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 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.