The Efficacy of Acarbose in the Treatment of Patients with Non–Insulin-Dependent Diabetes Mellitus: A Multicenter, Controlled Clinical Trial

  1. Jean-Louis Chiasson, MD;
  2. Robert G. Josse, MBBS;
  3. John A. Hunt, MBBS;
  4. Carol Palmason, ART;
  5. N. Wilson Rodger, MD;
  6. Stuart A. Ross, MBBS;
  7. Edmond A. Ryan, MD;
  8. Meng H. Tan, MD; and
  9. Thomas M. S. Wolever*, MD, PhD
  1. From the Hopital Hotel-Dieu de Montreal and the Universite de Montreal, Montreal, Canada; St. Michael's Hospital and the University of Toronto, Toronto, Canada; Lion's Gate Hospital and the University of British Columbia, North Vancouver, Canada; Miles Canada, Inc., Toronto, Canada; St. Joseph's Health Center and the University of Western Ontario, London, Canada; Foothills Hospital and the University of Calgary, Calgary, Canada; Walter C. Mackenzie Health Sciences Center and the University of Alberta, Edmonton, Canada; and Camp Hill Medical Center and Dalhousie University, Halifax, Canada. Requests for Reprints: Jean-Louis Chiasson, MD, Centre de Recherche/Hotel-Dieu de Montreal, 3850 Rue Saint-Urbain, Montreal, Quebec, Canada H2W 1T8. Acknowledgments: The authors thank Mrs. Susanne Bordeleau-Chenier for preparation of the manuscript. Grant Support: In part by a research grant from Miles Canada, Inc.
     
    Next Section

    Abstract

    Objective: To evaluate the long-term efficacy of acarbose, an α-glucosidase inhibitor, in improving glycemic control in patients with non–insulin-dependent diabetes mellitus.

    Design: A 1-year, multicenter, randomized, double-blind, placebo-controlled study.

    Setting: Seven university-affiliated, community-based, tertiary care diabetes clinics.

    Patients: 354 patients with non–insulin-dependent diabetes mellitus were recruited; 77 were being treated with diet alone, 83 with diet and metformin, 103 with diet and sulfonylurea, and 91 with diet and insulin. Patients in each treatment group were randomly assigned to either acarbose or placebo for 1 year. Eighty-seven percent of patients receiving acarbose and 92% of those receiving placebo were included in the efficacy analysis (n = 316).

    Measurements: At baseline and at 3-month intervals, levels of hemoglobin A1c (HbA1c), fasting and postprandial plasma glucose, fasting and postprandial serum C-peptide, and fasting serum lipids were measured.

    Results: Compared with placebo, acarbose treatment caused a significant decrease in the mean postprandial plasma glucose peak (90 minutes) in all four groups (19.0 ±0.4 mmol/L to 15.5 ±0.4 mmol/L; P < 0.001). Analysis of the postprandial plasma glucose incremental area under the curve showed that the change from baseline to the end of the treatment period differed for placebo and acarbose recipients by 4.73 mmol x h/L in the diet alone group (P < 0.001), 2.06 mmol x h/L in the metformin group (P = 0.01), 2.65 mmol x h/L in the sulfonylurea group (P < 0.001), and 3.13 mmol x h/L in the insulin group (P = 0.001). Corresponding decreases in HbA1c levels occurred; these were 0.9% in the diet alone group (P = 0.005), 0.8% in the metformin group (P = 0.011), 0.9% in the sulfonylurea group (P = 0.002), and 0.4% in the insulin group (P = 0.077). Acarbose did not significantly affect mean serum C-peptide or mean serum lipid levels.

    Conclusions: Acarbose improved long-term glycemic control in patients with non–insulin-dependent diabetes mellitus regardless of concomitant antidiabetic medication.

    *For author affiliations, current author addresses, and a listing of collaborators, see end of text.

    The high morbidity and mortality and the increasing health care costs associated with non–insulin-dependent diabetes mellitus [1-4] suggest that current treatment strategies are unsatisfactory. Persisting hyperglycemia has been linked to the development of diabetic complications [5] and to the exaggeration of insulin resistance and impaired insulin secretion that characterizes the pathophysiology of non–insulin-dependent diabetes mellitus [6-9].

    Diet therapy remains the cornerstone of the treatment strategies for non–insulin-dependent diabetes mellitus; it is intended to achieve and maintain ideal body weight and to reverse, at least partially, the metabolic abnormalities associated with the disease [10, 11]. Sustained weight loss, however, is rarely attained. When diet fails, oral hypoglycemic agents such as sulfonylurea or biguanide or both are added [12, 13]. These have been shown to increase fasting plasma glucose levels, but postprandial hyperglycemia persists in more than 60% of patients and probably accounts for sustained increases in hemoglobin A1c (HbA1c) levels [14]. Furthermore, secondary failure to oral agents is common, and more than 25% of patients with non–insulin-dependent diabetes mellitus require insulin to achieve acceptable glycemic control [15]. Insulin therapy, however, may promote weight gain and exacerbate insulin resistance [15]. Slowly absorbable or lente carbohydrates and high-fiber diets have been proposed as ways to delay glucose absorption and thus to blunt the postprandial increase in plasma glucose and insulin levels [16, 17]. Although these dietary manipulations have been shown to be effective, most patients find the regimen difficult to follow.

