These findings imply that there may be a narrow range of plasma concentrations below which sulfonylureas are ineffective and above which there is little additional, or even reduced, effect. However, no placebo-controlled therapeutic study relating the sulfonylurea dose to its blood-glucose-lowering effect has been done. To provide such information we performed a randomized, placebo-controlled, double-blind, crossover study examining the blood glucose lowering effect of glipizide, 10, 20, and 40 mg, given daily for 3 months each to 23 patients with noninsulin-dependent diabetes mellitus.

Methods

Top Methods Results Discussion Author & Article Info References

Twenty-four patients with noninsulin-dependent diabetes mellitus (11 women and 13 men) were randomized. Their mean age was 61 years (range, 47 to 71 years), body mass index was 27.5 (range, 20.5 to 38.3 kg/m²), and duration of disease was 8.2 years (range, 2 to 18 years). The mean HbA_1c was 8.2 (range, 5.8 to 11.0; reference range, 4.0% to 6.0%). Three patients had been treated with diet, 13 with sulfonylurea, and 8 with a combination of sulfonylurea and metformin.

Nine patients were treated for hypertension and two for coronary heart disease. Medications for these conditions were not changed during the study. One patient was withdrawn after 4 weeks of test medication due to worsening of his coronary heart disease.

Treatment Protocol

During a run-in period of 2 weeks, diabetes medication was stopped, and a placebo tablet given twice daily. Fasting plasma glucose was measured weekly at the outpatient department, and blood glucose was measured at home twice a week, four times daily before meals and at bedtime. Thereafter, treatment with glipizide was given in 4-day periods with 10, 20, or 40 mg daily, respectively. The 20-mg and 40-mg doses were divided in the morning before breakfast and before dinner. Home-monitored blood glucose measurements were done daily, and patients were seen every 4 days. Patients were allowed to remain in the study only if they did not have hypoglycemic reactions and if their fasting blood glucose did not fall below 3.5 mmol/L during this phase. None of the patients had hypoglycemic events or other adverse effects during this test period.

After the run-in period, the patients entered the schedule and were randomized to a double-blind, double-blind, cross-over study consisting of three periods of 3 months during which glipizide, either 10 mg (morning), 20 mg (morning and evening), or 40 mg (morning and evening) was given daily. The dose order for the 3-month periods was randomized within blocks of three consecutive patients to be either 10-20-40 mg, 20-40-10 mg, or 40-10-20 mg daily. Medication compliance was assessed by counting the number of tablets returned monthly. The patients had taken 92% to 100% of the prescribed tablets. All patients received similar dietary instructions and were instructed not to change diet or physical activities during the study period.

Blood samples for fasting blood glucose, HbA_1c, insulin, glipizide, cholesterol, high-density lipoprotein (HDL) cholesterol, and triglycerides were drawn each month after a 12-hour overnight fast. Samples for insulin were stored at –20°C until analyzed.

During the test periods, all patients measured blood glucose before breakfast, lunch, dinner, and at bedtime 2 days a week. The home-monitored blood glucose values were collected monthly, and the arithmetic averages of 1 month's values were recorded. The patients reported between 86% and 100% of the requested number of measurements.

After each treatment period, ß-cell function was tested by a test meal containing 20 g, whole wheat bread; 6 g, margarine; 40 g, cheese; 100 mL, low-fat milk; 120 mL, orange juice; and a cup of coffee; energy content, 625 kcal. After obtaining three basal samples at –30,-10, and 0 minutes for the measurement of serum glipizide, plasma glucose, and serum insulin concentrations, the test meal was ingested within 10 minutes, and further blood samples were drawn at 30-minute intervals for 4 hours.

