Quantification of the Contribution of GLP-1 to Mediating Insulinotropic Effects of DPP-4 Inhibition With Vildagliptin in Healthy Subjects and Patients With Type 2 Diabetes Using Exendin [9-39] as a GLP-1 Receptor Antagonist

Inhibitors of dipeptidyl peptidase 4 (DPP-4) are used in the treatment of type 2 diabetes. Vildagliptin (1) and sitagliptin (2) have been shown to raise plasma levels of intact glucagon-like peptide 1 (GLP-1) and glucose-dependent insulinotropic polypeptide (GIP) after nutrient stimulation. GIP, however, has been shown to be ineffective in augmenting insulin secretion in type 2 diabetes (3–5). Thus, GLP-1 is assumed to be the main mediator of DPP-4 inhibition and its effects on stimulating insulin and suppressing glucagon secretion (1,2). Nevertheless, other potential substrates for DPP-4 include gastrointestinal hormones, neuropeptides, cytokines, and other biologically active peptides (6). Whether GLP-1 is the only or even the main mediator of DPP-4 inhibition is not clear (7,8).

Exendin [9-39] is a GLP-1 receptor antagonist (9,10) that can be used in humans to block effects of endogenously secreted GLP-1 (11,12). GLP-1–mediated (inhibitable by exendin [9-39]) and non–GLP-1–mediated portions of the clinical effects of the DPP-4 inhibitor sitagliptin have been found in glucose reduction and insulin stimulation after oral glucose in patients with type 2 diabetes (13). We used a similar approach but with the DPP-4 inhibitor vildagliptin and a mixed meal in both patients with type 2 diabetes and healthy subjects. We hypothesized that a different DPP-4 inhibitor and different nutrient stimulus, which may lead to a different release pattern of gut hormones (14) and may be emptied from the stomach with a different kinetic profile (15), would change the results as much as the choice of subjects with hyperglycemia versus normoglycemia (16). Preliminary data have been presented (17).

The study protocol was approved by the ethics committee of the Medical Faculty, Georg-August-Universität Göttingen (Göttingen, Germany). Written informed consent was obtained from all participants.

Six healthy subjects participated in a pilot study examining the blockade of the insulinotropic effects of exogenous GLP-1 [7-36] amide by exendin [9-39]. Twenty-nine healthy subjects and 32 age-, sex-, and obesity-matched patients with type 2 diabetes participated in the main study.

Inclusion criteria for the patients with type 2 diabetes were treatment with either diet/exercise or metformin, age 30–75 years (inclusive), HbA_1c 6.5–9.0 (normal range <6.1%), fasting plasma glucose 6.0–11.0 mmol/L, BMI 20.0–35.0 kg/m², and absence of significant heart, kidney (serum creatinine ≤123 μmol/L in women and ≤132 μmol/L in men), liver (transaminases less than twofold the upper limit of normal), and gastrointestinal disease. Metformin treatment was continued throughout the study. Healthy control subjects were required to have a normal oral glucose tolerance test (75 g) and no first-degree relatives with type 2 diabetes or personal history of gestational diabetes.

GLP-1 [7-36] amide of good manufacturing practice quality was Clinalfa brand (Bachem Distribution Service GmbH, Weil am Rhein, Germany). Exendin [9-39] was custom synthesized at Novartis Pharmaceuticals. Infusions were prepared with 0.9% NaCl in 1% human serum albumin (CSL Behring, Marburg, Germany).

After a screening examination, patients were treated with vildagliptin (100 mg once daily) or placebo and participated in meal tests (day 9 and 10, respectively) in a crossover design. Between the two treatment periods was a ≥5-week washout period. Details of the pilot study evaluating the ability of exendin [9-39] to inhibit insulinotropic actions of exogenous GLP-1 are presented in Supplementary Fig. 1.

