Captopril Prevention Project (CAPPP)
6614 patients in primary care; 5-year follow-up. Outcome: diabetes by FPGX2 Heart Outcomes
Prevention Evaluation Project (HOPE) 267 centers in Canada, North and South America, and Europe
Captopril (N = 5183 nondiabetic); 50 mg qd or bid
Conventional (N = 5230 nondiabetic); (3-blockers and/or diuretics
Diuretic/beta-blocker strategy. ACE-inhibitors strategy. Calcium antagonist strategy
Ramipril (N = 2837
nondiabetic); 10 mg qd Placebo (N = 2883 nondiabetic) in 2 X 2 factorial with Vitamin E
Captopril: RR = 0.86 (0.74-0.99) vs. Conventional treatment of hypertension. Efficacy analysis: RR = 0.79 (0.67-0.94)
No difference in diabetes incidence by treatment group
Ramipril: RR = 0.66 (0.51-0.85) vs. placebo for incidence of new diabetes. No interaction with Vitamin E noted
Subgroup analysis of nondiabetic persons at baseline; more diabetes in captopril group at baseline; randomization by sealed envelope; dropouts 0.25%. Captopril group added thiazide diuretic in an unspecified percentage Subgroup analysis
Ramipril discontinued in 29%, placebo in 27 %; 18 % of placebo group used an ACE-inhibitor. Dropouts not stated. Side-effects primarily cough that differed from placebo
Freeman et al.
9297 subjects aged 55 yrs and older (x = 66); follow-up 5 years. Outcome: diabetes
Pravastatin (40 mg/day) vs. Pravastatin: multivariate HR = 0.7 Subgroup post-hoc analysis, requires
Placebo (0.50-0.99) vs. placebo confirmation. Absolute incidence rates not given by group; side-effects minimal. Approximately 30% withdrawals at 5 years in primary study cholesterol; 4.9-yr follow-up. Outcome: diabetes by modified ADA criteria
West of Scotland Coronary Prevention Study
Primary care practices; 5974men, (x =55 years), nondiabetic, elevated
I Calculated from data available in the paper, not by the authors; FSIGT = frequently sampled intravenous glucose tolerance test; HR = hazard ratio; RR = relative risk or risk ratio; ARR = absolute risk reduction (if not present, data were not available to calculate).
twice a day compared with placebo. Subjects had a fasting blood glucose level <6.2 mmol/l (112 mg/dl) and 1 h >9.4 mmol/l (170 mg/dl) or 2 h >6.7 mmol/l (120 mg/dl), levels that would be considered mostly in the normal or IGT range by current criteria. After an average follow-up of 14 months, fasting glucose levels of approximately 91 mg/dl (5.05 mmol/L) on tolbutamide were significantly lower than the approximately 100 mg/dl (5.56 mmol/L) levels on placebo (p <0.05), but there were no significant differences in 1, 2 or 3 h levels. Analysis was confusing, with some results 'adjusted for baseline levels', and others averaged over various time periods. Only 15 subjects in each group were tested at 14 months (23-25% drop-out rate) and changes in diabetes status to normal were not reported. This study is difficult to interpret, but appears negative by current criteria, since there were only small differences in fasting glucose, and no differences in 2 h glucose levels.
At around the same time, Camerini-Davalos, in another study of the effect of sulfonylureas on diabetes outcomes185, assigned 38 subjects with 'chemical diabetes' (normal fasting glucose, impaired oral tolerance, no symptoms, but criteria not stated - most would be current IGT) to 500 mg of tolbutamide twice daily and gave 59 subjects a placebo. Randomization or blinding were not specified. After variable follow-up of 10-24 months, the RR for continuing to have any degree of IGT was 0.73 (CI 0.45-1.18, ARR = 14/100 person-years; intention to treat, post-hoc analysis). The primary effect appeared to be a reduction in the sum of glucose levels among people who remained abnormal, without significant reduction in the percentage of people who remained glucose-intolerant. No weight change occurred between groups and triglyceride levels actually increased significantly on therapy. The numerous methodological problems with this study, and the use of early criteria and method of data presentation, make it difficult to interpret, though contemporary authors also considered it a negative result186.
Feldman and colleagues took advantage of the multiphasic screening program at Kaiser-Permanente in northern California to identify subjects with glucose intolerance187,188. People with chemical diabetes by Fajans and Conn criteria189 (most now classified as IGT) were randomly assigned to tolbutamide (n = 174, 500 mg twice daily), the biguanide phenformin (n = 91, 100 mg/day) or placebo (n = 85) in a double-blind design.
After five years of follow-up, the incidence of diabetes was 0.6% on tolbutamide, with a RR = 0.16 (CI 0.02-1.54; ARR ^3/100 person-years, intention to treat, not accounting for losses, post-hoc analysis), there were no cases among those treated with phenformin (p = 0.22 versus placebo), and the rate in placebo treated subjects was 3.5%188. The 2 h glucose decline below entry levels was greatest in people taking tolbutamide, but was closely followed by those on placebo, and in both these groups glucose levels had returned to baseline levels after 42 months and were not different from each other. People taking phenformin had almost no change in 2 h glucose levels over the period.
