Effects of Sitagliptin and Metformin Treatment on Incretin Hormone and Insulin Secretory Responses to Oral and “Isoglycemic” Intravenous Glucose

The study protocol was approved by the Georg August University Göttingen Ethics Committee before the study (registration number 1/4/08; date of approval: 10 October 2008). Written informed consent was obtained from all participants.

Twenty-one patients with type 2 diabetes participated in the current study. One patient withdrew due to dizziness and uncontrolled hyperglycemia. The characteristics of the 20 patients who completed the study are presented in Table 1. They were drug-naïve or treated with oral antihyperglycemic monotherapy before entering the study. Key inclusion criteria were a) Patients with type 2 diabetes not on antihyperglycemic medication or patients treated with metformin or a sulfonylurea (monotherapy), who were willing to discontinue their medication for the duration of the study; b) age in the range of 30–75 years (inclusive); c) BMI in the range of 25–35 kg/m² (inclusive); d) HbA_1c of ≥6.5 and ≤9.0% inclusive (if drug-naïve) or ≥6.0 and ≤8.5% (if being treated with metformin or sulfonyurea); e) fasting plasma glucose of ≥110 mg/dL (≥6.1 mmol/L) and ≤220 mg/dL (≤12.2 mmol/L) before and after a 2-week placebo run-in period; f) male or nonfertile female (women of childbearing potential using a medically approved birth control method could be included if they were willing to use the same method of contraception during the full course of the study); g) written informed consent to participate in the study.

At a screening visit, body height and weight were measured to calculate the BMI (Table 1), and blood was drawn in the fasting state for measurements of standard hematological and clinical chemistry parameters. Spot urine was sampled for the determination of albumin, protein, and creatinine by standard methods. Eligible patients entered a 6-week washout period if previously treated with oral antidiabetic agents. All patients entered a 2-week single-blind run-in period before randomized treatment was started.

After the run-in period, eligible patients entered the crossover study period, which consisted of four double-blind treatment periods lasting 6 days each (order randomized) with three 4-week (tolerance −3/+7 days) washout periods between treatments. The incretin effect was quantified after oral treatment with sitagliptin (100 mg daily), metformin (dose escalation day 1, 500 mg, daily; day 2, 500 mg twice daily; day 3, 500 mg three times daily; days 4 to 6, 500 mg four times daily, i.e., 2,000 mg/day), and the metformin/sitagliptin combination or matching placebo tablets.

The tests were performed in the morning after an overnight fast. On day 5 of each treatment period, an oral glucose challenge (75 g glucose and glucose oligomers; Roche O.G.T.) was given at 0 min. On day 6, 20% glucose was administered intravenously to copy the glycemic excursions obtained after the oral glucose (“isoglycemic” intravenous glucose infusion), as previously described (19). After basal blood specimens were drawn at –15 and 0 min, blood was taken at 15, 30, 45, 60, 90, 120, 180, and 240 min.

Glucose, insulin, C-peptide, total GLP-1 (C-terminally directed assay), intact GLP-1 (sandwich ELISA), total GIP (C-terminally directed assay), intact GIP (N-terminally directed assay), and glucagon were determined as previously described (19).

Integrated values (areas under the curve) were calculated using the trapezoidal rule. Incremental responses describe changes above baseline, whereas “total” responses describe the response above 0.

Insulin secretion rates were calculated from C-peptide concentrations using ISEC 3.4a software supplied by Dr. Roman Hovorka, London, U.K. (29). Population-derived coefficients of transition between compartments were used as described (30,31).

The incretin effect was calculated based on the integrated incremental responses (trapezoidal rule) of plasma insulin, C-peptide, or insulin secretion rates after oral and “isoglycemic” intravenous glucose administration. The difference was related to the respective response after oral glucose, which was taken as 100%. Therefore, incretin effects were expressed as the percent contribution to the total β-cell secretory response after oral glucose, as previously described (2,19).

Patient characteristics are reported as mean ± SD, and results are reported as mean ± SEM. The significance of any differences was assessed using repeated-measures ANCOVA or ANOVA, as appropriate, using Statistica 5.0 software (StatSoft Europe, Hamburg, Germany). When comparing experiments with the oral versus intravenous administration of glucose, the type of experiment was used as the independent variable. When the four experimental treatments were to be compared (baseline values or integrated responses), the treatments (sitagliptin or metformin, either yes or no) were used as independent variables. Mean values of the parameter of interest as determined with placebo treatment were imputed as the covariate. Results of ANCOVA are reported as P values for (A) differences by experiment, (B) differences over time, and (AB) differences due to the interaction of experiment and time. If a significant interaction was documented (P < 0.05), values at single time points were compared by one-way ANOVA for repeated measurements and Duncan's post hoc test. P values for the influence of treatment on basal and integrated parameters are presented for sitagliptin treatment, metformin treatment, or the interaction between sitagliptin and metformin treatment. A two-sided P value of <0.05 was taken to indicate significant differences.