On days 9 and 10 of treatment, study participants underwent a mixed meal (one scrambled egg, one slice of ham, 10 g of butter, two slices of toast, 20 g of strawberry jam, and 200 mL of unsweetened tea) test in the morning after fasting overnight. ¹³C-octanoic acid (110 μL/100 mg) was used as label (Supplementary Data).

Blood was drawn from an indwelling Teflon cannula inserted into a forearm vein and processed as previously described (18).

Glucose was measured (glucose oxidase method) with a Beckman Glucose Lab Analyzer 2 (Beckman Coulter, Munich, Germany). Insulin, C-peptide, glucagon, and total and intact GLP-1 and GIP were determined by specific immunoassays as previously described (18,19). Exendin [9-39] cross-reacted in the glucagon assay (by ∼0.014%), thus not allowing interpretation of glucagon measurements. Exendin [9-39] was determined by using an antibody raised in rabbits against exendin-4, as previously described (20).

Integration (area under the curve [AUC]) was carried out by using the trapezoidal rule. Data are presented as incremental changes above baseline or total AUCs, as indicated. Insulin secretion rates were calculated from C-peptide concentrations by using ISEC software, version 3.4a, supplied by R. Hovorka (London, U.K.) (21).

Subject characteristics are reported as mean ± SD and results as mean ± SEM. Statistical calculations were carried out as repeated-measures ANCOVA using Statistica Version 5.0 software (StatSoft [Europe], Hamburg, Germany). Experimental conditions (vildagliptin vs. placebo; exendin [9-39] vs. placebo) were used as independent fixed variables, and the respective baseline values of the dependent variable with placebo treatment were used as a covariate. If a significant influence of vildagliptin treatment or of exendin [9-39] was documented by P < 0.05 or by interaction of treatment and time (P < 0.05), values at individual time points were analyzed by ANCOVA. A two-sided P < 0.05 was taken to indicate significant differences. The primary end point reported was the ratio of the AUCs (total response including baseline values) of insulin secretory responses (insulin secretion rates calculated by deconvolution) relative to glucose concentrations as a modified insulinogenic index based on total responses over 4 h after meal ingestion. The originally prespecified primary end point (glucagon) could not be used because of cross-reactivity of exendin [9-39] in the glucagon immunoassay (Supplementary Data). Secondary end points were a similar ratio of insulin and C-peptide concentrations over glucose and the responses of glucose, insulin, C-peptide, insulin secretion rates, and total and intact GLP-1 and GIP.

Results from 32 of 34 patients with type 2 diabetes and 29 of 32 healthy control subjects originally recruited were analyzed (completers). Age, sex, and body mass were similar for both groups (Table 1).

Vildagliptin inhibited DPP-4 activity (Supplementary Table 1). In healthy subjects and patients with type 2 diabetes, vildagliptin treatment did not change the velocity of gastric emptying. However, exogenous exendin [9-39] (plasma concentrations shown in Supplementary Fig. 2) significantly accelerated gastric emptying in both groups (Supplementary Fig. 3 and Table 2).

Vildagliptin treatment increased intact, biologically active GLP-1 and GIP by two- to threefold and reduced integrated incremental responses of total GLP-1 and GIP slightly (Table 2 and Supplementary Fig. 4) in both groups. Both total and intact GLP-1 increased three- to sixfold, and both total and intact GIP increased modestly with exendin [9-39] (Table 2).

In healthy control subjects, neither glucose excursions nor meal-related insulin secretory responses were significantly influenced by vildagliptin treatment. Exendin [9-39] increased glycemic excursions and insulin secretory responses significantly (Fig. 1 and Table 2). In patients with type 2 diabetes, glucose concentrations were significantly lowered by vildagliptin treatment (Fig. 1, right panels), but integrated incremental glucose responses did not change significantly (Table 2). Exendin [9-39] administration induced a greater rise in glycemia after meal ingestion in patients with type 2 diabetes than in healthy control subjects. In patients, this was associated with a stimulation of insulin secretory responses with exendin [9-39] (Fig. 1 and Table 2).

