypothesis. Lesser elevation of liver fat, below the diagnostic threshold for NAFLD, can cause metabolic problems in leaner individuals, indicating the need to develop BMI-specific normal ranges. Across the range of BMIs, early type 2 diabetes has the same reversible pathophysiology.

Clinical Trial Registration Number: ISRCTN15177113

Supported by: We are grateful to Diabetes UK for grant funding

Disclosure: R. Taylor: Grants; Supported by grants from Diabetes UK. Lecture/other fees; academic lectures funded by Novo Nordisk, Lilly and Janssen. Non-financial support; RT is author of book 'Life Without Diabetes'.

111

Habitual intake of dietary methylglyoxal is associated with less low-grade inflammation, but also with impaired retinal microvascular function: The Maastricht Study

K. Maasen1, S.J.P. Eussen2, P.C. Dagnelie2, A.J.H. Houben1, C.A.B. Webers2, M.T. Schram1, T.T.J. Berendschot2, C.D.A. Stehouwer1, A. Opperhuizen3, M.M.J. van Greevenbroek1, C.G. Schalkwijk1;

1Department of Internal Medicine, School for cardiovascular diseases (CARIM), Maastricht University, Maastricht, 2CARIM School for Cardiovascular Diseases, Maastricht University, Maastricht, 3NUTRIM School of Nutrition and Translational Research in Metabolism, Maastricht University, Maastricht, Netherlands.

Background and aims: Dicarbonyls are highly reactive compounds and major precursors of advanced glycation endproducts (AGEs). Dicarbonyls are formed endogenously but also during food processing. Circulating dicarbonyls and AGEs are associated with inflammation and microvascular complications of diabetes, but the associations are currently unknown for dietary dicarbonyls. This study examined the associations of dietary dicarbonyl intake with low-grade inflammation and microvascular function.

Materials and methods: In 2793 participants of the Maastricht Study (60±8 yrs, 50% men, 26% T2DM), we estimated habitual intake of the dicarbonyls methylglyoxal (MGO), glyoxal (GO) and 3-deoxyglucosone (3-DG). Food Frequency Questionnaires were linked to our food composition database, including MGO, GO and 3-DG concentrations in >200 foods. Low-grade inflammation was determined as a z-score of plasma inflammation biomarkers (CRP, SAA, sICAM-1, IL-6, IL-8 and TNF-α). Microvascular function was determined in plasma as a z-score of endothelial function biomarkers (sVCAM, sICAM, eSelectin and vWF), in arterioles and venules of the retina as flicker light-induced dilation and diameters, and in skin as heat-induced skin hyperemic response. Cross-sectional associations of dietary dicarbonyls with markers of low-grade inflammation and microvascular function were investigated using linear regression adjusting for age, sex, glucose metabolism status (GMS), BMI, energy intake, smoking, alcohol, physical activity, education, triglycerides, systolic blood pressure, total cholesterol/HDL ratio and medication.

Results: Higher intake of MGO was associated with a lower z-score for inflammation, which was significant in the fully adjusted model (age, sex and GMS adjusted: -0.03 [-0.06;0.01]. Fully adjusted: -0.05 [-0.10;-0.01]). Intake of MGO was inversely associated with each individual inflammation biomarker, with strongest and significant associations for CRP and TNF-α. On the other hand, higher MGO intake was associated with impaired retinal venular dilation in the age, sex, and GMS adjusted model (-0.05 [-0.10;-0.004]), and remained so after full adjustment (-0.08 [-0.14;-0.02]). MGO intake was also associated with impaired retinal arteriolar dilation and skin hyperemia, although not significantly. We observed no association of dietary MGO with retinal diameters and plasma biomarkers of endothelial function. GO and 3-DG intake were not associated with any of the outcomes. The associations did not differ between individuals with NGM, prediabetes or T2DM (P int.>0.10).

Conclusion: Higher habitual intake of MGO was associated with less low-grade inflammation, but also with impaired retinal venular dilation. This suggests that food-derived MGO may induce anti-inflammatory effects, but also contributes to impaired retinal microvascular function, possibly via inflammation independent pathways. These associations in opposite directions require further investigation.

Supported by: NVWA, JPI HDHL, ERA-NET, ERDP, PoL, DMoEA, SDW, PSID, CVC, CARIM, CAPHRI, NUTRIM, SA, HFL, P, IS, JCBV, NNF, SAN

Disclosure: K. Maasen: None.

112

Very-low dose pre-meal whey protein microgels reduce postprandial glucose in type 2 diabetes: a randomised, placebo-controlled crossover study

O. Johansen1, I.J. Neeland2, R.L. Zagury3, B. Ahrén4, J. Neutel5, E. Perrin6, K. Reyes7, E. Berk8, M. von Eynatten1, L.H. de Gregório9;

1Global Clinical Development Cardiometabolism, Nestlé Health Science, Vevey, Switzerland, 2Case Western Reserve University School of Medicine, University Hospitals Cleveland Medical Center, Cleveland, USA, 3LPH (Human Performance Lab), Rio de Janeiro, Brazil, 4Lund University, Lund, Sweden, 5Orange County Research Center, Tustin, USA, 6Socar Research SA, Nyon, Switzerland, 7Nestlé Health Science, Bridgewater, USA, 8Metabolic health, Nestlé Health Science, Bridgewater, USA, 9IBPClin – Instituto Brasil de Pesquisa Clinica, Rio de Janeiro, Brazil.

Background and aims: Whey protein (WP) is found in dairy products, and are rich in branched chain amino acids and bioactive peptides that stimulate secretion of incretin peptides and insulin. Its applicability in routine nutraceutical clinical use has however been limited by 1) requiring a relatively high dose with a high caloric content, and 2) the need to take it well in advance of a meal. New technologies allowing for a more rapid absorption could enable use of a lower WP-dose, as well as allow to take the WP closer to the meal.

Materials and methods: In this single-intervention crossover study in individuals with drug-naïve or metformin-treated type 2 diabetes, we studied the effects of 10g WP (40 kcal) prepared with novel technology to enhance absorption (micelle-technology [WPM]), or placebo (0 kcal), provided as a 125 mL shot 15 min ahead of a 250 g pizza meal (622 kcal [29.0 g protein, 22.6 g fat, 72.6 g carbohydrates]). Postprandial (PP) glucose response over 4 hours, and incretin response (intact glucagon-like peptide [GLP]-1, peptide-YY [PYY], glucose-dependent insulinotropic polypeptide [GIP]) over 2 hours were assessed in blood, and the difference between WPM and placebo were assessed by comparing incremental areas under the curve (iAUC) between the two interventions.

Results: In total 26 individuals (14 females, mean [standard deviation] age 62.0 [8.3] years, baseline HbA1c 58 [12] mmol/mol /7.5% [1.1], eGFR 96.6 [25.7] ml/min/1.73m2, BMI 29.2 [4.8] kg/m2) completed both sequences. The pre-meal WPM shot significantly altered the early PP glucose trajectory, reducing the 2h iAUC by 22% (mean [95% CI] difference iAUC-30min-120min WPM vs placebo -29.43 [-55.45, -3.40] mg/dLxh, p=0.0283). The iAUC-30min-180min was similar (-31.58 [-68.43, 5.26], p=0.0896). A 66% increase in GLP-1 iAUC-30min-120min was observed (4.80 [2.19, 7.40] pmol/Lxh, p=0.0009), while responses for both PYY and GIP were similar between WPM and placebo (respectively, 5.21 [-1.14, 11.56] pmol/Lxh, p=0.1035, and 8.52 [-15.27, 32.31] pmol/Lxh, p=0.4668).

Conclusion: In sum, 10g WPM significantly reduces the early glycaemic response and significantly augments the GLP-1 response to a mixed meal in subjects with type 2 diabetes. These results support its use as a convenient pre-meal shot to improve PP metabolic profile.

figuream

Clinical Trial Registration Number: NCT04639726

Supported by: Nestlé Health Science

Disclosure: O. Johansen: Employment/Consultancy; Nestlé Health Science.

113

Investigation of sex-dependent effects of a one-year low-carb- vs low-fat intervention in patients with high-risk prediabetes - a randomised controlled trial

S. Kabisch1, A. Hustig1, U. Dambeck2, M. Kemper2, C. Gerbracht2, C. Honsek2, J. Machann3, M.A. Osterhoff2, A.F.H. Pfeiffer1;

1Campus Benjamin Franklin, Charité University Hospital Berlin, Berlin, 2Clinical Nutrition, German Institute of Human Nutrition, Nuthetal, 3Diagnostische Radiologie, Universitätsklinikum Tübingen, Tübingen, Germany.

Background and aims: Current guidelines point out, that diets of different macronutrient composition (low-carb vs. low-fat) are equally recommendable for prevention and therapy of T2DM, when achieving sufficient nutrient quality. Low-fat diets were the gold standard for more than 50 years, today a Mediterranean-style low-carb diet - low in meat, rich in fish and plant oils - is considered at least comparable, if not superior in reducing cardiovascular risks. Underlying risk factors are differentially affecting women and men; thus, RCTs need to assess their diet-induced metabolic outcomes under consideration of sex as well. Our DiNA-P study (Diabetes Nutrition Algorithms in Prediabetes) compares individualized two-phase low-carb- and low-fat diets. We present here data of the one-year intervention period, split up by sex.

Materials and methods: 267 subjects with high-risk prediabetes (age 59±10 years, BMI 32,5 ± 6,2 kg/m²; 38 % male) were identified by screening oGTT, characterized by insulin secretion deficit and/or insulin-resistant fatty liver disease, and 1:1 randomized to either low-carb or low-fat diet. They underwent an initial hypocaloric (1200-1500 kcal/d) dietary intervention for three weeks (< 40 g of carbs vs. < 30 kcal% of fat), followed by a moderate maintenance phase (1500-1800 kcal; < 40 kcal% vs. < 30 kcal% of fat) of 49 weeks. Clinical follow-up tests included oGTTs, routine lab, anthropometric assessment as well as MRI and 1H-MRS for body fat and liver fat quantification. Statistical analysis was conducted stratified by sex, assessed interaction effects between diet and sex, and was adjusted for differences in weight change.

Results: In the first diet phase, women showed a superior outcome in the low-carb group with respect to body weight, fasting glucose, blood pressure, triglycerides and LDL/HDL ratio, while in men, low-carb was inferior regarding levels of uric acid and 2-hours glucose (trend-wise). Effects on fasting glucose were characterized by a significant sex-by-diet interaction.

In the second diet phase, a significantly stronger reduction of body fat content was seen in men, only, but no other differences between the diets or sex and no significant interaction were determined. Compliance to the diets was considerably lower during maintenance phase, with the low-carb group showing clearly better long-term adherence compared to low-fat.

Conclusion: Only short-term benefits of the investigated diets seemed to be affected by a relevant sex-specific superiority for low-carb (women) and low-fat (men). After one year, both diets achieved comparable outcomes despite higher compliance to the low-carb diet. A modulation of dietary adherence or molecular metabolic mechanisms by sex needs to be evaluated by further analyses and upcoming trials.

Clinical Trial Registration Number: NCT02609243

Supported by: General Grant DZD

Disclosure: S. Kabisch: Grants; German Center of Diabetes Research (DZD).

114

The effect of dietary carbohydrate restriction beyond weight loss on health-related quality of life and cognition

N.J. Jensen1, H.Z. Wodschow1, M.J. Skytte1, A. Samkani1, A. Astrup2, J. Frystyk3,4, B. Hartmann5,6, J.J. Holst5,6, T.M. Larsen2, S. Madsbad7, F. Magkos2, K.W. Miskowiak8,9, J. Rungby1,10, S.B. Haugaard1, M.N. Thomsen1;

1Department of Endocrinology, Copenhagen University Hospital Bispebjerg, Copenhagen, 2Department of Nutrition, Exercise and Sports, University of Copenhagen, Copenhagen, 3Department of Endocrinology, Odense University Hospital, Odense, 4Department of Clinical Medicine, Aarhus University, Aarhus, 5Department of Biomedical Sciences, University of Copenhagen, Copenhagen, 6The Novo Nordisk Foundation Centre for Basic Metabolic Research, University of Copenhagen, Copenhagen, 7Department of Endocrinology, Copenhagen University Hospital Hvidovre, Hvidovre, 8Department of Psychology, University of Copenhagen, Copenhagen, 9Psychiatric Centre Copenhagen, Rigshospitalet, Mental Health Services, Copenhagen, 10Copenhagen Center for Translational Research, Copenhagen University Hospital, Bispebjerg and Frederiksberg, Denmark.

Background and aims: Carbohydrate restriction is emerging as a viable treatment strategy in type 2 diabetes, but its effect on health-related quality of life and cognitive function during a weight loss remains largely unknown. We aimed to evaluate the effect of weight loss induced by a carbohydrate-reduced high-protein (CRHP) diet on self-reported physical and mental health and cognition in patients with type 2 diabetes.

Materials and methods: In this randomized parallel trial, 72 adults with type 2 diabetes and overweight or obesity (mean±SD, HbA1c: 7.4±0.7% and BMI: 33±5 kg/m2) were randomly assigned 1:1 to CRHP diet (C30E%/P30E%/F40E%) or conventional diabetes (CD: C50E%/P17E%/F33E%) diet for 6 weeks. The two diets were intended to induce similar weight losses (~6%). Physical and mental component summary (PCS and MCS) scores were assessed from the short-form 36 questionnaire (SF-36), and global cognition, verbal memory, psychomotor speed and executive function from a neuropsychological test battery. Treatment differences were estimated from constrained linear mixed models using baseline adjustment.

Results: Both groups achieved a 5.8 kg (~6%) weight loss and improved PCS (median (IQR), CD: 2.7 (1.1;4.2)% and CRHP: 2.1 (0.7;3.7)%; both p<0.001). In addition, CRHP diet improved MCS (1.8 (-0.7;5.7)%, p<0.01) and 6 out of 8 domains of the SF-36 compared with 2 out of 8 after the CD diet. Global and specific domains of cognition did not change within or between groups, but CD scored better on the symbol digit modality test (SDMT) of psychomotor speed compared to CRHP (p<0.01; table 1).

Conclusion: Weight loss improves self-reported physical health independently of diet composition, and carbohydrate restriction may further benefit mental health, without adversely affecting overall cognition.

figurean

Clinical Trial Registration Number: NCT03814694

Supported by: Arla Foods amba and the Danish Dairy Research Foundation

Disclosure: N.J. Jensen: None.

OP 20 Keeping the balance in islet secretion

115

Re-internalised Phogrin/IA-2beta is targeted to multigranular bodies devoted to degradation of young insulin secretory granules

I. Kalaidzidis1,2, K.-P. Knoch1,2, J.M. Torkko1,2, A. Sönmez1,2, K. Ganss1,2, Y. Kalaidzidis3, M. Solimena1,2;

1Paul Langerhans Institute Dresden of the Helmholtz Center Munich, Dresden, 2German Center for Diabetes Research (DZD e.V.), Neuherberg, 3Max Planck Institute for Molecular Cell Biology and Genetics, Dresden, Germany.

Background and aims: Phogrin is a transmembrane protein of insulin secretory granules (SGs) and a target of autoimmunity in type 1 diabetes. Although it belongs to the receptor protein-tyrosine phosphatase family, it recognizes inositol phospholipids PI3P and PI(4,5)P2 as substrates. Its role and life-cycle in beta cells remains unclear. Previous studies indicated that upon SG exocytosis phogrin is incorporated into the plasma membrane, then endocytosed and recycled either to the Golgi complex or immature SGs. Here we revisited its lifecycle by using a time- and space- dependent CLIP/SNAP labelling approaches and high-resolution 3D microscopy.

Materials and methods: For age- and space-dependent conditional labelling of phogrin in rat insulinoma INS-1 cells a CLIP tag was inserted in its extracellular region after second convertase cleavage site. Imaging of living INS-1 cells expressing CLIP-phogrin was performed with a Nikon Eclipse Ti microscope equipped with a Andor iXON 897 EMCCD camera, a Yokogawa CSU-X spinning disk and a Plan Apochromat Oil objective (100x/NA1.45).

Results: INS-1 cells were incubated for 72h after transient transfection with insulin-SNAP and CLIP-Phogrin, and then with blocking non-fluorescent CLIP and SNAP permeable substrates, before being stimulated with 25mM glucose while being labelled with permeable fluorescent CLIPTMR and SNAPSiR substrates. After a 2h-long chase, cells were fixed, stained with various antibodies and imaged in 3D. At this time point newly-synthesized CLIPTMR-phogrin+ and insulin-SNAPSiR+ had left the ER and Golgi and colocalized by 75% within compact vesicular objects. <2% of CLIPTMR-phogrin+ objects were TGN38+ (Golgi marker), while ~15%, ~6% and ~2% of them were either ICA69+ (immature SGs), or EEA1+ or APPL1+ respectively, with the two latter being early endosome markers. Hence, CLIP-phogrin is properly sorted to the SGs. A 19°C block-and-release chase for SG biogenesis revealed that CLIPTMR-phogrin leaves the ER/Golgi ~30 min before insulin-SNAPSiR+, possibly explaining their apparent non-complete colocalization. Surprisingly, after 2h-chase ~15% of CLIPTMR-phogrin+ and 8% of insulin-SNAPSiR+ objects were also LAMP2+ (late endosome/lysosomal marker). We used the impermeable substrate CLIPSurface-647 to selectively label CLIP-phogrin at the cell surface after 25mM glucose stimulation. Quantitative analysis of deconvoluted images revealed that re-internalized CLIPSurface-647-phogrin first colocalized with EEA1, while later ~80% of it was found in multigranular bodies (MGB) which included multiple newly-synthesized insulin-SNAPSiR+ cores. Concomitantly, the colocalization of CLIPSurface-647-phogrin with LAMP2 and cathepsin D increased ~2 fold with a peak after 6h-chase.

Conclusion: After SG secretion, phogrin is endocytosed, but it is neither directly targeted to degradation, nor recycled to Golgi or immature SG for re-usage. Instead, we hypothesise that phogrin is part of a mechanism coupling insulin secretion with degradation of a pool of young granules (crynophagy).

Supported by: BMBF-DZD, IMI-RHAPSODY, IMI-INNODIA

Disclosure: I. Kalaidzidis: Grants; BMBF-DZD, IMI-RHAPSODY, IMI-INNODIA.

116

The importance of islet δ-cell ATP-sensitive K + channel (K ATP ) function for somatostatin secretion and islet hormone balance

T.G. Hill1, Q. Zhang2, A.I. Tarasov2, L.J. Briant2, C. Guida2, R. Terron Exposito1, P. Rorsman2, F.M. Ashcroft1;

1DPAG, University of Oxford, Oxford, 2OCDEM, University of Oxford, Oxford, UK.

Background and aims: Previous scRNAseq and RNAseq studies identified the ATP-sensitive K+ (KATP) channels within the islet δ-cells to be the same (Kcnj11 for Kir6.2 and Abcc8 for SUR1) and expressed at similar levels to those found in the β- and α-cells. However, the specific nature of the δ-cell KATP channels in driving glucose-induced somatostatin secretion (GISS) is unclear. We investigated the role of the KATP channels on δ-cell function and somatostatin (Sst) secretion by introducing a conditional gain-of-function Kir6.2-V59M point mutation into the MgATP-interacting pore-forming Kir6.2 subunits. This equates to a mutation associated with human neonatal diabetes.

Materials and methods: Insertion of the Kir6.2-V59M gain-of-function mutation specifically within δ-cells was achieved by crossing hemizygous Sst-iCre+/- mice with ROSA26-STOPloxP-Kir6.2+/- animals (δV59M mice). Cre/loxP-mediated excision of the STOP codon and Kir6.2-V59M expression was confirmed by the expression of an eGFP reporter situated behind the Kir6.2V59M sequence using qPCR. Intraperitoneal glucose (2 g/kg) and insulin (0.75 IU/kg) tolerance tests were performed on adult (12-14 wks) male/female δV59M mice. Sst-iCre+/-, ROSA26-STOPloxP-Kir6.2+/-, and WT littermates served as controls. The electrophysiology of δV59M islet cells was investigated via whole cell patch clamp. Sst, insulin, and glucagon secretion from δV59M and control animals exposed to various glucose concentrations was assessed by in situ pancreas perfusion (ISPP) and in static islet incubations.

Results: eGFP mRNA expression was detected within δV59M but not littermate control islets (δV59M: 0.23 ± 0.08; controls: n.d, relative to β-actin, n = 3), showing successful expression of the Kir6.2-V59M mutation. Adult mice displayed no adverse phenotype, difference in weight gain or ad libitum-fed blood glucose at 4-14 wks of age compared to controls. Adult male δV59M mice demonstrated significantly impaired glucose tolerance (AUC glucose excursion δV59M vs controls; 1800 ± 79 vs 1546 ± 67, p < 0.01, n = 10) and minor insulin insensitivity, whereas adult female mice revealed no change in glucose tolerance or insulin sensitivity compared to controls. Both male and female δV59M mice showed elevated plasma insulin in response to i.p. glucose. Male and female δV59M δ-cells were hyperpolarised in the presence of elevated glucose consequently causing a reduction in action potential firing compared to control cells. δV59M islets exhibited impaired GISS (ISPP % increase from 1 mM to 20 mM G δV59M vs controls: 95.5 ± 19.4 vs 150.5 ± 12.6, n = 3, p < 0.05) causing a loss of glucose-induced glucagon suppression and amplified 1st and 2nd phase glucose-stimulated insulin secretion, when compared to control islets (ISPP AUC δV59M vs controls 1 mM G to 20 mM G; 1st phase, 519 ± 81 vs 304 ± 46, p < 0.01; 2nd phase, 341 ± 60 vs 138 ± 23, p < 0.01, n = 4).

Conclusion: These data indicate that constitutive opening of the KATP channels specifically within Sst-expressing cells causes an in vivo and islet phenotype comparable to the reported global Sst-/- mouse, supporting an influential role for KATP channels in driving normal islet δ-cell function and Sst secretion. As with the β- and α-cells, these findings highlight the likely translational loss of normal islet δ-cell function in neonatal diabetic patients carrying the Kir6.2-V59M mutation.

Supported by: Novo Nordisk-Oxford Postdoctoral Fellowship; Leona M. and Harry B. Helmsley Charitable Trust

Disclosure: T.G. Hill: None.

117

Islet beta cell synchrony and second phase activity are governed by α cells upon physiological nutrient mixtures

M. Raoux1, M. Jaffredo1, K. Leal Fischer1, J. Gaitan1, A. Pirog2, S. Renaud2, J. Lang1;

1CBMN, CNRS UMR 5248, Univ. Bordeaux, Pessac, 2IMS, CNRS UMR 5218, Bordeaux INP, Univ. Bordeaux, Talence, France.

Background and aims: Pancreatic β cells are central in nutrient homeostasis and diabetes. Dynamic influences of non-β cells on β cell activity between and during meals and their putative deregulations in type 2 diabetes remain poorly understood. It is also crucial to decipher these functional interactions for cell therapies based on surrogate islets. α cells represent 20-35% of islet cells and have been mainly considered as β cell counter-regulators. We addressed their influence on β cell function in vivo and ex vivo in pre- and post-prandial situations with glucose and amino-acid mixtures.

Materials and methods: A model of induced α cell ablation in mice, termed GluDTR, was used (males, 13-26 weeks). In vivo i.p. glucose tolerance tests (IPGTT) were performed with and without a physiological mix of 19 amino acids (AAM). Islets were subsequently isolated and cultured on multi-electrode arrays (MEAs) for extracellular electrophysiological recordings of slow potentials (SPs), the β cell-specific coupling signals. Levels of β cell activity and synchrony were both monitored as published through SP frequencies and amplitudes, respectively.

Results: GluDTR islets contained insulin levels similar to WT (14.2±1.3 and 12.9±1.8 ng/islet respectively; N=3-4 mice), but were depleted in glucagon (4±1 vs 234±4 pg/islet). Fasting glycaemia were similar in WT and GluDTR mice (7.3±0.6 vs 7.8±0.5 mM respectively, N=11-15) as well as IP glucose tolerance. Co-administrations of glucose and AAM significantly improved glucose tolerance in WT (AUC glycaemia: p<0.001 for glucose with vs without AAM) but not in GluDTR mice, and glycaemia during the second phase were higher in GluDTR than in WT (p<0.05). Under these conditions, plasma C-peptide levels increased more in WT than in GluDTR mice (2.2±0.2 vs 1.5±0.2 fold respectively, p<0.05, N=3-6). In isolated islets analysed on MEAs, stimulations with glucose only from 3 (G3) to 6 (G6) or 8.2 mM (G8.2) triggered similar biphasic SP patterns between WT and GluDTR islets (n=80-106 islets; N=3-4 mice). In the additional presence of AAM, differences appeared between WT and GluDTR. AAM increased basal β cell activity and synchrony at G3 in WT but not in GluDTR islets. At G6, AAM increased similarly β cell activity and synchrony in WT and GluDTR islets in the first phase (p<0.0001, n=98-106, N=3), whereas the second phase was strongly increased in WT (p<0.0001) but not in GluDTR islets. At G8.2, AAM did not change the first phase in WT and GluDTR islets. During the second phase, β cell activity was increased by AAM in both WT and GluDTR islets (p<0.0001 and p<0.001 respectively), whereas β cell synchrony was stable in WT but decreased in GluDTR islets (p<0.05, n=80-102, N=3-4). Finally, inhibiting the action of glucagon-related peptides on GLP-1 receptors with exendin(9-39) in WT islets produced alterations of SP patterns similar to those observed in GluDTR islets.

Conclusion: These in vivo and ex vivo data provide a new model of β cell activity and synchronization upon physiological nutrient mixtures in which α cells are pivotal. Amino acids act directly on β in the first phase but mainly through α cells at low glucose and during the second phase with a strong influence on β cell synchrony. This regulation of intra-islet β cell networks by α cells will be further explored with increased spatial resolution by high-density MEAs.

Supported by: French Ministry of Research, ANR-18-CE17-0005 Diablo, FEDER Diaglyc

Disclosure: M. Raoux: None.

118

Pancreatic alpha and beta cells are globally phase-locked

H. Ren1,2, Y. Li1, C. Han3, Y. Yu1, B. Shi1, X. Peng3, S. Wu1, X. Yang1, L. Chen3, C. Tang1;

1Center for Quantitative Biology, Peking University, Beijing, 2Peking-Tsinghua Center for Life Sciences, Peking University, Beijing, 3Institute of Molecular Medicine, Peking University, Beijing, China.

Background and aims: The Ca2+ modulated pulsatile secretion of glucagon and insulin by pancreatic α and β cells plays a key role in glucose metabolism and homeostasis. However, how different types of islet cells couple and coordinate via autocrine and paracrine interactions to produce various Ca2+ oscillation patterns are still elusive.

Materials and methods: We generated transgenic mice to label α and β cells and designed a microfluidic device to facilitate long-term recording of islet Ca2+ activity at the single-cell level and simultaneously identifying different cell types in live islet imaging. We used T-distributed stochastic neighbor embedding and negative matrix factorization for spatial-temporal image analysis. Mathemitical modeling is used to link paracrine interaction and oscillation modes.

Results: we developed a microfluidic chip realized the quantitive observation of a long-term Ca2+ signal and showed heterogeneous but intrinsic Ca2+ oscillation patterns of islets upon glucose stimulation. Spatial-temporal analysis using T-distributed stochastic neighbor embedding and negative matrix factorization independently revealed that an oscillatory islet was composed of two well-separated out-of-phase cells groups. Combining with the α-Gcamp-β-Rcamp transgenetic mice, it was found that the α and β cells were globally phase-locked to various phase delays, causing fast, slow or mixed oscillations. For each oscillation cycle, the waiting time of α cells was fixed to 20s. And the waiting time of β cells determined the oscillation period. Since isolated single β cells only show slow oscillation with a period of 5 min, we have not followed the mathematical model of simplifying islet into single β cell in the past. The αβ-coupling oscillator islet model proposed in this work recaptured the oscillation modes and the quantitative relationship between phase and period. It was found the islet oscillation mode was tuned by the coupling strength of α to β cell. And the the waiting time of β cells was determined by glucagon concentration.

Conclusion: Our study highlights the importance of cell-cell interaction to generate stable but tunable islet oscillation patterns. And islet serve as the minimal unit to determine different oscillation modes.

figureao

Supported by: CMST, NNSFC, NKRDPC

Disclosure: H. Ren: None.

119

Investigating the presence of proglucagon-derived peptides in human pancreas

T. Mezza1, N. Wewer Albrechtsen2, G. Di Giuseppe1, C. Cefalo1, S. Moffa1, F. Cinti1, U. Capece1, S. Menchi1, G. Quero1, S. Alfieri1, A. Giaccari1, J.J. Holst2;

1Fondazione Policlicnico Universitario A. Gemelli IRCCS, Rome, Italy, 2University of Copenhagen, Copenhagen, Denmark.

Background and aims: Emerging evidence suggest production of GLP-1 in pancreatic islets. Prohormone convertase 1/3 (PC1/3), the enzyme responsible for GLP-1 cleavage from the proglucagon precursor, has been detected in rodent glucagon-producing cells, especially under β-cell stress conditions. However, only few α-cells are thought to produce GLP-1 in non-stressed conditions. Here, we evaluate whether fully processed active GLP-1 and N-terminally extended GLP-1 can be detected in biopsies obtained at partial pancreatectomy. The patients were carefully characterized regarding glucose tolerance: normal (NGT), impaired glucose tolerance (IGT), diabetes (DM)).

Materials and methods: We enrolled n=33 individuals with no known history of type 2 diabetes (18F/15M, age 66.2±9.29 yrs., BMI 25.1±4.74 kg/m2) scheduled for partial pancreatectomy periampullary tumors. To detect differences in the glucose tolerance and insulin secretion among subjects enrolled we performed a preoperative 75 gr OGTT and subjects were classified into n=9 NGT, n=14 IGT and n=10 T2D. Pancreas biopsies were collected during surgery. In extracts of frozen specimens of pancreatic tissue we measured chromogranin A (CgA), GLP-1 1-36, GLP-1 7-36, GIP, glucagon, insulin, c-peptide. Tissue proglucagon-derived peptides levels were also adjusted for CgA and expressed as percentage of CgA levels. Tissue measurements were correlated with patients’ clinical parameters.

Results: In the entire cohort, extractable levels of intact GLP-1 was 10 times lower compared to GLP-1 1-36 levels (mean levels of pancreatic GLP-1 1-36: 8.14 ± 1.41 pmol/g vs. mean levels of intact GLP-1: 0.81 ± 0.13 pmol/g). Further, CgA levels correlated to levels of GLP-1 1-36 (r=0.47, p=0.02) and intact GLP-1 (r=0.41, p=0.02), indicating that the expression of proglucagon-derived peptides is directly linked to the amount of endocrine tissue available in the biopsies. When subjects were classified according glucose tolerance and proglucagon-derived peptides levels adjusted for CgA, we observed similar levels of glucagon, while GLP-1 1-36 and intact GLP1 (p<0.01) levels were increased in T2D subjects compared to IGT and NGT. Moreover, we observed that intact GLP-1 tissue levels were positively correlated to in-vivo 2h glucose levels during OGTT (r=0.5, p=0.01)

Conclusion: Our data revealed that GLP-1 detected in human pancreas primarily consist of biological inactive GLP-1 1-36, while expression of intact GLP-1 is very low. We furthermore demonstrated that levels of intact GLP-1 were significantly increased in subjects with increased 2-h glucose levels. Our findings suggest that poor glucose metabolism is linked to increased islet levels of GLP-1, but the functional implication of this is still uncertain.

Supported by: EFSD Future Leaders Mentorship Programme for Clinical Diabetologists 2017 supported by an unrestricted educational grant from AstraZeneca

Disclosure: T. Mezza: None.

120

Modulation of cholesterol homeostasis via pancreatic LDL receptor alteration: impact on beta cell secretory activity

A. Marku, L. Da Dalt, A. Galli, N. Dule, D. Norata, A.L. Catapano, C. Perego;

Dept. of Pharmacological and Biomolecular Sciences, Università degli Studi di Milano, Milan, Italy.

Background and aims: the accumulation of cholesterol in pancreatic islets has been associated with reduced glucose-induced insulin secretion and cellular toxicity pointing to the critical role of cholesterol metabolism in pancreas. Cholesterol influx in β-cells mainly results from LDL uptake through the low-density lipoprotein receptor (LDL-R), abundantly expressed in these cells. The proprotein convertase subtilisin/kexin Type 9 (PCSK9) has been identified as the main target of LDL-R in liver. Interestingly, increased LDL-R expression and cholesterol content were observed in the pancreas of Pcsk9-deficient mice, suggesting a potential role for the protein in this tissue. Aim of this work was to understand the contribution of selective pancreatic PCSK9 production on β-cell function and to investigate the mechanisms involved.

Materials and methods: Pdx1Cre+/Pcsk9LoxP/LoxP (endocrine pancreas-selective Pcsk9 knock-out mice) and Pdx1Cre-/Pcsk9LoxP/LoxP mice were generated and fed to a standard or chow diet for 20 weeks, and the metabolic phenotype was characterized. The expression of LDL-R, the cholesterol accumulation and the insulin secretion were evaluated in islets from Pcsk9 deficient and control mice and in βTC3-cell lines treated with different levels of recombinant PCSK9 protein.

Results: pancreas specific Pcsk9-KO mice, as expected, lack detectable PCSK9 expression in pancreatic islets and showed decreased glucose clearance. The phenotype associated with impaired insulin secretion as insulin levels following fast and refeeding experiments were significantly lower than in littermates. In line with these findings, the glucose-stimulated insulin secretion (GSIS) test performed ex vivo resulted significantly decreased only in isolated islets from Pdx1Cre+/Pcsk9LoxP/LoxP mice. The analysis of pancreatic sections from littermates showed that LDL-R is prevalently present in β cells and its expression is increased in pancreas derived from specific Pcsk9-KO mice. Western blot analysis on βTC3-cell lines incubated with the PCSK9 recombinant protein showed a downregulation of the LDL-R, decreased BODYPY uptake and changes in the lipidomic profile. Studies are in progress to understand the role of PCSK9/LDL-R/cholesterol axis on β-cell function.

Conclusion: pancreatic Pcsk9 deficiency results in increased expression of LDL-R in β cells, thus leading to increased accumulation of cholesterol which impacts glucose-stimulated insulin secretion, resulting in hyperglycaemia, and impaired glucose tolerance. Our findings identify a new player in regulating pancreatic lipid homeostasis and insulin secretion.

Disclosure: A. Marku: None.

OP 21 The adipocentric angle

121

Yes-associated protein 1 in adipocytes plays an important role in glucose homeostasis

D.J.J. Han1,2, R. Aslam2, T. Ojha2, D.A. Yuen1,2, C.T. Luk1,2;

1Institute of Medical Science, University of Toronto, Toronto, 2Endocrinology and Metabolism, Keenan Centre for Biomedical Science, St. Michael's Hospital, Unity Health Toronto, Toronto, Canada.

Background and aims: Adipose tissue hypertrophy in obesity is associated with glucose intolerance and insulin resistance underlying type 2 diabetes. Furthermore, diabetes and obesity are associated with increased fibrosis formation in adipose tissue and metabolic dysregulation. Yes-associated protein 1 (YAP) is a transcription cofactor that promotes fibrosis in the lung, liver, and kidney. However, the role of YAP in adipocytes and glucose homeostasis is unknown. Therefore, the aim of this study was to elucidate the role of adipocyte YAP in glucose metabolism in vivo.

Materials and methods: To assess the impact of metabolic stress conditions on adipocyte YAP levels, we placed 6-week-old C57BL/6 mice on high-fat diet for 12 weeks. To further investigate the role of adipocyte YAP in glucose homeostasis, we generated novel adipocyte-specific YAP knockout mice (AdipoqYap-/-) using an adiponectin-Cre loxP recombination system. Littermate mice (AdipoqYap+/+) were used as controls and mice were fed chow or high-fat diet for 12 weeks at 6 weeks of age. Body weight, fasting blood glucose, and GTT were measured prior to sacrificing the mice. Mice were randomly assigned to groups by the experimenter; no formal blinding was used.

Results: YAP protein was increased in inguinal and perigonadal white adipose tissue from mice fed a high-fat diet for 12 weeks. This suggests that YAP is upregulated in adipose tissue with metabolic dysfunction and could potentially play an important role in the setting of obesity and glucose intolerance. When fed a high-fat diet, AdipoqYap-/- mice had significantly lower fasting blood glucose and improved glucose tolerance on glucose tolerance testing compared to littermate controls (n=8-10 mice/group, p-value<0.05, Student’s t-test). AdipoqYap-/- mice on high-fat diet also showed significantly lower levels of genes associated with fibrosis in inguinal and perigonadal white adipose tissue compared to littermate controls (n=5-6 mice/group, p-value<0.001, Student’s t-test). Overall, this suggests that disruption of YAP in adipocytes improves glucose tolerance and protects mice from high-fat diet-induced adipose tissue fibrosis.

Conclusion: Together, these data indicate that YAP increases in adipose tissue with weight gain and disruption of YAP in adipocytes improves glucose tolerance; suggesting that adipocyte YAP plays an important role in modulating glucose homeostasis under metabolic stress conditions. Past studies have shown that YAP is an important regulator of fibrosis development in various tissues. Our data show that knocking out YAP in adipocytes decreases the expression of key genes involved in the development of fibrosis. This suggests that adipocyte YAP also regulates fibrosis signalling in adipose tissue under metabolic stress conditions. Overall, our study identifies a potential novel therapeutic target for the treatment of adipose tissue fibrosis and obesity-associated glucose intolerance.

Supported by: NSERC Discovery Grant, CIHR Project Grant, BBDC Gales Family Charitable Foundation Pilot and Feasibility Grant, HSRLCE & BI-LILLY New Investigator Award, Heart and Stroke Foundation of Canada National New Investigator Award, J.P. Bickell Foundation Grant

Disclosure: D.J.J. Han: None.

122

Aberrant overexpression of HOTAIR inhibits abdominal adipogenesis through the epigenetic remodelling of genome-wide DNA methylation and transcription

F.-C. Kuo1, Y.-C. Huang2,3, P.-Y. Chen2,4;

1Division of Endocrinology and Metabolism, Department of Internal Medicine, Tri-Service General Hospital, National Defense Medical Center, Taipei, 2Institute of Plant and Microbial Biology, Academia Sinica, Taipei, 3Bioinformatics Program, Institute of Information Science, Taiwan International Graduate Program, Academia Sinica, Taipei, 4Bioinformatics Program, Taiwan International Graduate Program, National Taiwan University, Taipei, Taiwan.

Background and aims: Abdominal adiposity is strongly associated with diabetic and cardiovascular comorbidities. The long non-coding RNA HOTAIR (HOX Transcript Antisense Intergenic RNA) is an important epigenetic regulator, that is lowly expressed in abdominal subcutaneous adipose tissue (SAT) compared to gluteal SAT. HOTAIR also locates under the genome-wide association HOXC13 locus for human fat distribution. Hence, we aim to examine the phenotypic effects of HOTAIR overexpression on abdominal adipogenesis and link HOTAIR-mediated DNA methylation and transcriptome changes to identify its downstream regulated genes and functional pathways.

Materials and methods: The expression level of HOTAIR was compared among different fat-depots collected from six healthy, five morbidly obese, and five uremic subjects, correlated with dual-energy x-ray absorptiometry (DXA) defined regional adiposity. The human immortalized preadipocyte was used to assess the phenotypic effects of HOTAIR overexpression on abdominal adipogenesis. The integrative analysis of reduced representation bisulfite sequencing (RRBS) and RNA-sequencing was performed to identify putative HOTAIR-regulated genes and the associated signaling pathways. HOTAIR-repressed genes were further validated using RNA/chromatin immunoprecipitation with real-time qPCR and correlated with human body fat distribution.

Results: We found that the expression of HOTAIR was high in gluteal SAT, and low in arm/abdominal SAT and visceral (omental) adipose tissue. It could be aberrantly increased in uremic arm SAT. Notably, the lower expression of HOTAIR was correlated with higher abdominal adiposity in morbidly obese subjects, whereas a higher expression of HOTAIR was found to correlate with lower arm adiposity in uremic patients. HOTAIR overexpression in human immortalized abdominal preadipocyte remarkably suppresses the in vitro adipogenesis. We further identified 10 HOTAIR-mediated genes showing strong changes of DNA methylation associated with gene expression during abdominal adipogenesis, suggesting potential epigenetic regulation. Two HOTAIR-repressed genes, particularly SLITRK4 and PITPNC1, were further highlighted and validated with real-time qPCR and RNA/chromatin immunoprecipitation. Both presented an obesity-driven fat-depot specific expression pattern positively correlated with the central body fat distribution.

Conclusion: Our study indicates that HOTAIR is an important regulator for abdominal adipogenesis via intricate DNA methylation likely to associate with transcriptional regulation of specific genes, such as SLITRK4 and PITPNC1.

Supported by: MOST 109-2314-B-016-033; MOST 110-2314-B-016 -002 -MY3; TSGH-C03-110024

Disclosure: F. Kuo: None.

123

The type 2 diabetes gene RREB1 plays a role in high-fat diet induced adipogenesis in mice

G.Z. Yu1, L. Bentley2, N.A.J. Krentz3, R. Casero2, Y. Bai2, K.K. Mattis1, S. Holliman2, J.M. Torres4, A. Mahajan1,4, A.L. Gloyn1,3, R.D. Cox2;

1OCDEM, University of Oxford, Oxford, UK, 2MRC Harwell, Oxford, UK, 3Paediatrics, Stanford School of Medicine, Stanford, USA, 4Wellcome Centre for Human Genetics, University of Oxford, Oxford, UK.

Background and aims: Genome wide association studies have identified multiple independent signals for type 2 diabetes risk, glycaemic traits and ectopic fat distribution at the zinc finger transcription factor Ras-responsive element binding protein 1 (RREB1) locus. RREB1 is expressed in multiple diabetes relevant tissues, including adipose, but the impact of RREB1 loss in vivo and the mechanisms underlying these metabolic phenotypes remain unknown. The overall aim of this study was to determine the role of RREB1 in adipose tissue and its contribution in ectopic fat distribution.

Materials and methods: Global heterozygous Rreb1tm1b(EUCOM)Wtsi knockout C57BL/6N mice (homozygotes are embryonic lethal) were generated and characterised from 4 to 26 weeks of age on a high fat (HFD) and a matched low-fat (LFD) diet. Body weight (g), fat and lean mass (g) were measured, the latter using Echo-MRI. Tissue isolated from the inguinal adipose tissue depot of Rreb1+/- mice were subjected to histological analysis of cell size. Lipid formation using Oil Red O staining was measured in precursor adipocytes isolated from inguinal adipose tissues and grown and differentiated in vitro. Intraperitoneal insulin tolerance tests (IPITT) were performed to determine insulin sensitivity in Rreb1+/- mice.

Results: Rreb1+/- male mice were significantly protected from weight (p<0.0001) and fat mass (p<0.0001, Figure 1) gain compared to wild-type (WT) litter mates. Fat mass remained significantly lower in males (8-14 weeks) when adjusted for body weight (p<0.0001). Whilst the weights of multiple visceral adipose depots were reduced on a HFD in females (gonadal adipose tissues, p<0.0001; mesenteric adipose tissue, p=0.003; n=46). Adipocytes isolated from the inguinal adipose tissue depot were smaller in size compared to wild-type mice (p=0.0002), indicating that hypertrophic and hyperplastic processes may be altered in these mice. Precursor adipocytes isolated from the inguinal adipose tissue were less capable of differentiating into mature adipocyte as measured by lipid accumulation (p=0.03, n=12). Finally, IPITT showed reduced plasma glucose levels within the first 15 minutes following insulin injection in Rreb1+/- male mice compared to WT mice (p< 0.0001; n=30).

Conclusion: Our results demonstrate that RREB1 is important in fat deposition and glucose homeostasis exhibiting sex, diet and age effects and provide evidence that RREB1 is critical in adipocyte differentiation and lipid storage.

figureap

Supported by: Wellcome Trust; MRC; NIH

Disclosure: G.Z. Yu: None.

124

Deletion of CD44 promotes adipogenesis and insulin signalling in adipocytes

X. Weng, S.M. Warburton, C.K. Hennayake, L. Kang;

University of Dundee, Division of System Medicine, Dundee, UK.

Background and aims: Growing evidence suggest a close link between adipose fibrosis, inflammation, and insulin resistance in obesity. Hyaluronan, one of the main components of the extracellular matrix is increased in adipose tissue of obese and diabetic mice. CD44, the main hyaluronan receptor is associated with Type 2 diabetes from expression-based genome-wide association studies and its expression level in adipose tissue is positively correlated with adipose inflammation and insulin resistance. This study is to determine the role of CD44 in adipose function and insulin resistance.

Materials and methods: Stable, CD44-deficient 3T3-L1 cells were generated by Crispr Cas 9 technology using guide RNAs targeting the exon 3 of cd44 gene. The CD44 knockout (KO) cells were confirmed by Western blot and site mutations were determined by biallelic sequencing. Cells that were transfected but maintained normal level of CD44 protein were used as Crispr wildtype (WT) controls. Differentiation of 3T3-L1 cells to adipocytes was included by a cocktail of isobutylmethylxanthine, insulin, and dexamethosone. Mouse primary adipocytes were derived from stromal cells of the subcutaneous adipose tissue in 10 week old C57BL/6 WT and global CD44 KO mice. Adipogenesis was measured by Oil Red O staining and insulin sensitivity was measured by phosphorylation of Akt. Insulin resistance was induced by treating the cells with 250μM palmitic acid for 24 hours.

Results: CD44 gene expression decreased by 81% after differentiation in WT 3T3-L1 cells (P<0.05). Deletion of CD44 in 3T3-L1 cells increased adipogenesis as assessed by Oil Red O staining (3.24±0.86 arbitrary units vs 2.52±0.67 in 3T3-L1 naïve cells and 1.98±0.86 in Crispr WT cells) (P<0.05). Gene expression of the adipogenic markers PPARɣ and CEBPα were also consistently increased in the CD44 KO cells when compared with the control cells. Upon insulin stimulation, knocking out CD44 enhanced phosphorylation of AKT at S473 in differentiated 3T3-L1 adipocytes (2.1±0.8-fold increase, P<0.05). Palmitate acid induced a blunted response of Akt phosphorylation and P38 dephosphorylation in 3T3-L1 WT adipocytes, which was reversed in CD44 KO 3T3-L1 adipocytes (P<0.01). Consistent with the results in 3T3-L1 cells, primary adipose stromal cells isolated from CD44 KO mice displayed an enhanced adipogenic capacity and increased phosphorylation of Akt after insulin stimulation compared to those from the WT mice.

Conclusion: Deletion of CD44 promoted adipogenesis and improved insulin signalling in vitro in 3T3-L1 cells and mouse primary adipocytes. This study extends our knowledge of the role of CD44 in regulating adipocyte function, representing a potential target for mitigating adipose dysfunction in metabolic disorders.

Supported by: Diabetes UK and China Scholarship Council

Disclosure: X. Weng: None.

125

The role of CDKN2C in the regulation of human adipocyte metabolism

M. Vranic1, S. Hetty1, P.G. Kamble1, E. Holbikova1, H. Jernow1, S. Skrtic2,3, M.K. Svensson4, J.W. Eriksson1, M.J. Pereira1;

1Department of Medical Sciences, Clinical Diabetology and Metabolism, Uppsala University, Uppsala, 2Innovation Strategies & External Liaison, Pharmaceutical Technologies & Development, AstraZeneca, Gothenburg, 3Institute of Medicine at Sahlgrenska Academy, University of Gothenburg, Gothenburg, 4Department of Medical Sciences, Renal Medicine, Uppsala University, Uppsala, Sweden.

Background and aims: CDKN2C (Cyclin-Dependent Kinase Inhibitor 2C)/p18 is a member of the INK4 family of cyclin-dependent kinase inhibitors that controls cell cycle progression and has been shown to be involved in murine adipocyte cell differentiation. In a genome-wide association study CDKN2C gene polymorphisms have been associated with insulin resistance phenotypes and impaired peripheral adipose tissue (AT) storage capacity. This study aims to explore the role of CDKN2C in human adipocyte metabolism.

Materials and methods: Abdominal subcutaneous AT (SAT) needle biopsies were collected from 20 metformin-treated subjects with T2D and 20 sex, age- and BMI- matched healthy controls (20F/20M, age 58 ± 9 vs 58 ± 11 years, BMI 30.8 ± 4.6 vs 30.7 ± 4.9 kg/m2). Glucose control (HbA1c) was 6.6 ± 0.8 vs 5.6 ± 0.3 % (mean ± SD). Expression levels of CDKN2C were measured and correlated with expression of genes involved in lipid storage and differentiation and clinical markers of insulin resistance and obesity. Body fat volumes were measured by magnetic resonance imaging. Furthermore, CDKN2C expression was measured in paired samples of SAT and visceral AT (VAT) (15F/8M). CDKN2C was knocked out in human primary preadipocytes using CRISPR/Cas9 gene editing to study the effect on adipocyte proliferation and differentiation rates. Glucose uptake and lipolysis were done in differentiated adipocytes.

Results: CDKN2C mRNA expression levels in SAT were 30% lower in the T2D group compared to the control (P<0.05). In the T2D group, CDKN2C expression in SAT was negatively correlated with FFA AUC during OGTT (P<0.01), markers of dysglycemia (HbA1c, glucose AUC during OGTT; all P<0.05), insulin resistance (HOMA-IR, P<0.05), and visceral obesity (waist-hip ratio (WHR), VAT volume; all P<0.01), whereas it was positively correlated with SAT/VAT ratio (P<0.01) and the expression of genes promoting lipid storage (e.g. FASNPPARGFABP4, all P<0.01). In the control group, CDKN2C expression was negatively correlated with WHR and VAT volume and positively with the SAT/VAT ratio (all P<0.05). The CDKN2C expression was also down-regulated by 30% in VAT compared to SAT when measured in paired samples (P<0.05), and it correlated negatively with BMI and HOMA-IR in both SAT and VAT. In CRISPR/Cas9 experiments, the loss of CDKN2C did not affect human preadipocyte proliferation or differentiation rates when compared to wild type cells. Interestingly, basal and insulin-stimulated glucose uptake rates were decreased up to 30% in the knockout cells compared to wild type but no effect was seen on lipolysis rate (basal, isoproterenol-stimulated, or insulin-inhibited).

Conclusion: T2D patients have lower CDKN2C expression in SAT compared to healthy controls, and this expression is associated with markers of insulin resistance and visceral adiposity. Furthermore, our knockout results support a role of CDKN2C in human adipocyte glucose metabolism but not in the regulation of human adipocyte differentiation. Our findings suggest that downregulation of CDKN2C in T2D might contribute to insulin resistance in adipose tissue.

Supported by: H2020 Marie Sklodowska Curie ITN TREATMENT; AstraZeneca R&D; SSMF; EXODIAB; UU ALF

Disclosure: M. Vranic: None.

126

Zmat3 hypomethylation contributes to early senescence of adipose precursor cells from healthy individuals with a family history of type 2 diabetes

R. Spinelli1, P. Florese1, L. Parrillo1, F. Zatterale1, M. Longo1, M. Pastorino1, A. Desiderio1, G.A. Raciti1, C. Miele1, B. Gustafson2, A. Nerstedt2, U. Smith2, F. Beguinot1;

1URT GDD of the IEOS-CNR & Department of Translational Medicine, Federico II University of Naples, Naples, Italy, 2Lundberg Laboratory for Diabetes Research, Department of Molecular and Clinical Medicine, Sahlgrenska Academy, University of Gothenburg, Gothenburg, Sweden.

Background and aims: Senescence of adipose precursor cells (APC) impairs de novo adipogenesis and contributes to the age-related subcutaneous adipose tissue (SAT) dysfunction increasing risk of type 2 diabetes (T2D). First-degree relatives (FDR) of T2D individuals feature restricted adipogenesis in SAT, reflecting detrimental effects of APC senescence earlier in life and rendering FDR more vulnerable to T2D. Recent evidence indicates that epigenetics may contribute to these abnormalities. Our previous methylome analysis in APC from FDR and subjects with no family history of diabetes (CTRL) identified ZMAT3 as one of the top-ranked senescence-related genes which feature hypomethylation in the FDR subjects and associates with T2D risk. In this study, we investigated whether and how DNA methylation changes at ZMAT3 promote early APC senescence.

Materials and methods: APC were obtained from healthy and non-obese FDR (n = 12) and CTRL (n = 12) subjects (Table 1). A panel of senescence markers were assayed. Gene expression was assessed by both qPCR and western blot analyses. DNA methylation was assessed by bisulfite sequencing and its impact on transcriptional activity was further verified by luciferase assay. Gain-of-function experiments were performed to establish the role for ZMAT3 in inducing APC senescence.

Results: Senescence-related marks, including the acquisition of a senescence-associated secretory phenotype, were observed in APC from the FDR subjects. In FDR APC, reduced DNA methylation at the ZMAT3 gene caused ZMAT3 upregulation and accompanied development of the senescent phenotype. Interestingly, demethylation of ZMAT3 in the APC from CTRL donors led to both increased ZMAT3 expression and premature senescence. In addition, APC overexpressing ZMAT3 exhibited senescence and activation of the p53/p21 pathway similar to that of the APC from FDR individuals. Adipogenic differentiation was also inhibited in APC overexpressing ZMAT3. Indeed, ZMAT3 expression was enhanced in the poorly differentiated FDR APC while remaining unchanged in normally differentiated CTRL APC. Finally, in the FDR subjects, senolytic clearance of senescent APC was accompanied by increased DNA methylation and decreased expression of ZMAT3 as well as by improved adipocyte differentiation.

Conclusion: Our findings indicate that DNA methylation induces ZMAT3 upregulation in the FDR APC accompanied by acquisition of a senescent phenotype and impaired adipogenic differentiation, likely contributing to FDR insulin resistance and propensity to develop T2D.

figureaq

Disclosure: R. Spinelli: None.

OP 22 Understanding kidney disease in diabetes

127

Meta-analysis of whole exome and whole genome sequencing data for diabetic nephropathy in individuals with type 1 diabetes

J. Haukka1,2, A. Antikainen1,2, E. Valo1,2, V. Harjutsalo3, C. Forsblom1,2, N. Sandholm1,2, P.-H. Groop1,2, on behalf of the FinnDiane Study Group;

1Folkhälsan Research Center, Helsinki, 2Department of Nephrology, University of Helsinki and Helsinki University Hospital, Helsinki, 3National Institute for Health and Welfare, Helsinki, Finland.

Background and aims: Diabetic nephropathy (DN) is a major risk factor for severe kidney failure, end-stage renal disease (ESRD) and cardiovascular diseases in individuals with diabetes. The genetic landscape behind this complication is still relatively unexplored, although recent studies have discovered a few causal common variants. To find novel rare variants for DN, we performed meta-analysis of whole exome (WES) and whole genome sequencing (WGS) studies in individuals with type 1 diabetes (T1D) and extreme phenotypes of diabetic nephropathy.

Materials and methods: A total of 1100 individuals with T1D were recruited from the Finnish Diabetic Nephropathy Study (FinnDiane) to undergo WES and WGS. Post-QC, 493 individuals in WES and 583 in WGS were included in the analyses. In both cohorts, the cases had developed either macroalbuminuria or ESRD, whereas the controls had retained normal albumin excretion rate for at least 26 years on WES cohort and 35 years on WGS cohort. Samples were sequenced using Illumina platforms (WES: HiSeq2000 with >80% exome coverage, >20x capture, WGS: HiSeqX >30x capture). Variants were called jointly using Broad Institute’s best practices guidelines for Genome Analysis Toolkit, and they were filtered with 98% call rate and 10×10-50 HWE. Single variant meta-analysis was performed using rvtest and METAL, and we consider P-values < 1×10−6 as exome-wide significant. The gene-based SKAT-O meta-analysis was run with MetaSKAT. The SKAT-O analyses were performed for protein altering variants (PAV) and protein truncating variants (PTV) separately, and we consider P-value of 2.5 × 10−6 as gene-based test significance threshold.

