Choice of third oral diabetes drug can depend on each patient's needs

There is no clear choice for a third oral agent for controlling A1c levels above 7% following metformin and a sulfonylurea, researchers concluded from a meta-analysis of type 2 diabetics, so the best choice depends on the patient's need.

All the third-line agents decreased hemoglobin A1c levels about equally when added to metformin and a sulfonylurea, without any clear between-drug differences. Insulin was associated with more weight gain and hypoglycemia.

Researchers looked at primary change in A1c level, change in weight, and frequency of severe hypoglycemia in 18 trials that included 4,535 participants. Compared with placebo, drug classes did not differ in effect on A1c level (reduction ranging from −0.70% [95% credible interval {CrI}, −1.33% to −0.08%] for acarbose to −1.08% [CrI, −1.41% to −0.77%] for insulin). Results appeared in the May 17 issue of Annals of Internal Medicine.

In a pooled analysis (9 trials), the addition of a third agent led to a mean reduction of −0.96% in A1c level with statistically significant between-study heterogeneity (I²=63.7%; P=0.005). Change in A1c level was seen with each drug, varying from -0.6% for acarbose to −1.15% for thiazolidinediones.

Weight increase was seen with insulins (2.84 kg [CrI, 1.76 to 3.90 kg]) and thiazolidinediones (4.25 kg [CrI, 2.76 to 5.66 kg]). Weight loss was seen with glucagon-like peptide-1 agonists (−1.63 kg [CrI, −2.71 to −0.60 kg]), but they also were associated with more severe hypoglycemic reactions than any other drug class except insulin. Insulins (human and analogue) caused twice the absolute number of severe hypoglycemic episodes than noninsulin antihyperglycemic agents.

“It is common in clinical practice to initiate insulin therapy after failure of therapies of 2 oral antihyperglycemic agents,” the authors wrote. “In direct and network comparisons, insulins did not differ from other drug classes in their ability to decrease [hemoglobin] A1c levels, although the point estimate of effect was slightly greater for insulins in our analysis of trials directly comparing insulins with other drug classes.”