Efficacy of Benfluorex in Combination With Sulfonylurea in Type 2 Diabetic Patients

RESEARCH DESIGN AND METHODS

Eligibility criteria were type 2 diabetes with age ≥18 years, BMI of 25–40 kg/m², and A1C of 7–10% (inclusive) despite monotherapy with a sulfonylurea at the maximum tolerated dose for at least 4 months. All patients had to have either a history of gastrointestinal intolerance to metformin or a contraindication to its use, such as renal impairment or any cardiac or respiratory condition with a potential for tissue hypoxia (3). Exclusion criteria were severe renal impairment, an alanine aminotransferase or aspartate aminotransferase plasma level three times above the upper limit of normal, active proliferative retinopathy, uncontrolled high blood pressure (≥180/100 mmHg), or any disorder that could interfere with the study conduct or end point evaluation. Patients were withdrawn for lack of efficacy, which was defined as two fasting plasma glucose (FPG) measurements ≥15 mmol/l, after at least 1 month of study treatment.

A randomized, placebo-controlled, double-blind, parallel-group study, with an 18-week comparative treatment period, was conducted in 63 centers in seven countries. After a 2- to 4-week run-in period, patients were centrally randomly assigned to receive tablets containing either placebo or 150 mg benfluorex (Les Laboratoires Servier Industrie, Gidy, France), the dose being gradually increased over the first 3 weeks from one tablet per day to the recommended dose of one tablet three times daily. The dose of sulfonylurea was to be maintained throughout the study, except in patients with severe or repeated hypoglycemia. Subjects were asked to maintain their usual diet and physical activity. The primary efficacy end point was A1C. The secondary end points were FPG, fasting serum insulin (FSI), and lipid profile, each assessed at inclusion and at weeks 4, 10, and 18, and insulin resistance index evaluated at baseline and week 18. An interactive voice response system was used for centralized randomization, stratified by baseline A1C (≤8 or >8%) and geographic area. The protocol was approved by each institution’s ethics committee. The trial was conducted in accordance with the ethical principles stated in the Declaration of Helsinki 1964, as revised in Edinburgh, U.K., in October 2000. All subjects provided written informed consent.

Assessments

A1C, FPG, FSI, and standard biochemical, hematological, and lipid parameters were assessed centrally (MDS Pharma Services, Hamburg, Germany). Creatinine clearance was calculated according to the Cockcroft formula (10). A1C was assayed using an National Glycohemoglobin Standardized Program–certified high-performance liquid chromatography method (BioRad Variant I). FPG was assessed by a hexokinase method; all glucose and lipid assays were performed on a Hitachi 747 instrument (Boehringer Mannheim, Mannheim, Germany). FSI was measured by a specific radioimmunoassay; the insulin resistance index (homeostasis model assessment of insulin resistance [HOMA-IR]) was calculated using the classic homeostasis assessment model (11). Safety was assessed by adverse event spontaneous reporting, physical examination, recording of vital signs, laboratory tests, and 12-lead electrocardiogram at baseline and at week 18. Hypoglycemic events were recorded from the patient diary and were based on suggestive clinical symptoms only.

Statistical analysis

Sample size was estimated based on a 0.6% between-group difference in A1C and an assay standard deviation of 1.5%, using a two-sided Student’s t test for independent samples with 90% power, a type 1 error of 5%, and a 10% drop-out rate. The intention-to-treat (ITT) population was defined as all randomly assigned patients exposed to the study drug with a baseline value and at least one A1C value during treatment. Per-protocol patients were defined as patients who completed the study without a deviation affecting the A1C evaluation. For efficacy analyses, the difference between benfluorex and placebo treatment was studied as the change from baseline to last value with treatment using ANCOVA with baseline and geographic area as covariates. The within-group evolution between baseline and the last value during treatment was studied in each treatment group using a two-sided paired Student’s t test. Nonparametric approaches were performed for triglyceridemia and HOMA-IR analyses. Three subgroups were predefined according to regulatory requirements (12): A1C at inclusion >8%, age at selection >65 years, and creatinine clearance ≤80 ml/min. Safety analyses were performed on all patients exposed to at least one dose of study treatment. Final values for withdrawn patients corresponded to the last values during treatment. All statistical analyses were performed using SAS software (version 8.2; SAS, Cary, NC).