Sitagliptin, metformin, and the sitagliptin/metformin combination led to reductions in fasting glucose. Compared with placebo (8.2 ± 0.4 mmol/L), sitagliptin treatment alone reduced fasting plasma glucose to 7.4 ± 0.3 mmol/L (P < 0.0001), metformin treatment alone reduced fasting plasma glucose to 7.1 ± 0.2 mmol/L (P < 0.0001), and the combination of sitagliptin and metformin further reduced fasting plasma glucose to 6.5 ± 0.2 mmol/L (P < 0.0001). Fasting glucose concentrations were similar before oral and intravenous glucose administration for all experiments (P = 0.30–0.82; Fig. 1). After oral glucose administration, the integrated incremental glucose concentration was reduced, relative to placebo treatment, by 26.8% with sitagliptin (P < 0.0001), by 23.4% with metformin (P < 0.0001), and by 46.4% with the metformin/sitagliptin combination (P < 0.0001; Table 2).

Fasting total GLP-1 was increased by metformin treatment (P = 0.0001) and not changed by sitagliptin treatment (P = 0.62; Fig. 2). Total GLP-1 was stimulated by oral but not by “isoglycemic” intravenous glucose (P < 0.0001; Fig. 3C). Metformin treatment increased total GLP-1 responses significantly after oral (P = 0.0026) and after “isoglycemic” intravenous glucose (P = 0.0014) in the fasting state and after glucose administration (see asterisks in Fig. 3C). In contrast, sitagliptin treatment decreased total GLP-1 responses significantly, both without and with a metformin background (P < 0.0001; Fig. 3B and D and Table 3).

Although fasting intact GLP-1 was not significantly changed by metformin treatment (Fig. 2C), fasting intact GLP-1 increased with sitagliptin treatment (P < 0.0001; Fig. 2C). Likewise, intact GLP-1 responses were significantly increased by sitagliptin treatment before and after the oral glucose load (P < 0.0001; Fig. 3F). Again, intact GLP-1 concentrations were stimulated by oral but not by “isoglycemic” intravenous glucose.

Integrated incremental responses after oral glucose increased by 1.7-fold (both without and with a metformin background; Table 3).

Metformin treatment had no significant effect on fasting total or intact GIP (Fig. 2B and D and Fig. 3L and P). However, sitagliptin raised fasting intact GIP (P < 0.0001; Fig. 2D) as well as responses before and after oral glucose (Fig. 3O). As expected, GIP concentrations were raised considerably more after oral glucose stimulation than during “isoglycemic” intravenous glucose infusions, when incretin levels remained in the basal range. Thus, intact GLP-1 and intact GIP were raised slightly, but significantly, after sitagliptin treatment. The concentrations of intact GLP-1 remained in the “basal” (i.e., low picomolar) range (Fig. 2).

As with total GLP-1 (Fig. 3A–D), total GIP responses after oral glucose were significantly reduced with sitagliptin treatment by 28% (P < 0.0001; Fig. 3I–M and Table 3), but were not significantly changed by metformin treatment.

It was possible to closely match the glucose excursions after oral and intravenous glucose administrations (Fig. 1E–H). Thus, the conditions of “isoglycemia” necessary to accurately quantify the incretin effect were met. Irrespective of treatment, oral glucose elicited a significantly higher insulin secretory response, whether based on insulin, C-peptide, or the calculation of insulin secretion rates (Fig. 1I–U and Table 4).

