We read with interest the two studies conducted by Janka et al. (1) and Raskin et al. (2) and the accompanying editorial (3) regarding the initiation of insulin in patients with type 2 diabetes failing oral agents and noted the apparent discrepancy between the study results. We believe that the data of the two previous trials (1,2) may not be necessarily conflicting. One important factor that contributed to the more favorable HbA1c levels with insulin aspart 70/30 compared with insulin glargine in the study of Raskin et al. was the difference in insulin dosage between the two insulin regimens. Thus, at the end of the study, the mean daily insulin doses were ∼50% greater for the insulin aspart 70/30 group than for the glargine group (78.5 and 51.3 units, respectively). The magnitude of difference between the insulin doses remained essentially unchanged when expressed as units by weight (0.82 vs. 0.55 units/kg, respectively). The reasons for this substantial difference in insulin dose were not totally clear, particularly that patient characteristics were well balanced at the study entry, with nearly identical mean body weight, baseline HbA1c, and proportions of patients on metformin and pioglitazone in the two study groups (2). Failure of the investigators in the trial of Raskin et al. to increase the doses of glargine was unlikely attributed to the fear of hypoglycemia, since previous data have consistently shown that the peakless insulin glargine was associated with lower risk of hypoglycemia compared with traditional insulins (4). Moreover, there was enough room to increase glargine doses because the achieved mean fasting plasma glucose was 116 mg/dl, well above the study target of 80–110 mg/dl. The study of Raskin et al. was inevitably unblinded and financially supported by the manufacturer of insulin aspart 70/30. Therefore, the possibility that the investigators in the previous study were more aggressive in increasing the doses of insulin aspart compared with glargine to achieve better glycemic control with aspart could not be excluded. It would have been of interest if Raskin et al. had reported HbA1c values after adjusting for differences in insulin doses between the two insulin regimens.
Another factor that could have contributed to the variable results between the two trials (1,2) was the timing of injecting insulin glargine, which was administered in the morning in the study of Janka et al. (1) and at bedtime in the study of Raskin et al. (2). In one large trial, patients with type 2 diabetes who injected glargine in the morning had lower HbA1c values and lesser frequency of hypoglycemia than subjects who injected similar doses of glargine at bedtime (5).
