Evaluation of vertebral fracture assessment by dual X-ray absorptiometry in a multicenter setting

Summary  The utility of vertebral fracture assessment (VFA) by DXA to detect prevalent vertebral fracture in a multicenter setting was investigated by comparison to conventional radiography. While limited by lower image quality, overall performance of VFA was good but had a tendency to miss mild prevalent fractures. Introduction  In osteoporosis clinical trials standardized spine radiographs are used to detect vertebral fractures as a study endpoint. Lateral spine imaging with dual X-ray absorptiometry (DXA) scanners, known as vertebral fracture assessment (VFA) by DXA, presents a potential alternative to conventional radiography with lower radiation dose and greater patient convenience. Methods  We investigated in a multicenter setting the ability of VFA to detect fractures in comparison with conventional radiography. The study examined 203 postmenopausal women who had imaging of the spine by DXA and radiography. Three radiologists experienced in vertebral fracture assessment independently read the VFA scans and radiographs using the Genant semiquantitative method on two occasions. Conclusions  Analyzing the data from all readable vertebrae, the kappa statistic, sensitivity, and specificity ranged from 0.64–0.77, 0.65–0.84, and 0.97–0.98, respectively. Considering only moderate and severe fractures improved the kappa statistic (0.80–0.91) and sensitivity (0.70–0.86). While image quality of VFA is inferior to radiography, the detection of vertebral fractures using visual scoring is feasible. However, VFA underperformed due to unreadable vertebrae and reduced sensitivity for mild fractures. Nevertheless, VFA correctly identified most moderate and severe vertebral fractures. Despite this limitation, VFA by DXA provides an important tool for clinical research.