After oral glucose, there was no significant stimulation of insulin secretory responses (insulin, C-peptide, or insulin secretion rates) by sitagliptin or metformin treatment (Fig. 1K, L, O, P, S, and T), except for a slight increase of C-peptide with sitagliptin treatment (P = 0.022; Fig. 1O, asterisks). However, integrated incremental responses were higher with sitagliptin treatment (Table 3: insulin, P = 0.0014; C-peptide, P = 0.0032; insulin secretion rates, P = 0.0076), and not significantly changed with metformin treatment (Table 3). The stimulatory effect of sitagliptin was observed when used in monotherapy or in combination with metformin (Fig. 1 and Table 3). As a result of the glucose-lowering effects of sitagliptin and metformin treatments, these enhanced insulin secretor responses occurred at lower glucose concentrations. This applied to responses after oral glucose and “isoglycemic” intravenous glucose infusions because these were matched to those after oral glucose. Judging insulin secretory responses (insulin, C-peptide, insulin secretion rates) relative to the increment in glucose concentrations revealed a stimulation after sitagliptin treatment that occurred not only after oral glucose stimulation (when intact GLP-1 and GIP concentrations were in the “nutrient-stimulated” range of concentrations; Fig. 3) but also during “isoglycemic” intravenous glucose infusions (i.e., at just slightly elevated basal concentrations of intact GLP-1 and GIP; Figs. 2 and 3). Although responses were significantly lower with “isoglycemic” intravenous glucose infusions than after oral glucose stimulation, the pattern of stimulation by sitagliptin was similar, resulting in comparable relative increments in insulin secretory responses relative to glycemic rises for both ways of administering glucose (Table 4). When the same analysis was based in integrated incremental responses of insulin, C-peptide, or insulin secretion rates, and of glucose responses, the conclusions were unchanged (Supplementary Table 1).

Insulin secretion in response to oral glucose was higher than with “isoglycemic” intravenous glucose with all treatments (P < 0.0001; Table 2). Thus numerically, the incretin effect (placebo: 31 ± 5%; insulin secretion, calculated from C-peptide responses) was not significantly changed by sitagliptin (P = 0.84) or metformin treatment (P = 0.75, interaction P = 0.14; Table 2). The reason is that sitagliptin augmented insulin secretory responses (relative to glucose increments) not only after oral glucose stimulation but also with “isoglycemic” intravenous glucose infusions (Table 4).

Metformin raised fasting glucagon concentrations significantly (Supplementary Fig. 1). Glucagon concentrations were suppressed with oral and intravenous glucose and in a similar manner with placebo, sitagliptin, metformin, and the combination of sitagliptin and metformin treatment (Supplementary Figs. 1 and 2 and Table 4). Intravenous glucose led to a significantly earlier suppression of glucagon than did oral glucose (Supplementary Fig. 2).

The present analysis confirms that metformin leads to an augmented secretion of GLP-1, as previously shown after 2 days of treatment in nondiabetic (27) and in type 2 diabetic subjects (23). Furthermore, although DPP-4 inhibitor administration raises plasma concentrations of intact GLP-1 and GIP, the numerical contribution of the incretin effect after oral glucose is unchanged, as shown in our previous study using vildagliptin treatment (another DPP-4 inhibitor (19) and confirmed by Muscelli et al. (20) studying the effects of sitagliptin on mixed meal–stimulated insulin secretory responses compared with those after “isoglycemic” intravenous glucose. We believe this is explained by the fact that in the current study, as in our previous study (19) and that of Muscelli et al. (20), DPP-4 inhibitor administration stimulated insulin secretion not only after oral glucose (at high, nutrient-stimulated incretin hormone concentrations) but also at much lower incretin levels in the basal range (i.e., as typical for the absence of nutrient stimulation) (Fig. 3), and it did so both when studied in monotherapy and as an add-on to metformin.

This important combination had not been examined in the previous studies on this topic. This is best illustrated by analyzing the ratio of insulin secretory responses and glycemic increments, which—by definition—are very similar comparing oral glucose and “isoglycemic” intravenous glucose stimulation. The similarity of results obtained by looking at insulin, C-peptide, and insulin secretion rates (deconvolution) attests to the robustness of these analyses. The surprising observation is that the small but significant rises in fasting levels of intact, biologically active GLP-1 and GIP (Fig. 3 and Table 3) introduced by DPP-4 inhibition with sitagliptin are associated with significant augmentations in insulin secretion, not only after oral glucose stimulation but also with “isoglycemic” intravenous glucose infusions, resulting in no net change in the size of the incretin effect (Table 4). Thus, small variations in the low concentrations of basal incretin hormone levels may determine insulin secretory responses to a greater degree than previously thought. These findings challenge the view that incretins are important in the postload, nutrient-stimulated situation only, but have little if any influence as long as their plasma concentrations are low, as in the fasting state (16,17). However, we can only describe associations, and this cannot be used as a proof of causality.

Our finding is well compatible with the observation by Salehi et al. (32) that a receptor antagonist at the GLP-1 receptor is able to reduce hyperglycemia-induced insulin secretory responses in the absence of nutrient stimulation; that is, at low (“basal”) concentrations of incretin hormones, in particular GLP-1, again indicating a relatively strong insulinotropic effect at these low concentrations.

