Multiple methods are used to determine insulin sensitivity. Most commonly used in children are euglycemic-hyperinsulinemic clamp and frequently sampled intravenous glucose tolerance (FSIVGTT) with minimal modeling (MinMod). The parameters S(G) and S(I) of MinMod are related to insulin sensitivity and glucose effectiveness, respectively, but inappropriate modeling of glucose kinetics causes inaccuracies. Glucose tracer use may mitigate such inaccuracies, allowing use of multiple modeling approaches, including a 2-compartment model (2CMM). This study was designed to compare the 1-compartment model (1CMM) and 2CMM in a pediatric population. Twenty-three children were studied 4 times using FSIVGTT with [6,6] D(2) glucose. Glucose effectiveness and insulin sensitivity were calculated by 1CMM (S*(G1) and S*(I1)) and 2CMM (S*(G2) and S*(I2)). Indices were reliably estimated in 86 of 87 tests for 1CMM, but only in 49 for 2CMM. S*(G1) overestimated S*(G2), but they were positively related. S*(I1) and S*(12) were not different and were positively related. This suggests that inadequate modeling by the 1CMM has a smaller impact on glucose tolerance indices in children than in adults. Comparison with classical MinMod gave results analogous to those in adults (MinMod S(G) was larger than S*(G1), MinMod S(I) was smaller than S*(I1)). These results demonstrate significant differences in glucose effectiveness, but not insulin sensitivity, as measured by 1CMM and 2CMM in early adolescents. Thus, when insulin sensitivity is the primary interest, the 1CMM is more robust because its parameters are more reliably estimated than those of 2CMM and is the logical choice in pediatric population-based studies.
Copyright 2002, Elsevier Science (USA). All rights reserved.