Vertebral Fracture Prediction From MRI-based Vertebral Bone Quality Scores in Postmenopausal Women: A Longitudinal Cohort Study.

Abstract

A longitudinal cohort study. The study aims to assess the lumbar MRI-based vertebral bone quality (VBQ) score and dual energy x-ray absorptiometry (DXA) T score in a longitudinal cohort of postmenopausal women and to evaluate the performance of these scores in predicting VF over 5 years of follow-up. VBQ scores can be used for osteoporosis screening and identification of vertebral fractures (VFs). However, due to limitations in prior cross-sectional studies, definitive conclusions about the causal associations between the VBQ score and incident VF have not been drawn. Postmenopausal women with baseline DXA and lumbar MRI data were enrolled. Follow-up assessments were conducted over 5 years after the baseline examination; both anteroposterior- and lateral-view thoracolumbar spine radiographs were captured to evaluate incident VF. At baseline, correlation analyses between the VBQ score and osteoporosis or preexisting VF were performed. A Cox regression model for multivariate logistic regression was used to analyze the risk factors for incident VF. The hazard ratio (HR) was calculated. One hundred thirty-seven postmenopausal women were included. The VBQ score was moderately correlated with lumbar bone mineral density at baseline (r=0.4, P<0.001). Patients with incident VF had significantly lower T scores (-3.4 vs. -1.87, P<0.001) and higher VBQ scores (4.62 vs. 4.03, P<0.001) at baseline than those without incident VF. After adjusting for age, weight, creatinine clearance, and alkaline phosphatase levels, the HRs for the VBQ score and T score were 1.93 and 0.71, respectively. After adjusting for the T score, an increased VBQ score was independently associated with an increased risk of incident VF, with an HR of 1.87. The VBQ score was moderately correlated with the DXA T score and was a predictor of incident VF risk independent of bone mineral density in postmenopausal women.