Effects of Sitagliptin on Glycemia, Incretin Hormones, and Antropyloroduodenal Motility in Response to Intraduodenal Glucose Infusion in Healthy Lean and Obese Humans and Patients With Type 2 Diabetes Treated With or Without Metformin

Inhibition of dipeptidyl peptidase-4 (DPP-4) lowers glycemia by increasing intact glucose-dependent insulinotropic polypeptide (GIP) and glucagon-like peptide 1 (GLP-1) concentrations (1). In type 2 diabetes, the insulinotropic effects of GIP and GLP-1 are diminished, although the effect of GLP-1 is better preserved (2,3). GLP-1 also suppresses glucagon secretion (4), appetite, and energy intake (5) and slows gastric emptying (6,7). Therefore, the glucose-lowering effect of DPP-4 inhibitors in this disorder is likely to depend primarily on the actions of GLP-1 rather than of GIP.

Postprandial incretin secretion is regulated by the rate of nutrient delivery to the small intestine (8,9); GIP secretion increases linearly with increasing rates of intraduodenal (ID) glucose infusion, whereas the GLP-1 response is minimal at 1–2 kcal/min but substantially greater at 4 kcal/min (8,9). In both obesity (10,11) and type 2 diabetes (12,13), postprandial GLP-1 responses have been inconsistently reported to be reduced after oral nutrients and those of GIP to be intact. However, even in health, the overall rate of gastric emptying varies from 1 to 4 kcal/min (14), and patients with long-standing type 2 diabetes frequently have accelerated or delayed gastric emptying (15). Therefore, evaluation of incretin responses to nutrients and characterization of the effects of DPP-4 inhibition should be controlled for the rate of gastric emptying.

The majority of studies have reported a lack of effect of DPP-4 inhibitors on gastric emptying for unclear reasons (16–22). Although endogenous GLP-1 slows gastric emptying through suppression of antral motility and stimulation of pyloric contractions (6,7), the effect of DPP-4 inhibitors on these motor mechanisms has not been assessed. DPP-4 inhibitors appear to be weight neutral (23), but reports on their effect on appetite are limited.

Recently, it was shown that the combination of metformin and a DPP-4 inhibitor is more beneficial than either alone for optimizing glycemic control in type 2 diabetes (24). Metformin augments GLP-1 concentrations by uncertain means (25,26) and in mice, can increase expression of GLP-1 and GIP receptors in pancreatic islets (25). Although plasma DPP-4 activity is reportedly reduced in type 2 diabetic patients treated with metformin (27), in vitro studies have failed to show any direct effect of metformin on the catalytic activity of DPP-4 (27,28). Regardless, coadministration of metformin and sitagliptin results in an additive increase in plasma intact GLP-1 concentrations and improvement in postprandial glycemia in both healthy individuals and type 2 diabetic patients (26,29).

In the current study, we hypothesized that glucose lowering by DPP-4 inhibitors would be diminished in type 2 diabetic patients compared with healthy individuals in response to ID glucose infused at 2 kcal/min, a rate at which GIP appears to be the dominant incretin (8,9). We evaluated acute effects of sitagliptin on glycemia, incretin hormones, antropyloroduodenal (APD) motility, and appetite in healthy lean and obese subjects and type 2 diabetic subjects treated with metformin or placebo.

Twelve healthy lean subjects, 12 obese subjects without diabetes, and 12 subjects with type 2 diabetes (Table 1), all Caucasian males, were studied in a double-blind, randomized fashion after providing written informed consent. Type 2 diabetic subjects maintained an HbA_1c <7.5% (58 mmol/mol) on diet alone and had no micro- or macrovascular complications. The protocol was approved by the Royal Adelaide Hospital Human Research Ethics Committee and conducted in accordance with the Declaration of Helsinki.

