Use and Abuse of HOMA Modeling

To evaluate whether the homeostasis model assessment (HOMA) is a reliable surrogate measure of in vivo insulin sensitivity in humans.

The oral glucose tolerance test (OGTT) has often been used to evaluate apparent insulin release and insulin resistance in various clinical settings. However, because insulin sensitivity and insulin release are interdependent, to what extent they can be predicted from an OGTT is unclear. We studied insulin sensitivity using the euglycemic-hyperinsulinemic clamp and insulin release using the hyperglycemic clamp in 104 nondiabetic volunteers who had also undergone an OGTT. Demographic parameters (BMI, waist-to-hip ratio, age) and plasma glucose and insulin values from the OGTT were subjected to multiple linear regression to predict the metabolic clearance rate (MCR) of glucose, the insulin sensitivity index (ISI), and first-phase (1st PH) and second-phase (2nd PH) insulin release as measured with the respective clamps. The equations predicting MCR and ISI contained BMI, insulin (120 min), and glucose (90 min) and were highly correlated with the measured MCR (r = 0.80, P < 0.00005) and ISI (r = 0.79, P < 0.00005). The equations predicting 1st PH and 2nd PH contained insulin (0 and 30 min) and glucose (30 min) and were also highly correlated with the measured 1st PH (r = 0.78, P < 0.00005) and 2nd PH (r = 0.79, P < 0.00005). The parameters predicted by our equations correlated better with the measured parameters than homeostasis model assessment for secretion and resistance, the delta30-min insulin/delta30-min glucose ratio for secretion and insulin (120 min) for insulin resistance taken from the OGTT. We thus conclude that predicting insulin sensitivity and insulin release with reasonable accuracy from simple demographic parameters and values obtained during an OGTT is possible. The derived equations should be used in various clinical settings in which the use of clamps or the minimal model would be impractical.