Results: No variant or gene reached genome-wide significance after adjustments for sex, age at T1D onset and two first genetic principal components. The most significant SNP in single variant analysis was a synonymous variant rs10979729 in EPB41L4B p=6.13×10-6. However, it is in high LD (r2=0.94 in Finnish) with a missense variant rs2230794 (Ile830Met) in ELP1 with the 9th most significant P-value of 5.680 ×10-5ELP1 has been associated with kidney abnormality and renal insufficiency, offering a more plausible explanation for the region’s association with DN. The top gene in SKAT-O meta-analysis yielded a P-value 3.55×10-5 for NOM1, which includes 10 PAVs with MAF<0.10 across the two datasets. NOM1 was also the top gene associated to diabetic kidney disease in a common variant gene burden analysis from the type 2 diabetes knowledge portal.

Conclusion: Both the WES-WGS single variant and SKAT-O meta-analyses offer novel risk loci and plausible genes for DN, such as NOM1. However, due to the loss of statistical significance after Bonferroni correction, the results need further validation.

Disclosure: J. Haukka: None.

128

Long non-coding RNA from cells derived from urine in biopsy confirmed kidney disease with diabetes to differentiate diabetic kidney disease from non-diabetic kidney disease

S. Ghosh1, M. Basu1, A. Raychaudhuri2, N.P. Bhattacharyya1;

1IPGME & R CALCUTTA, Kolkata, 2Nephrology, IPGME & R CALCUTTA, Kolkata, India.

Background and aims: Renal involvement in Type 2 diabetes can be those due to diabetes per se (Diabetic Kidney Disease) or causes other than diabetes (Non-diabetic Kidney Disease NDKD). Currently available clinical, biochemical and radiological markers fail to differentiate the two accurately and renal biopsy remains gold standard for correct diagnosis. In animal model in those with diabetes and kidney disease it is observed that there is deregulation of several lncRNA. However, this has not be explored in humans with biopsy proven kidney disease in those with Type 2 diabetes. Aims: To determine expression of lncRNA in cells derived from urine (which mainly originate from the kidney) and to evaluate if this could be used as non-invasive markers to differentiateDKD from NDKD.

Materials and methods: We recruited consecutive patient with renal involvement (eGFR >30ml/min/m2 to <60ml/min/m2 and/or ACR>300mg/mcg or proteinuria >500 mg/24 hr. Patients without any contradiction for biopsy were subjected to biopsy.Histopathological classification was done as perInternational Society of Nephrology and the Renal Pathology Society (ISN/RPS) Classification Second morning urine sample in fasting state was collected for analysis.From 30 ml of freshly collected urine samples, RNA was isolated from urinary cell debris by commercially available kit. Concentration and purity of RNA was determined in spectrophotometer. Differential expression of 10 Long-Non coding RNAs (chosen from list of lncRNA found to be deregulated in previous animal model studies) were determined by Quantitative Real time PCR with respect to internal control. The fold changes were calculated using SDS software.Mann-Whitney U andReceiver operating characteristic (ROC) curves) were performed.

Results: 94 patients were included of whom 60(63.8%) were DKD, 19 (20.2%) were NDKD and 15 (15.9%) had both DKD &NDKD.We included only those with pure DKD or NDKD for comparison of expression of long non-coding RNA. Amongst the panel we found difference in level of expression of MALAT1(p<0.005), NEAT(p<0.05) and PVT1 (p<0.05).Amongst them MALAT1 had the best utility in discriminating DKD from NDKD(AUCs of 0.73 (95% CI: 0.5-0.88), p<0.005)).

Conclusion: Long-noncoding RNA expression in cells derived from urine may help distinguish DKD from NDKD. The result need validation in larger cohort.

Supported by: DST WB, RSSDI

Disclosure: S. Ghosh: Grants; Dept of Science & Technology, West Bengal, Research Society for the Study of Diabetes in India.

129

Novel cross-species transcriptional networks, effective genes and signalling pathways of diabetic nephropathy in human and mouse kidney

B.A. Bhat, T. Habib, I. Ahmed, S.S.P. Jeya, K.A. Fakhro, A.A. Akil;

Human Genetics, Sidra Medicine, Doha, Qatar.

Background and aims: Diabetic nephropathy (DN) or diabetic kidney disease is a serious kidney-related complication of chronic type 1 and type 2 diabetes. Mouse models are effective approach in describing the pathophysiology of (DN), however, it is incompletely summarizing disease presentations of human (DN). To describe the molecular similarities and differences between human and mouse (DN) and to maximize the potential of utilizing these models, we performed a cross-species comparison of kidney tissue transcriptional networks, enriched genes and biological pathways.

Materials and methods: Gene expression profiles for (DN) were downloaded from Gene Expression Omnibus (GEO) data repository for human and mouse and filtered only for diabetic and normal condition samples, normalized and batch corrected. Differentially expressed genes were obtained using bioconductor package “limma”. We performed modular repertoire analysis for each dataset separately, where differentially expressed genes are represented in fixed repertoire of transcriptional modules. To gain insight into the biological processes and pathways that could mediate the DN, KEGG enrichment analyses were performed on common DEGs between both datasets using the KEGG KAAS server. To identify the possible protein-protein interaction (PPI) between DEGs, STRINGDB was utilized. To identify critical TFs mediating genes involved in DN EncodeJASPAR and ChEAdatabases were utilized.

Results: Based on differential expression analysis, 1283 and 2677 genes were significantly dysregulated (p-value < 0.05) in mouse and human datasets respectively. Of the DEGs, a total of 908 and 2202 genes were significantly up-regulated in (DN) samples in both mouse and human datasets when compared with control samples. Interestingly, we found signalling pathways like PI3K-Akt, MAPK, and PPAR among top enriched pathways. Key genes mediating these signalling pathways are COL4A1, TLR4, FGFR2/4, ACE, PIK3CD, SPP1, CD36, FABP4, LCN2, and LPL. The PPI network of common DEGs between 2 contrast groups consisted of 22 genes and 45 interactions. Two topological features Maximal Clique Centrality (MCC) and degree were calculated to identify key nodes. Higher the two quantitative values of a gene, the more important it is in the PPI network. The top nodes ranked by degree and MCC were identified, which included COL4A1, SP1, NFKB1, TP53, STAT3, VIMLCN2, DNMT1 and HDAC1. The TF-gene interaction network consisted of 27 nodes (12 TFs and 15 genes). The top-ranked TFs were SMAD4, SMARCA5, SIN3A, POLR2A, E2F5, ZFX, and MAZ.

Conclusion: The discovered human-mouse shared transcriptional networks enriched genes and biological pathways are most likely contributing to disease progression in type 2 diabetic patients with nephropathy and in commonly used mouse models of DN.

Supported by: Sidra Medicine- Precision Medicine Program Grant – SDR#400149 Doha, Qatar

Disclosure: B.A. Bhat: None.

130

Soluble Nogo-B upregulates Tie1 receptor: implication for vegfa/vegfr2 signalling in diabetic nephropathy

C. Ricciardi, L. Gnudi;

School of Cardiovascular Medicine and Sciences - Vascular Biology Section, King's College London, London, UK.

Background and aims: The angiopoietin (angpt)/Tie2 system has been implicated in the permselective properties of the glomerular filtration barrier with angpt1 mediating Tie2 phosphorylation (activation) promoting capillary stability and angpt2 (competing for angpt1/Tie2 binding) preventing the action of angpt1. Recent reports have supported the notion that Tie1, an orphan receptor of the Tie receptor family, is required for angpt1 mediated activation of Tie2 phosphorylation and that could also facilitate the action of angpt2 as a Tie2 activator. We recently explored the role of the endoplasmic reticulum protein NogoB and its soluble (sNogoB) circulating isoform (~200 aa N-terminus of NogoB) in experimental animal models of diabetic kidney disease. NogoB and sNogoB bind, with their N-terminus, to their receptor NgBR, expressed in endothelial cells. NogoB is downregulated in diabetic glomeruli and overexpression of sNogoB improves diabetes-mediated albuminuria and prevents diabetes-mediated upregulation of VEGFA/VEGFR2 system, recognised pathway involved in diabetic glomerulopathy. Recent reports have suggested that Tie1 upregulation favours angpt1/2 Tie2 phosphorylation and inhibition VEGFA/VEGFR2 receptor system favouring vascular stability. We hypothesise that Tie1 is implicated in the renal protective response mediated by the sNogoB/NgBR system.

Materials and methods: VEGFR2 phosphorylation and Tie1 protein expression was studied with western immunoblotting in kidney cortex cell lysate of control and diabetic (streptozotocin induced) DBA2J mice overexpressing sNogoB in the circulation (adenoviral vector). In glomerular endothelial cells (GECs)(gift of Dr Satchell), with or without sNogoB overexpression, VEGFA-mediated VEGFR2 phosphorylation was studied by incubating starved GECs with VEGFA (50 ng/ml) for 15 min and ratio VEGFR2 phosphorylation/total VEGFR2 expression studied with western immunoblotting on total cell lysate. NgBR deletion in GECs was performed with siRNA technology and Tie1 expression studied with western immunoblotting.

Results: In-vivo, non-diabetic (ND) and diabetic (D) mice administered with AAV-sNogo-B were characterized by a 12-fold increase in circulating sNogoB when compared to controls (AAV-GFP) (ND-GFP vs ND-sNogoB and D-GFP vs D-sNogoB, p<0.001). sNogo-B overexpression in the circulation was paralleled by 4-5 times upregulation of Tie1 expression in both non-diabetic and diabetic mice (ND-GFP vs ND-sNogoB and D-GFP vs D-sNogoB, p<0.001, n=3-6 per group). Similarly, sNogoB overexpression in GECs upregulated the sNogo-B protein level in the supernatant by 10-fold and partially inhibited VEGFA-mediated VEGFR2 phosphorylation (p<0.05, n=4). Deletion of NgBR in GECs was paralleled by a near total downregulation of Tie1 (p<0.001, n=3-4).

Conclusion: sNogoB overexpression prevents the VEGFA-mediated VEGFR2 phosphorylation possibly by upregulating Tie1 receptor. Downregulation of Tie1 in NgBR deficient cells implicates the sNogo/NogoB-NgBR system in the regulation of the angpt Tie2/Tie1 receptor system. The sNogoB-mediated Tie1 upregulation represents a novel concept and opens new investigations looking at the potential protective role of sNogoB diabetic chronic vascular complications.

Supported by: KRUK

Disclosure: C. Ricciardi: None.

131

Novel compounds found to regulate VEGF-A splicing in diabetic podocytes

M.L. Ayine, Y. Liu, M. Stevens, S. Oltean;

University of Exeter, Exeter, UK.

Background and aims: Alternative splicing (AS) gives rise to multiple proteins from the same gene, by skipping or including exons or parts of them, or by keeping introns as coding sequences. The resulting AS isoforms may function differently from each other. Vascular endothelial growth factor A (VEGF-A) is an angiogenic protein. The alternatively spliced variant, VEGF-A165b, is formed when a distal 3’ splice site in exon 8 is selected. The anti-angiogenic and anti-permeability VEGF-A165b has reno-protective properties and has been shown to rescue kidney function in mouse models of diabetic nephropathy. The aim of this study was to investigate three novel compounds that regulate VEGF-A AS in podocytes exposed to a diabetic environment.

Materials and methods: Two of the VEGF-A AS regulatory compounds, trovafloxacin (10μM) and 10058-F4 (5-[(4-Ethylphenyl) methylene]-2-thioxo-4-thiazolidinone) (10μM), were identified from nine synthetic compounds to switch VEGF-A splicing. A third compound, delphinidin (10μg/ml), was found to be key compound regulating VEGF-A AS in a natural blueberry and seabuckthorn extract (DIAVIT). Human podocytes were exposed to a diabetic environment (glucose soup [GS]: 25 mM glucose, 1 ng/ml TNF-α, 1 ng/ml IL-6, and 100 nM insulin), in comparison to a normal glucose (5.5 mM glucose) and an osmotic (5.5 mM glucose + 19.5 mM mannitol) control, for 48 hours. RNA and protein were extracted for RT-PCR and Western blotting analysis of splice isoforms.

Results: Trovafloxacin (mean=2.48, ±0.45 SEM, *p=0.0212, N=17), 10058-F4 (mean= 4.20, ± 0.97 SEM, ****p<0.0001, N=18) and delphinidin (mean= 2.96, ± 1.01 SEM, *p<0.0446, N=9) significantly increased the anti-angiogenic VEGF-A165b relative to pro-angiogenic VEGF-A165a in podocytes exposed to a normal glucose environment at either the mRNA or protein level. Furthermore, 10058-F4 (mean= 5.75, ± 0.53 SEM, ***p=0.0008, N=9) and delphinidin (mean= 2.67, ± 0.64 SEM, *p<0.0269, N=11) were also found to increase the VEGF-A165b/VEGF-A165a ratio in podocytes exposed to a diabetic environment. Statistical analysis were done by using a one way ANOVA followed by a Tukey’s post-hoc analysis with p values <0.05 considered as significant. Regarding 10058-F4 and delphinidin, pilot data suggests that these compounds may influence the expression of Clk-1, a kinase known to regulate VEGF-A splice site selection, in diabetic podocytes. Additionally, delphinidin was found to significantly increase the phosphorylation of SRSF6, a splice factor known to promote VEGF-A165b splice site selection, in diabetic podocytes. 10058-F4 was also found to downregulate SRSF1 may be through c-myc inhibition.

Conclusion: We have identified three novel compounds that regulate VEGF-A AS to promote the expression of the reno-protective VEGF-A165b isoform in podocytes. This study will further investigate the mechanism of action of these three compounds regarding VEGF-A splicing regulation in diabetic podocytes. The final goal of this project is to identify whether these novel AS regulatory compounds can be used to develop new therapeutic strategies in diabetic nephropathy.

Supported by: Diabetes UK

Disclosure: M.L. Ayine: Grants; Diabetes UK. Lecture/other fees; Richard Bright VEGF research Trust.

132

Circulating tenascin-C levels predict renal progression in type 2 diabetes

D.T.W. Lui, C. Lee, C.Y.Y. Cheung, C.H.Y. Fong, W. Chow, Y. Woo, K.S.L. Lam;

Department of Medicine, The University of Hong Kong, Queen Mary Hospital, Hong Kong, Hong Kong.

Background and aims: Tenascin-C (TN-C) is an extracellular matrix glycoprotein highly expressed in inflammatory states. In individuals with type 2 diabetes, a condition associated with chronic low-grade inflammation, serum TN-C levels have been reported to correlate with baseline eGFR and albuminuria status, and predict major adverse cardiovascular events and mortality. Whether serum TN-C level is prospectively associated with renal progression in type 2 diabetes remains to be investigated. Therefore, we carried out this cohort study to address this issue.

Materials and methods: We conducted a nested case-control study involving participants recruited from the Hong Kong West Diabetes Registry (HKWDR), who had type 2 diabetes followed up regularly at the medical specialist clinics of the Hong Kong West Cluster since 2008. In this study, only those with baseline estimated glomerular filtration rate (eGFR) ≥60 mL/min and followed up for at least 1 year were included. Renal progression was defined by doubling of serum creatinine, which was equivalent to a 57% decline in eGFR. We compared TN-C level between the participants with and without renal progression, while controlling for underlying baseline demographic and clinical differences (glycemic control, eGFR and albuminuria status) and follow-up interval between two groups using propensity score matching. Each case of renal progression was matched to one control. Multivariable conditional logistic regression was employed to identify the independent factors associated with renal progression.

Results: A total of 760 individuals (380 cases and 380 controls) were included after propensity score matching. Their baseline characteristics were as follows: 57.1% men, mean age 63.0±12.0 years, BMI 26.5±5.0 kg/m2, glycated hemoglobin (HbA1c) 7.9±1.6%, duration of diabetes 12.2±8.7 years. The mean baseline eGFR was 82.7±14.4 mL/min with 42.2% of the cohort having albuminuria of <30mg/g (A1), 37.8% having albuminuria of 30-300mg/g (A2), and 20.0% having albuminuria of >300mg/g (A3). The mean follow-up period was 5.5±2.1 years. The baseline characteristics between individuals with and without renal progression were well-balanced in terms of age, sex, HbA1c, duration of diabetes, lipid profile, eGFR and albuminuria status, use of renin-angiotensin-aldosterone-system blockade, statin and aspirin. However, there were more prevalent cardiovascular diseases (p=0.049) and insulin users (p=0.038) in those with renal progression. Notably, despite comparable high-sensitivity C-reactive protein levels at baseline, we observed higher serum TN-C levels in those with renal progression compared to those without (44.2 ng/mL [95% CI 33.0-59.6] vs 41.7 ng/mL [95% CI 29.6-55.5], p=0.016). Multivariable conditional logistic regression analysis revealed that serum TN-C levels remained independently associated with renal progression (adjusted OR 1.39 [95% CI 1.03-1.89], p=0.034) after adjustments for prevalent cardiovascular diseases and insulin usage.

Conclusion: Our data would suggest serum TN-C level to be an independent predictor of renal progression among individuals with type 2 diabetes and relatively preserved kidney function.

Disclosure: D.T.W. Lui: None.

OP 23 Advances in insulin therapy

133

Ado09, a co-formulation of pramlintide and insulin A21G improves post-prandial glucose and body weight versus insulin aspart in type 1 diabetes

G. Meiffren1, G. Andersen2, R. Eloy1, C. Seroussi1, C. Mégret1, S. Famulla2, Y.-P. Chan1, M. Gaudier1, O. Soula1, J.H. DeVries2, T. Heise2;

1Adocia, Lyon, France, 2Profil Neuss, Neuss, Germany.

Background and aims: Pramlintide improves post-prandial glucose through delaying gastric emptying, reducing glucagon secretion, and promoting satiety. ADO09 is a co-formulation of pramlintide and insulin A21G. This study was a two-part, double-blind, randomised, 2-period cross-over trial comparing pre-meal ADO09 versus insulin aspart over 24 days in subjects with type 1 diabetes. Part A was conducted in participants using less than 40 U prandial insulin per day, part B in subjects using 40 to 75 U prandial insulin per day.

Materials and methods: During a 28 days run-in period, basal insulin was switched to insulin degludec and titrated. Each treatment period consisted of 3 inpatient days [baseline assessments including a mixed-meal-tolerance test (MMTT) on Day 1], followed by 3 outpatient weeks and a final inpatient MMTT on day 24. Blood glucose, glucagon, and kinetics of gastric emptying were analyzed, as were CGM-metrics. The two treatment periods were separated by a 5 to 7 days washout.

Results: 28 subjects were enrolled in part A and 16 in part B. BMI was higher in those using more prandial insulin, with a mean of 24.54±2.03 kg/m² in part A compared to 30.54±3.08 kg/m² in part B. Incremental plasma glucose AUCs during day 24 MMTT with ADO09 were reduced by >100% in the first 2hrs (p<0.001) in both parts vs insulin aspart and, after 4hrs by 39% (p=NS) in part A and 69% (p=0.027) in part B (figure). Gastric emptying was slower with ADO09 (Tmax part A: ADO09 2.47 h vs aspart 0.50h; part B: ADO09 1.89 h vs aspart 0.69 h) and glucagon levels lower with ADO09 over 0-2h (∆AUCGlucagon_0-2h part A: ADO09 3.29 pmol*h/L vs aspart 11.32 pmol*h/L; part B: ADO09 3.45 pmol*h/L vs aspart 11.66 pmol*h/L). Looking at CGM, BG improved significantly with ADO09 treatment (mean BG over 24h part A: -8.2 mg/dL with ADO09 vs aspart (p<0.001); part B: -7.0 mg/mL with ADO09 vs aspart (p=0.013)), and a higher Time-In-Range was observed (part A: +51 min (p=0.013); part B: +58 min (p=0.043) vs aspart). Time <70 mg/dL was slightly higher with ADO09 (part A: +10 min (p=0.046); part B: +13 min (p=NS)), as were hypoglycemic events (part A: ADO09 142 vs aspart 115; part B: ADO09 96 vs aspart 79). ADO09 reduced body weight in both parts, but the reduction was more pronounced in part B (part A -0.7 kg p=0.012 and part B - 1.6 kg (p=0.007) vs baseline), while insulin aspart had not such an effect (part A +0.0 kg (p=NS) and part B +0.4 kg (p=NS) vs baseline). Both treatments were well tolerated with more, but transient, gastrointestinal adverse events with ADO09 in both parts (24 vs 6 in part A and 11 vs 3 in part B), consistent with the known side effect profile of pramlintide. Overall treatment satisfaction measured with a questionnaire was improved with ADO09.

Conclusion: ADO09 was well tolerated and significantly improved post-prandial blood glucose control, hyperglucagonemia, CGM-metrics and body weight versus insulin aspart over 24 days across a wide range of prandial insulin doses.

figurear

Clinical Trial Registration Number: NCT03981627

DisclosureG. Meiffren: Employment/Consultancy; Adocia. Stock/Shareholding; Adocia.

134

Once weekly basal insulin Fc is safe and efficacious in patients with type 2 diabetes previously treated with basal insulin

J. Frias1, J. Chien2, Q. Zhang2, E. Chigutsa2, W. Landschulz2, P. Wullenweber2, A. Haupt2, C. Kazda2;

1National Research Institute, Los Angeles, 2Eli Lilly and Company, Indianapolis, USA.

Background and aims: Basal insulin Fc (BIF; LY3209590) is a novel, once-weekly, long-acting IgG Fc-fusion protein that is being assessed for the treatment of diabetes mellitus. The presented study evaluated the safety and efficacy of BIF compared to insulin degludec over 32 weeks in patients with type 2 diabetes mellitus previously treated with oral antidiabetic drugs and a basal insulin.

Materials and methods: The study design included 2 different dosing algorithms for BIF (BIF-A1 and BIF-A2) with two different fasting glucose (FG) targets of ≤7.8 mmol/L (BIF-A1) and ≤6.7 mmol/L (BIF-A2). Insulin degludec was titrated to a FG target of ≤5.6 mmol/L using a modified Riddle treat-to-target algorithm.

Results: Study participants (N=399) were randomized in a 1:1:1 ratio to 1 of 3 parallel treatment groups. The average age of participants was 60.2 years, baseline HbA1c was 65.2 mmol/mol (8.1%) and duration of diabetes 14.7 years. There were no statistically significant differences in demographics or baseline characteristics across the 3 treatment groups. Both BIF groups achieved non-inferiority (non-inferiority margin = 0.4%) for the primary endpoint of HbA1c change from baseline to Week 32 with a mean±SE reduction for BIF-A1, BIF-A2 and insulin degludec of 0.6±0.1%, 0.6±0.1% and 0.7±0.1%, respectively. In line with the different fasting serum glucose (FSG) targets, insulin degludec achieved greater FSG lowering from baseline as compared to the BIF arms. Similarly, both BIF dosing groups showed significantly fewer hypoglycaemic events compared to insulin degludec (all documented events as well as nocturnal events) when assessing events ≤3.9 mmol/L. Hypoglycaemic events <3.0 mmol/L (all documented events as well as nocturnal events) were not significantly different between the three dosing groups. Two severe hypoglycaemic events were reported in BIF-A2. The reported treatment-emergent adverse events and serious adverse events (SAEs) were balanced across the 3 treatment groups. Both BIF groups had a statistically significantly smaller increase in body weight compared to insulin degludec from baseline to Week 32.

Conclusion: In summary, BIF, when administered weekly according to either dosing algorithm, was noninferior to insulin degludec for glycaemic control as measured by change in HbA1c after 32 weeks with a lower rate of documented and nocturnal hypoglycaemia ≤3.9 mmol/L and less weight gain. Additionally, no safety signals were detected. While higher FG targets were chosen in this first Phase 2 study with BIF, the safety and tolerability results allow assessment of lower target glucose ranges in future trials. The results from this study support continued development of BIF as a once-weekly insulin treatment of diabetes mellitus.

Clinical Trial Registration Number: NCT03736785

Supported by: Eli Lilly and Company

Disclosure: J. Frias: Employment/Consultancy; Altimmune, Axcella, Boehringer Ingelheim, Coherus Therapeutics, Echosens, 89bio, Eli Lilly, Gilead, Intercept, Merck, Novo Nordisk, Sanofi. Grants; Allergan, AstraZeneca, Boehringer Ingelheim, BMS, Eli Lilly, Intercept, Janssen, Madrigal, Metacrine, Merck, NorthSea Therapeutics, Novartis, Novo Nordisk, Oramed, Pfizer, Poxil, Sanofi, Theracos. Other; Merck, Sanofi.

135

Similar hypoglycaemia duration with once-weekly insulin icodec vs insulin glargine U100 in insulin naive or experienced patients with type 2 diabetes

R. Silver1, M. Asong2, K. Begtrup2, S.R. Heller3, L. Liu2, J. Rosenstock4;

1Southern New Hampshire Diabetes and Endocrinology, Nashua, USA, 2Novo Nordisk A/S, Bagsværd, Denmark, 3University of Sheffield, Sheffield, UK, 4Dallas Diabetes Research Center at Medical City, Dallas, USA.

Background and aims: Insulin icodec (icodec) is a novel once-weekly basal insulin analogue in development. This post hoc analysis explored the duration of hypoglycaemia using double-blinded continuous glucose monitoring (CGM) (Dexcom G6®) data from two phase 2, randomized, open-label, treat-to-target 16-week trials.

Materials and methods: One trial compared three titration algorithms of icodec vs insulin glargine U100 (IGlar U100) in 205 insulin-naïve patients with type 2 diabetes (T2D), the other assessed 154 basal insulin-treated patients with T2D switching from any daily basal insulin to icodec with or without a 100% loading dose vs IGlar U100. In line with ATTD guidelines, a hypoglycaemic episode was defined as a CGM period of interstitial glucose (IG) <3.9 mmol/L for at least 15 min, ending when IG ≥3.9 mmol/L for at least 15 min.

Results: In the titration trial, the duration of hypoglycaemia was similar across all arms: median (interquartile range [IQR]) overall hypoglycaemic episode duration was 35.0 (20.0, 70.0) min for icodec titrations A (self-measured blood glucose target 4.4-7.2 mmol/L; adjusted ±21 U/week) and B (4.4-7.2 mmol/L ±28 U/week) and for IGlar U100 (4.4-7.2 mmol/L ±4 U/day), and 39.0 (24.0, 70.0) min for icodec titration C (3.9-6.0 mmol/L ±28 U/week). The distribution pattern of hypoglycaemic episodes by duration was similar across all treatment arms. Results were similar for nocturnal hypoglycaemic episodes. Similarly, in the switch trial, the duration of hypoglycaemia was similar between arms, irrespective of loading dose use: median overall hypoglycaemic episode duration (IQR) was 40.0 (20.0, 75.0) min for icodec with loading dose, 40.0 (25.0, 80.0) min for icodec without loading dose, and 35.0 (20.0, 60.0) min for IGlar U100. The distribution pattern of hypoglycaemic episodes by duration was similar across treatment arms. Similar results were seen for the nocturnal period.

Conclusion: In conclusion, CGM-based hypoglycaemic episode duration was similar with icodec vs IGlar U100 in insulin-naïve and insulin-experienced patients with T2D, regardless of titration algorithm or initial loading dose use.

Clinical Trial Registration Number: NCT03951805 and NCT03922750

Disclosure: R. Silver: None.

136

A first in human, single ascending dose study to assess the safety and tolerability, pharmacokinetics, and pharmacodynamics of AB101 in subjects with type 1 diabetes

B.K. Roberts1, X. Wang1, B. Franey2, M. Hernandez2, M. Hompesch2;

1Rezolute, Redwood City, 2ProSciento, Chula Vista, USA.

Background and aims: AB101 is a slow release microsphere form of polyethylene glycol-human recombinant insulin (peginsulin) for potential use as an ultra long acting basal insulin in patients with diabetes mellitus. In animals, including alloxan-induced diabetic mini pigs, AB101 resulted in slow onset, sustained, and peak less increases in insulin and corresponding glucose lowering for a week or more. This first in human study was designed to assess the safety, tolerability, pharmacokinetics (PK), and pharmacodynamics (PD) of single subcutaneous (SC) administration of AB101 in subjects with T1DM who were on background continuous SC insulin infusion (CSII).

Materials and methods: This study utilized the euglycemic clamp technique, CSII, and continuous glucose monitoring (CGM). Cohorts 1 thru 3 received single doses of AB101 (0.25, 0.5, or 1.0 mg/kg peginsulin) in sequential ascending dose fashion. Serial measurements of peginsulin were obtained. Three clamps of 24-36h duration were performed at cohort-specific timepoints adapted over the course of study to capture the time-action profile of AB101, between Day 3 and Day 30 relative to dosing. A follow-up safety visit occurred after clinic discharge.

Results: The mean (SD) age and BMI were 35.2 (10.3) and 25 (3.1), respectively; 69% and 88% of participants were male and caucasian, respectively. There were no serious adverse events (AEs) or discontinuations due to treatment emergent AEs, and no clinically significant changes in laboratory evaluations, ECGs, or vital signs. The PK profile showed a delayed onset of AB101 in humans compared to animals. The concentration-time profile demonstrated a dose-dependent increase in serum peginsulin concentrations beginning at Day 15 to Day 20 relative to dosing, with sustained insulin levels for a week or more. There was no evidence of initial or delayed sudden insulin release (burst). Corresponding to the temporal PK profile and supporting a time-action relationship, there was a trend toward dose dependent increases in glucose disposal as reflected by increases in glucose infusion rate (GIR), at clamp timepoints as described in the table. Additionally, AB101 resulted in decreased glycemic variability by CGM, and Cohort 3 consistently required less insulin during clamp run-in and non-clamp days (via CSII) compared to Cohorts 1 and 2.

Conclusion: Overall, single SC doses of AB101 in patients with T1DM were safe and well tolerated, with no evidence of sudden insulin release. Administration of AB101 resulted in slow onset and sustained insulin levels and activity for more than seven days with dose-dependent improvements in glycemic variability, background insulin requirements, and glucose disposal. However, at active doses, the necessary dose volume was greater than anticipated, and delayed or variable kinetics was observed. Formulation optimization would be necessary to to achieve the target product profile for a weekly basal insulin.

figureas

Disclosure: B.K. Roberts: Employment/Consultancy; Rezolute. Stock/Shareholding; Rezolute.

137

Final data from a long-term observational study of continuous intraperitoneal insulin infusion in a vulnerable population with diabetes

B. Gehr1, N. Oliver2,3, E. Renard4,5, D. Hilgard6, K. Mueller7, C. Rieger7, W. Mueller-Hoffmann7, A. Liebl1;

1m&i-Fachklinik Bad Heilbrunn, Bad Heilbrunn, Germany, 2Department of Metabolism, Digestion and Reproduction, Faculty of Medicine, Imperial College London, London, UK, 3Imperial College Healthcare NHS Trust, London, UK, 4Department of Endocrinology, Diabetes and Nutrition, Montpellier University Hospital, Montpellier, France, 5Institute of Functional Genomics, CNRS, INSERM, University of Montpellier, Montpellier, France, 6Kinder-Endokrinologie und –Diabetologie, Witten, Germany, 7Roche Diabetes Care GmbH, Mannheim, Germany.

Background and aims: Continuous intraperitoneal insulin infusion (CIPII) is an alternative method of insulin delivery in the management of people with type 1 or type 2 diabetes (T1D/T2D) requiring insulin, for whom subcutaneous insulin administration is not achieving target glucose, including people who do not tolerate subcutaneous insulin administration. The Accu-Chek® DiaPort system is intended for CIPII and was previously shown to be safe and effective. To assess long-term safety and performance of CIPII therapy with the Accu-Chek® DiaPort we conducted an observational post-market clinical follow-up (PMCF) study to capture real life information from people using the system.

Materials and methods: People aged from 6 years onwards with diagnosed T1D/T2D treated with insulin, either requiring the implantation of the Accu-Chek® DiaPort system, or with the system already implanted, were included. Participants routinely visited the clinic at least every 6 months for 24 months. Additional visits occurred if clinically required. There were no additional appointments or treatments for study purposes. After 24 months, additional safety data were captured at routine visits if the patient agreed to this. All recorded and derived variables are presented, stratified by treatment group and visits, using descriptive summary tables (continuous and ranked data: e.g. sample size, mean, SD).

Results: In 2019 we presented the first data from this study including 88.4 patient years. The most frequent indications for CIPII were unacceptable glucose fluctuations, hypoglycemia unawareness, and severe hypoglycemia, as well as subcutaneous insulin resistance, lipodystrophies and skin disorders. The first data showed that glycemic control (HbA1c and hypoglycemia) improved markedly. Furthermore, no additional unexpected risks were observed and no death occurred in this vulnerable patient population. The study finished at the end of 2020 reaching the goal of collecting data from over 100 patient years. The final data show a marked improvement of mean HbA1c between Visit 1 and last measurement from 8.2 to 7.65% without increasing the number of severe hypoglycemia. We will present further safety and efficacy as well as patient satisfaction data from this 2-5 years long observation time.

Conclusion: This will be the first presentation of safety, therapy efficacy and patient satisfaction data from 49 people with type 1 or type 2 diabetes treated with CIPII therapy using the Accu-Chek® DiaPort system.

Supported by: Roche Diabetes Care GmbH

Disclosure: B. Gehr: Honorarium; Participation in Roche Diabetes Care GmbH Advisory Boards.

138

Comparative effectiveness of insulin glargine 300 U/mL and insulin degludec 100 U/mL in insulin naive type 2 diabetes adults: the Restore-2 naive cohort

G.P. Fadini1, R. Buzzetti2, M. Larosa3, M.C. Rossi4, A. Nicolucci4, D. Cucinotta5;

1Department of Medicine, University of Padova, Padua, 2Department of Experimental Medicine, Sapienza University of Rome, Rome, 3Sanofi, Milan, 4CORESEARCH – Center for outcomes research and clinical Epidemiology, Pescara, 5Department of Medicine, University of Messina, Messina, Italy.

Background and aims: Second generation basal insulins (2BI) provide similar or improved efficacy with a better safety profile compared to first generation BI. Comparative real-world data on 2BI in European patients with type 2 diabetes (T2DM) are scant. The aim of the study was to assess comparative effectiveness and safety of 2BI [Insulin Glargine 300U/mL (Gla-300) vs. Insulin Degludec 100 U/mL(IDeg-100)] in naive T2DM patients at 6 months. Persistence of HbA1c reduction after 12 months from 2BI initiation was also assessed as long-term outcome.

Materials and methods: RESTORE-2 was a retrospective, non-inferiority, multicenter study, based on electronic medical records. All patients initiating Gla-300 or IDeg-100 in January 2017 - 2020 were 1:1 propensity score matched (PSM). Linear mixed models for repeated measures were applied to assess changes from baseline to 6 months in HbA1c and FBG levels, body weight and insulin dose, and changes in HbA1c levels after 12 months in the two groups. Incidence rates (IR) of hypoglyacemic events were compared between-group using Poisson regression models.

Results: Overall, 357 patients in each PSM cohort were identified. Participants had mean age of 69 years, diabetes duration of 14 years, and baseline HbA1c of 9.2% (77 mmol/mol). After 6 months, marked reductions in HbA1c were documented in Gla-300 and IDeg-100 group [-1.70% (95%CI -1.90;-1.50) vs. -1.69% (95%CI -1.89;-1.49)] without statistically significant between-group difference (p=0.49). The non-inferiority of Gla-300 vs. IDeg-100 was confirmed (margin of non-inferiority of 0.30%; between group mean difference at 6 months: 0.01%; 95%CI -0.29;0.27). Statistically significant within-group reductions were documented in FBG levels: -63 mg/dl (3.5 mmol/L) in Gla-300 group and -61 mg/dl (3.4 mmol/L) in IDeg-100 group (between group difference p=0.74). Minor changes in body weight were documented in both groups. No between-group differences were documented in insulin doses; after 6 months, the dose was 0.20 U/Kg in both groups. IR (episodes per patient-months) of blood glucose (BG) ≤70 mg/dl was 0.13 (95%CI 0.07;0.26) in Gla-300 group and 0.14 (95%CI 0.07;0.27) in IDeg-100 group (p=0.87) during 6 months. IR of BG <54 mg/dl was 0.02 (95%CI 0.01;0.05) in Gla-300 group and 0.02 (95%CI 0.01;0.04) in IDeg-100 group (p=0.49). No severe hypoglycaemic episodes were reported. HbA1c reduction from baseline to 12 months was greater in the Gla-300 group than in Ideg-100 group [-1.71% (95%CI (-1.94;-1.48) vs. -1.44% (95%CI (-1.67;-1.21)], although the between-group difference did not reach the statistical significance (p=0.052).

Conclusion: In adults with T2DM, initiating Gla-300 or IDeg-100 in real world practice was associated with similar improvements in glycemic control, with no weight gain, low hypoglycaemia rates, and no severe episodes. After 12 months, persistence of HbA1c reduction was documented in both groups.

Supported by: Sanofi

Disclosure: G.P. Fadini: Grants; AstraZeneca, Novo Nordisk, Mundipharma. Honorarium; Abbott, AstraZeneca, Boehringer, Lilly, Daichi-Sankyo, Mundipharma, Sanofi. Lecture/other fees; Abbott, AstraZeneca, Boehringer, Lilly, Mundipharma, Sanofi, Servier.

OP 24 Epidemiology of diabetes complications

139

Young-onset type 2 diabetes: clinical outcomes in Norwegian general practice

K.L. Tibballs, A. Jenum, E.S. Buhl;

General Practice, University of Oslo, Oslo, Norway.

Background and aims: People diagnosed with type 2 diabetes (T2D) early in adult life incur a high lifetime risk of complications and reduced life expectancy. General practitioners are responsible for the predominance of diagnosis and management of young-onset T2D (YOD) in Norway. In this study we aim to estimate the prevalence of YOD in Norway and study the relationships between age at diagnosis, duration of T2D and rates of macro- and microvascular complications.

Materials and methods: Cross-sectional data from general practice electronic medical records of 10 242 adults with T2D were collected in 2015. YOD was defined as diagnosis by age 40. We studied clinical outcomes by age at diagnosis, taking into account diabetes duration or current age. Associations were analysed by multivariate regression, using diabetes diagnosis after age 60 as baseline comparator.

Results: In our study sample, 11.8 % of those with a recorded year of diagnosis (n=9605 non-missing values) were diagnosed with T2D by age 40. In the unadjusted analysis, mean HbA1c (n=9346) was 50.8 mmol/mol among those diagnosed after age 60. Levels were on average 7.79 (6.98, 8.61) mmol/mol higher in YOD. In YOD, HbA1c showed a greater increase with age and diabetes duration. LDL (n=8105) and total cholesterol (n=8550) were also higher in YOD. Retinopathy (n=5880) was found in 8.1% of those diagnosed after age 60. In comparison, retinopathy in YOD had an OR of 3.87 (3.06, 4.88). After adjustment for diabetes duration, sex, highest education level, HbA1c, systolic blood pressure and LDL cholesterol, logistic regression analysis gave an OR of 1.89 (1.37, 2.59). In YOD, retinopathy arose earlier after diagnosis. Macrovascular disease was less frequent in YOD overall. Coronary heart disease had an OR of 0.24 (0.20, 0.30) in YOD. However, in a time to event model with age-based baseline frailty and adjusting for the same covariates as above, cox regression analysis gave a hazard ratio for coronary heart disease in YOD of 1.96 (1.53, 2.50). Macrovascular complications showed a stronger relation to current age than to diabetes duration.

Conclusion: All diabetes complications tended to increase more with age in young onset diabetes compared to older onset. Retinopathy showed the most severe development in YOD, with much higher prevalence and onset starting earlier after diabetes diagnosis. These findings may have implications for future guidelines and individualization of diabetes care.

figureat

Supported by: Katrina Tibballs is supported by AMFF with a PhD grant.

Disclosure: K.L. Tibballs: Grants; Allmennmedisinsk forskningsfond.

140

Diabetes complications among patients from metropolitan versus non-metropolitan cities in India: one year results of LANDMARC

S. Kalra1, A.K. Das2, S. Joshi3, A. Mithal4, K.M. Prasanna Kumar5, A.G. Unnikrishnan6, H. Thacker7, B. Sethi8, S. Chowdhury9, R. Ghosh10, S. Krishnan11, A. Nair10, D. Chodankar11, A.H. Zargar12, LANDMARC Study Group;

1Bharti Hospital, Karnal, 2Pondicherry Institute of Medical Sciences (PIMS), Puducherry, 3Lilavati Hospital, Mumbai, 4Medanta- The Medicity, Gurgaon, 5Centre for Diabetes and Endocrine Care, Bengaluru, 6Chellaram Diabetes Institute, Pune, 7Bhatia Hospital, Mumbai, 8Care Hospital, Hyderabad, 9IPGME and R and SSKM Hospital, Kolkata, 10Medical Affairs, Sanofi, Mumbai, 11Clinical Study Unit, Sanofi, Mumbai, 12Center for Diabetes & Endocrine Care, Srinagar, India.

Background and aims: As type 2 diabetes (T2D) progresses, it inflicts damage to the vasculature. Micro- and macro-vascular complications are the major causes of morbidity and mortality associated with T2D. In India, data on complications in patients with diabetes from metropolitan and non-metropolitan cities are unavailable. Hence, micro- and macro-vascular complications during the first year of LANDMARC, a 3-year nationwide prospective observational study, were evaluated in patients from metropolitan versus non-metropolitan cities.

Materials and methods: LANDMARC is the first nation-wide, prospective, long-term, multicenter, observational, and longitudinal study including patients with T2D. LANDMARC study included patients with T2D who were on ≥2 antihyperglycemic medications. Each participant is intended to be evaluated over the 3-year period (March 2018 to March 2021), comprising 7 visits at 6-months interval.

Results: Of the total 6236 enrolled patients, 2378 and 3858 were from metropolitan and non-metropolitan cities, respectively. Age, duration of T2D, and baseline A1C were similar across groups (Table). At 1-year, microvascular complications were significantly higher in those from non-metropolitan than metropolitan cities (19.08% vs. 10.89%; P<0.0001) (Table). Neuropathy was the most common microvascular complication reported in both the groups. Among macrovascular complications reported, acute coronary syndrome and heart failure were the most common and were significantly higher in participants from non-metropolitan cities (Table).

Conclusion: The present, first-of-its-kind data from India demonstrates that patients from non-metropolitan cities may have higher complications, particularly microvascular. Results from this ongoing study present patterns of disease progression among patients with T2D.

figureau

Clinical Trial Registration Number: CTRI/2017/05/008452

Supported by: Study funded by Sanofi.

Disclosure: S. Kalra: Honorarium; SK received honoraria/speaker fees from Eli Lilly, Novo Nordisk and Sanofi.

141

Risk factors, incident dementia, cognitive performance and structural brain abnormalities in individuals with type 2 diabetes

A.C.E. van Gennip1,2, C.D.A. Stehouwer1,2, The Maastricht Study, A. Singh-Manoux3,4, T.T. van Sloten1,2;

1Internal Medicine, MUMC+, Maastricht, Netherlands, 2School for Cardiovascular Diseases CARIM, Maastricht, Netherlands, 3INSERM U1153 Epidemiology of Ageing and Neurodegenerative diseases, Université de Paris, Paris, France, 4Epidemiology and Public Health, University College London, London, UK.

Background and aims: Type 2 diabetes is associated with increased risks of cognitive dysfunction and structural brain abnormalities. The extent to which risk factor modification can mitigate these excess risks is unclear. We investigated the associations between incident dementia, domain-specific cognitive performance and structural brain abnormalities among individuals with type 2 diabetes, according to the number of risk factors within target range, compared to controls without diabetes.

Materials and methods: Prospective data from UK Biobank of 87,856 individuals (n=10,663 diabetes; n=77,193 controls; baseline 2006-2010; dementia follow-up until February, 2018). Analysis was replicated using data from the Netherlands (the Maastricht Study; cohort with oversampling of type 2 diabetes; examination 2010-2019; cross-sectional data). Individuals with type 2 diabetes were categorized according to the number of seven selected risk factors within target range (nonsmoking; guideline-recommended levels of HbA1c, blood pressure, BMI, albuminuria, physical activity, diet). Outcomes were incident dementia, domain-specific cognitive performance, white matter hyperintensity volume and total brain parenchyma volume.

Results: After a mean follow-up of 9.0 years, 147 (1.4%) individuals with type 2 diabetes and 412 (0.5%) controls had incident dementia. Compared to controls, individuals with type 2 diabetes had a higher incidence of dementia (HR: 1.88 (95% CI: 1.55;2.27)). Among individuals with type 2 diabetes, excess dementia risk decreased stepwise for a higher number of risk factors within target range. Among individuals with type 2 diabetes who had 5 to 7 risk factors within target range, compared to controls (incidence rate per 1,000 person-years 0.62 (95%CI: 0.56; 0.68)), the absolute rate difference per 1,000 person-years for dementia was 0.20 (-0.11; 0.52) and the hazard ratio for dementia was 1.32 (0.89; 1.95). Similarly, differences in processing speed, executive function, and brain volumes were progressively smaller for a higher number of risk factors within target range; these results were replicated in the Maastricht Study.

Conclusion: Among individuals with type 2 diabetes, excess risk of dementia, lower cognitive performance and brain abnormalities decreased stepwise for a higher number of risk factors within target range.

Supported by: ERDF, PoL, SDW, PSID, CVC, CARIM, CAPHRI, NUTRIM, SA, HFL, JCBV, NNF, SAN, ZonMW, NCDC, MP, DHF

Disclosure: A.C.E. van Gennip: None.

142

Associations between chronic kidney disease, prior cardiovascular conditions and increased mortality in 36.303 type 1 diabetes patients between 2015-2017

L. Lyngfelt1, P. Wessman2, M. Miftaraj2, C. Zhou2, S. Franzén2,1, K. Eeg-Olofsson2,1, A.-M. Svensson2,1, B. Eliasson1;

1Institute of Medicine, Göteborg, 2National Diabetes Register, Göteborg, Sweden.

Background and aims: People with type 1 diabetes (T1D) still have an increased risk of premature death compared to the general population, usually caused by cardiovascular disease. The aim of this study was to present updated information on the significance of previous heart failure, kidney disease, and atherosclerotic vascular disease for the risk of mortality in a nationwide, registry-based cohort of T1D.

Materials and methods: All T1D patients in the Swedish National Diabetes Register (NDR) during the inclusion period between January 1, 2015 and December 31, 2017 were included (n=36 303). NDR data were linked to data from several national health registries through each patient’s unique personal identification number. The cumulative mortality was described using Kaplan-Meier curves, and evaluated for association to potential risk factors using a Multiple Cox regression model including age, gender, diabetes duration, HbA1c, BMI, SBP, DBP, albuminuria, previous cardiovascular disease (CVD), coronary heart disease, acute myocardial infarction or stroke, heart failure (HF) and eGFR stage as independent variables.

Results: Mean age at index day was 40.1 years, 55% were men and mean duration of diabetes 25.8 years. Record of previous CVD, was found in 8.7%, and prior HF in 1.7%. Normal kidney function (chronic kidney disease (CKD) stage G1; eGFR ≥90) was seen in 50%, and 70% were normoalbuminuric. The mean follow-up time was 3.3 years with a minimum patient follow-up time of 1 year. In total 1127 patients died, with an observed crude total mortality rate of 0.92 deaths/personyear.In patients with prior CVD the risk in all-cause mortality was HR (hazard ratio) 1.91 (95% CI 1.66-2.20) in multivariate analysis compared with persons free of CVD at baseline.In patients with HF adjusted HR was 1.92 (1.66-2.23).Patients with impaired kidney function (moderate, high and very high CKD stages) had HR 1.47 (1.19-1.80), 2.78 (2.11-3.66), and 3.80 (2.80-5.16), respectively, in multivariate analysis.Figure 1 shows cumulative mortality for people with T1D and varying degrees of CKD, with and without previous CVD and HF.

Conclusion: In our registry-based observational study, we found that for patients with T1D, a previous history of CVD, HF, and CKD all were associated with an increased risk of death, and a combination of risk conditions with substantially elevated risk.

figureav

Supported by: Sanofi

Disclosure: L. Lyngfelt: None.

143

Heart failure and renal complications in young- and usual-onset type 2 diabetes among white Caucasians from US and UK

S. Paul, J. Ling, O. Montvida;

University of Melbourne, Melbourne, Australia.

Background and aims: While people with T2DM are known to have increased risk of chronic kidney disease (CKD) and heart failure (HF), the population-level evidence on the risk of these events in young- and usual-onset T2DM in different healthcare setups is scarce. The aim of this study was to evaluate the risk of CKD and/or HF in White Caucasians with T2DM in two different healthcare setups following the same study design. We also evaluated the potential difference in risk dynamics between male and female across age groups.

Materials and methods: Using nationally representative electronic medical records of US and UK, 1491672 and 103233 White Caucasians diagnosed with T2DM between 2000 and 2018 in the age groups of 18-39 /40-49 / 50-59/ 60-69 years were identified. Date of T2DM diagnosis was considered as baseline, with first date of CKD or HF diagnosis (CKDHF) or end of follow-up from baseline considered as the time to event. The risk of CKDHF (hospitalisation or physician coded events) was evaluated in males and females in different age groups adjusting for time-varying covariates. Adjusted mean time to CKDHF was estimated using a propensity score based modelling approach, balancing and adjusting for confounders.

Results: At T2DM dx in US /UK, 50 /59% were male, 10 /13% had CKD, 2 /2% had HF, 48 /55% had hypertension, 39 /31% had dyslipidaemia, 13/15% had cardiovascular disease, and 20 /15% had microvascular disease. 24 /9% were on insulin and 34 /33% had depression prior to CKDHF or end of follow-up. With mean 4.9-5.1 /7.3-7.5 years of follow-up across all age groups in US /UK, 95% CIs for incidence rates per 1000PY for CKDHF were: (23-24) /(22-24) in 18-39 years, (37-38) /(33-35) in 40-49 years, (53-54) /(53-55) in 50-59 years and (82-83) /(83-86) in 60-69 years in US /UK. In the youngest age group in US /UK, compared to female, male had significantly higher CKDHF rate (Figure A) and 25% (HR CI: 1.20-1.29) /31% (HR CI: 1.14-1.53) higher adjusted risk. However, in the 50+ year groups, compared to female, male had significantly lower rate and 3-11 / 21-23% lower risk [range of HR CI: 0.88-0.98 /0.73-0.82] in US /UK. Adjusted years (CI) to CKDHF in 18-39 years group were 9.7 (9.2, 10.2) /8.9 (8.0, 9.7) in US/UK, only 4.2 / 2.6 years later compared to 5.5 (5.4, 5.6) /6.3 (6.2, 6.5) years in 60-69 years group (Figure B). Within age groups < 60 years, the time to event(s) was similar in US and UK.

Conclusion: While healthcare systems differ across countries, the risk paradigm for CKD and HF in White Caucasian males and females with young- and usual-onset T2DM is similar. This clearly suggests a common global approach in the proactive management of macro- and microvascular risk simultaneously in people with T2DM, particularly among young-onset T2DM who develop CKD or HF only 3-4 years later than the usual-onset.

figureaw

Disclosure: S. Paul: Employment/Consultancy; Novartis, Sanofi Avantis, GI Dynamics, Roche, AstraZeneca, Guangzhou Zhongyi Pharmaceutical and Amylin Pharmaceuticals LLC. Grants; Merck, Novo Nordisk, AstraZeneca, Hospira, Amylin Pharmaceuticals, Sanofi-Aventis and Pfizer.

144

Risk of long term HbA1c variability on cancer events and cause-specific death in 15,286 patients with type 2 diabetes (The Hong Kong Diabetes Register 1995-2019)

D. Mao, E. Lau, H. Wu, A. Yang, B. Fan, M. Shi, E. Chow, A. Kong, R. Ma, A. Luk, J. Chan;

The Chinese University of Hong Kong, Hong Kong, Hong Kong.

Background and aims: In high income areas, cancer is emerging as a leading cause of death in type 2 diabetes (T2D). The association of glycaemic variability (GV) with cancer and cause-specific death in type 2 diabetes (T2D) has not been well defined.

Materials and methods: In the Hong Kong Diabetes Register (1995-2019), we used clinic-based HbA1c values to calculate GV using 1) haemoglobin A1c variability score (HVS): percentage of HbA1c values varying by 0.5% compared with prior values, 2) standard deviation of HbA1c (SD_ HbA1c), and 3) SD independent of mean (SDIM) as GV indexes.

Results: We included 15,286 patients with ≥10 years of T2D before cancer occurred or censor date, ≥3 years of observation and ≥5 HbA1c measurements (51.7% male, age: 61.04±10.73 years, HbA1c: 7.54±1.63%, body mass index [BMI]: 25.65±3.92 kg/m2). We excluded 3 years of HbA1c values prior to cancer to avoid reverse causality. There were nonlinear relationships between HVS and cancer events as well as cause-specific death. In patients with HVS above the median (interquartile range) value of 42.31 (27.27, 56.28), every 1 SD increment of HVS increased the fully-adjusted hazard ratio (aHR) of all-site, breast and liver cancer by 1.15 (95% confidence interval: 1.04, 1.26), 1.44 (1.07, 1.94), and 1.37 (1.08, 1.74), respectively. The respective aHRs were 1.21 (1.06, 1.39), 1.27 (1.15, 1.40), and 1.15 (1.09, 1.22) for cancer, vascular, and noncancer nonvascular death. The significant association between HVS and clinical events was observed only in patients with above-median HVS. The aHRs of SD_HbA1c and SDIM for these events were consistent with that of HVS. Using obesity (BMI ≥25 kg/m2) and HVS median as risk stratifiers, the obese/high HVS group had aHRs of 1.42 (1.16, 1.73), 2.44 (1.24, 4.82), and 2.63 (1.45, 4.74) respectively for all-site, breast and liver cancer events versus the non-obese/low GV group. The respective aHRs were 1.45 (1.07, 1.96), 1.47 (1.12, 1.93), and 1.35 (1.16, 1.57) for cancer, vascular, and noncancer nonvascular death.

Conclusion: Obesity and GV were jointly associated with increased risk of cancer events and cause-specific death in T2D. Avoidance of glycaemic fluctuation and obesity might reduce the risk of cancer events and cause-specific death in T2D.

Supported by: This project was supported by the CUHK Direct Grant and CUHK Diabetes Research and Education Fund of the Department of Medicine and Therapeutics, Faculty of Medicine, CUHK.

Disclosure: D. Mao: None.

OP 25 Disparities and diversity in diabetes epidemiology

145

Marked and widening and socio-economic inequalities in prevalence of type 2 diabetes in Scotland

S.H. Wild1, J. Wang2, Scottish Diabetes Research Network epidemiology group;

1Usher Institute, University of Edinburgh, Edinburgh, 2Edinburgh Medical School, University of Edinburgh, Edinburgh, UK.

Background and aims: Socio-economic inequalities in prevalence of type 2 diabetes in Scotland were last described in 2007. Since then widening health inequalities in the United Kingdom in general have been reported. The aim of this study was to describe the association between socio-economic status and type 2 diabetes prevalence in Scotland in 2021 and to compare the findings with those reported in 2007.

Materials and methods: A February 2021 extract of the Scottish electronic population-based register of diagnosed diabetes and mid-year population estimates for 2019 (the most recent year for which data are available) were used to estimate diabetes prevalence for 35-84 year olds. The European Standard Population was used to generate age-standardised prevalence by sex and quintiles of an area-based measure of socio-economic status, the Scottish Index of Multiple Deprivation (SIMD) with Q1 and Q5 used to describe the most and least deprived fifths of the population respectively. Sex-specific relative risks for age-standardised diabetes prevalence for Q1 compared to Q5 were estimated.