Lean and obese subjects were studied twice (sitagliptin or control), separated by 3–14 days. Type 2 diabetic subjects were studied four times, during therapy with metformin 850 mg b.i.d. or placebo for 7 days each (first dose in the evening of day 1; sitagliptin or control on days 5 and 8), with a 14-day washout period between.

After a standardized beef lasagna meal (McCain, Victoria, SA, Australia) the evening before each study (∼1900 h), subjects fasted until laboratory testing at ∼0800 h. A manometric assembly (Dentsleeve International, Mississauga, ON, Canada) was inserted transnasally and positioned in the duodenum by peristalsis, with monitoring of antral and duodenal transmucosal potential difference (30). The assembly incorporated seven antral and six duodenal channels spaced at 1.5-cm intervals and a 4.5-cm pyloric sleeve sensor, each perfused with 0.9% saline (30). An additional infusion channel opened 12 cm beyond the pylorus.

An intravenous cannula was inserted for blood sampling. Sitagliptin (100 mg) or matching control (metformin 850 mg or matching placebo in type 2 diabetic subjects) was administered orally with 30 mL water (t = −30 min) followed after 30 min by an ID glucose infusion (60 g glucose dissolved in water to a volume of 240 mL) over 120 min (t = 0–120 min; 2 kcal/min). The catheter was subsequently removed, and subjects ate ad libitum from a cold buffet-style meal (t = 120–150 min) from which energy intake was calculated using Foodworks 3.01 (Xyris Software, Highgate Hill, QLD, Australia) (30). Venous blood was sampled frequently into ice-chilled EDTA tubes for plasma glucose and hormone measurements, with 10 μL/mL DPP-4 inhibitor (DPP4-010; Linco Research, St. Charles, MO) added to tubes for intact incretin measurements. Plasma was separated within 15 min and stored at –70°C for subsequent analysis.

Plasma glucose concentrations were measured by the glucose oxidase technique (YSI 2300 STAT Plus; Yellow Springs Instruments, Yellow Springs, OH). GLP-1, GIP, and glucagon analyses were performed as previously described (31,32). Intact GLP-1 was measured using a two-site ELISA (C-terminally directed GLP-1F5 catching antibody; N-terminally directed Mab26.1 detecting antibody) (31). Total GLP-1 was assayed using antiserum 89390, requiring the intact amidated C-terminus of the molecule and reacting equally with intact GLP-1 and the primary (N-terminally truncated) metabolite (31). Intact and total GIP were analyzed with the N-terminally and C-terminally directed antisera 98171 (31) and 80867 (32), respectively. Glucagon immunoreactivity was determined using the C-terminally directed antiserum 4305, which measures glucagon of pancreatic origin (31). Insulin was measured by ELISA (10-1113; Mercodia, Uppsala, Sweden).

Manometric pressures were digitally recorded (Flexisoft; Oakfield Instruments, Oxford, U.K.) and analyzed using custom-designed software (A.J. Smout, Academic Medical Center, Amsterdam, the Netherlands) to determine the number of isolated pyloric pressure waves (IPPWs) and antral and duodenal pressure waves over successive 15-min periods (30).

Area under the curve (AUC) for plasma glucose and hormone levels was calculated using the trapezoidal rule. The homeostasis model assessment of insulin resistance (HOMA-IR) was used to estimate insulin sensitivity (33). The AUC_insulin/AUC_glucose ratio was calculated to compare insulin concentrations while correcting for differences in blood glucose levels (33). Data were analyzed using the paired Student t test in healthy lean and obese groups, and two-factor repeated-measures ANOVA with sitagliptin and metformin as factors was used for type 2 diabetic subject data. Repeated-measures ANOVA with treatment and time as factors was also used for intragroup comparisons. Intergroup comparisons were performed using one-factor ANOVA. Post hoc comparisons, adjusted for multiple comparisons by Bonferroni correction, were performed if ANOVAs revealed significant effects. Relationships between variables were assessed by Pearson correlation analysis. All analyses were performed using SPSS version 19 (IBM Corporation, Chicago, IL) statistical software. A sample size of 12 subjects was calculated to have 80% power (at α = 0.05) to detect a difference in the AUC for blood glucose of 73 mmol/L · min with an SD of 82 mmol/L · min between sitagliptin and control in healthy lean and obese subjects (19) and to detect an additive glucose-lowering effect between metformin and sitagliptin in type 2 diabetic subjects (29). Data are presented as mean ± SEM; P < 0.05 was considered statistically significant.

