Defining the Role of GLP-1 in the Enteroinsulinar Axis in Type 2 Diabetes Using DPP-4 Inhibition and GLP-1 Receptor Blockade

Insulin release in response to oral glucose is substantially higher than in response to an isoglycemic glucose infusion. This phenomenon, termed the incretin effect (IE) (1), is mediated by two known gut-derived incretin hormones, glucagon-like peptide 1 (GLP-1) and glucose-dependent insulinotropic polypeptide (GIP). In healthy subjects, GLP-1 and GIP stimulate insulin secretion in a glucose-dependent manner and thereby contribute to postprandial euglycemia. Although the glucose-lowering and insulinotropic effects of GLP-1 are preserved to some extent in patients with type 2 diabetes (T2D), the GIP action is reported to be severely blunted or even abolished, as seen in studies using synthetic peptides (2–4).

Circulating GLP-1 and GIP are quickly inactivated by the ubiquitous enzyme dipeptidyl-peptidase-4 (DPP-4). Degradation-resistant GLP-1 receptor (GLP-1r) analogs have shown substantial glucose-lowering effects in diabetic patients (5,6). Although pharmacological actions of GLP-1 and its analogs have been tested rigorously, much less is known about the role of endogenous incretins. Exendin-9 (Ex-9), a widely used GLP-1r antagonist, is a useful tool to investigate the effects of endogenous GLP-1 (7).

Also, inhibition of the DPP-4 enzyme preventing GLP-1 and GIP degradation effectively lowers blood glucose. Although the dependence of DPP-4 action on GIP and GLP-1 has been demonstrated in mice (8), and DPP-4 inhibitors are now widely used to treat diabetes, there have been few mechanistic studies of these drugs in diabetic and nondiabetic humans. Particularly, the relative contribution of GLP-1 and GIP to insulin secretion and the IE under DPP-4 inhibition is not known.

We therefore determined the glucose-lowering effects of DPP-4 inhibition by using the DPP-4 inhibitor sitagliptin and analyzing the relative contribution of GLP-1. This was achieved by blocking its actions with a concomitant Ex-9 infusion. We furthermore used isoglycemic glucose infusions to study the IE and the effect of sitagliptin and Ex-9 on insulin secretion. In addition, we evaluated the effects of different treatments on gastric emptying (GE).

The study enrolled 27 Caucasian patients with T2D; of these, 24 (14 males) completed all four treatment periods (one consent withdrawal, and two patients with uncontrolled hyperglycemia discontinued the study). The 24 subjects had a mean duration of T2D of 6.6 ± 1.2 years and were in good glycemic control, with an average HbA_1c of 6.2 ± 0.2% (44 ± 2.2 mmol/mol). They were 61.6 ± 1.8 years old, with a mean BMI of 27.7 ± 0.9 kg/m². Patients with previous incretin-based therapies, thiazolidinediones, or insulin were excluded from the study. None had symptoms or a history of cardiac disease, gastrointestinal neuropathy, or evidence for nephropathy as assessed by microalbuminuria. A maximum of one oral antidiabetic drug was allowed, which was discontinued at least 1 week before study entry and withheld for the remainder of the study. All patients gave written informed consent, and the study protocol was approved by the University of Munich Institutional Review Board and the German Federal Institute for Drugs and Medical Devices.

The study was designed and conducted as a randomized, placebo-controlled, four-period, crossover study. Each of the four treatment periods consisted of 2 study days and was completed within 2 to maximal 4 days. On day 1, an oral glucose tolerance test (OGTT) containing 75 g of dextrose and 100 mg of ¹³C-acetate was performed over a 240-min period. On day 2, an isoglycemic glucose intravenous (ISO-IV) infusion, mimicking the glucose excursion of the OGTT%, was done to calculate the IE (9) (Supplementary Fig. 1).

The following four treatments were tested in each patient in a random fashion: 1) oral placebo and intravenous saline, 2) oral sitagliptin and intravenous saline, 3) oral placebo and intravenous Ex-9, and 4) oral sitagliptin and intravenous Ex-9. The treatment periods were separated by an interval of 4 to 14 days. Placebo or 100 mg sitagliptin was orally administered 1 day before the study and on study days 60 min before the experiment was conducted. Ex-9 infusion was started 60 min before the OGTT (t = 0) at a rate of 900 pmol/kg/min. This dose was shown to block >95% of the action of a pharmacological concentration of GLP-1 when infused together in a pilot study (10).

