Segment-specific prediction of vertebral collapse severity in osteoporotic fractures

Abstract

Background Osteoporotic vertebral compression fractures (OVCFs) predominantly cluster at the thoracolumbar junction (T11-L1). Recent studies suggest that both bone quality and muscle health significantly contribute to vertebral collapse. However, among numerous multidimensional imaging metrics, which one can most accurately predict the severity of vertebral collapse, and whether the precise load-sharing proportions between bone and muscle exhibit distinct segment-specific differences, remain rarely quantified in the clinical literature. Purpose To identify imaging metrics that best predict vertebral collapse severity and to quantify segment-specific contributions of bone and muscle quality at T11–L1. Methods This retrospective cohort study analyzed 223 postmenopausal patients with single-level OVCFs at T11–L1. The severity of vertebral collapse was quantified by the sagittal area compression ratio of the fractured vertebral body (SACR-FVB). Radiological parameters including bone mineral density (BMD), trabecular CT Hounsfield units (CT Hu), vertebral bone quality (VBQ), paravertebral muscle quality (PVMQ), and fat infiltration rate (FIR) were measured. LASSO regression selected predictors; segment-specific multivariate regression and LMG variance decomposition quantified proportional contributions; RCS modeled dose-response relationships. Results CT Hu and PVMQ were identified as significant independent predictors of SACR-FVB across all three evaluated segments (P < 0.05). LMG variance decomposition revealed that CT Hu consistently accounted for the largest proportion of the explained variance across all levels (68.2% at T11, 52.7% at T12, 56.3% at L1). However, the combined contribution of paraspinal muscle parameters (PVMQ and FIR) to the explained variance reached its maximum at the L1 segment (43.7%). Furthermore, RCS analyses confirmed that the structural vulnerability of the vertebral body manifests as a continuous gradient alongside local bone attenuation and progressive muscle deterioration. Conclusions CT Hu and PVMQ were identified as significant independent predictors of vertebral collapse severity across all thoracolumbar junction segments. Localized bone quality, measured by CT Hu, remained the primary structural determinant, while the combined contribution of paraspinal muscle parameters increased caudally. Continuous quantitative monitoring of bone and muscle imaging metrics may enable early identification of high-risk patients before vertebral collapse occurs. These findings underscore the importance of a segment-specific, multidimensional approach to individualized fracture risk assessment and intervention.