REASSESSMENT OF GLUCOSE EFFECTIVENESS AND INSULIN SENSITIVITY FROM MINIMAL MODEL ANALYSIS - A THEORETICAL EVALUATION OF THE SINGLE-COMPARTMENT GLUCOSE DISTRIBUTION ASSUMPTION

Minimal model analysis with the frequently sampled intravenous glucose tolerance test provides an effective way to measure two important met abolic parameters in vivo under non-steady-state conditions: glucose e ffectiveness (S-G) and insulin sensitivity (S-I). Two questions regard ing the validity of S-G and S-I have recently emerged. First, S-G from the minimal model is suspected to be overestimated. Second, the occur rence of S-I values indistinguishable from zero (''zero-S-I'') is not negligible in large clinical studies, and its physiological meaning is uncertain. In this study, we examined the significance of the assumed single-compartment glucose distribution embedded in the minimal model on the estimation of S-G and S-I. A more accurate two-compartment mod el was constructed by incorporating insulin action on hepatic glucose output and uptake into a previously validated construction. The two-co mpartment results were compared with the one-compartment minimal model results. It was shown that the one-compartment assumption contributes to a systematic deviation of S-G (slope = 0.54, y-intercept = 0.014 m in(-1) n=195 simulations). However, S-G from the minimal model was lin early correlated to S-G determined from the two-compartment model(r = 0.996). The one-compartment assumption also contributed to the occurre nce of zero S, values for insulin-resistant subjects. A similar linear relationship was found between S-I estimated by both the minimal mode l and the two-compartment model (slope = 0.58, y-intercept = -0.57 x 1 0(-4) min(-1) per pU/ml, r = 0.998). In conclusion, S-G and S-I from t he minimal model are not necessarily equivalent to values emanating fr om the more accurate two-compartment model. However, the very high cor relation between one-and two-compartment results suggests that the min imal model-derived S-G and S-I are dependable indexes of in vivo gluco se effectiveness and insulin sensitivity. Minimal model analysis' adva ntages of simplicity, minimal invasiveness, reasonable reflection of n on-steady-state glucose kinetics, and cost-effectiveness could in many cases outweigh the structural bias introduced by the model simplifica tion.