Body composition measured by dual-energy X-ray absorptiometry half-body scans in obese adults

Abstract

Dual-energy X-ray absorptiometry (DXA) has become a common measurement of human body composition. However, obese subjects have been understudied largely due to weight and scan area restrictions. Newer DXA instruments allow for heavier subjects to be supported by the DXA scanner, but the imaging area is still smaller than the body size of some obese subjects. In this study, we determined the validity of an automated half-scan methodology by comparing to the standard whole-body scans in a cohort of obese volunteers. Fifty-two subjects whose BMI>30 kg/m2 completed whole-body iDXA (GE Lunar) scans. The resulting scans were analyzed in three ways: the standard whole-body scan, total body estimated from the left side, and from the right side. Fat mass, nonbone lean mass, bone mineral content (BMC), and percent fat derived from each half scan were compared to the whole-body scans. Total fat mass, nonbone lean mass, or percent fat was comparable for the whole-body scans, left, and right side scans (>97% within individuals and >99.9% for the group). The BMC estimate using the right side scan was slightly but statistically higher than the whole-body BMC (approximately 30 g or 1%, P<0.001), while the left side scan BMC estimate was lower than the whole-body BMC by the same magnitude. No significant magnitude bias was found for any of the composition variables. We conclude that the new iDXA half-body analysis in obese subjects appears to be closely comparable to whole-body analysis for fat mass, nonbone lean mass, and percent fat.

Figure 1

Example of iDXA total-body composition image and a representation of the portions of the total-body image used to generate the right side and left side images.

Figure 2

Correlation plots between right side, left side, and whole-body scans (n = 52). Dashed lines represent line of identity.

Figure 3

Bland–Altman plots comparing the right side, left side, and whole-body scans. No significant magnitude bias was detected for any of the composition parameters. All r² < 0.033 (not significant, n = 52).