Surrogate indexes calculated from oral glucose tolerance tests (OGTTs) effectively predict the development of type 2 diabetes (T2D), suggests a US study. On the other hand, measures based on fasting insulin are more useful than those that do not measure insulin.
“[S]urrogate measures of insulin resistance can be useful for the assessment of insulin sensitivity and the prediction of diabetes risk in situations where more intensive methods, such as the hyperinsulinaemic–euglycaemic clamp (HEC), are not practical,” the researchers said. “Most indexes evaluated in our analysis have some utility for the prediction of future T2D.”
In this community-based longitudinal study of indigenous Americans, T2D risk was assessed among 2,260 participants followed for up to 14.5 years. Of these, 509 had T2D. The researchers estimated correlation coefficients (r) of all indexes with sensitivity (M) measured by HEC in 286 individuals. They also calculated indexes obtained from OGTTs using fasting and 2-h glucose and insulin measures.
The predictive performance of indexes was examined using hazard ratio per standard deviation and the area under the receiver operating curve (AUC).
Indexes based on OGTTs had the best performance: Matsuda showed the highest r with M (0.691), while Gutt and Cederholm exhibited the highest AUCs (0.728 and 0.728; p<0.05 compared with all other indexes). [J Clin Endoc Metab 2026;111:2251-2261]
Among indexes using fasting insulin levels only, the quantitative insulin sensitivity check index and the Homeostatic Model for Insulin Resistance performed best with equal r (0.644) and AUC (0.701).
Among those that did not require insulin measurement, the metabolic score for insulin resistance (METS-IR) and single-point insulin sensitivity index showed the highest r (0.597 and 0.595), while the METS-IR yielded the highest AUC (0.688).
Matsuda index
“In this study, HECs were performed on a subset of individuals, which allowed us to compare surrogate indexes to the gold-standard measure of insulin resistance/sensitivity, M,” the researchers said. “The Matsuda index had the strongest correlation with M compared to all others, and its correlation with M was statistically stronger than that for any of the other indexes.”
These findings support those of a meta-analysis by Otten and colleagues, which revealed that Matsuda was most associated with insulin sensitivity measures from the HEC. However, the statistical significance among correlations was not evaluated. [Diabetologia 2014;57:1781‐1788]
Similarly, a study by Lorenzo and colleagues reported a significantly stronger association between Matsuda’s index and M compared with other indexes. [J Clin Endocrinol Metab 2010;95:5082‐5090]
“Although the correlations with M of most of the indexes examined were highly statistically significant, the magnitude of the correlations was modest to moderate. The sensitivity of these indexes for detecting insulin resistance, defined by the lowest quartile of M, was also modest to moderate,” according to the researchers.
“Thus, there is some degree of misclassification, relative to the clamp measure, when using the indexes, even those based on the OGTT, to diagnose insulin resistance. As in many epidemiologic studies, the present analyses were based on a single measurement of the indexes, and the use of repeated measures may improve clinical utility,” they said.