Replacing DXA scanners: cross-calibration with phantoms may be misleading.

The high precision and stable calibration of dual X-ray absorptiometry (DXA) has led to its widespread use in clinical trials of new therapies for osteoporosis. Daily scanning of phantoms provides a check on the continuity of the bone mineral density (BMD) calibration and allows for correction of minor changes. However, because clinical trials may last up to 5 years, it is likely that the scanner will need replacing during the course of a study. We report the upgrade of a Hologic QDR-2000 to a QDR-2000plus. The new scanner was cross-calibrated with the old according to the manufacturer's instructions using the Hologic spine phantom. The accuracy of cross-calibration was checked by in vivo scans of patients in clinical trials and also using the European spine phantom (ESP). Daily QC scans with the Hologic phantom over the 6-week period of the in vivo study showed that the new scanner agreed within 0.2% with the old. However, in vivo scans at 10 sites in the spine, hip, total body, and forearm showed significant mismatch between the two systems with five sites differing by more than 2%. Results for the ESP phantom lay closer to the in vivo spine data than the Hologic phantom. The mismatch revealed was larger than anticipated and emphasized the importance of performing adequate in vivo cross-calibration studies whenever DXA systems are replaced. Typically, scans of 20-30 subjects are required at each scan site to achieve an accuracy of 1%. Such studies require careful planning and make significant demands on space, patient cooperation, scanning time, scientific resources, and, not least, the implementation of the findings.