An alternative explanation could be the involvement of other mediators in addition to GLP-1. Whether GIP contributes much in patients with type 2 diabetes may be viewed as highly unlikely, given the inability of exogenous GIP to stimulate insulin secretion in these subjects (3,4). Nevertheless, there appears to be some potential interaction at the level of GLP-1– and GIP-mediated effects on insulin secretion, as indicated by the fact that the endocrine pancreas in GLP-1 receptor knockout mice hyper-responds to GIP stimulation (33). Thus, hyporesponsiveness to one incretin may cause hyper-responsiveness to the other incretin, allowing for the possibility that another manipulation of the enteroinsular axis, the inhibition of DPP-4 activity, may have consequences regarding the ability of GIP to elicit insulin secretory responses. This, at present, is a mere hypothesis that needs experimental validation.

Another explanation could be a predominant action of DPP-4 inhibition on DPP-4 in the intestinal compartment, with little or no the importance of DPP-4 inhibition in the circulation (where it affects plasma concentrations of intact GLP-1 and GIP), as shown in mouse experiments by Waget et al. (34). This would require mediation by the autonomous nervous system rather than by changes in circulating incretin concentrations. However, establishing these mechanisms experimentally in human subjects will be difficult.

Muscelli et al. (20) reported, in addition to effects on insulin secretion, that sitagliptin treatment reduced the amount of glucose appearing in the general circulation, pointing to metabolic effects over and above the changes induced in insulin and glucagon secretion.

In principle, other mediators may include hitherto undiscovered gut peptides with incretin activity or neuropeptides that reach an effective concentration only when close to the neurons they act upon, which normally does not lead to high plasma concentrations, and thus are much more difficult to identify compared with incretin hormones with a more or less humoral mode of action. To our knowledge, there is no “hot” candidate among the known neuropeptides for this role. However, the existence of non–GLP-1–mediated mechanisms of action of DPP-4 inhibitors is also suggested from studies from our laboratory (35) as well as from other groups (36) indicating that clinical effects of DPP-4 inhibition can be partially but not fully blocked by exendin, a GLP-1 receptor antagonist [9–39].

A surprising finding was a small but significant rise in fasting glucagon concentrations with metformin treatment (Supplementary Fig. 1). The difference was small, and this may be a chance finding. Alternatively, the rise in fasting glucagon may be a response to the glucose-lowering effect of metformin (Fig. 1 and Table 1). It probably was not observed with sitagliptin (lowering glucose by a similar degree) because the elevation in intact GLP-1 (Figs. 2 and 3) induced a concomitant suppression in glucagon (1).

Our study confirms the feedback inhibition that elevated intact GLP-1 concentrations exert on L-cell secretion (19,37–39). Sitagliptin reduced total GLP-1 responses by ∼50% and total GIP responses to a somewhat lesser degree, ∼30%. This was not prevented by metformin cotreatment, despite metformin’s activity to augment GLP-1 secretion (both regarding basal concentrations [Fig. 2] and oral glucose-stimulated levels [Table 3]). This feedback effect, whereby increased intact GLP-1 levels restrain L-cell secretion, is one explanation of why DPP-4 inhibitors enhance intact GLP-1 concentrations by only 2- to 3-fold (15,39), even though, in the absence of DPP-4 inhibition, intact concentrations of GLP-1 are only 10–20% of the respective total concentrations (40), indicating a potential to increase these levels by 5- to 10-fold. This raises the possibility that interfering with the mechanism of this feedback inhibition may further enhance intact GLP-1 concentrations and potentially increase the clinical effectiveness of DPP-4 inhibitors. Nevertheless, despite total incretin levels being reduced, sitagliptin and, to a greater extent, the combination with metformin led to an increase in intact GLP-1 levels, which likely contributes to the greater clinical effectiveness of the DPP-4 inhibitor plus metformin combination (41). Thus, insulin secretory responses relative to the glycemic increment were highest with sitagliptin plus metformin treatment compared with either treatment alone (Table 4; Supplementary Table 1).

