Universal standardization for dual x-ray absorptiometry: patient and phantom cross-calibration results

The comparison of patient data among different dual x-ray absorptiometry (DXA) scanners is complicated because no universally accepted cross-calibration procedure or standard currently exists. This study was performed under the auspices of the International DXA Standardization Committee to establish appropriate cross-calibration parameters. Posteroanterior (PA) lumbar spine measurements of 100 women, ages 20-80 years (mean 52.6 +/- 16, range of BMD = 0.4-1.6 g/cm2) were obtained on a Norland XR26 Mark II, a Lunar DPX-L, and a Hologic QDR 2000 densitometer using standard procedures (pencil beam mode for all three scanners). Area, BMC, and BMD results from the different scanners were compared for all patients. In addition, the European spine phantom (ESP) and the European spine phantom prototype (ESP prototype), as well as standard phantoms from all three manufacturers, were evaluated on the three systems. To achieve universal scanner calibration, we used the intercept and slope of the patient's correlations and the value of the middle vertebra of the ESP as a reference point in a series of standardization formulas, and we have expressed the results as sBMD (mg/cm2). The correlations of the patients' spinal BMD values were excellent for each of the three scanner pairs. The average absolute difference in patient spinal BMD values (L2-4) between Hologic and Norland was 0.012 g/cm2 (1.3%); it was 0.113 g/cm2 (11.7%) between Hologic and Lunar and 0.118 g/cm2 (12.2%) between Norland and Lunar. The phantoms' regression lines approximated those of the patient regression lines, and the phantoms with only one measurement point were very close to the patients' regression lines. After applying the standardization formulas, the average absolute differences for the 100 patients were 28 mg/cm2 (2.7%) for Hologic/Norland, 23 mg/cm2 (2.2%) for Hologic/Lunar, and 29 mg/cm2 (2.8%) for Norland/Lunar. Average BMD results for the patients before correction were 0.972 mg/cm2 for Hologic, 1.100 g/cm2 for Lunar, and 0.969 g/cm2 for Norland. After correction, sBMD results for patients were 1045 mg/cm2 for Hologic, 1047 mg/cm2 for Lunar, and 1043 mg/cm2 for Norland. The standardization approach as performed in our study provided compatibility of DXA results obtained on different scanners