Effects of dapagliflozin and n-3 carboxylic acids on non-alcoholic fatty liver disease in people with type 2 diabetes: a double-blind randomised placebo-controlled study

Aims/hypothesis The EFFECT-II study aimed to investigate the effects of dapagliflozin and omega-3 (n-3) carboxylic acids (OM-3CA), individually or combined, on liver fat content in individuals with type 2 diabetes and non-alcoholic fatty liver disease (NAFLD). Methods This randomised placebo-controlled double-blind parallel-group study was performed at five clinical research centres at university hospitals in Sweden. 84 participants with type 2 diabetes and NAFLD were randomly assigned 1:1:1:1 to four treatments by a centralised randomisation system, and all participants as well as investigators and staff involved in the study conduct and analyses were blinded to treatments. Each group received oral doses of one of the following: 10 mg dapagliflozin (n = 21), 4 g OM-3CA (n = 20), a combination of both (n = 22) or placebo (n = 21). The primary endpoint was liver fat content assessed by MRI (proton density fat fraction [PDFF]) and, in addition, total liver volume and markers of glucose and lipid metabolism as well as of hepatocyte injury and oxidative stress were assessed at baseline and after 12 weeks of treatment (completion of the trial). Results Participants had a mean age of 65.5 years (SD 5.9), BMI 31.2 kg/m2 (3.5) and liver PDFF 18% (9.3). All active treatments significantly reduced liver PDFF from baseline, relative changes: OM-3CA, −15%; dapagliflozin, −13%; OM-3CA + dapagliflozin, −21%. Only the combination treatment reduced liver PDFF (p = 0.046) and total liver fat volume (relative change, −24%, p = 0.037) in comparison with placebo. There was an interaction between the PNPLA3 I148M polymorphism and change in liver PDFF in the active treatment groups (p = 0.03). Dapagliflozin monotherapy, but not the combination with OM-3CA, reduced the levels of hepatocyte injury biomarkers, including alanine aminotransferase, aspartate aminotransferase, γ-glutamyl transferase (γ-GT), cytokeratin (CK) 18-M30 and CK 18-M65 and plasma fibroblast growth factor 21 (FGF21). Changes in γ-GT correlated with changes in liver PDFF (ρ = 0.53, p = 0.02). Dapagliflozin alone and in combination with OM-3CA improved glucose control and reduced body weight and abdominal fat volumes. Fatty acid oxidative stress biomarkers were not affected by treatments. There were no new or unexpected adverse events compared with previous studies with these treatments. Conclusions/interpretation Combined treatment with dapagliflozin and OM-3CA significantly reduced liver fat content. Dapagliflozin monotherapy reduced all measured hepatocyte injury biomarkers and FGF21, suggesting a disease-modifying effect in NAFLD. Trial registration: ClinicalTrials.gov NCT02279407 Funding: The study was funded by AstraZeneca. Electronic supplementary material The online version of this article (10.1007/s00125-018-4675-2) contains peer-reviewed but unedited supplementary material, which is available to authorised users.


Inclusion criteria
1. Provision of informed consent prior to any study-specific procedures.
2. Men or women aged ≥40 years and ≤75 years with suitable veins for cannulation or repeated venipuncture.
3. Liver fat content as assessed by MRI >5.5%. 4. Type 2 diabetes diagnosed since ≥6 months in accordance with World Health Organization criteria. Diagnosis of type 2 diabetes could have been based on the following:  prior documentation in medical records of type 2 diabetes and/or  treatment with anti-hyperglycemic medications and/or diet and/or  random plasma glucose ≥11.1 mmol/l or fasting ≥7.0 mmol/l or HbA1c ≥48 mmol/mol (6.5%). 5. Antidiabetic therapy: stable (i.e. >1 months) metformin and/or sulfonylurea or nonpharmacological treatment.

Exclusion criteria
1. Involvement in the planning and/or conduct of the study.
2. Participation in another clinical study with an investigational product during the last 28 days.
3. Any condition when MRI is contraindicated such as, but not limited to, having a pacemaker or claustrophobia. 4. History of or presence of (as found at Visit 1) any clinically significant disease or disorder which, in the opinion of the investigator, may either put the patient at risk because of participation in the study, or influence the results or the patient's ability to participate in the study.