All subjects tolerated the study well.

In healthy lean and obese subjects, fasting plasma glucose concentrations (t = −30 min) did not differ between control and sitagliptin and remained unchanged immediately before ID glucose infusion (t = 0 min). During ID glucose infusion (t = 0–120 min), plasma glucose concentrations increased and were lower after sitagliptin than control (lean: P = 0.001; obese: P = 0.004) (Figs. 1A and 2A, Table 2).

In type 2 diabetic subjects, fasting glucose (t = −30 min) did not differ between control and sitagliptin during either placebo or metformin treatment but was reduced with metformin (ANOVA, metformin effect: P < 0.001). Plasma glucose remained unchanged at 0 min but increased during ID glucose infusion and was lower with metformin (ANOVA, AUC: P < 0.001), without any effect of sitagliptin or interaction between metformin and sitagliptin (Fig. 3A, Table 3). Both fasting plasma glucose (Table 1) and the AUC after ID glucose infusion with sitagliptin and control (Table 2) were greater in type 2 diabetic subjects during placebo treatment than in healthy lean and obese subjects (P < 0.001 for each), without any difference between the latter two groups.

In healthy lean and obese subjects, neither total nor intact GLP-1 concentrations differed between control and sitagliptin before ID glucose infusion. During ID glucose infusion, GLP-1 responses were minimal, but total GLP-1 was lower after sitagliptin than control in healthy lean subjects (P = 0.067 for AUC; P < 0.05 for ANOVA), without any difference in obese subjects (Figs. 1C and 2C, Table 2). In contrast, intact GLP-1 was higher after sitagliptin than control in both healthy lean and obese subjects (P = 0.016 and P < 0.001 for AUC, respectively) (Figs. 1E and 2E, Table 2).

In type 2 diabetic subjects, neither total nor intact GLP-1 concentrations differed between control and sitagliptin before ID glucose infusion but were greater with metformin than placebo (P = 0.016 and P < 0.001). During ID glucose infusion, GLP-1 responses were minimal, but total GLP-1 was greater with metformin than placebo (P = 0.001 for AUC), without any effect of sitagliptin or interaction between metformin and sitagliptin. In contrast, intact GLP-1 was greater with both metformin (P = 0.001 for AUC) and sitagliptin (P = 0.007 for AUC), without any interaction between them (Fig. 3C and E, Table 3).

During fasting, neither total nor intact GLP-1 concentrations differed among the three groups. On the control days, total GLP-1 during ID glucose infusion was less in obese than in healthy lean subjects (P < 0.05 for AUC), and intact GLP-1 was less in obese than in type 2 diabetic subjects (P < 0.05) (Table 2), whereas the AUC for total (but not intact) GLP-1 after ID glucose infusion in all groups combined was inversely related to BMI (r = −0.41, P = 0.04). After adjusting for BMI, the AUC for total GLP-1 did not differ among the groups. Neither total nor intact GLP-1 AUC was related to age. However, after adjusting for BMI, the AUC for intact GLP-1 tended to be positively related to age (r = 0.31, P = 0.06).

In healthy lean and obese subjects, neither total nor intact GIP concentrations differed between control and sitagliptin before ID glucose infusion (Table 2). During ID glucose infusion, GIP increased substantially, but total GIP concentrations were lower after sitagliptin than control in healthy lean subjects (P = 0.019 for ANOVA), without any differences in obese subjects (Figs. 1D and 2D, Table 2). In contrast, intact GIP was higher after sitagliptin than control in both healthy lean and obese subjects (P < 0.001 for both AUC and ANOVA) (Figs. 1F and 2F, Table 2).

