Spinal Instability Neoplastic Score Does Not Predict Future Vertebral Compression Fractures

Introduction: Understanding risk for metastatic vertebral compression fracture (VCF) is critical for all physicians and specialties involved in cancer patient care. While Spinal Instability Neoplastic Score (SINS) has been the standard for assessing instantaneous instability, its utility for predicting future VCF remains a topic of debate. We investigated both outcome correlations and predictive value of SINS in one of the largest series of patients/vertebra.

Methods: A multi-institutional, adult cohort of patients with tumor- infiltrated vertebral bodies, spanning T1 through L5, were included. Binary outcome of fracture/progression or no fracture/progression was recorded during 12-month follow-up. SINS score (binned as 0-6, 7-12, >12), subcomponents, demographic, and clinical data were collected. Univariate (Chi-squared) and multivariate (logistic regression) analysis was performed to evaluate attribute correlation with the binary outcome within 12 months. Subsequently, predictive modeling was performed using select SINS components with lasso regression in a random forest classifier. Models were evaluated using 10-fold cross-validation

Results: One-hundred-sixty-three patients with 973 vertebrae (120 fractures/progression) were included. The average time to outcome was 4.5 months. While steroid use, prior radiation, histology, and SINS scores were significant on univariate analysis (p < 0.05), only histology and SINS scores were significant on multivariate analysis (p < 0.05). Of the SINS components, feature selection found all subcomponents of “vertebral body collapse,” spinal alignment, posterolateral involvement, lytic tumor, mechanical pain, and junctional location were the most important features to predict outcome, along with SINS binned score. Classifier validation resulted AUC-ROC 0.669 (95% CI: 0.608-0.731), sensitivity of 0.203 (95% CI: 0.119-0.287), and specificity of 0.957 (95% CI: 0.941-0.973).

Conclusion : While SINS scores and subcomponents show correlation with future VCF/progression, our results show it does not achieve clinical utility performance for prediction. Future VCF prediction models should integrate a wider range of both clinical and radiographic biomarkers than those accounted for in SINS.