Addition of Benfluorex to Biguanide Improves Glycemic Control in Obese Non–Insulin-Dependent Diabetes

ORIGINAL ARTICLES

Addition of Benfluorex to Biguanide Improves Glycemic Control in Obese Non–Insulin-Dependent Diabetes: A Double-Blind Study versus Placebo Patrick Roger Jean Auclair Patrice Drain

ABSTRACT The oral antidiabetic benfluorex lowers insulin resistance in liver and muscle without stimulating insulin secretion; it is more effective than a biguanide in lowering insulin and triglyceride levels and does not elevate lactate. This double-blind multicenter community study compared the addition of benfluorex versus placebo to diet plus metformin in 127 uncontrolled obese type II diabetics. After a 2-month run-in on diet plus metformin, overall mean glucose was 7.7 mmol/L or greater (fasting) and/or 11 mmol/L or greater (2 h after 75-g oral glucose), with 20% or greater excess body weight [body-mass index (women/men): > 26.9/> 27.2 kg/m2). Patients were then randomized to adjuvant benflouorex (n 5 63; 150 mg t.i.d.) or placebo (n 5 64) for 90 days, with centralized biochemical monitoring. On benfluorex, glucose decreased from day 0 to

INTRODUCTION

B

enfluorex is a novel trifluoro compound which lowers elevated lipid and glucose levels.1–3 It decreases hypertriglyceridemia by approximately 35% and elevates HDL cholesterol.4–6

Department of Endocrinology and Nutrition (P.R.), Haut Le´veˆque Hospital, Pessac, IRIS (J.A.), 6 place des ple´iades, Courbevoie, and Biopharma (P.D.), 29 rue du pont, Neuilly-sur-Seine, France Reprint requests to be sent to: Prof. Patrick Roger, Hoˆpital du Haut Le´veˆque, Service d’Endocrinologie Nutrition, Avenue Magellan, 36600 Pessac, France. Journal of Diabetes and Its Complications 1999; 13:62–67  1999 Elsevier Science Inc. All rights reserved. 655 Avenue of the Americas, New York, NY 10010

day 90 (fasting: 8.99 6 2.76 to 7.81 6 2.32 mmol/L, p 5 0.002; 2 h post-load: 16.56 6 4.49 to 15.09 6 5.09 mmol/L, p 5 0.029) versus increases on placebo (intergroup p 5 0.003 and p 5 0.001, respectively). HbA decreased on benfluorex (7.47 6 1.44 to 7.12 6 1.13%; p 5 0.013) versus no change on placebo. Basal and stimulated insulin and C peptide did not change in either group. Body weight remained similar in both groups. Tolerability was good in both groups, with no increase in hypoglycemia on benfluorex. Adjuvant behfluorex improves glycemic control in obese type II diabetics uncontrolled by diet plus metformin. The combination is safe, welltolerated and suitable for introduction into routine practice. (Journal of Diabetes and Its Complications 13; 2: 62–67, 1999.)  1999 Elsevier Science Inc.

Animal studies show that it also decreases insulin resistance in liver and muscle,7–9 which may account for the improved glycemic and insulin profiles. These findings were confirmed in humans in three clinical pharmacology studies using the current reference method for evaluating insulin resistance, the hyperinsulinemic euglycemic clamp.10–12 As an increasing number of non–insulin-dependent diabetic patients respond poorly to conventional sulfonylurea or biguanide therapy, the addition of benfluorex to enhance insulin sensitivity is an obvious therapeutic option.13 Clinical studies have already shown that benfluorex improves the glycemic profile when 1056-8727/99/$–see front matter PII S1056-8727(98)00004-X

J Diab Comp 1999; 13:62–67

BENFLUOREX ADDED TO BIGUANIDE

TABLE 1. SCHEDULE OF INVESTIGATIONS Parameter Clinical examination Height Weight Waist/hip ratio Blood pressure Heart rate Hypoglycemia screen Blood biochemistry HbA1C PAI1 Fasting glucose Fasting insulin Fasting C peptide Glucose 2 h post-load Insulin 2 h post-load C peptide 2 h post-load Triglycerides Total cholesterol HDL cholesterol Apolipoprotein A1 and B Uric acid Lactate Creatinine SGOT SGPT gGT

