Effects of the Dipeptidyl Peptidase-IV Inhibitor Vildagliptin on Incretin Hormones, Islet Function, and Postprandial Glycemia in Subjects With Impaired Glucose Tolerance

This was a 12-week, double-blind, randomized, placebo-controlled, parallel-group study conducted in 28 sites in the U.S. (10), Spain (5), Finland (4), the U.K. (4), Sweden (3), and Germany (2). Each subject attended a prescreening visit (week −4) during which a 75-g oral glucose tolerance test (OGTT) was performed to determine eligibility. Subjects with confirmed IGT then attended the screening visit at week −2 during which inclusion/exclusion criteria were assessed. Eligible subjects were randomly assigned at week 0 (visit 3) to receive vildagliptin (50 mg q.d.) or placebo and attended two additional study visits at weeks 4 and 12.

The study enrolled male and female subjects aged 18–80 years with IGT (FPG <7.0 mmol/l and 2-h postchallenge glucose ≥7.8 but <11.1 mmol/l) and BMI of 23–45 kg/m². Females with childbearing potential were required to use a medically approved birth control method. Subjects were excluded if they had diabetes (other than a history of gestational diabetes), a history of serious cardiovascular disease, liver disease such as cirrhosis or chronic active hepatitis, or significant renal dysfunction. Any of the following laboratory abnormalities also precluded participation: alanine aminotransferase or aspartate aminotransferase >3 times the upper limit of normal, direct bilirubin >1.3 times the upper limit of normal, serum creatinine levels ≥220 μmol/l, clinically significant thyroid-stimulating hormone values outside the normal range, or fasting triglyceride levels >7.9 mmol/l.

Standard breakfast meal tests (500 kcal; 60% carbohydrate, 30% fat, and 10% protein) were performed after an overnight fast at baseline (week 0) and at week 12 (or study end point). Study medication was not given before the meal test at baseline but was given 15 min before the meal at week 12. Samples for determination of active GLP-1 and gastric inhibitory polypeptide (GIP), glucose, insulin, C-peptide, and glucagon were obtained at times −20, 0, 15, 30, 60, 90, and 120 min, with the meal beginning immediately after the time 0 sample and consumed within 15 min. A1C was measured at weeks 0 and 12; FPG was measured at −2, 0, 4, and 12 weeks.

Body weight and vital signs were measured at each study visit, and standard hematology and biochemistry laboratory assessments were performed at screening (week −2), week 0, and week 12. All adverse events were recorded and assessed as to their severity and possible relationship to the study medication as judged by the investigator. Subjects were provided with glucose monitoring devices and supplies and instructed on their use. Hypoglycemia was defined as symptoms suggestive of low blood glucose confirmed by self-monitored blood glucose measurement <3.1 mmol/l plasma glucose equivalent.

GLP-1 was measured at Wuxi PharmaTech (Shanghai, China) by enzyme-linked immunosorbent assay with NH₂-terminally directed antisera. GIP was measured at the Panum Institute (Copenhagen, Denmark) by a radioimmunoassay with an antibody (code 98171) specific for the NH₂-terminus (12). Accordingly, the intact, biologically active forms of the incretin hormones were measured. All other laboratory assessments were made by Covance (Indianapolis, IN, and Geneva, Switzerland). Assays were performed according to standardized and validated procedures according to good laboratory practice.

Insulin secretory rate (ISR) was estimated by deconvolution of C-peptide levels and expressed per square meter of body surface area (13). The total and incremental (Δ) areas under the curve (AUC) for GLP-1, GIP, glucose, insulin, glucagon, C-peptide, and ISR were calculated with the trapezoidal method for the 0- to 2-h postmeal time interval. Insulin secretion relative to glucose (ISR AUC_{0–2 h}/glucose AUC_{0–2 h}) was calculated as a measure of β-cell function. In addition, homeostasis model assessment of insulin resistance (HOMA-IR) and the meal-derived insulin sensitivity index (ISI) were calculated. The primary efficacy variable was the change from baseline to end point (week 12 or last available postbaseline value) in the prandial plasma glucose AUC_{0–2 h}. This and all other variables were analyzed with an ANCOVA model with treatment and predefined pooled center as the classification variables and baseline value as the covariate, using two-sided tests and a statistical significance level of 0.05. Prespecified subanalyses of the primary outcome variable were also performed based on baseline BMI (<30, ≥30, and ≥35 kg/m²), age-group (<65 and ≥65 years), and sex.

All participants provided written informed consent. The protocol was approved by the independent ethics committee/institutional review board at each study site, and the study was conducted in accordance with the Declaration of Helsinki using good clinical practice.

Table 1 reports the baseline demographics, metabolic characteristics, and disposition of all randomly assigned patients. The groups were well balanced at baseline, with A1C averaging 5.9% and FPG averaging 6.1 mmol/l. Subjects were predominantly Caucasian and obese, and ∼80% had IFG as well as IGT. A similarly high percentage of patients in each treatment group (>90%) completed the study, and the reasons for discontinuations were similar for the two groups. Further details regarding patient flow from screening to end point are provided in Fig. 1 of the online appendix (available at http://dx.doi.org/10.2337/dc07-1616).