    An alternative approach to the problem of postprandial hyperglycemia is to use competitive inhibitors of small intestine brush-border α-glucosidases such as acarbose [18-20]. Acarbose reduces postprandial plasma glucose and insulin responses [21, 22] and may improve metabolic control in non–insulin-dependent diabetes mellitus when combined with diet alone [23] or with sulfonylurea [24], but it was of no benefit when added to combined sulfonylurea and biguanide in one study [25]. However, most previous studies using acarbose were usually short or medium term, were often uncontrolled, and frequently had small numbers of patients [20]. No study has been done of acarbose in patients with non–insulin-dependent diabetes mellitus who were being treated with insulin. Therefore, the purpose of our study was to evaluate the long-term efficacy of acarbose in improving the metabolic control of patients with non–insulin-dependent diabetes mellitus that was not well controlled by diet alone, diet and metformin, diet and sulfonylurea, or diet and insulin. We also evaluated the safety and tolerability of acarbose during the 1-year study period; detailed data will be reported in a separate paper (Ross SA, Chiasson JL, Josse RG, Hunt JA, Palmason C, Rodger NW, et al. The safety of acarbose in the treatment of patients with NIDDM. A Multicenter Trial. In preparation), but a summary of our major findings is included in this report.

    Previous SectionNext Section

    Methods

    Patients with non–insulin-dependent diabetes mellitus of at least 6 months' duration, the diagnosis of which was based on World Health Organization criteria [26], were recruited from among the seven participating facilities: Camp Hill Medical Center, Halifax, Canada; Hopital Hotel-Dieu de Montreal, Montreal Canada; St. Michael's Hospital, Toronto, Canada; St. Joseph's Hospital, London, Canada; Walter C. Mackenzie Health Sciences Center, Edmonton, Canada; Foothills Hospital, Calgary, Canada; and Lion's Gate Hospital, Vancouver, Canada. Patients were recruited from the following four treatment groups: diet alone; diet and metformin; diet and sulfonylurea; and diet and insulin. Hemoglobin A1c levels at entry had to be greater than 7.0%, except in patients treated with diet alone, whose levels had to be greater than 6.5%. All patients had normal plasma creatinine levels and liver function test results. Patients with hypertension were included if their blood pressure was well controlled with antihypertensive medication; patients receiving therapy with β-blockers or thiazide diuretics were excluded. All patients with documented gastrointestinal disease and those taking medications likely to alter gut motility or absorption were excluded, as were patients taking medications to lower serum lipid levels.

    All patients were placed on a weight-maintaining diet [27] and continued to take their hypoglycemic medication. After a 6-week pretreatment period, the patients in each treatment group were randomly assigned to receive either acarbose or placebo for 1 year. Patients were asked to take study medication with the first bite of each meal three times daily. The initial dose was 50 mg and, if necessary, was titrated upward on subsequent visits to 100 mg, and finally to a maximum of 200 mg. The dose was increased if the postprandial plasma glucose level was greater than 10 mmol/L and was adjusted according to the patients' tolerance to the drug. If the postprandial plasma glucose level was 10 mmol/L or less, the concurrent hypoglycemic medication was decreased by approximately 25%. Drug and dietary compliance was verified by pill counts at each visit and by 3-day recall nutritional diaries every 3 months.

    The efficacy values were measured before random selection and every 3 months throughout the study period. At those visits, fasting blood samples were drawn for HbA1c, plasma glucose, serum C-peptide, and serum lipid profiles. The patients were then given a standard breakfast of Enrich (450 kcal; Ross Laboratories, Montreal, Canada); this liquid meal contains 55% carbohydrates (61% starch, 29% sucrose, and 10% soya polysaccharide), 30.5% lipids, and 14.5% proteins. Acarbose or placebo was taken with the first sip of the liquid meal. The test meal was ingested over 10 minutes, and blood samples were drawn at 60, 90, and 120 minutes for plasma glucose and serum C-peptide levels. The study was approved by the institutional review board at each facility, and all patients gave informed consent.

    Plasma glucose levels were measured at each facility using the hexokinase method [28]; all other values were measured in a laboratory in Toronto. The HbA1c levels were measured using the mini-column chromatography technique (Bio-Rad Laboratories, Hercules, California; normal range, 3.8% to 6.3%). Serum C-peptide levels were measured by radioimmunoassay [29]. Serum lipid levels were measured at the Lipid Research Laboratory at the University of Toronto, which is certified by the National Heart, Lung, and Blood Institute-Centers for Disease Control Lipid Standardization Program [30]. Samples were analyzed using the Technicon RA1000 and Technicon enzymatic reagents for total cholesterol (Technicon method SM4-0139G86), triglycerides (Technicon method SM4-0173G90), and triglyceride blank (reagent no. T01-2013-01) (Technicon-Miles, Mississauga, Ontario) [31]. The high-density lipoprotein (HDL) cholesterol level was measured as the cholesterol in the supernatant after precipitation of the non-HDL cholesterol using dextran sulfate magnesium chloride [31]. The low-density lipoprotein (LDL) cholesterol level was calculated for samples in which the triglyceride concentration was less than 4.5 mmol/L using the following formula: LDL = total cholesterol -(HDL + triglycerides)/2.2 [32].

    The safety and tolerability of acarbose were assessed every 3 months throughout the study. Complete blood count and biochemistry profiles were done to check for toxicity, and possible malabsorption was assessed by measuring serum vitamins A, D, folate, and B12 and the minerals Fe, Zn, Mn, Mg, Cu, Cd, and Se. When adverse events and side effects occurred, they were documented and a questionnaire on symptoms was administered. Hypoglycemic reactions (plasma glucose levels <3.5 mmol/L) were also documented. The results of the safety parameters will be presented in detail and discussed in a separate paper (Ross SA, Chiasson JL, Josse RG, Hunt JA, Palmason C, Rodger NW, et al. The safety of acarbose in the treatment of patients with NIDDM. A Multicenter Trial. In preparation.)