The experiments were conducted after a 12-h fasting period. An indwelling catheter was inserted into an antecubital vein for an intravenous infusion of Ex-9 or saline. A second catheter for blood sampling was inserted into the contralateral forearm. The hand of the respective arm was continuously warmed to exactly 40°C by using an infrared lamp regulated by a sensor-controlled biothermostat to obtain arterialized venous blood samples (“heated hand” technique [7]). At 60 min before the OGTT% (day 1, −60 min), Ex-9 or saline was infused and continued for 300 min (−60 to 240 min). At 0 min, the oral glucose solution was consumed within 5 min, and blood samples were drawn at regular intervals for the determination of glucose, insulin, c-peptide, glucagon, active GLP-1, and active GIP. A ¹³CO₂ breath test was also performed to monitor GE velocity (11,12). Breath samples for measuring of ¹³CO₂ exhalation were obtained before (−10 min) glucose ingestion and every 10 min thereafter (−10 to 240 min).

During the corresponding ISO-IV experiment (day 2), a variable glucose infusion was started at 0 min with a background infusion of saline. Blood glucose was monitored at 5-min intervals, and the glucose infusion was adjusted to match the glucose excursion during the OGTT. Blood samples were withdrawn regularly as indicated above.

Ex(9-39)acetate was purchased as a lyophilized sterile powder at pharmaceutical grade from Bachem (Clinalfa Products, Läufelfingen, Switzerland). Sitagliptin and its placebo were provided by Merck & Co., Inc., Rahway, NJ.

Glucose was measured using the glucose oxidase method (Glucose Analyzer; HemoCue GmbH, Ängelholm, Sweden). Blood samples were collected in chilled EDTA tubes containing 500 units of aprotinin and 50 μL diprotin A (3 mmol/L) per milliliter of blood. The blood samples were immediately placed in ice slurry and centrifuged within 30 min after withdrawal. The plasma was separated and stored at −30°C until assayed. The immunoreactivities of plasma insulin, plasma c-peptide, and GLP-1(7-36) were measured by sandwich immunoluminescence assays using specific monoclonal antibodies for capture and detection, as previously described (13): The GLP-1 assay cross-reacts 100% with human active GLP-1(7-36)-amide with no measurable cross-reactivity with GLP-1(7-37)-amide, GLP-1(9-36)-amide, GLP-2(1-33)-amide, GIP(3-42)-amide, glucagon(1-29)-amide, and Ex(9-39)-amide. The lower detection limit is 0.4 pmol/L. Intra- and interassay coefficients of variation (CV) are <6% and <15%, respectively. Glucagon was analyzed by commercially available radioimmunoassay kits (Linco Research, St. Charles, MO). Active (N-terminal) GIP immune-reactivity was measured using polyclonal antiserum #98171. The assay shows 100% cross-reactivity with human GIP(1-42), and no measurable cross-reactivity with human GIP(3-42), GLP-1(7-36)-amide, GLP-1(9-36)-amide, GLP-2(1-33), GLP-2(3-33), and glucagon. The lower detection limit is ∼5 pmol/L. Intra- and interassay CVs are <6% and <15%, respectively (14).

Power calculations were performed based on a two-tailed paired t test at the 5% significance level. A sample size of 24 subjects ensured a power of 90% to detect an 11% difference in the IE. This calculation was based on an intersubject CV of 0.20 (15). All values are shown as mean ± SEM. Blood glucose concentrations and the plasma concentrations of hormones before the OGTT or ISO-IV are given as absolute values. The excursions of blood glucose concentrations and plasma concentrations of hormones above the individual baseline levels (t = 0 min) after the OGTT and during the ISO-IV study were calculated as incremental area under the curve (AUC) according to the trapezoidal rule. The IE was calculated as the difference between the AUC of insulin, c-peptide, and the insulin-to-glucose ratio (IGR) after the OGTT and the matching ISO-IV experiment. Normality of distribution was assessed by the Kolmogorov-Smirnov test. Effects of sitagliptin or Ex-9 were analyzed using a two-way ANOVA for repeated measures (two-factor repetition) using oral medication (placebo or sitagliptin) and intravenous infusions (saline or Ex-9) as independent factors. If ANOVA indicated a significant interaction (i.e., the effect of the oral medication may depend on the effect of intravenous infusion or vice versa), a Student-Newman-Keuls multicomparison test was performed as a post hoc test. P < 0.05 was considered statistically significant.

