Insulin secretion rate during glucose stimuli: alternative analyses of C-peptide data

The ability to evaluate the pancreatic insulin secretion rate (ISR) is essential for a quantitative understanding of the glucose regulation system in man. Various approaches have been developed for evaluation of the ISR in vivo. The aim of this study was to compare input/output and compartmental models of C-peptide to reconstruct the ISR in response to both physiological and nonphysiological glucose stimuli in healthy humans. In particular we applied the nonparametric stochastic deconvolution and the C-peptide minimal model approaches to the graded up&down glucose infusion protocol, where glucose was infused at progressively increasing and then decreasing rates, and to the intravenous glucose tolerance test (IVGTT), where an impulse dose of glucose was administered. Our results show that the two models give virtually identical results when glucose and C-peptide (and thus ISR) profiles are smooth and regular, but when vigorous nonstationarities are present, like during the first 4 min of the IVGTT, the two ISR profiles are different (but not their areas under the curve). The C-peptide minimal model, albeit requiring, at variance with deconvolution, the knowledge of glucose data, has the advantage of providing quantitative indices of the beta-cell function, which is important in the parametric definition of different physiopathological states.