Control of Postprandial Hyperglycemia: Optimal use of short-acting insulin secretagogues

The study enrolled 20 subjects (10 men and 10 women) with non-insulin-requiring type 2 diabetes of at least 6-months duration. Volunteers attended a screening visit at which a history was obtained and physical, electrocardiogram, and screening laboratory tests were performed. Eligible patients were managed with diet and exercise or oral hypoglycemic agents and had reasonable glycemic control and a BMI <40 kg/m². Subjects were excluded if they had significantly abnormal vital signs, electrocardiogram tracings, or hematological, electrolyte, or liver laboratory results at screening. Other important exclusion criteria were pregnancy, chronic insulin treatment, medications that affect liver metabolism, active substance abuse, use of oral corticosteroids, known sensitivity to nateglinide or glipizide, or a history of gastroparesis. All patients gave written informed consent before participating in the study. The study protocol received approval from the University of New Mexico institutional review board.

This study had a randomized, placebo-controlled, crossover design. Participants were admitted overnight to the General Clinical Research Center for stabilization on four separate occasions. Subjects were fed a standardized snack at 10:00 p.m. and then fasted until morning. Participants’ usual diabetes medications were withheld on the morning of the studies. In random order, 120 mg nateglinide (15 min premeal), 10 mg glipizide (30 min premeal), 10 mg glipizide plus 120 mg nateglinide (15 and 30 min premeal, respectively), or placebo pills (15 and 30 min premeal) were administered in a double-blind fashion. After baseline blood draws, subjects ate a standardized 8-kcal/kg breakfast between 8:00 and 8:15 a.m. The breakfast was prepared in the metabolic kitchen and consisted of an English muffin, bacon, scrambled eggs, and a noncaloric decaffeinated beverage. A test meal for a 70-kg subject had an energy content of 560 kcal and a nutrient content of 24.5 g protein, 16.25 g fat, and 61.5 g carbohydrates. Arterialized venous blood was drawn for glucose, insulin, and C-peptide analysis at −0.05, 0, 0.5, 1, 2, 3, and 4 h relative to the meal.

Study measurements were performed in the General Clinical Research Center laboratory. Plasma was separated from blood elements by centrifugation immediately after sampling and frozen at −70°C for later determination. If blood glucose values were <4.4 mmol/l (80 mg/dl) as measured with a bedside glucometer, plasma glucose levels were determined immediately. Plasma glucose was measured using the glucose oxidase method (Analox Instruments USA, Lunenburg, MA), and serum insulin and C-peptide was measured using chemiluminescence with an Immulite instrument (DPC Cirrus, Randolph, NJ).

Excursions for the study measurements were expressed as the change from the baseline fasting blood level. Insulin and C-peptide excursions were corrected for the fasting level on the morning of the placebo studies. Areas under the curve (AUCs) for glucose, insulin, and C-peptide excursions were calculated using the linear trapezoidal rule. The primary efficacy variables assessed were 1) overall postprandial hyperglycemia, as determined by the glucose excursion AUC, and 2) peak postprandial glucose excursion. Secondary variables included plasma insulin and C-peptide levels. Data were analyzed using SAS (SAS Institute, Cary, NC). Parameters were compared among the different groups using ANOVA for repeated measures with post hoc Student’s t test for paired data. All data are reported as the means ± SEM.

A total of 10 men and 10 women (9 Caucasian, 8 Hispanic, 2 Native American, and 1 African-American) were studied. Age (mean ± SD), BMI, and duration of diabetes were 60 ± 10 years, 32 ± 5 kg/m², and 9 ± 8 years, respectively. The subjects had a mean HbA_1c of 7.2 ± 1.4%. Of the patients, 2 were managed with diet and exercise, and the remaining 18 were taking one or two oral hypoglycemic agents for diabetes control (n = 14 for sulfonylureas, n = 6 for metformin, and n = 3 for thiazolidinediones). No changes were made to the participants’ diabetes medications during the study period, apart from omission of usually prescribed medications on the study mornings.

