Computed tomography-based paravertebral muscle density predicts subsequent vertebral fracture risks independently of bone mineral density in postmenopausal women following percutaneous vertebral augmentation

Background The risk of subsequent vertebral fractures (SVF) after the primary vertebral fracture cannot be explained by lower bone mineral density (BMD) alone. Computed tomography (CT) measurements of paravertebral muscle density (PMD) are recognized radiographic markers used to predict physical function, fragile fractures. Aims This study aims to investigate the relationship between PMD and the risk of SVF in cohorts of postmenopausal women, and to determine if combining both PMD and BMD measures derived from CT can improve the accuracy of predicting SVF. Methods This study enrolled 305 postmenopausal women between the ages of 50 and 88 for 3 years of follow-up studies. Trabecular attenuation (Hounsfield units, HU) was measured at L1 level and muscle attenuation of paravertebral muscle at L3 level on preoperative lumbar CT scans to determine the L1 BMD and L3 PMD. Kaplan-Meier analysis was applied to evaluate SVF-free survival. The hazard ratios (HRs) of PMD for SVF events were estimated with the Cox proportional hazards model. The predictive values of L1 BMD and L3 PMD for SVF were quantified using the Receiver-Operating Characteristic (ROC) curve. Result During the 3 years of follow-up studies, 88 patients (28.9%) suffered an SVF. ROC curve analysis demonstrated that an L3 PMD threshold of 32 HU had a sensitivity of 89.8% and a specificity of 62% for the prediction of SVF. Kaplan-Meier analysis showed that L3 PMD <= 32 HU was significantly associated with lower SVF-free survival (p < 0.001; log-rank test). After adjusting for age, BMI, diabetes, postoperative osteoporosis treatment, handgrip strength, L1 BMD, multivariate analyses also indicated a persistent modest effect of L3 PMD on SVF-free survival. The area under the ROC curve of L3 PMD and L1 BMD, combined to predict the risk of SVF, was 0.790, which was significantly higher than the value for L1 BMD alone (0.735). L3 PMD and L1 BMD significantly improved the accuracy of SVF risk prediction compared with L1 BMD alone, which was confirmed by reclassification improvement measures. The inclusion of handgrip strength and postoperative osteoporosis treatment in the model further improved SVF prediction accuracy, and PMD remained significant in the model. Conclusion Decreased L3 PMD is an independent risk predictor of SVF. Combined CT-based L1 BMD and L3 PMD can significantly improve the accuracy of predicting the risk of SVF in postmenopausal women who have suffered prior osteoporotic vertebral fractures.