The effect of weight change on total body dual-energy X-ray absorptiometry: results from a clinical trial.

In the past decade dual-energy X-ray absorptiometry (DXA) scanning has assumed an important role in the evaluation of new treatments for osteoporosis. Although the spine and hip are the sites usually chosen for monitoring bone mineral density (BMD) changes, total body DXA is also of interest because of the comprehensive view it gives of the whole skeleton. However, recent studies have reported anomalies in total body DXA in subjects undergoing weight change, suggesting that the technique may not be valid in this circumstance. The present study evaluated total body DXA in a trial of cyclical etidronate therapy in which many subjects underwent significant weight change. The study population was 152 postmenopausal women who had spine, hip and total body DXA scans performed at baseline, 1 and 2 years. The total body scans were analyzed using two software options referred to as 'standard' and 'enhanced'. The following variables were studied: total body BMD, total body bone mineral content (BMC), and subregional BMD values for the following seven sites: lumbar spine, thoracic spine, pelvis, head, ribs, arms and legs. The percentage change from baseline was analyzed in a multivariate regression analysis to derive the treatment effect (defined as the difference in changes between the etidronate and placebo groups) and a coefficient that described the effect of weight change on the total body DXA variable. Mean weight change after 2 years was +1.1 kg (range -9.3 to +16.8 kg). Results for the weight change coefficient were significantly different from zero for five of nine total body variables using the standard analysis and seven of nine for the enhanced analysis with values (and standard errors) that varied from +0.67 (0.04)%/kg for standard total body BMC to -0.32 (0.11)%/kg for enhanced arm BMD. Results for the treatment effect at 2 years were significantly different from zero for total body BMD, total body BMC and for the lumbar spine, thoracic spine and pelvis BMD subregions, but were not significant for head, rib, arm or leg BMD. Findings for the standard and enhanced analyses agreed closely and the size of the treatment effect was related to the proportion of trabecular bone at the measurement site. We conclude that in a randomized study the effects of weight change can be corrected and total body DXA can give useful information about the response to treatment across the whole skeleton.