Heritability of Body Fat Using DXA

Abstract

Dear Madam: Hsu et al. (1) recently described heritability estimates of total body fat mass using DXA, citing that this area has not been well studied using direct measures of body composition. We agreed with their sentiments 8 years ago when we published the first paper to estimate the genetic variance of both total body and central abdominal fat measured by DXA, in a population of 436 British female twins (2). Our estimate of h2 of 60%, using both intraclass correlation coefficients and maximum likelihood estimates, is very similar to that of Hsu et al. We went on to confirm our findings in an Australian population (3), again with DXA. We could recommend to Hsu and colleagues a simple technique for measuring central abdominal fat using DXA. This technique could be determined from body composition measures as described in our papers (2, 3). Knowledge of sophisticated use of DXA allows for detailed phenotyping of subjects, which is imperative in studies of metabolic syndrome, moving away from indirect and biased measures such as BMI. Hsu et al. also claim that few studies include covariates such as smoking, diet, and lifestyle. We would respectfully refer them to our large body of work in the genetic epidemiology of total and central obesity and metabolic syndrome. Our analyses have included covariates such as smoking, hormone replacement, physical activity, and diet; therefore, our estimates of genetic and environmental influence are as robust as epidemiology allows. Using the monozygotic twin analyses, we have been able to estimate direct environmental effects of physical activity (4, 5, 6), diet (5), (7), smoking (8), and alcohol intake (9) on a number of metabolic syndrome phenotypes, including total and central fat, insulin resistance, arterial distensibility, and C-reactive protein (10), (11). These studies, by use of the monozygotic twins who were discordant only for the environmental factor of interest, are unique in controlling for genetic and all other measured environmental factors. Hsu et al.'s work, for the first time in US populations, confirms our own heritability estimates for total fat. We would encourage them to use a more detailed phenotype using DXA and derive central abdominal fat measures from their DXA data, particularly since central obesity plays such a pivotal role in the metabolic syndrome and diabetes.