Novel Body Composition Approaches with DXA: Comparison with Classical Total Body Scans

Abstract

Traditionally, total body (TB) DXA scans to assess body composition include the entire body, however novel abbreviated scan acquisition methods are available in new software. For example, total body less head (TBLH) is recommended given the head contains lean mass that cannot be altered by intervention. Recently, to reduce scan time and radiation exposure, a neck to knee (N-K) acquisition option is available. Finally, scans limited to the lower extremity might be useful to evaluate regional mass change after surgery/injury and in sarcopenia assessment. The purpose of this study was to evaluate comparability of measured and estimated body composition results using TBLH, N-K and lower extremity (LE; pelvis to toes) acquisition techniques. We hypothesized that body composition data with the 3 novel methods would not differ from the traditional TB method. Each subject had 4 scans: TB, TBLH, N-K and LE. A subset of 30 had repeat N-K and LE scans with repositioning between. A Lunar iDXA (GE, Madison, WI) with enCORE v18.0 was used to acquire and analyze scans. TB scans were acquired per ISCD guidance. TBLH and N-K scans were acquired using automated software features that limited the anatomy scanned; N-K estimates non-scanned leg mass. LE scans were started at L3 and auto stopped past the toes. Automated analysis, with manual correction, was used for all but LE scans which were analyzed manually by compressing all upper body ROIs above the pelvis then placing lower body ROIs at the top of pelvis, bisecting the femur neck and separating the legs. Pearson Correlation and Bias plots were used to compare regional lean, fat and BMC measurements from the 3 novel scans to TB. Precision determined by the ISCD Precision Calculator was compared to TB historical control by F-test. The study sample included 82 subjects, 41F/41M mean (SD) age 51.0 (18.4) years and BMI 25.5 (3.7) kg/m2. Mean lean, fat and bone mass were very similar among all approaches with some statistical, but likely not clinical, differences (Tables 1&3). Composition from all regions with the 3 novel methods was highly correlated with TB measurement, r=0.99-1.0, p< 0.001 with mean bias of 14-237g lean, 7-121g fat and 0.3-13.3g BMC (Table 2). N-K and LE precision was excellent ranging from 0.45-2.31 %CV (Table 3). These 3 novel approaches to either measure or estimate regional body composition are consistent with traditional TB measurement and may be considered as substitutes when appropriate.