Exclusion of trunk region reduces biological error but increases technical error of DXA lean soft tissue estimates from nonfasted assessments

Implementation of an overnight fast is a pervasive recommendation prior to body composition assessment, but this is not always feasible. Previous research has indicated that for dual-energy X-ray absorptiometry (DXA) scans, the trunk region may be particularly susceptible to biological error induced by food and fluid intake. This analysis explored the potential utility of excluding the trunk region from nonfasted DXA scans. Recreationally active adults were assessed by DXA after an overnight fast and again after consumption of standardized high-carbohydrate and low-carbohydrate diets. The effects of food consumption on total and appendicular lean soft tissue (LST; ALST) and ALST-derived skeletal muscle mass (SMM) were evaluated via analysis of variance, and metrics of reliability were calculated. In both conditions, the constant error of nonfasted assessments was slightly lower when the trunk was excluded (ALST and SMM: 0.7-1.2%; LST: 1.5%). However, in both conditions, the total error, standard error of the estimate, and limits of agreement were higher for ALST and SMM (total error: 2.4-3.0%; standard error of the estimate: 2.2-2.8%; and limits of agreement: 4.5-5.6%) than LST (total error: 2.1%; standard error of the estimate: 1.3-1.4%; and limits of agreement: 2.5-2.8%) when expressed relative to mean values. The added technical error due to demarcation of body regions for ALST and SMM appears to outweigh the removal of biological error due to exclusion of the trunk. Although elimination of the trunk region is theoretically appealing for nonfasted DXA assessments, it is apparently an inferior method compared with utilizing whole-body LST.