Assessment of circulating insulin using liquid chromatography-mass spectrometry during insulin glargine treatment in type 2 diabetes: Implications for estimating insulin sensitivity and beta-cell function

Aim: To determine the potential impact of the cross-reactivity of insulin glargine U-100 and its metabolites on insulin sensitivity and beta-cell measures in people with type 2 diabetes. Materials and Methods: Using liquid chromatography-mass spectrometry (LC-MS), we measured concentrations of endogenous insulin, glargine and its two metabolites (M1 and M2) in fasting and oral glucose tolerance test-stimulated plasma from 19 participants and fasting specimens from another 97 participants 12 months after randomization to receive the insulin glargine. The last dose of glargine was administered before 10:00 PM the night before testing. Insulin was also measured on these specimens using an immunoassay. We used fasting specimens to calculate insulin sensitivity (Homeostatic Model Assessment 2 [HOMA2]-S%; QUICKI index; PREDIM index) and beta-cell function (HOMA2-B%). Using specimens following glucose ingestion, we calculated insulin sensitivity (Matsuda ISI[comp] index) and beta-cell response (insulinogenic index [IGI], and total incremental insulin response [iAUC] insulin/glucose). Results: In plasma, glargine was metabolized to form the M1 and M2 metabolites that were quantifiable by LC-MS; however, the analogue and its metabolites cross-reacted by less than 100% in the insulin immunoassay. This incomplete cross-reactivity resulted in a systematic bias of fasting-based measures. By contrast, because M1 and M2 did not change following glucose ingestion, a bias was not observed for IGI and iAUC insulin/glucose. Conclusions: Despite glargine metabolites being detected in the insulin immunoassay, dynamic insulin responses can be used to assess beta-cell responsiveness. However, given the cross-reactivity of the glargine metabolites in the insulin immunoassay, fasting-based measures of insulin sensitivity and beta-cell function are biased.