D-60

D0

D45

D90

x

x x x x x x x

x

x

x x x x x

x x x x x

x

x x x x x x x x x x x x x x x x x x

x

x x x x x x x x x x x x x x x x x x

added to a sulfonylurea.6,14,15 Also, compared to a biguanide, benfluorex is more effective in lowering elevated insulin and triglyceride levels, and does not elevate lactate.16 We report a randomized double-blind study of the impact on glycemic and insulin parameters of adding benfluorex to a biguanide in poorly controlled non–insulin-dependent diabetes mellitus (NIDDM). METHODS Design. This was a community-based multicenter double-blind study in male and female diabetic patients aged 30–70 years providing informed written consent. The inclusion criteria after an open 60-day run-in on diet plus biguanide were the following: NIDDM treated with diet 1 metformin alone (850 mg 3 2); treatment compliance; stable inadequacy of glycemic control, shown by a treated blood glucose 11 mmol/L or less but 7.7 mmol/L (fasting) or more and/ or 11 mmol/L or more 2 h after a 75-g oral glucose load; difference of 2% or less between two HbA1C estimations 2 months apart; treated insulin 15 mU/mL or more (fasting) and/or 80 mU/mL or more 2 h postload; and a stable excessive body-mass index (BMI) of 26.9 kg/m2 or more (women) or 27.2 kg/m2 or more (men) (Table 1). Exclusion criteria were the following: maturity-onset

63

diabetes in the young; severe intercurrent illness; kidney or liver failure; severe hypertension (diastolic . 110 mm Hg and/or systolic . 165 mm Hg); chronic pancreatitis; and alcoholism. Non permitted concomitant treatments were b-blockers, steroids, oral contraceptives, and any other drug impacting on blood levels of glucose (in particular sulfonylureas and insulin), insulin, or lipids. Treatment with lipid-lowering drugs, diuretics, or angiotensin-converting enzyme inhibitors was permitted provided it had been received at the same dose for 6 months or more and was maintained unchanged throughout the study. At the end of the run-in, 127 patients were randomized to adjuvant placebo or benflouroex (benflouroex hydrochloride 150 mg; Mediatort, Les Laboratoires Servier, France-92415 Courbevoie) one tablet in the evening in week 1, one tablet at midday and in the evening in week 2, followed by one tablet morning, midday and evening from week 3 (150 mg t.i.d.) to their basic diet plus metformin therapy. The duration of combination therapy was 90 days with two visits at Descriptive statistics were used for pretreatment variables. Groups were compared at baseline using Student’s t test for independent samples (quantitative variables) and the x2 test of Fisher’s exact test depending on the sample size (qualitative variables). Efficacy was measured by comparing D90 values versus baseline (the D45 data were not used). Intergroup comparisons were performed using Student’s t test for independent samples and intragroup comparisons using Student’s t test for paired samples. All tests were twosided using a significance level of 0.05. RESULTS 127 patients were randomized to benflouorex plus biguanide (n 5 63) or placebo plus biguanide (n 5 64). The groups did not differ at baseline (Tables 2 and 3). Women were in the majority (57.1% and 54.7% in the benfluorex and placebo groups, respectively), and mean age was 55 years. Mean known duration of diabetes exceeded 7 years. The mean BMI exceeded 33 kg/m2 in both groups with waist/hip ratios of 0.95 6 0.07 and 0.94 6 0.09 in the benfluorex and placebo groups, respectively. Poor diabetic control was evidenced by mean blood glucose 9 mmol/L or more (fasting) and 16 mmol/L or more (2 h post-load) (Table 4). 30 and 29 patients in the benfluorex and placebo groups, respectively, failed to comply strictly with the insulin inclusion criterion as the results arrived after randomization. However, the mean fasting insulin exceeded 15 mU/mL, rising to 55.2 6 47.5 mU/mL and 52.5 6 48.7 mU/mL in the benfluorex and placebo groups, respectively, 2 h post-load. The mean HbA1C exceeded 7.5% in both groups. In the overall population, total cholesterol was in the upper normal range (5.80 6 1.01 mmol/L), HDL

64

ROGER ET AL.