In type 2 diabetic subjects, neither total nor intact GIP concentrations differed between control and sitagliptin before ID glucose infusion, and neither was altered by metformin. During ID glucose infusion, GIP increased substantially. Total GIP was greater with metformin (P = 0.014 for AUC), without any effect of sitagliptin or interaction between metformin and sitagliptin. In contrast, intact GIP was greater for sitagliptin than control (P = 0.008 for AUC), without any effect of metformin or interaction between metformin and sitagliptin (Fig. 3D and F; Table 3).

During fasting, neither total nor intact GIP concentrations differed among the three groups (Table 1), nor did total GIP during ID glucose infusion (Table 2). Intact GIP during ID glucose infusion was greater in type 2 diabetic subjects, without any difference between healthy lean and obese subjects (P < 0.01 for AUC). The AUC for total or intact GIP was not related to BMI; however, the AUC for intact GIP was positively related to age on the placebo but not sitagliptin days (r = 0.65, P < 0.001). This relationship remained significant when adjusting for the presence of type 2 diabetes (r = 0.43, P = 0.01).

In healthy lean and obese subjects, glucagon concentrations did not differ between control and sitagliptin before ID glucose infusion. After ID glucose infusion, glucagon declined comparably in both groups, without any effect of sitagliptin (Figs. 1G and 2G, Table 2).

In type 2 diabetic subjects, glucagon concentrations before ID glucose infusion were not affected by sitagliptin but were elevated with metformin (P = 0.025). During ID glucose infusion, glucagon declined from t = 30 min but was greater with metformin (P = 0.003 for AUC), without any effect of sitagliptin or interaction between metformin and sitagliptin (Fig. 3G, Table 3).

Glucagon concentrations were higher in type 2 diabetic than in healthy lean subjects both during fasting (P = 0.034) (Table 1) and after ID glucose infusion with control (P = 0.023 for AUC) and sitagliptin (P = 0.019 for AUC) (Table 2). Fasting glucagon was positively related to plasma intact GIP (r = 0.529, P = 0.001), even after adjusting for the presence of type 2 diabetes (r = 0.47, P = 0.005).

In healthy lean and obese subjects, neither fasting insulin nor HOMA-IR differed between control and sitagliptin. During ID glucose infusion, insulin increased more with sitagliptin than control (lean: P = 0.002, obese: P = 0.088 for AUC), and the AUC_insulin/AUC_glucose ratio was greater after sitagliptin than control in both groups (lean: P < 0.001, obese: P = 0.032) (Figs. 1B and 2B, Table 2).

In type 2 diabetic subjects, neither fasting insulin nor HOMA-IR differed between the control and sitagliptin days, and neither was altered by metformin. During ID glucose infusion, insulin was greater after sitagliptin than control (P = 0.049 for AUC), without any effect of metformin or interaction between metformin and sitagliptin. However, the AUC_insulin/AUC_glucose ratio was increased by metformin (P = 0.019) and tended to increase with sitagliptin (P = 0.065), without any interaction between them (Fig. 3B, Table 3).

Fasting insulin was greater in obese than in lean subjects (P = 0.005) and tended to be greater for type 2 diabetic than for lean subjects (P = 0.07), without any difference between obese and type 2 diabetic subjects (Table 1). HOMA-IR was greater in obese and type 2 diabetic subjects than in lean subjects (P < 0.05 for each), without any difference between obese and type 2 diabetic subjects (Table 1). During ID glucose infusion, serum insulin on control days was greater for obese than for type 2 diabetic subjects (P = 0.020) and tended to be greater for obese than for lean subjects (P = 0.073), without any difference between lean and type 2 diabetic subjects (Table 2). Both fasting insulin and the AUC during ID glucose infusion were positively related to BMI (r = 0.67 and 0.62, respectively, P < 0.001 for each). The increase in the AUC_insulin/AUC_glucose ratio was positively related to the rise in intact GIP (r = 0.33, P < 0.05), but not GLP-1, concentration.