    Sample size was calculated using a two-tailed α of 0.025 (Bonferroni correction for two primary efficacy variables), power of 90%; the mean HbA1c difference was 1% between acarbose and placebo with a standard deviation of 1.2%. The four groups were treated as four separate studies. The total number of patients needed was calculated to be 76 for each treatment group: 38 for acarbose and 38 for placebo; thus, 304 patients were needed. Allowing for an approximately 15% dropout rate, we estimated that at least 350 patients should be enrolled.

    The primary analyses were comparisons of changes in HbA1c and plasma glucose values (incremental area under the curve, or curve area) between baseline and end of treatment. The fasting plasma glucose level was analyzed as a secondary variable. Unpaired t-tests were done to obtain P values. Patients were considered invalid for efficacy analysis if they dropped out or if they required an increase in concomitant hypoglycemic medication within 60 days of random assignment. The concomitant hypoglycemic medication was increased if patients became symptomatic or if fasting plasma glucose levels increased to 15 mmol/L or more. If these events occurred after 60 days, intent-to-treat analysis was done using the data from the last valid visit. The data on HbA1c and the plasma glucose curve area were also analyzed over time using repeated-measures analysis of covariance. The model included the following independent factors: treatment and facility as main effects, treatment by facility interaction and baseline values as covariates, and values at 3, 6, 9, and 12 months as the dependent factors [33]. Incremental area under the curve was calculated using the trapezoidal rule after subtraction of the fasting value. A responder analysis was also done in which a responder was defined as a patient who, at the end of the study, had had a decrease in HbA1c level of at least 15% of baseline value or a decrease to less than 7% (absolute value) or both without any increase from baseline. A logistic regression analysis was then done to assess whether predicting factors for responder status could be found [34]. The data are expressed as the mean ±SE or as the mean with 95% CIs.

    Previous SectionNext Section

    Results

    Three hundred fifty-four patients participated in the study and were randomly assigned to receive either acarbose (n = 172) or placebo (n = 182). Seventy-seven of the recruited patients were being treated with diet alone, 83 with diet and metformin, 103 with diet and sulfonylurea, and 91 with diet and insulin Table 1. Of the patients receiving metformin, 74.7% were taking 1500 mg/d or more, 22.9% were taking 1000 mg/d, and 2.4% were taking 500 mg/d (mean daily dose, 1506 mg). Most patients (95.1%) receiving sulfonylurea were taking glyburide; only 3.8% were taking chlorpropamide, and 1% were taking tolbutamide. Of those patients taking glyburide, 74.5% were taking 15 mg/d or more, 18.3% were taking 10.0 to 12.5 mg/d, and 7.1% were taking 7.5 mg/d or less (mean daily dose, 16.6 mg). Three hundred twenty-six patients (92%) were white, and 28 were of other ethnic origins. The mean age for all patients was 57.4 ±0.5 years, the mean body mass index was 29.0 ±0.3, and the male-to-female ratio was 1.5:1.0 (211:143). The demographic characteristics are shown in Table 1 and were similar for all treatment groups, except that the mean known duration of diabetes was 5.2 ±0.6 years for the diet alone group, 8.8 ±0.6 years for the metformin group, 9.4 ±0.7 years for the sulfonylurea group, and 12.9 ±0.8 years for the insulin treatment group.

    View this table:
    Table 1. Demographic Characteristics of Patients at the Time of Random Selection*

    The overall dropout rate was 25%, similar to that of the acarbose (27%) and the placebo (23%) groups; almost 90% of all patients, however, were included in the efficacy analysis (Table 2). The most common reasons for discontinuation were gastrointestinal symptoms such as flatulence or diarrhea; these were cited by 19% of the patients treated with acarbose and 11% of placebo recipients and occurred early in the study. Only one patient in the acarbose group and eight patients in the placebo group were dropped from the efficacy analysis because their concomitant hypoglycemic medications had had to be increased within 60 days of random assignment.

    View this table:
    Table 2. Number of Patients by Treatment Group

    The mean total daily caloric intake at baseline was 1692 ±29 and 1661 ±27 kcal in the acarbose and placebo groups, respectively, and 48% of the intake was carbohydrates, mostly complex carbohydrates. These were well maintained throughout the study; this was reflected by stable body weight during the 12-month study period both for patients treated with acarbose (from 84.5 ±1.2 to 84.2 ±1.6 kg) and those treated with placebo (from 81.3 ±1.2 to 81.4 ±1.3 kg).

    Plasma Glucose

    The mean fasting plasma glucose level at baseline was 10.5 ±0.4 mmol/L for the diet alone group, 14.2 ±0.4 mmol/L for the metformin group, 11.9 ±0.3 for the sulfonylurea group, and 11.9 ±0.4 mmol/L for the insulin group (Table 3). Throughout the study period significant changes occurred in the fasting plasma glucose levels of acarbose recipients compared with placebo recipients in the diet alone (difference, 2.1 mmol/L; P < 0.001) and in the sulfonylurea (difference, 1.4 mmol/L; P = 0.013) groups only. No significant change was seen in the metformin or the insulin treatment groups.

