New Vertebral Fractures after Percutaneous Vertebroplasty for Painful Osteoporotic Vertebral Compression Fractures: A Clustered Analysis and the Relevance of Intradiskal Cement Leakage

Purpose: To perform clustered analysis of fracture-free probabilities of intact nontreated vertebrae after percutaneous vertebroplasty (PVP) in painful long-standing osteoporotic vertebral compression fractures (OVCFs) to determine risk factors for new vertebral fractures and estimate fracture-free probabilities of multiple intact nontreated vertebrae given their patient-and vertebra-specific covariate status. Materials and Methods: Informed consent and institutional review board approval were obtained. A total of 115 patients who underwent PVP for 216 painful long-standing OVCFs were prospectively followed up to detect new OVCFs during the 1st postoperative year. A total of 1031 intact vertebrae were available for clustered analysis of fracture-free probabilities by using a Cox proportional hazard frailty model. A clustered analysis takes clustering or correlation of fracture-free survival probabilities of individual intact vertebrae within one patient into account to improve estimates of fracture-free probabilities and risk factors. Relevant patient-and and vertebra-specific covariates were included. Volumetric analysis of intradiskal cement leakage was performed by using a receiver operating characteristic curve (ROC). Results: Three-and 12-month vertebral fracture-free probability was 97.0% and 94.5%, respectively. Strong patient-level risk factors included low bone mineral density (hazard ratio [HR], 0.53 per unit increase), high spinal deformity index (HR, 2.23 per five units increase), and low fracture age (HR, 0.52 per 2 months increase). Strong vertebra-specific risk factors were thoracolumbar localization (HR, 2.33), vicinity to the treated level (adjacent level HR, 3.53), and presence of intradiskal cement leakage (HR, 8.21). Fracture-free probabilities of individual vertebrae were clustered within a patient (ie, not independent) (P = .009). The predicted 1-year fracture-free probability of an individual vertebra could be as high as 99.8% or as low as 19.9% based on absence or presence of risk factors, respectively. Larger intradiskal cement leakage volumes were associated with a higher likelihood of occurrence of new adjacent OVCFs (area under the ROC curve, 0.70). Conclusion: New vertebral fractures after PVP were clustered within patients and depended heavily on the presence or absence of both patient-and vertebra-specific risk factors. Intradiskal cement leakage was a pronounced augmentation-related risk factor, for which a volumetric association was found. (C) RSNA, 2012