Results: Data on age, sex and SIMD were available for 255,764 people (98.9%) of 35-84 years of age with diagnosed type 2 diabetes giving an overall prevalence in this age-group of 8.3% in 2021 (as compared to 7.3% in 2007). Prevalence of type 2 diabetes was lowest in women aged 35-39 years from Q5 (0.5%) and highest in men aged 75-79 years from Q2 (22.4%). The figure shows that age-standardised prevalence was higher for men than for women for all SIMD quintiles and was inversely associated with socio-economic status. Relative risks for age-standardised diabetes prevalence for Q1 compared to Q5 have increased from 2.00 (95% CI 1.52-2.62) in 2007 to 2.48 (95% CI 2.15-2.89) in 2021 for women and from 1.58 (95% CI 1.20-2.07) in 2007 to 1.89 (95% CI 1.73-2.08) in 2021 for men.

Conclusion: Prevalence of type 2 diabetes in Scotland varies by age, sex and socio-economic deprivation. Socioeconomic inequalities in prevalence of type 2 diabetes in Scotland have widened between 2007 and 2021.

figureax

Disclosure: S.H. Wild: Other; The Scottish Diabetes Research Network is supported by National Health Service (NHS) Research Scotland, a partnership of Scottish NHS Boards and the Chief Scientist Office of Scottish Government.

146

Prevalence and characteristics associated with antidepressant and antipsychotic prescribing prior to diagnosis of type 2 diabetes in Scotland

C.R.L. Greene, S.H. Wild, H. Wu, C.A. Jackson;

Centre for Population Health Sciences, The University of Edinburgh, Edinburgh, UK.

Background and aims: The prescribing of antidepressant and antipsychotic drugs is increasing worldwide for a range of indications including mental ill-health. People with diabetes may be prescribed these drugs more often relative to the general population, due to the increased co-morbidity of diabetes and mental illness as well as the use of some types of antidepressants in treating neuropathy. The aim of this study was to describe the prevalence and patterns of antidepressant and antipsychotic drug prescribing prior to diagnosis of diabetes in Scotland.

Materials and methods: The Scottish Care Information-Diabetes Collaboration dataset (SCI-Diabetes) is a registry which includes information gathered routinely in primary and secondary care for almost all patients diagnosed with diabetes in Scotland. We used the SCI-Diabetes dataset to describe the prevalence and patterns of antidepressant and antipsychotic drug prescribing in adults with type 2 diabetes. We classified people as being prescribed antidepressant or antipsychotic drugs if they had received at least one prescription of these drugs in the four years prior to their diabetes diagnosis. Key characteristics were compared within the cohort by prescription history with P values for ANOVA for continuous variables and for chi-squared tests for categorical variables.

Results: Our cohort included 266,186 adults with type 2 diabetes. Of these, 22.5% were prescribed antidepressants, 5.3% were prescribed antipsychotics, and 6.6% were prescribed both antidepressants and antipsychotics. Of people who were prescribed antidepressants, 32.9% were prescribed a selective serotonin reuptake inhibitor, 30.5% were prescribed a tricyclic antidepressant, 9.9% were prescribed an antidepressant of a different subtype, and 26.8% were prescribed antidepressants from multiple subtypes. Of people who were prescribed antipsychotics, 80.4% were prescribed a first-generation antipsychotic, 14.2% were prescribed a second-generation antipsychotic, and 5.5% were prescribed antipsychotics from multiple subtypes. Compared to people not prescribed antidepressant or antipsychotic medication, a greater proportion of people prescribed antidepressants or antipsychotics were women, lived in more socioeconomically deprived areas, were current smokers, were obese, had hypertension, and had high total cholesterol. People prescribed antidepressant or antipsychotic medication were also more likely to have had a hospital admission for a psychiatric disorder.

Conclusion: Antidepressant and antipsychotic prescribing is common prior to diabetes diagnosis among people with type 2 diabetes in Scotland. In general, a prior prescription for these drugs is associated with a poorer risk factor profile. Further work is needed to investigate prescription patterns post-diabetes diagnosis and to determine whether use of these drugs influences the risk of macrovascular and microvascular complications of type 2 diabetes.

figureay

Disclosure: C.R.L. Greene: None.

147

Selection pressures on the ACE2 gene in a Scottish and South Indian type 2 diabetes population

C. Nangia1, S. Srinivasan1, V. Radha2, V. Mohan2, C.N.A. Palmer1;

1University of Dundee, Dundee, UK, 2Madras Diabetes Research Foundation, Chennai, India.

Background and aims: Host-pathogen interaction studies have identified ACE2 gene as the receptor of the SARS-COV2 spike protein, which is implicated in the pathogenesis of the COVID-19 pandemic. The infection has varying levels of severity within and between populations and between the sexes. Severity also varied in people with co-morbidities like Type 2 Diabetes (T2D). The current study thus aims to analyse whether selection pressures have played a role in the pathogenesis of the disease between a Scottish and South Indian population, and its association with T2D.

Materials and methods: The study populations consisted of Scottish T2D individuals from the Genetics of the Scottish Health Research Register (GoSHARE) study (n=6,681) and South Indian T2D individuals from the Madras Diabetes Research Foundation (MDRF) cohort (n=6,056). Pairwise calculation of FST was done on the X Chromosome. High FST Single Nucleotide Polymorphisms (SNPs) from the ACE 2 gene were taken for further analysis. This was followed by tests for selective sweep - Tajima’s D and Nucleotide Diversity. Candidate SNP association studies - sex stratified, and sex adjusted, were done on 13 phenotypes affecting T2D including age at onset, anthropometric measurements, blood pressure and lipids, in both the populations independently. Conditional analysis and gene expression studies were also done.

Results: 28 SNPs within the ACE2 gene had high FST values (FST >0.25). No evidence was found for a selective sweep as indicated by Tajima’s D (D > 0) and Nucleotide diversity (π>0) values. Candidate SNP association analysis showed significant association of these SNPs only in the South Indian population. P value significance was taken as p<0.05 since there was only one independent SNP. 27 SNPs were significantly associated with age at onset in males, 24 in the sex adjusted model (overall). With triglycerides, 16 SNPs were associated in males. 27 SNPs were associated with height, 23 with weight ,25 with waist hip ratio and 5 with BMI overall; 1 each with waist circumference in females and overall. The most significant hits from each association analysis are presented in Table1. Conditional analysis was done for age at onset and triglycerides (sex stratified). In both cases the top SNPs - rs2158083 and rs1978124, respectively, showed significance independent of the others. Gene expression studies showed a higher expression of the gene in males (testis).

Conclusion: The results showed an association of the alleles with only the South Indian cohort. No genotypic association was seen in the Scottish population. These diabetes related alleles have shown positive selection in the South Indian population indicating they may have been beneficial to them in the past. The results also indicate a possible male specific susceptibility to Covid. The results are yet to be confirmed in a Covid-19 infected cohort.

figureaz

Supported by: NIHR Global Health Research Unit on Global Diabetes Outcomes Research. Grant number 16/136/102

Disclosure: C. Nangia: None.

148

Sex, race, and ethnicity representativeness in cardiovascular outcomes trials in type 2 diabetes: a meta-epidemiological study

I. Avgerinos1, T. Karagiannis1, A. Liakos1, P. Kakotrichi1, A. Tsapas1,2, E. Bekiari1;

1Clinical Research and Evidence-Based Medicine Unit, Aristotle University of Thessaloniki, Thessaloniki, Greece, 2Harris Manchester College, University of Oxford, Oxford, UK.

Background and aims: Optimal sex, race, and ethnicity representativeness in clinical trials is necessary to enhance external validity of research findings. We performed a systematic review to evaluate the representation of participants in cardiovascular outcomes trials of type 2 diabetes based on sex, race, and ethnicity.

Materials and methods: We searched Pubmed, Embase, and the Cochrane Library for multi-regional randomised controlled trials assessing the effect of antidiabetic medications on major adverse cardiovascular events in adults with type 2 diabetes. Descriptive statistics were used to summarise demographic data, and chi-square test was used to compare the representation of all race and ethnic groups across eligible trials to the general distribution of these groups in the overall population with type 2 diabetes in the United States (US).

Results: We included 20 trials with 159715 patients. The median number of participating sites per trial was 592 (interquartile range [IQR] 391 to 680), the median number of participating countries per trial was 32 (IQR 23 to 39), and the median age of participants was 64.5 years (IQR 63.3 to 66.1 years). All trials reported demographic data on sex and race (white, black, asian, and other), whereas 12 trials reported data on ethnicity (Hispanic/Latino, not Hispanic/Latino). Across all trials, fewer participants were female comprising 35% (range 22% to 46%, n=56554) of all participants. Most participants were white (77%, n=123013), whereas asian (n=19462) and black (n=7141) participants represented 12% and 4% of the total population, respectively (Figure). In general, these percentages remained consistent in trials published from 2013 onwards (Figure). Compared to the general distribution of racial groups in the overall US population with diabetes, black patients were significantly underrepresented (4% vs 15%, p-value <0.05), whereas white patients were overrepresented (77% vs 57%, p-value <0.05) in cardiovascular outcomes trials. Hispanic/Latino participants represented 16% of all patients (n=18208), which is comparable to the representation of this ethnic group in the overall diabetic population in the US.

Conclusion: Over the last decade, black patients and women are being consistently underrepresented in large cardiovascular outcomes trials in type 2 diabetes. Targeted efforts to increase sex and race representativeness in diabetes research are needed to ensure generalisability of research findings and equity in health care provision.

figureba

Disclosure: I. Avgerinos: None.

149

GWAS reveals a novel locus associated with kidney function in people of Middle-Eastern decent

S.A. Mohamed, J. Tares, P.W. Franks, L. Bennet;

Department of Clinical Sciences Malmö, Lund University, Malmo, Sweden.

Background and aims: Chronic kidney disease is a significant public health problem that is increasing. Type 2 diabetes is one of the leading causes to kidney failure. Middle Eastern immigrants represent one of the largest immigrant populations to Europe today. Despite having a poor metabolic control and high risk for early onset insulin deficient type 2 diabetes and cardiovascular diabetic complications, they have better kidney function and lower mortality rates in all-cause mortality and cardiovascular-related mortality as compared to native Swedes. This study aimed to understand the underlying genetic basis of this difference by performing GWAS of eGFR and ten type 2 diabetes related traits on people of Iraqi ancestry living in Malmo.

Materials and methods: A total of 1201 residents of Malmo born in Iraq were genotyped, imputed using HRC reference panel, and tested for association with eGFR and ten type 2 diabetes-related traits using a linear mixed model.

Results: Out of the eleven phenotypes tested for association, novel loci for fasting glucose (in CAMTA1, NDUFA10, TRIO, WWC1, TRAPPC9, SH3GL2 and ABCC11), quantitative insulin-sensitivity check index (in METTL16), eGFR (in ERBB4), and HbA1C (in CAMTA1, ME1, PAK1 and RORA) was identified at a genome-wide significant level. eGFR also showed a strong signal in a previously reported locus (CST9). It was associated with 107 significant SNPs with an increasing effect on eGFR.

Conclusion: This study, in addition to identifying novel variants, also demonstrated the underlying reason behind the improved kidney function in the Middle Eastern population. The preserved kidney function may be a strong contributor to the survival benefits in Middle Eastern immigrants with type 2 diabetes.

Disclosure: S.A. Mohamed: None.

150

People with type 1 diabetes of African-Caribbean ethnicity are at increased risk of sight-threatening retinopathy

A. Mangelis1, S. Ayis2, A. Nirmalakumaran1, J. Collins3, L. Webster4, S.M. Thomas1, S. Mann4, J. Karalliedde1;

1King's College London, London, 2Medical statistics unit, King's College London, London, 3NHS England, London, 4SEL Diabetic Screening Programme, St Thomas’ Hospital London, London, UK.

Background and aims: There is limited information on the interaction between ethnicity and traditional risk factors on the development of sight-threatening diabetic retinopathy (STDR) in people with type 1 diabetes from ethnically diverse populations. The aim of our study was to describe the risk factors and features of an ethnically diverse cohort of people with type 1 diabetes who develop STDR. STDR was defined by the presence of any moderate to severe non proliferative or pre-proliferative diabetic retinopathy (R2) or proliferative diabetic retinopathy (R3) or maculopathy (M1) in either eye as per UK diabetes eye screening grading.

Materials and methods: Clinical and digital retinal imaging data from 1876 people (50% male, 72.3% Caucasian, 17.4% African-Caribbean, and 10.3% other) with type 1 diabetes as documented from primary care records, without retinopathy at baseline, attending diabetes eye screening in South East London were reviewed. Median duration of follow up was 6 years. Multivariable cox regression analyses was used to estimate hazard ratios for STDR.

Results: The median (interquartile range) age of the cohort was 29 (21, 41) years, duration of diabetes 6.0 (2.0, 12.0) years. Baseline HbA1c was 70.0 (56.8, 94.5) mmol/mol. Cumulative incidence of people with progression from no baseline retinopathy to STDR was 0.1% (95% CI 0.08-0.12) at 5 years, rising to 2.9% (2.5-3.3) at 10 years. The total number of people who developed STDR over 14 years of follow up was 359 (19.1 %). People with STDR had higher baseline HbA1c 77 (60.7, 97.8) vs. 69 (56, 93.9) mmol/mol, raised systolic blood pressure 121 (111, 133) vs. 119 (110, 129) mmHg, longer duration of diabetes 8 (4, 14) vs. 5 (2, 11) years and were more often of African-Caribbean origin (24% vs 15.6%), p<0.05 for all. There were no significant differences in other variables including lipid profiles, body mass index, albuminuria or degree of socio-economic deprivation. In multivariable cox regression analyses, hazard ratio (95% confidence intervals), HbA1c 1.007 (1.004, 1.011), duration of diabetes [1.44 (1.09-1.89) for patients with duration between 10-20 years and 1.52 (1.07-2.16) above 20 years, compared to those with duration of 0-9 years], and African-Caribbean ethnicity 1.43 (1.09-1.88) emerged as independent risk-factors associated with an increased risk of STDR, p< 0.05 for all.

Conclusion: People with type 1 diabetes of African-Caribbean ethnicity are at significantly greater risk of STDR. Further research is required to understand the mechanisms and reasons that may explain this novel observation.

Supported by: GSTT Charity

Disclosure: A. Mangelis: None.

OP 26 Beta cells: sensing, signalling and secreting

151

Gamma aminobutyric acid-induced calcium signalling in the primary cilium of islet beta cells

G. Sanchez1, C. Incedal1, J. Prada2, T. Dandekar2, O. Idevall-Hagren1;

1Dept of Medical Cell Biology, Uppsala University, Uppsala, Sweden, 2Dept. of Bioinformatics, Würzbung University, Würzburg, Germany.

Background and aims: The primary cilium is a rod-like structure that protrudes from the cell surface. It is studded with various receptors and is believed to be a specialised signalling organelle important for probing the local environment. Recent studies have shown that the absence of cilia on β-cells results in aberrant signalling and impaired insulin secretion. Although the primary cilia membrane is continuous with the plasma membrane, it presents with a unique set of receptors and effector proteins. These are selectively granted access to the primary cilium through a gate-keeping mechanism at the cilia base. The cilium is also a unique compartment for Ca2+and cAMP signalling. How these small molecules are prevented from freely diffusing between the cilia and cytoplasm is not known, although that would be required for a local signalling function. The best-characterized example of cilia signaling is the hedgehog pathway, which involves activation of the receptor Smoothened and is important for β-cell development and differentiation. This pathway also extensively cross-talk with other pathways to tune the responses, but the identity of these remains largely unknown. The aim of this study was to identify and characterise cilia-specific signaling pathways in β-cells.

Materials and methods: Cilia and cytosolic Ca2+ concentration changes were measured using custom-made biosensors and TIRF microscopy. Morphometric analysis of islet cell cilia and distribution of various receptors was performed on fixed islets by immunostaining followed by 3D confocal or STED microscopy.

Results: Transduction of mouse islets with an adenoviral vector encoding a cilia-targeted Ca2+ indicator enabled Ca2+ imaging within the cilia lumen. We found that elevation of the cytosolic Ca2+ concentration by either membrane depolarization of carbachol-mediated release from the ER failed to elevate the cilia Ca2+ concentration. In fact, rise of cytosolic Ca2+ resulted in a lowering of cilia Ca2+ concentration, which was particularly apparent in glucose-stimulated islets, where we observed anti-parallel Ca2+ oscillation in the two compartments. The isolation of the cilium against cytosolic Ca2+ involved active Ca2+ extrusion at the cilia base and was enhanced by membrane depolarization and suppressed by inhibition of the Na/Ca-exchanger. Extended imaging of unstimulated islets revealed prominent cilia Ca2+ “flashes” that were synchronized across the islet. Addition of low concentration of GABA (1-10 nM), the GABAB-receptor agonist Baclofen or inhibition of GABA degradation with Vigabatrin increased the occurrence of “flashes” 3-4-fold. Consistent with local GABA action on primary cilia, we find that GABAB-receptors are selectively enriched at the primary cilia base of both mouse and human islet β-cells and become mobile upon agonist binding. GABA stimulation was associated with an acute exit of the hedgehog pathway inhibitor Patched from the β-cell cilia, indicating functional crosstalk between the cilia GABA and hedgehog pathways.

Conclusion: We demonstrate that β-cell cilia are a unique Ca2+ compartment that is isolated against cytosolic Ca2+ changes, and we identify GABA and GABAB receptors as a cilia-selective signaling unit that crosstalk with the hedgehog pathway. Signal integration and modulation at the primary cilium may represent an important general mode of regulation of β-cell function within islets.

Supported by: Novo-Nordisk foundation Emerging Excellent Investigator award, Swedish Research Council

Disclosure: G. Sanchez: None.

152

Involvement of extracellular ATP signalling in the diabetogenic response of pancreatic beta cells

T. Brun1,2, D. Duhamel1,2, L. Oberhauser1,2, C. Jiménez-Sánchez1,2, C. Bartley1,2, V. Lavallard3, P. Maechler1,2;

1Dpt. of Cell Physiology & Metabolism, University of Geneva Medical Center, Geneva, 2Faculty Diabetes Center, University of Geneva Medical Center, Geneva, 3Cell Isolation and Transplantation Center, Geneva University Hospital, Geneva, Switzerland.

Background and aims: Chronic exposure of pancreatic beta-cells to elevated nutrient levels impairs their function. While the use of fructose as a sweetener is associated with obesity, fructose alone does not acutely stimulate insulin secretion, as opposed to the chief secretagogue glucose. Upon glucose stimulation, ATP and ADP are co-secreted with insulin and may activate purinergic receptors. We recently reported that chronic exposure to fructose induces extracellular ATP signalling in beta-cell, resulting in the potentiation of physiological glucose-stimulated insulin secretion (GSIS). This effect is mediated by the activation of the calcium-mobilizer purinergic P2Y1 receptor and is associated with the release of cellular ATP through pannexin-1 channels. However, little is known on extracellular ATP signalling responses to nutrient-rich metabolic stresses. Here, we investigated the expression pattern of genes implicated in the extracellular ATP signalling of beta-cells following metabolic stresses.

Materials and methods: INS-1E beta-cells or freshly isolated human islets were exposed for 3-4 days to glucose (5.5-25 mM), palmitate or oleate (0.4 mM), and fructose (5.5 mM). Following these treatments, we analyzed transcripts levels of several components of extracellular ATP signalling in INS-1E cells by qRT-PCR and in human islets by RNA-Seq, including Panx1, the ecto-ATPase Entpd3 and P2ry1. NTPD3 was also evaluated at the protein level by immunoblotting. The secretory response of INS-1E beta-cells was monitored during acute 8.3 mM glucose stimulation using online luminescence assay.

Results: Diabetogenic conditions reduced expression of the beta-cell enriched NTPD3 and P2Y1 in INS-1E cells (upon glucotoxicity) and in human islets (upon glucolipotoxicity), while PANX1 expression was preserved. Immunoblot analyses revealed that NTPD3 is an abundant glycosylated protein regulated by glucose in INS-1E cells and human islets. Addition of the ecto-ATPase inhibitor ARL67156 or the P2Y agonist 2MeSADP to naive INS-1E cells (i.e. cultured in standard 11 mM glucose media) potentiated the secretory response to 8.3 mM glucose, mimicking effects of chronic fructose treatment. Conversely, clearance of extracellular ATP and ADP to AMP by apyrase reduced stimulated secretion in naive cells and reversed the potentiated secretory response induced by fructose. INS-1E cells cultured at low 5.5 mM glucose exhibited poor responses to 8.3 mM glucose stimulation and neither chronic fructose treatment nor addition of ARL67156 and/or 2MeSADP did restore such blunted secretory response.

Conclusion: These results suggest that ectonucleotidases and purinergic receptors are key components of the ATP extracellular signalling during physiological conditions in pancreatic beta-cells and may represent putative target during pathophysiological conditions.

Disclosure: T. Brun: None.

153

A dual GLP-1/GIP agonist may encompass the beneficial effects of both incretins on pancreatic beta cell function in the absence of beta-arrestin2

N. Zaïmia, S. Costes, S. Dalle, G. Bertrand, M.A. Ravier;

IGF, Univ Montpellier, CNRS, INSERM, Montpellier, France.

Background and aims: Increasing the efficacy of GLP-1 receptor (GLP-1R) agonists by using dual GLP-1/GIP receptor agonist is a new strategy in the treatment of type 2 diabetes. GLP-1R and GIPR are G protein coupled receptors (GPCRs) known to be positively coupled to cAMP production and PKA/EPAC2 activations. Both GPCRs are also able to recruit the scaffold protein beta-arrestin2 (ARRB2), which may activate new signaling pathways such as the kinases ERK1/2 or Focal Adhesion Kinase (FAK). Our aim is to determine whether Tirzepatide, a dual GLP-1/GIP receptor agonist, has a beneficial effect superior to GLP-1 on primary pancreatic beta cells.

Materials and methods: Experiments were performed in beta cells from 4-month-old Arrb2-/- and Arrb2+/+ male mice. Endogenous PKA (AKAR3) and ERK1/2 (EKAR) activations were assessed by live cell imaging in mouse pancreatic beta cells after adenoviral infection with FRET-based sensors of interest. Insulin secretion was measured on isolated islets by homogenous time-resolved fluorescence. F-actin depolymerization was evaluated by phalloidin staining (Alexa Fluor 488-conjugated phalloidin) and the phosphorylation of FAK by immunofluorescence. GLP-1R endocytosis was assessed by immunofluorescence from 4% formaldehyde fixed and non-permeabilised beta cells.

Results: PKA activation and insulin secretion were significantly increased in response to 10pM-100pM GLP-1 (p<0.01) and maximally activated by 100nM-1nM. In contrast, GIP and Tirzepatide were only effective from 1nM (p<0.01), indicating, that Tirzepatide is less potent than GLP-1 on GLP-1R, as already reported in non-beta cells. In addition, GLP-1 induced PKA long lasting activation (>25 min) that was associated with a prolonged internalisation of GLP-1R, whereas the reversal of PKA activation upon GIP stimulation was rapid (<5min), suggesting slow versus fast recycling receptors, respectively. Surprisingly, Tirzepatide induced PKA long lasting activation similar to that of GLP-1, but without any GLP-1R internalisation. In Arrb2-/- beta cells, PKA and EPAC2 activations by GLP-1 or GIP were not affected despite a strong reduction of insulin secretion in response to GIP (~50%, p<0.01) that was caused by a reduction of F-actin depolymerisation (~50%, p<0.01) and FAK activation (~50%, p<0.01). Tirzepatide, like GLP-1, induced a similar insulin secretion, F-actin depolymerisation and FAK activation in Arrb2+/+ and Arrb2-/- beta cells. By contrast, ERK1/2 activation in response to GLP-1 was strongly reduced by ~50% (p<0.05) in Arrb2-/- beta cells while GIP and Tirzepatide recruited ERK1/2 independently of ARRB2.

Conclusion: In beta cells, Tirzepatide combines the beneficial effects of GLP-1 and GIP. Our study reports that this dual GLP-1/GIP agonist may overcome the functional consequences of the decrease in ARRB2 expression that can be observed in diabetogenic conditions.

Supported by: SFD research grant

Disclosure: N. Zaïmia: None.

154

Feeding inhibits the catabolic activity of glutamate dehydrogenase in mouse pancreatic beta cells as revealed by in situ assessment of enzyme activity

Y. Zhou, P. Maechler;

Department of cell physiology and metabolism, University of Geneva, Geneva, Switzerland.

Background and aims: Glutamate dehydrogenase (GDH) is a mitochondrial enzyme playing a key role in the control of insulin secretion. However, its regulation remains unclear, in particular regarding a putative oxidative catabolic activity consuming glutamate. Classical lysate preparations for enzymatic activity assessment might lose the in situ allosteric regulations governing GDH. Here, we developed an in situ redox-sensitive nitro blue tetrazolium (NBT) assay to investigate dehydrogenase activity in cryopreserved pancreatic sections. On subsequent sections, DTZ staining and immunohistochemistry provided corresponding mapping of the cells. The aim of the present study was to investigate in situ islet GDH activity in cryopreserved pancreatic sections.

Materials and methods: We applied the redox-sensitive nitro blue tetrazolium (NBT) assay to investigate in situ pancreatic islet GDH activity. On the same preparations, we measured activities of succinate dehydrogenase for positive control of mitochondrial metabolism, of GAPDH as a readout of glycolytic capacity, and of LDH. The study was conducted in mice of 12 weeks of age. As GDH negative control, we used beta-cell specific GDH knockout mice (Beta-Glud1-/-). Mice were fed a standard diet and sacrificed under different nutritional states: fed, 6-hour fasting, and overnight starvation. Pancreata were collected and stored at -80 C before the preparation and analyses of cryosections. On the same section series, insulin-containing cells were revealed by DTZ staining.

Results: Measured in situ, the NBT-based assay revealed strong activities of both GAPDH and succinate dehydrogenase in insulin-positive cells of cryopreserved islets isolated from fed mice. With prolonged fasting time, GAPDH activity was significantly reduced (p=0.014). LDH in situ activity was absent in islets, whatever the nutritional state. Specifically, in insulin-positive cells, the catabolic GDH flux to glutamate oxidation was activated upon overnight starvation, while such GDH activity was suppressed in fed conditions (p=0.0049). These GDH readouts were undetectable in Beta-Glud1-/- islets. In exocrine non-islet cells, GDH activity was not modulated by the nutritional state.

Conclusion: Overall, our study shows that commonly used assessment of GDH function does not properly reflect in situ activity. The catabolic activity of GDH was inhibited by feeding when beta-cells are stimulated, while glycolytic GAPDH and mitochondrial succinate dehydrogenase were active. The beta-cell disallowed LDH was silent in all tested conditions. The increased glutamate-consuming GDH activity observed upon starvation points to an energy-supporting role for GDH in such conditions.

Supported by: Swiss National Science Foundation

Disclosure: Y. Zhou: None.

155

Alterations in the expression of proteins involved in nuclear-cytoplasmic shuttling of cargo in pancreatic beta cells under the duress of chronic hyperglycaemia

A. Kowluru, S. Elebra, V. Thamilselvan, A. Harajli;

Wayne State University and VA Medical Center, Detroit, USA.

Background and aims: Recent evidence in the islet beta cell suggests critical regulatory roles for Rac1, a small G protein, in the pathogenesis of metabolic dysfunction under the duress of chronic hyperglycemia (HG). Evidence is also emerging to suggest increased nuclear translocation (i.e., mistargeting) of Rac1 in human islets, rodent islets and INS-1 832/13 cells following exposure to HG. Data from quantitative proteomics studies identified Karyopherin-α2 (KPNA2), a nuclear transport protein, as one of the interacting partners for Rac1 in pancreatic beta cells. Therefore, we determined the effects of HG conditions on the expression and subcellular distribution of specific nuclear-cytoplasmic shuttling proteins in pancreatic beta cells.

Materials and methods: Rat islets were isolated by the collagenase digestion method. Human islets were from Prodo Laboratories. INS-1 832/13 cells were cultured in the presence of low (2.5 mM) or high (20 mM) glucose for 24 hrs. for assessing the effects of HG. Nuclear and cytosolic fractions were isolated from INS-1 832/13 cells using the NE-PER Nuclear and Cytoplasmic Extraction Kit (Thermo Fisher Scientific). Expression of candidate proteins involved in nuclear-cytoplasmic shuttling of cargo was quantified by Western blotting and densitometry.

Results: KPNA2 is expressed in human islets, rat islets and INS-1 832/13 cells. Incubation of INS-1 832/13 cells with HG, but not mannitol (osmotic control), markedly increased (2.5-4-fold) the expression of KPNA2. HG conditions elicited minimal effects on the expression of karyopherin-β1. Subcellular fractionation studies indicated significant increase in KPNA2 expression in both cytosolic and nuclear fractions isolated from HG exposed INS-1 832/13 cells compared to cells exposed to basal glucose. A significant increase in nuclear association of Rac1 was also seen under these conditions. HG exposure conditions exerted no effects on the expression of Ran, a small G protein involved the shuttling of cargo between the nuclear and cytosolic compartments. However, a significant increase (2.6-fold) in the expression of Ran GTPase-Activating Protein 1 (RanGAP1), which converts Ran-GTP to Ran-GDP, was noted in beta cells exposed to HG. Lastly, HG exposure conditions did not significantly alter the expression of Regulator of Chromosome Condensation 1 (RCC1), a known guanine nucleotide exchange factor that mediates the conversion of Ran-GDP to Ran-GTP.

Conclusion: Our findings provide the first evidence for potential alterations in the expression of nuclear-cytoplasmic shuttling proteins in pancreatic beta cells under the duress of HG. Based on these data, we propose that HG conditions promote RanGAP1-mediated conversion of Ran-GTP to Ran-GDP, which might dictate directionality of nuclear transport by setting up the gradient of RanGAP1-regulated Ran-GDP level in the cytoplasm and RCC1-regulated Ran-GTP level in the nucleus. Molecular and pharmacological studies are underway to determine potential impact of these alterations in beta cells in the cascade of events involved in the transport and inappropriate localization of Rac1 in the nucleus for propagation of signals necessary for HG-induced metabolic dysfunction of the islet beta cell.

Supported by: VA, NIH

Disclosure: A. Kowluru: None.

156

What regulates insulin granule mobility in the submembrane space? Role of cAMP and adenine nucleotides

B. Gaus, I. Rustenbeck;

Institute of Pharmacology and Toxicology, Technische Universität Braunschweig, Braunschweig, Germany.

Background and aims: Submembrane insulin granules are in constant motion, the extent of which varies widely. While about 30% of the granules appear to be attached with very limited mobility, another 30% spend less than 1 s in the immediate submembrane space before disappearing again. The regulation and energetic requirement of granule mobility are largely unknown. Here, we have investigated the functional consequences of modifying cAMP levels and inhibiting the oxidative phosphorylation.

Materials and methods: MIN6-cells were cultured in DMEM-medium and the insulin granules were labelled by transient transfection with hIns-EGFP. The submembrane granules of constantly perifused cells were imaged by TIRF-microscopy and their number and mobility pattern were analyzed by an observer-independent evaluation program. The insulin secretion by statically incubated MIN6-cells was measured by ELISA, the ATP- and ADP-content was measured by luciferase luminometry and the cytosolic Ca2+ concentration ([Ca2+]i) by Fluo4-fluorescence.

Results: After an equilibration period of 45 min during which the cells were perifused with Krebs-Ringer-Medium containing 3 mM glucose, the cAMP levels were initially raised by perifusion with 1 nM GLP-1. Since this left the mobility pattern virtually unaffected, 75 μM IBMX was used. After a wash-out period the stimulation was repeated in the presence of 30 mM glucose. Again, the effect on the parameters of mobility was only modest, but the number of exocytoses was moderately increased. Acute inhibition of oxidative phosphorylation was induced by perifusion with either 10 μM CCCP or 5 mM Na-Azid or 4 μg/ml Oligomycin. Here, clear consequences for the granule mobility pattern appeared, but these were qualitatively and quantitatively different, depending on the inhibitor. The effects were reversible upon wash-out, corresponding to known kinetics of action. The most marked effects were produced by CCCP, which was also the most effective compound to lower the ATP/ADP ratio, to diminish the rates of exocytosis and at the same time to increase [Ca2+]i. Azide came close to CCCP in reducing the number of exocytoses and its effects showed the fastest reversibility. Oligomycin proved to be as effective as the other compounds to inhibit the secretion by potassium depolarization, even though it was the least effective to influence the granule mobility pattern, the ATP/ADP ratio, and [Ca2+]i.

Conclusion: The mobility pattern of submembrane granules does not appear to be regulated by cAMP, thus the known enhancing effect of cAMP on insulin secretion may not depend on the increased delivery of granules to sites of exocytosis. In contrast, the granule mobility pattern was markedly affected by inhibitors of the oxidative phosphorylation, but their effects in addition to the lowering of the ATP/ADP ratio makes it difficult to ascribe a specific role to the level of adenine nucleotides. The effects on mitochondrial function need to be studied in more detail to clarify their relation with granule mobility and exocytosis.

Supported by: DDG

Disclosure: B. Gaus: Grants; Deutsche Diabetes Gesellschaft.

OP 27 Prediction tools for outcomes in diabetes

157

Prediction models for future complications in type 1 diabetes

N. Al-Sari;

Steno Diabetes Center Copenhagen, Gentofte, Denmark.

Background and aims: Developing panel that predicts the risk of developing diabetes complications (DC) including diabetic nephropathy (microalbuminuria, macroalbuminuria or ≥30% decline in estimated glomerular filtration rate (eGFR)) and diabetic retinopathy (mild, moderate or severe).

Materials and methods: From a cohort of 537 clinically diagnosed adults with type 1 diabetes followed-up a median of 5.4 years at Steno Diabetes Center Copenhagen, we developed prediction models for DC progression. Exploratory analyses included non-matched participants and both tree and network based algorithms. A HbA1cand diabetes duration matched subset of the participants was classified into two groups: T1D stable (n=128), and T1D with progression to DC (n=189). Two random forest models were compared, first using clinical risk factors (16 features, Model 1) and using clinical risk factors with blood small molecule data (965 features, Model 2).

Results: The major contributors to the DC progression Model 1 included eGFR, Age, diabetes duration, BMI and blood triglycerides. The major contributors to Model 2 included eGFR, three metabolites (2,4-dihydroxy butyric acid (DHBA), 3,4-DHBA, creatinine) and a lipid species sphingomyelin d42:1. The machine-learning algorithm (Model 2) including 5 variables provided the best performance as assessed by net reclassification index (NRI, 5%) with an accuracy of 0.81 and the average receiver operating characteristic (ROC)= 0.85 curve with 6-fold cross validation (Fig.1).

Conclusion: A high-performing diabetes complications progression model was developed, with clinical variables and blood metabolites, to predict diabetes complications.

figurebb

Supported by: This work was supported by theNovo Nordisk Foundation (grant number NNF14OC0013659 PROTON Personalising treatment of diabetic kidney disease; internal funding was provided by Steno Diabetes Center Copenhagen, Gentofte, Denmark.

Disclosure: N. Al-Sari: None.

158

Validation of the classification for type 2 diabetes into five subgroups: a report from the ORIGIN trial

M. Pigeyre1, S. Hess2, M.F. Gomez3, O. Asplund3, L. Groop3, G. Pare1, H. Gerstein1;

1McMaster University, Hamilton, Canada, 2R&D, Translational Medicine & Early Development, Biomarkers & Clinical Bioanalyses, Sanofi-Aventis Deutschland GmbH, Frankfurt, Germany, 3Lund University Diabetes Centre, Department of Clinical Sciences, Lund University, Skåne University Hospital, Malmö, Sweden.

Background and aims: Data analyses from Swedish individuals with newly diagnosed diabetes have suggested that diabetes could be classified into five subtypes that differ with respect to the progression of dysglycemia and the incidence of diabetes consequences. We assessed this classification in a multiethnic cohort of participants with established and newly diagnosed diabetes, randomly allocated to insulin glargine versus standard care.

Materials and methods: 7,017 participants from the Outcome Reduction with Initial Glargine Intervention (ORIGIN) trial were assigned to the five predefined diabetes subtypes (namely, severe auto-immune diabetes (SAID), severe insulin-deficient diabetes (SIDD), severe insulin-resistant diabetes (SIRD), mild obesity-related diabetes (MOD), mild age-related diabetes (MARD)) based on the age at diabetes diagnosis, body mass index, glycosylated hemoglobin, fasting C-peptide levels and the presence of glutamate decarboxylase antibodies at baseline. Differences between diabetes subtypes in cardiovascular and renal outcomes were investigated using Cox regression models for a median follow-up of 6.2 years. We also compared the effect of glargine versus standard care on hyperglycemia, defined by having a mean post-randomization glycosylated hemoglobin ≥6.5%, between subtypes.

Results: The five diabetes subtypes were replicated in the ORIGIN trial and exhibited similar baseline characteristics in Europeans and Latin Americans, compared to the initially described clusters in the Swedish cohort. We confirmed differences in renal outcomes, with a higher incidence of events in the SIRD compared to the MARD subtype (i.e., chronic kidney disease stage 3A: HR=1.47, 95%CI [1.30-1.67]; stage 3B: HR=2.20 [1.80-2.69]; macroalbuminuria: HR=1.48 [1.17-1.86]). No differences were observed in the incidence of retinopathy and cardiovascular diseases after adjusting for multiple hypothesis testing. Diabetes subtypes also differed in glycemic response to glargine, with a particular benefit of receiving glargine (versus standard care) in the SIDD subtype compared to the MARD subtype, with a decreased risk of hyperglycemia by 14% (OR=1.37 [1.32-1.43] on glargine; OR=1.51 [1.45-1.58] on standard care; P for interaction subtype*intervention =0.001).

Conclusion: Cluster analysis enabled the characterization of five subtypes of diabetes in a multiethnic cohort. Both the incidence of renal outcomes and the response to insulin varied between diabetes subtypes. These findings reinforce the clinical utility of applying precision medicine to predict comorbidities and treatment responses in patients with diabetes.

Clinical Trial Registration Number: NCT00069784

Supported by: The ORIGIN trial and biomarker project were supported by Sanofi and the CIHR

Disclosure: M. Pigeyre: Other; The ORIGIN trial and biomarker project were supported by Sanofi and the Canadian Institutes of Health Research (CIHR).

159

Machine learning approaches for prediction of nocturnal hypoglycaemia in patients with type 1 diabetes in a hospital setting

V.B. Berikov1,2, R.M. Kozinetz1,2, J.F. Semenova1, V.V. Klimontov1;

1Laboratory of Endocrinology, Research Institute of Clinical and Experimental Lymphology - a branch of the Institute of Cytology and Genetics SB RAS, Novosibirsk, 2Laboratory of Data Analysis, Sobolev Institute of Mathematics, Novosibirsk, Russian Federation.

Background and aims: Nocturnal hypoglycaemia (NH) is a potentially dangerous and underestimated complication of insulin therapy. Minimizing the number of NH events requires reliable prediction. Machine learning algorithms operating on continuous glucose monitoring (CGM) data opens up new possibilities for hypoglycemia forecasting. In this study we aimed to establish an approach to the short-term prediction of NH in hospitalized patients with type 1 diabetes (T1D) based on sets of clinical and CGM data and machine learning techniques.

Materials and methods: We used a dataset of CGM records obtained from 406 adult T1D patients admitted to tertiary referral hospital. The NH was defined as an episode of interstitial glucose level <3.9 mmol/L for at least 15 min in the interval from 0 to 6 am. CGM-derived metrics, including mean glucose and glucose range, gradient, linear, quadratic and cubic trend coefficients, dynamic time wrapping similarity-based indices, and a number of glucose variability parameters, were used as potential NH predictors. At the next step, clinical parameters were added to the models. In total, we estimated 113 variables. Visualization of CGM data with t-distributed stochastic neighbor embedding technique was performed. The sub-sampling of mostly representative records without NH was applied to get more balanced distribution of classes. To this end, we used the dynamic time wrapping similarity metric between time series. Augmentation methodology, including perturbation with small Gaussian noise and weighted averaging of NH records, was applied for generating artificial CGM records. Random Forest method was used for the NH forecasting and estimation of the predictors. The quality of prediction was evaluated on a non-augmented test sample using Monte-Carlo cross-validation.

Results: Depending on predictor sequence length (30 minutes - 1 hour) and horizon of forecasting (5-30 minutes), we have analyzed 209-256 CGM records with at least one episode of NH and near 4∙104 records without NH. Cluster analysis has revealed that the CGM data were rather homogeneous: a number of closely spaced clusters have been found. The quality of forecasting did not depend significantly on predictor sequence length; therefore, we used 30 minutes as a baseline. The sensitivity and specificity varied from 98% and 97% respectively for a 5-minute horizon to 85% and 87% for a horizon of 30 minutes. However, the positive predictive values of the models were low (less than 5-10%) due to the unbalanced nature of the data. Mean and minimal glucose levels, linear trend coefficient and downward 5-minute gradient were the most important predictors assessed by Random Forest. Incorporating clinical data into the models, as well as the augmentation procedures, improved the quality of prediction for a horizon of 30 minutes giving an increase in sensitivity and specificity by 3-7%.

Conclusion: The results demonstrate that machine learning based on the clinical and CGM data is a promising approach for prediction of NH events in patients with T1D in a hospital setting.

Supported by: RSF (20-15-00057)

Disclosure: V.B. Berikov: None.

160

Non-invasive prenatal diagnosis of fetal genotype in pregnant women with GCK-MODY: the impact of precision medicine on antenatal care

A.E. Hughes1, J.A.L. Houghton2, A.J. Chakera3, G. Spyer4, M.H. Shepherd1,5, S.E. Flanagan1, A.T. Hattersley1,5;

1Institute of Biomedical and Clinical Science, Exeter, 2Exeter Genomics Laboratory, Exeter, 3Brighton and Sussex University Hospitals NHS Trust, Brighton, 4Torbay and South Devon NHS Foundation Trust, Torquay, 5Exeter NIHR Clinical Research Facility, Exeter, UK.

Background and aims: Birthweight and pregnancy outcomes in pregnancies affected by GCK-MODY are determined by fetal genotype. Fetuses who do not inherit the GCK variant from their mother are at risk of being large-for-gestational-age (LGA, birthweight >90th percentile for sex and gestational age), whereas fetuses who do inherit the maternal variant have normal birthweights. It is not certain if treatment of maternal hyperglycaemia is beneficial and it may reduce growth in fetuses with GCK-MODY. Accelerated growth of the fetal abdominal circumference (AC) >75th percentile for gestational age may indicate that the fetus has not inherited the maternal variant, but is imprecise. Non-invasive prenatal diagnosis (NIPD) is highly sensitive and specific for identifying fetal genotype and has recently become available. We aimed to investigate the impact of NIPD on management of pregnancies affected by GCK-MODY.

Materials and methods: We investigated 29 international pregnancies affected by maternal GCK-MODY who underwent NIPD at Exeter Genomics Laboratory between July 2019 and March 2021. Of these, 12 women had delivered and had available information on antenatal care and pregnancy outcome.

Results: Women were referred across pregnancy (median 11 weeks, range 5-32 weeks), with the fetal genotype reported within a median of 4 weeks (range 2-15 weeks) from receipt of the first maternal blood sample. Women referred in the first trimester (n=17) received a diagnosis by a median gestational age of 19 weeks (range 15-31 weeks). Of the 12 women who had delivered, 7 had an unaffected fetus and 5 had an affected fetus. Four women with an unaffected fetus underwent treatment for hyperglycaemia with insulin and/or metformin in pregnancy, but only two received treatment until delivery. Three women with an affected fetus received treatment with insulin or metformin in pregnancy, but they were stopped prior to delivery. Insulin was stopped in two women as a result of the affected fetal genotype reported by NIPD. An AC >75th percentile was common, present in 5 of the 7 unaffected fetuses and 4 of the 5 affected fetuses on at least one ultrasound scan performed after 20 weeks gestation. All women with an unaffected fetus had an induction or caesarean delivery before term, whereas all women with an affected fetus had a vaginal delivery, 3 of which were spontaneous. Two babies were born LGA and both were unaffected by GCK-MODY.

Conclusion: We have shown for the first time the impact of NIPD on the management of pregnancies affected by maternal GCK-MODY. Rapid and early diagnosis prior to the start of growth scans is possible and is more reliable; an AC >75th percentile was common in fetuses with and without a GCK-MODY variant, despite only two babies being LGA. We also showed how NIPD could be used to guide treatment decisions, as two women with an affected fetus had their insulin therapy stopped. Diagnosis of an unaffected fetus will help plan delivery decisions, where the risk of LGA is high. NIPD could be used to assess whether treatment of maternal hyperglycaemia reduces birthweight in unaffected fetuses as part of a prospective study, as this remains a key area of practice requiring more investigation.

Supported by: Wellcome Trust, Royal Society, NIHR, University of Exeter

Disclosure: A.E. Hughes: Grants; Wellcome Trust GW4-CAT PhD Fellowship, University of Exeter.

161

Validation of fear of hypoglycaemia screener: results from the T1D Exchange Registry

J. Liu1, J.-L. Poon2, J. Bispham1, M. Perez-Nieves2, A. Hughes1, K. Chapman1, B. Mitchell2, F. Snoek3, L. Fisher4;

1T1D Exchange, Boston, USA, 2Eli Lilly and Company, Indianapolis, USA, 3Department of Medical Psychology, Amsterdam University Medical Centers, Amsterdam, Netherlands, 4Department of Family and Community Medicine, University of California San Francisco, San Francisco, USA.

Background and aims: To examine reliability and validity of a newly developed fear of hypoglycaemia (FoH) screener as a practical and actionable tool for in-clinic use in adults with type 1 diabetes (T1D), in accordance with American Diabetes Association’s position on psychosocial care.

Materials and methods: In this validation study, adults with T1D were recruited from the T1D Exchange Registry to complete a draft screener online; potential items were previously identified from literature review, and interviews with health care professionals (HCPs) and people with T1D. Standard psychometric analyses assessed reliability and validity of the screener. The final FoH screener comprised 9 items assessing 2 domains - “worry” (6 items) and “behaviour” (3 items).

Results: The final sample comprised 592 adults with T1D (age 43.1 ± 15.3 years; duration of T1D 24.1 ± 15 years, 66.7% females, 91.6% White, 5.2% Hispanic, self-reported HbA1c 54.1 ± 13.1 mmol/mol [7.1% ± 1.2%]). Approximately 30% of participants reported severe hypoglycaemia in the past 12 months; 33.4% reported impaired awareness of hypoglycaemia. The FoH screener showed internal consistency (Cronbach’s α=0.88) and was highly correlated (r=0.71-0.75) with the Hypoglycemia Fear Survey (“worry” and “behaviour” subscales and total scores), confirming reliability. Construct validity of the FoH screener was demonstrated with significant correlations with depression (r=0.44), anxiety (r=0.47), Diabetes Distress Subscales (powerlessness, management distress, hypoglycaemia distress) (r=0.49-0.66). These correlations are considered moderate. Additionally, multivariable regression analysis showed that higher FoH screener scores were significantly associated with higher HbA1c (regression coefficient, β=0.04) and multi-morbidities (β=0.03).

Conclusion: This 9-item FoH screener demonstrated good reliability and validity. Further research is planned to assess clinical usability to help appropriately identify patients and assist effective HCP-patient conversations around FoH.

Supported by: Eli Lilly and Company

Disclosure: J. Liu: Employment/Consultancy; T1D Exchange.

162

Mental illness, ethnicity and civil status are associated with non-attendance in diabetic retinopathy screening among people with type 2 diabetes

G.B. Petersen1,2, S. Byberg1, M.V. Fangel2, D. Vistisen1, L.E. Joensen1, H. Vorum3, J.K. Kristensen2;

1Steno Diabetes Center Copenhagen, Gentofte, 2Center for General Practice Aalborg University, Aalborg, 3Department of Ophthalmology Aalborg University Hospital, Aalborg, Denmark.

Background and aims: Diabetic retinopathy can cause severe vision loss and blindness, if not detected early and treated. The key to early detection is screening, and in Denmark, a nationwide screening programme for diabetic retinopathy has been implemented. However, current participation in eye screenings is below the recommended levels. The aim of this study was to identify factors associated with non-attendance in screening for diabetic retinopathy among people with type 2 diabetes.

Materials and methods: We performed a retrospective, observational study using individual-level register data from 156,878 people with type 2 diabetes aged 40-70 years at diagnosis. Based on their eye screening history from 2013 till 2018 each person was characterised as either an attender (at least one registered screening) or a never-attender (no registered screening). We compared baseline characteristics between attenders and never-attenders using non-parametric tests. Eleven characteristics were included: Age, gender, ethnicity, place of residence (region), civil status, education, employment, income, diabetes duration, comorbidity and mental health. Among the attenders, we assessed 230,173 screening intervals and defined them as non-attender intervals if the person failed to participate in the eye screening within the recommended interval. Mixed-effects models were used to investigate the impact of the eleven characteristics on the likelihood of having a non-attender interval. Univariable and multivariable analyses including all characteristics were performed.

Results: A total of 42,068 (27%) individuals were identified as never-attenders, having no registered eye screening, and comparisons showed that attenders and never-attenders differed significantly on all included baseline characteristics. Compared to attenders, never-attenders were more frequently divorced, 30% vs. 22% (p<.0001), living in the Capital Region, 36% vs. 28% (p<.0001), mentally ill 10% vs. 5% (p<.0001) and had more comorbidities, 22 % vs. 17 % (p<.0001). For the remaining factors the quantitative differences were modest. Among the 230,173 screening intervals, 62,381 (27%) were identified as non-attender intervals. All characteristics except gender were significantly associated with the likelihood of having a non-attender interval in the univariate analysis. In the multivariate analyses the five factors with largest odd ratios (ORs) for non-attendance were mental illness (OR: 1.50 [1.42, 1.58], p<.0001), non-western descent (OR: 1.42 [1.35, 1.50], p<.0001), divorce (OR: 1.26 [1.22, 1.31], p<.0001), comorbidity (OR: 1.25 [1.19, 1.30], p<.0001) and region with highest ORs for the Capital Region and North Denmark Region.

Conclusion: This is the first national study to outline factors associated with non-attendance among people with type 2 diabetes in the Danish screening programme for diabetic retinopathy. Our findings suggest that both never-attendance and non-attendance are more common among people who are divorced and among people of poorer health. Additionally, non-attendance is more frequent among people of non-western decent. These population subgroups may benefit from targeted interventions aimed at increasing participation in eye screenings.

Disclosure: G.B. Petersen: None.

OP 28 Pathogenic mechanisms of complications

163

The gut microbiome composition is altered in long-standing type 1 diabetes and associated with disease-related complications

J.I.P. van Heck1, R. Gacesa2,3, R. Stienstra1,4, R.K. Weersma2, L.A.B. Joosten1, C.J. Tack1;

1Internal Medicine, Radboud UMC, Nijmegen, 2Department of Gastroenterology and Hepatology, University Medical Center Groningen, Groningen, 3Department of Genetics, University Medical Center Groningen, Groningen, 4Division of Human Nutrition and Health, Wageningen University & Research, Wageningen, Netherlands.

Background and aims: Type 1 diabetes (T1D) is associated with an increased risk for infections and an increased risk to develop long-term (micro and macro-)vascular complications. Although changes in the microbiome have been linked to the risk of development of T1D, not much is known about the role of the gut microbiome in long-standing type 1 diabetes. We therefore set out to determine differences in the gut microbiome of T1D patients compared to healthy controls and to associate the gut microbiome with diabetes-related parameters.

Materials and methods: 239 T1D patients were included with an average disease duration of 28,4 (15.6) years. Microbiome data were compared to a healthy cohort consisting of 2937 age-,sex- and BMI-matched individuals. Clinical characteristics and faecal samples were collected. Metagenomic shotgun sequencing was performed and the results were associated to T1D-related characteristics such as HbA1c, and macro- and microvascular complications.

Results: 43 bacterial taxa were significantly depleted in T1D, including S. Alistipes Putredinis (FDR=3.0x10-12). Furthermore, 37 bacterial taxa were significantly enriched in T1D, such as G. Clostridium (FDR=2.0x10-5). Interestingly, enriched species consisted mainly of opportunistic species (Clostridiales, Oscillibacter), whereas a reduction in commensal species was seen (Dorea sp, Bifidobacterium sp.). However, no significant difference in diversity of the gut microbiome was found between T1D patients and healthy controls. Several diabetes related factors displayed a significant association with changes in the gut microbiome. Glycaemic control, measured by HbA1c (ranging from 34 to 136 mmol/mol), and disease duration explained a significant part of variation in the gut microbiome (R2>0.010, FDR<0.05). HbA1c was also significantly associated with the presence of several microbial species. Furthermore, both micro-and macrovascular complications explained a significant part of variation in gut microbiome (R2>0.0075, FDR<0.05). Presence of nephropathy was strongly associated with several microbial species, consisting of Clostridiales (FDR<0.05). Macrovascular complications displayed similar associations with Clostridiales (p-value<0.05).

Conclusion: While the diversity is not affected, the composition of the gut microbiome in T1D patients differs significantly from the microbiome of healthy controls. Furthermore, the changes in the gut microbiome are associate with T1D related characteristics and vascular complications. These data suggest that the gut microbiome is not only important in the context of disease development, yet may also be involved in the development of diabetes-associated complications.

Supported by: TIMID project LSHM18057-SGF

Disclosure: J.I.P. van Heck: Grants; collaborative TIMID project: LSHM18057-SGF.

164

No association between haptoglobin genotype and cerebral small-vessel disease in type 1 diabetes

M.I. Eriksson1,2, A. Syreeni1,2, E.H. Dahlström1,2, N. Sandholm1,3, C. Forsblom1,3, D. Gordin1,3, T. Tatlisumak4,5, J. Putaala4, P.-H. Groop1,3, J. Martola6,7, L.M. Thorn1,8, on behalf of the FinnDiane Study Group;

1Folkhälsan Institute of Genetics, Folkhälsan Research Center, Helsinki, Finland, 2Research Program in Clinical and Molecular Metabolism, University of Helsinki, Helsinki, Finland, 3Department of Nephrology, University of Helsinki and Helsinki University Hospital, Helsinki, Finland, 4Department of Neurology, University of Helsinki and Helsinki University Hospital, Helsinki, Finland, 5Department of Neurology, Sahlgrenska University Hospital, Gothenburg, Sweden, 6Department of Clinical Neurosciense, Karolinska University Hospital and Karolinska Institute, Stockholm, Finland, 7Department of Radiology, University of Helsinki and Helsinki University Hospital, Helsinki, Finland, 8Department of General Practice and Primary Health Care, University of Helsinki and Helsinki University Hospital, Helsinki, Finland.

Background and aims: It has been proposed that susceptibility to cardiovascular disease in diabetes could partly be ascribed to a variation in the haptoglobin (Hp) gene. Haptoglobin is a protein that binds to free haemoglobin, inhibiting its oxidative activity. Although Hp1 allele has been suggested to associate with stroke and white matter hyperintensities (WMH) in type 1 diabetes, data on these associations are sparse. Our aim was to further evaluate the association between Hp and cerebral small-vessel disease (SVD).

Materials and methods: For this cross-sectional study, we included 179 individuals with type 1 diabetes from the Finnish Diabetic Nephropathy study. A total of 53% were female, mean age was 39±7 years, duration of diabetes 23±10 years, and HbA1c 65±12 mmol/mol. All participants underwent brain MRI and clinical investigation. Hp genotype was determined with two polymerase chain reactions showing the presence (Hp2) or absence (Hp1) of exons 5 and 6. Brain MRIs were assessed for WMHs, cerebral microbleeds (CMB), and lacunar infarcts as signs of SVD and analysed in relation to Hp genotype.

