Development of a nomogram model for prediction of new adjacent vertebral compression fractures after vertebroplasty

Abstract

Abstract Background Vertebroplasty is the main minimally invasive operation for osteoporotic vertebral compression fracture (OVCF), which has the advantages of rapid pain relief and shorter recovery time. However, new adjacent vertebral compression fracture (AVCF) occurs frequently after vertebroplasty. The purpose of this study was to investigate the risk factors of AVCF and establish a clinical prediction model. Methods We retrospectively collected the clinical data of patients who underwent vertebroplasty in our hospital from June 2018 to December 2019. The patients were divided into a non-refracture group (289 cases) and a refracture group (43 cases) according to the occurrence of AVCF. The independent predictive factors for postoperative new AVCF were determined by univariate analysis, least absolute shrinkage and selection operator (LASSO) logistic regression, and multivariable logistic regression analysis. A nomogram clinical prediction model was established based on relevant risk factors, and the receiver operating characteristic curve (ROC), calibration curve, and decision curve analysis (DCA) were used to evaluate the prediction effect and clinical value of the model. After internal validation, patients who underwent vertebroplasty in our hospital from January 2020 to December 2020, including a non-refracture group (156 cases) and a refracture group (21 cases), were included as the validation cohort to evaluate the prediction model again. Results Three independent risk factors of low bone mass density (BMD), leakage of bone cement and “O” shaped distribution of bone cement were screened out by LASSO regression and logistic regression analysis. The area under the curve (AUC) of the model in the training cohort and the validation cohort was 0.848 (95%CI: 0.786–0.909) and 0.867 (95%CI: 0.796–0.939), respectively, showing good predictive ability. The calibration curves showed the correlation between prediction and actual status. The DCA showed that the prediction model was clinically useful within the whole threshold range. Conclusion Low BMD, leakage of bone cement and “O” shaped distribution of bone cement are independent risk factors for AVCF after vertebroplasty. The nomogram prediction model has good predictive ability and clinical benefit.