Comparative study of the performances of X-ray and gadolinium 153 bone densitometers at the level of the spine,femoral neck and femoral shaft

We investigated the precision of gadolinium 153 dual photon absorptiometry (DPA) and quantitative digital radiography (QDR) bone densitometers by determining in vitro and in vivo coefficients of variation (CV) of bone mineral density (BMD). In vitro, the long-term CV of spine phantom BMD measured weekly for 40 weeks was 1.2% and 0.7% for DPA and QDR, respectively. Simulating soft-tissue thickness with water, the CV of 6 repeat measurements of spine phantom at depths from 0 to 27 cm in 1 cm steps (a total of 168 measurements) increased from 0.1 % at 0 cm of water to 2.5% at 27 cm for DPA, and from 0.2% at 0 cm to 1.4% at 27 cm for QDR; mean CV of the 28 series (0–27 cm) was higher for DPA (1.2%±0.8%, mean±SD) than for QDR (0.7%±0.6%;P<0.001). With the hip phantom, femoral neck BMD was determined, and the CV was also dependent on water thickness; mean CV of the 20 series (0–10 cm) was 2.1%±1.2% for DPA and 1.3%±0.9% for QDR (not significant). In vivo, at the spine level, with DPA, mean CV of BMD measured 6 times after repositioning in 6 healthy volunteers was 3.8%±1.9% and 2.1%±0.7% with¹⁵³Gd activity of 0.46Ci and 1 Ci, respectively (BMD range: 0.796 1.247 g/cm², no significant difference between the two groups). Both values were significantly higher (P < 0.05) than mean CV with QDR: 1.0%±0.5% (12 subjects, same conditions; BMD range: 0.811–1.124 g/cm², no significant difference with the two previous groups). At the femoral neck and shaft levels, the mean CV observed with QDR tended to be lower as compared with DPA (not significant). At the three sites, BMD values obtained with DPA and QDR in 62 patients were highly correlated. In conclusion, our results indicate that the higher precision obtained with QDR is particularly significant at the lumbar spine level, but large biological variations in soft tissue thickness can still influence the degree of precision of BMD measurement.