Table 1 summarizes the effects of oral sitagliptin and intravenous Ex-9 on fasting glucose and hormone concentrations. Treatment with sitagliptin led to a significant reduction of fasting blood glucose compared with oral placebo at −70 min (P = 0.004) and 0 min (P = 0.011). Although the blood glucose was significantly increased with placebo and sitagliptin treatment to 132 ± 5.4 (placebo + Ex-9) and 129 ± 4.5 mg/dL (sitagliptin + Ex-9) (P < 0.001 vs. saline), respectively, after 60 min of Ex-9 infusion (t = 0 min), the glucose levels remained significantly lower in the group receiving sitagliptin (P = 0.011). At 0 min, a small but significant reduction of fasting insulin (P = 0.006) but not c-peptide (P = 0.437) was observed in the sitagliptin group. Ex-9 infusion had no significant effect on fasting plasma insulin, but there was a significant reduction of c-peptide concentrations at 0 min (P = 0.023), despite higher levels of glycemia. Fasting glucagon concentrations increased significantly with Ex-9 infusion (P = 0.021) but were unchanged by sitagliptin administration. Sitagliptin, however, led to a significant increase of fasting active GLP-1 (P < 0.001) and GIP (P = 0.029), with and without Ex-9 infusion. Ex-9 further increased fasting levels of GLP-1 after 60 min of infusion during sitagliptin treatment (sitagliptin + saline vs. sitagliptin + Ex-9: 1.6 ± 0.1 vs. 2.7 ± 0.4 pmol/L, P < 0.001).

Blood glucose excursions during the ISO-IV studies matched blood glucose concentrations during the respective OGTT, with average R² values of 0.93 ± 0.009 (placebo + saline), 0.91 ± 0.01 (sitagliptin + saline), 0.94 ± 0.008 (placebo + Ex-9), and 0.94 ± 0.007 (sitagliptin + Ex-9), respectively (Supplementary Fig. 2).

During the OGTT, sitagliptin reduced the incremental AUC of the glucose excursion (P < 0.001) and also the peak incremental blood glucose (P < 0.001) compared with placebo (Table 2 and Fig. 1). Glucose time-to-peak tended to be delayed under sitagliptin (P = 0.067). Ex-9 increased glucose AUC and peak glucose concentration after the OGTT when given with or without the DPP-4 inhibitor and thereby reversed some of the glucose-lowering effects of sitagliptin. However, also during Ex-9, glucose AUC (P < 0.001) remained lower with sitagliptin compared with placebo, whereas peak glucose concentrations tended to be lower without reaching statistical significance (P = 0.057). Considering the incremental glucose AUC, the glucose-lowering effect of DPP-4 inhibition during the Ex-9 infusion was ∼50% of the sitagliptin effect during the background saline infusion (ΔAUC [placebo + saline − sitagliptin + saline] 3.4 g/dL × 240 min vs. ΔAUC [placebo + Ex-9 − sitagliptin + Ex-9] 1.7 g/dL × 240 min).