    View this table:
    Table 3. The Effect of Acarbose on Fasting Plasma Glucose Levels*

    The plasma glucose concentrations in response to the test meal, after 1 year of treatment, are shown in Figure 1. Compared with placebo, acarbose caused a significant decrease in the maximum postprandial plasma glucose concentration at 90 minutes from 17.7 mmol/L (95% CI, 15.9 to 19.4 mmol/L) to 13.2 mmol/L (CI, 11.4 to 15.1 mmol/L) for the diet alone group, from 19.3 mmol/L (CI, 17.4 to 21.1 mmol/L) to 15.8 mmol/L (CI, 14.0 to 17.6) for the metformin group, from 20.7 mmol/L (CI, 19.3 to 22.1 mmol/L) to 16.6 (CI, 15.5 to 17.8 mmol/L) for the sulfonylurea group, and from 18.4 mmol/L (CI, 17.5 to 19.4 mmol/L) to 15.7 mmol/L (CI, 14.7 to 16.6 mmol/L) for the insulin group. In all four groups, the postprandial plasma glucose concentrations at 60 and 120 minutes were also significantly lower in patients receiving acarbose than in those receiving placebo (P < 0.01). Figure 2 shows the postprandial plasma glucose response to the test meal as the incremental area under the curve throughout the study period. Treatment with acarbose resulted in a decrease in the plasma glucose incremental curve area in all four concurrent treatment groups when compared with placebo (P values between < 0.001 and 0.010 when analyzed by unpaired t-tests and < 0.001 when analyzed by repeated-measures analysis of covariance). A decrease of approximately 30% in incremental plasma glucose curve area had already been achieved after 3 months of treatment and was well maintained for as long as 12 months of treatment (Figure 2). Furthermore, the concomitant hypoglycemic medication was decreased in 17.7% of patients taking acarbose compared with 3% of patients taking placebo (in the metformin group, 6% of those taking acarbose had their medication decreased compared with 0% of those taking placebo; in the sulfonylurea group, it was 16% compared with 2%; in the insulin group, it was 36% compared with 6%).

    Figure 1. Data are expressed as means with 95% CIs. *  < 0.01, †  < 0.019, ‡  < 0.026. values are for differences between changes from baseline (time, 0) to end point (time, 12 months) for acarbose compared with placebo.
    View larger version:
    Figure 1. Data are expressed as means with 95% CIs. *  < 0.01, †  < 0.019, ‡  < 0.026. values are for differences between changes from baseline (time, 0) to end point (time, 12 months) for acarbose compared with placebo. The effects of acarbose (black circles) and placebo (○) after 12 months of treatment on postprandial plasma glucose concentrations in response to a test meal in groups of patients treated with diet alone (A), metformin (B), sulfonylurea (C), and insulin (D).PPPP
    Figure 2. Data are expressed as means with 95% CIs. *  ≤ 0.01. value is for differences between changes from baseline (time, 0) to end point (time, 12 months) for acarbose compared with placebo.
    View larger version:
    Figure 2. Data are expressed as means with 95% CIs. *  ≤ 0.01. value is for differences between changes from baseline (time, 0) to end point (time, 12 months) for acarbose compared with placebo. The effects of acarbose (black circles) and placebo (○) on postprandial plasma glucose incremental area under the curve in response to a test meal throughout the 1-year study period in patients treated with diet alone (A), metformin (B), sulfonylurea (C), and insulin (D).PP

    Glycosylated Hemoglobin

    The mean HbA1c level at baseline was 6.7% ±0.2% for the diet alone group, 7.8% ±0.2% for the metformin group, 8.0% ±0.2% for the sulfonylurea group, and 7.7% ±0.2% for the insulin group. At the time of random assignment, HbA1c levels were less than 7% in 70% of patients in the diet alone group, 36% of those in the metformin group, 31% of those in the sulfonylurea group, and 33% of those in the insulin group. After 12 months of treatment, the HbA1c levels were lower in patients receiving acarbose than in patients receiving placebo; the difference was 0.9% for the diet alone group (P = 0.005), 0.8% for the metformin group (P = 0.011), 0.9% for the sulfonylurea group (P < 0.002), and 0.4% for the insulin group (P = 0.077). When changes in HbA1c levels were analyzed by repeated-measures analysis of covariance, the P values for comparisons between acarbose and placebo in time were 0.030 for the diet alone group, 0.19 for the metformin group, less than 0.001 for the sulfonylurea group, and 0.011 for the insulin group. In all four groups, acarbose treatment was associated with a decrease in the mean HbA1c level, reaching maximum effect after 6 months of treatment and remaining at those levels for the rest of the study (Figure 3).

    Figure 3. Data are expressed as means with 95% CIs. *  ≤ 0.010, †  = 0.077. values are for differences between changes from baseline (time 0) to end point (time 12 months) for acarbose and placebo. ‡  = 0.01 by covariance analysis.
    View larger version:
    Figure 3. Data are expressed as means with 95% CIs. *  ≤ 0.010, †  = 0.077. values are for differences between changes from baseline (time 0) to end point (time 12 months) for acarbose and placebo. ‡  = 0.01 by covariance analysis. The effects of acarbose (black circles) and placebo (○) on hemoglobin A1c levels throughout the 1-year study period in groups of patients treated with diet alone (A), metformin (B), sulfonylurea (C), and insulin (D).PPPP

    Serum C-Peptide

    The mean fasting serum C-peptide levels at baseline were lower in the insulin group (404 ±36 pmol/L) than in the other three groups (720 ±34 pmol/L). The mean C-peptide response to the test meal at baseline as measured by incremental area under the curve was highest in the diet alone group (1275 ±144 pmol x h/L), lowest in the insulin group (434 ±50 pmol x h/L), and intermediate in the metformin and sulfonylurea groups (780 ±44 pmol x h/L). Although fasting serum C-peptide levels were not affected by acarbose, an upward trend was seen in the change in mean postprandial C-peptide incremental curve-area from baseline to end point: 98.8 pmol/L for acarbose compared with 46.9 pmol/L for placebo in the diet alone group, 323.9 compared with 119.1 for the metformin group, 272.9 compared with 59.5 for the sulfonylurea group, and 185.1 compared with 59.7 for the insulin group. These changes, however, were not statistically significant.

    Serum Lipid Profile

    The mean serum lipid profiles at baseline for all patients were 2.7 ±0.1 mmol/L for triglycerides, 5.6 ±0.1 mmol/L for total cholesterol, 4.1 ±0.1 mmol/L for LDL cholesterol, and 1.0 ±0.02 mmol/L for HDL cholesterol. These did not differ among the four groups and were not affected by acarbose treatment during the 12-month study period.