Drop-outs averaged 11%188 and no excess of cardiovascular mortality was seen.
The relatively low diabetes incidence, even among people on placebo, is probably a reflection of the criteria for the outcome of diabetes ('overt diabetes', probably requiring substantially elevated glucose levels), the relatively low blood glucose levels at entry, and the relatively young age of subjects. These results suggest that tolbutamide may lower the incidence of diabetes, though the lack of glucose lowering with phenformin suggests it is not likely to be effective, and it is no longer on the market due to adverse side-effects.
Keen and colleagues conducted a long-term randomized double-blind study of tolbutamide (500 mg twice daily) versus placebo (3 mg tolbutamide, twice daily), in a factorial design adding dietary restriction of carbohydrate up to a maximum of 110 g/day ('diet') or avoidance of simple sugars at the table ('no diet')190. The primary goal was to determine if cardiovascular disease endpoints were lessened (or increased) on tolbutamide, given the temporal relationship of the Bedford study with the University Group Diabetes Program (UGDP), which had reported increased CVD mortality in the tolbutamide group191. Subjects were seen twice a year for five years, then less frequently for the remaining five years.
In a series of reports spanning the study period190,191-194, the authors noted that diet had no effect within six months and that only about 50% of subjects took most of the assigned medication191. Compliance and continuation of treatment were not mentioned in the later reports. By the 8.5-year follow-up, it was clear that there was no benefit from tolbutamide versus placebo on the incidence of diabetes190 (RR = 1.11, CI 0.49-1.54), which was confirmed in the 10-year report using modern logistic analyses194. No excess of vascular events was seen in the tolbutamide group, in contrast to reports from the UGDP, but other adverse effects (e.g. hypoglycemia) were not reported. Thus, this reasonably conducted trial provides no evidence for the long-term efficacy of tolbutamide for diabetes prevention. Whether this is due to actual lack of efficacy or to limited drug exposure remains unclear.
Paasikivi186 randomly assigned 118 survivors of a first myocardial infarction (MI) at the Karolinska Institute in Stockholm, Sweden to 500 mg of tolbutam-ide, twice daily (n = 95) or placebo (n = 83) using odd/even allocation on birth date. The trial was blinded only to the patient, and the primary outcome was changes in intravenous glucose tolerance (IVGT), making it hard to compare with current standards.
Total mortality was slightly but not significantly lower on active drug (RR = 0.11, CI 0.36-1.39) but there were minimal changes in glucose outcomes. There were no differences in fasting glucose between treated and control groups over several follow-up examinations. While none of the subjects had overt diabetes at entry, 13% (tolbutamide) or 19% (placebo) had 'diabetic' IVGT results, categorized as low levels of glucose disappearance (K0) based on prior studies. This prevalence of abnormal K0 changed very little over the trial, which lasted up to five years, but was 2.9 years on average. None of the subjects remaining in the trial on tolbutamide developed overt diabetes, whereas three did so on placebo (p = 0.20, intention to treat, post-hoc analysis). Subjects with diabetic values for K0 did have improvements in their values, but none of the subjects with borderline or normal K0 showed changes. This trial, while interpreted as positive in subgroups by its author, was essentially negative by current intention-to-treat standards. There was a lower conversion to overt diabetes among treated subjects (0 versus. 3), but the trial was too small to detect a significant difference in this outcome or in total mortality at five years.
Investigators at the Joslin Clinic in Boston studied 120 men with chemical diabetes randomly allocated to placebo, chlorpropamide, tolbutamide, phen-formin or acetohexamide for five years195. All subjects had normal fasting glucose (<100 mg/dl, 5.5 mmol/L) at entry, and most would have had normal glucose tolerance or IGT by current criteria.
Over the five-year follow-up, between 30-50% of men dropped out of the trial. No significant differences in the percentage with normal glucose tolerance were seen between the placebo group and all other groups at any time, and there were no changes sequentially within any group toward improved OGTT results. The only exception was a small improvement at one year in the chlorpropamide group, which was not sustained from years 2 to 5. In addition, there were no improvements in insulin area during the OGTT, and no changes in lipids. This study, while suffering from small numbers, rather high losses to follow-up and univariate analysis, was otherwise reasonably conducted, and found no improvements on any drug regimen.
Sartor and colleagues conducted a community screening in Malmohus, Sweden from 1962 to 1965196. They conducted OGTTs on people who were glycosuric during screening, then sequentially randomized 267 men with IGT (by local criteria) to one of four groups: (1) diet alone, (2) diet plus 500 mg tolbutamide three times daily, (3) diet plus placebo or (4) no therapy. The first three groups underwent annual OGTT testing until the mid-1970s, and tol-butamide was unmasked in 1972, about two years before completion of follow-up.
