Efficacy of Benfluorex in Combination with Sulfonylurea in Type 2

Clinical Care/Education/Nutrition ORIGINAL ARTICLE

Efﬁcacy of Benﬂuorex in Combination With Sulfonylurea in Type 2 Diabetic Patients An 18-week, randomized, double-blind study

1 6 PHILIPPE MOULIN, MD, PHD DRAGOMIR KOEV, MD apy is being recognized as desirable in 2 7 MARIE ANDRE, MD RAY MOORE, MRCP type 2 diabetes. According to the patho- 3 8 HASAN ALAWI, MD VIOREL SERBAN, MD 4 2 physiological mechanism of the disease, LELITA C. DOS SANTOS, MD BRIGITTE PICANDET, MD 5 2 the combination of an insulin secreta- ABDUL K. KHALID, MD MARIE FRANCILLARD, MD, PHD gogue with an insulin sensitizer is the most common drug treatment proposed for type 2 diabetic patients, in addition to OBJECTIVE — The aim of this study was to demonstrate the superiority of benfluorex over diet, lifestyle counseling, and exercise. In- placebo as an add-on therapy in type 2 diabetic patients in whom diabetes is insufficiently sulin sensitizers are not a homogenous controlled by sulfonylurea monotherapy and who have a limitation for the use of metformin. drug class. For many years, metformin was the only drug of this type on the RESEARCH DESIGN AND METHODS — Type 2 diabetic patients with HbA1c (A1C) (7–10%) who were receiving the maximum tolerated sulfonylurea dose and had a contraindi- market; now thiazolidinediones are an cation to or poor tolerance of metformin were randomly assigned (double blind) to receive alternative. Both metformin and thiazo- benfluorex 450 mg/day (n ϭ 165) or placebo (n ϭ 160) for 18 weeks. The main efficacy criterion lidinediones have limitations on their use: was A1C, analyzed as the change from baseline to the end of treatment using ANCOVA with gastrointestinal intolerance is common with baseline and country as covariates. Secondary criteria were fasting plasma glucose (FPG), insulin metformin, and although serious complica- resistance, and plasma lipid level. tions such as lactic acidosis are very infre- quent, caution should be exercised in RESULTS — Both groups were similar at baseline in the intention-to-treat population. A1C significantly decreased with benfluorex from 8.34 Ϯ 0.83 to 7.52 Ϯ 1.04% (P Ͻ 0.001) and patients with cardiovascular, pulmonary, or tended to increase with placebo from 8.33 Ϯ 0.87 to 8.52 Ϯ 1.36% (NS), resulting in a mean renal insufficiency, which may be associ- adjusted difference between groups of Ϫ1.01% (95% CI Ϫ1.26 to Ϫ0.76; P Ͻ 0.001). The target ated with hypoxia or drug accumulation A1C (Յ7%) was achieved in 34% of patients receiving benfluorex versus 12% of patients (3). In prescribing thiazolidinediones, cau- receiving placebo. Significant between-group differences in favor of benfluorex were observed tion is required in patients with fluid reten- for mean FPG (Ϫ1.65 mmol/l) (P Ͻ 0.001) and for homeostasis model assessment of insulin tion, mildly elevated liver enzyme levels, or resistance. Overall tolerance was similar in both groups. Serious adverse events were more underlying cardiac insufficiency. Thus, al- frequent in the benfluorex group, without evidence of causality relationship. ternative drugs can be useful in patients CONCLUSIONS — Benfluorex as an add-on