Owing to significant differences in glucose excursions after the OGTT between the four treatment regimens, additional analysis of the IGR was performed. The results were added to the raw insulin and c-peptide data (Table 2 and Figs. 2 and 3). During background infusion of saline, treatment with sitagliptin versus placebo led to a significant increase of insulin (P < 0.001) and c-peptide (P = 0.032) concentrations after the OGTT despite lower glycemia. Consequently, the IGR was markedly enhanced by sitagliptin treatment (P < 0.001). Also, during the fasting glucose ISO-IV experiment, sitagliptin significantly enhanced insulin (P = 0.003) and c-peptide (P = 0.050) concentrations. Accordingly, the IGR was also significantly higher during intravenous glucose when sitagliptin was given versus saline. Ex-9 led to lower plasma insulin (P < 0.001) and c-peptide (P = 0.012) levels despite higher glucose excursions compared with saline infusion. This led to an even stronger suppression of the IGR by Ex-9 (P < 0.001). Sitagliptin treatment during the Ex-9 infusion restored some of the insulinotropic effects and led to significantly higher insulin (P < 0.001) and c-peptide (P = 0.032) concentrations. In the ISO-IV experiments matching the OGTTs during Ex-9 infusion, higher glucose concentrations resulted in slightly but significantly higher insulin levels (P = 0.044) compared with saline infusion (no Ex-9 was given during the ISO-IV studies). Accordingly, the IGR remained unchanged compared with the ISO-IV experiment matching for background saline infusion (P = 0.758).

Treatment with sitagliptin increased the IE (ΔAUC_{OGTT − ISO-IV}) significantly based on c-peptide concentrations (P = 0.043) and IGR (P = 0.017) but failed to reach significance when insulin concentrations were compared (P = 0.108; Table 2). Infusion with Ex-9 significantly reduced the IE regardless of whether the calculation was based on insulin (P = 0.005), c-peptide (P = 0.002), or IGR (P = 0.002). The relative reduction of the IE amounted to 40–50% of the IE calculated during saline infusion under sitagliptin and placebo treatment. There was no significant interaction between the oral medication and the intravenous infusion. Thus, Ex-9 did not abolish the IE. During GLP-1r antagonism, the IE accounted for 40% (insulin), 41% (c-peptide), and 50% (IGR) of the insulin excursion with placebo and for 39% (insulin), 42% (c-peptide), and 45% (IGR) with sitagliptin.

During the placebo and saline cotreatment, the OGTT led to a small glucagon increase during the first 60 min, followed by a suppression of glucagon for the rest of the study (Table 2 and Fig. 4). Inhibition of DPP-4 activity by sitagliptin led to a markedly stronger suppression of glucagon after the OGTT compared with placebo during the first 60 min (P = 0.006) and 120 min (P = 0.055). In contrast, GLP-1r blockade by Ex-9 infusion led to a significant increase of glucagon for 60 and 120 min during placebo and sitagliptin cotreatment. During a background infusion of Ex-9, sitagliptin was not able to significantly enhance the glucagonostatic effects compared with placebo (AUC 60 and 120 min; P = 0.347 and P = 0.221, respectively).

Active levels of GLP-1 and GIP were substantially and significantly increased (2.2- and 2.6-fold, respectively) with sitagliptin after the OGTT. This increase occurred with the background infusion of saline (GLP-1, P = 0.043; GIP, P < 0.001) and also with Ex-9 (GLP-1, P < 0.001; GIP, P = 0.003). Although GIP was significantly increased by sitagliptin, this elevation was less pronounced and statistically lower under concomitant Ex-9 (Table 2 and Fig. 4C). Independently of sitagliptin, Ex-9 infusion significantly increased the excursion of active GLP-1 (P = 0.011) but not that of active GIP (P = 0.600). There was an additive effect of cotreatment with sitagliptin and Ex-9 on the active GLP-1 excursion (approximately sevenfold increase compared with placebo + saline).

An increase of active GLP-1 with DPP-4 inhibition was seen not only after the OGTT but also during the ISO-IV fasting experiments (P < 0.001 compared with placebo). This paralleled the higher insulin levels with sitagliptin during fasting hyperglycemia. In contrast, active GIP did not increase when sitagliptin was given during the intravenous glucose experiment.

Compared with placebo, sitagliptin significantly delayed all parameters of the ¹³C-acetate breath test, indicating a prolongation of GE (Table 3 and Fig. 5): it increased the lag period (i.e., the time to maximal ¹³CO₂ exhalation; P = 0.005) and the exhalation half-time (P = 0.001), and decreased the maximal exhalation velocity (P < 0.001). Accordingly, the GE coefficient as a more general parameter was decreased with sitagliptin (P < 0.001). The effect of sitagliptin on GE was completely blocked by co-infusion of the GLP-1r antagonist Ex-9.

