Published in:
01-02-2014 | Short Communication
HOMA and Matsuda indices of insulin sensitivity: poor correlation with minimal model-based estimates of insulin sensitivity in longitudinal settings
Authors:
A. H. Xiang, R. M. Watanabe, T. A. Buchanan
Published in:
Diabetologia
|
Issue 2/2014
Login to get access
Abstract
Aims/hypothesis
Little is known about the performance of surrogates in assessing changes in insulin sensitivity over time. This report compared updated HOMA of insulin sensitivity (HOMA2-%S) and the Matsuda index from OGTTs with minimal model-based estimates of insulin sensitivity (SI) from frequently sampled IVGTTs (FSIGTs) in longitudinal settings and cross-sectional settings.
Methods
Two longitudinal studies were used: one a natural observational study in which 338 individuals were followed for a median of 4 years; one a clinical treatment study in which 97 individuals received pioglitazone treatment and were followed for 1 year. Pairs of OGTTs and FSIGTs were performed at baseline and follow-up. Correlations were computed. Impact of measurement uncertainty was investigated through simulation studies.
Results
Correlations between HOMA2-%S and SI from baseline or follow-up data were in the range reported previously (0.61–0.69). By contrast, correlations for changes over time were only 0.35–0.39. The corresponding correlations between the Matsuda index and SI were 0.66–0.72 for cross-sectional data and 0.40–0.48 for longitudinal change. Correlations for changes were significantly lower than the cross-sectional correlations in both studies (p < 0.03). Simulation results demonstrated that the reduced correlations for change were not explained by error propagation, supporting a real limitation of surrogates to fully capture longitudinal changes in insulin sensitivity.
Conclusions/interpretation
HOMA and Matsuda indices derived from cross-sectional data should be used cautiously in assessing longitudinal changes in insulin sensitivity.