Effect of sitagliptin on energy metabolism and brown adipose tissue in overweight individuals with prediabetes: a randomised placebo-controlled trial

Aims/hypothesis The aim of this study was to evaluate the effect of sitagliptin on glucose tolerance, plasma lipids, energy expenditure and metabolism of brown adipose tissue (BAT), white adipose tissue (WAT) and skeletal muscle in overweight individuals with prediabetes (impaired glucose tolerance and/or impaired fasting glucose). Methods We performed a randomised, double-blinded, placebo-controlled trial in 30 overweight, Europid men (age 45.9 ± 6.2 years; BMI 28.8 ± 2.3 kg/m2) with prediabetes in the Leiden University Medical Center and the Alrijne Hospital between March 2015 and September 2016. Participants were initially randomly allocated to receive sitagliptin (100 mg/day) (n = 15) or placebo (n = 15) for 12 weeks, using a randomisation list that was set up by an unblinded pharmacist. All people involved in the study as well as participants were blinded to group assignment. Two participants withdrew from the study prior to completion (both in the sitagliptin group) and were subsequently replaced with two new participants that were allocated to the same treatment. Before and after treatment, fasting venous blood samples and skeletal muscle biopsies were obtained, OGTT was performed and body composition, resting energy expenditure and [18F] fluorodeoxyglucose ([18F]FDG) uptake by metabolic tissues were assessed. The primary study endpoint was the effect of sitagliptin on BAT volume and activity. Results One participant from the sitagliptin group was excluded from analysis, due to a distribution error, leaving 29 participants for further analysis. Sitagliptin, but not placebo, lowered glucose excursion (−40%; p < 0.003) during OGTT, accompanied by an improved insulinogenic index (+38%; p < 0.003) and oral disposition index (+44%; p < 0.003). In addition, sitagliptin lowered serum concentrations of triacylglycerol (−29%) and very large (−46%), large (−35%) and medium-sized (−24%) VLDL particles (all p < 0.05). Body weight, body composition and energy expenditure did not change. In skeletal muscle, sitagliptin increased mRNA expression of PGC1β (also known as PPARGC1B) (+117%; p < 0.05), a main controller of mitochondrial oxidative energy metabolism. Although the primary endpoint of change in BAT volume and activity was not met, sitagliptin increased [18F] FDG uptake in subcutaneous WAT (sWAT; +53%; p < 0.05). Reported side effects were mild and transient and not necessarily related to the treatment. Conclusions/interpretation Twelve weeks of sitagliptin in overweight, Europid men with prediabetes improves glucose tolerance and lipid metabolism, as related to increased [18F] FDG uptake by sWAT, rather than BAT, and upregulation of the mitochondrial gene PGC1β in skeletal muscle. Studies on the effect of sitagliptin on preventing or delaying the progression of prediabetes into type 2 diabetes are warranted. Trial registration ClinicalTrials.gov NCT02294084. Funding This study was funded by Merck Sharp & Dohme Corp, Dutch Heart Foundation, Dutch Diabetes Research Foundation, Ministry of Economic Affairs and the University of Granada. Electronic supplementary material The online version of this article (10.1007/s00125-018-4716-x) contains peer-reviewed but unedited supplementary material, which is available to authorised users.

without sequelae. Thus, thirty participants completed the study. Also, one participant from the sitagliptin group was excluded from analysis, due to a distribution error where the participant received both sitagliptin and placebo as treatment.

Study design
Participants were enrolled in a randomised, double-blinded, placebo-controlled study that was and skeletal muscle were determined by PET/CT scan (Gemini TF-64, Philips Healthcare, Best, The Netherlands). In addition, thermoneutral and cold-exposed venous blood samples were collected. On the second day at the Leiden University Medical Center, a fasting skeletal muscle biopsy was taken from the vastus lateralis muscle followed by an OGTT. All measurements took place after participants had fasted for 10h overnight and had consumed a standardised dinner the night before. In addition, participants were asked to refrain from caffeine and alcohol intake or strenuous physical activity 24 h prior to the study days. Each week during the treatment period, participants measured their blood glucose and were contacted by the investigator to monitor compliance, adverse events or signs of hypoglycemia. In addition, participants were instructed not to alter their lifestyle during the study period. The study was conducted between March 2015 and September 2016.