The numbers of antral waves (AWs), duodenal waves (DWs), and IPPWs in response to ID glucose infusion (t = 0–120 min) did not differ among the groups by either control or sitagliptin days (Table 2). However, the number of AWs was less in healthy lean subjects (P = 0.032) and tended to be less in obese subjects (P = 0.10) after sitagliptin than control (Fig. 4, Table 2). The number of DWs was also less after sitagliptin in healthy lean subjects (P = 0.018) (Fig. 4, Table 2). The number of IPPWs did not differ between control and sitagliptin in healthy lean or obese subjects (Fig. 4, Table 2). There was no effect of metformin or sitagliptin on any parameter in type 2 diabetic subjects (Fig. 5, Table 3).

There was no effect of sitagliptin on energy intake in healthy lean or obese subjects (Table 2), but metformin suppressed energy intake in type 2 diabetic subjects (P = 0.040), without any effect of sitagliptin or interaction between metformin and sitagliptin (Table 3). On both the control and the sitagliptin study days, energy intake was greater in obese than in type 2 diabetic subjects (P < 0.05 for each), without differing between the other groups (Table 2).

There are substantial interindividual variations in the rate of gastric emptying and resultant stimulation of incretin hormones in both healthy and type 2 diabetic subjects (8,9,14,15), but the impact of this on glucose lowering by DPP-4 inhibitors has not been evaluated. In the current study, we standardized glucose entry to the small intestine at 2 kcal/min, a rate known to predominantly induce GIP rather than GLP-1 secretion (8,9). This experimental model allowed for between-group comparisons of incretin stimulation and the effects of DPP-4 inhibition in response to an identical small intestinal glucose stimulus.

As expected, sitagliptin increased plasma intact GIP concentrations substantially and intact GLP-1 concentrations minimally. However, glycemic excursions were attenuated with sitagliptin in healthy lean and obese subjects but not in type 2 diabetic subjects, without any effect on plasma glucagon. In contrast, in type 2 diabetic patients, metformin reduced fasting and glucose-stimulated glycemia associated with modest augmentation of total and intact GLP-1, total GIP, and glucagon concentrations, but the addition of sitagliptin did not reduce glycemia further. These observations support the concepts that in type 2 diabetes, 1) the capacity of GIP to lower blood glucose is markedly impaired, 2) the glucose-lowering efficacy of DPP-4 inhibitors depends on the rate of nutrient delivery into the small intestine and the resultant magnitude of GLP-1 stimulation, and 3) glucose lowering by metformin does not predominantly involve the incretin axis.

GIP secretion was preserved in both obese and type 2 diabetic subjects compared with healthy lean subjects, whereas the GLP-1 response was diminished in obese subjects but not in type 2 diabetic subjects. The latter is consistent with our previous report in BMI-matched healthy and type 2 diabetic subjects (9). Plasma intact GLP-1 concentrations were less in obese subjects than in type 2 diabetic subjects and tended to be less than in healthy lean subjects, whereas plasma intact GIP concentration was greatest in type 2 diabetic subjects and showed no difference between healthy lean and obese subjects. This discrepancy is likely to reflect differences in DPP-4 activity, which is reportedly increased in obesity (34) and decreased with aging (35). In support of this, we observed positive relationships of plasma intact GLP-1 and GIP levels with age on the control study days, particularly for GIP, probably because its concentrations were greater. The mechanism by which GLP-1 secretion is attenuated in obesity is unclear. Leptin resistance may play a role. In mice made leptin resistant by a high-fat diet, both basal and oral glucose–stimulated GLP-1 concentrations were decreased (36). Alternatively, the apparent volume of distribution of GLP-1 could be greater in obesity, resulting in greater dilution.