Figure 1 depicts plasma levels of active GLP-1, glucagon, insulin, and glucose at study end point. The hormone and glucose profiles during standard meal tests performed at baseline were very similar in the two groups and are not depicted. As shown in Fig. 1A, in subjects receiving placebo, active GLP-1 increased very modestly after food intake, whereas there was a marked and sustained increase in active GLP-1 in vildagliptin-treated subjects. This was also the case for plasma levels of active GIP (Fig. 2 of the online appendix). Plasma glucagon levels were substantially suppressed during the high-carbohydrate meal in subjects receiving vildagliptin relative to those receiving placebo (Fig. 1B). As shown in Fig. 1C, glucose levels increased after the meal in subjects receiving placebo and to a lesser degree in those receiving vildagliptin, and postmeal plasma insulin levels were superimposable in the two groups of subjects (Fig. 1D). However, as illustrated in the inset of Fig. 1D, insulin secretion relative to that of glucose (β-cell function) increased in vildagliptin-treated subjects and decreased modestly in subjects receiving placebo.

Table 2 summarizes the statistical comparisons of the total and incremental (suprabasal [Δ]) AUCs for incretin hormones, pancreatic hormones, and plasma glucose, as well as peak prandial glucose excursion, 2-h postprandial glucose level (PPG), and insulin secretion relative to that of glucose in subjects receiving vildagliptin or placebo. Relative to placebo, vildagliptin significantly increased the AUC_{0–2 h} and ΔAUC_{0–2 h} for active GLP-1 and GIP and significantly decreased both the AUC_{0–2 h} and the ΔAUC_{0–2 h} for glucagon and glucose. The increase in the ΔAUCs for GLP-1 and GIP represented >5- and nearly 2-fold increases, respectively. The ΔAUC for glucose decreased by ∼22% in vildagliptin-treated subjects and increased by ∼8% in those receiving placebo; thus, relative to placebo, vildagliptin decreased the glucose ΔAUC by ∼30%. There was a modest but significant decrease in the peak prandial glucose excursion (between-group difference of −0.6 mmol/l), but the decrease in postmeal 2-h PPG (between-treatment difference of −0.3 mmol/l) did not reach statistical significance.

Baseline BMI, age, and sex did not appear to influence the efficacy of vildagliptin to reduce postprandial glucose levels. In patients receiving vildagliptin, the mean changes from baseline in glucose AUC_{0–2 h} were similar in nonobese (−0.9 ± 0.3 mmol · l⁻¹ · h⁻¹), obese (−0.8 ± 0.3 mmol · l⁻¹ · h⁻¹), and severely obese (−0.8 ± 0.5 mmol · l⁻¹ · h⁻¹) subjects; in younger (−0.9 ± 0.3 mmol · l⁻¹ · h⁻¹) and older (−0.7 ± 0.3 mmol · l⁻¹ · h⁻¹) subjects; and in men (−0.8 ± 0.3 mmol · l⁻¹ · h⁻¹) and women (−0.9 ± 0.3 mmol · l⁻¹ · h⁻¹).

At baseline, HOMA-IR averaged 2.6 ± 0.2 in patients randomly assigned to vildagliptin and 2.8 ± 0.2 in those randomly assigned to placebo. HOMA-IR decreased during the 12-week treatment with vildagliptin (−0.2 ± 0.1) and to a somewhat lesser extent in patients receiving placebo (−0.1 ± 0.1); however, the between-group difference (−0.1 ± 0.2) was not statistically significant (P = 0.613). At baseline, the ISI averaged 5.2 ± 0.4 in patients randomly assigned to vildagliptin and 4.6 ± 0.3 in those randomly assigned to placebo. The ISI increased during the 12-week treatment with vildagliptin (+0.5 ± 0.2) and to a somewhat lesser degree in patients receiving placebo (0.2 ± 0.2). However, the between-treatment difference in the adjusted mean change from baseline in ISI (0.2 ± 0.3) was not statistically significant (P = 0.485).

The FPG at baseline was similar in the two treatment groups and did not change significantly by study end point. Baseline A1C was 5.9% in both treatment groups, and relative to placebo, this decreased significantly (−0.15%) during vildagliptin treatment (P < 0.001). Body weight was similar at baseline in the two treatment groups, and there was a nonsignificant trend for weight reduction in the vildagliptin treatment group.

One or more adverse events was reported by 49 subjects receiving vildagliptin (54.4%) and by 44 subjects receiving placebo (49.4%). No hypoglycemia was reported. A summary of the most common specific adverse events is provided in Table 1 of the online appendix. No specific adverse event occurred in more than four subjects in either group, and there were no notable differences in the adverse event profiles in subjects receiving vildagliptin or placebo. There was one serious adverse event in the vildagliptin group (congestive heart failure) and two serious adverse events in the placebo group (one instance of appendicitis and one instance of cellulitis). There were three discontinuations due to an adverse event in the vildagliptin group (two instances of headache and one instance of hypoesthesia) and two discontinuations due to an adverse event in the placebo group (one instance each of bronchospasm and eczema). There were no major changes from baseline to end point or between-treatment differences at end point for any biochemistry, hematology, or urinalysis parameter or vital sign, and no consistent trends over time were noted.