Indirect calorimetry, individualised cooling protocol and [ 18 F]FDG PET-CT scan
On the first measurement day an intravenous cannula was placed in antecubital vein for blood sampling during thermoneutral and mild cold conditions and injection of the [ 18 F]FDG tracer.
First detailed body composition was obtained by dual-energy X-ray absorptiometry, followed by placement of 14 wireless temperature sensors at 14 prescribed ISO-defined places [2].
Mean skin temperature was calculated as the average of all iButtons. Distal skin temperature was calculated as the average temperature of the hand and feet. Proximal skin temperature was defined as the average of the iButtons on the chest, abdomen, scapula and lower back.
Next participants were placed in a bed in semi-supine position between two water-perfused mattresses (BlanketRol® III Sub-Zero (CSZ) Products, Cincinnati, OH, USA). As described previously [3], the protocol started with 1 h of thermoneutrality (water temperature 32°C) at which thermoneutral REE was measured for 30 minutes using indirect calorimetry and a venous blood sample was drawn. After1 h at thermoneutrality, participants were progressively cooled until their shivering point or until they reached the minimum water temperature of 9°C. At this mild cold condition REE was measured for another 30 minutes and a second blood sample was collected. Oxygen consumption and carbon dioxide production were determined every minute. REE and respiratory quotient (RQ) were calculated and substrate utilization was assessed by calculation of lipid and glucose oxidation after correction for protein oxidation as described earlier [4]. Non-shivering thermogenesis (NST) was assessed by the difference in REE during mild cooling compared to thermoneutral conditions. Subsequently, 110 MBq of [ 18 F]FDG was injected intravenously and after 1 h of incubation, the PET/CT imaging protocol started with a low-dose CT scan (120 kV, 30 mAs), immediately followed by a PET scan (10 bed positions, 4 minutes per bed position) acquired according to EARL standards [5] to assess [ 18 F]FDG uptake by BAT, WAT and skeletal muscle from skull to pelvis. In the sitagliptin group one participant became claustrophobic inside the PET/CT scan and could therefore not finish this measurement.

Skeletal muscle biopsy and oral glucose tolerance test
On the second measurement day, a fasted skeletal muscle biopsy was taken from the vastus lateralis muscle according to the technique of Bergström [6]. Subsequently, the samples were frozen in liquid nitrogen and stored at -80° C until further analysis. After 1 h of rest, a glucose tolerance was assessed using a 75-g OGTT. A cannula was inserted in the antecubital vein for blood sampling and samples were drawn at respectively t = -10, 0, 10, 20, 30, 40, 50, 60, 90, 120 minutes after ingestion of the glucose drink. Serum was obtained, snap-frozen in liquid nitrogen and stored at -80°C until further analysis.
Insulin concentrations were measured using ELISA (Crystal Chem Inc., Elk Grove Village, IL, USA). The intra-assay coefficients of variability (CV) were 3.8%, 4.2%, 3.6%, 2.5% and 5.9% for triacylglycerol, total cholesterol, free fatty acids, glucose and insulin respectively. weighting the corresponding subclass diameters with their particle concentrations [9]. Details of the experimentation and applications of the NMR metabolomics platform have been described previously [10]. Data were analysed using SoftMaxPro 5.4.1 software. For the analysis of the OGTT, theAUC was calculated using the trapezoidal rule [11]. Incremental AUC was calculated by deducting the area below the baseline value from total AUCs. Insulin sensitivity was estimated using the Matsuda index [12]. The insulinogenic index (IGI; ΔI 0-30 /ΔG 0-30, where I is insulin and G is glucose) was used as a measure of early insulin secretion [13]. The oral disposition index (DI 0 ; [ΔI 0-30 /ΔG 0-30 ]/fasting insulin) was used to estimate beta cell function relative to the prevailing level of insulin resistance [14].

PET/CT scan analysis
[ 18 F]FDG uptake by BAT, WAT and skeletal muscle was determined from the [ 18 F]FDG PET/CT scan using Fiji ImageJ 1.51d (Beth Israel Deaconess Medical Center, Beth, Israel) [7] and analysed by two researchers (KJN, BMT) blinded to allocation. In the region of interest (ROI), the bilateral cervical and clavicular regions, mediastinal and paravertebral BAT areas were autocontoured using an set personalised standardised uptake value (SUVindiv) threshold with a tissue radiodensity between -190 and -10 Hounsfield units [15].
SUVindiv threshold was calculated with the following formula: 1.2/lean body mass (in kg)/body mass (in kg), according to the latest expert panel recommendations [16]. BAT metabolic volume (BMV) was measured in millilitres, BAT activity was reported in terms of SUV (the ratio of activity in kBq/mL within the ROI and the injected activity [kBq] per bodyweight [g]). Both SUV for the hottest single voxel (SUVmax) and mean SUV for all voxels within a BAT region (SUVmean) are reported. For WAT areas (subcutaneous and paracolic), skeletal muscles (sternocleidomastoid, longus colli, trapezius, deltoid, pectoralis major, psoas major, and gluteus maximus muscle) and reference tissues (liver, cerebellum and descending aorta), an SUV threshold was set at 0 and no Hounsfield units threshold was applied.     presented as mean (standard error of the mean). BAT: brown adipose tissue, BMV: BAT metabolic volume, SUV: standardised uptake value, WAT: white adipose tissue. Mixed model analysis was used for statistical comparison. * 0.002 < p < 0.05 week 0 vs week 12, not significant with Bonferroni corrected level of significance0.002 (alpha = 0.05 / 18). In the Sitagliptin group one subject was claustrophobic and did not completed the PET/CT scan.  Figure 2 The effect of sitagliptin on serum IDL and LDL particle concentration in overweight men with prediabetes. Serum was collected before (open circles / white bars/ Week 0) and after (closed circles / black bars/ Week 12) 12 weeks of treatment with placebo (n=15) or sitagliptin (n=14). NMR was used to measure serum IDL (a), large-(b), medium-(c) and small-sized (d) LDL particle concentration. In addition, mean LDL particle size (e) was determined. Data are presented as mean ± S.E.M., and as individual measurements. Mixed model analysis was used for statistical comparison. Bonferroni corrected level of significance is 0.01 (alpha = 0.05 / 5).