RESULTS

Demographic and baseline characteristics

A total of 325 patients were included: 165 were randomly assigned to benfluorex and 160 to placebo. There were 317 patients (97%) in the ITT population. Thirty withdrawals (9.2%) occurred, 21 for patients receiving benfluorex and 9 for patients receiving the placebo; 18 patients withdrew because of adverse events (12 receiving add-on benfluorex and 6 receiving the placebo), 2 patients withdrew because of lack of efficacy (1 in each group), and 10 patients withdrew for nonmedical reasons. There were 265 patients (81%) in the per-protocol population. The ITT and per-protocol populations were similar at inclusion. Baseline characteristics were broadly similar in the two groups (Table 1). Overall, 73% of patients had a metabolic syndrome [modified National Cholesterol Education Program (NCEP) Adult Treatment Panel III definition (13)] and 57% of patients had previous gastrointestinal intolerance to metformin. Concomitant treatment was being taken by 280 patients (86%), mainly ACE inhibitors and antihypertensive drugs. Lipid-lowering agents were being taken by 70 patients (22%), 54 of whom were receiving statins. Less than 5% of patients had their sulfonylurea dose modified. The subgroup of patients with A1C >8% (94 patients receiving add-on benfluorex and 91 receiving placebo), of age >65 years (72 patients receiving benfluorex and 84 on placebo), and of creatinine clearance ≤80 ml/min (65 patients receiving benfluorex and 80 receiving placebo) accounted for 57, 48, and 45%, respectively, of the included patients.

Efficacy

In the add-on benfluorex group, glycemic control was markedly improved by week 4 and continued to improve throughout the study (Fig. 1A). After 18 weeks, A1C decreased in the ITT population by −0.82% (P < 0.001) (Table 2). With add-on placebo, there was a trend toward an increase in A1C from weeks 4 to 18 (NS), resulting in a between-group difference of −1.01% (P < 0.001). A decrease in A1C of ≥1% was seen in 42.9% of patients receiving add-on benfluorex versus 14.7% receiving placebo (P < 0.001). A decrease in A1C of ≥0.5% was seen in 69.6% of patients receiving benfluorex versus 27.5% receiving placebo (P < 0.001). Target values of ≤7% were achieved by 34.2% of patients receiving benfluorex versus 11.5% of patients receiving placebo (P < 0.001), whereas 18.5% of patients achieved an objective of ≤6.5% with benfluorex versus 4.5% with placebo (P < 0.001). In patients with baseline A1C in the range of 7 to 8%, a target value of ≤7% was achieved by 47.1% of patients in the benfluorex group versus 14.9% in the placebo group (P < 0.001). Similar results were observed in the per-protocol population, with a −1.08% difference of A1C between the benfluorex and placebo groups. FPG consistently decreased with benfluorex, with a between-group difference after 18 weeks of −1.65 mmol/l (P < 0.001) (Table 2 and Fig. 1B). Similar changes over time in A1C and FPG were seen in the subgroups according to A1C >8%, age >65 years, and impairment of renal function. A slight decrease in weight was observed in both groups, with a nonclinically relevant difference between benfluorex and placebo: −1.3 kg with benfluorex versus −0.7 kg with placebo (P < 0.01). The superiority of benfluorex over placebo with respect to A1C was independent of the decrease in body weight: the estimated between-group difference in A1C from the covariance analysis performed with adjustment for weight evolution was −0.96% (P < 0.001).