Results: Genotype Hp1-1 was observed in 28 (16%), Hp1-2 in 78 (44%), and Hp2-2 in 73 (41%) of the participants. Participants with Hp1-1 differed from those with Hp2 alleles (Hp1-2 or Hp2-2) only regarding diastolic blood pressure (76±9 vs 80±6 mmHg, p=0.027). The same was seen in comparison across all three genotypes (p=0.019). In analysis of Hp1-1 vs Hp2 allele carriers, we detected no difference in the prevalence of SVD (36% vs 34%, p>0.999), or in any observed manifestation (WMH 29% vs 22%, p=0.595; CMB 25% vs 22%, p=0.904; lacunar infarcts 0% vs 3%, p>0.999). Hp1-1 was not associated with SVD in a logistic regression model adjusted for diastolic blood pressure, age, and diabetic retinopathy (OR 1.09 95% CI 0.44-2.61, p=0.855). When SVD manifestation were analysed separately, Hp1-1 was associated neither with WMH, CMB, nor lacunar infarcts. Correspondingly, comparison of all genotypes yielded no differences in prevalence of SVD (36% vs 37% vs 32%, p=0.758), nor in any observed manifestation (WMH p=0.117, CMB p=0.860, lacunar infarcts p=0.343). Furthermore, no association between the number of Hp2 alleles and SVD was observed.

Conclusion: Although the prevalence of white matter hyperintensities and cerebral microbleeds was high in our cohort of neurologically asymptomatic adults with type 1 diabetes, we observed no association between cerebral small-vessel disease and haptoglobin genotype.

Supported by: Folkhälsan Research Foundation, Academy of Finland, Stockmann Foundation, EVO governmental grants

Disclosure: M.I. Eriksson: Grants; Wilhelm and Else Stockmann Foundation, Otto-Malm foundation. Stock/Shareholding; Shareholder in BCB Medical.

165

Nox5 in circulating peripheral blood mononuclear cells: a potential biomarker in unstable diabetic vascular and renal disease

T.J. Block1, K.C. Sourris1, W.A. Khan1, P. Kantharidis1, J.C. Jha1, M.E. Cooper1, J. Shaw2, K. Jandeleit-Dahm1,3;

1Department of Diabetes, Monash University, Melbourne, Australia, 2Department of Cardiology, Alfred Health, Melbourne, Australia, 3Leibniz Chair, Heinrich Heine University, Dusseldorf North Rhine-Westphalia, Germany.

Background and aims: Oxidative stress derived from the human NADPH oxidase 5 (NOX5) plays a critical role in diabetic vascular and renal disease and is expressed and functionally active in monocytes and macrophages. Increased NOX5 expression has been demonstrated in atherosclerotic plaques in those with diabetes and coronary artery disease (CAD) as well as in the glomeruli and mesangial cells in kidney biopsies of diabetic patients. Therefore, we hypothesise that NOX5 expression in circulating peripheral blood mononuclear cells (PBMCs) is increased in patients with diabetes, particularly in those with unstable CAD and chronic kidney disease (CKD).

Materials and methods: 46 males aged 33-83 years underwent elective or emergency coronary angiography/angioplasty at the Alfred Hospital Catheter Laboratory. PBMCs were isolated from whole blood and processed for flow-cytometry to measure NOX5 protein. In parallel, NOX5 was measured in PBMCs by qPCR. To complement the in vivo findings, human macrophages (THP-1) were incubated in low (5 mM) and high glucose (25 mM). NOX5 expression and inflammatory markers were measured by qPCR.

Results: NOX5 protein expression in PBMCs was primarily driven by expression in monocytes (CD 45+/CD14+ cells) and was increased in diabetic and non-diabetic patients with CKD (29.5±4.4 versus 18.2±1.9 AU; p=0.0093) and in diabetic patients with CKD versus without CKD (28.4±4.3 versus 16.5±2.2 AU; p=0.03). CAD with acute presentation was associated with increased NOX5 expression versus elective presentation (26.9±3.3 vs 18.2±2.2 AU; p=0.02), particularly in diabetic patients presenting acutely versus electively (30.2±4.4 vs 15.2±2.4 AU; p=0.0046). A 4-fold upregulation of NOX5 gene was observed in patients with CKD versus patients without CKD irrespective of diabetes or CAD (p=0.018). In vitro, there was a 2-fold increase in NOX5, TNF-alpha, and Interleukin-6 expression in THP-1 cells exposed to high glucose.

Conclusion: The presence of CKD and unstable CAD appear to be the key factors for increased NOX5 protein and gene expression in circulating PBMCs in diabetic patients. These findings are consistent with postulated pathogenic mechanisms whereby NOX5 accelerates vascular and renal inflammation and fibrosis. Measurement of NOX5 in PBMCs may serve as a valuable prognostic marker in patients with clustering diabetic complications and lead to targeted interventions in patients at high cardiovascular and/or renal risk.

Supported by: Alfred Seeding Grant

Disclosure: T.J. Block: None.

166

Dicarbonyl stress alters mitochondrial protein homeostasis in endothelial cells

R. Bulkescher1, S. Herzig2, J. Szendrödi1, P.P. Nawroth1, J. Zemva1;

1Department of Internal Medicine I and Clinical Chemistry, University Hospital Heidelberg, Heidelberg, 2Helmholtz Center Munich, Neuherberg, Germany.

Background and aims: The main source for advanced glycation end products (AGEs) during hyperglycemic conditions is the reactive dicarbonyl methylglyoxal (MG), a byproduct of glycolysis. AGEs are formed by non-enzymatic glycation of proteins, lipids and DNA. MG-derived hydroimidazolone (MG-H1) is formed in a reaction with arginine side chains of proteins. This modification can lead to misfolding, aggregation and loss-of-function of the protein. Heat shock proteins (Hsp) are part of the protein quality control system (PQS). They are involved in maturation of newly synthesized polypeptides, refolding of misfolded proteins and maintenance of mitochondrial activity. We questioned the effect of MG on mouse cardiac endothelial cells (MCECs), and whether the loss of the stress-inducible Hsp70 (Hspa1a/Hspa1b) in a double knockout cell line (Hspa1a/Hspa1b KO) leads to reduced MG tolerance.

Materials and methods: mRNA was measured by qPCR and protein levels were measured by immunofluorescence. The data for the KO cells was obtained from three independent Hspa1a/Hspa1b KO cell clones.

Results: After treatment with 500 μM MG, Hspa1a/Hspa1b was up-regulated on the mRNA level (Hspa1a: mean= 1.57-fold, n= 4, p= n.s.; Hspa1b: mean= 1.78-fold, n= 4, p= n.s.) and on the protein level (mean= 1.56-fold, n= 3, p= n.s.) in wild-type cells (WT). The two mitochondrial chaperones Hspa9 and Hspd1 were both up-regulated after 500 μM MG treatment on the mRNA level (Hspa9: mean= 1.42-fold, n= 4, p= 0.029; Hspd1: mean= 2.11-fold, n= 4, p= 0.0054) and on the protein level (Hspa9: mean= 1.96-fold, n= 6, p= 0.000014; Hspd1: mean= 2.35-fold, n= 3, p= 0.0017). Compared to WT, treating Hspa1a/Hspa1b KO cells with 500 μm MG resulted only in a slight increase in mRNA expression of Hspa9 (mean= 1.52-fold, n= 3, p= n.s.) and Hspd1 (mean= 1.57-fold, n= 3, p= n.s.), whereas there was a significantly higher increase of Hspa9 (mean= 3.17, n= 3, p= 0.00044) and Hspd1 (mean= 3.84, n= 3, p= 0.00047) protein expression. The mitochondrial fission protein Drp1 was increased 1.39-fold on the mRNA level after 500 μM MG treatment (n= 3, p= 0.0097) in WT cells, indicating a disturbance in mitochondrial homeostasis. Drp1 mRNA levels in KO cells were also increased compared to WT (WT: mean= 1.39, n=3 vs. KO: mean= 2.11, n= 3, p= 0.0080), suggesting an aggravated mitochondrial disturbance. However, mitophagy related proteins Parkin, Bnip3 and Pink1 were all down-regulated (Parkin: mean= 0.45‑fold, n= 4, p= 0.0066; Bnip3: mean= 0.39-fold, n= 4, p< 0.000001; Pink1: mean= 0.69‑fold, n= 4, p= 0.0025).

Conclusion: In both, WT and Hspa1a/Hspa1b KO cells, mitochondrial stress was present after MG treatment. Drp1 was up-regulated, indicating increased mitochondrial fission and disturbance of mitochondrial homeostasis. As a compensatory mechanism, expression of the mitochondrial heat shock proteins Hspa9 and Hspd1 was significantly induced. In Hspa1a/Hspa1b KO cells, up‑regulation of Drp1 and of mitochondrial Hspa9 and Hspd1 was even stronger, indicating pronounced MG-induced mitochondrial stress in the absence of stress-inducible Hsp70.

Supported by: CRC1118

Disclosure: R. Bulkescher: None.

167

Insulin resistance associates with arterial stiffness in type 1 diabetes: a novel component of double diabetes

G. Llaurado1,2, A. Cano3, L. Albert3, I. Mazarico3, A. Romero3, O. Giménez-Palop3, S. Fernández-Veledo4,2, J. Vendrell4,2, J.-M. González-Clemente3,2;

1Department of Endocrinology and Nutrition, Hospital del Mar, Institut Hospital del Mar d'Investigacions Mèdiques (IMIM), Barcelona, 2Centro de Investigación Biomédica en Red de Diabetes y Enfermedades Metabólicas Asociadas (CIBERDEM), Instituto de Salud Carlos III, Madrid, 3Department of Endocrinology and Nutrition, Parc Taulí Hospital Universitari, Institut d’Investigació i Innovació Parc Taulí I3PT, Universitat Autònoma de Barcelona, Sabadell, 4Department of Endocrinology and Nutrition, Joan XXIII University Hospital, Institut d'Investigació Sanitària Pere Virgili (IISPV), Rovira i Virgili University, Tarragona, Spain.

Background and aims: The presence of insulin resistance (IR) in type 1 diabetes (T1D), known as double diabetes, increases the risk of both macro- and microvascular complications. When IR occurs, insulin impairs its metabolic effects while preserves its growth/mitogenic activities. In the arterial wall, IR could promote arterial stiffness (AS, or early vascular aging), a well-known cardiovascular risk factor, through several mechanisms (e.g. vasoconstriction, inflammation, hypertrophy, fibrosis -collagen synthesis-). The current study was aimed to evaluate the potential association between IR and AS in well -characterized cohort of people with T1D.

Materials and methods: 179 adults with T1D without established cardiovascular disease were evaluated. IR was assessed by the estimation of the glucose disposal rate -eGDR- in mg/kg/min as follows: eGDR=21.158-(0.09xWHR)-(3.407xHypertension)-(0.551xHbA1c). Subjects were categorized into four quartiles according to their eGDR (Q1:<6.1; Q2:6.1 to 8.6, Q3 8.6 to 10, and Q4 >10). AS was assessed by the measurement of aortic pulse wave velocity (aPWV).

Results: The characteristics of subjects included in the study are shown in Table 1, stratified by eGDR quartiles. aPWV increased in parallel with IR, i.e., it was negatively correlated with eGDR (r=-0.589; p<0.001). After adjustments for age, sex, diabetes duration, dyslipidemia and microvascular complications, the association between aPWV and eGDR still remained significant (R2: 0.546. β=-0.218; p=0.001), thus, being IR one of the main independent factors associated with the increase of AS. When each component of the eGDR equation were introduced in the same model separately, only WHR was significantly associated with aPWV (R2: 0.589. β=0.376; p<0.001).

Conclusion: Insulin resistance measured by eGDR is associated with arterial stiffness in people with T1D and no previous established cardiovascular disease. The main factor explaining this association seems to be abdominal obesity instead of glycemic control or hypertension.

figurebc

Supported by: PI15/00567. ISCIII

Disclosure: G. Llaurado: None.

168

Wnt regulation and collagen gene expression in the bone of type 2 diabetes elderly women

G. Leanza, F. Tramontana, F. Cannata, A. Piccoli, V. Viola, M. Faraj, R. Strollo, P. Pozzilli, N. Napoli;

Endocrinology and Diabetes, Campus Bio-Medico University of Rome, Roma, Italy.

Background and aims: Fragility fractures are a skeletal complication associated with type 2 diabetes (T2D), implying disability, hospitalization, impaired quality of life, and increased mortality. The interaction between bone metabolism, higher risk of fractures, and T2D is complex and not fully understood. Increased circulating sclerostin levels and accumulation of advanced glycation end-products (AGEs) are two potential mechanisms underlying low bone turnover and increased fracture risk. Recent data from our group have shown that T2D affects the expression of genes controlling bone formation (SOST and RUNX2) and that accumulation of AGEs not only higher in diabetics but it is associated with impaired bone microarchitecture. We hypothesized that T2D and increased SOST gene expression lead to a downregulation of Wnt β- catenin genes target (transcription factors TCF/LEF1) and AGEs accumulation is correlated to a downregulation of collagen (COL1A1) in bones from T2D subjects. The aim of this study was to investigate gene expression levels of TCF-LEF1, Wnt β -catenin inhibitor Dickkopf -1 (DKK-1), COL1A1, and to evaluate correlation analysis of AGEs bone content previously measured in the same population with above mentioned bone target genes to study underlying mechanisms of increased bone fragility in T2D.

Materials and methods: Bone tissues were obtained from femoral heads of 14 T2D (HbA1c 6.5 ± 1.7 %) and 21 non-diabetic postmenopausal women (Age >65 years) undergoing hip replacement surgery. Biopsies of trabecular bone were isolated from the surgical specimens. RNA was extracted with standard Tryzol protocol and gene expression of TCF-LEF1, DKK-1, and COL1a1 was measured by Real Time- PCR.

Results: Age (non-diabetic 73.2 ± 5.8 years vs diabetic 75.2 ± 8.5 years) and, BMI (non-diabetic 27.7 ± 5.6 kg/m2 vs diabetic 29.9 ± 5.4 kg/m2) did not differ between the two groups. We found that expression levels of TCF-LEF1 were significantly lower in T2D compared to non- diabetic subjects (p=0.002). DKK-1 was not different between groups (p=0.1083), however analysis of correlation showed that DKK-1 increases with age (r2=0.038; p=0.043) and HbA1c (r2=0.503; p=0.048) in T2D subjects. COL1A1 gene expression was lower in T2D compared to controls although significance was not fully reached (p=0.0564). Importantly, correlation analysis of AGEs bone content with LEF1 gene expression levels showed a negative association (r=-0.55; p=0.02), as well as for COL1A1 (r=-0.61; p=0.01).

Conclusion: Our data show for the first time that even in patients with a relatively good glycemic control, T2D affects the expression of Wnt β-catenin target genes and collagen. We also found that accumulation of AGEs in the bone of T2D is negatively associated with alteration in COL1A1 and TCF/LEF1 gene expression. We provide novel insights that may help understand the mechanisms underlying bone fragility in T2D.

Supported by: Internal grant, campus Bio-Medico University of Rome

Disclosure: G. Leanza: None.

OP 29 Understanding muscle and liver metabolism

169

Assessing the association between metabolic flexibility measured upon insulin stimulation and during incremental submaximal exercise

M. Bergman1, R.F. Mancilla1, Y.M.H. Bruls1,2, P. Schrauwen1, V.B. Schrauwen-Hinderling1,2, M.K.C. Hesselink1;

1Nutrition and Movement Sciences, Maastricht University, Maastricht, 2Radiology and Nuclear Medicine, Maastricht University Medical Center+, Maastricht, Netherlands.

Background and aims: Metabolic flexibility is defined as the capacity of substrate oxidation to adapt to substrate availability, commonly measured as the change in respiratory exchange ratio (RER). In obese, metabolically compromised individuals, metabolic flexibility is reduced, reflected by a blunted change from fat to glucose oxidation upon insulin stimulation during a hyperinsulinemic-euglycemic clamp. Reduced metabolic flexibility might be due to low mitochondrial oxidative capacity but may also merely be a reflection of reduced insulin-stimulated glucose uptake. As exercise of progressive intensity also requires a gradual switch from fat to glucose oxidation but is not dependent on insulin for its glucose uptake, we here investigate if metabolic flexibility (measured as the change in RER) during progressive exercise in metabolically compromised individuals is similar to the metabolic flexibility measured by hyperinsulinemic-euglycemic clamp. If so, this would suggest that mitochondrial function might be the main determinant of metabolic flexibility.

Materials and methods: Twenty-five (11 males and 14 females) overweight and obese individuals (BMI 31.7 ± 3.3 kg/m2; 64.7 ± 7.0 years) underwent an incremental submaximal cycling test at 30%, 50%, and 70% of the predetermined maximal power output (Wmax) and a two-step hyperinsulinemic-euglycemic clamp (10 and 40 mU/m2/min). Substrate oxidation upon insulin stimulation and during exercise was assessed by indirect calorimetry. Metabolic flexibility upon insulin stimulation was measured as the change in RER from the basal state to the high-insulin infusion state, while during incremental cycling, this was measured as the change in RER from 30% to 70% Wmax.

Results: Upon insulin stimulation, the RER showed an average increase of 0.12 ± 0.05 (from 0.78 ± 0.03 during the basal state to 0.89 ± 0.05 during the high-insulin state). During incremental cycling, the mean increase in RER was 0.13 ± 0.04 (0.80 ± 0.03 at 30% to 0.93 ± 0.03 at 70% Wmax). The change in RER upon insulin stimulation did not correlate with the change in RER measured during incremental submaximal cycling (r = -0.075, p = 0.723).

Conclusion: Although metabolic flexibility measured upon insulin stimulation quantitatively matches metabolic flexibility assessed during exercise of progressive intensity these variables did not correlate. This hints towards different underlying mechanisms controlling substrate oxidation upon insulin stimulation and progressive exercise.

Clinical Trial Registration Number: NCT03405545

Supported by: This project was supported by NWO and TiFN, a public-private partnership on precompetitive research in food and nutrition

Disclosure: M. Bergman: Grants; TiFN and NWO.

170

TSC22D4 interacts with Akt to regulate insulin sensitivity

B. Ekim Ustunel1,2, S. Demir1,2, G. Wolff1,2, A. Wieder1,2, J. Szendrödi1, S. Herzig2,3;

1Department of Internal Med. I and Clinical Chemistry, Heidelberg, 2Joint Heidelberg - IDC Translational Diabetes Program, Heidelberg, 3Helmholtz Center Munich, Helmholtz Diabetes Center HDC, Munich, Germany.

Background and aims: Transforming Growth Factor β-1 Stimulated Clone 22 D4 (TSC22D4) acts downstream of stress and metabolic signals playing role in glucose and lipid homeostasis. Particularly, hepatic TSC22D4 controls insulin sensitivity and regulates blood glucose levels. In diabetic mouse models, TSC22D4 promotes hyperglycemia and insulin resistance and strikingly, in obese human patients elevated hepatic TSC22D4 expression correlates with insulin resistance. These studies clearly establish a novel connection between TSC22D4 and metabolic regulation, yet the molecular mechanisms of this connection remains elusive. Protein Kinase B/Akt represents one of the very well-established components of the insulin signaling pathway that regulates gluconeogenesis and lipogenesis and contributes to metabolic homeostasis. Our preliminary data show that TSC22D4 interacts with Akt. The aim of this study is to investigate whether TSC22D4 plays role in metabolic regulation by interacting with Akt.

Materials and methods: We performed co-immunoprecipitation (co-IP) experiments to study the regulation of TSC22D4-Akt interaction. In order to map the domains required for TSC22D4-Akt interaction, we performed site-directed mutagenesis to delete conserved domains on TSC22D4. In order to understand the function of TSC22D4-Akt interaction in vivo, we generated Adeno-associated viruses (AAV) that contained the vector control, wild type (WT)-TSC22D4 or the TSC22D4 deletion mutant that cannot interact with Akt (ΔD2-TSC22D4). We introduced these AAVs to hepatocyte specific TSC22D4 knock out (TSC22D4Hep-/-) mice and subjected them to either chow diet or to high fat diet with high sucrose water and performed glucose and insulin tolerance tests (GTT and ITTs). We measured plasma insulin levels, liver triglycerides, and expression of gluconeogenic and lipogenic genes.

Results: We have shown that glucose and insulin stimulation impaired the TSC22D4-Akt interaction, whereas starvation or mitochondrial inhibition promoted it. Our experiments also indicate that together with its homodimerization domain (i.e. TSC box), TSC22D4 requires its intrinsically disordered region (D2) to interact with Akt. While deletion of D2 domain (ΔD2-TSC22D4) impaired the TSC22D4-Akt interaction; the D2 domain together with TSC box (D2+TSC) was sufficient to maintain the interaction. Furthermore, TSC22D4Hep-/- mice overexpressing ΔD2-TSC22D4 mutant in their livers performed better both in GTT and ITTs; and had lower insulin levels and an improved HOMA-IR index compared to mice overexpressing WT-TSC22D4. Additionally, mice overexpressing ΔD2-TSC22D4 tended to accumulate less triglycerides in their liver compared to mice with WT-TSC22D4 and had slightly lower expression of lipogenic genes. Expression of gluconeogenic genes, however, were not affected.

Conclusion: Here, we identify TSC22D4 as a novel Akt interacting protein. We show that not only glucose and insulin but also energy levels regulate the TSC22D4-Akt interaction. Hence, we define TSC22D4 as a novel signaling molecule that responds to metabolic signals by interacting with Akt and thereby regulates insulin sensitivity.

Supported by: DFG Grant to B.E.Ü (EK 108/1-1 ) and CRC 1118 to S.H.

Disclosure: B. Ekim Ustunel: None.

171

C2-ceramide recycling inhibits insulin signalling in muscle cells

C.L. Bandet1, S. Tan-Chen1, A. Blachnio-Zabielska2, J.-P. Pais-de-Barros3, P. Ferré1, F. Foufelle1, H. Le Stunff4, E. Hajduch1;

1INSERM U1138, Paris, France, 2University of Bialystok, Bialystok, Poland, 3INSERM U 1231, Dijon, France, 4CNRS U 9197, Orsay, France.

Background and aims: It is well documented that ectopic accumulation of saturated fatty acids (FA) induces lipotoxicity in muscle cells. Saturated FA are metabolized into ceramides (Cer) of which accumulation plays a central role in the development of muscle insulin resistance (IR). Studies involved in the discovery of Cer deleterious actions used short-chain Cer analogues such as C2-Cer (C2cer). Since C18-Cer is the predominant Cer species in myotubes and since Cer synthase 1 (CerS1)-derived C18-ceramide promotes insulin resistance, how can exogenous C2cer reproduce the harmful action of C18-Cer in myotubes? Our hypothesis is that C2cer is converted into longer chain Cer species that will mediate muscle using a salvage pathway by the action of ceramidase and CerS.

Materials and methods: We treated C2C12 myotubes with C2cer for 2h in the presence or absence of either a CerS inhibitor, Fumonisin B1 (FB1) or a ceramidase inhibitor (ceranib-2) before to quantify intracellular sphingolipids using UHPLC/MS/MS, and to evaluate the insulin response (Akt phosphorylation, GLUT4 translocation and glucose uptake). In addition, we used both Acetyl-CoA Carboxylase (ACC) and FA synthase (FAS) inhibitors to characterize whether FA produced from endogenous lipogenesis participate in the process.

Results: 2 h C2cer incubation of myotubes increased several endogenous long chain Cer species such as C14, C16, C18:1, C18, C24 and C24:1-Cer (p<0.05, n=5-6). To test whether generation of these long-chain Cer species from C2cer could happen through a C2cer de-acylation/re-acylation process, we treated myotubes with ceranib-2. After 2 h treatment, ceranib-2 alone induced a 50% accumulation in total endogenous Cer content, without alteration of insulin response (p<0.05, n=3-4). In contrast, inhibition of ceramidase activity with ceranib-2 reduced C2cer-induced endogenous Cer build-up and prevented C2cer to inhibit the insulin response in myotubes (p<0.05, n=3). To confirm a de-acylation/re-acylation process, we treated myotubes with FB1, and showed that it prevented the generation of endogenous Cer in response to C2cer, suggesting that sphingosine backbone of C2cer is re-used to produce other Cer species (p<0.05, n=3). FB1 pre-treatment also prevented the negative action of C2cer by restoring a normal insulin response (p<0.05, n=4). Endogenous FA used by CerS could be provided by the lipogenic pathway. To test it, myotubes were incubated with C2cer in the presence of 25 or 5 mM glucose. At 25 mM glucose, C2cer inhibited completely the insulin signal whereas at 5 mM glucose, C2cer action was blunted and the insulin signal remained unaffected (p<0.05, n=3). In addition, inhibition of de novo lipogenesis through the use of either a competitive inhibitor of ACC, 5-tetradecyloxy-2-furoic acid, or a FAS inhibitor, C-75, prevented the increase in total Cer content in response to C2cer. This was accompanied with the restoration of insulin action in myotubes treated with C2cer (p<0.05, n=3-6).

Conclusion: Our study provides important molecular mechanistic data showing for the first time that recycling of exogenous C2cer induced in muscle cells a loss in insulin sensitivity through production of endogenous Cer species. This recycling model also suggests a potential role of the uptake/recycling of circulating Cer, that are increased during obesity, in the installation of muscle insulin resistance.

Supported by: FDF, SFD

Disclosure: C.L. Bandet: Employment/Consultancy; fellowship from the French Ministry of Research.

172

The impact of different lipogenic diets on indirect pathway contributions to hepatic glycogen synthesis

A. Reis-Costa1, G.D. Belew1, L.C. Tavares1, M.J. Meneses2, J.G. Jones1;

1Center for Neuroscience and Cell Biology, Coimbra, 2Chronic Diseases Research Centre, Lisbon, Portugal.

Background and aims: We evaluated the impact of diet composition on hepatic glycogen metabolism in mouse models of diet-induced Non-Alcoholic Fatty Liver Disease (NAFLD). High fat and/or high sugar diets induce moderate weight gain and a mild form of NAFLD associated with impaired hepatic insulin action. In comparison to lipid, little is known about hepatic glycogen metabolism in this setting, in particular the indirect pathway contributions. We analysed the amount of glycogen synthesized in the liver, the contribution of the indirect pathway to glycogen synthesis, and the sources of indirect pathway carbons, including dietary fructose.

Materials and methods: For 18 weeks, 46 C57/BL6 male mice were assigned to 4 different diets: 12 to standard chow (SC), 12 to standard chow with sugar in the drinking water at 30% w/v (HS), 11 to a high-fat chow (HF) and 11 to high-fat chow with sugar in the drinking water at 30% w/v (HFHS). The sugar formulation was 55/45% fructose/glucose mimicking high-fructose corn syrup 55 (HFCS-55). During the final evening, deuterated water (2H2O) was administered and the fructose component of the HFCS-55 formulation was enriched to 20% [U-13C]fructose. Animals were allowed to feed ad libitum overnight. The following morning, livers were collected, and their glycogen extracted, digested and derivatized to mono-acetone glucose (MAG) for 2H- and 13C-NMR analysis. Total glycogen levels were assayed and the contribution of the indirect pathway to glycogen synthesis, including inputs from trioses phosphate (triose-P) and anaplerosis, were calculated using the 2H-enrichment of MAG positions 5 and 6. The contribution of fructose to the indirect pathway was estimated by quantifying 13C-enrichment of glycogen from [U-13C]fructose.

Results: Postprandial glycogen levels did not differ between the 4 diets. In HF mice, the indirect pathway accounted for 58±8% of overnight glycogen synthesis, a significantly higher fraction compared to the other three diets (SC = 32±8%, HS = 37±9%, HFHS = 40±4%, p<0,05). For both HF and SC mice, the majority of indirect pathway carbons were derived via anaplerosis, while for HS and HFHS mice, a substantial fraction of indirect pathway carbons were derived from triose-P sources, the entry point of fructose metabolites into the gluconeogenic pathway. Recruitment of HFCS-55 fructose by the indirect pathway was confirmed by the observation of [4,5,6-13C3]- and [5,6-13C2]glycogen isotopomers by 13C-NMR for HS and HFHS groups.

Conclusion: While postprandial hepatic glycogen levels were not affected by high fat and/or high sugar feeding, the sources of glycogen synthesis were strongly influenced by diet. High fat feeding promoted a high indirect pathway contribution, possibly compensating for impaired direct pathway activity secondary to hepatic insulin resistance. Although HFCS-55 provided additional indirect pathway precursors via fructose metabolites, direct pathway flux was preserved, possibly by fructose-mediated activation of glucokinase.

Clinical Trial Registration Number: DGAV, 0421/000/000/2013

Supported by: FCT-FEDER-02/SAICT/2017/028147; SPD-GIFT; UIDB/04539/2020; POCI-01-0145-FEDER-007440; REEQ/481/QUI/2006; RECI/QEQ-QFI/0168/2012; CENTRO-07-CT62-FEDER-002012; PTDC/BIA-BQM/28147/2017

Disclosure: A. Reis-Costa: None.

173

Transcriptional repression of the iron exporter ferroportin via the PI3K-AKT-Foxo1 signalling pathway may explain liver iron overload in patients with type 2 diabetes

R. Qiu1, N. Volk2, O. Marques1,3, S. Altamura1,3, M.U. Muckenthaler1,3;

1Department of Pediatric Hematology, Oncology and Immunology, University of Heidelberg, Heidelberg, 2Tissue Bank of the National Center for Tumor Diseases (NCT), Heidelberg, 3Molecular Medicine Partnership Unit, Heidelberg, Germany.

Background and aims: Iron and glucose homeostasis are tightly interlinked. Patients and mice with type 2 diabetes (T2DM) show elevated serum iron and ferritin levels, while patients with the iron overload disease hereditary hemochromatosis are prone to diabetes. In this study we investigated iron content and localization in the human diabetic liver and asked the question whether alterations in iron export via ferroportin (Fpn) explain altered iron homeostasis.

Materials and methods: Iron content and localization were analyzed in liver biopsies of 23 patients with T2DM and 25 patients without T2DM by Perl’s-Prussian-blue staining. Kupffer cells were identified by CD68 staining. To establish a cellular model of insulin resistance we applied palmitate (200 μM) and insulin (100 nM) to Hepa1-6 cells. Expression of mRNA and proteins involved in insulin signaling and iron homeostasis were analyzed by qPCR and western blotting. The PI3K and Foxo1 signaling pathways were blocked by pharmacological inhibition (Wortmannin and AS1842856, respectively) and RNAi.

Results: We show increased liver iron content in patients with T2DM compared to controls, whereby iron accumulates in hepatocytes, Kupffer cells or both. The percentage of CD68 positive Kupffer cells was not different. To study molecular mechanisms of iron accumulation in the diabetic liver, we established a cellular model of insulin resistance. We treated the hepatocytic cell line Hepa1-6 with palmitate and insulin, either alone or in combination. Insulin resistance was indicated by higher basal p-AKT levels and a blunted response to short-term insulin treatment. Similar to observations in T2D patients, insulin resistance in Hepa 1-6 cells was associated with cellular iron excess (e.g. increased ferritin, low transferrin receptor 1). Interestingly, treatment of insulin-resistant Hepa1-6 cells with exogenous iron (i.e. ferric ammonium citrate; FAC) resulted in aggravated iron accumulation. Systemic iron homeostasis is controlled by the hepcidin/ferroportin regulatory axis. We show that in conditions of insulin resistance the iron exporter ferroportin is downregulated at the mRNA and protein level in a hepcidin-independent manner. Reduced ferroportin transcription seems to be controlled by the PI3K signaling pathway, as inhibition by Wortmannin or silencing of AKT blocked the insulin-mediated Fpn response. Insulin inactivates the transcription factor FOXO1 via the AKT pathway. Consistently, the specific Foxo1 inhibitor AS1842856 as well as RNAi of Foxo1 reduced Fpn mRNA expression.

Conclusion: Our findings suggest that the iron exporter ferroportin is a novel target of the PI3K-AKT-Foxo1 signaling pathway in hepatocytes. We speculate that decreased expression of ferroportin may explain liver iron overload in patients with T2D.

Supported by: German Research Foundation (DFG - SFB1118)

Disclosure: R. Qiu: None.

174

Ciprofibrate decreases net hepatic glucose uptake and tends to decrease net myocardial glucose uptake in prediabetic male volunteers

V. de Wit-Verheggen1, F. Vanweert1, J. Raiko2, G. Schaart1, A. Gemmink1, E.B.M. Nascimento1, M.K.C. Hesselink1, J.E. Wildberger3, R. Wierts3, D. Montaigne4, B. Staels4, E. Phielix1, P. Schrauwen1, V.B. Schrauwen-Hinderling1,3, T. van de Weijer1,3;

1Department of Nutrition and Movement Sciences, School of Nutrition and Translational Research in Metabolism, Maastricht, Netherlands, 2Turku PET Centre, University of Turku and Turku University Hospital, Turku, Finland, 3Department of Radiology and Nuclear Medicine, Maastricht University Medical Center, Maastricht, Netherlands, 4Institut Pasteur de Lille, Lille, France.

Background and aims: Type 2 diabetes mellitus is characterized by reduced (hepatic) insulin sensitivity and reduced cardiac function. PPARαtreatment has beneficial effects on metabolism and regulates cardiac energy metabolism and function as established in pre-clinical animal studies. The objective of the current study was to determine whether the PPARα agonist ciprofibrate would improve cardiac and hepatic metabolism in volunteers with prediabetes.

Materials and methods: Ten male volunteers with prediabetes (BMI 29.1 ± 2.9 kg/m2, age 62.2 ± 9.3 years) were administered with 100 mg/day placebo and ciprofibrate in a randomized double-blind crossover study for 5 weeks. Investigations included cardiac and hepatic glucose uptake during a hyperinsulinemic euglycemic clamp by 18F-FDG positron emission tomography, cardiac function and structure with MRI and ultrasound, cardiac energy status (PCr/ATP ratio), whole body insulin sensitivity, intrahepatic lipid content, body composition; all analyzed by two-tailed paired sample t-tests with p < 0.05 considered as statistically significant.

Results: As expected, serum triglycerides were significantly decreased by ciprofibrate treatment indicating a therapy response in all volunteers (2.15 ± 1.32 mmol/l in placebo vs. 1.57 ± 1.05 mmol/l in ciprofibrate, p = 0.014). Ciprofibrate had no effect on fasting free fatty acids (FFA), cholesterol, fasting glucose and insulin levels. Ciprofibrate treatment decreased net hepatic glucose uptake (Ki 0.9 ± 0.2 versus 0.7 ± 0.2 ml/100ml/min in placebo and ciprofibrate, p = 0.039) and increased hepatic lipid content (7.1 ± 6.9 % versus 10.4 ± 10.0 % in placebo and ciprofibrate, p = 0.021). Myocardial net glucose uptake tended to decrease (0.049 ± 0.039 versus 0.040 ± 0.024 ml/100ml/min/mIU insulin in placebo and ciprofibrate, p = 0.098) upon ciprofibrate, but ciprofibrate had no effect on cardiac function and cardiac energy status. Whole body insulin sensitivity and body composition were unchanged upon ciprofibrate treatment.

Conclusion: This study indicates that 5 weeks of treatment with the PPARα ligand ciprofibrate decreases insulin-stimulated glucose uptake in the liver, with a similar tendency in the heart. The increase in liver fat was not associated with a decline in whole body insulin sensitivity, nor with reduced cardiac function parameters or cardiac energy status.

Clinical Trial Registration Number: NCT03662984

Supported by: B.S. and D.M. were supported by grants from Agence Nationale pour la Recherche (ANR-16-RHUS-0006-PreciNASH, ANR-10-LBEX-46, ANR TOMIS-Leukocyte: ANR-CE14-0003-01 and ANR CALMOS: ANR-18-CE17-0003-02). This work was (partially) funded by the National Center for Precision Diabetic Medicine – PreciDIAB (ANR-18-IBHU-0001; 20001891/NP0025517; 2019_ESR_11). B.S. is a recipient of an Advanced ERC Grant (694717). E.P. was supported by a senior fellowship from the Dutch Diabetes Foundation (Grant 2017.82.010). V.S. was supported by a grant from the European Research Council (ERC-2017-StG-759161). T.W. was supported by a junior fellowship by the Dutch Diabetes Foundation (Grant 2015.81.1833).

Disclosure: V. de Wit-Verheggen: None.

OP 30 GLP-1 receptor agonism: putative mechanisms of benefit

175

Combination very low dose sulphonylurea and DPP4 inhibitor have a potent glucose lowering effect through augmentation of beta cell function without increase in hypoglycaemia

R.L.M. Cordiner1, A. Mari2, K.F.E. Bedair1, E.R. Pearson1;

1Level 5, Mailbox 12, University of Dundee, Dundee, UK, 2Institute of Neuroscience - National Research Council, Padova, Italy.

Background and aims: The global prevalence of type 2 diabetes (T2DM) will surpass 600 million by 2035, with highest prevalence contributed by low and middle-income countries. We must readdress our use of cheaper generic therapies such as sulphonylureas (SU) and those due to come off patent such as DPP4 inhibitors. Negative aspects of SU, particularly hypoglycaemia, weight gain and reduced durability may be avoided by their use at low dose. We have previously shown that 20mg standard release gliclazide reduces plasma glucose through augmentation of the classical incretin effect, increased beta-cell glucose sensitivity and late phase incretin potentiation. We further hypothesise potential synergy between low dose SU in combination with a DPP4 inhibitor.

Materials and methods: 30 participants with T2DM (HbA1c <64mmol/l) treated with diet or metformin monotherapy were recruited to a single-centre, open-label, randomised crossover study. Participants completed four, 14-day blocks in a random order: control, gliclazide 20mg once daily (SU), sitagliptin 100mg (DPP4), or combination (SUDPP4). A 2-hour MMT was conducted at the end of each block. Beta cell function was modelled through the relationship between insulin secretion and glucose concentration. The primary outcome was the effect of treatment on beta cell glucose sensitivity. Continuous glucose monitoring (CGM) parameters were explored as a secondary outcome.

Results: Linear mixed model estimates showed a potent additive glucose lowering effect: mean glucose from area under the curve (mean 95% CI) (mmol/l): Control 11.5 (10.7 - 12.3), DPP4 10.2 (9.4 - 11.1), SU 9.7 (8.9 - 10.5), SUDPP4 8.7 (7.9 - 9.5) (p <0.001). Glucose sensitivity (pmol min-1 m-2mM-1) mirrored this additive effect: Control 71.5 (51.1 - 91.9), DPP4 75.9 (55.7 - 96.0), SU 86.3 (66.1 - 106.4), SUDPP4 94.1 (73.9 - 114.3) (p = 0.04 SUDPP4 Only). The plot of insulin secretion against glucose is shown in figure 1, which shows progressive increase in insulin secretion at the same glucose concentrations (i.e. increased glucose sensitivity) in favour of combination DPP4 and SU. Mean gliclazide concentrations (mean (SD)) (ng/ml) were: SU 662 (408), SUDPP4 603 (355) respectively (p = 0.31). Glucose time in range <3mmol/l on CGM (%) was unaffected: Control 1 (2 - 4), DPP4 2 (3 - 6), SU 1 (0 - 4), SUDPP4 3 (2 - 7) (p = 0.648).

Conclusion: Combination low dose gliclazide with a DPP4 inhibitor has potent glucose lowering effect through augmentation of beta cell function. Glucose reduction was achieved at gliclazide concentrations far below those achieved with standard therapeutic doses (~600ng/ml vs >5000ng/ml CMAX with 80mg gliclazide daily). A double-blind randomised controlled trial is merited to formalise efficacy and safety of this combination, which may avoid negative aspects of SU and provide pharmacoeconomic benefit in a world of rising costs of diabetes care.

figurebd

Clinical Trial Registration Number: NCT04192292

Supported by: This research is funded by the Wellcome Trust New Investigator Award held by ERP

Disclosure: R.L.M. Cordiner: None.

176

Emotional eating is associated with reduced sensitivity to the central effects of GLP-1 receptor agonist treatment

C.C. Van Ruiten1, J. Ten Kulve1, L. van Bloemendaal1, M. Nieuwdorp1, D. Veltman2, R.G. IJzerman1;

1Internal Medicine, Diabetes Center, Amsterdam University Medical Centers, Location VU University Medical Center, Amsterdam, 2Psychiatry, Amsterdam University Medical Centers, Location VU University Medical Center, Amsterdam, Netherlands.

Background and aims: The weight loss induced by GLP-1 receptor agonists (GLP-1RAs) is related to suppressed appetite signaling in the brain and increased satiety, which leads to a reduced food intake, via direct and indirect actions on the CNS. However, there is considerable variation in individual responses to GLP-1RAs. It is unclear why some patients respond with weight loss and others do not. It has been suggested that emotional eaters are less sensitive to the acute central effects of GLP-1RAs. Therefore, the aim of this study was to investigate if individuals with higher emotional eating scores are less sensitive to longer term effects of GLP-1 RA on central responses to food cues.

Materials and methods: We performed secondary analysis of a randomized crossover study in obese patients with type 2 diabetes (n = 20, mean age 59.3 ± 4.1 years, mean BMI 32 ± 4.7 kg/m2), consisting of two periods of 12-week treatment with either liraglutide 1.8 mg or insulin glargine. Using functional MRI, we assessed the relation between emotional eating and the effects of the GLP-1RA liraglutide after 10 days and 12 weeks of treatment on regional brain responses to visual food stimuli and to actual food.

Results: At baseline, higher emotional eating scores were associated with stronger brain responses to the anticipation of chocolate milk in the insula (right p=0.03, left p=0.05), and with less response to chocolate milk receipt in the insula (right p=0.004, left p=0.023). After 10 days of treatment, compared to insulin glargine, higher emotional eating scores were associated with less pronounced GLP-1RA induced reductions in brain responses to food pictures in the amygdala (p=0.01), bilateral insula (p=0.048) and left caudate nucleus (p=0.02). In addition, higher emotional eating scores tended to be associated with less pronounced GLP-1RA increases in brain responses to food reward in the left caudate nucleus and left insula. After 12 weeks of treatment, there were no significant associations between emotional eating scores and liraglutide-induced changes in CNS responses to food cues.

Conclusion: Our findings indicate that individuals with higher emotional eating scores, who are more vulnerable to cravings and less sensitive to rewarding properties of food, may be less sensitive to GLP-1RA treatment. These insights may help to optimize treatment strategies for obesity and to select patient groups with better efficacy of GLP-1RA treatment.

figurebe

Clinical Trial Registration Number: NCT01363609

Supported by: Novo Nordisk

Disclosure: C.C. Van Ruiten: Employment/Consultancy; Scientific Advisory Board of Caelus Pharmaceuticals, the Netherlands and Kaleido, USA. All payments were directly transferred to the nonprofit Amsterdam UMC. Grants; research grants from AstraZeneca, Eli Lilly&Co, and Novo Nordisk. All payments were directly transferred to the nonprofit Amsterdam UMC.

177

Liraglutide decreases postprandial fibroblast growth factor 19 and glucagon-like peptide 2, and increases postprandial cholecystokinin in individuals with obesity

A. Brønden1,2, H.H. Nerild1, C.C. Nexøe-Larsen1, P.H. Hellmann1, M. Baekdal1, I.M. Gether1, M.P. Gillum3, B. Hartmann4,3, L.B. Knudsen5, L.V. Jacobsen5, J.F. Rehfeld6, J.J. Holst4,3, T. Vilsbøll1,7, D.P. Sonne2,1, F.K. Knop1,7;

1Center for Clinical Metabolic Research, Herlev and Gentofte Hospital, Hellerup, 2Department of Clinical Pharmacology, Bispebjerg and Frederiksberg Hospital, Copenhagen, 3Novo Nordisk Foundation Center for Basic Metabolic Research, University of Copenhagen, Copenhagen, 4Department of Biomedical Sciences, Faculty of Health and Medical Sciences, University of Copenhagen, Copenhagen, 5Novo Nordisk, Bagsværd, Denmark, 6Department of Clinical Biochemistry, Rigshospitalet, Copenhagen, 7Steno Diabetes Center Copenhagen, Gentofte, Denmark.

Background and aims: Treatment with liraglutide as well as other glucagon-like peptide 1 (GLP-1) receptor agonists is associated with slightly increased risk of gallbladder-related disorders, which have been proposed to be a consequence of altered gallbladder motility; thus, liraglutide seems to delay postprandial gallbladder refilling. The gut hormones cholecystokinin (CCK), fibroblast growth factor 19 (FGF19) and glucagon-like peptide 2 (GLP-2) are known to regulate gallbladder motility, and changes in postprandial concentrations of these hormones could explain the altered gallbladder motility.

Materials and methods: In a single-centre, double-blinded, 12-week trial 52 participants with obesity were randomised 1:1 to once-daily subcutaneous liraglutide (escalated from 0.6 mg to 3.0 mg with 0.6-mg weekly increments) or placebo. We evaluated gallbladder dynamics using ultrasonograpy during 4-hour liquid meal tests (600 kcal, 23.7 g fat) at baseline, after the first dose of study drug and following 12 weeks of treatment and showed a liraglutide-induced deceleration of postprandial gallbladder refilling. Postprandial plasma responses of hormones known to modulate gallbladder motility (CCK, FGF19, GLP-2) were secondary endpoints. The primary endpoint of the study, maximum postprandial gallbladder ejection fraction, was reported previously.

Results: Baseline characteristics were similar between groups (50% male, age 47.6±10.0 years, body weight 99.0±15.7 kg, BMI 32.6±3.4 kg/m2 (mean±SD)). Compared to placebo, liraglutide reduced postprandial FGF19 responses after first dose (AUC 24.8 vs 48.0 ng/ml×min with treatment ratio (TR) [95% CI] 0.52 [0.39; 0.69]) and following 12 weeks of treatment (AUC 33.7 vs 48.5 ng/ml×min, TR 0.69 [0.52; 0.93]). Liraglutide also reduced postprandial GLP-2 responses (AUC 3,650 vs 4,894 pmol/l×min, TR 0.75 [0.62; 0.90]) following first dose as well as after 12 weeks (AUC 3,760 vs 4,882 pmol/l×min, TR 0.77 [0.60; 0.99]). Compared to placebo, liraglutide increased postprandial responses of CCK after first dose (AUC 762 vs 670 pmol/l×min (TR 1.14 [0.97; 1.33]) and following 12 weeks of treatment (AUC 873 vs 628 pmol/l×min (TR 1.39 [1.12; 1.73]).

Conclusion: Treatment with liraglutide caused increased postprandial plasma CCK concentrations and decreased plasma FGF19 and GLP-2 concentrations compared to placebo, which may explain the delayed postprandial gallbladder refilling observed in individuals with obesity treated with liraglutide.

Clinical Trial Registration Number: NCT02717858

Supported by: The study was supported by Novo Nordisk A/S

Disclosure: A. Brønden: None.

178

Short-term treatment with liraglutide does not improve cardiac diastolic function in patients with type 2 diabetes: a randomised double-blind placebo-controlled trial

A.S. Bojer1,2, M.H. Soerensen1, J. Bjerre3, P. Gæde1,2, N. Vejlstrup4, P.L. Madsen5,6;

1Department of Endocrinology and Cardiology, Slagelse Hospital, Slagelse, 2Institute of Regional Health Research, University of Southern Denmark, Odense, 3Department of Cardiology, Copenhagen University Hospital Bispebjerg, Copenhagen, 4Department of Cardiology, Rigshospitalet, Copenhagen, 5Department of Cardiology, Copenhagen University Hospital Herlev-Gentofte, Copenhagen, 6Department of Clinical Medicine, University of Copenhagen, Copenhagen, Denmark.

Background and aims: The cardioprotective effect of glucagon-like peptide-1 (GLP-1) receptor agonists in patients with type 2 diabetes is still not well explained. In these patients, diastolic dysfunction is significant and linked to outcome, and the cardioprotective effect of GLP-1 receptor agonists may be by improving diastolic function. This study aimed to investigate if short-term treatment of liraglutide a GLP-1 receptor agonist improves left ventricular diastolic function.

Materials and methods: In an investigator-initiated double-blind randomized placebo-controlled trial the effect of 18 weeks of treatment with liraglutide on diastolic function was assessed in type 2 diabetes patients and echocardiographic signs of diastolic dysfunction (echo-Doppler determined E/e´≥9 or/and lateral e´≤10 cm/sec). Primary outcomes were improved left ventricle filling (the early peak filling rate, ePFR) and left atrium ease of emptying (the passive emptying fraction, LAPEF), assessed by cardiac magnetic resonance imaging at rest and during chronotropic stress (glycopyrrolate 4 μg/kg; a cholinergic receptor antagonist increasing heart rate, and thereby inducing chronotropic stress without also affecting contractility). Secondary outcomes included left ventricular and left atrial volumes and systolic function, measures of aortic stiffness, and echocardiographic diastolic parameters.

Results: Forty patients were randomized to liraglutide s.c. 1.8 mg/day (n=20) or placebo (n=20). Liraglutide reduced HbA1c (-0.47% 95%CI (-0.88 to -0.06)) and weight (-2.9kg 95%CI (-4.6 to -1.2)), both p<0.03). Liraglutide did not change ePFR at rest -24±60 vs. -6±46 ml/sec, during stress 2±58 vs. -2±38 ml/sec, or the changes from rest and stress 12.9±72.5 vs. 4.7±104.0, all p>0.05. LAPEF decreased with liraglutide during stress (median(Q1,Q3)) -3.1(-9.0, 1.1) vs. 1.0(-2.9, 6.1) %, p=0.049, but no changes were evident at rest -4.3(-7.9, 1.9) vs. -0.6(-3.1, 2.2) %, p=0.19, or for the changes from rest to stress -1.7±8.4 vs. 0.8±8.2, p=0.4. All secondary outcomes were unchanged by liraglutide.

Conclusion: Short-term treatment with liraglutide did not improve diastolic function in patients with type 2 diabetes and echocardiographic signs of diastolic dysfunction. This suggests that the cardioprotective effect seen in long-term studies of liraglutide is not related to the improvement of left ventricular diastolic function.

figurebf

Clinical Trial Registration Number: NCT02655770

Supported by: During this work, ASB has received funding from the local research committee at NSR hospital, the regional research committee of Region Zealand [13–000835], and the Danish Heart Association [16-R107-A6790-22002, and 18-R125-A8444-22110]. Novo Nordisk supported the study by an unrestricted grant covering the costs of CMR scans and blood analyses. Further, Novo Nordisk provided free study medication and matching placebo pens. None of the funding sources played any role in the process of conduction, interpretation of results or publishing the study. The corresponding author had full access to all data and had the final responsibility for the decision to submit for publication.

Disclosure: A.S. Bojer: None.

179

Glucose-dependent insulinotropic polypeptide (GIP) contributes to sitagliptin-mediated improvement of beta cell function in patients with type 2 diabetes

S. Stensen1, L.S. Gasbjerg1, M.M. Rosenkilde2, B. Hartmann2, T. Vilsbøll3, J.J. Holst2, M.B. Christensen4, F.K. Knop1;

1Gentofte Hospital, Hellerup, 2Department of Biomedical Sciences, University of Copenhagen, Copenhagen, 3Steno Diabetes Center Copenhagen, Gentofte, 4Bispebjerg Hospital, Copenhagen, Denmark.

Background and aims: In patients with type 2 diabetes, dipeptidyl peptidase 4 (DPP-4) inhibitor treatment improves glycaemic control by raising the active levels of the insulinotropic gut hormones glucagon-like peptide 1 and glucose-dependent insulinotropic polypeptide (GIP). The potentiating effect of these hormones on glucose-stimulated insulin secretion seen in healthy persons was deemed lost in type 2 diabetes, as exogenous GIP was shown to have a poor or even absent insulinotropic effect in these patients. Using the highly selective GIP receptor antagonist GIP(3-30)NH2, we investigated endogenous GIP’s contribution to the insulinotropic effect of DPP-4 inhibition (assessed by insulin secretion rate (ISR) relative to plasma glucose) in patients with type 2 diabetes.

Materials and methods: In a double-blind, placebo-controlled, crossover study, 12 patients with type 2 diabetes underwent two randomised 12-13-day treatment courses with DPP-4 inhibitor (sitagliptin 100 mg once-daily) and placebo, respectively, with an interposed 1-3-week washout period. In the end of each treatment period, two randomised 5-hour liquid mixed meal tests with infusion of GIP(3-30)NH2 (1,200 pmol/kg/min) or saline (placebo) were performed.

Results: We included eight men and four women (mean ± SD; BMI 27.4 ± 2.6 kg/m2, HbA1c 54 ± 15 mmol/mol) in the study. During placebo treatment, GIP(3-30)NH2 lowered postprandial serum C-peptide (determined as difference in baseline-subtracted AUC (ΔbsAUC%) ± SEM: −31 ± 9%, p = 0.005) and increased postprandial plasma glucose excursions (ΔAUC% ± SEM: 7.3 ± 2.8%, p = 0.017) compared to saline. Sitagliptin increased concentrations of active GIP(1-42) (ΔAUC% ± SEM: 153 ± 21%, p < 0.0001) and lowered fasting plasma glucose from 8.7 to 7.6 mmol/L (p < 0.0004), whereas baseline-subtracted postprandial glucose excursions were unchanged compared to placebo treatment. We calculated the percentage reduction in AUCISR/AUCglucose ratio to evaluate the contribution of GIP to the insulinotropic effect of sitagliptin. The maximal potential of GIP, quantified as the AUC during DPP-4 inhibitor treatment including the contribution of GIP (sitagliptin treatment + saline) minus the physiological meal-response without GIP (placebo treatment + GIP(3-30)NH2) was specified to be 100% of what GIP could potentially mediate. Thus, GIP(3-30)NH2 caused a reduction in AUCISR/AUCglucose ratio equivalent to 37 ± 12% of the increments due to sitagliptin.

Conclusion: We demonstrate an insulinotropic and glucose-lowering effect of endogenous GIP in patients with type 2 diabetes and show that endogenous GIP is responsible for more than one third of the improved beta cell function observed during DPP-4 inhibitor treatment.

Clinical Trial Registration Number: NCT03845179

Supported by: EFSD and Novo Nordisk A/S Programme for Diabetes Research in Europe 2017, AP Moller Foundation and Novo Nordisk Foundation

Disclosure: S. Stensen: None.

180

Semaglutide reduces hsCRP levels across different treatment settings: post hoc analyses of SUSTAIN and PIONEER trials

O. Mosenzon1, M. Capehorn2, A. De Remigis3, S. Rasmussen3, P. Weimers3, J. Rosenstock4;

1Hadassah Medical Center, Jerusalem, Israel, 2Rotherham Institute for Obesity, Rotherham, UK, 3Novo Nordisk A/S, Søborg, Denmark, 4Dallas Diabetes Research Center at Medical City, Dallas, USA.

Background and aims: Limited information is available on the effects of glucagon-like peptide-1 receptor agonists on high-sensitivity C-reactive protein (hsCRP). This exploratory analysis aimed to determine the effect of the two formulations of semaglutide vs comparators on hsCRP.

Materials and methods: This analysis included trials where hsCRP data were available (SUSTAIN 3 and PIONEER 1, 2, 5). Subjects with type 2 diabetes (T2D; N=2,482) and chronic kidney disease (PIONEER 5 only) received once-weekly s.c. or once-daily oral semaglutide or comparators (see Figure). Mediation analyses assessed the direct and indirect effect of change in HbA1c and body weight (BW) on hsCRP; the percentage mediated by these parameters was calculated. These mediation analyses were performed using mixed models for repeated measurements, with change in hsCRP and change in the mediator as the outcomes. All hsCRP measurements were analysed by trial.

Results: Geometric mean baseline hsCRP was similar across trials (range 2.7-3.0 mg/L). Semaglutide significantly reduced hsCRP from baseline in all trials and vs all comparators (except vs placebo in PIONEER 5; Figure). Some of the semaglutide effect on hsCRP was mediated by a change in HbA1c (percentage mediated: 30-39%), whereas BW played a lesser role (5-35%), except in PIONEER 5 (HbA1c 26%; BW 50%).