J Diab Comp 1999; 13:62–67

TABLE 2. CLINICAL CHARACTERISTICS AT INCLUSION [m 6 sd (range)] Benfluorex (n 5 63) Age (years) Sex [n (%)] Women Men Duration of diabetes (months) Diabetic complication present Family history Height (cm) Weight (kg) BMI (kg/m2) Waist/hip ratio

Placebo (n 5 64)

p

56.1 6 9.6 (29–70)

55.0 6 9.2 (36–70)

0.532

36 (57.1%) 27 (42.9%) 94.9 6 84.3 (4–384) 23 (36.5%) 31 (49.2%) 164.4 6 9.9 (135–206) 89.1 6 15.7 (63–137) 33.1 6 6.0 (24.5–53.5) 0.95 6 0.07 (0.80–1.14)

35 (54.7%) 29 (45.3%) 82.2 6 74.3 (6–300) 18 (28.1%) 38 (59.4%) 164.5 6 9.2 (147–185) 89.3 6 16.4 (59.6–144.7) 32.9 6 5.0 (26.9–45.2) 0.94 6 0.09 (0.76–1.14)

0.781

cholesterol rather low (0.45 6 0.13 mmol/L), triglycerides moderately raised (2.04 6 1.27 mmol/L), and the apolipoprotein A1/B ratio greater than 1, (1.22 6 0.35). In the benfluorex and placebo groups, 12 and 16 patients respectively, were receiving lipid-lowering therapy at inclusion. There were 15/127 dropouts (11.8%): ten in the benfluorex group (due to an adverse event in four cases) and five in the placebo group (due to an adverse event in two cases). Efficacy was analysed in 53 and 59 patients in the benflouorex and placebo groups, respectively. Efficacy. Blood glucose: • Intention to treat population: mean fasting blood glucose in the benfluorex group decreased from D0 to D90 by 1.18 6 2.7 mmol/L (13%) versus an increase of 0.81 6 4.1 mmol/L on placebo (p 5 0.003; Table 4 and Figure 1). Mean blood glucose 2 h postload decreased from 16.6 6 4.5 to 15.1 6 5.1 mmol/L (p 5 0.029) on benfluorex vs an increase (p 5 0.018) on placebo (intergroup p 5 0.001; Figure 2). Mean HbA1C decreased from 7.47 6 1.44 to 7.12 6 1.13% on D90 on benfluorex (p 5 0.013) versus no change on placebo (Intergroup p 5 0,197; Figure 3). • Protocol population (n 5 23 and n 5 27 in the benfluorex and placebo groups, respectively): results were similar. Mean fasting blood glucose in the benfluorex group decreased by 1.96 6 3.3 mmol/L (p 5 0.010) versus no change on placebo (intergroup p 5 0.034). Mean blood glucose 2 h post-load decreased by 1.62 6 5.25 mmol/L on benfluorex (intergroup p 5 0.074; the difference was non significant due to

0.367

0.960 0.870 0.259

the decreased statistical power). Mean HbA1C decreased from 8.03 6 1.45 to 7.35 6 0.92% (p 5 0.003) on benfluorex vs no change (p 5 0.775) on placebo (intergroup p 5 0.044, despite the loss of statistical power). Insulin and C peptide: The groups did not differ. A decrease in mean values was observed in both parameters. Lipids: Mean total cholesterol decreased from D0 to D90 on benfluorex from 5.84 6 0.83 to 5.59 6 0.91 mmol/L (4%, p 5 0.017; Table 4 versus no change on placebo (intergroup p 5 0.045). There were no intergroup differences in HDL cholesterol or triglycerides. Mean apo A1 and B levels showed quantitatively similar decreases on benfluorex and quantitatively similar increases on placebo, with the result that apo A1/B ratios remained unchanged in both groups (Table 4). Mean body weight decreased similarly in both groups: 1.89 6 2.77 kg on benfluorex and 1.09 6 2.99 kg on placebo. Mean BMI underwent a similarly significant decrease in both groups. Adverse Events. 27 patients on benfluorex (42.9%) and 24 patients on placebo (37.5%) reported 1 or more adverse events (intergroup p 5 0.538). On benfluorex, there were more complaints of diarrhea (n 5 14), fatigue (n 5 8) and headache (n 5 6). All were moderate to mild, prompting dropout in only four patients on benfluorex and one on placebo. Hypoglycemic episodes were reported by only one benfluorex patient on D90 and by four and two placebo patients on D45 and D90, respectively. All were transient and moderate.