    Responder and Repeated-Measures Analysis

    A clinically relevant response to the study medication was defined as a decrease in HbA1c levels of at least 15% of baseline or a decrease to less than 7% (absolute value) or both without any increase. Because non–insulin-dependent diabetes mellitus is a progressive disease, an HbA1c level of less than 7% that does not increase during a 1-year period is also clinically relevant. A responder analysis was done to assess whether more of the patients receiving acarbose were responders compared with those receiving placebo. Overall, 52% of the patients who received acarbose were responders compared with 26% of those who received placebo Figure 4 A; Mantel-Haenszel, P < 0.001). If dropouts are classified as failures, 40% of the patients treated with acarbose and 18% of those treated with placebo were responders Figure 4 B; Mantel-Haenszel, P < 0.001).

    Figure 4. A. Study dropouts were carried over for analysis. B. All dropouts were considered failures. Data are expressed as percentage of patients with 95% CIs, and the number of patients who responded over the total number of patients are listed at bottom of the histograms. *  < 0.010, †  = 0.024, ‡  = 0.040. values are for differences between acarbose and placebo at the end of the study.
    View larger version:
    Figure 4. A. Study dropouts were carried over for analysis. B. All dropouts were considered failures. Data are expressed as percentage of patients with 95% CIs, and the number of patients who responded over the total number of patients are listed at bottom of the histograms. *  < 0.010, †  = 0.024, ‡  = 0.040. values are for differences between acarbose and placebo at the end of the study. The percentage of patients in four treatment groups who responded to acarbose or to placebo with decreases hemoglobin A1c levels by 15% from baseline or to less than 7% without any increase.PPPP

    Using the above definition for responders, we did a linear logistic regression analysis to identify factors that could predict the probability of success of acarbose treatment for the individual patient. When all patients were considered, the strongest relation was with the treatment itself; the odds for being a responder were 3.3 (CI, 2.0 to 5.5) in favor of acarbose compared with placebo (P < 0.001). No other variable could consistently predict response to the study drug.

    Safety Parameters

    Doses of acarbose as large as 200 mg three times daily had no toxic effect according to the results of hematologic and biochemical profiles, including liver function tests. They did not affect the serum concentrations of vitamins and minerals. No serious adverse event could be linked to acarbose; the only adverse events that could be linked to the drug were gastrointestinal symptoms. Compared with placebo, acarbose was more frequently associated with flatulence (73.2% for acarbose compared with 39.0% for placebo), diarrhea (43.6% compared with 20.3%), and abdominal cramps and discomfort (25.0% compared with 8.8%), but these symptoms were mostly of mild to moderate intensity. Because these gastrointestinal symptoms also occurred frequently in placebo recipients, they did not jeopardize the blinding of the study. Only four patients in the insulin group (one receiving acarbose and three receiving placebo) required correction of severe hypoglycemia.

    Previous SectionNext Section

    Discussion

    Our study was designed to test the long-term efficacy of acarbose in the treatment of patients with non–insulin-dependent diabetes mellitus. The drug is an α-glucosidase inhibitor that acts directly in the gastrointestinal tract to delay glucose absorption and thus to blunt the rapid increase in plasma glucose that follows a meal containing carbohydrates. In a 1-year, randomized, double-blind, parallel-group study, we compared acarbose with placebo in patients on four different treatment regimens: diet alone, diet and metformin, diet and sulfonylurea, and diet and insulin. The data show that acarbose improved glycemic control regardless of the treatment regimen used.

    This improvement cannot be attributed to changes in dietary habits, body weight, or concurrent hypoglycemic medications. The total caloric intake; the percentages of carbohydrates, lipids, and proteins; and the type of carbohydrates ingested remained unchanged throughout the study for both patients receiving acarbose and those receiving placebo in all four treatment groups. Although the study medication was titrated from 50 mg three times daily to a maximum of 200 mg three times daily, increase in or addition of oral hypoglycemic agents or insulin was not allowed. If such an increase or addition was required for the well-being of the patient after the first 2 months of the study, data gathered after the change in medication were dropped, and the data from the last valid visit were carried over for the final efficacy analysis. Such changes were more frequent in the patients receiving placebo than in those receiving acarbose (20% compared with 6%). Furthermore, in the diet and insulin group, the insulin dose was decreased by 15% or more in 36% of patients receiving acarbose but in only 6% of patients receiving placebo. Fewer patients receiving acarbose required extra medication to maintain adequate glycemic control; this suggests that acarbose might effectively slow the deterioration of glycemic control in patients with non–insulin-dependent diabetes mellitus.

    In general, the fasting plasma glucose levels remained the same or decreased in patients receiving acarbose and remained the same or increased in patients receiving placebo. Compared with placebo, patients receiving acarbose in the diet alone and the sulfonylurea groups had significantly lower fasting plasma glucose levels (P < 0.001 and P = 0.013, respectively), and a strong trend was seen in the metformin group (P = 0.107; (Table 3). No difference was seen in the insulin group. Although most studies in patients with non–insulin-dependent diabetes mellitus have not shown any effect of acarbose on fasting plasma glucose levels [25, 35-37], a few studies have shown small beneficial effects of the drug on those levels [23, 24]. In these studies, the initial fasting plasma glucose level was relatively low to start with (<10 mmol/L). The relatively small effect of acarbose on fasting plasma glucose levels is not unexpected because the drug delays the absorption of glucose from the gastrointestinal tract [18-20] and therefore should affect mainly the postprandial plasma glucose levels. The lowering effect of acarbose on fasting plasma glucose levels might be explained by decreased glucose toxicity with improvement in insulin sensitivity and β-cell response to glucose [9]. That this is occurring is suggested by the upward trend in postprandial C-peptide levels seen in patients who receive acarbose.

