Clinical Care/Education/Nutrition ORIGINAL ARTICLE Comparison of the Effects of Three Insulinotropic Drugs on Plasma Insulin Levels After a Standard Meal 1 2 LAWRENCE S. COZMA, MRCP KIRSTEN W. LANGENDORG, MSC both chronic (6) and acute hyperglycemia 1 3 STEPHEN D. LUZIO, PHD THOMAS PIEBER (7) have been shown to worsen the func- 1 1 GARETH J. DUNSEATH, BSC DAVID R. OWENS, MD, FRCP tion of the endothelium. Recent prospective studies have at- tempted to assess the effects of fasting and postprandial hyperglycemia independent OBJECTIVE — To compare the effects of repaglinide, glipizide, and glibenclamide on insulin of each other. Postprandial hyperglyce- secretion and postprandial glucose after a single standard 500-kcal test meal. mia might be more important than fasting RESEARCH DESIGN AND METHODS — A total of 12 type 2 diabetic patients with hyperglycemia in predicting CVD (8). However, despite indisputable evidence early diabetes (mean HbA1c of 6.1%) and 12 matched control subjects were enrolled in this randomized, double-blind, crossover trial. Subjects received placebo, 2 mg repaglinide, 5 mg that better diabetes control reduces the glipizide, and 5 mg glibenclamide in a random fashion during the trial. Administration of each incidence of microvascular complica- drug was followed by a single standard 500-kcal test meal. A washout period of 7–12 days existed tions, a reduction in the risk of developing between the four study visits. CVD has not been proven convincingly. Hemoglobin glycosylation is influ- RESULTS — All three drugs were equally effective on the total prandial insulin secretion (area enced by both fasting and postprandial under the curve [AUC] –15 to 240 min). However, clear differences were noted in the early glucose; the latter is more strongly corre- insulin secretion (AUC –15 to 30 min); both repaglinide and glipizide increased secretion in lated with HbA values (9). Therefore, nondiabetic subjects by ϳ61 and 34%, respectively, compared with placebo. In the diabetic 1c patients, the difference versus placebo was 37 and 47%, respectively. The difference between targeting the postprandial glucose level glipizide and glibenclamide reached significance in both groups of subjects, whereas repaglinide when elevated seems logical and has been was more effective than glibenclamide only in the healthy nondiabetic subject group. All three shown to achieve better control than fo- drugs were effective in decreasing total glucose AUC in the nondiabetic and diabetic population. cusing on the fasting glucose level alone In the nondiabetic subjects, however, repaglinide was significantly more effective than gliben- (10). clamide. The differences disappeared in the diabetic subjects, probably as a result of increased Postprandial hyperglycemia is gener- prevalence of insulin resistance in this group. ated by a combination of impaired pan- creatic insulin secretion, unsuppressed CONCLUSIONS — Repaglinide and glipizide but not glibenclamide significantly enhanced ␤ hepatic glucose production, and reduced the early insulin secretion in both nondiabetic and diabetic subjects with preserved -cell glucose uptake in the periphery (11). In- function after a single standard meal. sulin secretion in normal subjects has a Diabetes Care 25:1271–1276, 2002 characteristic biphasic pattern, with an early phase lasting Ͻ10 min after food ingestion followed by a more sustained later phase of insulin release, which par- ardiovascular disease (CVD) is the (3). Dysfunction of the vascular endothe- allels the glucose absorption from the gut most important cause of morbidity lium is present early in the history of dia- (12). In type 2 diabetes, there is a loss of and mortality in type 2 diabetes (1). betes (4) and has been implicated in C the early phase and a delayed, blunted, The classical risk factors of hypertension, atherogenesis. Insulin resistance and its smoking, increased LDL, and reduced associated features are a major determi- and consequently more prolonged late HDL with increased triglycerides (2) ac- nant of the abnormal endothelial function phase (13). These changes occur very count for Ͻ50% of the excess risk of CVD in the prediabetic stages (5). However, early in the natural history of this syn- drome, and the degree of blunting relates ●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●● to the fasting plasma glucose, the so- From the 1Diabetes Research Unit, University Hospital of Wales, College of Medicine, South Glamorgan, called “Starling curve” of the pancreas Wales; 2Novo Nordisk A/S, Bagsvaerd, Denmark; and 3Med Universitatsklinik Graz, Graz, Austria. (14). The early-phase loss contributes to a Address correspondence and reprint requests to Dr. L. S. Cozma, Diabetes Research Unit, University lack of early suppression of the glucagon Hospital of Wales, College of Medicine, First Floor, Academic Centre, Llandough Hospital and Community NHS Trust, Penlan Rd., Penarth, South Glamorgan CF642XX. E-mail: [email protected]. secretion after ingestion of carbohydrates Received for publication 4 November 2001 and accepted in revised form 26 March 2002. (15), which in turn leads to continuing Thomas R. Pieber has received honoraria for speaking engagements from Novo Nordisk A/S. hepatic glucose production and an accen- Abbreviations: AUC, area under the curve; CVD, cardiovascular disease; KATP, ATP-sensitive potas- tuation of the hyperglycemia (16). A loss sium. A table elsewhere in this issue shows conventional and Syste`me International (SI) units and conversion of the early phase of secretion has been factors for many substances. shown to cause postprandial glucose in- See accompanying editorial on p. 1472. tolerance in nondiabetic subjects (17). In DIABETES CARE, VOLUME 25, NUMBER 8, AUGUST 2002 1271 Effects of insulinotropic drugs on plasma insulin levels Table 1—Baseline characteristics cross-reacted Ͻ2% with insulin and ϳ100% with proinsulin. Type 2 diabetic subjects Nondiabetic subjects Statistical analysis n 12 12 Data were analyzed using SAS 6.11 soft- Age (years) 57.0 Ϯ 8.7 56.3 Ϯ 9.3 ware on a UNIX platform. Metabolic pa- Men/women (n) 7 (58) / 5 (42) 7 (58) / 5 (42) rameters with a normal distribution are BMI 29.5 Ϯ 3.2 29 Ϯ 1.9 presented as means with SD or 95% CI. Ϯ Ϯ ϭ HbA1c baseline 6.1 1.2* 4.6 0.4 (n 11)* Nonnormally distributed data parameters Duration of diabetes (years min–max) 2.6 (0.5–6.6) N/A are shown as median with minimum and Fasting glucose baseline (mmol/l) 7.83 Ϯ 0.7* 4.84 Ϯ 0.2* maximum also given. Area under the Data are means Ϯ SD and n (%). *P Ͻ 0.05. curve (AUC) was calculated using the trapezoidal rule. Fasting levels for glu- cose, insulin, and C-peptide were calcu- Ϫ type 2 diabetes, restoration of the early tients had good control with HbA1c lated by averaging premeal values ( 30, phase using short-acting insulin analogs Յ6.5%, two had borderline control with Ϫ20, Ϫ15, and 0 min). Insulin secretion significantly improves the glucose toler- HbA1c between 6.5 and 7.5%, and two was analyzed as early phase, terminal Ͼ ance by reducing the endogenous glucose had poor control with HbA1c 7.5%. phase, and total insulin secretion by cal- output (18). However, despite the inhib- The treatment period started within culating insulin AUCs for the first 30 min, itory effects on the hepatic glucose pro- 30 days after the screening visit and con- the last 120 min, and the total 240 min, duction, the impaired glucose utilization sisted of four visits with washout periods respectively. AUC and the maximal by the insulin-resistant tissues remains of 7–12 days between them. A follow-up plasma concentration (Cmax) were loga- unchanged. visit was conducted 7–12 days after the rithmically transformed to obtain nor- Among the agents available for man- last treatment day. The maximum dura- mally distributed data. The transformed agement of type 2 diabetes, only ␣-glyco- tion of participation for each subject was end points were compared across groups sidase inhibitors (19) and short-acting 77 days from the first visit to the last visit. using ANOVA for a crossover design ac- insulinotropic agents such as repaglinide After a 10-h overnight fast, an intra- counting for sequence of treatment, sub- (20) and nateglinide (21) have definitely venous cannula was inserted and saline ject (within a treatment sequence), visit been shown to impact the postprandial infusion was started. Each subject re- (period), and treatment. Wilcoxon’s glucose levels. ceived placebo, 2 mg repaglinide, 5 mg signed-rank test was used to compare the The purpose of this study was to com- glipizide, and 5 mg glibenclamide in a individual groups. To account for multi- pare the effects on the ␤-cell function of random fashion. Administration of the ple comparisons, the Bonferroni method three insulinotropic agents available for drug was followed 15 min later by a stan- was applied. Using an overall level of sig- the treatment of patients with type 2 dia- dard 500-kcal meal tolerance test (55% nificance of 0.05, the nominal level of sig- betes: repaglinide, glipizide, and gliben- carbohydrate, 30% fat, and 15% protein). nificance was 0.05/3 ϭ 0.02 with three clamide. Blood sampling for glucose, insulin, and comparisons. C-peptide was performed at –30, –20, RESEARCH DESIGN AND –15, and 0 min before the meal. Postmeal RESULTS — All 24 subjects who were METHODS samples were collected every 10 min dur- recruited completed the study. Demo- ing the first hour, every 15 min during the graphic and glycemic parameters of the Study design second hour, and hourly thereafter for a study population are shown in Table 1. This study was a randomized, double- total of 4 h. The only significant differences between blind, crossover trial performed at two the diabetic population and the control centers, one in the U.K. and the second in group were in the fasting glucose and the Austria. A total of 12 diet-treated type 2 Analytical methods HbA1c levels.