The United Kingdom Prospective Diabetes Study (UKPDS) revealed that use of metformin in patients with type 2 diabetes was associated with more favorable outcomes with respect to the development of diabetes-related end points, all-cause mortality and stroke. However, the UKPDS did not investigate how patients who have poorly controlled glucose levels while receiving oral agents should be treated with insulin. Yki-Järvinen and associates compared four antidiabetic regimens to evaluate their effects on glucose control, weight gain and hypoglycemic episodes in patients with type 2 diabetes (formerly known as non–insulin-dependent diabetes) inadequately controlled with sulfonylureas alone.

The one-year study comprised 96 patients randomized into four treatment groups. All patients in each group received intermediate-acting human isophane insulin, 100 IU per mL at 9 p.m., with the dose adjusted according to home glucose monitoring. In addition to the bedtime insulin, the four treatment groups received as follows: group 1, glyburide, one 3.5-mg tablet before breakfast and two 3.5-mg tablets before dinner, plus metformin placebo; group 2, metformin, two 500-mg tablets before breakfast and two 500-mg tablets before dinner, plus glyburide placebo; group 3, glyburide and metformin together, taken in the same dosages as those used in the previously described groups; and group 4, no oral therapy and a second injection of insulin before breakfast. Patients with severe cardiovascular disease or microvascular disease, and patients who had previously received insulin therapy were excluded from the study.

Patients were instructed to increase their insulin dose by 4 IU per day if the fasting glucose level exceeded 144 mg per dL (8 mmol per L) on three consecutive measurements and by 2 IU per day if the fasting glucose level exceeded 108 mg per dL (6 mmol per L) on three consecutive measurements. While the initial bedtime doses of insulin were similar in the four groups, they varied among groups after 12 months of treatment. Specifically, the bedtime insulin dosage averaged 24 IU per day in patients receiving glyburide, 36 IU per day in those receiving metformin, 20 IU per day in those receiving both oral drugs and 24 IU per day in those receiving insulin both in the morning and at bedtime.

Unlike patients in the other three treatment groups, patients receiving metformin and bedtime insulin had a progressive decrease in the hemoglobin A_1c level. After one year, the hemoglobin A_1c level in the metformin group averaged 7.2 percent. The frequency of hypoglycemic episodes in this group was also significantly lower than in the other three groups. During the 12 months of therapy, the mean number of symptomatic hypoglycemic episodes per patient was 1.8 in patients receiving metformin plus bedtime insulin, compared with 3.4 hypoglycemic episodes in the glyburide group, 3.3 episodes in the group receiving both oral drugs and 3.9 episodes in the group receiving two doses of insulin.

With respect to weight gain on the different regimens, patients who took metformin plus bedtime insulin did not gain weight during the one-year study. In contrast, weight gain averaged 3.9 kg (8.6 lb) in patients receiving glyburide plus bedtime insulin, 3.6 kg (7.9 lb) in patients receiving bedtime insulin and the two oral drugs, and 4.6 kg (10.1 lb) in patients taking morning and bedtime insulin.

The authors conclude that the combination of bedtime insulin and metformin prevented weight gain during insulin therapy and resulted in the best overall glycemic control. The authors suggest the improved glycemic control occurring in all of the treatment groups is likely to be the result of teaching patients to adjust their insulin dosage on the basis of home glucose monitoring.

In an accompanying editorial, Nathan notes that the pathophysiology of type 2 diabetes—insulin resistance compounded by falling insulin secretion—can be treated in many ways. Insulin resistance can be decreased with diet, weight loss, exercise or medications, while insulin levels can be increased with agents that increase endogenous insulin secretion or with insulin injections. Nathan states that the 20-year UKPDS, partly because of its design, did not provide clear answers about which therapies are most advantageous. While the current study showed that the combination of bedtime insulin and metformin resulted in lower hemoglobin A_1c levels, less weight gain and fewer episodes of hypoglycemia, it remains to be shown whether this combination would avert the worsening metabolic control that occurs over time in patients with type 2 diabetes. In addition, it is not known whether the added expense of such a combination is worth its weight-sparing effect. The UKPDS showed that modest weight gain does not diminish the decrease in microvascular complications that accompanies improved control of diabetes and may not be such a dangerous effect of treatment. However, Nathan concludes that the current study provides yet more support for the idea that intensive therapy goals can be achieved in patients with type 2 diabetes.