Laboratory Tests

Hemoglobin A_1c was determined by liquid chromatography; serum cholesterol and triglyceride levels were determined enzymatically (Boehringer Ingelheim, am Rhein, FRG), and HDL-cholesterol after ultracentrifugation [7]. Serum glipizide levels were determined by liquid chromatography [8]. Serum insulin [9] was determined by a double-antibody radio immunoassay (sensitivity of the assay, 5 pmol/L). The interassay coefficients of variation were 11%. Home-monitored blood glucose was measured four times a day, twice a week using the Glucometer II device and Glucostix test strips (Bayer Diagnostics, Leverkusen, FRG). All patients had been trained by a diabetes nurse, and the test results were collected monthly.

Statistical Methods

All values are presented as mean and SE. Differences between treatment periods were analyzed by analysis of variance using the Tukey HSD-test for multiple comparisons of correlated samples. Logarithmic values were used if data were not normally distributed. Alternatively, the Wilcoxon signed-rank test was used for pairwise comparisons. Correlations were calculated by least-squares regression analysis, and for data that were not normally distributed, the tests were either done after logarithmic transformation or Spearman rank correlation was used. Statistical analysis was done using the statistics program SYSTAT.

Results

Top Methods Results Discussion Author & Article Info References

Serum Glipizide

Monthly measured predose serum glipizide levels varied (P < 0.001) according to dosage from 33 nmol/L (10 mg), 170 nmol/L (20 mg), and 306 nmol/L (40 mg). Peak glipizide concentrations were reached 60 minutes after intake of the drug. The glipizide increases after intake of the 10-mg doses (10 mg in the morning or 10 mg in the morning and evening) were similar: 637 and 773 nmol/L, respectively. The increase after the 20-mg morning dose was higher, 1326 nmol/L, than after the 10-mg dose (P = 0.009).

Glycemic Control

The mean home-monitored blood glucose was higher at the end of the drug-free run-in period than during glipizide treatment (P < 0.006 to < 0.001; Figure 1, Table 1. The glipizide dose increase from 10 mg/d to 20 or 40 mg/d did not further improve glycemic control. The HbA_1c concentrations at the end of the three drug periods were 8.1%, 7.8%, and 7.7% [P > 0.2]. Home-monitored blood glucose concentrations from the preceding 3-month period correlated strongly with HbA_1c values at the end of each period (r = 0.82, P < 0.001).

View larger version (16K): [in this window] [in a new window]   Figure 1. Average home-monitored blood glucose. Values were measured before breakfast, lunch and dinner and at bedtime. n.s. = not significant.

When the patients were stratified by previous treatment, no difference was observed between patients treated with sulfonylurea alone (n = 16) and patients treated with sulfonylurea and metformin (n = 7). In particular, no effect could be observed by increasing the drug in either group. Four hypoglycemic events, verified by blood glucose measurements below 3.0 mmol/L, were reported by three patients, one patient taking 10 mg glipizide per day and two patients, 40 mg per day.

Beta-Cell Function

Beta-cell function was assessed by measuring serum insulin concentrations during a test meal after the drug-free run-in period and after each 3-month treatment period Figure 2, Table 2. The insulin response was smaller without than with glipizide (P = 0.001). The greatest insulin response was observed during treatment with 10 and 20 mg of glipizide daily. During treatment with 40 mg daily, the response was significantly weaker compared with 10 and 20 mg (P = 0.04 and 0.02, respectively).

View larger version (25K): [in this window] [in a new window]   Figure 2. Glucose and insulin response curve to a test meal. The glipizide amount given 30 minutes before the test meal was 10 mg after treatment with 10 and 20 mg/d and 20 mg after 40 mg/d (———0 mg; – · – 10 mg; — — — 20 mg; – – – – 40 mg). Each symbol represents the mean of 23 patients. For statistical comparisons, see Table 2..

The glucose response was higher after placebo treatment than after glipizide treatment Figure 2, Table 2; P = 0.02 to 0.03). The three treatment doses did not differ with respect to the glucose responses to the meal. The differences in the insulin responses correlated with differences in the glucose response during the latter part of the test meal (r = 0.334, see Table 2.