Conclusion: Semaglutide reduced hsCRP levels vs comparators in subjects with T2D, partially mediated indirectly via the effect on HbA1c and, to a lesser extent, BW; however, there may be a direct semaglutide effect. The observed reductions in hsCRP in subjects with T2D suggest a possible anti-inflammatory effect of semaglutide. Ongoing trials will provide further insights into the impact of semaglutide as an anti-inflammatory drug and its cardiovascular effects.

figurebg

Clinical Trial Registration Number: NCT01885208; NCT02906930; NCT02863328; NCT02827708

Supported by: Novo Nordisk A/S

Disclosure: O. Mosenzon: Employment/Consultancy; Advisory Board participation Novo Nordisk. Lecture/other fees; Novo Nordisk.

OP 31 Modelling diabetes long term complications

181

Mir-34a mediates progression of liver injury from NAFLD to fibrosis in diabetes associated liver fibrosis

Q. Su;

School of Biological Sciences, Queen's University Belfast, Belfast, UK.

Background and aims: Non-alcoholic fatty liver disease (NAFLD) is a multifactorial disease that encompasses a spectrum of pathological conditions, ranging from simple steatosis (NAFL), nonalcoholic steatohepatitis (NASH) and fibrosis/cirrhosis which could end up to hepatocellular carcinoma and liver failure. Diabetes is an independent risk factor in the development of liver fibrosis. However, the precise mechanisms by which diabetes worsens liver function is poorly understood. The aim of this study was to determine the mechanistic role of miR-34a at different disease stages of liver injury and wound healing in diabetic context.

Materials and methods: A mouse model that highly resembles the pathological development in patients with liver cirrhosis was established by treating the mice with a dose (200μg) of streptozotocin (STZ) at day 2 after born to induce diabetes and followed by feeding a high fat diet (HFD) after weaning to induce NAFLD and liver fibrosis. Liver tissues and primary hepatocytes were collected from the mice at the age of 6, 8 and 12 weeks and subjected to histological analysis, microRNA deep sequencing, q-RT-PCR and immunoblotting analysis.

Results: Histological analysis revealed the pathological development of hepatic steatosis at week 6, NASH at week 8 and liver fibrosis at week 12, a series of liver injury similar to clinical patients at different disease stage. Importantly, in the liver tissues from mice treated with STZ/HFD, lacking insulin induced significant upregulation of miR-34a which was further exacerbated by the subsequent HFD feeding. Upregulated miR-34a targeted the hepatic nuclear factor 4α (HNF4α), a verified true target of miR-34a, and inhibited expression of HNF4α at both mRNA and proteins (P<0.05). These pathological changes disturbed expression of genes involved in mitochondrial fatty acid β-oxidation and hepatic very low density lipoprotein (VLDL) metabolism, indicated by the decreased mRNA expression of PPARα and its downstream target genes CPT-1α and ACOX-1 and the upregulation of apolipoprotein B and MTP, respectively. Consequently, lipotoxicity generated from the accumulated lipids in hepatocytes induced metabolic inflammation and compromised hepatic insulin signaling. Moreover, upregulation of miR-34a further enhanced expression of pro-fibrogenic cytokines TGFβ1 and TGFβ2 in both hepatocytes and hepatic stellate cells (HSC) which stimulated expression of fibrosgenic genes, α-SMA and Col1A1, resulting in accumulation of extracellular matrix (ECM) components in the liver and inducing fibrosis. In Vitro, transfection of miR-34a into a mouse hepatocyte AML12 inhibited expression of insulin signaling molecules IRS-1 and PI3K (P<0.05). More importantly, incubation of the miR-34a transfected cells with insulin can inhibit activation of TGFβ signaling and mitigate fibrogenic gene expression, suggesting the essential role of insulin in preventing liver injury and fibrosis.

Conclusion: Our study, for the first time, demonstrates that upregulation of miR-34a induced by insulin deficiency and HFD target multiple metabolic pathways, including lipoprotein metabolism, oxidative stress and fibrogenesis, to induce liver fibrosis/cirrhosis via mediation of HNF4α. This novel finding may lend support to the development of miR-34a and HNF4α as pharmaceutical targets for the treatment of NAFLD and hepatic fibrosis.

Disclosure: Q. Su: None.

182

An siRNA strategy to silence apolipoprotein CIII in the fight against the metabolic syndrome

P. Recio-López, P.-O. Berggren, L. Juntti-Berggren, I. Valladolid-Acebes;

Molecular Medicine and Surgery, Karolinska Institutet, Stockholm, Sweden.

Background and aims: Apolipoprotein CIII (apoCIII) is increased in obesity-induced insulin resistance and type-2 diabetes. Emerging evidence support the advantages of small interfering RNAs (siRNAs) to target disease-causing genes. We aimed to develop a new RNA-based platform for in vivo silencing of apoCIII and investigate its protective effects against the metabolic syndrome.

Materials and methods: We used 12-week-old male and female B6.Cg-Lepob/ LepobJ (ob/ob) mice on a C57BL6/j background from our own breeding colony. Before start of treatment body weight (BW) and blood glucose (BG) were determined to randomize and assign the mice to two different groups: i) apoCIII-siRNA (n = 5); and ii) control-siRNA treatment (n = 5). The siRNAS were given intravenously (i.v.), at a dose of 0.008 mg/kg, for three consecutive days every 15 days for eight weeks. Plasma apoCIII, BW and BG were monitored during the study. After the last treatment plasma lipoprotein lipase (LPL) activity in pre- and post-heparin injected mice (0.2 international units per g of BW) was determined and intraperitoneal glucose tolerance test (IPGTT) and glucose-stimulated insulin secretion (GSIS) were performed. At the end of the study samples for lipid profiles, apoCIII gene expression and protein levels and off-target effects were taken. Data are expressed as mean±s.e.m. Statistical comparisons between the two groups of mice were performed using the Mann-Whitney U t-test. When two independent variables were considered, 2-ANOVA followed by Bonferroni´s post-hoc test was used. Statistical significance was defined as P < 0.05.

Results: Our results show that circulating apoCIII was progressively lowered upon administration of apoCIII-siRNA (apoCIII reduction: 47.8±4.5%; 2-ANOVA; F1,72=11.71; P < 0.001). Plasma LPL activity was higher in mice with reduced apoCIII levels (LPL activity fold-increase: 8.2±2.1; 2-ANOVA; F1,24=24.06; P < 0.0001). The increased plasma LPL activity resulted in lower levels of triglycerides (apoCIII-siRNA: 141.9±26.9 mg/dL; control-siRNA: 264.7±34.9 mg/dL Mann-Whitney U t-test; P < 0.05). Decreasing apoCIII induced a progressive reduction in weight gain (weight gain reduction: 8.8±2.1%; 2-ANOVA; F1,119=5.97; P < 0.05) and non-fasting BG levels (BG reduction: 35.8±6.8%; 2-ANOVA; F1,119=27.27; P < 0.0001), as well as improved IPGTT and GSIS (2-ANOVA; F1,35=34.33; P < 0.0001 and F1,35=11.86; P < 0.05, respectively). At the end of the study, it was confirmed that apoCIII gene and protein levels were reduced in liver from siRNA-treated mice (liver apoCIII mRNA reduction: 34.8±4.3%, P < 0.001; liver apoCIII protein reduction: 40.7±8.7%, P < 0.05; Mann-Whitney U t-test), compared to the control group. To test the specificity of the siRNA the expression of apoCIII was analyzed in duodenum, the second largest source of the apolipoprotein, and the levels were unaffected. Furthermore, since the apoCIII gene is located within a gene cluster with apoAI, apoAIV and apoAV we confirmed that there were no off-target effects of the siRNA on these apolipoproteins.

Conclusion: Our data demonstrate that obese, insulin resistent and hyperglycemic mice, treated during eight weeks with siRNA targeting the apoCIII gene, decrease in BW, improve their insulin sensitivity, lipid- and glucose homeostasis without any observed side effects. Therefore, our siRNA strategy to decrease apoCIII might become a new tool in the fight against the metabolic syndrome.

Supported by: SDA, KIF, SRC, NNF, FEPF, SRP-DKI, FKAWF, SJF, SICL, DWF, BvKF, SD, AZ, SAD, SSoEGM, ERC-EYLETS 834860

Disclosure: P. Recio-López: None.

183

Loss of Hsp70 leads to increased albuminuria in a STZ-induced diabetic mouse model

J. Zemva1, R. Bulkescher1, T. Poth2, I. Hausser2, C. Rodemer1, A. Erhardt1, J.G. Okun3, S. Herzig4,5, J. Szendrödi1,5, P. Nawroth1,5;

1Department of Internal Medicine I and Clinical Chemistry, University Hospital Heidelberg, Heidelberg, 2Department of Pathology, University Hospital Heidelberg, Heidelberg, 3Dietmar-Hopp-Stoffwechselzentrum, University Hospital Heidelberg, Heidelberg, 4Helmholtz Diabetes Center, München-Neuherberg, 5Deutsches Zentrum für Diabetesforschung, München-Neuherberg, Germany.

Background and aims: The stress-inducible heat shock protein 70 (Hspa1a/Hspa1b) is part of the protein quality control system. It is involved in the maturation of newly synthesized polypeptides, stabilization and refolding of misfolded proteins as well as disaggregation of protein aggregates. Posttranslational modifications, for example by glycation during hyperglycemia, can lead to misfolding, aggregation and loss-of-function of the protein. The resulting advanced glycation endproducts are associated with the development of diabetic complications. The aim of this study was to investigate the effect of the loss of Hspa1a/Hspa1b in a STZ-induced diabetic mouse model in terms of the development of diabetic nephropathy.

Materials and methods: Diabetes was induced at the age of 10 weeks and mice were sacrificed at the age of 28 weeks. Urine was collected for 24hrs just before organ collection and albumine and creatinine were measured via ELISA. Protein expression was analyzed via Western Blot and Immunohistochemistry (IHC). Morphologic changes of diabetic nephropathy were assessed by IHC and electron microscopy.

Results: Hspa1a/Hspa1b knockout mice (KO) did not differ from wild-type mice (WT) in the control or STZ-treated group in respect to bodyweight or blood glucose levels. However, STZ-treated KO mice (KO+STZ) revealed to have a significantly higher albumine/creatinine ratio (ACR) compared to STZ-treated wild-type mice (WT+STZ) and to controls (KO-STZ) (KO+STZ: 15.73 mg/g; WT+STZ: 9.83 mg/g; KO-STZ: 6.77 mg/g; vs. WT+STZ p=0.04, vs. KO-STZ p=0.03; n=3-5 per group). Preliminary data suggest that both, KO+STZ and WT+STZ, show atrophic proximal tubuli in the outer layers of the kidney parenchyma with thickening of the basal membrane. Furthermore, WT+STZ mice seem to have decreased Hspa1a/Hsapa1b protein expression compared to WT-STZ mice.

Conclusion: Loss of the stress-inducible Hspa1a/Hspa1b leads to increased ACR in a STZ-induced diabetic mouse model. Interestingly, both KO+STZ and WT+STZ mice showed atrophic proximal tubuli in the outer region of the kidney parenchyma. Furthermore, we observed a downregulation of Hspa1a/Hspa1b in WT+STZ mice, which is in line with preliminary human data, where Hspa1a/Hspa1b expression seems to be lost in the course of diabetic nephropathy.

Supported by: German Research Foundation (DFG, CRC1118)

Disclosure: J. Zemva: Grants; DFG (CRC1118).

184

Involvement of gut-hormones in regulating female reproductive function in obese and incretin receptor knockout animal models

D. Khan, O.O. Ojo, A. Sridhar, P.R. Flatt, R.C. Moffett;

School of Biomedical Sciences, Ulster University, Coleraine, UK.

Background and aims: Metabolism and reproduction are interdependent with the hypothalamus acting as major control centre. Recent studies suggest correlation between energy intake and reproductive dysfunction such as polycystic-ovary-syndrome (PCOS) and ovulatory disturbances. Metabolic peptides including glucose-dependent-insulinotropic polypeptide (GIP), glucagon-like peptide-1 (GLP-1) and Peptide YY (PYY) have been suggested to play a role in the regulation of reproductive function. The present study evaluates disturbed reproductive health, altered gut/reproductive hormone receptor expression and fertility outcome induced by high-fat feeding and global knockout (KO) of GLP-1 or GIP receptors in rodent models.

Materials and methods: Female Wistar rats (4-weeks-old) fed with high-fat diet for 20-weeks were used with regular monitoring of metabolic parameters. Consecutive estrous cycles were observed using vaginal smears and rats were bred with normal diet fed Wistar males to assess fertility. At week 20, gut and reproductive gene expression in ovaries and adrenals were assessed using PCR. In a second series, effects of global GIPR and GLP-1R deletion on estrous cycling and reproductive outcomes were evaluated in C57BL/6 mice.

Results: Female rats after 20 weeks of high-fat feeding displayed significantly (p<0.05 to p<0.001) increased body weight and plasma insulin without change of blood glucose and HbA1c. 50% of the high-fat fed Wistar rats had prolonged average cycle length (≥7days). Fertility outcome showed 48% reduction in litter size and 16% were unable to attain pregnancy after prolonged high-fat feeding. 35% of pups born to high-fat-fed rats were eaten by their mothers or born dead whilst this phenomenon was absent with control fed rats. H&E staining revealed morphological changes in the ovaries with significant (p<0.01) increases in the number of cysts. Ovarian expression of Amh, Npy2R and GcgR genes were downregulated (p<0.01) while Glp-1r and Insr (insulin R) genes were upregulated (p<0.05-p<0.001) in high-fat rats. Expression of Glp-1R, GiprGshrInsr, Amh, Esr-1, Npy2R and GcgR genes were also upregulated (p<0.01 to p<0.001) in high-fat fed rat adrenals. In the transgenic model study, female GIPR and GLP-1R KO mice exhibited significantly (p<0.05 and p<0.01) deranged estrous cycling compared to wild-type controls. 50% and 16% of female GIPR and GLP-1R KO mice respectively produced litters with wild-type males across three breeding cycles. Consistent with functional role of incretin receptors in pregnancy outcome, litter size was significantly (p<0.001 to p<0.05) decreased in GIPR-/- and GLP-1R-/- mice.

Conclusion: High-fat feeding and incretin hormone receptor deletion disrupts reproductive function in females. This suggests important interactions between gut and reproductive hormones at level of ovaries and adrenals. Taken together with previous observations, these data suggest that incretin receptor modulation could represent a novel means for treating female reproductive disorders.

Supported by: DUK RD Lawrence Fellowship and UU strategic funding

Disclosure: D. Khan: None.

185

Adiporon improves endurance capacity and decreases ectopic lipid deposition in middle-aged obese mice

C. Selvais;

Institut de recherche experimentale et clinique/endocrinologie, diabète et nutrition, UCLouvain, Woluwe-Saint-Lambert, Belgium.

Background and aims: Obesity and ageing go hand in hand with loss of endurance capacity, insulin-resistance and features of the metabolic syndrome. Two additional burdens associated with this syndrome are non-alcoholic fatty liver disease (NAFLD)/steatohepatitis (NASH) and myosteatosis, whose severity increases with age. Myosteatosis is known to negatively correlate with muscle performance. Adiponectin (ApN) is an insulin-sensitizing, lipid-lowering hormone, which is decreased in the metabolic syndrome. Muscle and liver are two of its main target tissues. The aim of this study is to investigate whether an active ApN receptor agonist, AdipoRon (AR) could protect the muscle and the liver of middle-aged obese mice.

Materials and methods: Three groups of male mice were studied up to 14 months of age: one received a normal diet (ND), another a high fat diet (HFD) and the last a HFD combined with AR given orally (30 mg/kg/d scaled up to 50 mg/kg during the last part of the study; HFD+AR). Treadmill tests and micro-computed tomography (MCT) were carried out in vivo. Tissue analyses were performed by (immuno)histochemistry ex vivo.

Results: AR did not markedly alter diet-induced-obesity. Yet, this treatment, even when given at a low dose, rescued exercise endurance of obese mice during treadmill tests (P < 0.05), while improving insulin sensitivity only at the higher dose (P < 0.05). Density of dorsal muscles and liver, which was measured in vivo before the end of the study by MCT, was decreased in obese mice, suggesting fatty infiltration. This decrease was less pronounced in the HFD+AR group. These data were confirmed by (immuno)histochemistry. AR significantly decreased steatosis and cellular ballooning in the liver (P < 0.05), thus decreasing the NAFLD activity score. AR strikingly reversed (and actually even over corrected) intramyocellular lipid (IMCL) accumulation either due to ageing in oxidative fibers (types I and IIb, soleus) or to HFD in glycolytic ones (types IIx and IIb, extensor digitorum longus) (P < 0.05). Size of subsarcolemmal lipid droplets, known to be associated with adverse metabolic outcomes, was reduced as well.

Conclusion: In conclusion, AR enhances muscle endurance in obese mice, an event that precedes the improvement in insulin sensitivity. AR also protects obese middle-aged mice against burdens associated with the metabolic syndrome: NASH and myosteatosis, the effects on muscle being particularly impressive.

Supported by: FNRS, SFD

Disclosure: C. Selvais: None.

186

Analysis of the hypothalamic transcriptome controlling counter-regulatory responses to hypoglycaemia

J. Castillo-Armengol1, A. Rodriguez Sanchez-Archidona2, C. Fledelius1, B. Thorens2;

1Novo Nordisk A/S, Måløv, Denmark, 2Center for Integrative Genomics (CIG), University of Lausanne, Lausanne, Switzerland.

Background and aims: The development of hypoglycaemia during the treatment of diabetes mellitus by insulin injection is a major risk and one of the principal difficulties associated with the maintenance of normoglycaemia. After a hypoglycaemic episode, the probability of having subsequent episodes of hypoglycaemia increases. These defects in the counter-regulatory response (CRR) to hypoglycaemia are due to the suppression of the hormonal secretion of glucagon and catecholamines to restore normal glucose levels. The role of the central nervous system (CNS) is essential to activate it, but the modulators involved in this process and the mechanisms underlying the response to hypoglycaemia are poorly understood. In this context, we aim to identify novel regulators of hypoglycaemia counter-regulation in the murine brain in the context of diabetes and recurrent hypoglycaemia.

Materials and methods: A model of type-2 diabetes mellitus (T2DM) in mice exposed to acute (AH) or recurrent hypoglycemia (RH) was used. Counter-regulation was assessed by glucagon secretion. Single nuclei RNA sequencing (snRNAseq) was performed in the hypothalami of AH and RH mice. Differential expression and pathway analyses were performed in an overall dataset and in the different cell types resulting from the clustering identification.

Results: Recurrent hypoglycemia driven by insulin injections significantly impaired glucagon secretion in T2DM mice (AH, 74.96 ± 8.42 pg/mL vs RH, 49 ± 6.57 pg/mL, p<0.05). We studied the transcriptional profile of the hypothalami from AH and RH T2DM mice through snRNAseq. This data revealed 299 differentially expressed genes between AH and RH (FDR<0.05); 274 of these mRNAs were part of the neuron subpopulation. A subsequent analysis was done using Gene Ontology (GO) and Kyoto Encyclopedia of Genes and Genomes (KEGG) to determine the associated functions of these differentially expressed genes. This analysis revealed that the there was a strong association of these genes to mechanisms controlling synapse organization, regulation of membrane potential, neurotransmitter secretion and vesicle-mediated transport suggesting that RH impairs the connectivity of the cells altering the brain function in response to hypoglycaemia.

Conclusion: Here, we first established a mouse model of RH and impaired glucagon secretion in T2DM. Transcriptomic analysis of hypothalami revealed that RH and defective insulin-induced glucagon secretion were associated with important changes in genes controlling synaptic plasticity; surprisingly no changes in the expression of metabolic or glucose signaling-related mRNAs were detected. Thus, this investigation suggests that hypoglycemia-associated autonomic failure may be caused by a general defect on synaptic activity rather than in glucose sensing pathways.

Supported by: Novo Nordisk A/S, Innovative Medicines Initiative 2 Joint Undertaking (JU) under grant agreement No 777460.

Disclosure: J. Castillo-Armengol: Grants; It has received funding from the Innovative Medicines Initiative 2 Joint Undertaking (JU) under grant agreement No 777460.

OP 32 Benefits of GLP-1: from traditional to non-traditional complications

187

Liraglutide may induce impaired diastolic heart function by activation of sympathetic tonus: A group effect of this class of drugs?

S.B. Haugaard1, P. Kumaraturai2, C. Anholm3, O.W. Nielsen2, A. Sajadieh2;

1Endocrinology, Bispebjerg University Hospital, Copenhagen, 2Cardiology, Bispebjerg University Hospital, Copenhagen, 3Endocrinology, Hillerød University Hospital, Hillerød, Denmark.

Background and aims: Glucagon-like-peptide 1 receptor agonists (GLP1-RAs) are a group of drugs not only used in treatment of type 2 diabetes but also to treat obesity. The widely used GLP1-RAs semaglutide and liraglutide have been shown to increase heart rate, maybe due to increased sympathetic activation. We have shown that liraglutide reduces heart rate variability, reduces diastolic heart function (e’), and increases heart rate in obese people with newly diagnosed type 2 diabetes and stable coronary heart disease. We suggest that activation of the sympathetic tonus as reflected in increased heart rate and decreased heart rate variability may be associated with impaired diastolic function of the heart.

Materials and methods: Out of 41 people with stable coronary artery disease and newly diagnosed type 2 diabetes 30 people completed a 12 plus 12 weeks over-crossed placebo-controlled clinical trail with 2 weeks of washing out of antidiabetic therapy before and between the intervention periods. The people taking part in the study undertook 24 hours HOLTER monitoring and a trans-thoracic echocardiography at weeks 0, 12, 14 and 26. They were treated by metformin as backbone and liraglutide titrated to 1.8 mg q.d. and corresponding placebo.

Results: Heart rate increased by 11% (p=0.003) and heart rate variability (SDNN) decreased independent of the increase in heart rate by 25% (p<0.001) during treatment with liraglutide. Diastolic filling of the heart as measured by e’ was impaired by 10% (p<0.02) during liraglutide therapy. It was observed that the increase in mean heart rate during 24 hours was strongly associated with decrease in e’ (R2= 16%, p<0.01). The decrease in SDNN during sleep (from midnight to 2 am.) was also associated with decrease in e’ (R2=10%, p<0.02).

Conclusion: Activation of the sympathetic tonus during therapy with the GLP1-RA, liraglutide may result in increased heart rate and reduced heart rate variability. Our data suggest an association between increased heart rate and reduced heart rate variability and impaired diastolic function of the heart in obese people with stable coronary heart disease and newly diagnosed type 2 diabetes undergoing therapy with liraglutide. It should be addressed whether this observation is a group effect of GLP1-RAs, and if so, whether drug classes with anti-chronotropic and lusitropic effects will abolish this effect on heart rate and heart rate variability with possibly positive effects on diastolic function of the heart of people with type 2 diabetes.

Clinical Trial Registration Number: NCT01595789

Supported by: Novo Nordisk A/S, Danish Heart Foundation, A.P. Møller Foundation

Disclosure: S.B. Haugaard: Grants; From Novo Nordisk A/S.

188

Liraglutide reduces cardiac adipose tissue in type 2 diabetes: results from the LiraFlame randomised controlled trial

T. Hansen1, I. Rasmussen1, E. Zobel1, R. Ripa2,3, B. von Scholten1,4, V. Curovic1, A. Kjaer2,3, P. Rossing1,3;

1Steno Diabetes Center Copenhagen, Gentofte, 2Department of Clinical Physiology, Nuclear Medicine & PET and Cluster for Molecular Imaging, Rigshospitalet, Copenhagen, 3University of Copenhagen, Copenhagen, 4Novo Nordisk A/S, Soeborg, Denmark.

Background and aims: An increased amount of cardiac adipose tissue is associated with a higher risk of cardiovascular disease and mortality in persons with type 2 diabetes. Recent studies have shown that treatment with GLP-1 receptor agonists can reduce the risk of major adverse cardiovascular events. We hypothesized that 26 weeks of treatment with liraglutide was associated with a reduction in cardiac adipose tissue as compared to placebo.

Materials and methods: We conducted a randomized placebo-controlled, double-blind, parallel clinical study, where participants with type 2 diabetes were randomized to treatment with liraglutide 1.8 mg/day or placebo for 26 weeks. Computed tomography was performed before and at end-of-treatment to evaluate the cardiac adipose tissue volume, which was automatically quantified. We report the results of a secondary endpoint evaluating the change in cardiac adipose tissue.

Results: A total of 102 participants were included and randomly assigned to receive liraglutide (n=51) or placebo (n=51). The participants had a mean age of 66.4 (SD 8.2) years, 15.7% were females and median diabetes duration was 10.9 [IQR 5.7 - 18.2] years. Baseline clinical characteristics were balanced between the two treatment groups, except for triglycerides. From baseline to end-of-treatment mean change in HbA1c was -5.1 (95% CI -8.1, 2.0) mmol/mol in the liraglutide group and -0.1 (-1.9, 1.7) mmol/mol in the placebo group. The mean change in body weight was -3.7 (-4.8, -2.6) kg in the liraglutide group and -0.18 (-0.76, 0.40) kg in the placebo group. At baseline, the mean (SD) cardiac adipose tissue volume was comparable between the liraglutide and the placebo group [232.6 (112.8) vs 227.0 (103.2) ml, p=0.80]. From baseline to end-of-treatment the mean cardiac adipose tissue change was -11.5 (95% CI: -17.6, -5.4) ml in the liraglutide group (p<0.001) and -0.01 (-5.3, 5.3) ml in the placebo group (p=1.00) (Figure 1). The reduction in cardiac adipose tissue was significantly larger in the liraglutide group compared to the placebo group (mean difference: -11.4 (-19.4, -3.3) ml, p=0.005).

Conclusion: Treatment with liraglutide for 26 weeks was associated with a reduction in cardiac adipose tissue compared to placebo, suggesting a possible mechanism of the cardioprotective benefits of liraglutide observed in outcome studies.

figurebh

Clinical Trial Registration Number: NCT03449654

Supported by: NN

Disclosure: T. Hansen: None.

189

Positive impact of liraglutide on pulmonary function in patients with type 2 diabetes: data from the randomised cross-over LIRALUNG study

C. López-Cano1, A. Ciudin2, E. Sánchez1, F. Tinahones3, A. Soto4, S. Pellitero5, F. Barbé6, M. Dalmases6, M. García-Ramírez2, A. Gaeta6, R. Martí1, C. Hernández2, R. Simó2, A. Lecube1;

1Endocrinology and Nutrition, Hospital Universitari Arnau de Vilanova, Lleida, 2Endocrinology and Nutrition, Hospital Universitari Vall d'Hebron, Barcelona, 3Endocrinology and Nutrition, Hospital Universitario Virgen de la Victoria, Málaga, 4Endocrinology and Nutrition, Hospital Universitario A Coruña, A Coruña, 5Endocrinology and Nutrition, Hospital Universitari Germans Trias i Pujol, Badalona, 6Pneumology, Hospital Universitari Arnau de Vilanova, Lleida, Spain.

Background and aims: There is experimental evidence that glucagon-like peptide 1 receptor (GLP-1R) agonists ameliorates lung fibrosis and stimulates surfactant production. Therefore, our aim was to evaluate the effect of liraglutide (a GLP-1R agonist) on pulmonary function and the serum levels of surfactant in type 2 diabetes.

Materials and methods: A double-blind, randomized, crossover, placebo-controlled clinical trial comprising 76 patients with a baseline forced expiratory volume in first second <90% of predicted. Liraglutide was administered for 7 weeks (2 weeks of titration plus 5 weeks at 1.8 mg daily). This short duration was intentional in order to minimize weight loss as a potential confounding factor. Lung function parameters included: FEV1, FVC, peak expiratory flow (PEF), maximum mid-expiratory flow (FEF25-75). Serum surfactant protein D was also assessed.

Results: Liraglutide significantly reduced HbA1c [Δ: -1.2% (95%CI: -1.5 to -0.9)] and the BMI [Δ: -0.5 kg/m2 (-0.8 to -0.2)] and exerted a positive impact on forced vital capacity [ΔFVC: 5.2% of predicted (0.8 to 9.6)] in comparison with placebo. No differences in the other pulmonary variables were observed between groups. Participants under liraglutide treatment also experienced a decrease in serum surfactant protein D [196.4 (128.2 to 271.4) to 169.6 (108.1 to 233.6), p=0.038]. The absolute change in FVC correlated with the final serum surfactant protein D in participants receiving liraglutide (r=-0.313, p=0.036). Stepwise multivariate regression analysis showed that the final serum surfactant protein D independently predicted changes in FVC.

Conclusion: Liraglutide exerts a positive impact on FVC and surfactant in patients with type 2 diabetes.

figurebi

Clinical Trial Registration Number: NCT02889510

Supported by: Novo Nordisk S.A. (Investigator Sponsored Study)

Disclosure: C. López-Cano: Grants; This study was supported by a grant from Novo Nordisk S.A. (Investigator Sponsored Study).

190

Effect of subcutaneous semaglutide on features of the metabolic syndrome in patients with non-alcoholic steatohepatitis

L.L. Gluud1, I. Bakulin2, S. Ladelund3, P.N. Newsome4, A. Rendon3, A.-S. Sejling3, M.E. Tushuizen5, K. Cusi6;

1Copenhagen University Hospital Hvidovre, Hvidovre, Denmark, 2Mechnikov North-Western State Medical University, Saint-Petersburg, Russian Federation, 3Novo Nordisk A/S, Søborg, Denmark, 4University Hospitals Birmingham NHS Foundation Trust, Birmingham, UK, 5Leiden University Medical Center, Leiden, Netherlands, 6University of Florida, Gainesville, USA.

Background and aims: Non-alcoholic fatty liver disease and non-alcoholic steatohepatitis (NASH) are closely linked to insulin resistance, visceral obesity and features of the metabolic syndrome. Treatment options for patients with NASH are currently limited. Here, the effects of semaglutide on parameters associated with metabolic syndrome are reported in patients with NASH.

Materials and methods: In a 72-week phase 2 trial, 320 patients with NASH and fibrosis stage F1-F3 were randomised to subcutaneous semaglutide (0.1, 0.2 or 0.4 mg once daily) or placebo. Treatment with semaglutide 0.4 mg led to a significantly greater proportion of patients having NASH resolution without worsening of fibrosis (primary endpoint). Changes in metabolic and inflammatory parameters are reported here for all randomised patients during the in-trial period; HbA1c and fasting plasma glucose are reported in patients with T2D only, and HOMA-IR and adipose tissue insulin resistance index (Adipo-IR: fasting plasma insulin × free fatty acids) in patients not treated with insulin at baseline. Correlations with primary endpoint (NASH resolution) are reported in all patients on-treatment at week 72 (semaglutide 0.4 mg and placebo combined).

Results: A total of 320 patients were randomised to treatment (semaglutide 0.1 mg, n=80; 0.2 mg, n=78; 0.4 mg, n=82; or placebo n=80), of whom 62% (n=199) had type 2 diabetes and 49% (n=158) were treated with insulin at baseline. Semaglutide 0.4 mg was associated with significantly greater improvements than placebo in waist circumference, HbA1c, fasting plasma glucose, HOMA-IR, Adipo-IR, HDL cholesterol, triglycerides and high-sensitivity C-reactive protein (Table). Changes in non-HDL cholesterol and blood pressure were not significantly different between semaglutide 0.4 mg and placebo. Improvements in waist circumference, HbA1c (both p<0.001), fasting plasma glucose (p=0.006), HOMA-IR (p=0.017), Adipo-IR (p=0.015) and HDL cholesterol (p<0.001) were correlated with achieving NASH resolution without worsening of fibrosis (primary endpoint).

Conclusion: Once-daily subcutaneous semaglutide 0.4 mg in patients with NASH, in addition to weight loss and improved glycaemic control, resulted in significant improvements in multiple features of the metabolic syndrome. Changes from baseline in metabolic parameters were correlated with resolution of steatohepatitis.

figurebj

Clinical Trial Registration Number: NCT02970942

Supported by: Funded by Novo Nordisk A/S

Disclosure: L.L. Gluud: Employment/Consultancy; Novo Nordisk (consultancy only). Grants; Novo Nordisk, Alexion, Gilead. Lecture/other fees; Novo Nordisk, Norgine.

191

Multi-target engagement effect of a novel long-acting Glucagon/GIP/GLP-1 triple agonist (HM15211) in animal model of NASH

J. Choi, J. Lee, J. Kim, H. Kwon, E. Park, J. Lee, D. Kim, Y. Kim, I. Choi;

Hanmi Phaarm. Co., Ltd, Seoul, Republic of Korea.

Background and aims: NASH is a complex disease progressed by multiple mechanism. To date, no approved drug is available due to unsatisfied efficacy. Of note, recent studies demonstrate advantages of targeting multiple aspects of this disease. To aid multi-targeting and provide a novel treatment option with improved efficacy, HM15211, a long-acting Glucagon/GIP/GLP-1 triple agonist, has been developed. Here, enhanced anti-steatosis effect by HM15211 were investigated in diet induced obese mice (DIO mice) and potential benefits of HM15211 was evaluated by comparing its therapeutic effects with NASH drug candidates in high-fat diet induced mouse model of NASH.

Materials and methods: To investigate the anti-steatosis effect of HM15211, HM15211 was administered to DIO mice and liraglutide was used as comparator. After 4 weeks treatment, hepatic triacylglycerol (TG), and blood TG and cholesterol were measured. To unveil the MoAs for improved hepatic lipid metabolism by HM15211, expression level of genes involved in β-oxidation and de novo lipogenesis was determined by qPCR analysis. Next, to evaluate potential benefits of HM15211 on NASH, AMLN-diet induced NASH mice were administered either with HM15211 or obeticholic acid (OCA) for 12 weeks and histological analysis and blood ALT level as well as hepatic TG analyzed at the end of treatment. Additional study was performed to compare therapeutic effect of HM15211 with incretin analogs such as acylated GLP-1 or GLP-1/GIP in AMLN-diet induced NASH mice, followed by histological analysis.

Results: In DIO mice, greater reduction in hepatic TG (-63.6%, -80.6% vs. vehicle for liraglutide, HM15211) and blood TG (-33.4%, -73.7% vs. vehicle for liraglutide, HM15211) were confirmed for HM15211 treatment compared to liraglutide, which was well correlated with favorable reprogramming of hepatic lipid metabolism-related gene expression. In AMLN mice, HM15211 treatment more efficiently normalized hepatic lipid contents (-48.9%, -93.0% vs. vehicle for OCA, HM15211) and steatosis score (2.9, 2.1, 0.1 for vehicle, OCA, HM15211) compared to OCA treatment. Consistently, HM15211 treatment significantly reduced histological score for lobular inflammation (1.6, 1.1, 0.9 for vehicle, OCA, HM15211) and ballooning (1.4, 0.7, 0.0 for vehicle, OCA, HM15211) along with blood ALT (546, 349, 138 U/L for vehicle, OCA, HM15211) even greater than OCA treatment. Notably, while all individuals (7/7) treated with HM15211 achieved NASH resolution criteria, only 14.3% (1/7) achieved it after OCA treatment. Similarly, HM15211 treatment was associated with greater reduction in all sub-components of NAS when compared to acylated GLP-1 or GLP-1/GIP treatment in AMLN mice.

Conclusion: Considering more benefits of HM15211 over other incretin analogs and FXR agonist for NASH treatment, HM15211 may be a novel therapeutic option for NASH. Efficacy study in biopsy proven NASH patients is ongoing to assess the clinical relevance of these finding.

Disclosure: J. Choi: None.

192

In patients with fatty liver, higher fasting GLP-1 levels are associated with increased insulin resistance and reduced beta hydroxybutirate

G. Mocciaro1, F. Carli1, M. Gaggini1, C. Barbieri1, B. Patricio1, C. Rosso2, A. Armandi2, E. Lembo3, E. Bugianesi2, G. Mingrone3, A. Gastaldelli1;

1Cardiometabolic Risk Unit, Institute of Clinical Physiology, CNR, Pisa, 2Department of Medical Sciences, Division of Gastroenterology and Hepatology, University of Turin, Turin, 3Department of Internal Medicine, Catholic University Sacro Cuore, Rome, Italy.

Background and aims: Non-alcoholic fatty liver disease (NAFLD) is a spectrum of diseases ranging from simple steatosis to more aggressive forms and is a major risk factor for type 2 diabetes (T2DM). Treatment with glucagon-like peptide 1 (GLP-1) receptor agonists ameliorates NAFLD. However, less is known about GLP-1 levels in NAFLD. Thus, we investigate whether fasting levels of GLP-1 were able to distinguish NAFLD patients with different severity and metabolic phenotypes.

Materials and methods: We studied 40 nondiabetic adults with NAFLD, diagnosed by liver biopsy (n=25) or ultrasound (Table). Besides clinical biochemistry tests (Table) we measured metabolic fluxes by tracer infusion, HOMA-IR, hepatic insulin resistance (Hep-IR = endogenous glucose production•insulin), adipose tissue IR (AT-IR = NEFA•insulin), GLP-1 by ELISA and lipidomic profile by LC-MS.

Results: The cohort comprised two groups one with BMI ≤ 30 and one with BMI ≥ 35 (Table). Despite the remarkable difference in BMI, glucose, GLP-1, lipids, liver enzymes, HOMA-IR and Hep-IR and NAFLD Activity Score (NAS) were not different while insulin and AT-IR were higher in obese (Table). To assess the impact of fasting GLP-1 on disease status, we divided our cohort according to the median values of GLP-1 into a low group (GLP-1 < 57.5 pmol/l) and a high group (GLP-1 ≥ 57.5 pmol/l). The two groups had similar age and BMI (Table 1). However, insulin, liver enzymes, and NAS were significantly increased in the high GLP-1 group as compared to the low GLP-1 group (Table). Compared to the low GLP-1, the high GLP-1 group showed increased NAS and insulin resistance as measured by HOMA-IR, Hep-IR and AT-IR (Table). We assessed the circulating metabolite β-hydroxybutyrate as a marker of hepatic lipid mitochondrial oxidation that was reduced in the high vs. low GLP-1 group (0.06 ± 0.04 vs. 0.11 ± 0.11 p<0.05). Lastly, preliminary lipidomic analyses highlighted among the best lipid discriminants the triacylglycerol with saturated and monounsaturated fatty acids this being in agreement with the elevated de novo lipogenesis observed in IR.

Conclusion: In NAFLD patients, increased fasting GLP-1 levels associate with a worst metabolic phenotype as shown by higher NAS, circulating liver enzymes, and insulin resistance.

figurebk

Supported by: Funded by: Horizon2020 under grant agreement: no 634413, Epos

Disclosure: G. Mocciaro: None.

OP 33 Diabetic foot problems: from prediction to treatment

193

Metabolomic risk predictors of diabetic foot ulcers

J. Hedegaard Andersen1, T. Suvitaival1, K. Trošt1, S. Theilade1,2, I. Mattila1, A. Rasmussen1, M. Frimodt-Møller1, P. Rossing1,2, C. Legido-Quigley1, T.S. Ahluwalia1,2;

1Steno Diabetes Center Copenhagen, Gentofte, 2University of Copenhagen, Copenhagen, Denmark.

Background and aims: Diabetic foot ulcers (DFUs) are among the most severe complications associated with diabetes. The severity is highlighted by the evidence of increased mortality and morbidity and more than 70% of all amputations being preceded by a DFU. In recent years, the main strategy of dealing with DFUs has shifted towards prevention. Even though prevention has been successful in lowering the incidence of DFUs, the lifetime risk of DFUs for individuals with diabetes is still 15-33% warranting research on risk markers with the potential of predicting DFUs. In this study, we investigated the association between circulating plasma metabolites and DFUs in subjects with type 1 diabetes (T1D).

Materials and methods: Plasma metabolites (n=75) and clinical characteristics from 637 individuals with T1D recruited from Steno Diabetes Center Copenhagen as part of a cross-sectional study were assessed. Baseline characteristics, DFU diagnoses at baseline and longitudinal data on development of DFU were retrieved from electronic patient journals. Associations between single metabolites and DFU were evaluated by linear regression analyses at baseline and by Cox proportional hazards model at follow-up. Models were fitted with and without adjustments (age, gender, body mass index, systolic blood pressure, cholesterol, HbA1c, smoking, statin, triglycerides, eGFR and urinary albumin excretion) and corrected for multiple testing.

Results: Participants had a mean age of 54 (IQR 46, 62) years, 55% were male (n=348), diabetes duration 35 (25, 44) years, HbA1c 64 (56, 72) mmol/mol and eGFR 85 (64, 102) ml/min/1.73m2. In total 19 participants had DFU at baseline, and a further 108 developed DFU during a median follow-up of 10 years. In the crude model, 11 metabolites at baseline were associated with future risk of DFU. After adjustment, higher levels of ribonic acid exhibited a significantly elevated risk of future DFUs (HR 1.38(1.06-1.8) p<0.05). Figure 1 shows DFU-free survival of participants stratified by ribonic acid at baseline measurement.

Conclusion: In this study, we identified several circulating metabolites associated with future risk of DFU in individuals with T1D. After adjustment for potential confounders and multiple testing, a sugar derivate (i.e. ribonic acid) retained significant association to future development of DFUs. Previous studies have demonstrated that ribonic acid is related to other complications (i.e. kidney disease and retinopathy) and this study adds to theses findings as well as the growing evidence of predicting DFUs via measurements of plasma metabolites.

figurebl

Disclosure: J. Hedegaard Andersen: None.

194

IDR-1018 peptide improves wound healing in vitro and when topically applied to skin wounds in a model of type 1 diabetes

M. Petkovic1,2, M.V. Mouritzen1, N. Molchanova1, K. Qvist1, P. Nyeng1, E.C. Leal2, E. Carvalho2, L.T. Dalgaard1, H. Jenssen1;

1Roskilde University, Roskilde, Denmark, 2Center for Neuroscience and Cell Biology, Coimbra, Portugal.

Background and aims: The role of skin antimicrobial peptides in wound healing, highly dysregulated in diabetes, is under intense investigation. We aimed to evaluate the impact of synthetic peptide IDR-1018 in wound healing, using an in vitro assay followed by pathway analysis of proteomics data. In addition, the effect of topical dermal treatments of IDR-1018 on wound healing kinetics, angiogenesis, and autophagy during wound healing in diabetic mice was evaluated.

Materials and methods: Human keratinocyte cells (HaCaT) were cultured in DMEM medium and Human Umbilical Vein Endothelial cells (HUVEC) were grown in endothelial cell basal medium. For scratch migration assays, cells were stimulated with 25 μg/mL IDR-1018, media, 500 ng/mL EGF, or 25 μg/mL free amino acids. Endothelial tube-formation assay was done to estimate in vitro angiogenesis. As an inflammation model, TNFα stimulated HaCaT cells underwent proteomics followed by pathway analysis for Gene ontology (GO: BP, GO: CC) categories. Diabetes was induced in C57BL/6 mice using low dose streptozotocin injections for 5 consecutive days. Wounds were topically treated with 25 μg, 12.5 μg IDR-1018 or saline. Wound-healing kinetics were evaluated up to day 10. Angiogenesis, inflammation, and autophagy of ubiquitinated cargo in murine diabetic skin were assessed by immunofluorescence for PECAM-1, TNFα and SQSMT1 (p62) in mouse skin sections 10 days after full thickness wounding.

Results: There were no proliferative effects of IDR-1018 on HaCaT nor HUVEC cells. The IDR-1018 treatment had an inhibitory effect on HaCaT cell migration (0.58-fold ± 0.28, NS), whereas the migration of the HUVEC cells was enhanced (2.3-fold ± 1.1, NS). In the tube-formation assay IDR-1018 increased the number of branches (101 branch points ± 13) compared with untreated HUVEC cells (57 branch points ± 14, p<0.0001). TNFα levels in skin were increased by high dose IDR-1018 (2-fold ± 0.6, p<0.05). In terms of biological processes, the majority of significantly altered GOBP categories were related to over-representation of proteins targeting to ER (FDRlog10=4.5) and cellular protein catabolic processes (FDRlog10=4.2) while majority of GO:CC were related to cell-cell (FDRlog10= 2.1) and cell-substrate adherence (FDRlog10=2.8). The high dose of IDR-1018 (25 μg/wound) promoted faster wound closure on days 9 and 10 in mice wounds (23 % ± 14 and 11 % ± 8, p<0.05) compared to saline (43 % ± 17 and 25 % ± 17, p<0.05). The number of blood vessels was elevated with low dose (2.6-fold ± 1.3, p<0.05). Autophagy was elevated in high dose (18±3 cells) compared to low dose (10±1 cells, p<0.01) and compared to saline (13±2 cells, p<0.05).

Conclusion: IDR-1018 peptide promoted angiogenesis in HUVEC cells. Moreover, exogenous delivery of IDR-1018 peptide to wounds accelerated wound closure in diabetic mice, leading to an enhanced vascularization and a decreased autophagy. Our findings suggest that antimicrobial peptide IDR-1018 may improve the healing outcome under diabetic conditions.

Supported by: EFSD/Novartis European Research Programme in Microvascular Complications of Diabetes 2015, EASD/DDA/SPD

Disclosure: M. Petkovic: None.

195

Protein tyrosine phosphatase 1B inhibition promotes diabetic wound healing via activation of the antioxidant enzyme heme oxygenase 1

E.C. Leal1,2, A. Figueiredo1,2, D. Santos1,2, M. Delibegovic3, E. Carvalho1,2;

1Faculty of Medicine, Polo I, Piso 1, Center for Neuroscience and Cell Biology, University of Coimbra, Coimbra, Portugal, 2Institute of Interdisciplinary Research, University of Coimbra, Coimbra, Portugal, 3University of Aberdeen, Aberdeen, UK.

Background and aims: Diabetic foot ulcers are a leading cause of hospital admissions for people with diabetes in the developed world. Protein tyrosine phosphatase 1B (PTP1B) is a negative regulator of the insulin signaling pathway and it is upregulated in diabetes and diabetic foot ulcers. Evidence has emerged that PTP1B is involved in the pathology of diabetic ulcers which indicates that its inhibition may be a potential therapeutic approach. Our main aim was to evaluate the role of PTP1B inhibition in diabetic wound healing and study the underlying mechanisms.

Materials and methods: Diabetes in male C57BL/6 mice was induced with streptozotocin (STZ), i.p. injections (50 mg/Kg), for 5 consecutive days. After 6 weeks of diabetes, two 6 mm excision wounds per mouse were created in the dorsum and the wounds were treated topically, twice a day, up to day 3 with: vehicle (control); 1 μg of Trodusquemine (PTP1B inhibitor); 1 μg of Protoporphyrin IX zinc (II) (heme oxygenase 1 inhibitor - HO1i) and PTP1Bi+HO1i. The wounds were measured every day up to day 10 post wounding. The skin was collected at day 3 and 10 post wounding. Immunohistochemistry was used for the detection of M1 macrophages, with CD68 and TNF-alpha, or M2 macrophages, with CD68 and CD206. Immunohistochemistry was also used to evaluate the vascularization of the skin with the endothelial cell marker CD31, the proliferation with ki67, and the levels of heme oxygenase 1 (HO1). The levels of PTP1B were measured by western blot and the activity of HO1 was evaluated by the production of bilirubin. The production of reactive oxygen species (ROS) was measured with dihydroethidium (DHE).

Results: PTP1B levels were significantly increased in unwounded (474±150 % of control, p<0.01) and wounded skin (day 3 post wounding, 398±68 % of control, p<0.01) of diabetic mice when compared to healthy animals. The treatment with PTP1B inhibitor improved diabetic wound healing progression when compared to non-treated wounds (day 10, 1.1±0.2% and 7.2±1.4% of original wound, p<0.01). Moreover, PTP1B inhibition decreased the inflammatory environment in the wounds with a decrease in M1/M2 ratio (day 3, 0.7±0.1 and 1.6±0.2, p<0.01). In addition, the increase in angiogenesis (147.4±5.5 % of control, p<0.01) and cell proliferation (141.1±5.9 % of control, p<0.01), after 10 days of wound induction, was also observed. The oxidative stress, in diabetic wounds after 3 days post wounding, was significantly decreased by PTP1B inhibition (68.1±6.1 % of control, p<0.01) and the increase in HO1 levels (219.4±20.7 % of control, p<0.01) and activity (170.4±15.7 % of control, p<0.01) was also observed. These effects of the inhibition of PTP1B were reverted by HO1 inhibition.

Conclusion: PTP1B inhibition promotes wound healing in diabetes with decrease in the inflammatory environment and the oxidative stress which improves the regenerative capacity of the skin with an enhance in angiogenesis and proliferation, through an increase of the antioxidant defense HO1. This study suggests that PTP1B is a target of interest for the treatment of diabetic foot ulcers.

Supported by: Diabetes UK, SPD/GIFT

Disclosure: E.C. Leal: None.

196

A M1/M2-macrophages-regulating new drug for diabetic foot ulcers with poor-controlled HbA1c risk factor in an International Phase 3 study

M.-L. Kuo1, S.-C. Chang2;

1Oneness Biotech Company Limited, Taipei, 2Department of Surgery, Taipei Medical University-Shuang Ho Hospital, Ministry of Health and Welfare, New Taipei, Taiwan.

Background and aims: The diabetic foot ulcers (DFUs) has been a life-threatening and complication of diabetes mellitus. DFU chronicity is attributed by the stalled healing progress in inflammation stage due to hyperglycemia triggering overabundance of inflammatory macrophages and the dysregulated M1/M2 macrophages in the lesion. Restoring the balance of M1/M2 macrophages plays a critical role in orchestrating the healing process of DFUs. Recognizing the therapeutic potential by regulating M1/M2-macrophage ratio in DFUs, we aim to evaluate the efficacy and safety of ON101, a topical new drug with M1/M2-macrophage-regulating mechanism versus a hydrocolloid dressing in a multicentre, randomized, controlled, evaluator-blind phase 3 study in 21 clinical/medical centers across the US, China, and Taiwan by following Guidance for Industry Chronic Cutaneous Ulcer and Burn Wounds - Developing Products for Treatment, the International Conference on Harmonization guidelines, and investigational new drug programs.

Materials and methods: 236 eligible patients with post-debrided DFUs between 1 - 25 cm2, present for ≥4 weeks and Wagner Grade 1 or 2, were randomized 1:1 to receive ON101 or the hydrocolloid dressing for up to 16 weeks followed by a 12-week follow-up. Standardized digital photo-taking, study materials, standard of care instruction and blinded evaluation were implemented thoroughly. The primary outcome was incidence of complete healing, defined as complete re-epithelization at 2 consecutive visits during the treatment period assessed by the full analysis set (all participants with post-randomization data collected). The secondary outcome was time to complete ulcer healing. Safety outcomes included assessment of the incidence of adverse events, clinical laboratory values, and vital signs.

Results: Baseline demography and medical history between the two groups were well balanced. The incidence of complete healing as the primary endpoint in overall group in full analysis set was 60.7% by ON101 and 35.1% by the comparator during the treatment period (p=0.0001). Subgroup analysis on the primary efficacy variable was also conducted on DFU patients with poor-controlled baseline HbA1c risk factor (defined ≧9% according to definition of ADA) and found a significant odds ratios in favor of the ON101 group (p=0.0353). Time to complete ulcer healing as the secondary endpoint was noted faster in the ON101 group (p=0.002). No clinically significant changes or differences between 2 treatment groups in hematology, biochemistry (including HbA1c and fasting glucose), or vital signs. Related treatment of emergent adverse events was reported in 7 patients (5.7%) in the ON101 group and 5 (4.4%) from the comparator group. None of the serious adverse events was related to study drug while there was a case of osteomyelitis reported to be related to the comparator.

Conclusion: Hyperglycemia is an underlying cause of the chronicity of DFUs where M1-to-M2 macrophage transition delays and inflammation stage is prolonged. ON101, recently approved by Taiwan Food and Drug Administration, has been shown with clinically robust efficacy and well-tolerated safety profile in overall DFUs, as well as in DFU patients with poor glycemic control. This indicates that ON101 provides not only a new prospect in DFU management but a differentiated approach towards active-healing.

Clinical Trial Registration Number: NCT01898923

Supported by: Trial sponsorship by Oneness Biotech Company Limited

Disclosure: M. Kuo: Employment/Consultancy; Oneness Biotech Company Limited.

197

Long-term outcomes of autologous cell therapy, angioplasty and conservative therapy in patients with chronic limb-threatening ischaemia and diabetes

M. Dubsky1,2, J. Husakova1,2, R. Bem1, V. Fejfarova1, R. Jarosikova1, A. Jirkovska1, V. Woskova1;

1Institute for Clinical and Experimental Medicine, Prague, 2First Faculty of Medicine, Charles University, Prague, Czech Republic.

Background and aims: Long-term clinical outcomes of revascularization, especially by autologous cell therapy (ACT), in diabetic patients with chronic limb-threatening ischemia (CLTI) remain unclear. The aim of our study was to compare the mortality and amputation rates of patients with diabetic foot (DF) and CLTI treated by ACT with patients treated by repeated percutaneous transluminal angioplasty (re-PTA) and patients treated conservatively.

Materials and methods: One-hundred and thirty patients with DF and CLTI (defined as transcutaneous oxygen pressure - TcPO2 ˂ 30 mmHg after unsuccessful standard revascularization) treated in our foot clinic over 7 years were enrolled into the study. Forty-five patients were treated by ACT, 43 patients underwent re-PTA and 42 patients were treated conservatively and formed the control group. Mortality and major amputation rate were assessed over a 5-year follow-up period.

Results: Patients in all groups did not differ significantly in demographic characteristics. The frequency of comorbidities (hypertension, ischemic heart disease, end-stage kidney disease) also did not differ significantly among the groups. Patients in ACT and control groups had significantly more severe angiographic findings according to Graziani classification than the re-PTA group (5.0±0.9 and 5.1±0.8 vs. 3.5±1.1, p˂0.001), but there were no differences in baseline values of TcPO2 among all groups. The rate of major amputation after 5 years was significantly lower in ACT and re-PTA groups in comparison with control group (28.9% and 20.9% vs. 64.2%, p=0.011 and 0.002 respectively). There was a trend to lower mortality in ACT group and significantly lower mortality in re-PTA group in comparison with control group (35.6% and 25.6% vs. 61.9%, p=0.09 and 0.012 respectively).

Conclusion: Our study showed significantly lower long-term amputation rate and increased survival in patients treated by ACT and re-PTA in contrast to patients treated conservatively. Bone-marrow derived autologous cell therapy is a promising method for the treatment of CLTI in diabetic patients, comparable with re-PTA, even in patients with no-option CLTI.

Supported by: the Ministry of Health of the Czech Republic, grant no. 00023001

Disclosure: M. Dubsky: None.

198

Diabetes is not associated with major amputation after open vascular surgery for chronic limb-threatening ischaemia: a nationwide propensity score analysis

E. Lilja1, A. Gottsäter1, M. Miftaraj2, J. Ekelund2, B. Eliasson3, A.-M. Svensson2, M. Zarrouk1, S. Acosta1;

1Department of Clinical Sciences, Lund University, Malmö, 2National Diabetes Register, Centre of Registers, Gothenburg, 3Department of Molecular and Clinical Medicine, Institute of Medicine, University of Gothenburg, Gothenburg, Sweden.

Background and aims: The risk of major amputation is higher after urgently planned endovascular therapy for chronic limb-threatening ischemia (CLTI) in patients with diabetes mellitus (DM). The aim of this nationwide cohort study was to compare outcomes between patients with and without DM following urgently planned open revascularization for CLTI from 2010 to 2014.

