We have several comments on the study of Barnett et al. (1) regarding the addition of premeal inhaled human insulin (INH) versus metformin as adjunctive therapy in patients with type 2 diabetes failing sulfonylureas.
First, the authors mentioned that decreases in 2-h postprandial glucose at week 24 in the INH and metformin groups were similar. However, inspection of data in Table 1 revealed that reduction of 2-h postprandial hyperglycemia was more pronounced in the INH group compared with the metformin group (75 and 59 mg/dl, respectively) (1). The difference between the adjusted mean change in 2-h postprandial hyperglycemia was −11.4 mg/dl with a 95% CI of −18.6 to −4.19 mg/dl (i.e. a CI that did not include zero). We believe this difference was statistically significant, and contributed to the slightly lower HbA1c (A1C) level of 0.22 percentage points recorded in the INH arm compared with the metformin arm. Second, in the same table, the units of fasting and 2-h postprandial glucose should be written in milligrams per deciliter and not millimoles per liter. Third, the mean fasting plasma glucose values achieved with the INH (172 mg/dl) and with metformin (169 mg/dl) after 24 weeks of intervention were much higher than the study glycemic target of 80–140 mg/dl (1). Furthermore, only a small percentage of patients, 23 and 25% in the metformin and INH groups, respectively, achieved the A1C goal of <7% recommended by the American Diabetes Association (2). Thus, in this diabetic population poorly controlled on sulfonylureas (mean baseline A1C 9.7%), neither the addition of metformin nor premeal INH was the ideal approach to optimize glycemic control. In a similar trial conducted by Rosenstock et al. (3), including patients with type 2 diabetes poorly controlled (mean baseline A1C 9.2%) on two oral agents (mainly metformin and sulfonylureas), the proportion of subjects who attained A1C <7% after the addition of premeal INH for 12 weeks was 32%, slightly better than the current study (1), but still far from being perfect. We speculate whether insulin-induced hypoglycemia was the main limiting factor that prevented the investigators (1) from increasing the insulin dose. In this respect, it would be interesting to know whether hypoglycemia associated with the use of INH occurred more commonly in specific timing or was unpredictable, implying possible inconsistency of bioavailability of the INH.
Overall, the studies of Barnett et al. (1) and Rosenstock et al. (3) suggested that premeal INH plus one (1) or even two (3) oral agents may not be a sufficient therapy in diabetic patients with “very high” A1C level. A more aggressive strategy including a basal/bolus insulin regimen added to oral agents may be more effective in this challenging diabetic population frequently encountered in clinical practice. This strategy clearly deserves further evaluation in clinical trials.