    On the other hand, postprandial plasma glucose concentrations were significantly affected by acarbose in patients in all four treatment groups (Figure 1). This is consistent with the data in studies in which acarbose was added to diet alone [23] or to sulfonylurea [24, 35, 37]. Our study, however, is the first long-term, controlled study to show a beneficial effect of acarbose on postprandial plasma glucose levels in patients with non–insulin-dependent diabetes mellitus who are being treated with biguanide or insulin. This makes acarbose an effective antidiabetic drug that can be added as adjunct therapy to any of the currently available treatments for patients with non–insulin-dependent diabetes mellitus. Also, the maximum effect of acarbose on postprandial plasma glucose was already reached 3 months into the study and was well maintained for the rest of the 12-month treatment period (Figure 3). Three months into the study, 60% of patients were still receiving no more than 100 mg three times daily; this raises the question of whether the maximal dose of 200 mg three times daily is necessary to achieve adequate glycemic effect. This question is important because the gastrointestinal side effects are dose-related [20].

    The clinical relevance of the effect of acarbose on postprandial plasma glucose is highlighted by the significant decrease in HbA1c levels seen in patients in all four treatment groups (Figure 3). The mean effect of a drug in the overall study population, however, will underestimate the true efficacy of the medication because some patients respond and others do not. For this reason, a responder analysis of the data showed that more than 50% of all patients treated with acarbose responded to the drug, regardless of their concurrent antidiabetic medication, compared with 26% of patients who received placebo (Figure 4); Mantel-Haenszel, P < 0.001). The response to acarbose in patients treated with diet alone and in those treated with diet and sulfonylurea is similar to that found in the studies of Hanefeld and colleagues [23] and Reaven and colleagues [24]. It differs, however, from the data reported by Buchanan and colleagues [37], which did not show any improvement in HbA1c levels in patients treated with diet alone. The difference can be explained by the fact that these patients were withdrawn from sulfonylurea therapy before starting acarbose treatment [38]. Our study is the first long-term, controlled study to show an improvement in HbA1c levels in patients when acarbose is added to biguanide or insulin. The maximum decrease in HbA1c levels was reached at 6 months and remained constant for the rest of the 12-month treatment period. The longer half-life of HbA1c explains the slower change in time when compared with plasma glucose. In the patients treated with acarbose, the decrease in HbA1c leveled off at slightly more than 7%, a result similar to that obtained in the Diabetes Complication Clinical Trial study for an experimental group of patients with insulin-dependent diabetes mellitus who were receiving intensive insulin therapy [38]. It is also similar to the decrease in HbA1c levels seen in patients with non–insulin-dependent diabetes mellitus in the United Kingdom Prospective Diabetes Study [14].

    Our study did not show that acarbose had any effect on mean fasting and postprandial serum C-peptide levels, although an upward trend in postprandial C-peptide levels was seen. Although the reduction in plasma glucose levels by acarbose treatment in healthy persons has always been associated with decreased plasma insulin levels [39, 40], the results in patients with non–insulin-dependent diabetes mellitus have conflicted [24, 35-37]. This is probably due to the heterogeneity of the study population in relation to the concurrent treatment and the β-cell responsiveness. Although metformin has no effect on β-cells [13], sulfonylurea stimulates them [12] and insulin treatment suppresses them [41]. Furthermore, in our study population, the different treatment groups probably represented different stages in the progression of non–insulin-dependent diabetes mellitus. This is supported by the differences in the known duration of diabetes among patients from each treatment group. Duration ranged from a mean of 5 years in the diet alone group to 9 years in the sulfonylurea and the metformin groups to 13 years in the insulin group. It is therefore likely that these patients represent different stages on the “Starling” curve of insulin secretion in response to glucose in the appearance and progression of non–insulin-dependent diabetes mellitus [8].

    Acarbose also did not affect the mean fasting serum lipid profile in any of the four groups. Although some studies in patients with non–insulin-dependent diabetes mellitus have shown that acarbose could reduce total serum triglyceride levels [42, 43], others have not been able to reproduce that effect [44, 45]. Because the serum lipid profiles were normal or nearly normal in most of our patients, major changes due to acarbose were not expected.

    Finally, our study shows that acarbose is a safe drug for treating patients with non–insulin-dependent diabetes mellitus. Over the 1-year study period, acarbose did not affect hematologic, biochemical, or liver function test results. Our study also strongly suggests that the drug does not induce nutritional malabsorption because the serum concentrations of vitamins and minerals remained unchanged in patients treated with acarbose. The most common side effects were flatulence and diarrhea, and these were mostly mild to moderate in severity. Although they caused discontinuation of therapy in 19% of patients treated with acarbose, they were also responsible for the study withdrawal of 11% of placebo recipients. Furthermore, acarbose did not increase the incidence of hypoglycemia. Acarbose is therefore safe and well tolerated by patients with non–insulin-dependent diabetes mellitus.

    Acarbose provides physicians with the first new pharmacologic concept in the treatment of diabetes mellitus since the advent of sulfonylurea and biguanide in the mid-1950s. It is advantageous because it can be used as first-line treatment with diet alone or as an adjunct medication to metformin, sulfonylurea, or insulin. Our study also emphasizes the importance of postprandial plasma glucose as a determinant of the HbA1c leveland thus of the degree of glycemic control.

