Glucose effectiveness and insulin sensitivity from the minimal models: consequences of undermodeling assessed by Monte Carlo simulation.

The unlabeled (cold) minimal model (MM) and the labeled (hot) minimal model (HMM) are a powerful tool to investigate in vivo metabolism from a standard intravenous glucose tolerance test (IVGTT) or hot IVGTT (HIVGTT). They allow to estimate metabolic indexes of the glucose-insulin system, namely glucose effectiveness (GE) and insulin sensitivity (IS) (of uptake and production those of MM, and of uptake only those of HMM). Here, the consequences of the single-compartment glucose kinetics approximation used in the MM's are investigated via Monte Carlo simulation, using a physiologic reference model (RM) of the system. RM allows to generate noisy synthetic plasma concentrations of glucose, tracer glucose, and insulin during IVGTT and HIVGTT, which are then analyzed with MM and HMM. The MM and HMM GE and IS are then compared with the RM ones. Results of 400 runs show that: 1) correlation of MM GE with the RM index is weak; 2) MM IS is well correlated with the RM index, but severely underestimates it; 3) HMM clearance rate is correlated with RM clearance; and 4) HMM IS is well correlated and only slightly overestimates the RM index. These results demonstrate that GE of MM is most affected by the single-compartment approximation and the indexes of HMM are more robust than those of MM.