Comparison of insulin sensitivity assessment indices with euglycemic-hyperinsulinemic clamp data after a dietary and exercise intervention in older adults

Multiple indices to assess insulin sensitivity calculated from mathematical equations based on fasting blood parameters or oral glucose tolerance data have been developed. Although these indices have frequently been validated using euglycemic-hyperinsulinemic clamp data, the utility of each equation in measuring change in insulin sensitivity over time remains uncertain. We examined change in insulin sensitivity in response to a 12-week diet and exercise intervention in 31 older men and women with impaired glucose tolerance using a euglycemic-hyperinsulinemic clamp and 10 commonly used insulin sensitivity equations. Mean glucose disposal as calculated from clamp data was significantly higher after the intervention compared with baseline (5.92 ± 0.38 vs 5.18 ± 0.30 mg · kg fat free mass−1 · min−1, P = .013). In contrast, none of the examined indices indicated a significant change in insulin sensitivity over time (all P > .3). A limits of agreement approach to compare insulin sensitivity calculated from each equation with the measure of glucose disposal from the clamp indicated overall imperfect agreement between measures (agreement limits ranged from ±2.48 to ±4.23 mg · kg fat free mass−1 · min−1) despite significant bivariate correlations between indices and clamp data. The wide variability in the 95% prediction limits of agreement among equations suggests that these equations vary substantially from a euglycemic-hyperinsulinemic clamp in their ability to assess insulin sensitivity. Despite the observed limited agreement using this statistical approach, changes in several calculated indices were significantly correlated with changes in clamp data, suggesting that these indices may have some utility in tracking improvements in insulin sensitivity. Further research is necessary to examine agreement between indices and clamp data in larger, more heterogeneous populations and in response to other interventions where the magnitude of change in insulin sensitivity may be larger.