While the primary analysis was conducted as an efficacy analysis ('per protocol'), i.e. comparing compliant and non-compliant subjects, it is possible to reconstruct an intention-to-treat result. This indicated no significant difference between tolbutamide (10.2% diabetes at 10 years), placebo (12.5%) and diet only (14.0%). Group 4 above, the 'no therapy' group was made up of an unspecified retrospective sample of 59 men originally randomized to no therapy in which the cumulative incidence of diabetes was 28.8%. Drop-outs were about 11.2% of total subjects.
Only limited conclusions are possible from this study, given its small size and efficacy analysis. Based on intention-to-treat analysis, there was no effect of tolbutamide therapy over 10 years, compared to placebo or diet alone, though the overall incidence was slightly lower (RR = 0.82, CI 0.27-2.50 versus placebo; RR = 0.73, CI 0.25-2.14 versus diet, post-hoc analyses). In the efficacy analysis, none of the subjects continuing to take tolbutamide developed diabetes at 10 years, an observation which has caused many to be hopeful of the potential benefit of the sulfonylureas 10.
A later follow-up of these subjects has been published197. Several useful clarifications appear here. First, it appears that the IGT criteria used in the 1960s included many normal subjects. On average, 79% of subjects in the groups randomized to tolbutamide, placebo or diet only had normal glucose tolerance by WHO criteria, and 18% had WHO IGT, while 3% had diabetes. Cumulative mortality rate ratios for the tolbutamide-treated men (compared to non-treated men) up until 1987 (~22 years) were 0.66 (CI 0.39-1.10) for all causes of death, and 0.42 (CI 0.16-1.12) for ischemic heart disease (IHD) deaths. Thus, it appears that the long-term IHD mortality in the tolbutamide group was lower than in men not taking it, for up to 10 years, rather than increased (as suggested by the UGDP results) among people with diabetes191,198. No additional insights concerning diabetes incidence were published.
A small pilot study was conducted in Oxford, the UK using the sulfonylurea glicazide (40 mg twice daily) versus placebo44. In the same pilot, a 'healthy living' group was included, and two-year follow-up of this portion of the study were reported44 as noted above. Only the six-month follow-up of the drug group has been reported. Subjects with abnormal fasting glucose levels or glucose infusion tests with results said to be similar to WHO IGT were included. Six were given glicazide and eight were given placebo in blinded fashion.
At six months, fasting glucose and 1 h CIGMA glucose were decreased significantly without change in insulin sensitivity, but with improved beta-cell function. No significant changes were noted in the placebo group. One month after stopping the drug, people were retested, with reversion of fasting glucose to baseline levels. The authors concluded that drug therapy, to be effective, must be continued. This pilot led to the development of the Fasting Hyper-glycemia Study summarized below, since larger numbers were needed with a longer follow-up period.
The Fasting Hyperglycemia Study (FHS) II was designed to identify high-risk subjects for prevention therapy using lifestyle and/or glicazide56. The basic design was described previously under combined lifestyle trials. In the pharmacological portion of the FHS, 227 subjects with IFG were randomly allocated to glicazide, 80 mg twice daily (n = 112) or control (n = 115), with 58 of the 115 controls given placebo tablets, and 57 given no tablets199.
At one year of follow-up in the drug group, fasting glucose and HbA1c were significantly lower. However, there was no change in the proportion of people with diabetes. Two-hour glucose and the area under the OGTT curve were both higher in the drug group at one year, though they were lower or showed no change in the control group. Beta-cell function improved slightly in the drug group, but there were no changes in insulin sensitivity in either group. Mild symptoms of hypoglycemia were twice as common among glicazide-treated subjects, though they were no different at one year. After six years, there were only 188 subjects available for analysis. Fewer subjects in the glicazide group (3.2%) developed overt diabetes (two fasting plasma glucose values >180 mg/dL or 10 mmol/L); than in the control group (10.8%) (p = 0.047; ARR = 7.6/100 person-years)200. However, there were no differences in the rate of development of WHO diabetes (FPG >140 mg/dL or 7.8 mmol/L or OGTT) and no differences in Homeostasis Model Assessment (HOMA) derived measures of resistance or secretion. The authors also conducted a two-month drug washout in a subsample, which found small increases in FPG (5.4 mg/dL, 0.3 mmol/L, p = 0.022), but no difference in two hour glucose, HbA1c, or the proportion with diabetes between drug and control groups.
Was this article helpful?
Do You Suffer From High Blood Pressure? Do You Feel Like This Silent Killer Might Be Stalking You? Have you been diagnosed or pre-hypertension and hypertension? Then JOIN THE CROWD Nearly 1 in 3 adults in the United States suffer from High Blood Pressure and only 1 in 3 adults are actually aware that they have it.