J Diab Comp 1999; 13:62–67

BENFLUOREX ADDED TO BIGUANIDE

65

TABLE 3. BLOOD BIOCHEMISTRY AT INCLUSION [m 6 sd (RANGE)] Parameter Fasting glucose (mmol/l) Glucose 2 h post-load (mmol/l) Insulin (mU/ml) Insulin 2 h post-load (mU/ml) Fasting C peptide (pmol/ml) C peptide 2 h post-load (pmol/ml) Total cholesterol (pmol/ml) HDL cholesterol (mmol/l) Triglycerides (mmol/l) Apolipoprotein A1 (g/l) Apolipoprotein B (g/l) A1/B ratio HbA1C (%) Hemostasis : PAI1 (ng/ml)

Benfluorex (n 5 63)

Placebo (n 5 64)

9.40 6 3.05 (4.66–20.09) 17.26 6 4.83 (5.10–28.25) 17.41 6 8.79 (5.40–47.50) 55.20 6 47.45 (8.50–293.40) 1.08 6 0.39 (0.39–2.11) 2.62 6 1.33 (0.45–7.60) 5.91 6 1.00 (3.84–9.59) 0.43 6 0.13 (0.14–0.74) 2.05 6 0.84 (0.76–4.23) 1.47 6 0.30 (0.62–2.11) 1.29 6 0.24 (0.77–1.93) 1.18 6 0.35 (0.54–2.23) 7.59 6 1.43 (4.50–11.20) 49.27 6 30.78 (7.50–142.10)

9.13 6 2.81 (4.32–19.53) 16.38 6 5.38 (5.55–34.57) 18.30 6 20.35 (2.60–168.20) 52.47 6 48.66 (9.60–261.30) 1.06 6 0.47 (0.34–3.27) 2.47 6 1.26 (0.87–7.48) 5.69 6 1.02 (3.89–8.23) 0.46 6 0.13 (0.18–0.77) 2.03 6 1.59 (0.58–12.73) 1.45 6 0.23 (0.86–2.07) 1.21 6 0.28 (0.61–2.04) 1.26 6 0.35 (0.75–2.59) 7.82 6 1.38 (5.26–12.60) 48.99 6 34.81 (3.00–130.00)

DISCUSSION As part of a larger evaluation of adjuvant benfluorex in obese poorly controlled NIDDM, the study was run in parallel with a matching investigation of benfluorex 1 sulfonylurea. Selection criteria, outcome measures and analytical strategy were identical. However, the two studies differed in the basic treatment (biguanide vs sulfonylurea) and the sample size required to answer each question. For this reason, they should be considered as separate studies, with the present report addressing the addition of benfluorex to a biguanide. The study as performed largely complied with the protocol apart from two deviations. The first was the fasting blood glucose at inclusion which often exceeded the set upper limit (11 mmol/L). This is readily explained by poor diabetic control at inclusion and is not a major parameter invalidating the study conclusions. The second concerned the hyperinsulinemia criterion, the purpose of which was to confirm that benfluorex lowers blood glucose via a mechanism other than the stimulation of insulin secretion. Overall compliance with this requirement was poor, the main reason being that the results from the central laboratory often became avail-

p 0.600 0.333 0.750 0.750 0.814 0.507 0.248 0.189 0.912 0.709 0.078 0.211 0.349 0.968

able only after patients were already included. In a community study, it is difficult to require patients to return at a short interval to confirm as selection criterion. Moreover the criterion is itself open to question given the intrinsic variability in insulin levels over time : insulin status cannot be categorized on the basis of a single data point. The main purpose of the hyperinsulinemia criterion was to confirm an already documented pharmacological finding; it was not a predictor of treatment response. For these reasons the deviation should not be considered major in terms of the clinical (as opposed to the purely pharmacological) benefit of adding benfluorex to basic antidiabetic therapy. Though previous studies had shown that benfluorex improved glycemic control when added to standard sulfonylurea therapy,15 this is to our knowledge the first double-blind controlled study of benfluorex plus biguanide in a large population. The results show clear evidence of improved diabetic control. The selection criteria over the 2-month run-in ensured that NIDDM in the study population was both stable and poorly controlled. Adjuvant benfluorex induced highly significant glucose load, while fasting and stimulated insulin

66

ROGER ET AL.

