Evidence from candidate gene studies suggests that obesity may modify genetic susceptibility to type 2 diabetes and dyslipidaemia. On an aggregate level, gene–obesity interactions are expected to result in different heritability estimates at different obesity levels. However, this hypothesis has never been tested.

The present study included 2,180 British female twins. BMI was used as an index of general obesity. Outcome measures were insulin sensitivity (indexed by quantitative insulin-sensitivity check index [QUICKI]) and fasting plasma lipid profile. Structural equation modelling was used to test whether BMI interacted with latent genetic and environmental effects to impact on the outcome measures.

Genetic influences on triacylglycerol increased with BMI (p < 0.001) whereas the unique environmental influence on QUICKI decreased with BMI (p < 0.001), resulting in a higher heritability estimate for both measures at higher BMI levels. This was further illustrated by stratified analysis in twin pairs concordant for normal weight and twin pairs concordant for overweight. Heritability was 19 percentage points higher for triacylglycerol (p < 0.001) and 31 percentage points higher for QUICKI (p < 0.01) among twins concordant for overweight than among twins concordant for normal weight. BMI had no moderator effect on the latent genetic and environmental factors for total cholesterol and HDL-cholesterol.

Our results suggest that the expression of genes influencing triacylglycerol and insulin sensitivity can vary as a function of obesity status. The substantial increases in the genetic contribution to the total variance in insulin sensitivity and triacylglycerols at higher BMIs may prove extremely valuable in the search for candidate genes.