Improved Meal-Related β-Cell Function and Insulin Sensitivity by the Dipeptidyl Peptidase-IV Inhibitor Vildagliptin in Metformin-Treated Patients With Type 2 Diabetes Over 1Year

This was a multicenter, randomized, double-blind, placebo-controlled study that compared the effects of 52-week treatment with vildagliptin (50 mg once daily) and placebo in patients with type 2 diabetes continuing a stable dosage of metformin (1,500–3,000 mg/day). Male or infertile female patients aged ≥30 years diagnosed with type 2 diabetes at least 6 months before enrollment and treated with a stable dosage of metformin for ≥3 months could be included. Prerandomization A1C while on metformin monotherapy was required to be between 7.0 and 9.5%, inclusive, and baseline BMI was to be between 20 and 35 kg/m², inclusive. Patients were excluded if they had a history of type 1 or secondary forms of diabetes, significant diabetes complications, clinically significant cardiovascular abnormalities or other diseases affecting carbohydrate metabolism. Patients were also excluded if they had fasting triglyceride levels >5.1 mmol/l or were treated with any drug considered possibly able to affect results or their interpretation. A total of 107 patients were included in the study. They were randomized to vildagliptin (50 mg once daily)/metformin (VM, n = 56) or placebo/metformin (PM, n = 51). After a 12-week core study, patients and study sites were asked to continue throughout a 40-week extension period. Forty-two VM and 29 PM patients participated in this extension. The treatment with vildagliptin versus placebo was masked both in the core study and in the extension phase. The patient disposition of the study population and the reasons for drop outs have been reported previously (9). A total of 57 patients completed the 52-week study period with all tests (31 in the VM group and 26 in the PM group). Table 1 shows the clinical characteristics of these patients.

Before randomization and at 12, 24, and 52 weeks after randomization, a standardized meal test was given. The patients thereby fasted overnight and study drug was administered 30 min before consumption of a standardized 465-kcal breakfast meal. The meal was consumed within 15 min, and samples were obtained at specific time points. Here we report the insulin and C-peptide responses to the breakfast ingestion. The clinical results, including A1C data and results on insulinogenic index, along with safety and tolerability data have been published previously (9).

All samples were analyzed at a central laboratory (Medical Research Laboratories International, Zaventem, Belgium) using standardized procedures. Insulin and C-peptide were measured by radioimmunoassay (Boehringer Mannheim, Mannheim, Germany) and glucose was measured with a glucose oxidase technique.

Areas under the concentration curves (AUC) of glucose, insulin, and C-peptide during the standardized meal tests were calculated with the trapezoidal rule. Prehepatic insulin delivery was estimated as the suprabasal 30-min AUC of C-peptide divided by the 30-min increase in circulating glucose. Insulin sensitivity was estimated from a model of glucose clearance, which provides OGIS, a validated index of insulin sensitivity (12). The adaptation index, which gives a figure of the ability of the β-cell to adapt insulin secretion to the ambient insulin sensitivity (13), was calculated as the product between prehepatic insulin delivery and OGIS. We also determined the conventional homeostasis model assessment of insulin resistance (HOMA-IR) from the fasting levels of glucose and insulin (14). Data and results are reported as means ± SE and statistically evaluated with ANOVA; comparisons were performed both within the respective groups before versus after 52 weeks of treatment and between the two groups. The relationship between change in insulin secretion and change in A1C after 52 weeks of treatment was assessed using the Spearman correlation coefficient.

Written informed consent was obtained from all patients and renewed before participation in the extension. The institutional review board/independent ethics committee at each site approved the protocol. The study was conducted with good clinical practice in accordance with the Declaration of Helsinki.

