We read with interest the results of the A1chieve study (1), a noninterventional, observational, 24-week study examining the safety and effectiveness of insulin analogs (detemir, aspart, biphasic aspart 30 [all Novo Nordisk, Bagsvaerd, Denmark]) alone or in combination in routine clinical use in 66,726 people with type 2 diabetes. The mean change in HbA1c level was −2.1% (SD 1.7%) (–23 mmol/mol [19 mmol/mol]) in insulin-naive participants and −1.8% (1.7%) (–19 mmol/mol [19 mmol/mol]) among insulin users. The factor explaining most of the variance in HbA1c change was baseline HbA1c level, which accounted for approximately half or more of the improvement in HbA1c levels in univariate analysis and around two-thirds in multivariate analysis.
Baseline insulin regimen had limited predictive power in the A1chieve study, both in insulin-naive people (r2 = 0.007) and prior insulin users (r2 = 0.004). This was quite unexpected: Intuitively, the more aggressive the insulin treatment, the greater may be the HbA1c response. This was what we observed in our analysis of 87 randomized controlled trials (RCTs), with 135 arms and 38,803 patients (2). The weighted variance (r2) using baseline HbA1c as independent variable was 0.485, a little lower than that reported by Home et al. (1). However, there was a significant variation among insulin regimens that ranged from r2 = 0.178 for the basal to r2 = 0.719 for the basal-bolus insulin regimen. The baseline HbA1c level of the entire A1chieve cohort was 9.5% (1.7%) (80 mmol/mol [19 mmol/mol]), somewhat higher of the 8.7% (median, interquartile range 8.4–9.1%) found in the 87 RCTs.
As the authors correctly stated, previous analyses between studies have noted that baseline HbA1c level is a strong predictor of response to glucose-lowering agents, including insulin. In the A1chieve study, it is stated that for each increase of 1.0%-units (11 mmol/mol) of baseline HbA1c level there was a 0.7–0.8%-units (8–9 mmol/mol) greater fall. This estimate was very similar to what we found in another analysis comprising 137 study arms and 39,100 patients with type 2 diabetes treated with insulin analogs. At baseline HbA1c 7.5–8%, the response to insulin (any analog, any regimen) was −0.61% (8 arms with 1,250 patients), increasing to −1.29% at baseline HbA1c 8.5–9% (55 arms with 21,350 patients) and to −2.29% at baseline HbA1c 9.5–10% (16 arms with 6,317 patients). The difference in HbA1c response was 0.7% between lower and middle and 1% between middle and higher baseline HbA1c levels. Ironically, the median duration of the studies included in our analysis (3) was 24 weeks (interquartile range 20–26 weeks), exactly the same as the A1chieve study. This was reassuring as predictors of HbA1c responses derived from meta-analyses of published RCTs are similar, at least in the present case, to those emerging in routine clinical care around the world. At the very least, a look at the baseline HbA1c level may be enough.
Article Information
Duality of Interest. K.E. and D.G. have received funding from Eli Lilly & Company and Sanofi for themselves or their institutions for research, educational, and advisory activities. No other potential conflicts of interest relevant to this article were reported.
