Estimation of prehepatic insulin secretion: comparison between standardized C-peptide and insulin kinetic models

Our aim was to compare traditional C-peptide-based method and insulin-based method with standardized kinetic parameters in the estimation of prehepatic insulin secretion rate (ISR). One-hundred thirty-four subjects with varying degrees of glucose tolerance received an insulin-modified intravenous glucose tolerance test and a standard oral glucose tolerance test with measurement of plasma insulin and C-peptide. From the intravenous glucose tolerance test, we determined insulin kinetics parameters and selected standardized kinetic parameters based on mean values in a selected subgroup. We computed ISR from insulin concentration during the oral glucose tolerance test using these parameters and compared ISR with the standard C-peptide deconvolution approach. We then performed the same comparison in an independent data set (231 subjects). In the first data set, total ISRs from insulin and C-peptide were highly correlated (R(2) = 0.75, P < .0001), although on average different (103 +/- 6 vs 108 +/- 3 nmol, P < .001). Good correlation was also found in the second data set (R(2) = 0.54, P < .0001). The insulin method somewhat overestimated total ISR (85 +/- 5 vs 67 +/- 3 nmol, P = .002), in part because of differences in insulin assay. Similar results were obtained for fasting ISR. Despite the modest bias, the insulin and C-peptide methods were consistent in predicting differences between groups (eg, obese vs nonobese) and relationships with other physiological variables (eg, body mass index, insulin resistance). The insulin method estimated first-phase ISR peak similarly to the C-peptide method and better than the simple use of insulin concentration. The insulin-based ISR method compares favorably with the C-peptide approach. The method will be particularly useful in data sets lacking C-peptide to assess beta-cell function through models requiring prehepatic secretion.