Background: The American Diabetes Association (ADA) recommends lifestyle modification and metformin (Glucophage) for primary treatment of type 2 diabetes mellitus. Most patients, however, will eventually require additional antidiabetic medication to reach treatment goals. The ADA recommends the use of sulfonylureas or insulin if an additional medication is needed. Other categories of antidiabetic medications, including thiazolidinediones (e.g., pioglitazone [Actos]), glucagon-like peptide-1 (GLP-1) analogues, glinides, alpha-glucosidase inhibitors, and dipeptidyl-peptidase-4 (DPP-4) inhibitors are not supported by the ADA guidelines. To evaluate the effectiveness of these other drugs in combination with stable doses of metformin, Phung and colleagues constructed a mixed-treatment comparison meta-analysis.

The Study: The authors conducted a systematic search of Medline and Cochrane CENTRAL from 1950 through January 2010 to identify studies that (1) did not use insulin; (2) evaluated patients who had inadequate control on maximum-tolerated doses of metformin; (3) were parallel-design randomized controlled trials; (4) compared the addition of one antidiabetic drug to metformin with another antidiabetic drug or with placebo; (5) followed patients for 12 to 52 weeks after randomization; and (6) reported A1C levels. Validity was assessed for each study with the Jadad scale to measure inherent bias. The end points measured included the mean change in A1C level, number of patients meeting the A1C goal of less than 7 percent, incidence of weight loss, and incidence of hypoglycemic episodes.

Results: Of the 45 full-text articles reviewed, 31 (representing 27 randomized controlled trials) were eligible for inclusion. When added to metformin, each class of antidiabetic medication showed statistically significant reductions in A1C level compared with placebo. Overall, each medication class was more effective than placebo in achieving the A1C goal, with some subgroup differences. Sulfonylureas, glinides, thiazolidinediones, and DPP-4 inhibitors were effective regardless of the starting A1C level; alpha-glucosidase inhibitors and GLP-1 analogues were also effective if the starting A1C level was 8 percent or more. Similarly, sulfonylureas, glinides, thiazolidinediones, and DPP-4 inhibitors were effective regardless of the study duration, whereas the alpha-glucosidase inhibitors and GLP-1 analogues were significantly more effective than placebo only in studies lasting longer than 24 weeks.

The adverse effect profiles were different among classes. Sulfonylureas, glinides, and thiazolidinediones were associated with an approximate 4.4-lb (2.0-kg) weight gain. Alpha-glucosidase inhibitors and DPP-4 inhibitors were weight-neutral, and the GLP-1 analogues were associated with statistically significant weight loss (approximately 3.8 lb [1.7 kg]). Hypoglycemic events also varied; sulfonylureas and glinides were associated with increased risk of hypoglycemia, whereas the other classes did not show any hypoglycemic tendencies compared with placebo.

Conclusion: The authors conclude that all classes of noninsulin antidiabetic drugs effectively lower A1C levels when used as second-line treatment with metformin. Weight gain, risk of hypoglycemia, cost, and other comorbidities should help determine which drug to select.