J Diab Comp 1999; 13:62–67

TABLE 4. BLOOD GLUCOSE AND LIPID IMPACT OF ADJUVANT BENFLUOREX Parameter Fasting glucose (mmol/l) Post-load glucose (mmol/l) Fasting insulin (mU/ml) Post-load insulin (mU/ml) Fasting C peptide (pmol/ml) Post-load C peptide (pmol/ml) HbA1C (%) Total cholesterol (mmol/l) HDL cholesterol (mmol/l) Triglycerides (mmol/l) Apolipoprotein A1 (g/l) Apolipoprotein B (g/l) A1/B ratio

Group

N

P B P B P B P B P B P B P B P B P B P B P B P B P B

59 53 59 52 59 53 59 51 59 53 59 52 59 53 59 53 59 53 59 53 59 53 59 53 59 53

D0 8.96 8.99 16.00 16.56 18.93 16.74 53.90 57.32 1.08 1.05 2.51 2.75 7.77 7.47 5.65 5.84 0.46 0.43 2.05 1.95 1.44 1.47 1.20 1.27 1.25 1.20

6 6 6 6 6 6 6 6 6 6 6 6 6 6 6 6 6 6 6 6 6 6 6 6 6 6

2.44 2.76 4.51 4.49 20.99 8.63 49.84 50.80 0.47 0.38 1.28 1.41 1.19 1.44 0.94 0.83 0.12 0.13 1.65 0.79 0.23 0.31 0.25 0.23 0.31 0.36

D90

p*

6 6 6 6 6 6 6 6 6 6 6 6 6 6 6 6 6 6 6 6 6 6 6 6 6 6

0.133 0.002 0.018 0.029 0.286 0.319 0.174 0.309 0.676 0.173 0.979 0.162 0.624 0.013 0.891 0.017 0.510 0.797 0.678 0.965 0.178 0.045 0.105 0.135 0.499 0.896

9.77 7.81 17.57 15.09 16.14 15.52 47.23 51.72 1.06 0.97 2.51 2.50 7.69 7.12 5.68 5.59 0.44 0.43 2.06 1.89 1.48 1.42 1.25 1.23 1.22 1.20

3.53 2.32 6.86 5.09 9.58 10.05 45.99 37.19 0.42 0.42 1.46 1.02 1.47 1.13 0.92 0.91 0.11 0.12 1.02 0.82 0.27 0.29 0.25 0.24 0.30 0.35

D90–D0 0.811 21.179 1.577 21.471 22.785 21.217 26.675 25.596 20.022 20.081 20.003 20.247 20.08 20.35 0.03 20.25 20.01 20.00 0.01 20.06 0.04 20.05 0.05 20.05 20.02 0.00

6 6 6 6 6 6 6 6 6 6 6 6 6 6 6 6 6 6 6 6 6 6 6 6 6 6

4.09 2.69 4.98 4.71 19.87 8.80 37.25 38.90 0.41 0.42 0.94 1.25 1.20 0.98 0.78 0.68 0.09 0.08 1.42 0.57 0.19 0.18 0.23 0.19 0.30 0.23

p† 0.003 0.001 0.584 0.882 0.462 0.256 0.197 0.045 0.491 0.786 0.011 0.018 0.693

* p, intragroup D90 vs D0. †

p, intergroup D90 vs D0.

and C peptide levels underwent similar changes in both groups. Further evidence of improved glycemic control was the significant reduction in HbA1C levels on benfluorex vs no change on placebo. Other results suggest that benfluorex enhances peripheral insulin sensitivity in NIDDM without stimulating insulin secretion.17–19

FIGURE 1 Fasting blood glucose. Intergroups D90 (day 90)–D0 (day 0): p 5 0.003.

Total cholesterol decreased significantly on benfluorex. However, triglycerides remained largely unchanged in both groups, conflicting with recent evidence of a decrease in triglycerides on benfluorex.2

FIGURE 2 Blood glucose 2 h post-load.

