Diagnostic sensitivity of peripheral quantitative computed tomography measurements at ultradistal and proximal radius in postmenopausal women.

ABSTRACT

Peripheral quantitative computed tomography (pQCT) is a bone densitometry technique that is able to provide real volumetric bone density values not only of the total but also of trabecular and cortical bone separately. Normal reference curves were constructed with cross-sectional data obtained in 275 postmenopausal women (50-85 years), measured at 4% of the ulnar length (ultradistal region), and data for total, trabecular, and cortical bone density were obtained. In these postmenopausal subjects, continuously significant (p < 0.0001) age-dependent declines in bone density of 1.14%, 1.1%, and 0.57% for total, trabecular, and cortical bone, respectively, were observed while similar declines of 0.9%, 0.9%, and 0.4% per year since menopause, respectively, were found. The estimated mechanical stability index also showed linear dependencies with decreases of 0.84%/year and 0.6%/year since menopause (p < 0.0001). A more proximal acquisition at 15% of the ulnar length, an almost pure cortical region, resulted in linear declines of 0.41%/year and 0.27%/year (p < 0.0001) for the cortical bone and the mechanical stability index with significant changes of -0.27% and -0.23% per year, respectively, since menopause. Covariance analysis showed similar age dependencies of the different bone indices obtained in both regions of interest except for the stability index. A significant size adaptation of the bone with age was also observed, which was seen in the relationships of the trabecular and cortical bone areas to age and to bone density. Diagnostic sensitivity of all parameters for established osteoporosis was assessed by receiver operating characteristic (ROC) curves, comparing 99 patients with at least one fracture to the reference population. The area under these curves was highest in the ultradistal pure trabecular density of the radius (75%), followed by stability index (72%) and the area of cortical bone (65%) of the proximal site. No distinguishing power was seen for the cortical bone density values obtained in either the ultradistal (51%) or proximal radius (52%).