Current issues of ACP Journal Club are published in Annals of Internal Medicine


Metformin alone or combined with glyburide improved glycemic control in poorly controlled NIDDM

ACP J Club. 1996 Mar-April;124:45. doi:10.7326/ACPJC-1996-124-2-045

Source Citation

DeFronzo RA, Goodman AM, and the Multicenter Metformin Study Group. Efficacy of metformin in patients with non-insulin-dependent diabetes mellitus. N Engl J Med. 1995 Aug 31;333:541-9.



To determine whether metformin is well tolerated and improves glycemic control in non-insulin-dependent diabetes mellitus (NIDDM) that is poorly controlled by sulfonylurea or diet.


2 randomized, double-blind, placebo-controlled trials.


26 clinical centers in the United States.


Eligible patients had poorly controlled NIDDM (established according to history and fasting plasma glucose levels), were obese (120% to 170% of ideal weight), were aged 40 to 70 years, and had normal renal and liver function. Exclusion criteria were symptomatic diabetes or cardiovascular disease, diastolic blood pressure (BP) > 100 mm Hg, concurrent disease, insulin use within 6 months, high alcohol use, and illicit drug use. 289 patients (mean age 53 y, 57% women) treated with diet alone were in study 1; 632 patients (mean age 55 y, 53% women) treated with a sulfonylurea were in study 2.


Patients in study 1 received 8 weeks of diet control and 5 weeks of metformin titration. 146 patients were assigned to placebo, and 143 were assigned to metformin starting at 850 mg/d and increased every 2 weeks until 2550 mg was reached or the plasma glucose level was < 7.8 mmol/L. Patients in study 2 continued to receive or began receiving glyburide, 5 mg twice daily for the first week and then 10 mg twice daily for 4 weeks. Patients were then assigned to glyburide, 10 mg twice daily, and metformin placebo (n = 209); metformin and glyburide placebo (n = 210); or metformin and glyburide, 10 mg twice daily (n = 213). Metformin doses were increased until either 2550 mg was reached or the fasting plasma blood glucose was ≤ 7.8 mmol/L. Study medication was given for 29 weeks in both studies.

Main Outcome Measures

Fasting plasma blood glucose levels, glycosylated hemoglobin levels, lipid levels, BP, and adverse effects.

Main Results

In study 1, patients given metformin, compared with placebo recipients, had a lower blood glucose level (10.6 vs 13.7 mmol/L, P < 0.001) and a lower glycosylated hemoglobin level (7.1% vs 8.6%); were more likely to have a plasma glucose level ≤ 7.8 mmol/L (22% vs 6%); had fewer treatment failures (1% vs 12%); and had lower lipid levels. The groups did not differ for weight loss or BP. In study 2, metformin alone had an effect similar to that of glyburide alone. Patients given metformin and glyburide compared with patients given glyburide alone had lower mean fasting blood glucose (10.5 vs 14.5 mmol/L, P < 0.001); glycosylated hemoglobin (7.1% vs 8.7%, P < 0.001); and lipid levels. The groups did not differ for adverse effects.


Metformin alone or with glyburide was well tolerated and reduced fasting glucose, glycosylated hemoglobin, and lipid levels in obese patients with poorly controlled NIDDM.

Source of funding: Lipha Pharmaceuticals, Inc.

For article reprint: Dr. R.A. DeFronzo, Diabetes Division, University of Texas Health Sciences Center, 7703 Floyd Curl Drive, San Antonio, TX 78284, USA. FAX 210-567-6554.


DeFronzo and colleagues have shown that healthy obese patients with NIDDM and plasma glucose levels of 8.5 to 19.4 mmol/L respond to maximal doses of metformin (alone or with glyburide) with a mean 3.3 to 3.9 mmol/L reduction in plasma glucose levels, a mean 1.5% reduction in glycosylated hemoglobin levels, and a mean 5% reduction in low-density lipoprotein cholesterol levels. What are we to do with these interesting results?

It is clear from this study that for a few patients—those without renal, liver, or cardiac disease—metformin is a valuable new tool for improving glycemia; indeed, as many as 25% of treated patients achieved a target fasting glucose level of 7.8 mmol/L. Because this study only addressed metabolic outcomes, it did not address the following health outcome-related questions: Do oral agents (alone or in combination) do more good than harm? For patients in whom sulfonylureas fail, is the addition of metformin better than switching to insulin? In fact, only 2 randomized trials have addressed these issues in healthy, low-risk patients: the much-discussed University Group Diabetes Program (1) and the nearly finished United Kingdom Prospective Diabetes Study (UKPDS) (2).

For now, we cannot conclude that tighter glucose control with pills will have the same benefit (and risks) as intensive insulin therapy in patients with insulin-dependent diabetes mellitus (3). For example, some currently unrecognized long-term toxic effects of pills may offset the advantage of improved glycemia. We must await the results of the UKPDS.

John S. Kizer, MD
Arthur T. Evans, MDUniversity of North Carolina at Chapel HillChapel Hill, North Carolina, USA


1. University Group Diabetes Program. Diabetes. 1970;19(Suppl 2):785-830.

2. United Kingdom Prospective Diabetes Study (UKPDS) 13. BMJ. 1995;310:83-8.

3. Diabetes Control and Complications Trial Research Group. N Engl J Med. 1993;329:977-86.