therapy was superior to placebo in lowering with intolerance of or contraindications to A1C with a between-group difference of 1% in type 2 diabetic patients whose disease was metformin and/or thiazolidinediones. insufficiently controlled with sulfonylurea alone and in whom metformin was contraindicated or Benfluorex (2-[[(1RS)-1-methyl-2-[3-(tri- not tolerated. fluoromethyl)phenyl]ethyl]-amino]ethyl benzoate hydrochloride) is a compound Diabetes Care 29:515–520, 2006 with both lipid-lowering and antihypergly- cemic actions. Benfluorex is almost com- he incidence and progression of mi- tes Study (UKPDS) in 1998, recom- pletely absorbed after oral administration. It crovascular complications in type 2 mended target values for HbA1c (A1C) in is fully metabolized by a circulating esterase T diabetes are strongly associated with type 2 diabetes were reduced from 7 to and mainly eliminated in urine. It was ini- the degree of hyperglycemia. Following 6.5% or even lower (1,2). To achieve such tially developed in patients with metabolic the results of the U.K. Prospective Diabe- goals, early initiation of combined ther- syndrome, mainly for European market. In ●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●●● further preclinical and clinical studies, ben- From the 1Endocrinology Department, Cardiovascular Hospital, University Claude Bernard, Lyon, France; fluorex has been shown to improve insulin the 2Institut de Recherches Internationales Servier, Courbevoie, France; 3Schwerpunkt Diabetologie Klinik, sensitivity, decrease hepatic glucose pro- Saarlouis, Germany; the 4University Hospital of Coimbra, Coimbra, Portugal; the 5Universiti Kebangsaan, 6 7 duction, and improve aerobic glucose utili- Kuala Lumpur, Malaysia; the University of Sofia, Sofia, Bulgaria; the Umhlanga Hospital, Durban, South zation in skeletal muscle (4–7). Benfluorex Africa; and the 8Spitalul Judetean, Timisoara, Romania. Address correspondence and reprint requests to Prof. P. Moulin, Service d’Endocrinologie—Unite 11, decreases gluconeogenesis by affecting the Hoˆpital Cardio-Vasculaire Louis Pradel, 28 avenue Doyen Lepine, 69677 Bron Cedex, France. E-mail: expression of genes encoding enzymes in- [email protected]. volved in both glucose and fatty acid metab- Received for publication 2 August 2005 and accepted in revised form 20 November 2005. olism. The mechanisms of these metabolic Abbreviations: FPG, fasting plasma glucose; FSI, fasting serum insulin; HOMA-IR, homeostasis model assessment of insulin resistance; ITT, intention to treat; NCEP, National Cholesterol Education Program; actions in liver and muscle differ from those UKPDS, U.K. Prospective Diabetes Study. of metformin: in particular, benfluorex de- A table elsewhere in this issue shows conventional and Syste`me International (SI) units and conversion creases gluconeogenesis by inhibition of factors for many substances. ␤-oxidation (4). These findings prompted © 2006 by the American Diabetes Association. The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby the development of the product in overt marked “advertisement” in accordance with 18 U.S.C. Section 1734 solely to indicate this fact. type 2 diabetes. A small-scale exploratory

