Pediatric in vivo cross-calibration between the GE Lunar Prodigy and DPX-L bone densitometers.

Dual energy x-ray absorptiometry (DXA) machine cross-calibration is an important consideration when upgrading from old to new technology. In a recent cross-calibration study using adult subjects, close agreement between GE Lunar DPX-L and GE Lunar Prodigy scanners was reported. The aim of this work was to cross-calibrate the two machines for bone and body composition parameters for pediatrics from age 5 years onwards. One-hundred ten healthy volunteers aged 5-20 years had total body and lumbar spine densitometry performed on DPX-L and Prodigy densitometers. Cross-calibration was achieved using linear regression and Bland-Altman analysis. There was close agreement between the machines, with r2 ranging from 0.85 to 0.99 for bone and body composition parameters. Paired t-tests demonstrated significant differences between machines that were dependent on scan acquisition mode, with the greatest differences reported for the smallest children. At the lumbar spine, Prodigy bone mineral density (BMD) values were on average 1.6% higher compared with DPX-L. For the total body, there were no significant differences in BMD; however, there were significant differences in bone mineral content (BMC) and bone area. For small children, the Prodigy measured lower BMC (9.4%) and bone area (5.8%), whereas for larger children the Prodigy measured both higher BMC (3.1%) and bone area (3.0%). A similar contrasting pattern was also observed for the body composition parameters. Prodigy values for lean body mass were higher (3.0%) for small children and lower (0.5%) for larger children, while fat body mass was lower (16.4%) for small children and higher (2.0%) for large children. Cross-calibration coefficients ranged from 0.84 to 1.12 and were significantly different from 1 (p<0.0001) for BMC and bone area. Bland-Altman plots showed that within the same scan acquisition modes, the magnitude of the difference increased with body weight. The results from this study suggest that the differences between machines are mainly due to differences in bone detection algorithms and that they vary with body weight and scan mode. In general, for population studies the differences are not clinically significant. However, for individual children being measured longitudinally, cross-over scanning may be required.