5. Diagnosis or signs of type 1 diabetes (e.g. history of positive islet antibodies).
10. Intolerance or allergy to dapagliflozin or any other sodium-glucose co-transporter 2 inhibitor (SGLT2i) or any other substance in the tablets.
11. Use of dapagliflozin or any other SGLT2i within the last 4 weeks prior to Visit 1.
12. Use of insulin or glucagon-like peptide-1 therapy or oral antidiabetic drugs other than metformin or sulfonylurea within the last 4 weeks prior to Visit 1.
13. Use of fish oil, other eicosapentaenoic acid (EPA)-or docosahexaenoic acid (DHA)containing supplements, or EPA-and/or DHA-fortified foods within 4 weeks from Visit 1, or during the study.
14. Ongoing weight-loss diet (hypocaloric diet) or use of weight-loss agents, unless the diet or treatment has been stopped at least 3 months before screening and that the patient has had a stable body weight (+/-3 kg) during the 3 months before screening.
15. Use of flax seed, perilla seed, hemp, spirulina or blackcurrant oils within 1 month from study start and during the study until study end. 16. Any clinically significant abnormalities in clinical chemistry, haematology or urinalysis results as judged by the investigator. This includes signs of liver disease other than non-alcoholic fatty liver disease that motivated further investigations or treatment based on clinical judgment.
17. Recent history (past 12 months) of drug abuse or alcohol abuse. Alcohol abuse is defined as >14 drinks per week (1 drink = 35 cl beer, 14 cl wine or 4 cl hard liquor) or as judged by the investigator.
18. Women who are pregnant, lactating or planning to become pregnant during the study period, or women of childbearing potential who are not using acceptable contraceptive methods. A woman is considered of childbearing potential if she is not surgically sterile or is less than 1 year since last menstrual period. Acceptable contraceptive methods were: combined (oestrogen-and progesterone-containing) hormonal contraception associated with inhibition of ovulation (oral, intravaginal, transdermal), progesterone-only hormonal contraception associated with inhibition of ovulation (oral, injectable, implantable), intrauterine device, intrauterine hormone-releasing system, bilateral tubal occlusion and vasectomized.
19. Any other condition the investigator believed would interfere with the patient's ability to provide informed consent, comply with study instructions, or which might confound the interpretation of the study results or put the patient at undue risk. 20. Plasma donation within 1 month of screening or any blood donation/blood loss >500 ml during the 3 months prior to Visit 1 or during the study.

Blood analyses
Fasting blood samples were taken at baseline and at the end of treatment in the morning before the intake of the investigational products. Plasma glucose levels were analysed using a hexokinase enzymatic method; GLUC3 (Glucose HK Gen. 3) reagent kit (Roche Diagnostics, Indianapolis, IN, USA). Non-esterified fatty acids (NEFA) were analysed using an enzymatic colorimetric assay; NEFA HR (2) test kit (WAKO Chemicals, Richmond, VA, USA). Both glucose and NEFA analyses were performed using the Roche Modular and Cobas Analyzer. Serum levels of total cholesterol and triglycerides were measured using the Cholesterol Gen 2 (CHOL2) reagent and the triglyceride reagent, respectively, from Roche Diagnostics. High-density-lipoprotein (HDL) and low-densitylipoprotein (LDL) cholesterol were measured using direct HDL and LDL cholesterol methods, HDLC3, third-generation reagents and LDL-C plus, second-generation assay (Roche Diagnostics). Levels of cholesterol and triglycerides were measured on the Roche Modular and Cobas Analyzers. Beta-hydroxybutyrate plasma levels were analysed with an enzymatic colorimeteric assay (LiquiColor, Stanbio Laboratory, Boerne, TX, USA). Uric acid was measured using an enzymatic colorimetric assay (ABX Pentra Uric acid CP, Horiba ABX, Montpellier, France). Measurement of HbA1c levels utilized the principles of ionexchange high-performance liquid chromatography. All Variant II and Variant II Turbo Hemoglobin A1c reagents were manufactured by Bio-Rad (Hercules, CA, USA). Plasma insulin levels were measured using Access Ultrasensitive Insulin assay, a simultaneous onestep immunoenzymatic (sandwich) assay (Beckman Coulter Inc., Brea, CA, USA). Serum Cpeptide levels were analysed by a two-site sandwich immunoassay methodology, ADVIA Centaur C-Peptide ReadyPack (Siemens Healthcare Diagnostics, Tarrytown, NY, USA).
