Influence of region of interest and bone size on calcaneal BMD: implications for the accuracy of quantitative ultrasound assessments at the calcaneus

There is considerable technological diversity among quantitative ultrasound (QUS) devices used to assess osteoporosis. Because the distance between the transducer and the footplate remains constant, the location of the calcaneus measured will vary with foot size. This study was designed to quantify the variation in bone mineral density (BMD) between a manufacturer's region of interest (ROI_M), which is fixed relative to the footplate, and an anatomical region of interest (ROI_A), which is defined as 20% of calcaneal length. The effect of foot length and width on QUS variables measured using two Food and Drug Administration cleared QUS devices, the Sahara (Hologic) and the Achilles+ (Lunar) was assessed. 26 healthy subjects (12 male and 14 female), aged 22-54 years (35.6 +/- 10 years) and with foot lengths of 21.5 cm to 29.7 cm (25.1 +/- 2.3 cm) were recruited. QUS assessments were performed at the right calcaneus. In addition. a Hologic 4500 densitometer was used to measure the BMD of the calcaneus in the ROI_M and ROI_A. The sizes of the ROIs were approximated to the sizes of the transducers of the Sahara and Achilles+ devices. The results showed a significant difference in BMD between the two ROI locations for the Sahara device (BMD 0.642 +/- 0.135 g cm(-2) vs 0.616 +/- 0.114 g cm(-2), p = 0.014), but no significant difference was found in BMD between the two locations for the Achilles device (BMD 0.661 +/- 0.120 g cm(-2) vs 0.662 +/- 0.123 g cm(-2), p = 0.818). At the ROI_A, there was a significant difference in BMD between the two QUS devices (p < 0.001). The correlation between QUS variables and BMD was slightly higher for the ROI_M (r = 0.68-0.79, since this is site-matched) than the ROI_A (r = 0.59-0.70) for the Achilles device, while for the Sahara device the correlations were r = 0.35-0.40 and r = 0.51-0.54, respectively. The smaller ROI of the Sahara device resulted in more than 50% of the subjects having BMD differences of greater than 5% between the ROI_A and the ROI_M, compared with only 20% of the subjects on the Achilles device. ROIs containing cortical bone edge and other soft tissues were found in 58% of cases for the Achilles device and 46% of cases for the Sahara device. The greatest differences occurred in very small and very large feet. Calcaneal length correlated significantly with Sahara speed of sound (SOS), and heel width correlated significantly with Achilles SOS. Heel width also correlated significantly with Sahara broadband ultrasound attenuation (BUA) but not Achilles+ BUA. These results suggest that variation in ROI and bone size might affect the accuracy of QUS measurements, since the calcaneus is heterogeneous both in terms of its external geometry and its internal structure and density.