The present work provides the first evidence regarding the effects of a DPP-4 inhibitor in a pre-diabetic population. This study clearly established the fact that the mechanisms underlying the clinical efficacy of vildagliptin in patients with type 2 diabetes are also operant in pre-diabetes, as demonstrated by markedly increased postprandial incretin hormone responses of >5- and nearly 2-fold increases in the incremental AUCs for GLP-1 and GIP, respectively. These effects were associated with improvements in both β-cell function (increased insulin secretion relative to glucose) and α-cell function as measured by a reduction in the inappropriate glucagon release in response to a high-carbohydrate meal and, consequently, decreased prandial glucose excursions (∼30% reduction in ΔAUC for glucose). Twelve-week treatment with vildagliptin also modestly but significantly decreased A1C despite normal baseline levels, although there was no effect on FPG.

The findings of the present study in pre-diabetes regarding incretin hormones and islet function agree qualitatively with earlier studies performed in patients with type 2 diabetes, although quantitative comparisons are problematic because of differences in experimental design. Thus, in all studies in which meal tests were performed, vildagliptin greatly increased postmeal plasma levels of active GLP-1 (4,14) and GIP (4,14), decreased inappropriate glucagon secretion (4,14), and increased insulin secretion relative to glucose (4) as well as other measures of β-cell function (14,15), irrespective of dose used, treatment duration, or concomitant oral antidiabetic medication.

Vildagliptin has now been studied in a very broad spectrum of subjects, from those with pre-diabetes to patients with diabetes and mild hyperglycemia (16). It also has been studied in drug-naive patients with type 2 diabetes and moderate to severe hyperglycemia (2,3,17) and as an add-on to metformin (18), pioglitazone (19), or insulin (20). In all of these studies, vildagliptin decreased FPG, PPG, and A1C, and the magnitude of the change was proportional to the baseline value. Indeed, the effects of vildagliptin on measures of glycemic control in this study were proportional to the degree of dysregulation exhibited by the IGT study population. The reduction in A1C, although modest (a decrease of 0.15% from a baseline of 5.9%), is noteworthy because it is known that A1C correlates with total, cardiovascular disease, and ischemic heart disease mortality even within the range of normal levels (21).

Although vildagliptin has been reported to improve measures of insulin sensitivity in patients with type 2 diabetes (14), the modest trends toward decreased HOMA-IR and increased ISI observed in the present study were not statistically significant compared with placebo. This finding is probably explained by the very modest degree of hyperglycemia in the study participants, the lack of effect on FPG, and the fact that these subjects with IGT were not exceptionally insulin resistant at baseline (fasting insulin levels <70 pmol/l) (Fig. 1D).

OGTTs were performed initially (before study randomization) only to ascertain the glucose tolerance status of the study participants; this was assessed with a more physiological meal test during the study. Perhaps it may be perceived as a limitation of the present study that a repeat OGTT was not performed at the study end point to determine whether this DPP-4 inhibitor “normalized” glucose tolerance; however, an OGTT after only 12 weeks would have provided very limited information regarding “diabetes prevention” and would only reflect the blood glucose–lowering effects of vildagliptin. The results from the meal tests in this study can serve, however, as the rationale for future trials testing whether prolonged administration of vildagliptin in pre-diabetic subjects would delay the diagnosis of diabetes, as has been shown for rosiglitazone (10), metformin (9), and acarbose (8).

With regard to the potential for diabetes prevention or disease modification, incretin-based therapies are considered promising because of the preclinical evidence demonstrating that GLP-1, GLP-1 receptor agonists, and DPP-4 inhibitors (11) can inhibit apoptosis and promote β-cell proliferation, thereby increasing β-cell mass. The present findings confirm that the mechanism of action of vildagliptin is fully manifest in pre-diabetic subjects, with an incidence of adverse events comparable to that with placebo and with no weight gain. A long-term diabetes prevention trial with vildagliptin appears to be justifiable, although it should be noted that there is limited clinical experience with DPP-4 inhibitors and the potential liabilities of this mechanism may become apparent with more extensive patient exposure.

In summary, in pre-diabetic subjects, a 12-week treatment with vildagliptin (50 mg q.d.) markedly increased postmeal levels of active GLP-1 and GIP, improved both α- and β-cell function, decreased postprandial hyperglycemia, and decreased A1C levels. Vildagliptin was well tolerated and weight neutral and did not cause hypoglycemia. Therefore, we conclude that vildagliptin is a good candidate that warrants further investigation to explore its full potential in pre-diabetes and would be suitable for testing in future diabetes prevention trials.

This study was funded by Novartis Pharmaceuticals Corporation.

The authors gratefully acknowledge the investigators and staff at the 28 participating sites, Bernard Reimund of Novartis for the overall management of this trial, and Beth Dunning Lower for editorial assistance and helpful discussion. A list of investigators is provided in the online appendix.