Ex-9, when compared with saline infusion, moderately accelerated ¹³CO₂ exhalation in the early phase after the OGTT, indicated by shortening of the lag period (P = 0.004) and an increase of the maximal exhalation velocity (P = 0.007). However, Ex-9 influenced neither the exhalation half-time nor the GE coefficient.

Here, we investigated the role of GLP-1 on glucose metabolism after an OGTT with and without enhancing endogenous incretin levels by using the DPP-4 inhibitor sitagliptin and by blunting GLP-1 action using the GLP-1r antagonist Ex-9. We enrolled 24 subjects with well-controlled T2D in a randomized, placebo-controlled, crossover study design to define the relative contribution of GLP-1 to insulin secretion, glucagon suppression, and GE.

A number of studies using Ex-9 to block GLP-1 action have shown that Ex-9 is a potent inhibitor of the GLP-1r in vitro (16,17) and in vivo in animals (18,19) and humans (7). In a pilot study in healthy subjects, Ex-9 at 900 pmol/kg/min suppressed >95% of the insulinotropic effect of pharmacological doses of GLP-1 (0.4 and 1.2 pmol/kg/min) (10). Therefore, we believe we have achieved maximally possible GLP-1r blockade during our experiments with Ex-9 infusion.

After ingestion of 75 g of glucose, DPP-4 inhibition enhanced GLP-1 and GIP by ∼2.5-fold and led to a significant improvement in glucose tolerance (20). In contrast, Ex-9 impaired oral glucose tolerance, as has been observed in several previous studies in healthy subjects (21–25) and patients with T2D (13,26). The reduced glucose excursion under sitagliptin was accompanied by a significant increase of insulin and c-peptide concentrations and resulted in a higher IGR. Blunting GLP-1r action resulted in significantly lower insulin and c-peptide levels, despite higher ambient glucose concentrations, thereby significantly lowering IGR. Under Ex-9 infusion, sitagliptin treatment was able to restore some but not all of the impaired glucose tolerance. The glucose-lowering effect of DPP-4 inhibition during background infusion of Ex-9 amounted to ∼50% of that during background of saline infusion. Also, after ingestion of oral glucose, excursions of insulin and c-peptide as well as of the IGR were significantly higher with sitagliptin during saline infusion compared with Ex-9 infusion. A clear insulinotropic effect of sitagliptin remained even under GLP-1r blockade. This demonstrates for the first time that GLP-1r–dependent and –independent mechanisms are both involved in the glucose-lowering effects of DPP-4 inhibition in patients with T2D. These findings in humans are consistent with the mechanism of action determined in mice with genetic deletion of the incretin receptors. In two elegant studies of animals with targeted deletion of one or both incretin receptors, pharmacological DPP-4 inhibition reduced glycemia in animals with GLP-1r or GIPr deficiency, but not when both receptors were deleted (8,27).

Two previous studies investigated the contribution of GLP-1 to the IE in diabetic patients using duodenal meal perfusion and an oral meal during hyperglycemic clamp conditions, respectively (13,26). In contrast to these rather artificial models of postprandial metabolism, the use of a standard OGTT and a glucose ISO-IV study enabled us to determine the effect of DPP-4 inhibition and GLP-1r antagonism on the IE for the first time. Sitagliptin significantly enhanced the IE. However, sitagliptin increased not only insulin during the OGTT but also insulin levels during the respective fasting hyperglycemic clamp, during which there was no enteral stimulation for the release of gut hormones. This leads to an underestimation of the IE under DPP-4 inhibition. A recent study by Vardarli et al. (28), using the DPP-4 inhibitor vildagliptin, showed no numerical increase of the IE owing to a parallel increment of insulin after the oral and intravenous glucose tests. Although this conflicts with the findings reported here, these differences may be due to the better glucose control in our cohort (HbA_1c 6.2 vs. 7.7%) and the acute versus chronic dosing of the DPP-4 inhibitor. A possible explanation for the increase of insulin during intravenous glycemia is the significant increment in active GLP-1 levels in our study during treatment with sitagliptin even when patients were fasting. In another study using an intravenous glucose tolerance test in fasting patients, D’Alessio et al. (29) demonstrated an insulinotropic effect of vildagliptin with barely increased incretin hormone concentrations. Thus, alternate mechanisms, such as protection of GLP-1 within the local gut environment facilitating a gut–brain axis of GLP-1 signaling with glucose-lowering effects, may be responsible for some of the insulinotropic effects of DPP-4 inhibition (30–32).