Other Variables

No statistical changes were observed in serum cholesterol, HDL-cholesterol, or triglycerides during the treatment periods (Table 3). Body weight also did not change during the study. Routine hematologic, kidney function, and liver enzyme tests were done before the test period and after each 3-month treatment period. During drug treatment, no adverse effects were observed in serum values of aspartate aminotransferase, alkaline phosphatase, bilirubin, lactate dehydrogenase, creatinine, urea, sodium, potassium, and chloride.

Discussion

Top Methods Results Discussion Author & Article Info References

Sulfonylurea treatment improved mean blood glucose and fasting blood glucose measured at home. During the entire study, the fasting blood glucose monitored either at home or at the clinic showed a strong correlation to the HbA_1c concentration, indicating that both measurements provided good estimates of overall glucose control. To mimic the clinical recommendations, the 10-mg/d dose was given once daily whereas the 20-mg/d and 40-mg/d doses were divided. This allowed comparison of the effects of 10 and 20 mg in the morning as well as of a second dose in the evening. Even though glipizide levels increased by increasing dose, neither a morning dose increase from 10 to 20 mg nor addition of an evening dose of 10 or 20 mg improved insulin or glucose levels.

Because individual responses differed so markedly between patients, it is possible that a dose-dependent effect was masked. In addition, the study patients were relatively hyperglycemic and may have represented cases unlikely to respond to sulfonylurea treatment. We do not believe, however, that these reasons explain our findings.

First, the results were analyzed by stratifying the patients into groups on the basis of results during the run-in period. Stratification for fasting plasma glucose or average home-monitored glucose, HbA_1c, meal-induced insulin response, body mass index, serum lipids, or mode of treatment before entering the study did not reveal any differences, and the differences between the test doses were identical for all variables tested. Second, the sulfonylurea effect in patients with mild disease was not better, indicated by low HbA_1c values, showing that the effect was independent of pretreatment glycemia. Finally, a second dose of glipizide in the evening did not further improve fasting insulin or glucose concentrations.

These findings indicate that when the glipizide dose is increased over a certain threshold around 10 mg, the resulting higher plasma levels do not further improve glucose control. In keeping with previous studies [10, 11], it is also apparent that dividing the glipizide dose does not further improve glucose control.

Despite the dose-dependent increase in plasma glipizide concentrations, the rise in plasma insulin concentrations and the reduction in plasma glucose concentrations were smaller with the higher glipizide doses than with the 10-mg dose. This finding may relate to the mode of action of sulfonylureas. These drugs seem to stimulate insulin secretion by a receptor-like mechanism. They bind to selective membrane sites and help to close cAMP-dependent potassium channels. This, in turn, results in calcium ion influx and exocytosis of insulin granules [12]. Saturation of this binding process at higher sulfonylurea concentrations could explain the lack of effect of higher sulfonylurea dosages. This agrees with the finding that glipizide dose increase from 15 to 25 mg did not improve, but rather impaired, glucose control [5]. Long-term treatment with another sulfonylurea, tolazamide, has been shown to desensitize the ß cell to sulfonylurea activation [13]. The limited efficacy is not unique to glipizide. Experiments with glyburide suggest that the insulinotropic effect reaches a plateau at plasma glyburide concentrations of 100 to 200 nmol/L, corresponding to a dose of about 10 mg/d [6].

Apart from the lack of efficacy of the currently recommended maximum sulfonylurea dose, a dose increase from 20 to 40 mg/d would double the treatment costs. Second, although not observed in the current study, the side effects may increase with increasing dosage with some sulfonylureas. Finally, high doses may impair ß cell function [5, 13].

Given these findings, it seems rational to reconsider recommended maximum doses of sulfonylurea. From the clinical point of view, rather than increasing the dose when moderate doses of sulfonylurea do not work, it is probably better to consider insulin therapy.