Risk factors for refracture by different cement leakage after percutaneous vertebroplasty in patients with thoracolumbar compression fractures

To investigate the risk factors for cement leakage-related recurrent fractures in osteoporotic vertebral compression fractures and analyze their correlation with postoperative recurrent fractures in both operated and adjacent vertebrae following percutaneous vertebroplasty (PVP). We retrospectively screened 412 patients who underwent unilateral PVP at the Sixth Affiliated Hospital of Xinjiang Medical University between January 2019 and November 2024. Based on follow-up findings, patients were divided into two groups: (1) fracture group (n = 91) and (2) non-fracture group (n = 321). Univariate, LASSO regression, and multivariate logistic regression analyses were employed to identify risk factors for cement leakage after PVP.The dataset was randomly split into training and test sets (7:3 ratio), and a nomogram prediction model for cement leakage was constructed using multivariate logistic regression results. Model performance was evaluated using receiver operating characteristic (ROC) curves, Hosmer-Lemeshow calibration curves, and decision curve analysis. (i) Among the 412 included patients, significant differences were observed between the fracture and non-fracture groups in bone mineral density (BMD), preoperative Cobb angle, cement volume, vertebral cleft sign, fracture-to-surgery interval, and cement leakage (P < 0.05). (ii) LASSO regression identified six predictor variables for cement leakage, which were further analyzed using multivariate logistic regression. The results confirmed that BMD, preoperative Cobb angle, cement volume, vertebral cleft sign, fracture-to-surgery interval, and cement leakage were independent risk factors for recurrent fractures after PVP (P < 0.05). (iii) The ROC curve analysis demonstrated strong predictive performance, with an area under the curve of 0.839 (95% CI 0.780-0.897) for the training set and 0.846 (95% CI 0.759-0.932) for the test set. The Hosmer-Lemeshow calibration curves indicated good fit (training set, P = 0.062; validation set, P = 0.081). Decision curve analysis supported the clinical utility of the nomogram in predicting re-fracture risk. This study highlights multiple risk factors for recurrent fractures after PVP. A nomogram incorporating six key predictors was developed, offering a practical tool for assessing re-fracture risk in PVP patients.