DIABETES CARE, VOLUME 29, NUMBER 3, MARCH 2006 515 Benfluorex in combination with sulfonylurea trial demonstrated that in overweight dia- ing either placebo or 150 mg benfluorex using a two-sided Student’s t test for in- betic patients whose diabetes was poorly (Les Laboratoires Servier Industrie, Gidy, dependent samples with 90% power, a controlled by a sulfonylurea as mono- France), the dose being gradually in- type 1 error of 5%, and a 10% drop-out therapy, the combination with benfluorex creased over the first 3 weeks from one rate. The intention-to-treat (ITT) popula- over 3 months led to a significant improve- tablet per day to the recommended dose tion was defined as all randomly assigned ment in blood glucose control (8). A recent of one tablet three times daily. The dose of patients exposed to the study drug with a large-scale, 6-month trial in diet-failed type sulfonylurea was to be maintained baseline value and at least one A1C value 2 diabetic patients demonstrated the effi- throughout the study, except in patients during treatment. Per-protocol patients cacy of benfluorex versus placebo in mono- with severe or repeated hypoglycemia. were defined as patients who completed therapy, with a significant 0.86% reduction Subjects were asked to maintain their the study without a deviation affecting the in A1C and a good safety profile (9). The usual diet and physical activity. The pri- A1C evaluation. For efficacy analyses, the most common side effects are gastrointesti- mary efficacy end point was A1C. The difference between benfluorex and pla- nal disorders, asthenia, somnolence, dizzi- secondary end points were FPG, fasting cebo treatment was studied as the change ness, and headache. These data, together serum insulin (FSI), and lipid profile, from baseline to last value with treatment with the assumption of a potential benefit of each assessed at inclusion and at weeks 4, using ANCOVA with baseline and geo- benfluorex in combination with a sulfonyl- 10, and 18, and insulin resistance index graphic area as covariates. The within- urea, led to the design of the present study evaluated at baseline and week 18. An in- group evolution between baseline and the to confirm the antidiabetic properties of teractive voice response system was used last value during treatment was studied in benfluorex and to demonstrate its superior- for centralized randomization, stratified each treatment group using a two-sided ity over a placebo as add-on therapy in type by baseline A1C (Յ8orϾ8%) and geo- paired Student’s t test. Nonparametric ap- 2 diabetic patients whose diabetes is insuf- graphic area. The protocol was approved proaches were performed for triglyceride- ficiently controlled with sulfonylurea by each institution’s ethics committee. mia and HOMA-IR analyses. Three monotherapy. In Europe the first-line ther- The trial was conducted in accordance subgroups were predefined according to apy for this group of patients involves the with the ethical principles stated in the regulatory requirements (12): A1C at in- combination of a sulfonylurea and met- Declaration of Helsinki 1964, as revised clusion Ͼ8%, age at selection Ͼ65 years, formin; patients who could not be given this in Edinburgh, U.K., in October 2000. All and creatinine clearance Յ80 ml/min. combination due to a contraindication or subjects provided written informed con- Safety analyses were performed on all pa- intolerance to metformin were selected for sent. tients exposed to at least one dose of study this first pivotal trial. treatment. Final values for withdrawn pa- Assessments tients corresponded to the last values dur- RESEARCH DESIGN AND A1C, FPG, FSI, and standard biochemi- ing treatment. All statistical analyses were METHODS — Eligibility criteria were cal, hematological, and lipid parameters performed using SAS software (version type 2 diabetes with age Ն18 years, BMI were assessed centrally (MDS Pharma 8.2; SAS, Cary, NC). of 25–40 kg/m2, and A1C of 7–10% (in- Services, Hamburg, Germany). Creati- clusive) despite monotherapy with a sul- nine clearance was calculated according RESULTS fonylurea at the maximum tolerated dose to the Cockcroft formula (10). A1C was for at least 4 months. All patients had to assayed using an National Glycohemoglo- Demographic and baseline have either a history of gastrointestinal in- bin Standardized Program–certified characteristics tolerance to metformin or a contraindica- high-performance liquid chromatogra- A total of 325 patients were included: 165 tion to its use, such as renal impairment phy method (BioRad Variant I). FPG was were randomly assigned to benfluorex or any cardiac or respiratory condition assessed by a hexokinase method; all glu- and 160 to placebo. There were 317 pa- with a potential for tissue hypoxia (3). Ex- cose and lipid assays were performed on a tients (97%) in the ITT population. Thirty clusion criteria were severe renal impair- Hitachi 747 instrument (Boehringer withdrawals (9.2%) occurred, 21 for pa- ment, an alanine aminotransferase or Mannheim, Mannheim, Germany). FSI tients receiving benfluorex and 9 for pa- aspartate aminotransferase plasma level was measured by a specific radioimmu- tients receiving the placebo; 18 patients three times above the upper limit of nor- noassay; the insulin resistance index (ho- withdrew because of adverse events (12 mal, active proliferative retinopathy, un- meostasis model assessment of insulin receiving add-on benfluorex and 6 receiv- controlled high blood pressure (Ն180/ resistance [HOMA-IR]) was calculated ing the placebo), 2 patients withdrew be- 100 mmHg), or any disorder that could using the classic homeostasis assessment cause of lack of efficacy (1 in each group), interfere with the study conduct or end model (11). Safety was assessed by ad- and 10 patients withdrew for nonmedical point evaluation. Patients were with- verse event spontaneous reporting, phys- reasons. There were 265 patients (81%) drawn for lack of efficacy, which was de- ical examination, recording of vital signs, in the per-protocol population. The ITT fined as two fasting plasma glucose (FPG) laboratory tests, and 12-lead electrocar- and per-protocol populations were simi- measurements Ն15 mmol/l, after at least diogram at baseline and at week 18. Hy- lar at inclusion. Baseline characteristics 1 month of study treatment. poglycemic events were recorded from were broadly similar in the two groups A randomized, placebo-controlled, the patient diary and were based on sug- (Table 1). Overall, 73% of patients had a double-blind, parallel-group study, with gestive clinical symptoms only. metabolic syndrome [modified National an 18-week comparative treatment pe- Cholesterol Education Program (NCEP) riod, was conducted in 63 centers in Statistical analysis Adult Treatment Panel III definition (13)] seven countries. After a 2- to 4-week Sample size was estimated based on a and 57% of patients had previous gastro- run-in period, patients were centrally ran- 0.6% between-group difference in A1C intestinal intolerance to metformin. Con- domly assigned to receive tablets contain- and an assay standard deviation of 1.5%, comitant treatment was being taken by