It is noteworthy that this cohort of T2D subjects all showed a substantial IE (∼60%), despite previous reports demonstrating an impairment of the IE in patients with T2D (33). This is probably due to the good glycemic control in our subjects, as indicated by the low HbA_1c levels. This is in accordance with previous studies suggesting that β-cell responsiveness to incretin hormones strongly depends on general β-cell function (34–36). GLP-1r blockade led to an ∼50% reduction of the IE in our T2D patients. Thus, the contribution of endogenous GLP-1 to the IE after an OGTT is very similar to those of our previous study using a duodenal mixed meal (13). During treatment with sitagliptin, Ex-9 reduced the IE by ∼50%; however, a significant IE remained with the DPP-4 inhibitor despite GLP-1r blockade. This underlines the notion that in T2D subjects, additional factors other than GLP-1 also contribute to the IE at DPP-4 inhibitor action. Moreover, it shows for the first time that this GLP-1–independent incretin action is under regulation of DPP-4. One may speculate about the existence of hitherto unknown insulinotropic signal peptides that are released postprandially. But evidently, GIP is such a factor that may partly explain the remaining effects under GLP-1r blockade seen here, despite reports of strongly attenuated insulinotropic effects of synthetic GIP in diabetic patients (2,4). Other studies have demonstrated at least a partial recovery of GIP’s insulinotropic action after reducing glucose toxicity with intensive insulin treatment (3). Because a specific GIPr antagonist is not available for human use, explanations for the remaining IE beyond GLP-1 remain speculative.

Suppression of glucagon is an important mechanism by which GLP-1 maintains glucose homeostasis in healthy subjects (7,24) and diabetic patients (13). In our study, the OGTT stimulated glucagon release during the first 60 min and suppressed it thereafter. Treatment with sitagliptin strongly and significantly suppressed glucagon release, and this contributed to the lower glucose excursion. GLP-1r blockade with Ex-9 substantially increased glucagon levels after the OGTT. Most importantly, the glucagonostatic effect of sitagliptin was absent under GLP-1r blockade. We were therefore able to show that endogenous GLP-1 has an important role in suppression of glucagon in subjects with T2D. It seems very likely that the glucagonostatic effect of DPP-4 inhibition is mediated exclusively by GLP-1.

Active levels of GLP-1 and GIP were increased as expected under DPP-4 inhibition. Interestingly, the GIP AUC was lower with Ex-9 than with saline when sitagliptin was given before the OGTT. We believe that this was due to the faster rate of GE during Ex-9, which resulted in a higher and earlier peak of GIP, followed by a more rapid decline. GIP secretion depends on the rate of glucose absorption in the duodenum and upper jejunum and not the mere presence of nutrients in the small intestine (37). The higher initial emptying rate may have escaped the duodenal glucose absorption rate (∼1.44 kcal/min [38]), thus exceeding the capacity of GIP secretion. In accordance, a duodenal perfusion of glucose at 3 compared with 2 kcal/min did not further increase GIP secretion during the first 60 min (39). The amount of glucose remaining in the stomach 1 h after the OGTT, however, may have been too small to stimulate GIP secretion to the same extent as with the slower emptying rate under sitagliptin alone, resulting in a decreased GIP AUC. However, there was an additional increase of active GLP-1 concentrations by Ex-9. This well-known effect has been described under GLP-1r blockade in healthy subjects and patients with type 1 or type 2 diabetes (13,24–26,40). It occurs during the postprandial but not in the fasting state and is believed to be due to negative feedback regulation of GLP-1 release in intestinal L-cells and the lack of this feedback during GLP-1r blockade. Particularly, the high GLP-1 levels during the combination of sitagliptin and Ex-9 underline the importance of a sufficiently high concentration of Ex-9 to minimize residual effects of circulating GLP-1 during such experiments.