The fasting plasma glucose level was lower immediately prebreakfast (time = 0) in the glipizide plus nateglinide study condition (7.1 ± 0.5 mmol/l) compared with placebo (7.7 ± 0.5 mmol/l, P = 0.02) (Table 1). Postmeal glucose profiles did not differ significantly between glipizide, nateglinide, or glipizide plus nateglinide (Fig. 1A). The peak postprandial glucose excursion above baseline was higher with placebo (6.1 ± 0.5 mmol/l) than glipizide (4.3 ± 0.6 mmol/l, P = 0.002), nateglinide (4.2 ± 0.4 mmol/l, P = 0.001), or glipizide plus nateglinide (4.1 ± 0.5 mmol/l, P = 0.001). The AUC for the glucose excursion was also higher with placebo (14.1 ± 1.8 mmol/h · l) compared with glipizide (6.9 ± 2.4 mmol/h · l, P = 0.004), nateglinide (9.7 ± 2 mmol/h · l, P = 0.02), or glipizide plus nateglinide (5.6 ± 2.2 mmol/h · l, P = 0.002). At 4 h postmeal, plasma glucose levels were not significantly different between the nateglinide (9 ± 0.9 mmol/l) and placebo (9.6 ± 1 mmol/l) groups, and 4-h glucose levels for both of these groups remained significantly higher than the fasting baseline glucose level (7.8 ± 0.6 mmol/l, P = 0.04). For glipizide (7.6 ± 0.9 mmol/l) or glipizide plus nateglinide (7.4 ± 0.8 mmol/l), the 4-h postmeal plasma glucose was significantly lower than either placebo (P = 0.003 vs. glipizide, P = 0.006 vs. glipizide plus nateglinide) or nateglinide (P = 0.02 vs. glipizide, P = 0.009 vs. glipizide plus nateglinide) and had returned to premeal glucose levels (7.4 mmol/l for glipizide, 7.1 mmol/l for glipizide plus nateglinide). Using ANOVA, we found no study arm sequence effect on the glucose response.

Peak and integrated insulin excursions were higher with glipizide (peak 566 ± 130 pmol/l, AUC 1,556 ± 349 pmol/h · l) than placebo (peak 401 ± 71 pmol/l, P = 0.01; AUC 1,097 ± 211 pmol/h · l, P = 0.009) or nateglinide (peak 493 ± 125 pmol/l, P = 0.02; AUC 1,364 ± 231 pmol/h · l, P = 0.04). The combination of glipizide plus nateglinide produced similar integrated insulin levels compared with glipizide or nateglinide alone. Absolute or baseline corrected peak or integrated plasma insulin excursion levels did not differ between nateglinide and placebo. However, nateglinide (682 ± 22) pmol/h · l) did result in significantly higher insulin levels in the first 2 h postmeal [AUC insulin _{(0–2 h)}] compared with placebo (481 ± 25) pmol/h · l, P = 0.02). Compared with placebo, early insulin secretion as measured by the insulin levels at 30 min postmeal was significantly higher when either of the two oral hypoglycemic agents was given (overall P = 0.001). This difference persisted for glipizide or glipizide plus nateglinide when the insulin levels were corrected for the ambient plasma glucose levels. Using ANOVA, we found no differences in the insulin data when subjects were treated with sulfonylurea compared with nonsulfonlyurea therapy between study days. C-peptide changes postmeal mirrored the changes in insulin levels (Fig. 1C). The highest C-peptide excursion levels were measured after the administration of glipizide, and the lowest levels were measured after placebo.

Six episodes of postprandial hypoglycemia (plasma glucose <3.3 mmol/l) occurred. The same three participants experienced hypoglycemia during the glipizide-only and glipizide-plus-nateglinide studies. For two of these subjects, diabetes management consisted of diet and exercise only. Two of the three hypoglycemic episodes in the glipizide-only studies were symptomatic and required treatment with oral glucose. The hypoglycemic values were recorded at 2–3 h postmeal. There were no episodes of severe hypoglycemia (plasma glucose <2.2 mmol/l).