Sitagliptin suppressed total GLP-1 and GIP concentrations in healthy lean subjects, consistent with negative feedback on incretin secretion by the intact peptides (37–39). This effect was diminished in obesity and type 2 diabetes, which might be detrimental because GIP acts on adipocytes to enhance fat deposition and impair insulin sensitivity (40).

The very modest increase in intact GLP-1 concentration after sitagliptin in each group, together with the lack of any glucagonostatic effect, suggests that GLP-1 is unlikely to play a major glucoregulatory role in the current model. The lack of glucose lowering with sitagliptin in type 2 diabetic subjects, therefore, supports the hypothesis of a defective glucoregulatory capacity of endogenous GIP in this group and is consistent with the marked impairment of glucose lowering by sitagliptin in type 2 diabetic patients after oral glucose during GLP-1 antagonism with exendin-(9-39) (20). Although insulin response to exogenous GIP can be partly restored with strict glycemic control in type 2 diabetes, its insulinotropic effect is not associated with improvement in glucose disposal during hyperglycemic clamp studies (41). In a group of patients with relatively well-controlled type 2 diabetes, exogenous GIP was shown to exert a modest effect on glucose disposal when blood glucose was clamped at ∼12 mmol/L, but it had no effect on fasting hyperglycemia (∼8 mmol/L), suggesting a glucose-dependent effect of GIP to lower glycemia (42). That the insulinotropic effect of GIP deteriorates with progression of glycemic control in type 2 diabetes does not support a major contribution of GIP to reduce glycemia with DPP-4 inhibitors. Both the obese and the type 2 diabetic subjects showed comparable insulin resistance, so the failure of the type 2 diabetic subjects to achieve a compensatory insulin response to ID glucose infusion indicates impaired β-cell function despite augmented intact GIP (2,3). The fact that DPP-4 inhibitors lower blood glucose concentrations in type 2 diabetes in other settings highlights the importance of endogenous GLP-1; when the latter is specifically stimulated by dietary strategies, such as consuming a d-xylose preload before a meal (18), the glucose-lowering capacity of DPP-4 inhibitors is augmented. To establish this concept further, it would be of interest to modify the current model to investigate the effects of DPP-4 inhibition at a rate of ID glucose infusion (e.g., 4 kcal/min) known to stimulate a much greater endogenous GLP-1 response (9).

Consistent with previous reports (25,26), plasma total and intact GLP-1 concentrations increased during metformin treatment in type 2 diabetic subjects. Metformin might increase preproglucagon gene expression in the intestine (26), but this has not been a consistent observation (25). Metformin appears not to stimulate L cells directly (43), but it could do so by way of neural pathways (43) or changes in intestinal glucose or bile acid transport (44). Our observation that GIP level is modestly increased with metformin has been reported (45), but not consistently (25,26), although the mechanism remains to be established. As expected, metformin lowered both fasting and postglucose glycemia. The fact that its stimulation of the incretins was modest and that sitagliptin did not potentiate its effects in the present model suggests that nonincretin mechanisms predominate, and indeed, the glucose-lowering effect of metformin persists in GLP-1 and/or GIP receptor knockout mice (25). Alternative actions of metformin include enhancement of insulin-mediated peripheral glucose disposal, suppression of hepatic glucose production (46), and antagonism of hepatic glucagon signaling (47). The increase in plasma glucagon that we observed after metformin could reflect a reactive response to the latter (48). The increase in the AUC_insulin/AUC_glucose ratio after metformin is likely due to improved β-cell responsiveness resulting from attenuation of glycemia (49).