    Study Collaborators: B. MacIntyre, RN; J. Salmond, RN; and K. Travers, RDt (Halifax, Nova Scotia, Canada); J.-M. koe, MD; M. Verdy, MD; F. Ducros, RN, BSc; H. Langelier, RDt; and D. Poisson, RN (Montreal, Quebec, Canada); T. Siddigi, BSc; Y. Strasberg, RN; and N. Zello, RDt (Toronto, Ontario, Canada); G. Becks, MD; G. Tevaarwerk, MD; B. Cubberley, RDt; W. Prescob, RDt; and E. Rose, RN (London, Ontario, Canada); S. Imes, RDt, MSc and G. Scrivens, RN (Edmonton, Alberta, Canada); R. Jamal, RDt (Calgary, Alberta, Canada); I. Byers, RDt and J. Tyson, RN (Vancouver, British Columbia, Canada); J. Mukherjee, PhD and C. Bartlett, MMath (Toronto, Ontario, Canada).

    Previous Section
     

    References

    1. 1.↵
      Tan MH, Wornell CM. Diabetes mellitus in Canada. Diabetes Res Clin Pract. 1991; 14(Suppl 2):S3-8.
    2. 2.↵
      Pirart J. Diabetes mellitus and its degenerative complications: a prospective study of 4400 patients observed between 1947 and 1973. Diabetes Care. 1978; 1:168-88.
    3. 3.↵
      Morrish NJ, Stevens LK, Head J, Fuller JH, Jarrett RJ, Keen H. A prospective study of mortality among middle-aged diabetic patients (the London Cohort of the WHO Multinational Study of Vascular Disease in Diabetics) II: Associated risk factors. Diabetologia. 1990; 33:542-8.
    4. 4.↵
      Huse DM, Oster G, Killen AR, Lacey MJ, Colditz GA. The economic costs of non–insulin-dependent diabetes mellitus. JAMA. 1989; 262: 2708-13.
    5. 5.↵
      Skyler JS. Relation of metabolic control of diabetes mellitus to chronic complications. In: Rifkin H, Porte D, eds. Diabetes Mellitus, Theory and Practice. New York: Elsevier; 1990:856-68.
    6. 6.↵
      DeFronzo RA, Ferrannini E. Insulin resistance. A multifaceted syndrome responsible for NIDDM, obesity, hypertension, dyslipidemia, and atherosclerotic cardiovascular disease. Diabetes Care. 1991; 14:173-94.
    7. 7.↵
      Ward WK, Beard JC, Halter JB, Pfeifer MA, Porte D Jr. Pathophysiology of insulin secretion in non–insulin-dependent diabetes mellitus. Diabetes Care. 1984; 7:491-502.
    8. 8.↵
      DeFronzo RA, Bonadonna RC, Ferrannini E. Pathogenesis of NIDDM. A balanced overview. Diabetes Care. 1992; 15:318-68.
    9. 9.↵
      Rossetti L, Giaccari A, DeFronzo RA. Glucose toxicity. Diabetes Care. 1990; 13:610-30.
    10. 10.↵
      Henry RR, Wallace P, Olefsky JM. Effects of weight loss on mechanisms of hyperglycemia in obese non–insulin-dependent diabetes mellitus. Diabetes. 1986; 35:990-8.
    11. 11.↵
      Henry RR, Brechtel G, Griver K. Secretion and hepatic extraction of insulin after weight loss in obese noninsulin-dependent diabetes mellitus. J Clin Endocrinol Metab. 1988; 66:979-86.
    12. 12.↵
      Groop LC. Sulfonylureas in NIDDM. Diabetes Care. 1992; 15:737-54.
    13. 13.↵
      Bailey CJ. Biguanides and NIDDM. Diabetes Care. 1992; 15:755-72.
    14. 14.↵
      UK Prospective Diabetes Study (UKPS). VIII. Study design, progress and performance. Diabetologia. 1991; 34:877-90.
    15. 15.↵
      Genuth S. Insulin use in NIDDM. Diabetes Care. 1990; 13:1240-64.
    16. 16.↵
      Brand JC, Colagiuri S, Crossman S, Allen A, Roberts DC, Truswell AS. Low-glycemic index foods improve long-term glycemic control in NIDDM. Diabetes Care. 1991; 14:95-101.
    17. 17.↵
      Jenkins DJ, Wolever TM, Jenkins AL, Taylor RH. Dietary fibre, carbohydrate metabolism and diabetes. Mol Aspects Med. 1987; 9:97-112.
    18. 18.↵
      Hillebrand I. Pharmacological modification of digestion and absorption. Diabet Med. 1987; 4:147-50.
    19. 19.↵
      Puls W, Krause HP. Delay of carbohydrate absorption by inhibitors of intestinal α-glucosidases. Adv Exp Med Biol. 1979; 119:341-6.
    20. 20.↵
      Clissold SP, Edwards C. Acarbose. A preliminary review of its pharmacodynamic and pharmacokinetic properties, and therapeutic potential. Drugs. 1988; 35:214-43.
    21. 21.↵
      Madar Z. The effect of acarbose and miglitol (BAY-M-1099) on post-prandial glucose levels following ingestion of various sources of starch by nondiabetic and streptozotocin-induced diabetic rats. J Nutr. 1989; 119:2023-9.
    22. 22.↵
      Sailer D. The influence of a new glycoside hydroxylase inhibitor on blood glucose and serum insulin concentration in non–insulin treated diabetes. Horm Metab Res. 1981; 13:55-8.
    23. 23.↵
      Hanefeld M, Fischer S, Schulze J, Spengler M, Wargenau M, Shollberg K, et al. Therapeutic potentials of acarbose as first-line drug in NIDDM insufficiently treated with diet alone. Diabetes Care. 1991; 14:732-7.
    24. 24.↵
      Reaven GM, Lardinois CK, Greenfield MS, Schwartz HC, Vreman HJ. Effect of acarbose on carbohydrate and lipid metabolism in NIDDM patients poorly controlled by sulfonylureas. Diabetes Care. 1990; 13(Suppl 3):32-6.
    25. 25.↵
