Dual-energy X-ray absorptiometry measurements of fat and lean masses in subjects with eating disorders

OBJECTIVE: The main objective was to use a dual-energy X-ray absorptiometer (DXA) to examine the total-body and regional fat and lean composition of soft tissue in subjects with and without eating disorders initially and after weight change. It was necessary also to study the effects of differences of calibration of different models of DXA scanner. DESIGN: A total of 175 women with eating disorders, including anorexia nervosa (AN) and bulimia nervosa, and 43 age-matched controls were measured for soft-tissue composition with a pencil-beam Hologic QDR 1000W scanner and results converted to be equivalent to those from a fan-beam Hologic QDR 4500A, using previously determined crosscalibration factors. Some measurements were repeated at 6 and 12 months. RESULTS: The baseline body composition of the patients covered a continuous range of fat proportions. Implausibly low fat proportions in some of the AN subjects were corrected by conversion to 4500 equivalents. The relationship between total lean mass and fat mass could be fitted equally well by a linear or linear/log regression. The relationship between leg and trunk fat was best fitted by a polynomial regression. There were weight changes in either direction in some of the subjects. The fat proportion in the total changed mass was a mean of 55%, higher in the legs and lower in the trunk, but not different between weight gainers and losers or clinical groups. The proportion was dependent on the initial fat proportion. CONCLUSIONS: Relatively small differences in fat/lean calibration of DXA scanners may lead to anomalous results in very anorexic subjects and corrections are necessary in comparing results from different instruments. Concerns expressed about preferential trunk fat accumulation during weight recovery are not well founded. Previous claims of a relationship between fat proportion in regained weight and the amount of the weight gain are not justified.