Early insulin response and insulin sensitivity are equally important as predictors of glucose tolerance after correction for measurement errors

Abstract 

Aims

: We estimated measurement error (ME) corrected effects of insulin sensitivity (M/I), from euglycaemic insulin clamp, and insulin secretion, measured as early insulin response (EIR) from oral glucose tolerance test (OGTT), on fasting plasma glucose, HbA1c and type 2 diabetes longitudinally and cross-sectional.

Methods

: In a population-based study (n=1128 men) 17 men made replicate measurements to estimate ME at age 71 years. Effect of 1 SD decrease of predictors M/I and EIR on longitudinal response variables fasting plasma glucose (FPG) and HbA1c at follow-ups up to 11 years, were estimated using uncorrected and ME-corrected (with the regression calibration method) regression models.

Results

: Uncorrected effect on FPG at age 77 years was larger for M/I than for EIR (effect difference 0.10mmol/l, 95% CI 0.00;0.21), while ME-corrected effects were similar (0.02mmol/l, 95% CI −0.13;0.15mmol/l). EIR had greater ME-corrected impact than M/I on HbA1c at age 82 years (−0.11%, −0.28; −0.01%).

Conclusions

: Due to higher ME effect of EIR on glycaemia is underestimated as compared with M/I. By correcting for ME valid estimates of relative contributions of insulin secretion and insulin sensitivity on glycaemia are obtained.

Keywords: Insulin secretion, EIR, Insulin sensitivity, Regression dilution bias