Estimation of Total-Body and Regional Soft Tissue Composition From DXA Bone Densitometry of the Lumbar Spine and Hip

The percent fat in soft tissues adjacent to the lumbar spine and proximal femur, which are required parameters in the calculation of bone mineral by conventional dual-energy X-ray absorptiometry (DXA) of the spine and hip, were analyzed for their relationship with the total-body soft tissue-scanning derivatives. The cohort (N = 149), consisting of stable actively treated male human immunodeficiency virus HIV-positive patients, was split in half to obtain prediction equations with one half to be validated by the other half. Prediction equations for the dependent variables total-body fat, total-body lean mass, trunk fat, total arm + leg fat, and leg fat were derived by step-down multiple regression. A Bland-Altman comparison of the predicted and observed values showed that the limits of agreement were too large to be clinically helpful. The correlations of the ratio of adjacent spine/hip fat with ratios of trunk/arm + leg fat and trunk/leg fat, markers of peripheral lipoatrophy in HIV, were 0.725 and 0.780, respectively. The 3 ratios were compared with the clinical diagnosis of the presence or absence of peripheral lipoatrophy by receiver operating characteristic analysis. The area under the curve was 0.720 for adjacent spine/hip fat ratio and 0.655 and 0.699 for trunk/arm + leg fat and trunk/leg fat, respectively; they were not significantly different. In conclusion, for male HIV-positive patients, the difference between predicted values and actual values rendered limits of agreement that were too wide to be clinically acceptable. The ratio of percent fat in the lumbar spine region to percent fat in the proximal femur region reflected the presence of peripheral lipoatrophy as effectively as the trunk/peripheral fat ratio that was derived from the total-body scan.