516 DIABETES CARE, VOLUME 29, NUMBER 3, MARCH 2006 Moulin and Associates

Table 1—Baseline characteristics in the randomized population versus 27.5% receiving placebo (P Ͻ 0.001). Target values of Յ7% were Benfluorex Placebo achieved by 34.2% of patients receiving benfluorex versus 11.5% of patients re- n 165 160 ceiving placebo (P Ͻ 0.001), whereas Ϯ Ϯ Age (years) 62.9 10.8 64.8 10.3 18.5% of patients achieved an objective of Sex (men/women) 86/79 68/92 Յ6.5% with benfluorex versus 4.5% with Diabetes duration (years) 6.8 Ϯ 5.8 7.5 Ϯ 6.1 Ͻ ² Ϯ Ϯ placebo (P 0.001). In patients with BMI (kg/m ) 29.5 3.7 29.3 3.7 baseline A1C in the range of 7 to 8%, a Presence of metabolic syndrome (%) 72.1 73.1 target value of Յ7% was achieved by Hypertension (%) 68.5 70.6 47.1% of patients in the benfluorex group Sitting systolic blood pressure (mmHg) 139.2 Ϯ 13.3 138.7 Ϯ 14.0 versus 14.9% in the placebo group (P Ͻ Sitting diastolic blood pressure (mmHg) 80.9 Ϯ 7.3 80.0 Ϯ 7.2 0.001). Similar results were observed in Diabetes complications the per-protocol population, with a Coronary heart disease (%) 35.2 37.5 Ϫ Nephropathy (%) 12.1 11.9 1.08% difference of A1C between the Neuropathy (%) 14.5 15.0 benfluorex and placebo groups. FPG con- Ophthalmologic (%) 10.9 12.5 sistently decreased with benfluorex, with A1C (%) 8.32 Ϯ 0.83 8.32 Ϯ 0.87 a between-group difference after 18 Ϫ Ͻ FPG (mmol/l) 9.87 Ϯ 2.54 9.67 Ϯ 2.39 weeks of 1.65 mmol/l (P 0.001) (Ta- HOMA-IR (index)* 6.62 Ϯ 7.99 6.35 Ϯ 7.95 ble 2 and Fig. 1B). Similar changes over Creatinine clearance (ml/min) 93.2 Ϯ 33.5 86.0 Ϯ 28.6 time in A1C and FPG were seen in the Creatinine clearance Յ80 ml/min (%) 60.6 50.0 subgroups according to A1C Ͼ8%, age Total cholesterol (mmol/l) 5.54 Ϯ 1.04 5.49 Ϯ 1.13 Ͼ65 years, and impairment of renal func- LDL cholesterol (mmol/l) 3.57 Ϯ 0.80 3.55 Ϯ 0.89 tion. A slight decrease in weight was ob- HDL cholesterol (mmol/l) 1.27 Ϯ 0.31 1.28 Ϯ 0.28 served in both groups, with a Triglycerides (mmol/l) 2.35 Ϯ 2.01 2.16 Ϯ 1.33 nonclinically relevant difference between Data are means Ϯ SD unless otherwise indicated. *HOMA-IR was calculated for the ITT population. benfluorex and placebo: Ϫ1.3 kg with benfluorex versus Ϫ0.7 kg with placebo (P Ͻ 0.01). The superiority of benfluorex 280 patients (86%), mainly ACE inhibi- tors 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 Ͼ8% (94 patients receiving add-on benfluorex and 91 receiving placebo), of age Ͼ65 years (72 patients receiving benfluorex and 84 on placebo), and of creatinine clearance Յ80 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 Ϫ0.82% (P Ͻ 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 differ- Figure 1—Changes in ence of Ϫ1.01% (P Ͻ 0.001). A decrease A1C (A) and FPG (B) over Ն time in the ITT population. in A1C of 1% was seen in 42.9% of pa- Data are means Ϯ SEM. tients receiving add-on benfluorex versus ***P Ͻ 0.001 with be- 14.7% receiving placebo (P Ͻ 0.001). A tween-group test. F, ben- decrease in A1C of Ն0.5% was seen in fluorex group; E, placebo 69.6% of patients receiving benfluorex group.