In healthy control subjects, insulinogenic indices after meal ingestion were not changed significantly by vildagliptin treatment (Fig. 2). In contrast, in patients with type 2 diabetes, both vildagliptin treatment and exogenous exendin [9-39] had a significant influence on insulinogenic indices based on insulin, C-peptide, and insulin secretion rates after meal ingestion (Fig. 2). Insulinogenic indices significantly increased with vildagliptin treatment and decreased with exendin [9-39]. Expressing the reduction in insulinogenic indices due to exendin [9-39] as a percentage of the value determined without exendin [9-39] surprisingly indicated a lower contribution of GLP-1 after vildagliptin treatment (Fig. 2G–I).

With consideration that insulinogenic indices after placebo treatment with exendin [9-39] represent physiological responses minus any contribution of GLP-1 and that values after vildagliptin treatment in the absence of exendin [9-39] show the full DPP-4 inhibitor treatment effect, including actions mediated by GLP-1 (Table 2), the difference between these conditions defines the maximum that GLP-1 could contribute to the effects of DPP-4 inhibitor treatment. The true reduction in insulinogenic indices after vildagliptin treatment was thus expressed relative to this maximum estimate (= 100%): It was found to be ∼50% of the potential maximum whether based on the measurement of insulin, C-peptide, or insulin secretion rates (Fig. 2J).

No adverse events emerged from treatment relative to either vildagliptin treatment or administration of exogenous exendin [9-39].

The current study quantified the contribution of GLP-1 to the glucose-lowering effects of DPP-4 inhibition. Like Aulinger et al. (13), we used exendin [9-39] (9,10) to estimate the proportion due to GLP-1. Aulinger et al. used higher doses of exendin [9-39] (900 pmol ⋅ kg⁻¹ ⋅ min⁻¹) than we did (500 pmol ⋅ kg⁻¹ ⋅ min⁻¹ from −1 to 0 h and 350 pmol ⋅ kg⁻¹ ⋅ min⁻¹ thereafter). A pilot experiment indicated that the present regimen blocked the insulinotropic effects of a pharmacological dose of exogenous GLP-1 (Supplementary Fig. 1). Different from Aulinger et al., we used vildagliptin instead of sitagliptin and a mixed meal rather than oral glucose. Aulinger et al. studied subjects with type 2 diabetes with an HbA_1c of 6.2 ± 0.2% vs. 7.2 ± 0.5% in the present study; we also studied a nondiabetic control group. Aulinger et al. started DPP-4 inhibitor treatment 1 day before the experiments (after two daily doses), whereas we treated for 9–10 days. Nevertheless, the results are comparable.

Incretin hormone responses were similarly influenced by DPP-4 inhibition and exendin [9-39] in the study by Aulinger et al. (13) and the current study. In present study, no differences were found between healthy subjects and patients with type 2 diabetes (22).

In Aulinger et al. (13), gastric emptying of oral glucose (liquid) was mainly influenced by sitagliptin (no significant effect of GLP-1 receptor blockade), whereas in the current study, only exendin [9-39] significantly accelerated gastric emptying of solid components of a mixed meal without a significant influence of vildagliptin treatment, indicating a decelerating effect of endogenous GLP-1 on gastric emptying as previously described (23). This effect has an impact on the interpretation of the present results because accelerated gastric emptying led to greater glycemic excursions with exendin [9-39]. Aulinger et al. and the current study focused on the effects of DPP-4 inhibitor treatment on insulin secretion. Given the glucose-lowering effect of DPP-4 inhibitor treatment and the rise in glycemic excursions caused by GLP-1 receptor blockade, insulin secretory responses need to be interpreted relative to the glycemic rise after the stimulus (insulinogenic index). In both studies, the insulin/glucose ratio was significantly increased by DPP-4 inhibition. The GLP-1–mediated proportion was assessed as follows: The oral glucose/meal test with exendin [9-39] defines insulin secretion in the absence of GLP-1 effects. The experiments with DPP-4 inhibition define the therapeutic effect, including GLP-1– and non–GLP-1–mediated proportions; the difference between both defines the maximum range that could potentially be mediated by GLP-1. The experiment with DPP-4 inhibition and exendin [9-39] divides this range into the proportion mediated by GLP-1. The residual difference then represents the estimate of the non–GLP-1–mediated proportion (Supplementary Fig. 5). For both the glycemic excursions after oral glucose (13) or the mixed meal (Supplementary Fig. 5A and B) and the insulinogenic indices (Supplementary Fig. 5C and D), a proportion of the therapeutic effect is mediated by GLP-1, and another proportion cannot be inhibited by exendin [9-39]. Candidate mediators include GIP (24), oxyntomodulin (25), pituitary adenylate cyclase–activating peptide (26), and stromal cell-derived factor-1α (27).

