Genetic and baseline metabolic factors for incident diabetes and HbA(1c) at follow-up: the healthy twin study.

Abstract

We investigated baseline anthropometric/metabolic traits predicting incident diabetes, genetic/environmental relationships between these traits and HbA1c at follow-up and the contribution of genetics, covariates and environments to variance in HbA(1c) at follow-up and incident diabetes. Nondiabetic twins (n = 869) and their family members (n = 949) were followed over 3.7 ± 1.4 years (44.3 ± 12.8 years of age); baseline anthropometric/metabolic traits were measured. Fasting plasma glucose and HbA(1c) were measured at follow-up. Incident diabetes was defined as HbA(1c) ≥6.5% or fasting plasma glucose ≥7 mmol/L. Age-adjusted incident diabetes was 4.9% in men and 4.1% in women. Odd ratio for incident diabetes was 2.34-2.40, 1.25-1.28, 1.22-1.27 and 1.89 per standard deviation of baseline fasting plasma glucose, white blood cell (WBC), triglycerides and waist circumference, respectively, in multivariate generalized estimating equation models (p < 0.05). Age-adjusted and sex-adjusted heritability was 0.85 for diabetes and 0.72 for HbA(1c). In bivariate analyses adjusted for age, sex and body mass index at baseline, HbA1c at follow-up showed significant genetic and environmental correlations with baseline glucose (0.44, 0.17), significant genetic correlation with baseline waist circumference (0.16) and triglycerides (0.30) and significant environmental correlation with baseline WBC (0.09). Variance in HbA1c at follow-up and incident diabetes was explained by genetics (33% and 28%, respectively), covariates (36% and 48%, respectively), shared environments (7% and 0%, respectively) and errors (24% and 24%, respectively). High values for baseline fasting plasma glucose, WBC, triglycerides and waist circumference are independent risk factors for incident diabetes. While genetic influences strongly contribute to variance in HbA1c at follow-up and incident diabetes, these risk factors significantly contribute to the remaining variance.