DIABETES CARE, VOLUME 29, NUMBER 3, MARCH 2006 517 Benfluorex in combination with sulfonylurea over placebo with respect to A1C was in- episode required external assistance or sion. They had inadequate glycemic dependent of the decrease in body hospitalization. Headache was reported control, with a mean A1C of 8.32%. weight: the estimated between-group dif- by 2.4% of patients receiving benfluorex This study demonstrates that the ference in A1C from the covariance anal- and by 2.5% receiving placebo. Other ad- agent studied, benfluorex, as add-on ther- ysis performed with adjustment for verse events were reported in fewer than apy behaved throughout the trial as an weight evolution was Ϫ0.96% (P Ͻ 2% of patients. Very few adverse events active glucose-lowering drug that signifi- 0.001). were considered related to the study drug cantly and markedly improved glycemic Insulin resistance as assessed by (6.6% of patients receiving benfluorex vs. control. After 18 weeks, the mean A1C HOMA-IR decreased over the study by 1.3% receiving placebo, mostly diarrhea was significantly reduced by Ϫ0.82 Ϯ 20% (from 6.62 Ϯ 7.99 at baseline to or abdominal pain). Serious adverse 1.04% in the benfluorex group with a be- 4.87 Ϯ 3.67) with benfluorex and by 5% events (fatal and nonfatal) occurred in 11 tween-group difference of Ϫ1.01%. This (from 6.35 Ϯ 7.95 to 5.93 Ϯ 5.35) with patients. Two fatal events were reported reduction in A1C was seen in each sub- placebo, with a statistically significant be- in the benfluorex group, without any ob- group, especially in patients with a base- tween-group difference (P Ͻ 0.01) (Table vious causal relationship: one patient line A1C Ͼ8%, with a between-group 2). In terms of the lipid profile (Table 2), committed suicide (4 weeks after a isch- difference of Ϫ1.10%. More than 40% of with benfluorex, LDL cholesterol de- emic cerebrovascular attack) and the the patients receiving benfluorex (vs. creased by 6% and triglycerides by 7%. other died as a result of cerebrovascular 15% receiving placebo) had an A1C de- More than one-third (35.5%) of the pa- hemorrhage. One patient in the placebo crease Ն1%, and a significantly larger tients treated with benfluorex had a favor- group had two independent serious non- number of patients than in the placebo able evolution of their LDL cholesterol: fatal adverse events (unstable angina and group achieved a target A1C Յ7% or even from Ն130 mg/dl at baseline to below this hyperglycemia). Nine patients in the ben- a target of Յ6.5%. This amplitude of A1C level at the end of the study [cutoff point fluorex group had 10 serious nonfatal ad- decrease has been demonstrated to be of as used in the NCEP recommendations verse events: diabetic neuropathy, significant clinical benefit. In the UKPDS (14)] compared with 14.3% of patients pyelonephritis and ovarian cyst, nephro- 35, a 1% reduction in A1C corresponded receiving a placebo. At the end of the lithiasis, cerebrovascular accident, lum- to a 37% reduction in risk of microvascu- study, 34.5% of patients receiving benflu- bar radiculopathy, biliary neoplasm, lar complications and a 21% reduction in orex had plasma triglycerides Ͻ2.2 cardiac failure after inappropriate termi- the risk of death related to diabetes (15). mmol/l (baseline Ն2.2 mmol/l) versus nation of diuretic therapy, pleural metas- Although the majority of patients did not 24.1% of patients receiving placebo. The tasis, and abdominal pain. Of these, the have any major plasma lipid abnormali- changes from baseline in the benfluorex last three led to patient withdrawal. Only ties at baseline, some mild improvements group were Ϫ0.27 Ϯ 0.73 mmol/l for cal- for