Biomarkers of insulin resistance and their performance as predictors of treatment response in adults with risk factors for type 2 diabetes.

Insulin Resistance (IR) is a component of the pathogenesis of type 2 diabetes mellitus (T2DM), and risk factor for cardiovascular and neurodegenerative diseases. Amino acid and lipid metabolomic IR surrogate assays have utility for predicting future T2DM risk. Whether these biomarkers accurately track IR changes following exercise therapy has not been established. In the present study we evaluated the ability of two distinct insulin assays to detect exercise-training induced change in IR. We then evaluated metabolomic and clinical phenotypes previously reported to predict IR status or T2DM risk, in the pre-intervention samples from three independent cohorts; META-PREDICT (MP, n=179), STRRIDE-AT/RT (S2, n=116) and STRRIDE-PD (S-PD, n=149). Multiple strategies, including Bayesian projective prediction were used to assess the utility of validated biomarkers for fasting insulin and HOMA2-IR and to predict change in fasting insulin following exercise or lifestyle interventions. Both insulin assays demonstrated high performance against international standard insulin (R²>0.99), while the correlation between values generated in fasting clinical samples was far less congruent (R²=0.65). Only the high-sensitivity ELISA assay detected the influence of exercise status on fasting insulin. While combined clinical/lipoprotein/amino-acid models were statistically significant in linear modelling they were unable to diagnose baseline IR-status in older adults. Furthermore, models built using the same metabolomic and clinical parameters were unable to predict treatment responses in examined cohorts. Thus, only direct measure of insulin is currently suitable for monitoring IR longitudinally, and choice of insulin assay is critical for reliable detection of key environmental influences on IR.