A1C decreased in the VM group from 7.6 ± 0.1 to 7.1 ± 0.1% (P = 0.004) and increased in the PM group from 7.7 ± 0.1 to 8.3 ± 0.2% (P = 0.012), the between-group difference being −1.0 ± 0.2% (P < 0.001). In 7 of the 31 patients in the VM group, A1C reached the target of 6.5% (23%) after 52 weeks, whereas A1C did not reach 6.5% in any patient in the PM group after 52 weeks. Fasting glucose was reduced after 52 weeks in the VM group from 9.6 ± 0.3 to 9.1 ± 0.3 mmol/l (P = 0.012). In the PM group, fasting glucose was 10.1 ± 0.4 mmol/l before treatment and 10.4 ± 0.5 mmol/l after 52 weeks (NS), the between-group difference being −0.9 ± 0.3 mmol/l (P = 0.016). In contrast, fasting insulin or C-peptide levels did not change significantly in any of the two groups. Thus in the VM group, fasting insulin was 71 ± 8 pmol/l before treatment and 72 ± 8 pmol/l after 52 weeks (NS) and in the PM group, fasting insulin was 84 ± 10 pmol/l before treatment and 71 ± 6 pmol/l after 52 weeks (NS). Based on fasting glucose and insulin levels in the VM group, HOMA-IR was 30.9 ± 4.2 mmol/l glucose × pmol/l insulin before treatment and 29.1 ± 3.7 mmol/l glucose × pmol/l insulin after treatment (NS) and the corresponding values in the PM group were 39.4 ± 5.7 before treatment and 33.4 ± 3.3 mmol/l glucose × pmol/l insulin after 52 weeks (NS). Further, in the VM group, fasting C-peptide was 0.29 ± 0.02 nmol/l before treatment and 0.31 ± 0.02 nmol/l after 52 weeks (NS) and in the PM group, fasting C-peptide was 0.30 ± 0.02 nmol/l before treatment and 0.28 ± 0.02 nmol/l after 52 weeks (NS).

Figure 1 shows the glucose, insulin, and C-peptide responses during the standardized breakfast at baseline and after 52 weeks of treatment in the VM and PM groups, respectively. AUC_glucose was reduced in the VM group from 557 ± 68 to 372 ± 77 mmol/l 240 min after 52 weeks (P < 0.001), whereas AUC_glucose was increased in the PM group from 539 ± 79 to 610 ± 89 mmol/l 240 min (P < 0.001). The between-group difference in AUC_glucose was −256 ± 23 mmol/l 240 min (P < 0.001). AUC_insulin was increased in the VM group (from 26.2 ± 2.8 to 30.3 ± 2.8 pmol/l 240 min, P = 0.016) but reduced in the PM group (from 28.7 ± 3.6 to 24.3 ± 3.0 pmol/l 240 min, P = 0.036); the between-group difference was 8.5 ± 2.6 pmol/l 240 min (P = 0.018). Similarly, AUC_C-peptide was increased in the VM group (from 60.8 ± 5.4 to 71.6 ± 5.6 pmol/l 240 min, P = 0.015) and reduced in the PM group (from 59.6 ± 8.0 to 40.2 ± 4.9 pmol/l 240 min, P = 0.021); the between-group difference was 30.2 ± 7.7 pmol/l 240 min (P = 0.019).

Figure 2 shows meal-related β-cell function, insulin sensitivity, and adaptation index at baseline and after 12, 24, and 52 weeks of treatment with VM or PM. Insulin secretion from the suprabasal 30-min AUC of C-peptide after breakfast divided by the 30-min increase in glucose was increased in the VM group from 0.036 ± 0.02 to 0.042 ± 0.003 (pmol/l 30 min)/(mmol/l) after 52 weeks (P = 0.012) but decreased in the PM group from 0.037 ± 0.002 to 0.032 ± 0.002 (pmol/l 30 min)/(mmol/l) (P = 0.036). The between-group difference in 52-week change in insulin secretion was 0.011 ± 0.03 (pmol/l 30 min)/(mmol/l) (P = 0.018). Also, the OGIS was increased in the VM group (from 248 ± 10 to 279 ± 9 ml · min⁻¹ · m⁻² after 52 weeks (P = 0.012) but was not significantly altered in the PM group (from 242 ± 7 to 237 ± 8 ml · min⁻¹ · m⁻², NS). The between-group difference in 52-week change in insulin sensitivity was 27 ± 4 ml · min⁻¹ · m⁻² (P = 0.036). Adaptation index (insulin secretion × insulin sensitivity) increased in the VM group from 9.3 ± 0.6 to 11.4 ± 0.8 after 52 weeks (P = 0.003) and decreased in the PM group from 8.8 ± 0.5 to 7.3 ± 0.6 after 52 weeks (P = 0.017); the between-group difference in 52-week change in adaptation index was 3.2 ± 0.1 (P = 0.040). The change in adaptation index across the entire study group showed a negative correlation with change in A1C (r = −0.39, P = 0.04; Fig. 2), i.e., the highest increase in insulin secretion showed the highest reduction in A1C.

We have previously shown that when added to metformin treatment, vildagliptin (previously LAF237) is effective at improving glycemic control for 52 weeks in patients with type 2 diabetes in association with favorable tolerability and safety (9). Since DPP-4 inhibition prevents the inactivation of GLP-1, enhanced prandial levels of active GLP-1 during DPP-4 inhibition are seen in subjects with type 2 diabetes (8). This would suggest that effects of GLP-1 might underlie the antidiabetic actions of DPP-4 inhibition. However, the mechanisms of the improved glycemic control after long-term DPP-4 inhibition remain to be established. In this study, we explored whether vildagliptin during 52-week treatment, when added to metformin, affects β-cell function and insulin sensitivity.