Insulin resistance as assessed by HOMA-IR decreased over the study by 20% (from 6.62 ± 7.99 at baseline to 4.87 ± 3.67) with benfluorex and by 5% (from 6.35 ± 7.95 to 5.93 ± 5.35) with placebo, with a statistically significant between-group difference (P < 0.01) (Table 2). In terms of the lipid profile (Table 2), with benfluorex, LDL cholesterol decreased by 6% and triglycerides by 7%. More than one-third (35.5%) of the patients treated with benfluorex had a favorable evolution of their LDL cholesterol: from ≥130 mg/dl at baseline to below this level at the end of the study [cutoff point as used in the NCEP recommendations (14)] compared with 14.3% of patients receiving a placebo. At the end of the study, 34.5% of patients receiving benfluorex had plasma triglycerides <2.2 mmol/l (baseline ≥2.2 mmol/l) versus 24.1% of patients receiving placebo. The changes from baseline in the benfluorex group were −0.27 ± 0.73 mmol/l for calculated LDL cholesterol and −0.05 ± 2.08 mmol/l for triglycerides (median −0.12 mmol/l), with a significant between-group difference in favor of benfluorex (P < 0.001 for LDL cholesterol and P = 0.027 for triglycerides). No change in HDL cholesterol was observed.

Safety and tolerability

A similar proportion of patients in each group (53.0% receiving benfluorex and 51.3% receiving a placebo) reported at least one adverse event. Gastrointestinal disorders were the most common adverse events, occurring in 15.1% of patients in the benfluorex group and 10.0% of patients in the placebo group, and mostly involved diarrhea (6.0% of patients receiving benfluorex vs. 1.9% receiving placebo). Two patients treated with benfluorex were withdrawn because of diarrhea (1.2%). In patients with known previous intestinal intolerance to metformin, the incidence of gastrointestinal disorders was the same with benfluorex and placebo (15.2 vs. 15.3%, respectively).

Hypoglycemic symptoms were more frequently reported with sulfonylurea and benfluorex (24 episodes in 14 patients) than with sulfonylurea and placebo (13 episodes in 6 patients). No episode required external assistance or hospitalization. Headache was reported by 2.4% of patients receiving benfluorex and by 2.5% receiving placebo. Other adverse events were reported in fewer than 2% of patients. Very few adverse events were considered related to the study drug (6.6% of patients receiving benfluorex vs. 1.3% receiving placebo, mostly diarrhea or abdominal pain). Serious adverse events (fatal and nonfatal) occurred in 11 patients. Two fatal events were reported in the benfluorex group, without any obvious causal relationship: one patient committed suicide (4 weeks after a ischemic cerebrovascular attack) and the other died as a result of cerebrovascular hemorrhage. One patient in the placebo group had two independent serious nonfatal adverse events (unstable angina and hyperglycemia). Nine patients in the benfluorex group had 10 serious nonfatal adverse events: diabetic neuropathy, pyelonephritis and ovarian cyst, nephrolithiasis, cerebrovascular accident, lumbar radiculopathy, biliary neoplasm, cardiac failure after inappropriate termination of diuretic therapy, pleural metastasis, and abdominal pain. Of these, the last three led to patient withdrawal. Only for abdominal pain was there thought to be a causal relationship; the others reasons were considered to be related to each patient’s medical history or concomitant pathologic conditions. There were no significant changes in the electrocardiogram or biological safety chemical parameters, including liver enzyme levels. Mean sitting diastolic and systolic blood pressures were similar in the benfluorex group and the placebo group, with no clinically relevant changes at the end of the study.

CONCLUSIONS

This is the first large-scale study of the efficacy and safety of benfluorex as add-on therapy in type 2 diabetic patients whose diabetes was suboptimally controlled with sulfonylureas at their maximum tolerated dose. The design versus placebo, the good compliance with the protocol procedure and with treatment (drop-out rate <10% with no patients lost to follow-up), and the consistency of results across the analyses make the analysis of the results reliable. Moreover, the patient population was representative of patients with type 2 diabetes in terms of disease duration (mean ± SD 7.1 ± 6.0 years), BMI (mean 29 kg/m²), presence of metabolic syndrome (in 73% of patients), and the presence of diabetic cardiovascular complications and hypertension. They had inadequate glycemic control, with a mean A1C of 8.32%.