The current study has some limitations. We cannot interpret the glucagon results because of some cross-reaction of exendin [9-39] in the assay. The differences in glycemic excursions and insulin secretory responses elicited by vildagliptin treatment and exendin [9-39] are small, so for an exact quantification, larger numbers of subjects may be necessary. Also of interest would be to examine longer treatment periods than 2 days as in Aulinger et al. (13) or 9–10 days, as in the present study.

In conclusion, this study suggests that not all insulinotropic effects introduced by DPP-4 inhibition (vildagliptin treatment) in patients with type 2 diabetes are mediated by GLP-1 as previously suggested by Aulinger et al. (13).

Acknowledgments. The authors acknowledge the expert technical assistance of S. Lessmann, B. Nawrodt, and K. Rotenberger (Diabeteszentrum Bad Lauterberg) and Lene Albæk and Sofie Pilgaard (University of Copenhagen). PAREXEL acknowledges laboratory assistance of M. Nickel and clinical conduction of K. Reseski, T. Körnicke, and H. Egri (PAREXEL, Berlin, Germany).

Duality of Interest. This study was supported by a grant from Novartis Pharmaceuticals, Basel, Switzerland. M.A.N. has been a member on advisory boards or has consulted with AstraZeneca, Boehringer Ingelheim, Eli Lilly and Company, GlaxoSmithKline, Hoffmann-La Roche, Menarini/Berlin-Chemie, Merck Sharp & Dohme, Novo Nordisk, and Versatis. He has received grant support from Eli Lilly and Company, Menarini/Berlin-Chemie, Merck Sharp & Dohme, Novartis Pharmaceuticals, and Ypsomed. He has also served on the speakers’ bureau of AstraZeneca, Boehringer Ingelheim, Eli Lilly and Company, Hoffmann-La Roche, Menarini/Berlin-Chemie, Merck Sharp & Dohme, and Novo Nordisk. J.K. and M.B. are employees of PAREXEL. J.J.H. has received honoraria as a member of advisory boards from GlaxoSmithKline, Novo Nordisk, and Zealand Pharma. He has consulted with Novo Nordisk and has received grant support from Novo Nordisk and Merck Sharp & Dohme. C.F.D. has received lecture and consultant fees from Bristol-Myers Squibb, Boehringer Ingelheim, Eli Lilly and Company, Merck Sharp & Dohme, Novo Nordisk, and Novartis Pharmaceuticals, and her spouse is employed by Merck Sharp & Dohme and holds stock in Merck. Y.L.H., L.K., and J.F. are employees of Novartis Pharmaceuticals and are eligible for stocks and stock options. No other potential conflicts of interest relevant to this article were reported.

Author Contributions. M.A.N. contributed to the data research, discussion, and writing, review/editing, and final approval of the manuscript. J.K. contributed to the discussion and writing, review/editing, and final approval of the manuscript. L.D.K., J.J.H., and C.F.D. contributed to the data research, discussion, and review/editing and final approval of the manuscript. M.B. contributed to the review/editing and final approval of the manuscript. Y.L.H. contributed to the data research and review/editing and final approval of the manuscript. L.K. and J.F. contributed to the discussion and review/editing and final approval of the manuscript. M.A.N. 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.