Regional body composition: cross-calibration of DXA scanners--QDR4500W and Discovery Wi

Differences exist in body composition assessed by dual-energy X-ray absorptiometers (DXAs) between devices produced by different manufacturers and different models from the same manufacturer. Cross-calibration is needed to allow body composition results to be compared in multicenter trials or when scanners are replaced. The aim was to determine reproducibility and extent of agreement between two fan-beam DXA scanners (QDR4500W, Discovery Wi) for body composition of regional sites. The sample was: 39 women 50.6 +/- 9.6 years old with BMI 26.8 +/- 5.5 kg/m(2), body fat 33 +/- 7%. Four whole body scans (two on each device) were performed over 3 weeks. Major variables were fat mass, nonosseous lean mass, and bone mineral content (BMC) for the truncal and appendicular regions. Extent of agreement was assessed using Bland and Altman plots. Both devices demonstrated good precision with mean test-retest differences close to zero for fat mass, nonosseous lean mass, and BMC of the truncal and appendicular regions. Evaluation of interdevice agreement revealed significant differences for truncal and appendicular BMC, nonosseous lean mass, and fat mass. The greatest interdevice difference was for truncal fat mass (0.69 +/- 0.60 kg). Differences in truncal and appendicular fat mass increased in magnitude at higher mean values. Furthermore, differences in truncal and appendicular fat mass were strongly related to BMI (R = -0.61, R = -0.55, respectively). In conclusion, in vivo cross-calibration is important to ensure comparability of regional body composition data between scanners, especially for truncal fat mass and for subjects with higher BMI.