Visual assessment of vertebral deformity by X-ray absorptiometry: a highly predictive method to exclude vertebral...

The accurate identification of prevalent vertebral fractures is important in both the clinical and research setting as they are associated with increased risk of further fracture and irreversible clinical consequences. This study reports a direct comparison of prevalent vertebral deformity identification using X-ray absorptiometry (XA) scans, acquired on a dual-energy X-ray absorptiometry (DXA) machine, and conventional radiographs in a diverse group of 161 postmenopausal women, ranging from healthy subjects with normal bone mineral density (BMD) to osteoporotic subjects with multiple vertebral deformities. Deformities were identified by a trained operator by visual assessment of the XA scans (VXA) and semiquantitatively by an experienced radiologist on the conventional radiographs (XSQ). Subjects were recruited prospectively and were triaged according to their VXA results into normal, equivocal and definite deformity groups. VXA and XSQ demonstrated good agreement (96.3%, K = 0.79) in classifying vertebrae as normal or deformed in the 1978 of 2093 vertebrae deemed analyzable on both the XA scans and conventional radiographs. VXA showed good sensitivity (91.9%) in the identification of moderate/severe XSQ deformities and an excellent negative predictive value (98.0%) was produced when VXA was used to distinguish subjects without vertebral deformities from those with possible or definite deformities on a per subject basis. The majority of disagreement between the two methods resulted from different classification of mild wedge and endplate deformities and the poor visualization of upper thoracic vertebrae on the XA scans. Agreement improved, particularly on a per subject basis, when analysis was restricted to the vertebral levels from L4 to T7. Visual triage of XA scans by a trained operator would seem to be swift, convenient and cost-effective method, with excellent negative predictive value, to distinguish subjects with very low risk of vertebral deformities from those with possible deformities. These 'normal' subjects can then be excluded prior to performing conventional radiographs and further time-consuming and costly methods of vertebral deformity assessment such as XSQ by an experienced radiologist and/or quantitative morphometry. VXA may prove useful in the clinical evaluation of patients at risk of osteoporosis as an adjunct to BMD scans or in the selection of subjects for osteoporosis-related clinical trials.