Sarcopenia Versus Systemic Inflammation as Predictors of New Vertebral Fractures After Vertebroplasty or Kyphoplasty: A Retrospective Cohort Study

Background: Osteoporotic vertebral compression fractures (OVCFs) are among the most 11 common fragility fractures in the elderly. Although vertebroplasty and kyphoplasty provide effective pain relief, new vertebral fractures remain a significant concern postoperatively. Imaging parameters associated with sarcopenia and systemic inflammatory markers have been individually associated with fracture risk, but their combined predictive value in the postoperative period has not been adequately defined. Methods: This retrospective cohort study included 166 patients who underwent vertebroplasty or kyphoplasty for OVCFs with a follow-up period of at least 12 months. Cross-sectional area (CSA) and density (HU) of the Psoas muscle were measured at the L3 mid vertebral level on preoperative CT. Preoperative hematological indices (NLR, PLR, LMR, SII, lymphocyte count, hemoglobin, and MPV) were recorded. The primary outcome was the development of a new vertebral fracture. Group comparisons were performed using Mann–Whitney U tests with Benjamini–Hochberg correction. Logistic regression identified independent predictors. Internal validation was performed using bootstrap optimism correction (1000 iterations) and 10-fold cross-validation. Calibration was assessed using the Hosmer–Lemeshow test and calibration plots. Results: Forty-nine patients (29.5%) developed a new fracture. After multiple comparison correction, Psoas 25 HU (BH-adj p < 0.001, r_rb = −0.810), Psoas CSA (BH-adj p < 0.001, r_rb = −0.622), NLR (BH-adj p = 0.016), lymphocyte count (BH-adj p = 0.009), and hemoglobin (BH-adj p = 0.033) showed significant differences between groups. SII did not remain significant after multiple-comparison correction (BH-adjusted p = 0.092). In multivariate logistic regression, only Psoas CSA (OR = 0.403, 95% CI 0.230–0.708, p = 0.002) and Psoas HU (OR = 0.825, 95% CI 0.770–0.885, p < 0.001) remained independently significant. The parsimonious model, with adequate calibration (Hosmer–Lemeshow p = 0.524), achieved an optimism-adjusted AUC of 0.918 (10-fold CV AUC = 0.924). A Psoas HU threshold of 20.50 yielded 79.6% sensitivity and 94.9% specificity. Conclusions: CT-derived Psoas muscle mass and quality are strongly associated with new vertebral fractures after percutaneous vertebral augmentation procedures in this retrospective cohort and showed stronger independent predictive performance than systemic inflammatory markers. These readily accessible imaging biomarkers can aid in risk stratification, although the proposed threshold requires externally validation before clinical implementation.