abdominal pain was there thought to were observed, particularly for LDL cho- culated LDL cholesterol and Ϫ0.05 Ϯ be a causal relationship; the others rea- lesterol. An interesting finding is the con- 2.08 mmol/l for triglycerides (median sons were considered to be related to each comitant improvement in other metabolic Ϫ0.12 mmol/l), with a significant be- patient’s medical history or concomitant parameters, especially the insulin resis- tween-group difference in favor of benflu- pathologic conditions. There were no sig- tance index and plasma triglycerides, orex (P Ͻ 0.001 for LDL cholesterol and nificant changes in the electrocardiogram which have been shown to be indepen- P ϭ 0.027 for triglycerides). No change in or biological safety chemical parameters, dent risk factors for coronary heart dis- HDL cholesterol was observed. including liver enzyme levels. Mean sit- ease in the type 2 diabetic patient ting diastolic and systolic blood pressures population, whereas no deleterious effect Safety and tolerability were similar in the benfluorex group and was observed on weight and blood pres- A similar proportion of patients in each the placebo group, with no clinically rel- sure (12). Finally, the reduction in A1C group (53.0% receiving benfluorex and evant changes at the end of the study. with benfluorex in this study is consistent 51.3% receiving a placebo) reported at with the findings of previous studies: least one adverse event. Gastrointestinal CONCLUSIONS — This is the first means Ϯ SEM Ϫ0.86 Ϯ 0.17% (P Ͻ disorders were the most common adverse large-scale study of the efficacy and safety 0.001) after 29 weeks of monotherapy in events, occurring in 15.1% of patients in of benfluorex as add-on therapy in type 2 294 patients whose diabetes was inade- the benfluorex group and 10.0% of pa- diabetic patients whose diabetes was sub- quately controlled by diet alone (9) and, tients in the placebo group, and mostly optimally controlled with sulfonylureas at in a less powered study (68 patients), involved diarrhea (6.0% of patients re- their maximum tolerated dose. The de- Ϫ0.66% after 12 weeks of combined ceiving benfluorex vs. 1.9% receiving sign versus placebo, the good compliance treatment with sulfonylurea and benflu- placebo). Two patients treated with ben- with the protocol procedure and with orex (8). The magnitude of the effect of fluorex were withdrawn because of diar- treatment (drop-out rate Ͻ10% with no benfluorex on blood glucose control is rhea (1.2%). In patients with known patients lost to follow-up), and the con- comparable with those observed with previous intestinal intolerance to met- sistency of results across the analyses other recently registered drugs. Thus, tri- formin, the incidence of gastrointestinal make the analysis of the results reliable. als using rosiglitazone in combination disorders was the same with benfluorex Moreover, the patient population was with a sulfonylurea showed between- and placebo (15.2 vs. 15.3%, respectively). representative of patients with type 2 di- group differences versus placebo of Ϫ1.0 Hypoglycemic symptoms were more abetes in terms of disease duration (mean Ϯ and Ϫ0.9% for A1C (16,17). frequently reported with sulfonylurea SD 7.1 Ϯ 6.0 years), BMI (mean 29 kg/m2), As expected, episodes suggestive of and benfluorex (24 episodes in 14 pa- presence of metabolic syndrome (in 73% of hypoglycemia were more frequent with tients) than with sulfonylurea and pla- patients), and the presence of diabetic car- benfluorex than with placebo, although cebo (13 episodes in 6 patients). No diovascular complications and hyperten- overall there were not many of these hy-