Sitagliptin was associated with suppression of antral and duodenal motility in healthy lean subjects, with a tendency for fewer AWs in obese subjects, but had no effect in type 2 diabetic subjects. Gastric emptying is driven by antral and duodenal contractions acting against pyloric resistance. Because pyloric motility was not altered in any group, the modest suppression of antroduodenal motility by sitagliptin in healthy lean subjects may not have been sufficient to affect gastric emptying (19). Although sitagliptin might have a larger effect on IPPWs if endogenous GLP-1 levels were greater, we recently demonstrated that gastric emptying was not affected by sitagliptin in type 2 diabetic patients, even in the setting of augmented endogenous GLP-1 secretion (18). This might relate to the effects of peptide YY (PYY), whose degradation from PYY1-36 to PYY3-36, which has more potent effects to slow gastric emptying, is prevented by DPP-4 inhibition (50). The current data are thus in keeping with the majority of evidence that DPP-4 inhibition does not influence gastric emptying (17–19,21,22), although a recent study reported slowing of oral glucose emptying after sitagliptin in type 2 diabetic patients (20). Metformin was reported to slow gastric emptying in mice (25), but we did not observe any effect on APD motility in the current experimental setting.

Sitagliptin had no effect on energy intake, consistent with its neutral effect on weight during long-term trials (23). Again, this could be due to the modest stimulation of GLP-1 and/or the potential concomitant decrease in PYY3-36, which also has satiating effects (51). In contrast, metformin suppressed energy intake in type 2 diabetic patients consistent with its known anorectic effect and capacity to induce weight loss (52). This reduction was not associated with nausea and is probably GLP-1 independent because the increase in plasma GLP-1 after metformin was modest, and moreover, metformin suppresses food intake comparably in GLP-1 receptor knockout and wild-type mice (25).

This study has several limitations. The number of subjects was relatively small; however, effects were consistent among subjects, so increasing the sample size would be unlikely to change the outcomes substantially. The type 2 diabetic subjects were older than the subjects in the other groups. Age may not influence incretin concentrations (53) or insulinotropic actions (3), but it may affect insulin sensitivity and incretin metabolism (35).

In conclusion, the current observations are consistent with defective glucoregulatory capacity of endogenous GIP in type 2 diabetes. Further, the findings indicate that the effect of DPP-4 inhibition on glycemia is likely to depend on the release of GLP-1, which may require a threshold of gastric emptying of glucose >2 kcal/min.

Acknowledgments. The authors thank Kylie Lange (Centre of Research Excellence in Translating Nutritional Science to Good Health, University of Adelaide) for expert statistical advice and Ada Lam (Royal Adelaide Hospital Pharmacy) for the preparation and blinding of the treatments.

Funding. Merck, Sharp & Dohme supplied the study drugs and matching controls and placebos. This work was also supported by the National Health and Medical Research Council of Australia (grant no. 627129).

Duality of Interest. This investigator-initiated study was funded by Merck, Sharp & Dohme. C.F.D. has received consultant/lecture fees from companies with an interest in developing and marketing incretin-based therapies (Bristol-Myers Squibb, Boehringer Ingelheim, Eli Lilly, Merck, Novartis, and Novo Nordisk), and her spouse is employed by Merck and holds Merck stock. M.H. has participated on the advisory boards and/or in symposia for Novo Nordisk, Sanofi, Novartis, Eli Lilly, Merck, Boehringer Ingelheim, and AstraZeneca and has received honoraria for these activities. C.K.R. has received research funding from Merck, Eli Lilly, and Novartis. No other potential conflicts of interest relevant to this article were reported.

Author Contributions. T.W. contributed to the study design and coordination, subject recruitment, data collection and interpretation, statistical analysis, drafting of the manuscript, and critical review and final approval of the manuscript. J.M., M.J.B., and H.C. contributed to the data collection and critical review and final approval of the manuscript. C.F.D. performed the glucagon and incretin hormone assays and contributed to the data interpretation and critical review and final approval of the manuscript. K.L.J. and M.H. contributed to the study concept, data interpretation, and critical review and final approval of the manuscript. C.K.R. contributed to the study concept and design, data interpretation, and critical review and final approval of the manuscript. C.K.R. is the guarantor of this work and, as such, had full access to all the data in the study and takes responsibility for the integrity of the data and the accuracy of the data analysis.