In conclusion, treatment of type 2 diabetes patients with metformin increases GLP-1 secretion. Treatment with the DPP-4 inhibitor sitagliptin raises plasma concentrations of intact GLP-1 and GIP in the basal (fasting) state and after stimulation with oral glucose. This gives rise to augmented insulin secretory responses after oral glucose and “isoglycemic” intravenous glucose infusions, resulting in the numerical size of the incretin effect being unchanged. Overall, our results indicate a prominent insulinotropic effect of small variations in incretin hormone levels in the low, “basal” range, or alternatively, other mediators; that is, hitherto undescribed peptides that are subject to degradation/inactivation by DPP-4 and have the ability to stimulate insulin secretion at elevated glucose concentrations.

Acknowledgments. The excellent technical assistance of Brigitte Nawrodt, Ute Buss, and Sabine Schminkel is acknowledged (Diabeteszentrum Bad Lauterberg). The authors thank Sabine Ciossek for secretarial assistance, Sofie Pilgaard and Lene Albæk for expert technical assistance, and Michelle Valentin for logistical support.

Duality of Interest. This study was supported by a grant from Merck Research Laboratories (Rathway, NJ). Merck Sharp & Dohme (Germany) supported this study financially (investigator-initiated study). I.V. has received lecture honoraria from Berlin Chemie AG; has received travel support to an international diabetes meeting from Boehringer Ingelheim Pharma GmbH & Co. KG (Ingelheim, Germany), and MSD GmbH (Haar, Germany); and has received a consultancy fee from Eli Lilly and Company and Lilly Deutschland GmbH (Bad Homburg, Germany). C.F.D. was an advisory panel member for Eli Lilly and Company and Merck Sharp & Dohme Limited; has received consultancy fees from Eli Lilly and Company and Merck Sharp & Dohme Limited; has received lecture fees from Novo Nordisk A/S, Merck Sharp & Dohme Limited, Eli Lilly and Company, Boehringer Ingelheim Pharmaceuticals Inc., and Bristol-Myers Squibb Company; and her spouse is employed by Merck Sharp & Dohme Limited. J.J.H. was a board member for Merck Sharp & Dohme Limited and Novo Nordisk A/S and has received consultancy fees from Novo Nordisk A/S. M.A.N. was member of the advisory panel for Berlin Chemie AG, Amylin Pharmaceuticals, Inc., Boehringer Ingelheim International GmbH (Ingelheim, Germany), Eli Lilly and Company, GlaxoSmithKline, MSD GmbH, Merck Sharp & Dohme Limited, Novo Nordisk Pharma GmbH (Mainz, Germany), Sanofi Deutschland GmbH, Versartis, Inc., and Intarcia Therapeuticals, Inc.; has received consultancy fees from Amylin Pharmaceuticals, Inc., AstraZeneca (Mjölndal, Sweden), Berlin Chemie AG, Boehringer Ingelheim (Ingelheim, Germany), Diartis Pharmaceuticals, Inc. (Redwood City, CA), Eli Lilly and Company, GlaxoSmithKline, Hoffman La Roche (Basel, Switzerland), Intarcia Therapeutics, Inc., MSD GmbH, Novartis Pharmaceuticals Corporation, Janssen Global Services (Titusville, NJ), Novo Nordisk A/S, Sanofi Pharma (Bad Soden, Germany), and Versartis, Inc.; has received research support from Berlin Chemie AG, Boehringer Ingelheim (Ingelheim, Germany), Novartis Pharma (Basel, Switzerland/Nürnberg, Germany), MSD GmbH, Metacure, AstraZeneca (Södertälje, Sweden), GlaxoSmithKline, Roche Pharma AG (Grenzach/Wyhlen, Germany), Novo Nordisk Pharma GmbH (Mainz, Germany), and ToleRx; has received speaking honoraria from AstraZeneca (Mjölndal, Sweden), Berlin Chemie AG, Boehringer Ingelheim (Ingelheim, Germany), Lilly Deutschland GmbH (Bad Homburg, Germany), MSD GmbH, Novartis Pharma (Basel, Switzerland), Novo Nordisk A/S, and Roche Pharma AG (Grenzach/Wyhlen, Germany); and has other relations to Diabate/Boehringer Ingelheim (Ingelheim, Germany), MSD GmbH, Incretin Expert Program/Lilly Deutschland GmbH (Bad Homburg, Germany), and Medscape LLC (New York, NY). No other potential conflicts of interest relevant to this article were reported.

Author Contributions. I.V. and M.A.N. researched data; contributed to discussion; and wrote, reviewed, and edited the manuscript. E.A., C.F.D., and J.J.H. researched data, contributed to discussion, and reviewed and edited 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.

Prior Presentation. This study was presented at the 72nd Scientific Sessions of the American Diabetes Association, Philadelphia, PA, 8–12 June 2012.