Materials and methods: Out of 1537 individuals registered in the Swedish Vascular Registry, 569 were registered in the National Diabetes Register. A propensity score adjusted Cox regression analysis was conducted to compare outcome between the groups with and without DM. Median follow-up was 4.3 years and 4.5 years for patients with and without DM, respectively.

Results: Patients with DM more often had foot ulcers (p=0.034) and had undergone more previous amputations (p=0.001) at baseline. No differences in mortality, cardiovascular death, major adverse cardiovascular events (MACE), or major amputation were observed between groups. The incidence rate of stroke was 70% higher (95% CI 1.11-2.59; p= 0.0137) and the incidence rate of AMI 39% higher (95% CI 1.00-1.92; p=0.0472) among patients with DM in comparison to those without.

Conclusion: Open vascular surgery remains a first-line option for a substantial part of patients with CLTI, especially for limb salvage in patients with DM. The higher incidence rate of stroke and AMI among patients with DM following open vascular surgery for infrainguinal CLTI require specific consideration preoperatively with the aim of optimizing medical treatment to improve cardiovascular outcome postoperatively.

Supported by: Region Skåne, Hulda Almroth foundation, ALF

Disclosure: E. Lilja: None.

OP 34 SGLT2 inhibition: putative mechanisms of benefit

199

The SGLT2 inhibitor ertugliflozin causes a switch of cardiac substrate utilisation leading to reduced cardiac mTOR-signalling, unfolded protein response and apoptosis

P.A. Mann1, J. Möllmann1, B.M. Klinkhammer2, P. Droste2, B. Peter2, N. Marx1, M. Lehrke1;

1Internal Medicine I, UK Aachen, Aachen, 2Pathology, Aachen, Germany.

Background and aims: SGLT2 inhibitors reduce hospitalization for heart failure in patients with and without diabetes. The underlying mechanisms remain incompletely understood but might relate to the induction of a fasting like response with low blood glucose and insulin levels and increased ketone bodies. The study aimed to investigate underlying signaling pathways.

Materials and methods: Cardiac hypertrophy was induced by transverse aortic constriction (TAC) surgery in 20-week-old C57Bl/6J mice. Mice were treated with the SGLT2 inhibitor ertugliflozin (225 mg/kg chow diet) or vehicle for a period of 10 weeks.

Results: Ertugliflozin significantly improved left ventricular systolic and diastolic function (dp/dtmax p<0.001 and dp/dtmin p<0.01; by millar catheter with dobutamine stress) and reduced myocardial fibrosis (p=0.17) and hypertrophy (p=0.09). This was paralleled by the expected fasting like response with lower glucose and insulin levels (HOMA-IR p<0.05) and increased ketone body concentrations (p<0.05). As a consequence cardiac insulin signaling (AKT-phosphorylation Thr(308) was reduced (p<0.01) by ertugliflozin with less insulin-dependent glucose transporter GLUT4 expression (p<0.05) while fatty acid transporter CD36 (p<0.001) and the ketone body catabolizing key enzyme beta-hydroxybutyrate dehydrogenase BDH-1 were increased (p<0.01) in addition to AMPK-signaling (p<0.01). This led to downstream inhibition of the mTOR pathway with reduced p70S6K (p<0.05)-, 4E-BP1 (p<0.05)- and ULK1 (p<0.01)-phosphorylation. MTOR signaling critically mediates cardiac hypertrophy, endoplasmic reticulum stress, unfolded protein response (UPR) and adverse cardiac remodeling. Consistently, we found ertugliflozin to reduce ATF6 (p<0.05) and elf2α phosphorylation (p=0.0611) as well as downstream signaling (ATF4 p<0.01; CHOP p<0.001). This let to reduced caspase 3 (p<0.05), collagen I (p<0.01) and IL-1b expression indicating less apoptosis, fibrosis and left ventricular remodeling with consequential reduction of BNP expression (p<0.001) in response to SGLT2 inhibition.

Conclusion: The SGLT2 inhibitor ertugliflozin improves left ventricular function in a murine model of cardiac hypertrophy. Mechanistically, this was associated with a metabolic switch of cardiac substrate utilization with reduced cardiac insulin- and increased cardiac AMPK-signaling leading to reduced cardiac mTOR-signaling, unfolded protein response and apoptosis.

figurebm

Supported by: ML received grants and personal fees from MSD

Disclosure: P.A. Mann: Grants; Michael Lehrke from MSD. Lecture/other fees; Michael Lehrke received fees from MSD.

200

Empagliflozin induced white adipocyte browning and modulated mitochondrial dynamics in KK Cg-Ay/J mice and mouse adipocytes

L. Chen1, L. Xu1,2, X. Liu1, T. Li1, X. Li1, M. Xue1, B. Sun1;

1Laboratory of Hormones and Development, Tianjin Key Laboratory of Metabolic Diseases, Chu Hsien-I Memorial Hospital & Tianjin Institute of Endocrinology, Tianjin Medical University, Tianjin, 2Department of Endocrinology, the First Hospital of Shanxi Medical University, Shanxi Medical University, Taiyuan, Taiyuan, China.

Background and aims: White adipose tissue (WAT) browning is a promising target to prevent and/or treat obesity. Empagliflozin has emerged as an agent with weight-loss potential in clinical and in vivo studies, but the mechanisms underlying its effect are not fully understood. Here, we investigated whether empagliflozin could induce WAT browning and mitochondrial alterations in KK Cg-Ay/J(KKAy) mice and explored the relevant mechanisms involved on its effects.

Materials and methods: Eight-week-old male KKAy mice were administered empagliflozin or saline for 8 weeks and compared with control C57BL/6J mice. Mature 3T3-L1 adipocytes were treated in the presence or absence of empagliflozin. Mitochondrial biosynthesis, dynamics, and function were evaluated by gene expression analyses, fluorescence microscopy, and enzymatic assays. The roles of adenosine monophosphate-activated protein kinase (AMPK) and peroxisome proliferator-activated receptor-γ coactivator-1-alpha (PGC-1α) were determined through AICAR (5-Aminoimidazole-4-carboxamide1-β-D-ribofuranoside)/Compound C and RNA interference, respectively.

Results: Empagliflozin substantially reduced the bodyweight of KKAy mice. Mice treated with empagliflozin exhibited elevated cold-induced thermogenesis and higher expression of uncoupling protein 1 (UCP1) and other brown adipose tissue signature proteins in epididymal and perirenal WAT, which was indicative of browning in these WAT depots. At the same time, empagliflozin enhanced fusion protein mitofusin 2 (MFN2) expression, while decreasing the fission markers phosphorylated dynamin-related protein 1(Ser616) (pDRP1(Ser616)) in epididymal WAT and perirenal WAT. Empagliflozin also increased mitochondrial biogenesis and fusion, improved mitochondrial integrity and function, and promoted browning in 3T3-L1 adipocytes. Furthermore, we found that AMPK signaling activity was indispensable for empagliflozin-induced browning and mitochondrial biogenesis and that PGC-1α was required for empagliflozin-induced fusion.

Conclusion: Our results suggest that empagliflozin is a promising anti-obesity treatment, whose activity made immediate through WAT browning, mitochondrial biogenesis, and regulated mitochondrial dynamics.

Supported by: This work was supported by the National Natural Science Foundation of China (no. 81970697, 81470187) and a grant from the Natural Science Foundation of Tianjin (no. 18JCYBJC26100 and 18JCZDJC35500).

Disclosure: L. Chen: None.

201

SGLT2 inhibition improves beta cell function and glucose tolerance, but does not affect glucose or FFA uptake in skeletal muscle

J.H. Voigt1, K.M. Lauritsen2, S.B. Pedersen2, T.K. Hansen1, N. Møller1, N. Jessen1, L.C. Gormsen3, M.C. Laurenti4, C.D. Man5, A. Vella4, E. Søndergaard1;

1Steno Diabetes Center Aarhus, Aarhus, Denmark, 2Department of Endocrinology and Internal Medicine, Aarhus University Hospital, Steno Diabetes Center Aarhus, Aarhus, Denmark, 3Department of Nuclear Medicine and PET Centre, Aarhus University Hospital, Aarhus, Denmark, 4Endocrine Research Unit, Department of Endocrinology, Diabetes and Nutrition, Mayo Clinic Rochester, Rochester, USA, 5Department of Information Engineering, University of Padua, Padua, Italy.

Background and aims: SGLT2 inhibition induces an insulin-independent lowering of plasma glucose. This leads to reduced insulin levels and an increase in circulating free fatty acids (FFA). Isolated elevation of FFA reduces insulin sensitivity in skeletal muscle, whereas SGLT2 inhibition improves insulin sensitivity, presumably by a reduction in glucotoxicity. This apparent paradox could be explained by a reduction in glucotoxicity, but other factors relating to skeletal muscle substrate utilization may contribute to improvement in insulin sensitivity. Therefore, we aimed to investigate the effects of SGLT2 inhibition on glucose metabolism and skeletal muscle substrate metabolism. We measured the effects of SGLT2 inhibition on beta cell function and glucose tolerance together with substrate utilization and intracellular signalling in skeletal muscle.

Materials and methods: 13 metformin-treated individuals with type 2 diabetes were randomized to once-daily empagliflozin 25 mg or placebo for four weeks in a crossover design with a one-week wash-out between the two treatment periods. At the end of each treatment period participants were studied after an overnight fast.Insulin secretion and action were measured using the oral minimal model. Skeletal muscle glucose and FFA uptake was measured using 18F-FDG PET/CT and 11C-palmitate PET/CT, respectively. Regulation of intracellular signalling pathways involved in lipid and glucose metabolism were measured in biopsies from the vastus lateralis muscle.

Results: Four weeks of empagliflozin treatment increased total beta cell responsivity (ɸ) [20 ± 8 vs. 14 ± 9 10-9min-1 (p < 0.01)]. In addition, empagliflozin improved glucose tolerance [SI: [9 (IQR: 10.7) vs. 3.28 (IQR: 4.82) 10-4 dL/kg/min per μmol/mL (p < 0.01)], glucose effectiveness [GE: 2.6x10-2 ± 2.9x10-3 vs. 2.4 x10-2 ± 2.8x10-3 dL/kg/min, (p=0.02)], and the disposition index (DItotal) [275 (IQR: 187) vs. 65.6 (IQR: 54.4) 10-14 dL/kg/min2 per pmol/L (p < 0.01)]. Empagliflozin did not affect FFA or glucose uptake in skeletal muscle. Protein content of AKT, GLUT4, CD36 and HK-II was unaffected by empagliflozin and no effect was observed on skeletal muscle lipoprotein lipase activity.

Conclusion: 4 weeks of empagliflozin treatment improves beta-cell function. Glucose tolerance is also improved, at least in part by the insulin-independent loss of glucose in the urine. Empagliflozin does not appear to have major effects on glucose and FFA uptake or intracellular pathways involved in substrate uptake and insulin signalling in skeletal muscle in the postabsorptive state.

Clinical Trial Registration Number: EUDRA-CT nr: 2017-001779-22

Disclosure: J.H. Voigt: None.

202

Urinary proteomics and the effect of dapagliflozin treatment in persons with type 2 diabetes and diabetic kidney disease: a randomised crossover trial

T.K.E. Rönkkö1, M.K. Eickhoff1, H. Mischak2, P. Rossing1,3, F. Persson1, T.S. Ahluwalia1,3;

1Complications Research, Steno Diabetes Center Copenhagen, Gentofte, Copenhagen, Denmark, 2Mosaiques diagnostics GmbH, Hannover, Germany, 3University of Copenhagen, Copenhagen, Denmark.

Background and aims: Nearly 40% of persons with type 2 diabetes develop diabetic kidney disease which places a major economic burden on modern health care systems. Sodium-Glucose-Co-Transporter-2 inhibitors (SGLT2i) have emerged as a novel way to treat persons with type 2 diabetes and diabetic kidney disease. However, the mechanisms by which SGLT2i work, remain partly unclear. The current study aims to better understand SGTL2i function by observing the changes induced by SGLT2i treatment in the urinary proteome.

Materials and methods: The study is a double-blinded, randomized, placebo-controlled, crossover trial in which 32 participants with type 2 diabetes and diabetic kidney disease were treated with 10 mg of dapagliflozin for 12 weeks or matching placebo on top of standard treatment. All participants had albuminuria (Urine Albumin-to-Creatinine Ratio ≥ 30 mg/g) and received Renin-angiotensin-aldosterone system blockade treatment. The urinary proteomics data from after placebo and dapagliflozin periods were analyzed using Wilcoxon signed-rank test adjusted for multiple testing using Benjamini-Hochberg method. All results with p < 0.05 were considered statistically significant.

Results: The participants had a mean (SD) age of 63 (8) years, 88% males, diabetes duration 15.9 (4.7) years, BMI 33.7 (5.4) kg/m2, HbA1c 10.8 (3.2)%, median (IQR) Urine Albumin-to-Creatinine Ratio 154 (94-329), eGFR 85.5(19.1) ml/min/m2, respectively. 36 urinary peptide fragments belonging to 18 unique proteins changed significantly due to dapagliflozin, of which 24 decreased and 12 increased multifold. Concentrations of albumin and alpha-1-antitrypsin fragments decreased, while collagen alpha-1 (I) & (II) and collagen alpha-3 (I) fragments increased, as expected and in confirmation with previous findings. Decreases were also seen in the concentrations of alpha-1-glycoprotein, cornulin, and peptidase inhibitor 16 fragments, while increases were seen in keratin (type II) cytoskeletal 1, apolipoprotein C-III, and E3 ubiquitin-protein ligase fragments. All of the changes except the changes in two of the collagen alpha-1(III) chain fragments were observed when the urinary proteome of 50 healthy controls was compared to that of 110 participants with diabetic kidney disease and type 1 diabetes from an independent cohort, further supporting the role of these peptides in the development of diabetic kidney disease.

Conclusion: We identified and validated differential urinary peptide patterns in response to SGLT2i (dapagliflozin) on individuals with type 2 diabetes and diabetic kidney disease. The findings highlight the potential renoprotective effects of SGLT2i and help elucidate the response pathways involved in SGLT2i treatment.

Clinical Trial Registration Number: NCT02914691

Supported by: AZ supports the study by a research grant

Disclosure: T.K.E. Rönkkö: Grants; Research grant from AstraZeneca.

203

Effect of SGLT2 inhibition on ketone bodies in patients with stable chronic heart failure

R. Pietschner1, J. Kolwelter2, A. Bosch1, D. Kannenkeril1, C. Ott1,3, M. Schiffer1, S. Achenbach2, R.E. Schmieder1;

1Department of Nephrology and Hypertension, University Hospital Erlangen, Friedrich-Alexander-University Erlangen-Nürnberg (FAU), Erlangen, 2Department of Cardiology and Angiology, University Hospital Erlangen, Friedrich-Alexander-University Erlangen-Nürnberg (FAU), Erlangen, 3Department for Nephrology and Hypertension, Paracelsus Medical University, Nürnberg, Germany.

Background and aims: Recent studies indicate that sodium glucose cotransporter (SGLT)-2 inhibition increases levels of ketone bodies in the blood in patients with type 1 and 2 diabetes. Other studies suggest that in patients with chronic heart failure (CHF), increased myocardial oxygen demand can be provided by ketone bodies as a fuel substrate. Experimental studies report that ketone bodies, specifically beta-hydroxybutyrate (β-OHB) may increase blood pressure by impairing endothelium-dependant relaxation, thereby leading to increased vascular stiffness. In our study we assessed whether the SGLT 2 inhibition with empagliflozin increases ketone bodies in patients with stable CHF and whether such an increase alters blood pressure and vascular function.

Materials and methods: In a prospective, double blind, placebo controlled, parallel-group single centre study 71 patients (67.6 ± 8.9 years) with CHF (left ventricular ejection fraction 39.0 ± 8.2 %) were randomized (2:1) to the SGLT-2 inhibitor empagliflozin 10mg orally once daily or to placebo. After a run-in phase off any SGLT-2 inhibitors for at least 10 weeks, we evaluated blood pressure with a 24h ambulatory blood pressure (ABP) monitoring device, vascular stiffness parameters by the SphygmoCor system (AtCor Medical, Sydney, NSW, Australia) and fasting metabolic parameters, including β-OHB by an enzymatic assay (Beckman Coulter DxC 700 AU). The same measurements were repeated 12 weeks after treatment. In 21 of the 74 patients levels of β-OHB were beneath the lower border of our assay (<0.05mmol/l) therefore being excluded from the subsequent analysis.

Results: In patients with stable CHF, treatment with empagliflozin (n=36) was followed by an increase of β-OHB by approximately 33 % (p=0.017), as well as a significant fall in 24 h systolic (p=0.038) and diastolic (p=0.003) ABP, weight reduction (p=0.003) and improvement of central systolic blood pressure (p=0.008) and central pulse pressure (p=0.008). The increase in β-OHB was related to a lower decrease of empagliflozin-induced 24h systolic (r=0.321, p=0.069) and diastolic (r=0.516, p=0.002) ABP and less improvement of central systolic blood pressure (r=0.470, p=0.009) and central pulse pressure (r=0.391, p=0.033). No significant changes were seen in any of these parameters in the placebo group (n=17).

Conclusion: In patients with stable CHF ketone bodies as assessed by β-OHB increased after treatment with empagliflozin. This increase led to an attenuation of the beneficial effects of empagliflozin on blood pressure and vascular parameters.

figurebn

Clinical Trial Registration Number: NCT03128528

Supported by: Boehringer Ingelheim

Disclosure: R. Pietschner: Grants; supported by Boehringer Ingelheim.

204

Effects of SGLT2 inhibition on lipid storage and lipolysis in adipose tissue in type 2 diabetes

K.M. Lauritsen1, J.H. Voigt1, S.B. Pedersen1, T.K. Hansen2, N. Møller1, N. Jessen2, L.C. Gormsen3, E. Søndergaard1;

1Department of Endocrinology and Internal medicine, Aarhus University, Aarhus N, 2Steno Diabetes Center Aarhus, Aarhus University, Aarhus N, 3Department of Nuclear Medicine and PET, Aarhus University, Aarhus N, Denmark.

Background and aims: SGLT2 inhibition induces an insulin-independent reduction in plasma glucose causing increased lipolysis and subsequent lipid oxidation by energy-consuming tissues. However, it is unknown whether SGLT2 inhibition also affects lipid storage in adipose tissue.The objective of this study was to elucidate the effects of SGLT2 inhibition on the balance between storage and release of fatty acids in adipose tissue.

Materials and methods: 13 individuals aged 50-70 years with type 2 diabetes treated with metformin were investigated in a randomized, double-blind, placebo-controlled crossover trial after a four-week intervention period with once daily empagliflozin 25 mg and placebo.The main outcome measures were adipose tissue fatty acid uptake, lipolysis rate and clearance measured by 11C-palmitate PET/CT and adipose tissue glucose uptake measured by 18F-FDG PET/CT. Protein and gene expression of pathways involved in lipid storage and lipolysis were measured in biopsies of abdominal subcutaneous adipose tissue.

Results: Subjects were weight stable during the study (94.6±9.6 vs. 95.2±9.7 kg, p=0.15). SGLT2 inhibition did not affect FFA uptake in abdominal subcutaneous adipose tissue (0.31±0.23 vs. 0.25±0.15 μmol/100g/min, p=0.32), but increased FFA uptake in visceral adipose tissue by 27% (p<0.05). Both the lipolysis and FFA clearance rate remained unaffected by SGLT2 inhibition. Interestingly, SGLT2 inhibition reduced GLUT4 protein (9.51x106±4.58x106 vs. 12.8x106±6.46x106 arbitrary units (AU), p=0.03) and mRNA content (0.49 (95% CI 0.36-0.62) vs. 0.66 (95% CI 0.55-0.79) AU, p=0.01) in abdominal subcutaneous adipose tissue (figure). However, adipose tissue glucose uptake was not different in subcutaneous (0.35±0.36 vs. 0.77±0.88 μmol/g/min, p=0.16) or visceral adipose tissue (1.08±1.03 vs. 0.99±0.87 μmol/g/min, p=0.79) after SGLT2 inhibition compared to placebo.

Conclusion: SGLT2 inhibition reduces GLUT4 gene and protein expression in abdominal subcutaneous adipose tissue, indicating a reduced channeling of substrate for lipid storage.

figurebo

Clinical Trial Registration Number: EudraCT nr.: 2017-001779-22

Supported by: The Novo Nordisk Foundation, the Danish Diabetes Academy supported by the Novo Nordisk Foundation, the Health Research Fund of Central Denmark Region and Steno Diabetes Center Aarhus.

Disclosure: K.M. Lauritsen: None.

OP 35 Omics and more for type 2 diabetes and complications

205

CpG sites associated with insulin resistance and related novel variants suggest a possible mechanism linking insulin resistance and cardiometabolic traits

N. Fragoso-Bargas1,2, S. Lee-Ødegård2, J.O. Opsahl1, L. Sletner2,3, A.K. Jenum1, L.C. Groop4, E. Qvigstad1,2, G.-H. Moen2,5, K.I. Birkeland1,2, R.B. Prasad4, C. Sommer1;

1Oslo University Hospital, Oslo, Norway, 2University of Oslo, Oslo, Norway, 3Akershus University Hospital, Oslo, Norway, 4Lund University Diabetes Centre, Malmö, Sweden, 5The University of Queensland Diamantina Institute, Brisbane, Australia.

Background and aims: Insulin resistance increases during pregnancy and reaches similar levels seen in type 2 diabetes (T2D), providing a valuable T2D model. We aimed to identify methylation patterns in CpG sites related to insulin resistance, explore their associations with genetic variants, and the relationship of these variants with cardiometabolic traits.

Materials and methods: In EPIPREG, 294 European (EUR) and 162 South Asian (SA) pregnant women had both DNA methylation (EPIC beadchip) and HOMA-IR data at gestational week 28±2. We performed an EWAS of HOMA-IR using linear mixed models, adjusted for age, smoking, cell composition, and using ethnicity as a random effect, accepting a 5% false discovery rate (FDR). We performed methylation quantitative trait loci (mQTL) analysis in cis (1Mb) and trans using imputed genotypes (EUR=187 and SA=135). The mQTLs were performed separately in EUR and SA for each of the six CpG sites. We used linear models adjusted for age, smoking and cell composition as covariates, p<5e-08 was considered significant. Next, we selected specific mQTLs to test for association with HOMA-IR and other cardiovascular related phenotypes using simple linear models. The selection was done as follows: For the CpG sites that showed mQTLs in both ethnic groups, the common SNP with the smallest p-value in both ethnicities was chosen, if there were no common mQTLs between CpG sites across ethnicities; we selected the most significant SNP from each ethnic group.

Results: Six CpG sites were inversely associated with HOMA-IR (Table 1). The results persisted after adjustment for BMI. We observed mQTLs for 5/6 CpG sites in EUR and 3/6 in SA (Table1). In EUR, rs16893889 was positively associated with HOMAI-IR (p=0.039), Gestational diabetes mellitus (GDM) (p=0.048), and both systolic (p=0.001) and diastolic blood pressure (p=0.001). Both rs34964576 and rs1108902 were negatively associated with HOMA-IR (p=0.039 and p=0.039, respectively) and Fasting glucose (p=0.020 and p=0.043, respectively). rs7006759 showed positive associations with HOMA-IR (p=8.46e-06), fasting glucose (p=0.015), GDM (p=0.012), BMI (p=0.002), systolic (p=0.004) and diastolic (p=0.002) blood pressure. In SA, rs16893889 was positively associated with HOMA-IR (p=0.042), rs1108902 with diastolic blood pressure (p=0.030), and rs6641912 with HOMA-IR (p=0.044) and HbA1c (0.043).

Conclusion: We discovered six CpG sites associated with insulin resistance, independently of BMI. The mQTLs were associated with HOMA-IR and other related variables, suggesting that insulin resistance and related traits may be explained by an interplay between genotype and DNA methylation.

figurebp

Supported by: NDA and HSØ

Disclosure: N. Fragoso-Bargas: None.

206

Novel biomarkers for glycaemic deterioration in type 2 diabetes: an IMI RHAPSODY study

R.C. Slieker1, L.A. Donnelly2, L. Lopez-Noriega3, G.N. Giordano4, M. Åkerlund4, IMI-RHAPSODY, I. Pavo5, V. Lyssenko6, C. Legido Quigley7, L. Groop4, M. Ibberson8, J.W.J. Beulens9, L.M. 't Hart1, E.R. Pearson2, G.A. Rutter3;

1LUMC, Leiden, Netherlands, 2University of Dundee, Dundee, UK, 3Imperial College London, London, UK, 4Lund University, Lund, Sweden, 5Eli Lilly Regional Operations GmbH, Vienna, Austria, 6University of Bergen, Bergen, Norway, 7Steno Diabetes Center Copenhagen, Gentofte, Denmark, 8SIB Swiss Institute of Bioinformatics, Lausanne, Switzerland, 9Amsterdam UMC, Amsterdam, Netherlands.

Background and aims: Type 2 diabetes is a progressive multifactorial disease which presently affects >400m worldwide, with numbers expected to increase to >700 m by 2045. Biomarkers for the disease, which provide a deeper understanding of the disease process, are therefore eagerly sought.

Materials and methods: We have deployed a multi-omics approach in large cohorts of patients with existing type 2 diabetes to identify biomarkers for disease progression across three molecular classes, metabolites, lipids and proteins.

Results: A Cox regression analysis for association with time to insulin requirement in 2,973 patients in the DCS, ANDIS and GoDARTS cohorts identified homocitrulline, isoleucine and 2-aminoadipic acid, as well as the bile acids glycocholic and taurocholic acids, as predictive of more rapid deterioration. Increased levels of eight triacylglycerol species, and lowered levels of the sphingomyelin SM 42:2;2 were also predictive of disease progression. Of ~1,300 proteins examined in two cohorts, levels of GDF-15/MIC1, IL-18RA, CRELD1, NogoR, FAS, and ENPP7 were associated with faster progression, whilst SMAC/DIABLO, COTL1, SPOCK1 and HEMK2 predicted lower progression rates. Strikingly, identified proteins and lipids were also associated with diabetes incidence and prevalence in external replication cohorts. Implicating roles in disease compensation, NogoR/RTN4R improved glucose tolerance in high fat-fed mice and tended to improved insulin signalling in liver cells whilst IL-18R antagonised inflammatory IL-18 signalling towards nuclear factor kappa-B in vitro. Conversely, high NogoR levels led to islet cell apoptosis.

Conclusion: This comprehensive, multi-disciplinary approach thus identifies novel biomarkers with potential prognostic utility, provides evidence for new disease mechanisms, and identifies potential therapeutic avenues to slow diabetes progression.

Supported by: IMI-RHAPSODY

Disclosure: R.C. Slieker: None.

207

Mirror effects of rare OPRD1 variants on the aetiology of type 2 diabetes and obesity

S. Meulebrouck1, G. Quéniat1, M. Baron1, M. Canouil1, M. Derhourhi1, B. Balkau2, G. Charpentier3, S. Franc3, M. Marre4, R. Roussel4, R. Scharfmann5, P. Froguel1, A. Bonnefond1;

1CNRS UMR8199, EGID, Lille, 2Inserm U1018, CESP, Villejuif, 3CERITD, Evry, 4Centre de Recherche des Cordeliers, Paris, 5INSERM U1016, Institut Cochin, Paris, France.

Background and aims: Opioid consumption (i.e. opium or heroin) affects metabolic homeostasis with contradictory effects (i.e. by decreasing adiposity, but increasing hyperglycemia), but the mechanisms linking opioid consumption and metabolic alterations are unknown. In rodents, the δ opioid receptor (DOP), encoded by OPRD1, is mainly expressed in the central nervous system, although in humans, RNA-seq analyses demonstrated that OPRD1, in contrast to other opioid receptor genes, is also expressed in pancreatic islets and β cells. Importantly, DOP as an inhibitory G-protein coupled receptor is a privileged drug target. Here, we aimed to perform large-scale functional genetics to decipher the putative link between OPRD1 mutations and metabolic disorders.

Materials and methods: OPRD1 was sequenced in 6,971 adult participants. To assess the functional activity of each variant, we performed SRE-luciferase assays on HEK293 cells overexpressing each variant, followed by DOP activation by its agonists deltorphin II (DII) or DPDPE during 5 hours. In parallel, we performed western blots and immunofluorescence assays on these cells to assess expression and cellular localization of each mutant. We then performed association studies between each cluster of rare gain- (GoF) or loss-of-function (LoF) variants, and various metabolic traits assessed in our cohort. We also performed association analyses between the frequent GoF variant encoding p.I52V and metabolic traits in 34,812 individuals from the AMP T2D Knowledge Portal. Finally, in human pancreatic EndoCβH1 cells, we performed a glucose-stimulated insulin secretion assay coupled to DOP activation by DII and OPRD1 overexpression.

Results: We identified 31 rare variants (minor allele frequency [MAF] < 1%) and 3 frequent variants (MAF ≥ 1%). Through luciferase assays, we identified 7 GoF variants increasing DOP activity in response to both agonists, including the frequent p.I52V variant, and 12 rare LoF variants decreasing it. Immunofluorescence assays showed that all the mutants were effectively expressed and localized at the plasma membrane, except for two LoF mutants (p.P14R and p.G36E). Western blots showed that these mutants tended to have a lower expression than wild-type DOP, while the other DOP mutants were expressed at the same magnitude as wild-type DOP. Association analyses highlighted that LoF variants were associated with increased overweight/obesity risk (P=0.0054; OR=11) but decreased hyperglycemia risk (P=0.054; OR=0.23), whereas GoF variants were associated with improved lipid metabolism. Association analyses performed in the AMP T2D Knowledge Portal confirmed this mirror effect, as the p.I52V variant that is only frequent in Africans, associates with increased type 2 diabetes risk (P=3.6×10-6; OR=2), but also with lower body mass index (P=0.0038; β=-0.37±0.18), and improved lipid metabolism. Finally, in EndoCβH1 cells overexpressing OPRD1 (as wild-type cells poorly expressed OPRD1), DOP agonist DII significantly inhibits glucose-stimulated insulin secretion.

Conclusion: This study highlights DOP as a major metabolic link between opioids and type 2 diabetes. DOP agonists and/or antagonists should be considered as new tools to improve metabolic homeostasis.

Disclosure: S. Meulebrouck: None.

208

Multi-phenotype association analysis reveals shared biological pathways between type 2 diabetes and depression

J.G. Maina1, Z. Balkhiyarova2, M. Kaakinen2, A. Nouwen3, I. Prokopenko2;

1INSERM UMR 1283, CNRS UMR 8199, University of Lille, Lille, France, 2Department of Clinical and Experimental Medicine, University of Surrey, Guildford, UK, 3Department of Psychology, Middlesex University, London, UK.

Background and aims: Type 2 diabetes (T2D) and depression are two common multi-factorial diseases with a growing global disease burden. Epidemiological evidence suggests a bi-directional relationship between the two. Furthermore, genome-wide association analysis (GWAS) results of the two diseases show a positive genetic correlation between T2D and depression. Various hypotheses suggest common pathways involved in the shared predisposition to T2D and depression, however, we are currently unable to adequately explain their shared pathophysiology.

Materials and methods: Using the UK Biobank, we employed the MTAG software to perform a multi-phenotype GWAS (MP-GWAS) of T2D and depressive phenotypes to increase the power for loci identification, improve effect size estimates and provide suggestions for multi-phenotype effects, such as pleiotropy. We compared two diagnostic criteria for depression: clinically diagnosed major depressive disorder (MDD) (n = 91,796; 5,357 cases, 86,439 controls) and depressive symptoms based on self-report assessed using the PHQ-9 questionnaire (n = 153,567) and evaluated how each performs in a MP-GWAS with T2D (n = 482,985; 19,344 cases, 463,641 controls). We applied on the results functional transcriptome-wide analysis (TWAS) based on various tissue prediction models in the GTEx database.

Results: In single-phenotype GWAS of T2D, depressive symptoms and MDD in the UK Biobank we identified 92, three and zero genome-wide significant loci (P-value=5x10-8) respectively. In contrast, MP-GWAS increased the number of significantly associated loci for depressive symptoms from three to eight, while none were identified for MDD in MP-GWAS. Seven of the eight loci identified for depressive symptoms after MP-GWAS were established T2D loci including TCF7L2, CDKAL1 and IGF2BP2. Moreover, the risk alleles at these shared loci were the same for T2D and depressive symptoms. TWAS results revealed enrichment of significant gene associations in the same tissues for depressive symptoms and T2D including the frontal cortex (CDKAL1), whole blood (HLA-DRB1), adipose subcutaneous tissue (IRS1), and muscle (HLA-DRA). These genes are involved in immune system and glucose-insulin signalling pathways, important for the shared pathogenesis of T2D and depression.

Conclusion: In this first large-scale multi-phenotype of T2D and depressive phenotypes, we show that depressive symptoms based on self-report perform better than MDD in identification of potential multi-phenotype effects with T2D. The shared loci between depressive symptoms and T2D support a role of insulin function and signalling and, immune system pathways in the pathophysiology of T2D and depression boosting our understanding of their shared pathogenesis.

Supported by: WCRF 2017/1641, LONGITOOLS H2020-SC1-2019-874739, PreciDIAB ANR-18-IBHU-0001

Disclosure: J.G. Maina: None.

209

Serum magnesium is inversely associated with heart failure, atrial fibrillation and microvascular complications in type 2 diabetes

L.J. Oost1, A.A.W. van der Heijden2, E.A. Vermeulen3, C. Bos1, P.J.M. Elders2, R.C. Slieker4, S. Kurstjens1, M. van Berkel5, J.G.J. Hoenderop1, C.J. Tack6, J.W.J. Beulens4, J.H.F. de Baaij1;

1Department of Molecular Physiology, Radboud University Medical Center, Nijmegen, 2Department of General Practice and Elderly Care Medicine, Amsterdam UMC, Amsterdam, 3Department of Nephrology, Amsterdam UMC, Nijmegen, 4Department of Epidemiology & Data Science, Amsterdam UMC, Amsterdam, 5Department of Laboratory Medicine, Radboud University Medical Center, Nijmegen, 6Department of Internal Medicine, Radboud University Medical Center, Nijmegen, Netherlands.

Background and aims: In this study, we investigated whether serum magnesium (Mg2+) was prospectively associated with macro- or microvascular complications in people with T2D, and whether this association was mediated by glycemic control (HbA1c).

Materials and methods: This study was performed in 4,348 participants of the Diabetes Care System cohort, a prospective cohort of people with T2D. We analyzed the association of serum Mg2+ per 0.1 mmol/L increment with macrovascular disease and mortality (acute myocardial infarction (AMI) (n=91), coronary heart disease (CHD) (n=169), heart failure (HF) (n=155), cerebrovascular accident (CVA) (n=118), peripheral arterial disease (PAD) (n=113)), atrial fibrillation (AF) (n=157), and microvascular complications (n=1,285) (chronic kidney disease (CKD) (n=1,093), diabetic retinopathy (n=114) and diabetic foot complications (n=1,147) using Cox regression, adjusted for confounders. Mediation analysis was carried out to assess whether glycemic control mediated these associations.

Results: The average baseline serum Mg2+ was 0.80 ± 0.08 mmol/L, 9.1% of the participants had hypomagnesemia (<0.7 mmol/L Mg2+). Adjusted for cardiovascular confounders, serum Mg2+ was during 6.2 years of follow-up prospectively, inversely, associated with HF 0.76 (95% CI: 0.62; 0.93) and AF 0.59 (95% CI: 0.49; 0.72). The association of serum Mg2+ with AF was independent of fatal and non-fatal HF. The associations with AMI, CHD, CVA and PAD were not significant. Serum Mg2+ was during 5.1 years of follow-up prospectively, inversely, associated with overall microvascular events with a hazard ratio of 0.85 (95% CI: 0.78; 0.91), 0.89 (95% CI: 0.82; 0.96) for CKD, 0.77 (95% CI: 0.61; 0.98) for diabetic retinopathy and 0.85 (95% CI: 0.78; 0.92) for diabetic foot. Glycemic control partly mediated the associations of serum Mg2+ with overall microvascular events, diabetic retinopathy and diabetic foot, but not with macrovascular events.

Conclusion: Serum Mg2+ is prospectively, inversely, associated with the risk to develop HF, AF, and with the occurrence of microvascular endpoints, CKD, diabetic retinopathy and foot complications, in T2D. HbA1c partly mediated the association of serum Mg2+ with microvascular complications.

Supported by: This research was funded by grants from the Netherlands Organization for Scientific Research (NWO Veni 016.186.012), the Dutch Diabetes Research Foundation (2017-81-014), and the NIGRAM2+ consortium. The NIGRAM2+ collaboration project is co-funded by the PPP Allowance made available by Health~Holland, Top Sector Life Sciences & Health, to stimulate public-private partnerships (LSHM17034) and the Dutch Kidney Foundation (16TKI02). J.W.J. Beulens is supported by a ZonMW NWO-Vidi grant (91 71 8304).

Disclosure: L.J. Oost: None.

210

Genome-wide meta-analysis and omics integration identifies novel genes for diabetic kidney disease

N. Sandholm1,2, J.B. Cole3,4, V. Nair5, X. Sheng6, H. Liu6, N. vanZuydam7, E. Ahlqvist8, D. Fermin5, M. Kretzler5, K. Susztak6, J.N. Hirschhorn3,4, J.C. Florez3,9, P.-H. Groop1,2, GENIE consortium;

1Folkhalsan Research Center, Helsinki, Finland, 2Department of Nephrology, University of Helsinki and Helsinki University Hospital, Helsinki, Finland, 3Programs in Metabolism and Medical and Population Genetics, Broad Institute, Cambridge, USA, 4Department of Pediatrics, Boston Children’s Hospital, Boston, USA, 5Department of Internal Medicine, University of Michigan, Ann Arbor, USA, 6Departments of Medicine and Genetics, University of Pennsylvania, Philadelphia, USA, 7Science for Life Laboratory, Uppsala University, Uppsala, Sweden, 8Department of Clinical Sciences, Lund University, Malmö, Sweden, 9Department of Medicine, Harvard Medical School, Boston, USA.

Background and aims: Diabetes is the leading cause of kidney disease, with heritability studies suggesting a substantial genetic contribution to diabetic kidney disease (DKD). Our aim was to identify novel genetic factors and genes contributing to DKD, by performing meta-analysis of previous genome-wide association studies (GWAS) on DKD from the DNCRI SUMMIT consortia, and by integrating the results with various renal transcriptomics datasets.

Materials and methods: We performed GWAS meta-analyses for ten phenotypic definitions of DKD, including up to 26,785 individuals with either type 1 or type 2 diabetes. Meta-analysis results were integrated with estimated quantitative trait locus (eQTL) data from human glomerular (N=119) and tubular (N=121) samples to perform transcriptome-wide association study (TWAS) with MetaXcan. Gene expression was studied in human transcriptomics data from nephrectomy samples (433 tubule and 335 glomerulus samples) and kidney biopsies from Pima Indian cohort (67 glomerular and 48 tubule-interstitial tissues) and association was tested with relevant pathological phenotypes.

Results: A novel low frequency intronic variant (rs72831309) in the TENM2 gene was associated with combined chronic kidney disease (CKD) and DKD phenotype (odds ratio 2.08, 95% confidence interval 1.62 - 2.67, p = 9.82×10-9). Gene-level analysis with MAGMA and PASCAL identified 10 genes associated with DKD phenotypes (PTPRN - RESP18, COL20A1, DCLK1, EIF4E, GPR158, INIP - SNX30, LSM14A, and MFFp < 2.3×10-6). TWAS indicated that tubular AKIRIN2 expression was higher in DKD than in diabetic controls (p = 1.1×10-6).

For many of the identified genes, renal expression was associated with relevant pathological phenotypes: For example, TENM2 expression in tubular nephrectomy samples correlated with higher eGFR (r = 0.287, p = 2.34×10-9) and less fibrosis (r = -0.292, p = 4.68×10-9). Also, tubular DCLK1 expression correlated with fibrosis both in the nephrectomy samples (r = 0.345, p = 1.55×10-12), and in kidney biopsies (r = 0.52, p = 0.0003). Finally, tubular SNX30 expression correlated with higher eGFR (r = 0.341, p = 7.63×10-13) and with lower level of fibrosis (r = -0.534, p < 2×10-16). In line with this, renal SNX30 expression was associated with higher eGFR also in a TWAS in the general population (p=0.046). Furthermore, the top SNPs within the DCLK1, AKIRIN2SNX30 and 3 other gene regions were significant methylation quantitative trait loci (mQTL) in 188 healthy kidney samples (p<2.2×10-11).

Conclusion: GWAS meta-analysis and integration with renal transcriptomics data points to novel genes contributing to pathogenesis of DKD.

Supported by: JDRF

Disclosure: N. Sandholm: None.

OP 36 Optimising insulin therapy

211

Impact of the fasting plasma glucose titration target on the success of basal insulin titration in insulin-naive patients with type 2 diabetes: a systematic analysis

J. Wolters, D. Wollenhaupt, M. Abd El Aziz, M.A. Nauck;

Katholisches Klinikum Bochum gGmbH, St. Josef Hospital, Bochum, Germany.

Background and aims: Basal insulin titration algorithms differ in their fasting plasma glucose (FPG) targets. We compared beneficial (glycaemic control) and adverse outcomes (hypoglycaemia, weight gain) of basal insulin titration performed with different FPG titration targets (TT).

Materials and methods: A PubMed literature search retrieved 42 publications reporting clinical trials on titrating basal insulin in 17643 insulin-naïve patients with type 2 diabetes on a defined background of oral glucose-lowering medications (OGLM) and reporting relevant outcomes. 61 individual study arms were grouped by FPG TT (a: ≤ 5.0; b: 5.01-5.6; c: ≥ 5.61 mmol/l). For all subgroups, weighted means and their standard deviations were calculated for baseline and end-of-treatment FPG, HbA1c, target achievement, incidence of any or severe hypoglycaemic events, and insulin doses, and compared by ANOVA and post hoc comparisons or contingency table analysis.

Results: Achieved FPG and HbA1c after titration (study end) were lower with more ambitious titration targets (a. 6.17 [6.11; 6.24]; b. 6.67 [6.63; 6.71]; c. 6.93 [6.79; 7.06] mmol/l and a. 7.08 [7.05; 7.11]; b. 7.19 [7.18; 7.21]; c. 7.32 [7.26; 7.37] %; p < 0.01 for all comparisons). Accordingly, HbA1c target achievement (≤ 7.0 or 6.5 %) was highest with FPG TT a, intermediate with b, and worst with c (p < 0.01). The basal insulin dose achieved by titration was highest in subgroup a. (p < 0.01). The incidence of (severe or any) hypoglycaemic episodes was not higher with lower FPG TTs, and body weight gain (1 to 2 kg across all subgroups) was not more prominent as well. However, overall, only 29.1 % (95 % CI: 28.5-29.8 %) reached their individual FPG TT.

Conclusion: Aiming for a lower FPG titration target improves glycaemic control without increasing the risk for adverse events like hypoglycaemia or increased body weight. More stringent titration should take more patients to ambitious targets.

Disclosure: J. Wolters: None.

212

Advancing therapy in basal insulin users with type 2 diabetes: better clinical outcomes with iGlarLixi vs premix BIAsp 30 in the SoliMix trial

C. Trescoli1, J. Rosenstock2, R. Emral3, L. Sauque-Reyna4, V. Mohan5, S. Al Sifri6, N. Lalic7, A. Alvarez8, P. Picard9, N. Demil10, M. Bonnemaire11, R.J. McCrimmon12;

1Hospital Universitario de La Ribera, Alzira, Spain, 2Dallas Diabetes Research Center at Medical City, Dallas, USA, 3Ankara University Faculty of Medicine, Ankara, Turkey, 4Instituto de Diabetes Obesidad y Nutrición S.C., Cuernavaca, Morelos, Mexico, 5IDF Centre of Excellence in Diabetes Care & ICMR Centre for Advanced Research on Diabetes, Chennai, India, 6Al Hada Military Hospital, Taif, Saudi Arabia, 7Clinical Center of Serbia, Belgrade, Serbia, 8Sanofi, Buenos Aires, Argentina, 9Ividata, Levallois-Perret, France, 10Sanofi, Chilly-Mazarin, France, 11Sanofi, Paris, France, 12University of Dundee, Dundee, UK.

Background and aims: iGlarLixi, a fixed-ratio combination of insulin glargine 100 U/mL and lixisenatide, offers an alternative treatment advancement option to premix insulin for uncontrolled basal insulin-treated T2D. SoliMix, an open-label, multicentre study, was the first to compare these two insulin coformulations.

Materials and methods: SoliMix randomised adults with T2D and HbA1c 58-86 mmol/mol (7.5-10.0 %) on basal insulin + oral antihyperglycaemic drugs (OADs; metformin ± sodium-glucose co-transporter-2 inhibitors) to once-daily iGlarLixi or twice-daily premix biphasic insulin aspart 30 (BIAsp 30; 30% insulin aspart + 70% insulin aspart protamine). The two primary efficacy endpoints were non-inferiority in HbA1c reduction or superiority in bodyweight change from baseline to Week 26 of iGlarLixi vs BIAsp 30.

Results: Baseline characteristics were similar between those randomised to iGlarLixi vs BIAsp 30 (49% vs 51% female; mean BMI 29.7 vs 30.0 kg/m2; mean HbA1c 71 vs 70 mmol/mol [8.6 vs 8.6 %]). In both groups, mean age was 59.8 years and mean duration of diabetes was 13.0 years. Both primary efficacy endpoints were met (Table), as well as iGlarLixi superiority vs BIAsp 30 in HbA1c reduction. Mean ± SD HbA1c reduced from 71 ± 7 mmol/mol (8.6 ± 0.7 %) for iGlarLixi and 70 ± 7 mmol/mol (8.6 ± 0.7 %) for BIAsp 30 at baseline to 56 ± 12 mmol/mol (7.3 ± 1.1 %) and 58 ± 11 mmol/mol (7.5 ± 1.0 %) at Week 26. Significantly more participants reached HbA1c target (<53 mmol/mol [<7 %]) without weight gain, and HbA1c target without weight gain and without hypoglycaemia, in the iGlarLixi group than in the BIAsp 30 arm. Incidence (Table) and rates (data not shown) of hypoglycaemia (ADA Level 1 or 2) were lower with iGlarLixi vs BIAsp 30. Gastrointestinal adverse events were more common with iGlarLixi (10.4%) vs BIAsp 30 (2.3%).

Conclusion: Based on better glucose control with weight benefit and less hypoglycaemia, once-daily iGlarLixi is a favourable alternative to twice-daily premix BIAsp 30 for advancing therapy in people with T2D uncontrolled on basal insulin + OADs.

figurebq

Clinical Trial Registration Number: 2017-003370-13

Supported by: Sanofi

Disclosure: C. Trescoli: None.

213

Real-world persistence, adherence, healthcare resource utilisation, and costs in people with type 2 diabetes switching from basal insulin (BI) to 2 nd - vs 1 st -generation BI

E. Wright1, J. Gill2, S. Huse3, X. Li2, T. Reid4, F. Zhou2;

1Charlotte Area Health Education Center, Charlotte, 2Sanofi US, Bridgewater, 3Evidera, Lexington, 4Diabetes Center at Mercyhealth System, Janesville, USA.

Background and aims: Basal insulin (BI) is a mainstay of treatment for type 2 diabetes (T2D). However, many people need to change their BI for a variety of clinical, personal, or economic reasons. This retrospective, real-world observational study using the Optum Clinformatics claims database compared outcomes in people switching from prior BI to either 2nd-generation (2nd-gen) BI (insulin glargine 300 U/mL [Gla-300]) or 1st-generation (1st-gen) BI (insulin glargine 100 U/mL [Gla-100] or insulin detemir [IDet]).

Materials and methods: Data were included from adults (≥18 years) with T2D who had received prior BI (NPH, Gla-100, IDet) in the 6-month baseline period, and switched to either the 2nd-gen BI, Gla-300 or a 1st-gen BI (Gla-100 or IDet) (treatment switch=index date) between April 1, 2015 and August 31, 2019. Participants were followed from index date for 12 months or plan disenrollment or death. Cohorts were propensity score matched (PSM) on baseline demographic/clinical characteristics. Outcomes were persistence (days on treatment without discontinuation), adherence (proportion days covered), and all-cause, diabetes-related, and hypoglycaemia-related healthcare resource utilisation and costs.

Results: After PSM (n=3077/cohort; mean age 68 years, 52% female), cohorts were well balanced except for 1 variable (hospitalisation) with standardised difference >0.1, which was adjusted in models as a covariate. During the 12-month follow-up period, participants who received Gla-300 vs 1st-gen BI had greater persistence with (adjusted p=0.0001) and adherence to (adjusted p=0.0006) therapy (Table). All-cause hospitalisations and emergency room (ER) visits were significantly lower for Gla-300 vs 1st-gen BI (both p<0.0001; Table), as were diabetes-related hospitalisations (21.5 vs 29.1 per 100 patient-years [P100PY]) and ER visits (54.8 vs 74.2 P100PY]), and hypoglycaemia-related ER visits (2.9 vs 5.7 P100PY); all p<0.0001. Costs for all-cause hospitalisations, ER visits and total healthcare were numerically lower for Gla-300 vs 1st-gen BI (Table), as were diabetes-related hospitalization ($5,626 vs $6,210), ER ($1,087 vs $1,543), and total healthcare costs ($22,613 vs $25,165). Similarly, hypoglycaemia-related hospitalization ($199 vs $333), ER ($46 vs $92) and total healthcare costs ($713 vs $1,326) were also numerically lower for Gla-300 vs 1st-gen BI, while pharmacy costs were numerically higher (all-cause: Table; diabetes-related: $7,093 vs $5,178).

Conclusion: Switching to Gla-300 was associated with significantly better persistence, adherence, and—despite numerically higher pharmacy costs—lower all-cause healthcare resource utilisation versus switching to a 1st-gen BI, in people with T2D previously treated with BI therapy. This is the first study to assess these outcomes in tandem.

figurebr

Supported by: Sanofi US

Disclosure: E. Wright: Employment/Consultancy; Abbott Diabetes Care, Bayer, Boehringer Ingelheim, Lilly, Sanofi. Honorarium; Abbott Diabetes Care, Avion, Bayer, Boehringer Ingelheim, Lilly, Sanofi. Lecture/other fees; Abbott Diabetes Care, Avion, Bayer, Boehringer Ingelheim, Lilly, Sanofi. Stock/Shareholding; Abbott. Other; Investigator: Abbott Diabetes Care.

214

Glycaemic improvement in 3,436 people with type 1 diabetes using the Omnipod DASH® Insulin Management System over first 90 days of use

G. Aleppo1, D. J. DeSalvo2, F. Lauand3, L. M. Huyett4, A. Chang4, T. Vienneau4, T. T. Ly4;

1Division of Endocrinology, Metabolism and Molecular Medicine, Northwestern University, Chicago, USA, 2Baylor College of Medicine, Houston, USA, 3Insulet, Paris, France, 4Insulet, Acton, USA.

Background and aims: Outcome data describing real-world use of various devices by people with type 1 diabetes (T1D) are important to support clinical decision-making. This retrospective study characterized patient-reported clinical outcomes of people with T1D in the United States before (baseline) and 90 days after (follow-up) initiation of the tubeless Omnipod DASH® Insulin Management System from October 2018 until February 2021.

Materials and methods: This was a retrospective analysis of an existing dataset of patient information regularly collected by Clinical Service Managers (CSMs) across the US as part of standard Omnipod DASH initiation procedures. CSMs include certified diabetes educators, registered dieticians, and registered nurses who work with patients, caregivers, and HCPs to support users of the Omnipod DASH therapy. Clinical data were collected directly from the patient’s medical record or self-reported if medical records were unavailable. The primary outcome was change in HbA1c levels from baseline to follow-up. Secondary outcomes were change in total daily dose (TDD) of insulin and frequency of hypoglycemic events (HE) per week (#/week <70 mg/dL). Outcomes were assessed by prior treatment modality (MDI or CSII), and by age (children: <18y, adults: ≥18y).

Results: The cohort of Omnipod DASH users included 3,436 patients. Children (n=1,020) were aged (mean±SD) 10±4y and 48% female, with diabetes duration 2.4±2.8y. Adults (n=2,416) were aged 44±17y and 60% female, with diabetes duration 17±14y. Both cohorts consisted primarily of prior MDI users (80% of children, 63% of adults), with some prior CSII users (10% of children, 23% of adults). At follow-up, significant and clinically relevant reductions in HbA1c levels were demonstrated for the study population, regardless of age group. The overall change in HbA1c was -0.8±1.9% for children and -0.9±1.6% for adults (both p<0.0001) (Table 1). The change in HbA1c for prior MDI users was ‑0.9±2.0% in children and -1.0±1.7% in adults (both p<0.0001). In prior CSII users, the change in HbA1c was -0.3±1.2% in children (p>0.05) and -0.6±1.1% in adults (p<0.0001) (Table 1). For children, TDD remained roughly the same after initiating the system (change: ‑0.9±11.3U/d, p>0.05), while adults had a significant reduction in the amount of insulin used (change: -12.5±30.2U/d, p<0.0001). A reduction of HE frequency was seen regardless of the age group The self-reported HE frequency decreased significantly by -1.4±2.7 and -1.6±3.2 episodes per week in children and adults, respectively (both p<0.0001).

Conclusion: These results show that in this large cohort of patients with T1D initiating the Omnipod DASH System, there were significant reductions in HbA1c, TDD, and number of HE after 90 days of use across age groups and/or compared to prior treatments.

figurebs

Supported by: Insulet

Disclosure: G. Aleppo: None.

215

Pharmacodynamics, pharmacokinetics, safety, and tolerability of INS068 vs insulin degludec in type 1 diabetes at steady state: a phase I, randomised, double-blind, cross-over-trial

M. Hernandez1, B. Franey1, J. Wang2, Y. Li2, B. Zhang2, M. Hompesch1;

1ProSciento, Chula Vista, USA, 2Jiangsu Hengrui Medicine Co., Ltd, Shanghai, China.

Background and aims: To assess steady state (SS) pharmacokinetics (PK) and pharmacodynamics (PD), safety and tolerability of INS068, a novel long-acting basal insulin analog, vs insulin degludec (IDeg) in patients with type 1 diabetes mellitus (T1DM).

Materials and methods: In this single-center, randomized, double-blind, cross-over study, patients with T1DM were randomized (1:1:1:1:1:1) to a sequence of 2 treatment periods with once-daily subcutaneous INS068 or IDeg for 9 days at one of three doses (0.4, 0.6 or 0.8 U/kg), separated by a 7 to 21-day washout period. Patients at all doses underwent a 42-h euglycemic clamp after the last dose of each treatment period. The primary endpoint was molar dose ratio with INS068 vs IDeg, as assessed by the area under the glucose infusion rate (GIR) time curve during a dosing interval (tau) at SS (AUCGIR,tau,ss).

