Comparison of in-vivo body composition using two Lunar dual-energy X-ray absorptiometers

Objective: To compare in-vivo composition analysis between two dual energy X-ray absorptiometers, a DPX and a DPX/L, from the same manufacturer (LUNAR), pre(Study A) and post(Study B) hardware changes on both absorptiometers. Design: Comparison of (1) quality assurance (QA) data: air-counts low (38 keV), air-counts high (70 keV), air-counts ratio, percent spillover, R-delrin: and (2) total body compartments: total body tissue (TBTSS), total body fat (TBF), percent total body fat (%TBF), total body lean (TBLEAN), total body bone mineral content (TBBMC) and total body bone mineral density (TBBMD), between the two absorptiometers. Setting: Centre for Bone and Body Composition Research, University of Leeds. Subjects: Study A, 14 normal subjects and Study B, a different cohort of 19 normal subjects, were scanned on both machines on the same day. Results: In Study A, large significant differences were observed in the QA parameters between the two machines. The DPX, air-counts low and air-counts high, being 25% and 22% lower than the DPX/L. The Bland-Altman method of analysis indicated that the DPX was significantly higher for TBTISS (0.3 kg), %TBF (2%) and TBF (1.4 kg) and correspondingly lower for TBLEAN (-1.0 kg). No significant difference was observed in TBBMC. After the hardware changes (Study B) a marked reduction in the differences in QA air-counts was observed. The DPX air-counts low was now 1% higher and air-counts high 8% lower than the DPX/L. The DPX had now only small significant negative differences for %TBF (-0.6%) and TBF (- 0.4 kg) and a small significant positive difference for TBLEAN (0.4 kg), compared to the DPX/L. TBBMC difference although slightly increased, was still non-significant. Conclusions: The closer agreement observed in the QA parameters after the hardware changes was associated with a reduction in the mean differences, 95% of the mean differences and limits of agreement of the comparison of body composition analysis from the Lunar machines using the Bland-Altman method. The study indicates that the QA limits set for bone mineral analysis may require more stringent limits for body composition.