New long-acting insulin analogs: from clamp studies to clinical practice.

The NPH era started in 1946 and ended nearly 60 years later with the introduction to the market of the long-acting analogs, glargine in 2000 (1) and detemir in 2004 (2). More recently, a secondgeneration acylated long-acting insulin analog, degludec (3), has becomeavailable in some countries outside the U.S. In this issue of the journal, Becker et al. (4) report on pharmacokinetics (PK) and pharmacodynamics (PD) of a new long-acting insulin analog, glargine U300 (Gla-300). This is the same glargine molecule concentrated three times (300 units in 1 mL). It was approved by the U.S. Food and Drug Administration and received a positive opinion from the European Medicines Agency at the end of February 2015. Becker et al. (4) compared Gla-300 to the original glargine (Lantus, U-100; Gla-100) in a clamp experiment in subjects with type 1 diabetes (T1D). They studied the right population (absent endogenous insulin secretion) in which the observed PK/PD after subcutaneous (s.c.) injection of an insulin preparation are specifically attributable to the injected insulin. When clamp studies are instead run in subjects without diabetes or with type 2 diabetes (T2D), the PK/PD results are confounded by endogenous insulin secretion. Becker et al. (4) correctly studied the subjects after 1 week of treatment at “steady state.” However, they did use a fixed dose of basal insulin (0.4 units/kg/day glargine, either Gla-300 or Gla-100). This approach has the advantage of simplicity but the disadvantage of inducing a pharmacological overinsulinization with risk of hypoglycemia as the previously daily basal insulin dose used by the subjects with T1D was lower (mean 0.30 units/kg/day). Indeed, the majority of subjects needed a robust glucose infusion rate (GIR) during the 6 h prior to injection of glargine (time zero) to prevent hypoglycemia, and several subjects had baseline plasma insulin concentrations higher before than after glargine clamp dosing. These baseline conditions, partly due to the automated biostator for glucose clamping (5), make it more difficult to interpret PK and PD. However, the experiments of Becker at el. may still serve for interpretation of PK/PD data of Gla-300 versus Gla-100 as the baseline GIR and plasma insulin concentrationswere similarly elevated in both conditions. After s.c. injection, plasma insulin concentration was smoother in the first 12 h and more sustained in the second 12 h with Gla-300 versus Gla-100. The PD mirror effect was a smoother and longer glucodynamic effect with Gla-300. This is possibly explained by the fact that the three times more concentrated insulin Gla-300 results in a two-thirds lower s.c. spheric depot with a surface area reduced by 50% as compared with Gla-100, with consequent slower and more prolonged insulin absorption of Gla-300. In fact, in 2000, Jørgensen et al. (6) showed slower insulin absorption from s.c. tissue with more concentrated NPH insulin in a pig model. In addition to the averaged PK/PD, Becker et al. (4) provide us with raw data and clamp profiles on all subjects studied. This is important because it shows how large the interindividual differences in insulin absorption and action are. Of note, a measure of dispersion (SD or interquartile range) is still not given in the graph on the primary outcome, the GIR over time, whereas it is given for insulin levels over time. Nevertheless, the data package provided for insulin Gla-300 is much more comprehensive than those provided for the recent new long-acting analog insulin degludec (7) and the investigational insulin peglispro (8). The hypothesis that Gla-300 might be more beneficial than Gla-100 in basal insulin replacement in T1D and T2D by approaching nocturnal near-normoglycemia with less risk for hypoglycemia has been tested in the EDITION studies (9–16). In all EDITION studies, Gla-300 appears noninferior to Gla-100 in lowering A1C while conferring risk reduction of hypoglycemia primarily at night in T2D (9–12). In a study with 549 subjects with T1D, no convincing hypoglycemia difference was reported (13), although