Estimation of beta-cell sensitivity from intravenous glucose tolerance test C-peptide data. Knowledge of the kinetics avoids errors in modeling the secretion.

Parametric models of insulin secretion are used to measure indexes of beta-cell function from plasma C-peptide concentration during an intravenous glucose tolerance test (IVGTT). Since the models have been usually assessed against plasma C-peptide data, both secretory and kinetic parameters need to be simultaneously estimated. However, undesired compensations between the two sets of parameters may arise. In this study, in order to evaluate IVGTT insulin secretion models, we have analyzed IVGTT data from seven normal subjects for whom individual C-peptide kinetics were known from a separate experiment. Three different beta-cell models have been examined: the minimal model M1 (Diabetes 37:223-231, 1988); a variation of a published model, M2 (Math Biosci 27:319-332, 1975); and a new one, M3. A two-compartment model was used to describe C-peptide kinetics. The results suggest the inadequacy of M1 since kinetic parameter estimates were consistently biased versus the known individual values, and systematic errors were present in the prediction of C-peptide data when kinetic parameters were fixed to the known individual values. M2 performs better than M1 since it reproduces C-peptide data satisfactorily when the individually known description of the kinetics is adopted. M3 retains the second-phase description of M2 but improves the description of first-phase release. M3 is thus proposed to reconstruct the insulin secretion time course and to estimate parameters of first- and second-phase sensitivity to glucose. We also show the robustness of M3, i.e., standard values of C-peptide kinetic parameters can be used when individual values are not available without a loss of accuracy in the estimated secretion parameters. Finally, the shortcomings of using a simplified single-compartment description of C-peptide kinetics are discussed.