This study demonstrates that the agent studied, benfluorex, as add-on therapy behaved throughout the trial as an active glucose-lowering drug that significantly and markedly improved glycemic control. After 18 weeks, the mean A1C was significantly reduced by −0.82 ± 1.04% in the benfluorex group with a between-group difference of −1.01%. This reduction in A1C was seen in each subgroup, especially in patients with a baseline A1C >8%, with a between-group difference of −1.10%. More than 40% of the patients receiving benfluorex (vs. 15% receiving placebo) had an A1C decrease ≥1%, and a significantly larger number of patients than in the placebo group achieved a target A1C ≤7% or even a target of ≤6.5%. This amplitude of A1C decrease has been demonstrated to be of significant clinical benefit. In the UKPDS 35, a 1% reduction in A1C corresponded to a 37% reduction in risk of microvascular complications and a 21% reduction in the risk of death related to diabetes (15). Although the majority of patients did not have any major plasma lipid abnormalities at baseline, some mild improvements were observed, particularly for LDL cholesterol. An interesting finding is the concomitant improvement in other metabolic parameters, especially the insulin resistance index and plasma triglycerides, which have been shown to be independent risk factors for coronary heart disease in the type 2 diabetic patient population, whereas no deleterious effect was observed on weight and blood pressure (12). Finally, the reduction in A1C with benfluorex in this study is consistent with the findings of previous studies: means ± SEM −0.86 ± 0.17% (P < 0.001) after 29 weeks of monotherapy in 294 patients whose diabetes was inadequately controlled by diet alone (9) and, in a less powered study (68 patients), −0.66% after 12 weeks of combined treatment with sulfonylurea and benfluorex (8). The magnitude of the effect of benfluorex on blood glucose control is comparable with those observed with other recently registered drugs. Thus, trials using rosiglitazone in combination with a sulfonylurea showed between-group differences versus placebo of −1.0 and −0.9% for A1C (16,17).

As expected, episodes suggestive of hypoglycemia were more frequent with benfluorex than with placebo, although overall there were not many of these hypoglycemic events as a consequence of the sulfonylurea treatment and of the improvement in glycemic control demonstrated in the benfluorex group. The downward shift in glucose levels over 18 weeks combined with the known improvement in insulin sensitivity may explain the small increase in the rate of hypoglycemia. Previous studies have shown that benfluorex does not have any effect on insulin secretion by itself.

This trial was specifically carried out in patients for whom metformin use was inappropriate due to known intolerance or a contraindication. According to this patient selection strategy, it was easy to establish that there is obviously no overlap between poor gastrointestinal tolerance to metformin and gastrointestinal side effects (mainly diarrhea) related to benfluorex; the proportion of patients who did not tolerate metformin before the trial and who developed such a side effect was 15% with both benfluorex and placebo treatment. Moreover, in the whole study population, gastrointestinal effects remained moderate, resulting in eight patients withdrawing (five in the benfluorex group and three in the placebo group). Overall tolerance was similar in both groups. Serious adverse events were more frequent in the benfluorex group but without evidence of any causality relationship. No concern was raised in previous studies with 5,093 patients taking benfluorex from 2 weeks to 3 years.

This trial confirms the efficacy of benfluorex in achieving targets for glycemic control and suggests the possibility of using benfluorex as a first-line insulin sensitizer, particularly for an unmet medical need when metformin is not tolerated or contraindicated. To properly confirm and evaluate the clinical relevance of this new authentic antidiabetic agent, it will be of interest to perform additional trials to assess by direct comparison the efficacy and safety profile of benfluorex versus other recent insulin sensitizers such as the thiazolidinediones in type 2 diabetic patients.