Total (esterified and free plasma levels) EPA and DHA concentrations in plasma were measured by Covance Laboratories Inc. (Madison, WI, USA), on behalf of AstraZeneca. The method used was liquid chromatography with tandem mass spectrometric detection (LC-MS/MS), which was validated over the range 1.00-250 µg/ml, and used appropriate stable label internal standards. EPA and DHA were extracted from plasma after digestion using liquid-liquid extraction, evaporated under nitrogen and subsequently reconstituted and analysed by LC-MS/MS. The assay had a within-assay and between-assay coefficient of variation of less than or equal to 15%.
PNPLA3 genotyping The single nucleotide polymorphism in the PNPLA3 gene rs738409 was investigated in the 80 participants (n = 20 in each group) who gave informed consent for genetic testing. DNA extraction from whole EDTA-treated blood and genotyping using quantitative real-time PCR were performed by Tepnel Pharmaceutical Services (Hologic Ltd, Livingston, UK).
MRI MRI was used to quantify liver lipids using a proton density fat fraction technique utilizing a spoiled three-dimensional gradient, six-echo gradient echo in the axial plane covering the liver in a single breath-hold. The water-fat image reconstruction was performed including T2* and a multi-peak lipid spectrum in the signal model [1]. The liver was segmented manually by a trained operator from the axial slices using the software ImageJ (https://imagej.nih.gov/ij/). The border of the liver was avoided to reduce partial volume effects. The liver fat content was determined by the median of the fat fraction values inside the delineated total liver volume. Liver volume was assessed using a dedicated T1-weighted gradient echo, single echo, single breath-hold scan with high resolution and spectral fat suppression. The full liver was segmented by a trained operator using the semi-automated segmentation software Smartpaint (http://www.cb.uu.se/~filip/SmartPaint/). The coefficient of variation for repeated examinations and analyses of liver PDFF was 5.3% as determined by test-retest scanning and analysis of 10 healthy volunteers.
For determination of abdominal adipose tissue volumes, a 21-slice, 8 mm slice thickness, 3echo gradient echo axial scan, positioned at the L4/L5, was performed in a single breath-hold.
Water and fat images were reconstructed, and visceral adipose tissue (VAT) and subcutaneous adipose tissue (SAT) volumes were extracted using an automated method. Manual inspection was performed, and results were corrected when needed by an experienced operator. The coefficient of variation for repeated examinations and analyses of SAT and VAT was 0.7% and 2.3%, respectively, as determined by test-retest scanning and analysis of 10 healthy volunteers. and 4-hydroxy Nonenal-d3 [4-HNE-D3]), were all purchased from Cayman Chemical (Ann Arbor, MI, USA). Standards of 2,3-dinor-8-iso-PGF2-alpha and 8-iso-PGF2-alpha, as well as the corresponding isotopic labelled internal standard 8-iso-PGF2-alpha-D4, were all purchased from Cayman Chemical.

Acylcarnitine analysis
We used a simplified sample pretreatment protocol described by Peng et al. [4] where we injected the supernatant of the protein-precipitated plasma sample directly onto the LC-MS system without a drying step. Briefly, plasma samples (50 µl LC-MS analysis was performed using a system comprising an ultra-high-performance liquid chromatography (UHPLC) system (1290 Infinity binary pump, 1290 Infinity autosampler with thermostat and 1290 Infinity thermostated column compartment, all Agilent Technologies) coupled to a 6460 triple quadrupole mass spectrometer (Agilent Technologies).