      Rodier M, Richard JL, Monnier L, Mirouze J. Effect of long term acarbose (Bay g 5421) therapy on metabolic control of non insulin dependent (type II) diabetes mellitus. Diabete Metab. 1988; 14:12-4.
    26. 26.↵
      WHO Study Group on Diabetes Mellitus. Diabetes mellitus: Report of a WHO Study Group. Geneva: World Health Organization; 1985:10-4.
    27. 27.↵
      Nutritional recommendations and principles for individuals with diabetes mellitus. Diabetes Care. 1993; 16(Suppl 2):22-9.
    28. 28.↵
      Bergmeyer HU, Bernt E, Schmidt F, Stock H. D-glucose. Determination with hexokinase and glucose-6-phosphatase dehydrogenase. In: Bergmeyer HU, Gawehn K, eds. Methods of Enzymatic Analysis. New York: Academic Press; 1974:1196-201.
    29. 29.↵
      Kuzuya T, Saito T, Yoshida S, Matsuda A. Human C-peptide immunoreactivity (CPR) in blood and urine—evaluation of a radioimmunoassay method and its clinical applications. Diabetologia. 1976; 12:511-8.
    30. 30.↵
      Manual of Laboratory Operations: Lipid Research Clinics Program. Bethesda, Maryland: National Heart and Lung Institute, National Institutes of Health; 1982.
    31. 31.↵
      Warnick GR, Benderson J, Albers JJ. Dextran sulfate-Mg2+ precipitation procedure for quantitation of high-density-lipoprotein cholesterol. Clin Chem. 1982; 28:1379-88.
    32. 32.↵
      Friedwald WT, Levy RI, Frederickson DS. Estimation of the concentration of low-density lipoprotein cholesterol in plasma, without use of the preparative ultracentrifuge. Clin Chem. 1972; 18:499-502.
    33. 33.↵
      Winer BJ. Statistical Principles in Experimental Design. New York: McGraw-Hill; 1971.
    34. 34.↵
      Hosmer DW, Lemeshow S. Applied Logistic Regression. New York: John Wiley; 1989.
    35. 35.↵
      Van Gaal L, Nobels F, DeLeeuw I. Effects of acarbose on carbohydrate metabolism, electrolytes, minerals and vitamins in fairly well-controlled non–insulin-dependent diabetes mellitus. Z Gastroenterol. 1991; 29:642-4.
    36. 36.↵
      Sachse G, Willms B. Effect of the α-glucosidase-inhibitor BAY-g-5421 on blood glucose control in sulfonylurea-treated diabetics and insulin-treated diabetics. Diabetologia. 1979; 17:287-90.
    37. 37.↵
      Buchanan DR, Collier A, Rodrigues E, Millar AM, Gray RS, Clarke BF. Effectiveness of acarbose, an α-glucosidase inhibitor, in uncontrolled non-obese non–insulin-dependent diabetes. Eur J Clin Pharmacol. 1988; 34:51-3.
    38. 38.↵
      The Diabetes Control and Complications Trial Research Group. The effect of intensive treatment of diabetes on the development and progression of long-term complications in insulin-dependent diabetes mellitus. N Engl J Med. 1993; 329:977-86.
    39. 39.↵
      Gerard J, Jandrain B, Pirnay F, Pallikarakis N, Krzentowski G, Lacroix M, et al. Utilization of oral sucrose load during exercise in humans. Effect of the α-glucosidase inhibitor acarbose. Diabetes. 1986; 35:1294-301.
    40. 40.↵
      Jenkins DJ, Taylor RH, Goff DV, Fielden H, Misiewicz JJ, Sarson DL, et al. Scope and specificity of acarbose in slowing carbohydrate absorption in man. Diabetes. 1981; 30:951-4.
    41. 41.↵
      Liljenquist JE, Horwitz DL, Jennings AS, Chiasson JL, Keller U, Rubenstein AH. Inhibition of insulin secretion by exogenous insulin in normal man: demonstration by C-peptide assay. Diabetes. 1978; 27:563-70.
    42. 42.↵
      Lardinois CK, Greenfield MS, Schwartz HC, Vreman HJ, Reaven GM. Acarbose treatment of non–insulin-dependent diabetes mellitus. Arch Intern Med. 1984; 144:345-7.
    43. 43.↵
      Akazawa Y, Koide M, Oishi M, Azuma T, Tashiro S. Clinical usefulness of acarbose and fiber in the treatment of diabetes mellitus. Therapeutics. 1982; 36:848-9,870-5.
    44. 44.↵
      Buber V. Effect of acarbose in non–insulin-dependent diabetic patients during long-term treatment. In: Creutzfeldt W, ed. Proceedings, First International Symposium on Acarbose: Effects on Carbohydrate and Fat Metabolism, Montreux, October 8-10, 1981. Amsterdam and Princeton: Excerpta Medica; 1982:422-6.
    45. 45.↵
      Noyon R, Pagano Mirani-Oostdijk C, van Gent CM, Frolich M, Hillebrand I, Terpstra J. Long-term effect of acarbose on diurnal serum triglyceride, glucose, insulin and adipose tissue lipoprotein lipase levels in patients with primary endogenous hypertriglyceridaemia, with or without type II diabetes. Neth J Med. 1986; 29:157-64.
    « Previous | Next Article »Table of Contents

    This Article

    1. Ann Intern Med December 15, 1994 vol. 121 no. 12 928-935
    1. AbstractFREE
    2. Figures
    3. » Full Text
    4. Full Text (PDF)

    Rapid Responses

    1. Submit a response
    2. No responses published

    Navigate This Article

    Current Issue

    1. November 3, 2009, 151 (9)
    1. Alert me to current issues