We used a ¹³C-actetate breath test that reliably measures GE of liquids (11). GE was slightly accelerated under GLP-1r blockade; vice versa, sitagliptin moderately delayed GE. This action must be solely mediated by GLP-1 because coadministration of Ex-9 completely abolished this effect. This indicates GLP-1 as an inhibitory regulator of GE in patients with T2D. As has been shown in several previous studies (41–44), GE is a determinant of the early postprandial rise of blood glucose. Also in the present study, the acceleration of GE during Ex-9 was associated with a higher and earlier glucose peak, whereas the deceleration under sitagliptin was accompanied by a lower and later peak. Two other studies using scintigraphy and addressing the role of endogenous GLP-1 on GE in healthy subjects showed conflicting results without an effect (24) or inhibition of GE by endogenous GLP-1 (23). Acute administration of the DPP-4 inhibitor vildagliptin delayed scintigraphically measured GE in diabetic patients (45), whereas no effect was found after a 2-day dosing with sitagliptin in healthy subjects (46) or a 10-day treatment with vildagliptin in T2D subjects (47). Clearly, endogenous GLP-1 is a determinant of gastroduodenal motility (25,48). However, most of the evidence suggests that GLP-1 is not a major regulator of GE and that its effects may depend on meal size and composition as well as on the ambient glucose concentrations and the presence of diabetes. Even a moderate increase in glycemia within the physiological range slows GE (49). One may speculate on tachyphylaxis under continuous DPP-4 inhibition, but no solid data exist controlling for GE under long-term treatment. The only study addressing this issue used pharmacological doses of GLP-1 and demonstrated that in the short-term, GE was further inhibited by GLP-1 (50).

A limitation of this study is that there remains uncertainty about the capability of Ex-9 to completely block GLP-1rs that are not easily accessible by the circulation (i.e., within the gut or in the central nervous system) (32). Also, the acute administration of the DPP-4 inhibitor does not allow us to uncritically extend our findings to chronic treatment. The cohort in our study was relatively lean and had good glycemic control; thus, their physiology may not be representative for other patients with T2D. We used an OGTT as the classic test meal for glycemic control in T2D. With regard to GE, the use of glucose as the test meal may not allow us to extend our findings to solid meals. However, we believe that despite these inherent limitations, we have provided important new insights into how glucose is regulated by endogenous GLP-1 and how DPP-4 inhibitors mediate improvements in postprandial glycemia.

In the presented study, we have shown for the first time that after an OGTT, endogenous GLP-1 regulates glucose homeostasis in patients with T2D by multiple effects: an increase in insulin secretion thereby maintaining a substantial IE, by suppression of glucagon, and an inhibition of GE. Treatment with the DPP-4 inhibitor sitagliptin reduced the glycemic excursion after an oral glucose challenge by means of augmenting insulin, suppressing glucagon, and slowing GE. Furthermore, sitagliptin enhanced the IE. It increased levels of active GLP-1 during fasting, leading to higher insulin concentrations even during intravenous hyperglycemia. Although the effects of sitagliptin on glucagon and GE were abolished by GLP-1r blockade, a partial but considerable effect on glucose-lowering and insulinotropic actions was maintained. This suggests that DPP-4–sensitive factors beyond circulating GLP-1 substantially contribute to the IE in these well-controlled T2D patients. GIP may be a likely candidate. Further research is necessary discerning these effects to fully understand the glucose-lowering actions of sitagliptin and other DPP-4 inhibitors.

Acknowledgments. The authors thank Gerald Spöttl, Department of Internal Medicine II, Ludwig-Maximilians University of Munich, for technical support, and Rita Schinkmann and Silke Knopp from the Department of Internal Medicine II, Ludwig-Maximilians University of Munich for their excellent technical assistance.

Duality of Interest. These studies were supported by an unrestricted educational grant of Merck & Co., Inc., Rahway, NJ. No other potential conflicts of interest relevant to this article were reported.

Author Contributions. B.A.A. wrote the manuscript and discussed the data. A.B. and G.K. researched data. J.d.H. and B.G. contributed to discussion and reviewed and edited the manuscript. J.J.H. measured active GIP, contributed to discussion, and reviewed and edited the manuscript. J.S. researched data and wrote the manuscript. J.S. is the guarantor of this work, and, as such, had full access to all the data in the study and takes full responsibility for the integrity of data and the accuracy of data analysis.