Chromatography was performed on a Poroshell 120 HILIC (100 mm × 2.1 mm i.d., 2.7 µm particle size, 120 Å pore size, Agilent Technologies) at a column temperature of 25°C. A sample volume of 10 l was injected onto the column at a flow rate of 0.3 ml/min. Eluant A consisted of acetonitrile:water (95:5), and eluent B consisted of 10 mM ammonium acetate with 0.2% formic acid. Gradient elution was performed with the following programme: 15% B to 26% B in 4.5 minutes, wash 50% B during 1 minute followed by re-equilibration at 15% B for 3 minutes. The total cycle time of the run was 9 minutes. The instrument settings of the 6460 triple quadrupole mass spectrometer were as follows: drying gas temperature, 325°C; drying gas flow, 9 l/min; nebulizer pressure, 35 psi; capillary voltage, 4000 V, sheath gas temperature 350°C; sheath gas flow, 11 l/min, fragmentor voltage, 110 V; cell accelerator voltage, 4 V; and dwell time, 35 ms. Quadrupoles were working at unit resolution.
Calibration samples were made by spiking a matrix of water/methanol (1:1) with known quantities of acylcarnitine standards and were subjected to the same treatment as the samples described in the sample preparation section above. A six-point calibration curve was obtained by plotting the peak area ratios of corresponding analyte to the internal standard against their theoretical concentrations. The MassHunter Quantitative Analysis Software (version B.06.00, Agilent Technologies) was used for constructing the linear regression analysis with 1/x weighting and for the determination of analyte concentration in the samples.

Aldehyde analysis
The volatility and intrinsically low response of aldehydes in electrospray ionization present analytical challenges; these can be circumvented by chemical derivatization that enables LC-MS/MS measurements to be made with high sensitivity. The derivatization method was based on the protocol described by Matsouka et al. [5] in which aldehydes are condensed with CHD and ammonium ions to form water-soluble adducts. Briefly, both the calibration standards and the plasma samples (200 µl) were pipetted into 2 ml microcentrifuge tubes (Eppendorf, Hamburg, Germany) and were subjected to the addition of 125 µl of CHD solution, 125 µl of acetonitrile and 20 µl of internal standard. The mixtures were put into a Thermomixer comfort (Eppendorf GmbH) and heated to 60°C for 60 minutes at a shaking frequency of 750 rpm.
After the reaction, the tubes were placed on ice for 1 minute. Next, 600 µl of acetonitrile was added to each tube in order to precipitate the proteins. The tubes were subjected to centrifugation (12,500 rpm, 8°C) for 10 minutes, and the supernatants were subjected to solidphase extraction (SPE) using a polymeric reversed-phase approach in a 96-well plate format (Strata-X 33 µm; 60 mg/well, Phenomenex, Torrance, CA, USA). Briefly, the SPE well plate was activated with 2 ml of methanol and reconditioned with 1 ml of 75% aqueous acetonitrile before use. The calibration standards and the samples were loaded onto the SPE plate, and the flow-through fractions were collected into glass vials. The fractions were dried under gentle nitrogen gas flow at -35°C (Techne Sample Concentrator, Staffordshire, UK) and finally reconstituted in 400 µl of 20% acetonitrile.
The analysis was performed on an LC-MS system comprising a UHPLC system (1290 Infinity binary pump, 1290 Infinity autosampler with thermostat and 1290 Infinity thermostated column compartment, all Agilent Technologies) coupled to a 6490 triple quadrupole mass spectrometer (Agilent Technologies). Chromatography was performed on an Eclipse RRHD column (100 mm × 2.1 mm i.d., C18, 1.8 µm particle size, 300 Å pore size; Agilent Technologies) at a column temperature of 45°C. A sample volume of 20 L was injected onto the column at a flow rate of 0.35 ml/min. Eluant A consisted of 0.1% formic acid in water, and eluent B consisted of acetonitrile with 0.1% formic acid. Gradient elution was performed with the following programme: 12% B to 55% B in 5.5 minutes, wash 95% B during 1.5 minutes followed by re-equilibration at 12% B for 3.5 minutes. The total cycle time of the run was 10.5 minutes.
The SPE well plate was conditioned with 1.2 ml of 2% formic acid in methanol followed by equilibration with 1.2 ml of water. The samples were loaded onto the SPE plate and washed with 1.2 ml of water followed by 1.2 ml of 20% methanol in water and then 100% acetonitrile. The SPE columns were dried for 30 s, and analytes were eluted with 1. and dwell time, 50 ms. Quadrupoles were working at unit resolution.