518 DIABETES CARE, VOLUME 29, NUMBER 3, MARCH 2006 Moulin and Associates aaaeetmtso h hnefredmnsbsln o h ihngopaayi n siae ftedfeec o efloe iu lcb fad of placebo minus benfluorex for difference the of estimates and analysis within-group the for baseline test. minus end *Nonparametric for change the of estimates are Data 2— Table 1 vrl % 6 8.34 161 (%) overall A1C A1C Age raiieclearance Creatinine P mo/)199.89 159 (mmol/l) FPG OAI idx 5 6.62 157 (index) HOMA-IR D mo/)193.60 149 (mmol/l) LDL D hlseo mo/)111.25 151 (mmol/l) cholesterol HDL rgyeie mo/)112.26 151 (mmol/l) Triglycerides poglycemic events as a consequence of Ͼ Ͼ the sulfonylurea treatment and of the im- 5yas()7 8.28 70 (%) years 65 %()9 8.93 93 (%) 8% provement in glycemic control demon- hne ngyei n ii aaeesi h T population ITT the in parameters lipid and glycemic in Changes strated in the benfluorex group. The downward shift in glucose levels over 18 weeks combined with the known im-

Յ provement in insulin sensitivity may ex- 0m/i % 28.17 62 (%) ml/min 80 plain 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

n inappropriate due to known intolerance or a contraindication. According to this patient selection strategy, it was easy to

aeieFnlChange Final Baseline establish that there is obviously no over- Ϯ Ϯ Ϯ Ϯ Ϯ Ϯ Ϯ Ϯ Ϯ lap between poor gastrointestinal toler- .37.52 0.83 .37.78 0.53 .07.42 0.80 .67.38 0.76 .78.67 2.57 .94.87 7.99 .03.33 0.80 .81.23 0.28 .22.21 1.62 ance to metformin and gastrointestinal side effects (mainly diarrhea) related to efloe Placebo Benfluorex benfluorex; the proportion of patients who did not tolerate metformin before the Ϯ Ϯ Ϯ Ϯ Ϯ Ϯ Ϯ Ϯ Ϯ trial and who developed such a side effect 1.04 1.11 0.89 0.97 2.46 3.67 0.72 0.30 2.18 was 15% with both benfluorex and placebo treatment. Moreover, in the whole study population, gastrointestinal effects Ϫ Ϫ Ϫ Ϫ Ϫ Ϫ Ϫ Ϫ remained moderate, resulting in eight pa- .2(.8 5 8.33 156 (0.08) 0.82 .5(.1 98.96 89 (0.11) 1.15 .6(.0 28.31 82 (0.10) 0.86 .8(.2 88.32 78 (0.12) 0.78 .2(.0 5 9.71 156 (0.20) 1.22 .5(.0 5 6.35 150 (0.50) 1.75 .7(.6 5 3.52 152 (0.06) 0.27 .3(.2 5 1.28 152 (0.02) 0.03 Ϫ

.1 5 2.11 152 0.11* tients 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 n without evidence of any causality relationship. No concern was raised in previous studies with 5,093 patients taking aeieFnlChange Final Baseline benfluorex from 2 weeks to 3 years. Ϯ Ϯ Ϯ Ϯ Ϯ Ϯ Ϯ Ϯ Ϯ This trial confirms the efficacy of ben- .78.52 0.87 .68.90 0.56 .78.28 0.87 .78.59 0.87 .910.22 2.39 .55.93 7.95 .93.56 0.89 .81.25 0.28 .22.13 1.32 fluorex in achieving targets for glycemic control and suggests the possibility of using benfluorex as a first-line insulin sensitizer, particularly for an unmet medical utdgopmasadaeepesda means as expressed are and means group justed Ϯ Ϯ Ϯ Ϯ Ϯ Ϯ Ϯ Ϯ Ϯ need when metformin is not tolerated or .601 (0.11) 0.19 1.36 1.39 1.24 .402 (0.15) 0.27 1.34 .805 (0.23) 0.51 2.88 5.35 .200 (0.05) 0.04 0.92 0.31 .80.05* 1.18 contraindicated. To properly confirm and evaluate the clinical relevance of this new authentic antidiabetic agent, it will be of Ϫ Ϫ Ϫ Ϫ interest to perform additional trials to as- .6(0.15) 0.06 .3(0.13) 0.03 .2(0.65) 0.42 .3(0.02) 0.03 sess by direct comparison the efficacy and safety profile of benfluorex versus other recent insulin sensitizers such as the thiazolidinediones in type 2 diabetic patients. Ϫ Ϫ Ϫ Ϫ Ϫ Ϫ Ϫ Ϫ .1(0.13) 1.01 .0(0.18) 1.10 .1(0.17) 0.81 .6(0.20) 1.16 .5(0.27) 1.65 .8(0.07) 0.28 .1(.3 NS (0.03) 0.01 .6(0.07) 0.16 Ϫ 0.81* Between-group Acknowledgments— This trial was sponsored

difference by Institut de Recherches International Servier, Ϯ

Dfrbaseline. for SD Courbevoie, France, and supervised by the au- thors, who were national coordinators. Ͻ Ͻ Ͻ Ͻ Ͻ Ͻ Ͻ Ͻ Parts of this work were presented as an oral P 0.001 0.001 0.001 0.001 0.001 0.01 0.001 0.05 value presentation at the 41st annual meeting of the European Association for the Study of Diabe- tes (EASD), Athens, Greece, September 2005.