J Diab Comp 1999; 13:62–67

BENFLUOREX ADDED TO BIGUANIDE

67

(types II B et IV) pre´sentant une glyce´mie normale, ou une anomalie de la tole´rance au glucose, ou un diabe`te insulinode´pendant. Sem Hoˆp Paris 60:2643–2651, 1984. 5. Velussi M: Inte´reˆt de l’association du benfluorex a` l’insuline ou aux sulfamides hypoglyce´miants. Vie Med 28:17–21, 1986. 6. Brun JM: Efficacite´ antidiabe´tique du benfluorex. Donne´es cliniques. Presse Med 21:1344–1352, 1992. 7. Brindley DN: Mode d’action du benfluorex. Donne´es re´centes. Presse Med 21:1330–1335, 1992. 8. Geelen MJ: Mechanisms responsible for the inhibitory effects of benfluorex on hepatic intermediary metabolism. Biochem Pharmacol 32:1765–1772, 1983. 9. Lacour F, Espinal J, Arnaud O, Duhault J: Improvement in glucose tolerance of insulin resistant rats after chronic or acute administration of benfluorex. Life Sci 53:1525– 1529, 1993. FIGURE 3 Glycosylated hemoglobin (HbA1c).

However, triglyceride levels in our study were approximately 2 mmol/L, i.e., patients were not manifestly dyslipidemic, which may explain the results observed. The apo A1/B ratio remained unchanged in both groups, as each parameter in the ratio underwent a proportional decrease on benfluorex and a proportional increase on placebo. Tolerability was satisfactory in both groups. In particular, the addition of benfluorex did not increase the risk of hypoglycemia. The prevalence of adverse events was similar in both groups, though transient diarrhea was more frequent on benfluorex. The evidence from this randomized double-blind study in a large and rigorously characterized population confirms the benefit of adding benfluorex to a biguanide (metformin) in obese poorly controlled NIDDM and suggests that this is a therapeutic strategy deserving introduction into routine practice. REFERENCES 1. Arnauld O, Nathan C: Antiobesity and lipid-lowering agents with anit-diabetic activity, in Bailey CJ, Flatt PR (eds), New Antidiabetic Drugs. London, Smith-Gordon, pp. 133–142, 1990. 2. Richard JL: Insulinore´sistance, diabe`te non insulinode´pendant et approches the´rapeutiques. Diabe´tologie 2: 137–140, 1996. 3. Ravel D, Laudignon N: Research prospects with benfluorex. J Diabetes Complications 10:246–254, 1996. 4. Bertolini S: E´tude a` long terme de l’efficacite´ du benfluorex (Mediator) chez 83 patients hyperlipide´miques

10. Erkelens DW: A novel agent in treatment of syndrome X. Diabetes 40 (supp 1):253A, 1991. 11. Vigili de Kreutzenberg S, Riccio A, De Biasi L, Da Tos V, Del Prato S: Short term treatment with benfluorex improves insulin action in non-insulin dependent patients. Diabetes 40 (supp 1):344A, 1991. 12. Bianchi R, Bongers V, Bravenboer B, Erkelens DW: Effects of benfluorex on insulin resistance and lipid metabolism in obese type II diabetic patients. Diabetes Care 16:557–559, 1993. 13. DeFronzo RA, Prato SD: Insulin resistance and diabetes mellitus. J Diabet Complications 10:243–245, 1996. 14. Parodi FA: E´tude clinique multicentrique de l’efficacite´ et de la tole´rance du benfluorex chez le diabe´tique dyslipide´mique. Vie Me´d 1, 1986. 15. Stucci N, de Gregoris P, Lavielle R, Tomasi F: Therapeutic benefit of benfluorex in type II diabetic patients treated with sulfonylureas. J Diabet Complications 10:267–273, 1996. 16. Broussole C, Coste M, Guillon-Metz F, Peltier C, Noe¨l G, Richard J-L, Roger P, San Marco JL, Schutz D, Simonin R, Vannereau D: Comparaison de l’action antidiabe´tique du benfluorex et de la metformine. E´tude multicentrique. Sem Hoˆp Paris 64:3153–3165, 1988. 17. Bianchi R, Bongers V, Bravenboer B, Erkelens DW: Benfluorex decreases insulin resistance and improves lipid profiles in obese type 2 diabetic patients. Diabetes Metab Rev 9(suppl 1):29S–34S, 1993. 18. Riccio A, Vigili de Kreutzenberg S, Dorella M, Debiasi L, Marescotti M-C, Tiengo A, Del Prato S: Mechanism(s) of the blood glucose lowering action of benfluorex. Diabetes Metab Rev 9(suppl 1):19S–27S, 1993. 19. Bianchi R, de Vries DE, Bravenboer B, Erkelens DW: Effect of benfluorex in addition to insulin therapy in obese type II diabetic patients with secondary failure to conventional oral treatment. Diab Nutr Metab 9:81–88, 1996.