Aims: With the increasing availability of new drugs for the treatment of insulin resistance in patients with Type 2 diabetes, simple methods for their identification is an important challenge. The aim of our study was to compute a new algorithm for estimating insulin resistance in a routine clinical setting.
Methods: Clinical data and blood samples were collected from 4265 Type 2 diabetic patients from 149 clinical sites. A clinical algorithm to estimate insulin resistance was developed by stepwise multiple regression analysis. The new generated score was compared with the HOMAIR-score, calculated from fasting insulin and glucose levels measured in a central laboratory. In a subgroup of 48 patients, the score was verified against a frequently sampled intravenous glucose tolerance test with subsequent modified minimal model analysis according to Bergman.
Results: Multiple regression analysis revealed fasting blood glucose, BMI, triglycerides and HDL as the most powerful predictors of insulin resistance which were used for further computation of the IRIS II score. A significant overall correlation was found between the HOMAIR-score and the new clinical IRIS II score (r = 0.42; P < 0.0001). Compared with HOMAIR, the new score revealed a specificity of 0.95, a sensitivity of 0.34 and a positive predictive value of 0.95. This was in good agreement with the subset analysis of the intravenous glucose tolerance test, where a sensitivity of 0.37 and a specificity of 0.85 of the IRIS II score was calculated. Patients with insulin resistance according to the IRIS II score revealed an increased odds ratio for overall vascular complications (1.28; 1.11-1.46; P < 0.001).
Conclusions: The new IRIS II score can identify insulin resistance in Type 2 diabetic patients with high predictive value and high specificity.