Results: 98 (57.1% male, age 34.8 ± 11.7 yrs, BMI 25.3 ± 2.9 kg/m2, HbA1c 7.6 ± 1.2% [mean ± SD]) of 99 randomized patients received treatment. The molar dose ratio with INS068 vs IDeg based on AUCGIR,tau,ss was 0.85 (95% CI: 0.75-0.93) using a linear mixed effects model. AUCGIR,tau,SS and GIRmax,SS were similar for INS068 and IDeg (with a time to GIRmax,SS of 12-13 h post-dose) over the dose range of 0.4-0.8 U/kg (Fig 1A). Duration of action of INS068 was in line with IDeg, lasting for > 42 h at all doses (Fig 1B). Fluctuation of GIR at SS were comparable with INS068 vs IDeg across all doses based on peak-to-trough fluctuation (PTFGIR,tau,SS; range of least-squares geometric mean, 0.54-0.86 vs 0.46-0.77) and average fluctuation (AUCFGIR,tau,SS; range of geometric mean, 0.12-0.17 vs 0.10-0.17). The glucose lowering effect of INS068 was even between the 1st and 2nd half of the dosing interval at SS (AUCGIR,12-24h,SS/AUCGIR,tau,SS; range of geometric mean, 0.46-0.50 for INS068 vs 0.49-0.50 for IDeg). PK parameters (AUCtau,SS and Cmax,SS) at SS were proportional to dose within 0.4-0.8 U/kg for INS068 and IDeg. The mean half-life of elimination (t1/2,SS) was estimated as 21 h for INS068 and 26 h for IDeg across all doses. The accumulation ratio for INS068 as compared with IDeg was 1.7 vs 1.6 for AUC and 1.6 vs 1.4 for Cmax. Adverse events (AEs) were mostly mild, with no deaths and only one case of serious AE reported (hypoglycemia [IDeg], reported after patient withdrawal). Treatment-related AEs occurred in 17.2% of patients with INS068 and 13.3% with IDeg, with the most common (incidence ≥2%) being hyperglycemia (6.9% vs 2.2%), injection site pain (3.4% vs 4.4%) and headache (3.4% vs 4.4%). Hypoglycemic episodes were reported in 85.1% of patients with INS068 vs 86.7% with IDeg.

Conclusion: INS068 was well-tolerated and demonstrated generally similar PD and PK profiles to IDeg at SS. The glucose‐lowering effect and exposure of INS068 were dose proportional at doses of 0.4-0.8 U/kg.

figurebt

Disclosure: M. Hernandez: Employment/Consultancy; ProSciento.

216

Glycaemic control with once weekly basal insulin Fc in persons with type 2 diabetes using continuous glucose monitoring in a phase 2 study

C. Kazda1, J. Chien1, Q. Zhang1, E. Chigutsa1, W. Landschulz1, P. Wullenweber1, A. Haupt1, J. Frias2;

1Eli Lilly and Company, Indianapolis, 2National Research Institute, Los Angeles, USA.

Background and aims: Basal insulin Fc (BIF; LY3209590) is a novel, once-weekly, long-acting IgG Fc-fusion protein assessed for the treatment of diabetes mellitus. A 32-week study evaluating the safety and efficacy of BIF vs degludec in persons with type 2 diabetes mellitus previously treated with a basal insulin showed HbA1c non-inferiority of BIF vs degludec with significantly fewer hypoglycaemic events (≤3.9 mmol/L (70 mg/dL)). Here we present continuous glucose monitoring (CGM) data derived by Dexcom G6, allowing a more detailed assessment of glycaemic control of BIF vs degludec.

Materials and methods: The study included 2 dosing algorithms for BIF with different fasting glucose (FG) targets: ≤7.8 mmol/L (140 mg/dL) (BIF-A1) and ≤6.7 mmol/L (120 mg/dL) (BIF-A2). Degludec was titrated to FG ≤5.6 mmol/L (100 mg/dL). Subjects were randomized to 1 of the 3 arms.

Results: Subject (N=399) mean age was 60.2 years and baseline HbA1c was 65.2 mmol/mol (8.1%). For the entire 32 weeks, the percent of 24 hours in range, hyperglycaemia and hypoglycaemia was similar for the 3 arms (Figure). At Week 32, total duration of hypoglycaemia was similar across 7 days post-injection for BIF-A1 and A2, showing that duration of hypoglycaemia is independent of day post-injection.

Conclusion: CGM data confirm that BIF showed similar glycaemic control vs degludec despite higher FG targets and numerically lower time in hypoglycaemia. The flat pharmacokinetic profile enables near peak less insulin concentrations without an increase in hypoglycaemia risk at highest exposure.

figurebu

Clinical Trial Registration Number: NCT03736785

Supported by: Eli Lilly and Company

Disclosure: C. Kazda: Employment/Consultancy; Eli Lilly and Company. Stock/Shareholding; Eli Lilly and Company.

OP 37 Cytokine storm in type 1 diabetes: from signalling to interventions

217

Specific alterations in the STAT1/STAT6 axis may contribute to beta cell loss in type 1 diabetes

K. Afi Leslie, S.J. Richardson, N.G. Morgan, M.A. Russell;

University of Exeter Medical School, Exeter, UK.

Background and aims: Loss of the anti-inflammatory transcription factor, Signal Transducer and Activator of Transcription (STAT)-6, coupled with sustained up-regulation of the pro-inflammatory transcription factor, STAT1, have been reported in the islets of subjects with type 1 diabetes (T1D). These concerted events could play an important role in promoting β-cell loss during the autoimmune attack, by exacerbation of pro-apoptotic signalling. However, the possibility that the pathways regulated by STAT1 and STAT6 may interact coordinately to regulate β-cell viability, has not been investigated. This scenario has been studied in the present work.

Materials and methods: Cultured INS-1E rodent and human EndoC βH1 β-cells were employed, with sections of pancreas from control subjects and those with recent-onset T1D. Knockdown of target proteins was achieved using interference RNA techniques while qRT-PCR and western blotting were employed to monitor gene and protein expression, respectively. Interferon-responsive promoter activity was studied using a luciferase reporter assay in β-cells. Target proteins were detected in human pancreas sections by immunohistochemistry.

Results: Treatment of β-cells with the anti-inflammatory cytokine, interleukin-13 (IL-13) resulted in the activation of STAT6 (via increased phosphorylation) and this was accompanied by a marked loss of STAT1 protein (control: 1AU; STAT1: 0.5, p<0.05). By contrast, targeted knockdown of STAT6 was associated with a 2-fold increase STAT1 expression (p<0.05) and this correlated with a large increase in STAT1 activity as judged by reporter assays (6 fold; p<0.001). To study these relationships further, we monitored the influence of STAT6 on the expression of SIRPα, a protein previously identified as a regulator of β-cell viability and which is induced by STAT6. Immunoprecipitation of SIRPα from β-cells resulted in significant pulldown of STAT1 suggesting a direct interaction between the two. Conversely, when SIRPα was knocked down selectively, the levels of STAT1 increased by 18-fold (p<0.001) implying a second level of regulation, mediated via SIRPα-dependent alterations in STAT1 gene expression. In parallel, we also monitored the expression of a further set of proteins (including SOCS1, PIAS2, SMAD2, SHP1, SHP and RNF2 which act to negatively regulate STAT1 signalling in β-cells and found that exposure to IL-13 resulted in their up-regulation (by up to 7.5-fold; p<0.001). The increased production of a series of these molecules (SOCS1, SHP1, SHP2 and SMAD2) were clearly dependent on STAT6 activation since they were no longer up-regulated following STAT6 knockdown. To verify the importance of these responses in T1D, immunohistochemical analysis was undertaken in pancreas sections from matched control subjects with recent-onset T1D. This confirmed that STAT1 was increased in the residual insulin-containing islets in T1D in parallel with a reduction in STAT6 and negative regulators, including PIAS2 (control β-cells: 64.5.4±1.7AU, T1D: 48.01±1.1; p<0.001) and SHP2 (control β-cells: 148.4±5.1AU, T1D: 51.9±2.3AU; p<0.001).

Conclusion: The loss of STAT6, SIRPα and other negative regulators of STAT1 signalling in the β-cells in T1D may exacerbate pro-inflammatory signalling and promote β-cell loss during autoimmune attack. This suggests that selective targeting of the STAT6 pathway might offer a therapeutic opportunity to minimise pro-apoptotic signalling and consequent β-cell loss during the progression of T1D.

Supported by: EFSD Programme Award EFSD/JDRF/Lilly European Programme in Type 1 Diabetes Research 2020

Disclosure: K. Afi Leslie: None.

218

Type III interferons are expressed in human pancreas at type 1 diabetes onset and induce immunostimulatory and antiviral activities in human beta cells

E.E. Ringqvist1, S. Tuomela1, M. Mazur1, K. Burdsall1, A. Parajuli1, H. Sork1, L. Krogvold2, K. Dahl-Jørgensen2, C. Matthews3, I. Gerling4, M. Flodström-Tullberg1;

1Center for infectious medicine, Karolinska Institutet, Huddinge, Sweden, 2Department of Paediatrics, Oslo University Hospital, Ullevål, Norway, 3Department of Pathology, University of Florida, Gainesville, USA, 4University of Tennessee Health Science Center, Memphis, USA.

Background and aims: Type I and II interferons (IFNs) have been proposed to contribute to beta cell destruction in human type 1 diabetes. Type III IFNs (IFN-λ 1-4) constitute a group of more recently described IFNs that are produced by both immune and parenchymal cells during inflammation and infection. Type III IFNs are known to induce a gene transcriptional signature similar to that of type I IFNs and they have been demonstrated to contribute to antiviral defense. We have previously shown that type III IFNs are expressed by human islets infected with Coxsackievirus B (CVB) in vitro and that type III IFNs in parallel with the induction of an IFN-induced gene transcriptional signature protects human islets from CVB infection. If type III IFNs play a role in type 1 diabetes has previously not been investigated. Moreover, the direct effect of type III IFNs on beta cells remains unexplored. In the present study, we aimed to investigate if IFNλ are expressed in the human pancreas at diabetes onset and to describe the effects that IFNλ have on beta cells by in-depth proteome, islet transcriptome and immune marker analysis. We also investigated the function of IFNλ induced pathways in β-cells with regards to immune status and antiviral defence.

Materials and methods: EndoC-βH1 cells were exposed to IFNs, poly I:C or infected with CVB serotypes. Proteome analysis was performed with HiRIEF LC-MS, and selected genes were quantified by qPCR. IFNλ expression, viral infection, immune activation markers and HLA-ABC expression was analyzed by FACS. Laser-captured islets of biopsies from 5 newly diagnosed T1D patients and 18 control pancreas donors were analyzed by Affymetrix Gene Arrays

Results: The proteome of IFN- λ1 or IFN- λ2 exposed EndoC-βH1 cells revealed 93 and 119 differentially expressed proteins compared to mock exposed cells. The abundance of a large set of proteins corresponding to known interferon stimulated genes (ISGs) were similarly increased by the two cytokines. The cell surface expression of MHC class I on EndoC-βH1 cells was induced by IFN-λ1 and IFN-λ2, as were key antiviral proteins as RIG-I and MxA. EndoC-βH1 cells were productively infected by all six CVB serotypes. The Coxsackie and Adenovirus Receptor (CAR) was expressed on the beta cells and blocking antibodies to the receptor attenuated CVB infection. IFN-λ1 or IFN-λ2 treatment strongly reduced cellular permissiveness to CVB infection. Finally, we discovered that the genes encoding IFN-λ1/2 showed increased expression in islets from diabetic individuals compared to healthy controls.

Conclusion: Type III IFNs are expressed in the human pancreas at type 1 diabetes onset. Type III IFNs increase the expression of MHC class I and activate antiviral defense in human beta cells. Collectively, our studies show that type III IFNs are expressed in the islets at disease onset and that this family of IFNs induce the expression of MHC I and ISGs, markers of human type I diabetes. We also show that type III IFNs have antiviral activity. In summary, these results highlight a potentially important immunomodulatory function of type III IFNs during development of type 1 diabetes.

Supported by: Novo Nordisk Foundation, The Swedish Research Council, The Strategic Diabetes Research Program at Karolinska Institutet

Disclosure: E.E. Ringqvist: None.

219

Using the HALO image analysis platform to study pancreas pathology and insulitis in young people with recent-onset type 1 diabetes

R.C. Wyatt1, P. Leete1, M. Padilla2, M. Yang2, M. Bogdani3, G. Deutsch4, C. Flaxman1, M. Atkinson2, I. Kusmartseva2, N.G. Morgan1, S.J. Richardson1;

1Islet Biology Exeter (IBEx), University of Exeter, Exeter, UK, 2nPOD Pathology Core, University of Florida, Gainesville, USA, 3Benaroya Research Institute at Virginia Mason, Seattle, USA, 4Seattle Children's Hospital, Seattle, USA.

Background and aims: Worldwide, <600 type 1 diabetes (T1D) pancreata are available for research and, of these, fewer than 80 are from individuals who were <10y at onset with <1y disease duration. We have accessed the majority of these samples and used the digital pathology platform, HALO, to study insulitis and β-cell loss in multiple islets across each individual. The aim was to gain a more complete understanding of the progression of the disease in young children.

Materials and methods: Serial sections from 56 T1D cases (from the Exeter Archival Diabetes Biobank [EADB] & Seattle Children’s Hospital [SCH] collection) and 8 non-diabetic controls (EADB) were stained for insulin/glucagon/CD45, CD8/CD20/glucagon and CD4/CD20/glucagon. Pancreas samples from 55 non-diabetic donors collected by the Network for Pancreatic Organ Donors with Diabetes (nPOD) were used as additional controls. Islets with ≥15 CD45+ cells were defined as inflamed. The average number of islets/mm2 tissue and total numbers of CD45+, CD8+ T-, CD4+ T- and CD20+ B-cells in the islet and acinar tissue were calculated. We also examined the intensity of islet insulitis by calculating immune cell densities (CD45+ cells/mm2 islet).

Results: 7646 islets from the 56 T1D cases and 3508 islets from 8 EADB control subjects, respectively were analysed initially. Of these, 34.6% (T1D) and 99.9% (controls) contained insulin and 14% and 0.03% respectively, were inflamed. CD20+ B-cell, CD4+ and CD8+ T-cell numbers were assessed in 35 cases (including all 6 from SCH) and in all 8 EADB controls. Cases with an average of <3 B cells/insulin-containing islet (ICI) (defined as type 1 diabetes endotype 2 [T1DE2], n=15) had a median age at diagnosis of 15.5y and a median 32.9% residual ICIs, 10.8% of which were inflamed. Mean CD45+ cell and CD20+ B-cell numbers/ICI were 8.48±2.0 and 0.20±0.0.4, respectively. Cases with ≥3 B cells/ICI (T1DE1, n=20) had a median age at diagnosis of 6y, a median of 10.3% residual ICIs, 77.9% of which were inflamed. Mean CD45+ cell and B-cell numbers/ICI were 63.7±9.3 and 28.1±7, respectively, differing significantly from T1DE2 in all criteria (p<0.05). Islet inflammation was more intense in T1DE1 donors than in those with T1DE2 as the median number of CD45+ cells/mm2 ICI was 496.2 (IQR 98.1, 1191) vs 0 (IQR 0, 179.2; p<0.0001). The average number of islets/mm2was reduced in all 56 T1D donors (irrespective of endotype) compared with the 63 age-matched controls (p<0.0001).

Conclusion: Individuals with young-onset T1D (<7y) display a markedly different profile of insulitis from those diagnosed ≥13y, consistent with the existence of disease endotypes. The youngest children (T1DE1) have fewer residual ICIs at diagnosis and are their islets infiltrated with large numbers of T- and B-cells than those who are older at diagnosis (T1DE2). Type 1 diabetes pancreata also have fewer islets/mm2 of tissue compared with controls, which could contribute to early loss of glucose tolerance.

Supported by: DUK 16/0005480; JDRF CDA 5-CDA-2014-221-A-N

Disclosure: R.C. Wyatt: None.

220

Ag019 ActoBiotics as monotherapy or in association with teplizumab in recent-onset type 1 diabetes was safe and demonstrated encouraging metabolic and immunological effects

C. Mathieu1, S. Blomme2, S. Caluwaerts2, L. Haazen2, L. Steidler2, K. Van Huynegem2, J. Vermeiren2, K. Cerosaletti3, A. Wiedeman3, S.A. Long3, E. Serti4, AG019-T1D-101 Trial Investigators, P. Rottiers2, G.T. Nepom3, K.C. Herold5;

1UZ Leuven, Leuven, Belgium, 2Precigen ActoBio, Gent, Belgium, 3Benaroya Research Institute, Seattle, USA, 4Immune Tolerance Network, Bethesda, USA, 5Yale University, New Haven, USA.

Background and aims: AG019, a genetically modified Lactococcus lactis secreting human proinsulin and interleukin-10, is a novel, antigen-specific immune-modulating therapy for Type 1 Diabetes Mellitus (T1D). A Phase 1b/2a study is ongoing to assess the safety and tolerability, and potential treatment effects of AG019 alone and in association with anti-CD3 monoclonal antibody (teplizumab).

Materials and methods: Individuals with recent-onset T1D were treated with AG019 monotherapy (Phase 1b; 2 or 6 capsules BID for 8 weeks, n=19) or AG019 /teplizumab combination therapy or double-placebo (Phase 2a; 6 capsules BID for 8 weeks and 12 days teplizumab infusions, n=17). Safety was assessed as the incidence of treatment emergent adverse events (TEAEs). C-peptide area under the curve (AUC) levels were calculated after a mixed-meal tolerance test and preproinsulin (PPI)- and islet-specific CD4+ and CD8+ T-cell responses were measured using a Class II islet- and PPI-peptides activation assay and a Class I pooled tetramer assay, respectively.

Results: An interim analysis was performed on all Phase 1b monotherapy participants (10 adults and 9 adolescents; up to 12 months after treatment start) and on 14 Phase 2a combination therapy participants (12 adults (4:1) and 2 adolescents (open label); up to 6 months after treatment start). AG019 was well tolerated and safe when administered for 8 weeks as monotherapy or in association with teplizumab. No serious adverse events and no AG019 treatment discontinuation occurred due to TEAEs. Most TEAEs reported were mild (72.3%) and sometimes moderate (24.3%). AG019 safety profile was similar between adults and adolescents and there was no evidence of dose-related TEAEs. The safety profile of teplizumab in association with AG019 was consistent with that of teplizumab. C-peptide (mean AUC versus baseline) stabilized (coefficient of variation on at least 3 visits ≤ 9.7%) or increased at month 6 in 58% (7/12) of monotherapy participants aged 17 or older. In a subset analysis, AG019 monotherapy was associated with an antigen-specific immune response at 3 months, including a trend towards increased frequency of PPI-specific Tr1-cells and islet-reactive Treg-cells (4/4 adults), and a decreased mean frequency of PPI-specific CD8+ T-cells (p=0.08, 5 adults / 5 adolescents). C-peptide stabilized or increased at month 6 in 70% (7/10) of combination therapy participants aged 17 or older. In combination therapy, trends towards an increased frequency of PPI- and islet-reactive Tr1 cells (2/5 and 3/5 adults, respectively), and decreased mean frequency of PPI-specific CD8+ T-cells (p=0.2, 7 adults / 1 adolescent) was seen at 3 months.

Conclusion: AG019 was safe and well tolerated, and may favorably modulate autoimmune T-cell responses. The metabolic and immunological data suggest that AG019, alone or associated with teplizumab, may have the potential to be effective in preserving insulin-production in recent-onset T1D. AG019 has the convenience of oral administration and data suggests an opportunity for an improved effect with prolonged treatment duration.

Clinical Trial Registration Number: NCT03751007

Disclosure: C. Mathieu: Employment/Consultancy; Consultant for Precigen Actobio.

OP 38 Novel agents

221

The novel GIP, GLP-1, and glucagon triple receptor agonist LY3437943 exhibits robust efficacy in preclinical models of obesity and diabetes

T. Coskun, J.S. Moyers, W. Roell, L. O'Farrell, A. Regmi, A.D. Showalter, K.W. Sloop, D.B. Wainscott, F.S. Willard, J. Ficorilli, O. Cabrera, S. Urva, F. Norouziyan Cooper, J. Alsina-Fernandez, H. Qu;

Eli Lilly and Company, Indianapolis, USA.

Background and aims: The ever-growing prevalence of obesity and its associated comorbidities (type 2 diabetes, NASH/NAFLD) is driving the need to discover new therapies for improving metabolic health. Recently, multi-receptor agonists have offered promise for meeting this need. Here, we characterize LY3437943, a novel single agent tri-agonist at the glucose-dependent insulinotropic polypeptide (GIP), glucagon-like peptide-1 (GLP-1), and glucagon (Gcg) receptors (R).

Materials and methods: All in vitro assays in HEK293 or endogenous cells were performed in the absence of albumin to allow direct comparison to native peptides without the confounding influence of albumin binding. Intravenous glucose tolerance tests (IVGTT) were performed in anesthetized Wistar rats. The effects of LY3437943 on gastric emptying, body weight, food intake, and energy expenditure were determined in diet-induced obese (DIO) C57/Bl6 mice.

Results: Pharmacologic analysis of LY3437943 in cAMP assays using recombinant cell lines expressing the individual receptors indicated a potency balance favouring GIPR agonism (1.7- and 2.5-fold less potent at the GLP-1R and GcgR, respectively, but 7-fold more potent at the GIPR; all potencies in relation to the native ligands). In endogenous cells, LY3437943 regulated adipocyte lipolysis and hepatocyte glucose output. In vivo studies demonstrated regulation of multiple metabolic endpoints. Acute treatment with LY3437943 dose-dependently inhibited semi-liquid gastric emptying in mice and enhanced glucose dependent insulin secretion in rat IVGTT experiments. Chronic studies in DIO mice reduced food intake and body weight (45% weight loss primarily via reduced fat mass) superior to other GIPR and GLP-1R agonists. In these experiments, LY3437943 lowered blood glucose and plasma insulin, indicating improved insulin sensitivity. Additionally, chronic administration improved biomarkers of liver health, decreasing both plasma alanine aminotransferase and liver triglycerides. Rodent and cynomolgus monkey PK modeling also suggested the potential for weekly dosing in humans.

Conclusion: Taking these findings together, LY3437943 is a novel tri-agonist at the GIPR, GLP-1R, and GcgR, producing superior weight loss and glycaemic control compared with other incretin receptor-targeting molecules and offers additional benefit for liver health. These findings prompt evaluation of the potential clinical benefit of LY3437943 in individuals with obesity and metabolic diseases.

Supported by: Eli Lilly and Company

Disclosure: T. Coskun: Employment/Consultancy; Eli Lilly and Company. Stock/Shareholding; Eli Lilly and Company.

222

Novel GIP/GLP-1/glucagon receptor agonist LY3437943: a first in human dose study in healthy subjects

C.T. Benson, C. Loghin, Y. Du, M.K. Thomas, T. Coskun, S. Urva, Z. Milicevic;

Eli Lilly and Company, Indianapolis, USA.

Background and aims: Multifunctional incretins are in clinical development for several metabolic conditions. Novel LY3437943 has potent agonist activity on glucose-dependent insulinotropic polypeptide (GIP), glucagon-like peptide-1 (GLP-1) and glucagon receptors. The primary objective of this randomised, double-blind, placebo-controlled, Phase 1, first in human study was to assess safety and tolerability of single-ascending doses of LY3437943.

Materials and methods: Forty-five healthy participants were randomised (6:2) to subcutaneous LY3437943 (6 rising dose levels) or placebo. Vital signs, ECGs, laboratory data and adverse events (AEs) were monitored to assess safety and tolerability. LY3437943 pharmacokinetics (PK) as well as change from baseline in fasting insulin, C-peptide and weight were measured. Appetite was assessed using a visual analog scale (VAS).

Results: The most common treatment-emergent AEs were gastrointestinal, including vomiting (with higher doses), abdominal distention and nausea, which were dose-dependent, mostly mild in severity, occurred within 4 days of dosing and resolved within a week of onset. Dose-dependent increases in heart rate and decreases in systolic blood pressure were observed, which returned to near baseline by Day 29. Mean terminal half-life of LY3437943 ranged from 134-165 h (~7 days) across the 6 doses, supporting once-weekly dosing. Dose-dependent increases in mean fasting insulin and C-peptide, with maximum levels observed at 24 and 48 h, returned to near baseline by Day 15. Dose-dependent weight loss was statistically significant with the 3 highest doses vs placebo (up to 3.5 kg at the highest dose). Weight loss was maintained up to Day 43 following single administration of the two highest doses. Overall VAS score significantly increased with higher doses compared with placebo, reflecting decreased appetite.

Conclusion: Triple-agonist peptide LY3437943 had a safety and tolerability profile similar to other incretins in Phase 1 trials with PK and pharmacodynamic outcomes that support further clinical evaluation.

Clinical Trial Registration Number: NCT03841630

Supported by: Eli Lilly and Company

Disclosure: C.T. Benson: Employment/Consultancy; Eli Lilly and Company. Stock/Shareholding; Eli Lilly and Company.

223

Safety and pharmacokinetic study of CPL207280, a novel GPR40 receptor agonist, after a multiple-dose in healthy volunteers

K. Bazydlo-Guzenda1, K. Jarus-Dziedzic2, A. Gierczak-Pachulska1, K. Bus-Kwasnik3, P.J. Rudzki1, M. Wieczorek1;

1Celon Pharma SA, Kazun Nowy, 2BioResearch Group Sp. z o.o., Kajetany, 3Łukasiewicz Research Network, Warsaw, Poland.

Background and aims: G-protein-coupled receptor 40 (GPR40) is a free fatty acid receptor mainly expressed in pancreatic β-cells and mediates signalling from free fatty acid (FFA) to modulate glucose-stimulated insulin secretion (GSIS). GPR40 emerged as an anti-diabetic target and was proposed as a novel treatment modality for type 2 diabetic (T2D) patients. First in class developed and evaluated in clinical trials GPR40 agonist - fasiglifam, was great hope for diabetic society - both patients and clinicians - as it might have offered a new modality for the treatment of diabetes. It was supposed to supersede a popular sulfonylurea as it showed limited hypoglycemia risk and it appeared as a potential competitor to insulinogenic GLP-1 analogues. However, it caused liver toxicity in phase 3 of clinical trials in T2D patients thereby its development was stopped. This example suggested that toxicity might be a hallmark of the entire class and hence quenched both the research field and hopes for a new secretagogue to be launched to the market in the nearest future. However, later some papers appeared, suggesting hepatotoxicity related more closely to the molecule than to the target. The aim of this Phase 1 clinical study is to determine the safety, tolerability and pharmacokinetic (PK) profile of CPL207280 a new GPR40 agonist - in healthy subjects in fasting condition.

Materials and methods: This was a multiple-dose, randomized, double-blind, placebo-controlled, ascending dose study in healthy volunteers with administrations once daily over 2 week. CPL207280 was administrated orally as a prolonged-release tablet at four doses: 60, 120, 240, 480 mg among 32 subjects (n = 8 per cohort, with randomization 3:1) who met all the inclusion and none of the exclusion criteria. Blood samples for pharmacokinetic analysis were collected on days 1, 8, 14. On the other days, samples were collected just before and 1 hour after CPL207280 administration and up to 48 hours following the last dosing. CPL207280 concentration in human plasma was measured using HPLC/MS/MS. Safety assessment included adverse events (AE) reporting, clinical laboratory tests, vital signs, physical examination, electrocardiography (ECG).

Results: The administration of CPL207280 was generally safe and well-tolerated with no serious AEs. All adverse events were classified as not related to the study product. No increase in the tested hepatic parameters (ALT, AST, ALP, bilirubin, creatinine) was observed. There were no hypoglycaemia episodes during the study. CPL207280 plasma concentration-time profile of all subjects was determined during the study. CPL207280 mean Cmax values during all dosing days were in the range of 106 - 500 ng/mL for cohort 1 and 4, respectively, and were observed at 0.33-4 hours after administration. The CPL207280 exposure increased in a dose-dependent (between the cohorts) and in a time-dependent (inside the cohorts, between day 1 and 14) manner.

Conclusion: The administration of CPL207280 for 14 days was safe and well-tolerated. The pharmacokinetic profile of CPL207280 confirms findings from the single administration study in healthy volunteers, supports once-daily administration and justify further development of the therapy for patients suffering from T2D.

Clinical Trial Registration Number: NCT04622111

Supported by: The study was co-financed by the National Centre of Research and Development (grant no. POIR.01.01.01.-00-0334/17)

Disclosure: K. Bazydlo-Guzenda: Employment/Consultancy; Celon Pharma employee.

224

DA-1241 a novel GPR119 agonist: Safety, tolerability, pharmacokinetics, and pharmacodynamics: Part 2 of multiple ascending dose study in type 2 diabetes patients

M. Hompesch1, B. Franey1, M. Grimm1, D. Lee2, J. Jeong2, M.-K. Kim2;

1ProSciento, Chula Vista, USA, 2Dong-A ST, Seoul, Republic of Korea.

Background and aims: DA-1241 is a novel small molecule selective GPR119 agonist. In preclinical studies DA-1241 enhanced metabolic hormones, and improved metabolic control characteristics. The primary objective of Part 2 was to assess safety and tolerability of multiple once daily oral doses of DA-1241 versus placebo and Sitagliptin (SG) in Type 2 Diabetes Mellitus (T2DM). Secondary objectives were to establish pharmacokinetic (PK) and pharmacodynamic (PD) characteristics.

Materials and methods: Part 2 was a double blind placebo and SG controlled study with three sequential cohorts of T2DM (n=25/cohort) blinded and randomized (3:1:1) to receive DA-1241: 25, 50 or 100 milligram (mg; n=15/cohort), placebo (n=5/cohort) or SG (n=5/cohort) single daily oral doses for 56 days.

Results: 84 subjects were enrolled in Part 2, 3 subjects were withdrawn due to Treatment Emergent Adverse Events (TEAEs; 1 active, 2 placebo), 72 subjects completed the study. There were no relevant demographic imbalances (age 56.8 ± 7.3, BMI 29.4 ± 3.3). Doses tested were generally safe and well tolerated. The most frequent TEAEs were mild GI side effects (nausea, diarrhea, abdominal pain), all resolved spontaneously. Day 56 PK Cmax and AUC0-tau parameters were dose proportional. Fasting plasma glucose trended towards improvement for all DA-1241 doses, as did incremental glucose exposure after Mixed Meal Tolerance Tests (MMTTs; Change from Baseline to Day 56 iAUC0-4h: 25mg: 18.5±89.2, 50mg: -5.1±86.8, 100mg: -38.6±58.7, Placebo: 34.4±145.9, SG: -23.9±132.8) and time spent at blood glucose < 180 mg/dL. HbA1c trended towards improvement across dosing groups, as did body weight. In pooled analyses of the data on Day 56, glucose AUC0-4h values during MMTT were well correlated with HbA1c (%) and glucose AUC0-24h values assessed using a continuous glucose monitoring system (r = 0.626 and 0.685, respectively for 72 subjects; p<0.001). Total GIP, GLP-1 and PYY were increased at day 56, consistent with the mechanism of action of DA-1241. Among them, after single administration of DA-1241 tablet on Day 1, plasma excursion (AUC0-4h) of total GLP-1 during MMTT showed a distinctive positive correlation with both the AUC0-24h and Cmax of DA-1241 (r = 0.4999 and 0.3834, respectively for total 46 patients; p<0.05).

Conclusion: This phase 1b study in T2DM showed favorable safety, tolerability and PK profiles of DA-1241. Biomarkers and PD data confirmed the mechanism of action and showed favorable efficacy data trends.

Clinical Trial Registration Number: NCT03646721

Disclosure: M. Hompesch: Employment/Consultancy; ProSciento. Stock/Shareholding; ProSciento.

OP 39 Glucagon metabolism in humans

225

Different patterns of glucose- and glucagon-stimulated insulin secretion in new diabetes subphenotypes

K. Prystupa1,2, K. Kantartzis1,3, M. Heni1,2, L. Fritsche1,3, O.-P. Zaharia4,3, I. Yurchenko4,3, K. Bodis4,3, S. Trenkamp4,3, V. Burkart4, J. Szendrődi4,3, N. Stefan1,2, A.L. Birkenfeld1,2, A. Fritsche1,2, M. Roden4,3, R. Wagner1,2;

1Department of Internal Medicine IV, Division of Endocrinology, Diabetology and Nephrology, University of Tübingen, Tübingen, 2Institute of Diabetes Research and Metabolic Diseases (IDM) of the Helmholtz Centre Munich, Tübingen, 3German Center for Diabetes Research (DZD), Munich-Neuherberg, 4Institute for Clinical Diabetology, German Diabetes Center, Leibniz Center for Diabetes Research at Heinrich-Heine-University Düsseldorf, Düsseldorf, Germany.

Background and aims: Glucagon-like peptide-1 (GLP-1) improves insulin secretion in people with type 2 diabetes by affecting GLP-1 receptors on beta-cells. Recent data indicate that glucagon mostly exerts its insulinotropic effects via these GLP-1 receptors. We hypothesized (i) that the rise of insulin secretion after glucagon administration reflects a GLP-1 effect on beta-cells and (ii) that the GLP-1 effect differs between the novel clusters (subgroups) of diabetes.

Materials and methods: We compared insulin response during intravenous Glucose Tolerance Test (IVGTT) and Glucagon Stimulation Test (GST) in 562 people with recent-onset diabetes and 190 controls from the German Diabetes Study (GDS), and used the known genetically mediated incretin resistance by the variant rs7903146 in TCF7L2 as a benchmark. Furthermore, we compared GST and IVGTT data between the diabetes subgroups. Linear mixed models were used to compare the effects of age, BMI, insulin resistance, and fasting glucose on insulin response, given as fold change C-peptide, in IVGTT and GST. Multivariable linear regressions were performed to assess the effect of the TCF7L2 genotype, diabetes status, and the interaction of risk-genotype and glucose dynamics on insulin response.

Results: Intravenous glucagon bolus elicited a higher insulin response despite less increase in blood glucose, compared to glucose injection during IVGTT (fold change C-peptide 2.4±0.9 vs. 1.8±1.3, p<0.001; blood glucose 129.8± 23.9 vs. 245.9± 40.6 mg/dl, p<0.001). Glucagon-enhanced insulin secretion was less impacted by glycemia, aging, obesity and insulin resistance than glucose-stimulated insulin secretion (pinteraction≤0.002). Incretin resistance was evident in GST by reduced glucose-dependent insulin release in rs7903146 risk allele carriers (p=0.038). The risk allele was negatively associated with insulin release during both tests in severe insulin resistant diabetes (SIRD) (all p<0.05), but positively in mild age-related diabetes (MARD) (β=0.07, p=0.033).

Conclusion: These results from large comprehensively phenotyped human cohort are consistent with previous preclinical data that glucagon acts as a partial agonist of the GLP-1 receptors in beta-cells. We show that diabetes subgroups have distinct impairments in insulin secretion and that the TCF7L2 risk genotype exhibits different effects on insulin secretion in these diabetes subgroups.

Clinical Trial Registration Number: NCT01055093

Disclosure: K. Prystupa: None.

226

Increased glucagon sensitivity in totally pancreatectomised patients

I. Rix1,2, A. Lund1, L.F. Garvey1, K. Owen2, C. Hansen3, G. van Hall4,5, J.J. Holst5,6, T. Vilsbøll1,7, F.K. Knop1,7;

1Center for Clinical Metabolic Research, Gentofte Hospital, Hellerup, 2Zealand Pharma A/S, Søborg, 3Department of Surgery, Rigshospitalet, Copenhagen, 4Clinical Metabolomics Core Facility, Department of Clinical Biochemistry, Rigshospitalet, Copenhagen, 5Department of Biomedical Sciences, Faculty of Health and Medical Sciences, University of Copenhagen, Copenhagen, 6Novo Nordisk Foundation Center for Basic Metabolic Research, Faculty of Health and Medical Sciences, University of Copenhagen, Copenhagen, 7Steno Diabetes Center Copenhagen, Gentofte, Denmark.

Background and aims: Glucagon secretion and signalling are affected in several metabolic disorders including different types of diabetes. The aim of this study was to investigate the effect of exogenous glucagon on glucose, amino acids, and lipid metabolism in participants deprived of endogenous glucagon and/or insulin.

Materials and methods: We included nine patients who had undergone total pancreatectomy (PX, age 64.2 [mean] years; BMI 22 kg/m2), eight patients with C-peptide negative type 1 diabetes (T1D, age 61.6 years; BMI 23.9 kg/m2) and nine healthy controls (age 62.0 years; BMI 23 kg/m2) in a randomised, double-blinded, placebo-controlled, crossover study. Participants received a 2.5-hour i.v. infusion of glucagon (4 ng/kg/min) and placebo (saline), respectively, under fasting conditions with usual basal insulin administration.

Results: Glucagon infusion increased plasma glucagon to high physiological levels in all groups (PX 67.6±3.8 [mean±SEM] pmol/l; T1D 64.1±3.0 pmol/l; and controls 68.7±3.7 pmol/l) with no difference between groups (p=0.655). Compared with placebo, glucagon infusion increased insulin levels significantly in the controls (p<0.001). Baseline-subtracted glucose levels increased in all groups compared to placebo (PX 7.1±0.4 mmol/l, p<0.001; T1D 2.8±1.2 mmol/l, p=0.060; and controls 1.2±0.2 mmol/l, p<0.001; Figure). In the PX group, baseline-subtracted AUC (bsAUC) for glucose was 2.2 and 3.5-fold higher compared to the T1D group (p=0.015) and the control group (p<0.001), respectively. PX patients were characterised by 1.9 and 1.8-fold elevated fasting plasma levels of amino acids compared to the T1D group (p<0.001) and control group (p<0.001), respectively (Figure). In the PX group, glucagon reduced bsAUC for total amino acids by 2.8 and 2.3-fold compared to the T1D group (p=0.003) and the control group (p=0.008), respectively. In the healthy controls, glucagon infusion reduced bsAUC for glycerol (p<0.001) and palmitate (p=0.002), whereas no changes were observed in the diabetic groups. Glucagon did not change plasma levels of LDL, HDL or total cholesterols, or triglycerides in any of the groups.

Conclusion: We conclude that PX patients in terms of amino acid turnover and endogenous glucose production are hypersensitive to exogenous glucagon; most likely because of upregulation of glucagon receptive mechanisms in response to their lack of pancreatic glucagon secretion.

figurebv

Clinical Trial Registration Number: NCT03526445

Supported by: Innovation Fund Denmark, Novo Nordisk Foundation, A.P. Møller Foundation, Aase and Ejnar Danielsen Foundation, Inge & Per Refhall's Research Grant, Augustinus Foundation

Disclosure: I. Rix: Employment/Consultancy; Zealand Pharma A/S.

227

Splanchnic and leg glucagon metabolism in healthy adults

A. Basu1, R. Ruchi1, Y. Yadav1, A. Weaver1, L. Wilkins1, M. Schiavon2, C. Cobelli2, C. Dalla Man2, A. Pandey3, K. Johnson3, S. Renuse3, R. Basu1;

1University of Virginia, Charlottesville, USA, 2University of Padova, Padova, Italy, 3Mayo Clinic, Rochester, USA.

Background and aims: Regional and systemic glucagon metabolism is poorly understood in humans. To study post absorptive glucagon turnover, we applied isotope dilution technique combined with organ catheterization to estimate splanchnic and leg glucagon metabolism in healthy nondiabetic adults.

Materials and methods: After IRB approval and informed consent, subjects admitted to the Interventional Radiology Unit after an overnight fast, had catheters placed in the right femoral artery and vein under aseptic precautions. Under fluoroscopic guidance, a hepatic venous catheter was positioned through the femoral vein. 13C15N Glucagon tracer was infused intravenously to measure glucagon turnover, and Indocyanine Green infused via the femoral artery sheath to measure splanchnic and femoral plasma flows. After a basal sampling period, exogenous glucagon was infused @ 0.65 ng/kg/min (medium) for 60 minutes followed by 1.5 ng/kg/min (high) for another 60 minutes to determine the effects of plasma glucagon concentrations spanning the physiological range on regional glucagon extraction and turnover. Samples were collected periodically from the femoral artery, femoral vein, and hepatic vein for 13C15N Glucagon (tandem MS/MS), glucagon (Mercodia), glucose (YSI) and indocyanine green (UVA-Vis) concentrations. Splanchnic and leg glucagon extraction and glucagon turnover was calculated using regional plasma flow, 13C15N Glucagon concentrations and steady state equations.

Results: Data from 8 healthy (age 23.1 ± 2.9 yrs., 4 females, BMI 26.6 ± 3.5 kg/m2, fasting plasma glucose 4.7 ± 0.5 mM; HbA1c 5.0 ± 0.2 %) subjects analyzed thus far is shown. There was substantial fractional glucagon extraction across the splanchnic (23.5 ± 12.4 vs. 30.4 ± 16.3 vs. 23.4 ± 13.1 %: basal vs. medium vs. high) and leg (35.9 ± 17.9 vs. 28.5 ± 8.0 vs. 22.4 ± 7.9 %: basal vs. medium vs. high) regions during arterial glucagon concentrations spanning the physiological range (26.8 vs. 39.9 vs. 66.4 pg/ml: basal vs. medium vs. high; Fig 1: upper panel). Interestingly, systemic rate of appearance of glucagon rapidly dropped to undetectable levels with exogenous glucagon administration (Fig 1: lower panel).

Conclusion: These initial data in healthy nondiabetic adults demonstrate substantial fractional glucagon extraction across the splanchnic and leg regions and that exogenous glucagon administration inhibits endogenous glucagon production.

figurebw

Supported by: NIH-R01-DK-085516 to AB; NIH-R01-DK-029953 to RB

Disclosure: A. Basu: None.

228

A 2-year follow-up of RYGB surgery in obesity and type 2 diabetes: enhanced responses of multiple hormones to oral glucose but not i.v. arginine challenge

G. Fanni, P. Katsogiannos, M.J. Pereira, J.W. Eriksson;

Clinical Diabetology and Metabolism, Department of Medical Sciences, Uppsala University, Uppsala, Sweden.

Background and aims: Bariatric surgery, in particular roux-en-Y gastric bypass (RYGB), is known to greatly improve glycemic control in patients with type 2 diabetes, even to its remission. This effect is largely independent of the magnitude of weight loss, and endocrine and neural responses are likely to be involved. We aimed at shedding light on the effects of RYGB on the dynamic endocrine crosstalk contributing to this metabolic shift. A systematic assessment of insulin, glucagon, gut hormones, growth hormone (GH) and the hypothalamus-pituatary-adrenal (HPA) axis following fasting and nutrient challenges, respectively, was performed repeatedly up to 2 years post-RYGB.

Materials and methods: Our cohort includes 13 patients with T2DM (up to ten years of disease duration, age 18 to 65 y/o, BMI 30-45 kg/m2) undergoing RYGB who were compared to 6 patients with standard of care but no RYGB (randomized design, but controls not included in the present analyses). We analyzed plasma samples for insulin, glucagon, GLP-1, GIP, ACTH and cortisol (PYY and GH ongoing) on four occasions: 1) before surgery; 2) 4 weeks; 3) 24 weeks; 4) and 104 weeks post-RYGB; and in three conditions assessed sequentially on the same day: 1) in the morning after an overnight fast; 2) during i.v. arginine challenge, 20 g; 3) during OGTT 75 g, 3h.

Results: Following RYGB there was a reduction of fasting levels of insulin, glucagon, GLP-1 and GIP (p<0.001, p<0.02, p<0.001, p=0.05, respectively; figure) as well as during arginine challenge (p<0.001, p=0.22, p<0.001, p=0.03). The relative stimulating effect of arginine at 3 minutes was not altered by RYGB and remained intact throughout all visits. During OGTT, we identified a more rapid rise for GIP levels following RYGB but with essentially unchanged AUC. There was also a marked increase for glucagon, GLP-1 (see figure) and, surprisingly, for ACTH and cortisol (all p<0.01). All post-RYGB effects were sustained throughout the 2-year follow-up.

Conclusion: There was a stimulating effect of i.v. arginine on insulin, glucagon, GLP-1, and GIP, and this effect was maintained following RYGB, albeit with overall lower absolute levels. Following RYGB, we found raised OGTT-stimulated secretion of glucagon, GLP-1, ACTH, and cortisol and a more rapid GIP secretion pattern, and this was sustained over 2 years. Thus, oral glucose stimulation of multiple hormones, with pituitary, adrenal, pancreatic islet and gut origin, were similarly amplified following RYGB, and this was sustained over at least 2 years. in contrast, the fold effect following arginine stimulation remained unchanged after RYGB. Dynamic regulation of multiple hormones may contribute to the beneficial effects of RYGB on glycemic control and energy balance in obesity and type 2 diabetes.

figurebx

Clinical Trial Registration Number: NCT02729246

Supported by: Svenska DS, Ernfors F, Exodiab, Akademiska SH.

Disclosure: G. Fanni: Grants; Swedish Diabetes Foundation, Ernfors Foundation, Exodiab (Excellence of Diabetes Research in Sweden), Novo Nordisk Foundation, ALF grants Uppsala University Hospital.

OP 40 Protecting the kidney in diabetes

229

Protective impact and potential mechanisms of Elabela on DKD via ß-arrestins

M. Shi1, D. Tan2, Y. Chen1, W. Gu1, H. Zhang1;

1The Affilated Huai’an No.1 People’s Hospital of Nanjing Medical University, Huai’an, 2Endocrinology, Huai’an Clinical College of Xuzhou Medical University, Huai’an, China.

Background and aims: ELABELA (ELA), an endogenous ligand for apelin receptor (APJ), is decreased in plasma of DKD patients. The effects of ELA on relieving diabetic renal lesion have not been reported. This study was performed to explore the protective impact and potential mechanisms of ELA on DKD via β-arrestins.

Materials and methods: Sixteen eight-week-old male db/db mice were randomly divided into diabetic nephropathy group (db/db group) and intervention group (db/db+ELA group) with administration of ELA (5mg ∙ kg-1 ∙ day-1) for 8 weeks. 8 age-matched male db/m mice were recognized as normal control group (db/m group). Both db/db and db/m group received equivalent normal saline injection for 8 weeks. Body weights and blood glucose levels were measured once a week. After 8 weeks, blood and urine samples were collected. Serum creatinine and urinary albumin to creatinine ration (UACR) were examined. The kidneys were taken after sacrificing mice. The morphology change of kidney tissues were assessed by HE and PAS staining. The expression of β-Arrestin1/2 was detected by immunohistochemistry. The levels of fibronectin (FN), collagen type IV (Col-IV) and β-arrestin1/2 in kidney tissues were examined by western blotting.

Results: (1) With the intervention of ELA, UACR, serum creatinine and the ratio of kidney weight to body weight of db/db mice were significantly decreased (P<0.05). (2) Compared with db/m mice, db/db mice exhibited renal tubular epithelial cells edema, decreased ratio of non-lumen area to total area, thickened basement membrane and increased glycogen accumulation, however, those pathological abnormality were mitigated by ELA. (3) The levels of FN, Col-IV and β-Arrestin-1/2 in db/db mice were higher than db/m mice, while ELA treatment suppressed the expression of FN, COL-IV and β-Arrestin-1/2 (P<0.05).

Conclusion: ELA may play a role in protection from diabetic kidney damage via down-regulating the expression of β-Arrestin-1/2 and inhibiting the secretion of extracellular matrix.

Supported by: National Natural Science Foundation of China Grant Award (81700723), BK20191213

Disclosure: M. Shi: None.

230

Chop-ASO ameliorates glomerular and tubular damage on top of ace-inhibition in diabetic nephropathy

R. Klar1, K. Shahzad2, S. Fatima2, M. Al-Dabet2,3, I. Gadi2, H. Khawaja2, A. ElWakiel2, P.P. Nawroth4, P.R. Mertens5, S. Michel1, F. Jaschinski1, B. Isermann2;

1Secarna Pharmaceuticals GmbH & Co. KG, Planegg, Germany, 2University Hospital Leipzig, Leipzig, Germany, 3American University of Madaba, Amman, Jordan, 4German Diabetes Center, University of Heidelberg, Heidelberg, Germany, 5Otto-von-Guericke University, Magdeburg, Germany.

Background and aims: Diabetic nephropathy (dNP) is the leading cause of end-stage renal disease globally. Despite promising therapeutic developments dNP still lacks an efficient therapy halting or even reversing the disease. Maladaptive endoplasmatic reticulum (ER)-stress signaling has been linked to dNP and is mediated by the transcription factor C/EBP homologous protein (Chop) that is hard to target by conventional modalities like small molecule inhibitors or antibodies.

Materials and methods: We identified locked nucleic acid (LNA)-modified antisense oligonucleotides (ASOs) specifically targeting Chop (Chop ASO) and investigated therapeutic efficacy in vivo and in vitro. Renal functional parameters, morphological changes and differentially regulated genes and pathways were assessed in db/db mice treated with Chop ASO alone or on top of ACE-inhibition (ACEi).

Results: Chop ASO efficiently reduced Chop protein levels in diabetic mice in early (8 weeks aged db/db mice) and late stages (16 weeks aged db/db) of experimental dNP. Therapy with Chop ASO in early stage dNP led to a reduction of albuminurea, glomerular damage and interstitial fibrosis to a similar extent than ACEi. Combined intervention significantly reduced interstitial fibrosis as compared to each intervention alone. Intervention in late stage dNP with a combination of Chop ASO + ACEi resulted in superior therapeutic efficacy with regard to reduction of mesangial matrix accumulation and glomerular basement membrane thickness compared to the respective monotherapies. Strikingly, in contrast to ACEi, Chop ASO alone or in combination with ACEi showed therapeutic effects in the tubular compartment as measured by tubular diameter, apoptosis and expression of the damage marker KIM-1. Gene set enrichment analysis after gene expression analysis revealed the disease gene sets Hyperglycemia, Tubulointerstitial Fibrosis and Hypertensive Disease being the top three pathways regulated by Chop ASO therapy. Slc5a2 (SGLT2), the target of the increasingly used SGLT2 inhibitors was one of the genes downregulated under Chop ASO therapy. In line with the in vivo findings in mice, we observed a normalization of Chop expression and a reduction of apoptosis in glucose-stressed, Chop-ASO-treated human renal cells.

Conclusion: LNA-modified ASOs targeting the transcription factor Chop efficiently reduced renal Chop expression in a mouse model of dNP. The therapeutic intervention with Chop ASO especially at late stages of dNP provide an added benefit on top of the current standard of care (ACEi). Our studies demonstrate that ASO-based therapies that efficiently reduce maladaptive Chop expression in kindey diseases have the potential to provide a novel treatment option for patients in need.

Supported by: IS-67/8-1 IS-67/11-1 CRC 1118/B07 CRC854/B26 SH849/1-2 SH849/4-1 361210922/GRK2408/P7&P9 361210922/GRK2408/P5

Disclosure: R. Klar: Employment/Consultancy; Secarna Pharmaceuticals GmbH & Co. KG.

231

SGLT2 inhibition prevents acute kidney injury by inhibiting glucose transport and mTORC1 activity in proximal tubule cells

A. Kogot-Levin1, L. Hinden2, Y. Riahi1, I. Abramovich3, E. Gottlieb3, A. Rothner2, L. Kadosh1, T. Vaisid4, J. Roitelman4, E. Cerasi1, J. Tam2, O. Mosenzon1, G. Leibowitz1;

1Sec. of Endocrinology, Hadassah Hebrew University, Jerusalem, 2Obesity and Metabolism Laboratory, Institute for Drug Research, School of Pharmacy, Faculty of Medicine, The Hebrew University of Jerusalem, Jerusalem, 3Department of Cell Biology and Cancer Science, Rappaport Faculty of Medicine, Technion - Israel Institute of Technology, Haifa, 4Institute of Lipid and Atherosclerosis Research, Sheba Medical Center, Tel Hashomer, Tel Aviv, Israel.

Background and aims: Patients with diabetes are prone to develop acute kidney injury (AKI), which markedly increases the risk for end-stage kidney disease. SGLT2 inhibitors (SGLT2i) exert robust kidney protection in patients with diabetic kidney disease (DKD). Previous reports suggested that SGLT2i treatment might increase the risk for AKI, with subsequent publication of a "black box" warning by the FDA. However, randomized-controlled clinical trials showed that treatment with SGLT2i is associated with decreased, rather than increased incidence of AKI. We have recently reported that SGLT2i prevent DKD by inhibiting glucose transport and mTORC1 activity in renal proximal tubule cells (RPTCs). Herein, we studied whether a similar mechanism operates in the context of drug-induced AKI.

Materials and methods: Eight-week old wildtype mice were injected intraperitoneally with a single dose of folic acid (FA; 250 mg/kg) and were pretreated with or without dapagliflozin for 5 days. Mice were sacrificed 2 or 14 days following FA injection and analyzed for the expression of markers of tubular injury, fibrosis and inflammation, and glycolytic enzymes. mTORC1 activity was analyzed by Western blotting and immunofluorescence for phospho-S6. Cre-lox system was used to generate RPTC-specific Raptor knockout mice.

Results: AKI developed 2 days after FA administration, evidenced by increased plasma creatinine and urea nitrogen levels and increased gene and protein expression of tubular injury, fibrosis and inflammation markers, along with increased expression of glycolytic enzymes, DNA damage repair (phosphorylated-p53) and autophagy. FA-induced tubular injury resulted in tubular atrophy and dilatation, and interstitial fibrosis by Masson's trichrome staining at 14 days. SGLT2i prevented these alterations and the late development of tubular atrophy and fibrosis. AKI was associated with marked activation of RPTC mTORC1 activity, which was prevented by pre-treatment with dapagliflozin. Heterozygous Raptor KO in RPTCs attenuated the activation of mTORC1 by FA and inhibited mTORC1-TGFβ-mediated fibrosis.

Conclusion: SGLT2i are highly effective in protecting the kidney from drug-induced kidney injury, which is more prevalent in patients with diabetes. The protective effects of SGLT2i are mediated via inhibition of glucose transport and prevention of mTORC1-TGFβ induced kidney fibrosis.

Disclosure: A. Kogot-Levin: None.

232

Acute effects of dapagliflozin on renal oxygenation and perfusion in type 1 diabetes with albuminuria: a randomised, double-blind, placebo-controlled crossover trial

J.C. Laursen1, N. Søndergaard Heinrich1, J.M.L. de Melo1, B. Haddock2, I.K.B. Rasmussen1, F. Safavimanesh1, C. Stevns Hansen1, J. Størling1, H.B. Wiberg Larsson2, P.-H. Groop3, M. Frimodt-Møller1, U.B. Andersen2, P. Rossing1;

1Steno Diabetes Center Copenhagen, Copenhagen, Denmark, 2Rigshospitalet, Copenhagen University Hospital, Copenhagen, Denmark, 3Folkhälsan Research Center, Helsinki, Finland.

Background and aims: Inhibitors of the sodium-glucose cotransporter 2 (SGLT2) slow the progression of diabetic kidney disease, possibly by reducing the proximal tubule transport workload with subsequent improvement of renal oxygenation. We aimed to test this hypothesis in individuals with type 1 diabetes and albuminuria.

Materials and methods: A randomised, double-blind, placebo-controlled, crossover trial with a single dose of 50 mg of the SGLT2 inhibitor dapagliflozin and placebo in random order, separated by a two-week washout period. Magnetic resonance imaging (MRI) was used to assess renal R2* (a low value corresponds to a high tissue oxygenation), renal perfusion (arterial spin labelling) and renal artery flow (phase contrast imaging) at baseline, three- and six hours from tablet ingestion. Exploratory outcomes, including baroreflex sensitivity, peripheral blood oxygen saturation and peripheral blood mononuclear cell mitochondrial oxygen consumption rate, were evaluated at baseline and 12 hours from medication.

Results: Between February 3, 2020 and October 23, 2020, 31 individuals were screened, and 19 eligible individuals were randomised. Three dropped out before receiving any of the interventions and one dropped out after receiving only placebo. We included 15 individuals (33% female) in the per-protocol analysis with a mean age of 58 (SD 14) years, median urinary albumin creatinine ratio of 46 [IQR 21-58] mg/g and an eGFR of 73 (32) ml/min/1·73m2. The mean changes in renal cortical R2* from baseline to six hours were for dapagliflozin -1·1 (SD 0·7) s-1 and for placebo +1·3 (0·7) s-1, resulting in a difference between interventions of -2·3 s-1 [95% CI -4·0 to -0·6]; p=0·012. No between-intervention differences were found in any other MRI outcomes, physiological parameters or exploratory outcomes. There were no adverse events.

Conclusion: Dapagliflozin improved renal cortical R2* without changing renal perfusion or blood flow. This suggests improved renal cortical oxygenation due to a reduced tubular transport workload in the proximal tubules. Such improved oxygenation may in part explain the long-term beneficial renal effects seen with SGLT2 inhibitors.