We previously demonstrated that the so-called insulinogenic index (relating the increase in insulin to the increase in glucose after meal ingestion) was improved after 52-week treatment with vildagliptin in metformin-treated patients with type 2 diabetes (9). In the present study, we proceeded by estimating insulin secretion by determining the C-peptide responses to meal ingestion in relation to the increase in glucose. We found that meal-related insulin secretion was augmented in the VM group compared with the PM group. The most pronounced action of vildagliptin was seen during the first 12 weeks of treatment. However, the increase in insulin secretion was sustained throughout the study period. This suggests that vildagliptin improves β-cell function when added to metformin in subjects with type 2 diabetes. This would be similar as the action of GLP-1 to increase insulin secretion under a variety of conditions (2). An important fact that has been discovered during recent years is that insulin secretion has to be judged in relation to insulin sensitivity because there is an inverse relation between these variables (11). To accomplish this, we also evaluated insulin sensitivity by estimating glucose clearance in relation to the prevailing insulin levels during the meal challenge. This was performed by calculating the OGIS index of insulin sensitivity (12). Its validity has been confirmed both against the hyperinsulinemic-euglycemic clamp in healthy, obese, and type 2 diabetic subjects (12) and against insulin sensitivity assessed with the intravenous glucose test (15,16). OGIS has already been used in several other studies (17,18). In the present study, OGIS was significantly increased in the VM group, showing improved insulin sensitivity by the DPP-4 inhibition. It has previously been proposed that short-term GLP-1 administration does not improve insulin sensitivity in healthy subjects (19) or in subjects with type 2 diabetes (20). On the other hand, long-term (6 weeks) subcutaneous administration of GLP-1 has been shown to enhance insulin sensitivity as judged by the hyperinsulinemic-euglycemic clamp technique in subjects with type 2 diabetes along with the improved glucose metabolism (21). Similarly, it has also been shown that long-term (12 weeks) DPP-4 inhibition improves insulin sensitivity in a diabetes model in rats (22). Our findings suggest that DPP-4 inhibition improves insulin sensitivity after long-term treatment in patients with type 2 diabetes. However, it cannot be ruled out that this may be an indirect action through improvement of the glycemia. It may be of interest that in spite of improved OGIS in the VM group, fasting insulin or HOMA-IR did not change significantly. This may suggest that DPP-4 inhibition more readily improves the dynamic insulin sensitivity in relation to meal intake than fasting (static) insulin sensitivity.

Considering further the inverse relation between insulin sensitivity and insulin secretion (11), an index relating the changes in posthepatic insulin levels to insulin sensitivity by multiplying the two variables was introduced by Bergman et al. (23) in the early 1980s and later explored in detail by Kahn et al. (24). This so-called disposition index is a valid index to understand the relation of the posthepatic insulin response to insulin sensitivity and has been demonstrated to be reduced in diabetic subjects (11). However, a more appropriate descriptor characterizing the relation between insulin secretion and insulin sensitivity requires a more direct measurement of prehepatic insulin secretion based on C-peptide. We have previously introduced such an index, called the adaptation index, by multiplying the sensitivity of glucose of C-peptide response by insulin sensitivity (13). The adaptation index thus displays the direct adjustment of the β-cell function to changes of insulin sensitivity and has been shown to be reduced in subjects with impaired glucose tolerance (13). Here we show that the adaptation index was augmented in the VM group when compared with the PM group, suggesting that vildagliptin augments β-cell function independent from changes in insulin sensitivity.

In conclusion, this study demonstrates amelioration of the β-cell function in type 2 diabetic subjects after vildagliptin. The main effect was observed already within the initial 12-week treatment and was sustained throughout the 52-week study period. The correspondingly ameliorated glucose tolerance and the correlation between change in adaptation index and change in A1C after 52 weeks suggests that an important mechanism underlying the improved glycemic control following 52-week treatment with vildagliptin is improved action of the β-cell.

B.A. is grateful for support from the Swedish Research Council (grant no. 6834), the Swedish Diabetes Association, Albert Påhlsson Foundation, Region Skåne, and Faculty of Medicine, Lund University.

The authors gratefully acknowledge Mona Landin-Olsson, Lund; Per-Anders Jansson, Göteborg; Maria Svensson, Umeå; Suad Efendic, Stockholm; Per Olof Ohlsson, Karlstad; and Ibe Lager, Kristianstad; as well as other participating physicians, research nurses, and the patients who participated in the study.