This study demonstrates that the acute premeal administration of 10 mg glipizide has equal efficacy compared with 120 mg nateglinide in controlling postbreakfast hyperglycemia in type 2 diabetes. These results are in contrast to a previous study by Hollander et al. (16), who found that the long-acting sulfonylurea glyburide was less effective than nateglinide in stimulating early insulin release and in limiting mealtime glycemic excursions. However, in Hollander et al.’s study, nateglinide was administered at the optimal time premeal, whereas glyburide (which has a delayed onset), was also administered 10 min before the test meals. Previous studies suggest that glyburide, like glipizide, may be more effective when given before rather than with breakfast (23). In the current protocol, glipizide was administered 30 min before the test meals as recommended in the package insert (24,25). In this setting, glipizide alone or glipizide plus nateglinide blunted the postprandial glucose rise and facilitated the return of glucose to the premeal baseline level by 4 h after a standardized mixed-composition meal. In doing so, glipizide resulted in a lower postprandial glucose nadir, causing hypoglycemia in 3 of 20 (15%) of the study subjects. Two of these three hypoglycemic episodes occurred in drug-naive patients, whereas only one hypoglycemic episode was recorded in a subject previously treated with oral hypoglycemic agents. More prolonged postprandial sampling may have demonstrated a further fall in glucose levels to below fasting levels in more of the subjects when treated with glipizide.

In contrast to glipizide, plasma glucose levels had not fallen to fasting levels by 4 h postmeal when nateglinide was administered before the test meal. It is possible that continued postprandial sampling may have allowed time for subjects to return to baseline fasting glucose levels in the nateglinide arm. However, this longer sampling time would not reflect the typical duration of the interprandial period between breakfast and lunch for most patients. The postprandial glucose nadir was higher with nateglinide than with glipizide. However, no subject experienced hypoglycemia after preprandial dosing of nateglinide.

Glipizide caused a more pronounced and sustained postmeal insulin secretory response than nateglinide. However, when comparing the insulin profiles of glipizide and nateglinide, there was no difference in insulin levels achieved during the first 30 min postmeal. Therefore, when glipizide is administered 30 min premeal, it is as effective as nateglinide in stimulating early insulin release in subjects with type 2 diabetes. However, the proportion of insulin secreted during the first 2 h postmeal was more pronounced with nateglinide than with glipizide. These differences in insulin secretion explain the higher incidence of postprandial hyperglycemia with glipizide and the waning effectiveness of nateglinide in the control of late postprandial hyperglycemia.

The coadministration of nateglinide and glipizide resulted in increased early insulin release compared with nateglinide alone but did not significantly change the postprandial glucose profile compared with the administration of either agent. Because sulfonylureas and nateglinide act at the same K⁺ATPase channels on the pancreatic β-cell, it is possible that competitive antagonism occurred when the two agents were given together. The C-peptide profiles paralleled the postprandial insulin profiles, suggesting that all of the plasma insulin measured was released directly from the pancreas and that neither agent altered the peripheral availability or hepatic metabolism of insulin, as has been suggested in previous studies with glipizide (21).

In conclusion, we have demonstrated that glipizide taken 30 min premeal can ameliorate the defect in early phase insulin release, which plays a critical role in the control of postprandial hyperglycemia in type 2 diabetes. Moreover, glipizide has a glucose-lowering efficacy similar to nateglinide when administered acutely before a standardized solid meal. Postmeal hypoglycemia may be more frequent with glipizide compared with nateglinide, particularly in drug-naive diabetic subjects. However, the insulin secretagogue and glucose-lowering effects of nateglinide appear to be insufficient to provide sustained postprandial glycemic control lasting until the next scheduled mealtime dose. We hypothesize that the time of administration and pharmacokinetic properties of individual insulin secretagogues are major determinants of their effectiveness in the control of postmeal hyperglycemia. When administered 30 min premeal, the generic sulfonylurea glipizide has equal efficacy to nateglinide in the control of postmeal glucose excursions. Therefore, the clinical decision to use glipizide versus nateglinide should be based on factors other than the control of postprandial hyperglycemia in type 2 diabetes. These factors include the relative cost of the medications, the presence of sulfonylurea allergy, the hazards of delayed hyperinsulinemia (such as postmeal hypoglycemia and potential weight gain), and the relative inconvenience of taking a medication 30 versus 10 min before meals.

This research was supported by the University of New Mexico General Clinical Research Center (National Institutes of Health, National Center for Research Resources, General Clinical Research Center Grant 5 MO1-RR00997).

The authors acknowledge the expert assistance of Carolyn King in the preparation of this manuscript for publication.