Clinical Trial Registration Number: The study is registered in the EU Clinical Trials Register (EudraCT 2019-004557-92), on ClinicalTrials.gov (NCT04193566), and is completed.

Supported by: The study was funded by the Novo Nordisk Foundation grant PROTON personalising treatment of diabetic nephropathy (NNF14OC0013659).

Disclosure: J.C. Laursen: None.

OP 41 Building the evidence base for new devices

233

Cambridge hybrid closed-loop in very young children with type 1 diabetes: a multi-national 4-month randomised trial

J. Fuchs1,2, J.M. Allen1, C.K. Boughton1, M.E. Wilinska1,2, A. Thankamony2, C. de Beaufort3,4, E. Froehlich-Reiterer5, T.M. Kapellen6, B. Rami-Merhar7, S.E. Hofer8, F.M. Campbell9, J. Sibayan10, L. Bocchino10, R. Hovorka1,2, KidsAP Consortium;

1Wellcome Trust-MRC Institute of Metabolic Science, University of Cambridge, Cambridge, UK, 2Department of Paediatrics, University of Cambridge, Cambridge, UK, 3Clinique Pédiatrique, Centre Hospitalier de Luxembourg, Luxembourg, Luxembourg, 4Department of Paediatric Endocrinology, UZ-VUB, Brussels, Belgium, 5Department of Pediatric and Adolescent Medicine, Medical University of Graz, Graz, Austria, 6Hospital for Children and Adolescents, Leipzig, Germany, 7Department for Pediatrics and Adolescent Medicine, Medical University of Vienna, Vienna, Austria, 8Department of Pediatrics I, Medical University of Innsbruck, Innsbruck, Austria, 9Department of Paediatric Diabetes, Leeds Children's Hospital, Leeds, UK, 10Jaeb Centre for Health Research, Tampa, USA.

Background and aims: We assessed safety and efficacy of Cambridge hybrid closed-loop compared with sensor-augmented pump therapy over 16 weeks in very young children with type 1 diabetes (T1D).

Materials and methods: In a multi-national, multi-centre, randomised crossover trial, children aged 1 to 7 years with T1D on insulin pump therapy were recruited from seven centres across the UK, Austria, Luxembourg and Germany. Participants underwent two 16-week periods comparing the CamAPS FX hybrid closed-loop system (CL) with sensor-augmented pump therapy (control) in random order. The primary endpoint was the between group difference in time in target glucose range (3.9 to 10.0mmol/L) measured by sensor glucose levels. Key secondary endpoints were analysed hierarchically and included time in hyperglycaemia (>10.0mmol/L), glycated haemoglobin, mean sensor glucose, and time in hypoglycaemia (<3.9mmol/L). Analysis was by intention-to-treat.

Results: We randomised 74 participants: mean (±SD) age 5±2 years, 58% (n=43) male and baseline HbA1c 7.3±0.7%. The proportion of time glucose was in the target range was 8.7 percentage points (95% CI 7.4 to 9.9) higher during CL compared to control period (p<0.001). Time with glucose >10.0mmol/L was 8.5 percentage points lower (95% CI 7.1 to 9.9; p<0.001) during CL than during control period. Mean glucose was 0.7mmol/L lower (95% CI 0.5 to 0.8; p<0.001), and glycated haemoglobin 0.4% lower (95% CI 0.3 to 0.5; p<0.001) with CL than with control therapy. Time in hypoglycaemia (<3.9mmol/L) was similar between CL and control periods (p=0.74). Mean closed-loop usage was 93±8% over 16 weeks. One severe hypoglycaemia event occurred in CL and none in the control period. There were no other serious adverse events.

Conclusion: The Cambridge hybrid closed-loop system is safe and significantly improves glycaemic control in very young children with T1D, without increasing time in hypoglycaemia.

figureby

Clinical Trial Registration Number: NCT03784027

Supported by: EC Horizon 2020, JDRF, NIHR Cambridge BRC, Wellcome Strategic Award, Dexcom

Disclosure: J. Fuchs: Grants; European Commission Horizon 2020 (731560), JDRF, National Institute for Health Research Cambridge Biomedical Research Centre, Wellcome Strategic Award (100574/Z/12/Z). Other; Dexcom supplied discounted CGM devices.

234

Cambridge hybrid closed-loop in children and adolescents with type 1 diabetes: a multicentre 6-month randomised trial

C.K. Boughton1, L. Kanapka2, R. Hovorka1, DAN05 Consortium;

1Wellcome Trust-MRC Institute of Metabolic Science, University of Cambridge, Cambridge, UK, 2The Jaeb Center for Health Research, Tampa, USA.

Background and aims: We compared safety and efficacy of Cambridge hybrid closed-loop algorithm with usual care over 6 months in children and adolescents with type 1 diabetes.

Materials and methods: In an open-label multicentre, multinational, single-period, parallel randomised controlled trial, participants aged 6-18 years using insulin pump therapy were assigned to closed-loop insulin delivery or usual care (control) for 6 months. We used the Cambridge algorithm in two consecutive hardware iterations, FlorenceM (modified Medtronic 640G pump, Medtronic Guardian 3 sensor, algorithm on locked-down smartphone) or CamAPS FX (Sooil Dana RS pump, Dexcom G6 sensor, algorithm on unlocked smartphone). Primary endpoint was HbA1c at 6 months.

Results: We randomised 133 participants, 65 to closed-loop and 68 to control (mean±SD baseline HbA1c 8.2±0.7% vs 8.3±0.8%). At 6 months mean HbA1c was 0.32% lower with closed-loop compared to control (95%CI 0.04 to 0.59; p=0.02). Closed-loop usage was low and variable (40% [26, 53]; median [IQR]) with FlorenceM due to unreliable hardware, and consistently high (93% [88, 96]) with CamAPS FX. In the cohort using CamAPS FX (n=21), HbA1c was 1.05% lower (95%CI 0.67 to 1.43; p<0.0001) and time in target range 3.9-10.0mmol/L was 15.0 percentage points higher (95%CI 8.0 to 22.1; p=0.0001) compared to control (n=25), without an increase in hypoglycaemia (p=0.15). Severe hypoglycaemia and DKA events were similar between closed-loop and control.

Conclusion: Cambridge hybrid closed-loop algorithm is safe and significantly improves glycaemic control in children and adolescents with type 1 diabetes. Efficacy relies on consistently high usage of closed-loop as demonstrated with CamAPS FX.

figurebz

Clinical Trial Registration Number: NCT02925299

Supported by: National Institutes of Health and National Institutes of Diabetes, Digestive and Kidney Diseases (UC4 DK108520)

Disclosure: C.K. Boughton: None.

235

Effect of intermittent-scanning CGM to glycaemic control including hypoglycaemia and quality of life of patients with type 1 diabetes (ISCHIA study)

T. Murata1, N. Sakane2, A. Kuroda3, J. Miura4, Y. Hirota5, K. Kato6, M. Toyoda7, R. Kouyama8, K. Kouyama9, A. Shimada10, S. Kawashima11, Y. Matoba12, K. Nishimura13, K. Hosoda14, The ISCHIA study group;

1Department of Clinical Nutrition, NHO Kyoto Medical Center, Kyoto, 2Division of Preventive Medicine, NHO Kyoto Medical Center, Kyoto, 3Diabetes Therapeutics and Research Center, Tokushima University, Tokushima, 4Diabetes Center, Tokyo Women’s Medical University, Tokyo, 5Division of Diabetes and Endocrinology, Kobe University, Kobe, 6Diabetes Center, NHO Osaka National Hospital, Osaka, 7Division of Nephrology, Endocrinology and Metabolism, Tokai University, Isehara, 8Division of Endocrinology and Metabolism, Tsuchiura Kyodo General Hospital, Tsuchiura, 9Department of Diabetes and Metabolism, NHO Hyogo-Chuo National Hospital, Sanda, 10Department of Endocrinology and Diabetes, Saitama Medical University, Moroyama, 11Kanda Naika Clinic, Osaka, 12Department of Diabetes, Endocrinology and Metabolism, NHO Kokura Medical Center, Kitakyushu, 13Department of Preventive Medicine, NCVC, Suita, 14Division of Diabetes and Lipid Metabolism, NCVC, Suita, Japan.

Background and aims: To investigate the effect of intermittent-scanning CGM (isCGM) combined with structured education focused on the frequent scanning of the sensor (10 times/day or more) and the utilization of the trend arrow to prevent hypoglycemia in patients with type 1 diabetes mellitus, we conducted a crossover study comparing isCGM (Intervention) and the self-monitoring of blood glucose (SMBG) accompanied by masked retrospective CGM (Control).

Materials and methods: The ISCHIA study was a randomized crossover study of type 1 diabetes mellitus patients who used multiple daily injections. The study included a 28-day run-in period, an 84-day sequence 1 (Intervention or Control), a 28-day washout period, and an 84-day sequence 2 (Control or Intervention). Participants continued SMBG three times per day or more throughout the study period in order to comply with the approved labeling of isCGM in Japan that demands adjunctive use of isCGM to SMBG. The primary endpoint was the decrease in the time spent in hypoglycemia (<70 mg/dL) per day.

Results: Out of 104 participants, 102 participants were randomized and 93 participants completed the study. The mean age of the participants was 51.4±15.3 years, 52.7% of the participants were female, the mean HbA1c was 7.3±0.7%, and the rate of isCGM-naïve participants was 46.2%. The time spent in hypoglycemia per day was significantly reduced in the Intervention arm (2.42±1.68 hour/day) compared to the Control arm (3.10±2.28 hour/day) (P = 0.012). Time-in-range (TIR [70-180 mg/dl]) was not significantly different between the two arms (14.54±2.66 hour/day vs. 13.75±2.45 hour/day). Time-above-range (>180 mg/dl) was 7.03±3.12 hour/day vs. 7.15±3.50 hour/day, respectively. The change of glycoalbumin (GA) was not significantly different between the two arms (0.3±2.3% vs. 0.1±2.3%). Frequency of severe hypoglycemia (SH) was not significantly different between the two arms (2.1% vs. 6.5%). No severe adverse events related to the study excluding SH occurred. Frequency of non-severe adverse events related to the study was not significantly different between the two arms (17.0% vs. 17.4%).

Conclusion: The usage of isCGM combined with structured education significantly reduced the time spent in hypoglycemia without the deterioration of TIR or GA.

Clinical Trial Registration Number: jRCT1052180075

Supported by: AMED, Japan IDDM Network

Disclosure: T. Murata: None.

236

Evaluation of accuracy and safety of the next generation 180-day long-term implantable Eversense CGM system: the PROMISE Study

S. Garg1, D. Liljenquist2, B. Bode3, M. Christiansen4, T. Bailey5, R. Brazg6, D. Denham7, A.R. Chang8, H. Akturk1, A. Dehennis9, K.S. Tweden9, F.R. Kaufman9;

1University of Colorado, Aurora, 2Rocky Mountain Diabetes Center, Idaho Falls, 3Atlanta Diabetes Associates, Atlanta, 4Diablo Clinical Research, Walnut Creek, 5AMCR Institute, Escondido, 6Rainier Clinical Research Center, Renton, 7Clinical Trials of Texas Inc, San Antonio, 8John Muir Health, Concord, 9Senseonics, Germantown, USA.

Background and aims: Use of Continuous Glucose Monitoring Systems (CGM) is being rapidly adopted as standard of care for insulin-requiring patients with diabetes. The PROMISE study was designed to evaluate the accuracy and safety of the next generation implantable Eversense CGM System for up to 180 days.

Materials and methods: The study was prospective, multi-center, unblinded, involving 181 subjects with T1D or T2D at 8 sites in the United States. Accuracy was evaluated by comparing CGM to reference values from a YSI 2300 glucose analyzer (YSI) during 10 clinic visits from day 1-180, which included hyperglycemia and hypoglycemia challenges. Eighty-five (85) subjects had 1 sensor and ninety-six (96) subjects had 2 sensors inserted in the arm. Counting 2 replaced sensors, there were a total of 279 sensors inserted (558 insertions/removals). The CGM prompts 2 calibrations/day until day 21, after it prompts 1 calibration/day. The incidences of device-related and sensor insertion/removal procedure-related serious adverse events up to 180 days post-insertion were evaluated. Precision between two implanted sensors was measured by paired ARD (PARD) and percent coefficient of variation (PCV).

Results: The accuracy analyses were based on 49,613 matched pairs over 180 days. Percent CGM readings within 15/15% and 20/20% of YSI values were 85.6% and 92.9%, respectively, and the overall MARD was 9.1%. MARDs across different glucose ranges from 40-400 mg/dL (2.2-22.2 mmol/L) were <9.4% (Table). The MARD across individual visit days between days 1-180 were <11.0% with highest MARD of 11.0% (day 1) and lower MARDs of 8.4% (day 30), 7.7% (day 60) and 8.2% (day 90). The CGM System readings within 20/20% by day were 89% (day 1), 91% (day 7), 92% (day 14), 94% (day 22), 95% (day 30), 96% (day 60), 94% (day 90), 93% (day 120), 93% (day 150), and 90% (day 180). The evaluation for precision yielded a PARD of 10.1% and PCV of 7.1%, demonstrating high precision between the two sensors in an individual patient. The confirmed hypoglycemia detection rates at 70 mg/dL (3.8 mmol/L) and 60 mg/dL (3.3 mmol/L) were 93% and 87%, respectively. Hyperglycemia detection rate at 180 mg/dL (10 mmol/L) was 99%. There were no device or insertion/removal procedure-related serious adverse events. Two mild skin infections occurred. All sensors were removed during the initial removal procedure.

Conclusion: We conclude that next generation Eversense long-term implantable CGM System has sustained accuracy and safety up to 180 days with primarily one calibration/day.

figureca

Clinical Trial Registration Number: NCT03808376

Supported by: The study was funded by Senseonics, Inc

Disclosure: S. Garg: None.

OP 42 Cardiovascular disease: predictors and outcomes

237

Respective predictive value of glycation gap and silent myocardial ischaemia for cardio-vascular events in asymptomatic patients with type 2 diabetes

P. Valensi, M. Nguyen, I. Banu, S. Chiheb, E. Cosson;

Jean Verdier hospital, Bondy, France.

Background and aims: Protein glycation is involved in diabetic complications. It can be evaluated by the glycation gap (GG), difference between glycation in the erythrocyte (HbA1c) and the plasma compartments (fructosamine). In a diabetic population, an association has been reported between GG and history of cardio-vascular events (present in 30% of the population) and mortality. The present study aimed to analyse the predictive value of GG for major cardiac (MCE) or cardio-vascular (MCVE) events in patients with type 2 diabetes (T2D) free of cardio-vascular history or symptom who had been assessed for silent myocardial ischemia (SMI).

Materials and methods: We included 741 T2D patients who had concomitant HbA1c and fructosamine measurements. Using the strong correlation between HbA1c and fructosamine we calculated predicted HbA1c and GG (= measured HbA1c - predicted HbA1c). The patients had a stress myocardial scintigraphy to detect SMI, with a coronary angiography in those with SMI.

Results: SMI was present in 221 patients, 83 of them had significant coronary stenoses (CS) on angiography. Among the 695 patients who could be followed (during 5.4±3.6 years), a MCE (acute coronary syndrome, congestive heart failure, secondary coronary revascularization or cardiac death) occurred in 65 patients, and a MCVE (MCE or amputation or peripheral revascularization) in 107 patients. The patients were separated in 2 groups: GG ≥1% (n=154) or <1% (n=587). The prevalence of SMI and CS, and the incidence of MCE (12.6% and 8.5% respectively; p=0.147) and MCVE (18.9% and 14.5%, p=0.195) did not differ significantly between the 2 groups. The incidence of MCE and MCVE was higher in the patients with SMI compared to those without (16.7% vs 6.2% and 26.2% vs 10.7%; p<0.0001 for both comparisons). In Cox logistic regression, SMI (OR 2.3[1.4-3.8], p<0.001) but not GG predicted MCE and MCVE.

Conclusion: These results do not support the predictive value of GG for cardio-vascular events in T2D patients free of cardio-vascular history but confirm the good predictive value of SMI.

Disclosure: P. Valensi: None.

238

Validation study for coronary heart disease (CHD) among diabetes patients based on automatic retinal image analysis (ARIA) method

D.R. Owens1, J. Lee2, Y. Qu2, Y. Zhou3, L. Ding3, S. Liu4, W. Yuan3, M. Chen4, J. Zhang3, J.M. Rafferty5, R.L. Thomas1, B. Zee2;

1Diabetes Research Group, Swansea University, Swansea, UK, 2Centre for Clinical Research and Biostatistics, Chinese University of Hong Kong, Hong Kong, China, 3Department of Acupuncture and Moxibustion, Shenzhen Traditional Chinese Medicine Hospital, Shenzhen, China, 4Department of Cardiovascular disease, Shenzhen Traditional Chinese Medicine Hospital, Shenzhen, China, 5Secure Anonymised Information Linkage databank, Swansea University, Swansea, UK.

Background and aims: Coronary heart disease (CHD) is common age-related disease that constitute a growing health and social burden, especially for long-term management of diabetes patients. Although the existing diagnostic tools for this disease is reliable, their complexity and high cost hinder their use for screening the general population, while risk engines based on traditional risk factors have shown poor performance. Thus, there is an urgent need for new, simple and inexpensive high-accuracy techniques for use as complementary screening tools. In this study, we employed data from Hong Kong for training a risk assessment model for coronary heart disease, and evaluate if retinal image analysis can be used as risk assessment tool for diabetes complications management in UK.

Materials and methods: Patients in the training data were enrolled from the Cardiology Department of a Traditional Chinese Medicine Hospital (TCM) from December 2017 to September 2019. Baseline demographic and medical information were collected from participants’ hospital medical records. Retinal images of both eyes of each participant were taken within six months of admission. A total of 388 subjects were included in this study. Among them, 272 subjects were used for training of the classification model. They included 132 CHD patients obtained from TCM, and 140 controls obtained from a lifestyle study with healthy individuals. A set of 116 subjects with 56 CHD and 60 controls were used as testing data. All subjects have their retinal images taken by a non-mydriatic fundus camera (Canon CR2). A fully automatic retinal image analysis for CHD risk (ARIA-CHD) was developed to estimate retinal microvascular characteristics and incorporate machine-learning technique to derive an overall estimation of CHD risk. An external testing data include 43 CHD patients from UK and 61 controls from the lifestyle study.

Results: The training analysis has very high accuracy with sensitivity and specificity of 92.9% and 91.7% respectively. The 10-fold cross validation using support vector machine (SVM) approach has sensitivity of 91.5% and specificity of 88.0%. A validation analysis using 61 control subjects from CCRB, 56 CHD data from TCM, and 43 CHD from UK achieved high accuracy in classification with mean CHD probabilities of 0.28 (95% CI of 0.24, 0.32), 0.73 (95% CI of 0.69, 0.77), and 0.73 (95% CI of 0.69, 0.76) in the control group and CHD groups of the lifestyle study and UK respectively. The 95% confidence intervals for two CHD groups were overlapping but significantly different from control group, indicating good performance in classification.

Conclusion: Method to estimate CHD risk based on retinal images has been developed and validated. ARIA-CHD is a fast, convenient and non-invasive alternative for the community pre-screening program for CHD. The retinal images in UK have similar accuracy for CHD detection using the model developed by retinal images from Hong Kong and TCM.

Disclosure: D.R. Owens: None.

239

Cardiovascular and renal diseases in type 1 compared with type 2 diabetes: a nationwide observational study

D. Angoulvant, G. Fauchier, C. Semaan, A. Bisson, J. Herbert, P. Ducluzeau, L. Fauchier;

Centre Hospitalier Universitaire de Tours, Tours, France.

Background and aims: Type 1 diabetes (T1D) and type 2 diabetes (T2D) increase risks of cardiovascular (CV) and renal disease compared with diabetes-free populations. There are only few studies comparing T1D and T2D for the risk of these clinical events. We examined these issues in a nationwide analysis in France.

Materials and methods: All patients aged >=18 seen in French hospitals in 2013 with at least 5 years of follow-up were identified and categorized by their diabetes status. A total of 50,623 patients with T1D (age 61.4±18.6, 53% male) and 425,207 patients with T2D (age 68.6±14.3, 55% male) were followed over a mean period of 4.3±2.1 years (median 5.3, interquartile 2.8-5.8 years). Prevalence and event rates of myocardial infarction (MI), heart failure (HF), ischemic stroke, chronic kidney disease (CKD), all-cause death and CV death were assessed with age stratification of 10-year intervals. Cox regression analyses were used to estimate risk with adjustment on sex and age.

Results: The age and sex-adjusted prevalence of CV diseases was higher in T2D for ages above 40 years whereas the adjusted prevalence of CKD was more common in T1D between ages 18 and 69 years and higher in T2D for ages above 80 years. During 2,033,239 person-years of follow-up, there were 27,497 patients with MIs (yearly rate 1.4%), 24,892 with ischemic strokes (yearly rate 1.2%), 100,769 with incident HF (yearly rate 5.4%), 65,928 with incident CKD (yearly rate 3.4%) and 197,858 deaths (yearly rate 9.7%) including 49,026 CV deaths (yearly rate 2.4%) were recorded. Age and sex-adjusted event rates comparing T1D versus T2D showed that MI risk was increased for ages above 60 (1.2-fold for T1D versus T2D) and HF between ages 18-29 and above 60 years (1.1-1.4-fold). Adjusted risk of ischemic stroke did not markedly differ between T1D and T2D. Risk of incident CKD was 1.1-2.4-fold higher in T1D between ages 18-49 and above 60 years. The all-cause death risk was 1.1-fold higher in T1D at age >=60 years, the cardiovascular death risk being 1.1-fold higher in T1D between 60 and 69 years.

Conclusion: The adjusted prevalent burden and risk of incident renal disease are greater among patients with T1D compared with T2D patients across most ages. Although the prevalent burden of cardiovascular diseases may be lower in T1D than in T2D, patients with T1D may have a higher risk of incident MI, HF, all-cause death and cardiovascular death at middle-older ages, highlighting their need for improved prevention.

Disclosure: D. Angoulvant: None.

240

A longitudinal study on the incidence of cardiovascular events in a population of Northern Italy

E. Bianco1,2, M. Dei Cas1, M. Bignotto1, C. Morano1, C. Rigoldi1,2, G. Trevisi1,2, C. Berra3, P. Zermiani1, M. Zuin1,2, R. Paroni1, F. Folli1,2, P. Battezzati1,2;

1Department of Health Sciences, Università degli Studi di Milano, Milano, 2ASST Santi Paolo e Carlo, Milano, 3IRCCS MultiMedica, Milano, Italy.

Background and aims: The CA.ME.LI.A (CArdiovascular risks, MEtabolic syndrome, Liver, and Autoimmune disease) study is a project promoted by University of Milan in 2009-2011, aimed to identify risk factors for cardiovascular (CV), metabolic and liver diseases in a population representative of the Northern Italy. Here we aim to assess whether impaired glucose metabolism (IGM) and/or overweight/obesity (OWO) are independently associated to an increased CV risk.

Materials and methods: Time 0 was the enrolling in the study; the observation ended on August 30th, 2017, or at the first CV event (recovery or death), or with the loss of the subject to the follow-up. The study of the incidence of events was conducted through the Kaplan-Meier method using log rank test. A univariate analysis of CV events and continuous variables was conducted with Cox time-dependent models to a single variable. The variables that, to the univariate analyses carried out by means of the log rank test or of a time-dependent model of Cox to a single variable (continuous variables), were significantly associated with the incidence of CV events, were introduced in a multivariate time-dependent Cox model to identify those that had an independent prognostic value. The population (n=2545, 1257 men) was stratified into 6 categories: 1. Normal glucose tolerance (NGT) (FG<110 mg/dL)/ BMI≤24.9 kg/m2 (Normal weight, NW); 2. NGT/ BMI ≥25 kg/m2 (overweight or obese, OWO); 3. Impaired fasting glucose (IFG) (FG 110-125 mg/dL)/NW; 4. IFG/OWO; 5. Diabetes (DM) (FG≥126 mg/dL)/NW; 6. DM/OWO.

Results: During the follow-up, 163 CV events occurred with an incidence rate of about 1.5 cases per 100 year-patient; men had a higher number of events as compared to women (106 vs 57; 4.2% vs 2.2%, p=0.001, Fig.1). DM had a grater incidence of events compared both to IFG and NGT (IFG 13% vs DM 21%, p=0.02; NGT 5%, p<0.0001). Even OWO was associated with higher CV risk (NW 3% vs OWO 10%, p<0.0001). When a IGM coexisted with OWO, the incidence of CV events further increased, with DM/OWO being higher than in IFG/OWO (24% vs 15% p=0,0001); in NGT/NW the incident was 3%. Considering the association between IGM and OWO, we found that there wasn’t difference in the number of events between only DM and DM/OWO (p=0.09). Instead, there was a difference in CV events between OWO and DM/OWO (p<0.0001) and between OWO and IFG/OWO (p=0.0016).

Conclusion: Survival curves shows that in both sexes OWO, IFG and DM show higher mortality than NW and NGT. Among OWO patients, the concomitant presence of diabetes leads to a higher risk of events or death: DM/OWO have a higher incidence of CV events than only OWO. Also IFG increases mortality compared to NGT subjects as well as IFG/OWO have significantly more CV events than OWO only. These data suggest that not only DM but also IFG subjects need to be treated to prevent CV.

figurecb

Supported by: Regione Lombardia, MIUR, Dipartimento di Scienze della Salute dell’Università degli Studi di Milano

Disclosure: E. Bianco: None.

OP 43 Genes, epigenes and telomeres in type 1 diabetes

241

Birth weight, BMI in adulthood and latent autoimmune diabetes in adults: a Mendelian randomisation study

Y. Wei1, Y. Zhan1, J.E. Löfvenborg1, T. Tuomi2,3, S. Carlsson1;

1Karolinska Institutet, Stockholm, Sweden, 2Helsinki University and University Hospital, Folkhalsan Research Center, Helsinki, Finland, 3Lund University, Malmö, Sweden.

Background and aims: Latent autoimmune diabetes in adults (LADA) is a hybrid form of diabetes with adult onset and a pathogenesis that includes both type 1-like and type 2-like features. Observational studies found that both lower birth weight and higher adult BMI were associated with an increased risk of LADA. The present study aimed to revisit these findings using two-sample Mendelian randomization (MR).

Materials and methods: Forty-four independent SNPs of genome-wide significance were selected as instruments for birth weight from a genome-wide association (GWAS) study including individuals of European descent from UK Biobank and 30 other studies. 820 independent SNPs were identified as instruments for BMI in adulthood from a GWAS study including individuals of European ancestry from UK Biobank and the GIANT (Genetic Investigation of ANthropometric Traits) study. The selected instruments explained 1.62% variance in birth weight and 7.43% variance in adult BMI. Summary statistics for the SNP-LADA associations were extracted from the GWAS study conducted in 2634 LADA cases and 5947 population controls of European ancestry. Inverse-variance weighted (IVW) method was used to examine the causal association, supplemented by several sensitivity analyses, namely robust IVW, weighted median, MR-Egger, MR-PRESSO (Mendelian randomization pleiotropy residual sum and outlier test), and a conservative analysis excluding hypothetically pleiotropic SNPs.

Results: One SD decrease in genetically predicted birth weight was associated with a 69% higher risk of LADA (OR: 1.69, 95% CI: 1.13-2.53). All sensitivity analyses showed the same direction of association as found in the main analysis. MR-Egger (P for directional pleiotropy: 0.119) and MR-PRESSO (P for global pleiotropy: 0.173) indicated no directional pleiotropy. One SD increase in genetically predicted BMI in adulthood was associated with a 40% (OR: 1.40, 95% CI: 1.14-1.71) higher risk of LADA, and sensitivity analyses also showed similar risk estimates. MR-PRESSO detected two outliers for adulthood BMI but detected no distortion in risk estimates by the outliers.

Conclusion: These findings provide genetic support for a causal link between low birth weight and adult overweight in relation to LADA.

figurecc

Supported by: CSC, VR, FORTE, and Novo Nordisk Foundation

Disclosure: Y. Wei: None.

242

The influence of HLA haplotype, number of injections and administration route on the effect of GAD-specific immunotherapy in type 1 diabetes

U. Hannelius1, C. Beam2, J. Ludvigsson3;

1Diamyd Medical, Stockholm, Sweden, 2Department of Biomedical Sciences, Western Michigan University Homer Stryker M.D. School of Medicine, Kalamzoo, USA, 3Region Östergötland and Division of Pediatrics, Linköping University, Crown Princess Victoria Children´s Hospital, Linköping, Sweden.

Background and aims: Genetic risk for T1D is largely associated with variations in HLA genes as is individual disease pathology. In line with this, we have previously shown that immunomodulatory treatment with GAD-alum has a positive effect specifically in individuals carrying the high-risk HLA haplotype DR3-DQ2. In this study we sought to further investigate the influence of HLA, number of injections and administration route on the clinical effect of GAD-alum treatment in preserving endogenous insulin production in individuals with recent-onset T1D.

Materials and methods: We combined individual-level data (n=627) from four placebo controlled randomized GAD-alum clinical trials in individuals recently diagnosed with type 1 diabetes. We estimated the treatment effect at 15 months from baseline on C-peptide retention and HbA1c, using mixed model repeated measures including terms for subgroups defined by HLA haplotypes DR3-DQ2 and DR4-DQ8 as well as number of doses. The effect of administration route (subcutaneous vs intralymphatic) was evaluated using a Bayes method.

Results: Analysis showed a treatment effect ratio compared to placebo of 1.18 in the full population (P=0.001), 1.36 in individuals carrying HLA DR3-DQ2 (n=313; 95% CI 1.18, 1.57; P<0.0001), and 1.57 in individuals carrying HLA DR3-DQ2 but not DR4-DQ8 (n=149; 95% CI 1.30, 1.89; P<0.0001). The effect was highest in subjects receiving a higher number of injections, three or four, with an estimated treatment effect ratio of 1.48 (95% CI 1.26,1.74; P<0.0001) in individuals carrying DR3-DQ2, and 1.77 in individuals carrying DR3-DQ2 but not DR4-DQ8 (95% CI 1.43, 2.19; P<0.0001). GAD-alum treatment also had a significant effect on lowering HbA1c compared to placebo, again highest in subjects receiving a higher number of doses where an effect of -4.79 mmol/mol was seen in individuals carrying the DR3-DQ2 haplotype (95% CI -8.08, -1.50; P<0.01) and -6.84 mmol/mol individuals carrying DR3-DQ2 but not DR4-DQ8 (95% CI -11.71, - 1.97, P<0.01). There was a 98% and 99% probability that three intralymphatic injections were superior to three subcutaneous injections regarding effects on C-peptide retention and HbA1crespectively.

Conclusion: These analyses highlight the importance of genetics, dosing regimen and route in immunotherapeutic treatment of type 1 diabetes and a clinically relevant potential for using intralymphatic GAD-alum.

figurecd

Disclosure: U. Hannelius: Employment/Consultancy; I am employed by Diamyd Medical that develops GAD-alum. Stock/Shareholding; I own shares in Diamyd Medical that develops GAD-alum.

243

Epigenome-wide association study identifies differentially methylated DNA sites for diabetic kidney disease in type 1 diabetes

E.H. Dahlstrom1,2, L.J. Smyth3, A. Syreeni1,2, J. Killner3, G.J. McKay3, N. Sandholm1,2, C. Forsblom1,2, A. Maxwell3, A. McKnight3, P.-H. Groop1,4;

1Folkhälsan Research Center, Helsinki, Finland, 2Research Program for Clinical and Molecular Metabolism, University of Helsinki and Helsinki University Hospital, Helsinki, Finland, 3Molecular Epidemiology Research Group, Queen’s University Belfast, UK, 4Department of Nephrology University of Helsinki and Helsinki University Hospital, Helsinki, Finland.

Background and aims: Epigenetic modification such as DNA methylation impact the transcription of genes and it can have long-term effects on disease. The aim of this study was to identify differentially methylated CpG sites (dmCpGs) in whole-blood DNA, associated with diabetic kidney disease (DKD) in individuals with type 1 diabetes.

Materials and methods: This study included 651 individuals with DKD and 651 individuals with long-duration type 1 diabetes and no evidence of DKD matched for age, sex and diabetes duration from the Finnish Diabetic Nephropathy study (FinnDiane, n=800) and the United Kingdom-Republic of Ireland (UK-ROI, n=502). An epigenome-wide analysis of 862,927 CpGs sites assessed with the Illumina MethylationEPIC BeadChip was performed using the RnBeads and Bioconductor R packages for each cohort separately. Results from both cohorts were meta-analysed based on p-values and weighted by sample-size using the METAL software.

Results: Altogether six hypermethylated and 25 hypomethylated sites were associated (PFDRadj≤1.0x10-8) with DKD after adjusting for age, sex, and cell-type heterogeneity (Figure 1). Of these, cg17944885 (PFDRadj =2.0x10-44) and cg25544931 (PFDRadj =2.3x10-37), located between ZNF788 and ZNF20 on chromosome 19, were the most significant hypermethylated sites across both cohorts. This site has previously been associated with kidney disease, also in diabetes. Strongest hypomethylation was identified for dmCpGs in LOC101927438 (cg05710777, PFDRadj=3.0x10-18), ROD1 (cg00008629, PFDRadj =5.4x10-15) and NME7 (cg08150816, PFDRadj =8.2x10-15), of which none have previously been associated with kidney disease. Further adjustment for smoking status reduced the number of epigenome-wide significant dmCpGs to 18, but the top-ranked dmCpGs remained. After additional adjustment for diabetes duration, BMI, HbA1c, HDL cholesterol and triglycerides, seven dmCpGs remained (FDRadjp≤1.0x10-8) across both cohorts. The top-ranked dmCpGs located in genes including ROD1NME7SLC27A3SLC1A5 and LOC101927438 remained consistently associated across each of the analyses.

Conclusion: We find that whole blood DNA methylation in individuals with type 1 diabetes are significantly associated with DKD at several dmCpGs, even after adjustment for major confounders, with the majority demonstrating a hypomethylation profile.

figurece

Supported by: NIH (1R01DK105154-01A1) GENIE II

Disclosure: E.H. Dahlstrom: None.

244

Telomeres do not always shorten over time in people with type 1 diabetes: a FinnDiane substudy

A. Syreeni1,2, L. Carroll3, S. Mutter1,2, A.S. Januszewski3, C. Forsblom1,4, M. Lehto1,2, P.-H. Groop2,4, A.J. Jenkins3;

1Folkhälsan Research Center, Helsinki, Finland, 2Research Program for Clinical and Molecular Metabolism, Faculty of Medicine, University of Helsinki, Finland, 3NHMRC Clinical Trial Centre, The University of Sydney, Sydney, Australia, 4Department of Nephrology, University of Helsinki and Helsinki University Hospital, Helsinki, Finland.

Background and aims: Telomeres are repetitive sequences at chromosome ends that maintain chromosome integrity. Shorter telomeres in white blood cells (WBC) are associated with, and predictive of, poor health outcomes, including in relatively small studies in people with type 1 diabetes (T1D). Both genetic and acquired factors modulate telomere length (TL) and their shortening rate. There are cellular mechanisms to maintain and even increase TL. Our aim was to characterise changes in telomere length in adults with T1D over time, and the associated clinical variables.

Materials and methods: Adults with T1D in the Finnish Diabetic Nephropathy (FinnDiane) cohort (n=618) with WBC DNA from two time-points (median [range] follow-up 6.8 [2.8-17.7] years were studied. Relative TL (rTL) was measured by quantitative PCR, with intra- and inter-assay CVs of 1.8% and 4.0%, respectively. Both samples (in triplicate) from a subject were on the same plate. rTL change was calculated as % change per annum (p.a.) and a total change of <4% (assay CV) was considered stable. Central obesity was defined as waist-height-ratio ≥0.5. Albuminuria was defined as AER ≥30 mg/24 h in two out of three consecutive urine collections.

Results: At baseline, mean (SD) age and T1D duration was 36.7 (12.1) and 22.2 (11.6) years, 49.2% were male, and 23.7% had albuminuria. Baseline rTL correlated inversely with age (r= -0.093, p=0.021) and was shorter in men vs. women ( -0.38 vs. -0.094, p=0.0011, age-adjusted). Comparing the 1st and 3rd baseline rTL tertile, the 1st (shortest rTL) tertile was older (+3.5 years), had more men (+19.9%), and individuals with albuminuria (+9.2%). In adjusted models (sex, age), the shortest rTL group was prescribed more antihypertensive medication (AHT, 38.5 vs. 23.8%, p=0.019), had higher BMI (25.8 vs. 24.6 kg/m2, p=0.003), more central obesity (55.8 vs. 33.0%, p=2.8×10-4), lower eGFR (81.4 vs. 90.4 mL/min/1.73 m2, p=0.004), but no difference in HbA1c or frequency of albuminuria. Baseline rTL was a strong determinant of rTL change (r=-0.66, p=5.4×10-78). Over time, telomeres shortened in 342 (55.3%) subjects, lengthened in 247 (40.0%) individuals and remained stable in 4.7%. In the 1st vs. 2nd plus 3rd baseline rTL tertiles, rTL %-change was median (IQR) +5.3 (-0.9, 15.6) % p.a. vs. -3.3 (-7.3, 0.6) % p.a., p=3.0×10-21. Higher baseline BMI and central obesity were associated (linear) with increasing rTL after adjustment for sex and age. Association with BMI and rTL change remained significant after adjusting for baseline rTL (Beta=0.24, p=0.030). Those in whom rTL increased (vs. those whom rTL decreased) were more likely to start AHT during follow-up (24.5 vs. 17.4%, p=0.047, adjusted for sex, age, and baseline AHT). HbA1c and change in albuminuria were not associated with TL change.

Conclusion: In a T1D longitudinal study, rTL inversely correlated with age, and over median 6.8 years, did not shorten in 44.7%. Associates of change in TL over time were higher baseline BMI, central obesity, rTL and the start of AHT during follow-up. The biology of these relationships and clinical significance of TL change over time in people with T1D merit further investigation.

Supported by: This study was supported by Albert Reynold Travel Fellowship (LC) and Australian NHMRC Practitioner Fellowship (AJJ).

Disclosure: A. Syreeni: None.

OP 44 Diabetes around the clock!

245

The influence of residual beta cell function upon free-living postprandial and nocturnal glycaemic control in individuals with type 1 diabetes

A.C. Shaw1, G.S. Taylor1, T.J. McDonald2, R.A. Oram2, J.A.M. Shaw3, D.J. West1;

1Population Health Sciences Institute, Newcastle Upon Tyne, 2Royal Devon and Exeter NHS Foundation Trust, Exeter, 3Translational and Clinical Research Institute, Newcastle Upon Tyne, UK.

Background and aims: Residual β-cell function can persist many years after diagnosis of type 1 diabetes (T1D), indicated by low concentrations of stimulated endogenous insulin and C-peptide. Residual β-cell function is associated with improved free-living glycaemic control, measured by continuous glucose monitoring (CGM). However, it is unclear which glucose excursions throughout the day are improved by retaining low levels of functional β-cells. We explored whether β-cell function affects nocturnal and postprandial glycaemic control.

Materials and methods: Individuals with T1D >1 year (n=66, M/F: 35/31, age: 41 ± 12, diabetes duration: 21 ± 13 years, HbA1c: 58 ± 10 mmol/mol, BMI: 25 ± 3 kg/m2) completed a postprandial urine C-peptide creatinine ratio test (UCPCR), wore a CGM, and kept a food and insulin diary for 7 days in free-living conditions. Participants were split by UCPCR into groups: undetectable (Cpepund [UCPCR < 0.001 nmol/mmol], n = 36), low (Cpeplow [0.001 - 0.19], n = 13), and high (Cpephigh [≥ 0.2], n = 19). CGM outcomes were compared between groups using one-way ANOVA for 24hrs (00:00-00:00 hrs), nocturnal (00:00-06:00 hrs), and postprandial (0-300 mins) periods.

Results: Data are presented in Table 1 as means ± SD (significance accepted at p ≤ 0.05). 24hrs: Cpephigh spent more time in euglycaemia (3.9-10mmol/L) (72 ± 17%) than Cpeplow (62 ± 13%, p = 0.2), and significantly more time than Cpepund (58± 20%, p = 0.028) overall. Cpephigh spent significantly less time in hyperglycaemia (>13.9mmol/l) than Cpepund (p < 0.001). SD glucose was significantly lower in Cpephigh than Cpepund (p <0.001). Nocturnal: Cpephigh spent significantly more time in euglycaemia (76 ± 20%) than Cpeplow (60 ± 20%, p = 0.025) and Cpepund (58 ± 20%, p = 0.002) overnight. Cpephigh spent significantly less time in hypoglycaemia (<3.9mmol/l) (p = 0.023) and hyperglycaemia (10-13.9mmol/l) (p = 0.036) and had lower mean and SD glucose than CpepundPostprandial: Cpephigh spent more time in euglycaemia (68 ± 22%) than Cpeplow (53 ± 23%, p = 0.2), and significantly more time than Cpepund (51± 22%, p = 0.045) postprandially. There were no group differences in mean, SD, or CV glucose (p > 0.05).

Conclusion: T1D with higher residual β-cell function demonstrate improved free-living glycaemic control, specifically with greater protection from nocturnal and postprandial dysglycaemia. Further research, patient education, and clinical management should focus on improving nocturnal and postprandial glycaemia. This is of particular importance in those with no residual β-cell function.

figurecf

Supported by: Award to DW by the DRWF

Disclosure: A.C. Shaw: None.

246

Diurnal rhythms in the human skeletal muscle metabolome are altered in insulin resistant individuals

J.-F. Harmsen1, M. van Weeghel2,3, J. Wefers1, J. Hoeks1, R.H. Houtkooper2, P. Schrauwen1;

1Nutrition and Movement Science, Maastricht University Medical Center, Maastricht, 2Laboratory Genetic Metabolic Diseases, Amsterdam UMC, Amsterdam, 3Core Facility Metabolomics, Amsterdam UMC, Amsterdam, Netherlands.

Background and aims: Up to ~10-20% of the human muscle transcriptome, lipidome and metabolome have been shown to display 24h rhythmicity. These diurnal changes in muscle metabolism likely facilitate transitions between glucose and fat metabolism in the fed and fasted state, respectively. As insulin resistant (i.e. pre-diabetes) older individuals are characterized by glucose intolerance and metabolic inflexibility, we hypothesized that diurnal rhythms in muscle metabolism are disturbed in this population. Therefore, we here compared diurnal rhythms of the skeletal muscle metabolome between insulin resistant older individuals and young, healthy volunteers, measured under controlled 24h conditions.

Materials and methods: Five sequential skeletal muscle biopsies were obtained from 12 older insulin resistant men (65±9 years, BMI: 30±3 kg/m²) and 12 young healthy men (22±2 years, BMI: 22±2 kg/m²) over the course of 24h, i.e. at 8AM, 1PM, 6PM, 11PM and 4AM. Volunteers were provided with three standardized meals, and biopsies were taken directly before meals to prevent direct postprandial effects. Muscle metabolites were determined in these muscle biopsies using UPLC/HRMS-based semi-targeted metabolomics. Individual metabolites were classified as rhythmic according to 24h cosinor analysis. If metabolites were classified as rhythmic in both groups, the Circacompare mixed model was used to detect differences in mean levels and amplitude and to detect phase shifts. For non-rhythmic metabolites and metabolites that were only rhythmic in one group, a 2-way repeated measures mixed-effects model was used to detect differences over time between groups.

Results: Out of 115 metabolites that could be detected in both groups, 48 were rhythmic in the young healthy men and 51 in the older insulin resistant adults. 35 were rhythmic in both groups. The 13 metabolites that were only rhythmic in young men included e.g. coenzyme A, adenosine, asparagine and kynurenic acid. The 16 metabolites that were only rhythmic in insulin resistant men included e.g. ADP, glucose, glycerate, citric acid, histidine, lactate and uric acid. From the 35 shared rhythmic metabolites, most metabolites displayed higher mean levels in the young healthy men, such as hydroxybutyrate. In contrast, ADP-ribose, flavin adenine dinucleotide (FAD), malate and methionine had higher mean levels in insulin resistant men. FAD and uridine diphosphate were phase delayed, i.e. peaked at a later time of the day, in insulin resistant men. Interestingly, differences in metabolite levels between groups were most pronounced during the night at 4AM, with 38 metabolites being significantly different between groups.

Conclusion: Diurnal rhythms in the muscle metabolome were altered in insulin resistant men with some metabolites losing but others gaining rhythmicity. Differences in the muscle metabolome were especially apparent during the night, which might be linked to the reduced switch from glucose to fat oxidation observed upon insulin resistance. Future studies should explore how diurnal rhythms in the muscle metabolome can be modified and which metabolites, that are affected during the night, can be mechanistically linked to the metabolic inflexibility.

Clinical Trial Registration Number: NCT03733743

Supported by: This work is partly financed by the Netherlands Organization for Scientific Research (TOP 40-00812-98-14047 to P.S.). We acknowledge the support from the Netherlands Cardiovascular Research Initiative: an initiative with the support of the Dutch Heart Foundation (CVON2014-02 ENERGISE).

Disclosure: J. Harmsen: None.

247

Diurnal pattern of meal tolerance and insulin sensitivity in type 2 diabetes

Y. Yadav1, D. Romeres2, C. Cobelli2, R. Carter3, A. Basu1, C. Dalla Man2, R. Basu1;

1Division of Endocrinology, University of Virginia, Charlottesville, USA, 2University of Padova, Padova, Italy, 3Department of Health Sciences Research, Mayo Clinic, Rochester, USA.

Background and aims: We have previously shown that individuals with type 2 diabetes (T2DM) have nocturnal hyperglycemia due to increased endogenous glucose production (EGP). The objective of this study was to assess the diurnal regulation of carbohydrate metabolism and insulin sensitivity in individuals with and without T2DM.

Materials and methods: 19 T2DM [8F; Age: 60±11 yrs; BMI:32±5 kg/m2] and 19 anthropometrically matched non-diabetic (ND) [11 F; Age: 53±12 yrs; BMI:29±5 kg/m2 ] subjects were studied during breakfast (B), lunch(L) and dinner (D) with identical mixed meals (75 gm carbs) on three consecutive days in randomized Latin square design. The triple tracer technique was used to estimate meal rate of appearance (MRa), endogenous glucose production (EGP), and rate of glucose disappearance (Rd). Indices of Insulin sensitivity (SI), insulin secretion (Φ) and β-cell responsivity (Disposition Index: DI) were estimated using the oral glucose minimal model and the C-peptide model, respectively.

Results: Fasting and post meal glucose concentrations (iAUC) were lower (p< 0.01) in ND than T2DM subjects after all three meals. In contrast, insulin iAUC was higher (p<0.05) in ND vs. T2DM subjects after L and D but not after breakfast. Of note insulin concentrations achieved following D was significantly lower in T2D than what was observed at B, L. MRa did not differ between ND and T2DM subjects for all meals. However, fasting EGP was higher (p<0.02) and post prandial EGP was less suppressed (p<0.05) in T2DM than ND subjects during all three meals. Rd was significantly higher (p<0.01) in ND than T2DM subjects after all three meals. SI was significantly higher (p<0.05) in ND than T2DM subjects during B and D but though numerically higher did not reach statistical significance during L (p=0.07). Φ and DI were significantly higher (p<0.01) in ND than T2DM subjects for all three meals.

Conclusion: These data demonstrate 1) the presence of a diurnal pattern of carbohydrate tolerance in T2DM which differs from what is observed in ND subjects and 2) impairment of insulin sensitivity and beta cell function in T2D responsible for impaired glucose tolerance throughout the day. We conclude that the lower DI in T2D coupled with lower Rd and higher EGP following dinner may be partly responsible for the nocturnal hyperglycemia previously observed.

figurecg

Supported by: NIH RO1 DK029953 to RB, NIH R01 DK085516 to AB from National Institute of Health

Disclosure: Y. Yadav: None.

248

The effect of melatonin on incretin hormones: results from experimental and randomised clinical studies

E. Stistrup Lauritzen1, J. Støy2, C. Bæch-Laursen3, N. Grarup3, T. Hansen3, N. Jessen2, N. Møller4, B. Hartmann3, J. Holst3, U. Kampmann2;

1Medical/Steno Research Laboratory, Aarhus University, Aarhus, 2Steno Diabetes Center Aarhus, Aarhus University Hospital, Aarhus, 3Novo Nordisk Foundation Center for Basic Metabolic Research, University of Copenhagen, Copenhagen, 4Medical/Steno Research Laboratory, Aarhus University Hospital, Aarhus, Denmark.

Background and aims: Glucose homeostasis is under circadian control through both endocrine and intracellular mechanisms with several lines of evidence suggesting that melatonin affects glucose homeostasis. We aimed to evaluate the acute in-vivo and in-situ effects of melatonin on secretion of the incretin hormones, GLP-1 and GIP, and their impact on β-cell insulin secretion.

Materials and methods: A human randomized, double-blinded, placebo-controlled crossover study combined with confirmatory in-situ studies of perfused rat intestines and pancreases were performed. Fifteen healthy male participants were examined 2 x 2 times: An oral glucose tolerance test (OGTT) was performed on day one and an isoglycemic intravenous glucose infusion (IIGI) replicating the blood glucose profile of the OGTT day was performed on day two. This pair of study days were completed two times during either melatonin (10 mg capsules administered orally each hour for four hours, M) or placebo (identical placebo capsules, P) treatment. For the in-situ studies, six rat intestines and four rat pancreases were perfused arterially with perfusion buffer ± melatonin. The intestines were concomitantly perfused with glucose through the luminal compartment.

Results: In humans, GIP levels were significantly reduced during an OGTT after melatonin treatment compared with placebo (iAUC difference 0.74±0.3 μM x min, p=0.047) (Panel A). There was a significant time x treatment interaction (ANOVA p=0.003), and post-hoc analysis revealed a difference at t = 40, 90, and 120 minutes (M: 33±3, 34±4, and 28±3 vs P: 39±3, 39±4, and 36±4 pM, P=0.02, 0.02, and <0.001). This effect of melatonin on GIP secretion was also observed in the perfused rat intestines (ANOVA time x treatment p=0.003) (Panel B) in which GLP-1 secretion was also impaired by arterial melatonin infusion (ANOVA time x treatment p<0.001). Despite a decrease in GIP levels, the in-vivo glucose-stimulated insulin secretion was unaffected by melatonin (p=0.78). Likewise, melatonin perfusion of the rat pancreas did not change glucose-stimulated insulin secretion (p=0.77).

Conclusion: Melatonin reduced GIP secretion during an oral glucose challenge in healthy young men but did not affect the insulin secretion. Reduced GIP secretion was confirmed in an in-situ model of the rat intestine.

figurech

Clinical Trial Registration Number: NCT03204877

Supported by: The Novo Nordisk Foundation (NNF16OC0022280) and The Danish Diabetes Academy (DDA1022 8201)

Disclosure: E. Stistrup Lauritzen: None.

OP 45 What surgery can do for you

249

Sleeve gastrectomy suppresses hepatic glucose production and increases hepatic insulin clearance independent of weight loss

D. Ben-Zvi1, A. AlKurd2, B. Moalem1, A. Permyakova3, H. Israeli1, A. Bardugo1, L. Hefetz1, M. Bergel1, A. Haran1, S. Azar3, J. Tam3, R. Grinbaum2, R. Ben-Haroush Schyr1;

1Dev. Biology and Cancer Research, Hebrew University of Jerusalem, Jerusalem, 2Hadassah Medical Center, Jerusalem, 3Institute for Drug Research, Hebrew University of Jerusalem, Jerusalem, Israel.

Background and aims: Bariatric surgeries induce weight-loss which is associated with an improvement in hepatic steatosis and reduction in hepatic glucose production. It is not clear whether these outcomes are entirely due to weight-loss, or whether the new anatomy imposed by the surgery contributes to the improvement in the metabolic function of the liver.

Materials and methods: We performed vertical sleeve gastrectomy (VSG) on obese mice provided with a high-fat high-sucrose diet and compared them to diet and weight-matched sham-operated mice (WMS).

Results: 40 days after surgery, VSG-operated mice displayed lesser hepatic steatosis compared to WMS. By measuring the fasting glucose and insulin levels in the blood vessels feeding and draining the liver we showed directly that hepatic glucose production was suppressed after VSG in fasting animals. Insulin levels were elevated in the portal vein of VSG-operated mice but not in venous or arterial blood. indeed, over 70% of the insulin was cleared in the liver of VSG-operated mice, while less than 50% was cleared in WMS-operated mice. The hepatic expression of genes associated with insulin clearance was upregulated as well. We repeated the experiment in lean mice and observed that portal insulin and glucagon are elevated, but only insulin clearance is increased in VSG-operated mice. Therefore the liver of VSG-operated mice experiences hyperinsulinemia and hyperglucagonemia.

Conclusion: Direct measurement of glucose and insulin in the blood entering and leaving the liver shows that VSG affects glucose and insulin metabolism through weight-loss-dependent and independent mechanisms.

figureci

Supported by: ERC StG for DBZ; ISF for JT

Disclosure: D. Ben-Zvi: Grants; ERC StG H2020.

250

Beta cell mass in type 2 diabetes before and after gastric bypass surgery, measured as pancreatic uptake of radiolabeled exendin in human

L.N. Deden1, M. Boss2, E.J. Hazebroek1, F.J. Berends1, M. Gotthardt2;

1Bariatric Surgery, Vitalys, Rijnstate Hospital, Arnhem, 2Nuclear medicine, Radboud University Medical Center, Nijmegen, Netherlands.

Background and aims: An approach for in vivo beta cell imaging is targeting the glucagon-like peptide-1 (GLP-1) receptor by radiolabeled exendin-4. Currently, the role of beta cells in the onset, course and remission type 2 diabetes mellitus (T2DM) is not clear. Beta cell mass (BCM) and function (BCF) might be related to T2DM remission after gastric bypass surgery (RYGB). The aim of this study is examining BCF and pancreatic uptake of 68Ga-exendin-4 (EX) before and after RYGB in patients with T2DM.

Materials and methods: Thirteen patients were included between December 2017 and September 2019. Arginine stimulation test (AST), oral glucose tolerance test (OGTT) and EX PET/CT were performed pre- and one year post-RYGB. Total pancreatic uptake of EX per injected activity (kBq/MBq) was measured quantitatively on PET/CT as marker for BCM.

Results: Preliminary analysis in nine patients with complete follow-up was performed. Six were female, mean age of 54 years and mean duration of T2DM was 12 years. Preoperatively, six patients were on insulin therapy (104±53IU/day) and three on metformin (1-2g/day). Postoperatively, average BMI and HbA1c decreased from 39±4.7 to 27±3.7 kg/m2 and from 63±10 to 47±17 mmol/mol, respectively. Preoperatively, pancreatic uptake of EX was lower in the insulin than the metformin group: 1.55 [range: 0.53-2.66] vs 2.95 [range: 2.47-3.20] kBq/MBq (p=0.017). In addition, the c-peptide response during OGTT was smaller in the insulin as compared to the metformin group (1.1 vs 2.9 nmol/l, p=0.015). Postoperatively, in the insulin group, one patient had complete remission (i.e. no antidiabetics and normal HbA1c), two patients had little improvement (insulin or sulfonylurea treatment and unchanged HbA1c) and three patients had improvements in between. The mean pancreatic uptake increased to 2.29 [1.86-3.06] kBq/MBq (p=0.025) in the insulin group. This seems to be related to the degree of improvement; relative increase was 56-340% in three patients with most improvement, and 8-30% in patients with least improvement. In the metformin group all patients had complete remission. The mean pancreatic uptake remained stable or decreased, with relative changes ranges from -47% to +13%.

Conclusi