Complex relation between increasing fat mass and decreasing high density lipoprotein cholesterol levels: evidence from a population-based study of premenopausal women.

High density lipoprotein (HDL) cholesterol levels can be used to predict cardiovascular disease risk in women. To better understand variability in HDL cholesterol levels, the authors examined the relation with three domains (body size and type, sex hormone status, and carbohydrate metabolism) in a cross-sectional population-based 1993-1994 study of 402 premenopausal women from Tecumseh, Michigan. They found that these domains explained 27% of the total variation in HDL cholesterol levels; waist-to-hip ratio was the term that explained the highest proportion of variability (6% after fat mass, sex hormone binding globulin, and insulin levels were added to the model). In analyses restricted to women whose body mass index was > or = 32 kg/m2, which constituted 19% of this population, neither body mass index nor fat mass was a significant predictor of variability in HDL cholesterol levels. Significant variables were insulin levels, waist-to-hip ratio, and smoking. This finding suggests that there is a saturation of the relation between increasing fat mass and lower HDL cholesterol levels, as evidenced by the lack of a relation between the two among the heaviest women. Meanwhile, among the heaviest women, increasing insulin levels and a higher waist-to-hip ratio remained predictors of low levels of HDL cholesterol.