DIABETES CARE, VOLUME 29, NUMBER 3, MARCH 2006 519 Benﬂuorex in combination with sulfonylurea

the blood glucose lowering action of ben- National Cholesterol Education Program References fluorex. Diabetes Metab Rev 9 (Suppl. 1): (NCEP) Expert Panel on Detection, Eval- 1. European Diabetes Policy Group 1999: A 19S–27S, 1993 uation, and Treatment of High Blood desktop guide to type 2 diabetes mellitus. 8. Stucci N, De Gregoris P, Lavielle R, To- Cholesterol in Adults (Adult Treatment Diabet Med 16:716–730, 1999 masi F: Therapeutic benefit of benfluorex Panel III). JAMA 285:2486–2497, 2001 2. American Diabetes Association: Stan- in type II diabetic patients treated with 14. Grundy SM, Cleeman JI, Merz NB, Brewer dards of medical care in diabetes. Diabetes sulfonylureas. J Diabetes Complications 10: B Jr, Clark LT, Hunninghake DB, Paster- Care (Suppl. 1)28:S4–S36, 2005 267–273, 1996 nak RC, Smith SC Jr, Stone NJ, for the 3. Howlett HC, Bailey CJ: A risk-benefit as- 9. Del Prato S, Erkelens DW, Leutenegger sessment of metformin in type 2 diabetes Coordinating Committee of the National M: Six-month efficacy of benfluorex vs. Cholesterol Education Program: Implica- mellitus. Drug Saf 20:489–503, 1999 placebo or metformin in diet-failed type 2 4. Kohl C, Ravel D, Girard J, Pegorier JP: tions of recent clinical trials for the Na- diabetic patients. Acta Diabetol 44:20–27, tional Cholesterol Education Program Effects of benfluorex on fatty acid and glu- 2003 Adult Treatment Panel III guidelines. Cir- cose metabolism in isolated rat hepato- 10. Cockcroft DW, Gault MH: Prediction of culation 110:227–239, 2004 cytes: from metabolic fluxes to gene creatinine clearance from serum creati- 15. U.K. Prospective Diabetes Study Group: expression. Diabetes 51:2363–2368, 2002 nine. Nephron 16:31–41, 1976 5. Bianchi R, Bongers V, Bravenboer B, Er- 11. Matthews DR, Hosker JP, Rudenski AS, Association of glycaemia with macrovas- kelens DW: Effects of benfluorex on insu- Naylor BA, Treacher DF, Turner RC: Ho- cular and microvascular complications of lin resistance and lipid metabolism in meostasis model assessment: insulin re- type II diabetes: prospective observa- obese type II diabetic patients. Diabetes sistance and beta-cell function from tional study (UKPDS 35). BMJ 321:405– Care 16:557–559, 1993 fasting plasma glucose and insulin con- 412, 2000 6. De Feo P, Lavielle R, De Gregoris P, Bolli centrations in man. Diabetologia 28:412– 16. Rosiglitazone Prescribing Information 2004. GB: Anti-hyperglycaemic mechanisms of 419, 1985 GlaxoSmithKline Clinical Trial Register— benfluorex in type II diabetes mellitus. Di- 12. The European Agency for the Evaluation Clinical Trial Information for Study abetes Metab Rev 9 (Suppl. 1):35S–41S, of Medicinal Products: Evaluation of Med- 49653/015 1993 icines for Human Use. London, May 2002 17. Rosiglitazone prescribing information 2004. 7. Riccio A, Vigili de Kreutzenberg S, Dorella (CPMP/EWP/1080/00) GlaxoSmithKline Clinical Trial Register— M, Da Tos V, De Biasi L, Marescotti MC, 13. NCEP Adult Treatment Panel III: Execu- Clinical Trial Information for Study Tiengo A, Del Prato S: Mechanism(s) of tive summary of the third report of the 49653/096

520 DIABETES CARE, VOLUME 29, NUMBER 3, MARCH 2006