椎体后凸成形治疗骨质疏松性椎体压缩骨折后新发骨折危险因素预测模型的建立

A prediction model for new fracture risk factors after percutaneous kyphoplasty for osteoporotic vertebral compression fractures

背景:目前对于椎体后凸成形治疗后新发骨折的研究较多,但是如何将其风险直观地呈现以便于临床应用,这类研究尚少。目的:探究椎体后凸成形治疗骨质疏松性椎体压缩骨折后新发椎体骨折的危险因素,并建立预测术后再骨折发生的列线图。方法:选择2017年12月至2020年6月徐州医科大学附属医院收治的骨质疏松性椎体压缩骨折患者439例,男107例,女332例,年龄(71.38±9.73)岁,根据椎体后凸成形治疗后1年内有无新发骨折分为新发骨折组(n=57)和无新发骨折组(n=382)。采用单因素和多因素分析确定术后发生新骨折的危险因素。使用KNN法对缺失值进行多重插补,并用SMOTE方法对样本过采样均衡。将数据随机分为训练集(75%)和测试集(25%),根据多因素分析结果建立了预测再骨折发生风险的列线图,并使用受试者工作特征曲线(ROC)、校准曲线和决策曲线分析来评估其性能,制作网络计算器给临床医生带来更便捷的交互体验。结果与结论:(1)单因素分析显示,两组间骨密度T值、既往骨折史、脑血管疾病、受伤时间及术后椎体Cobb角比较差异有显著性意义(P<0.05);多因素分析显示,脑血管疾病[OR=35.742,95%CI(9.377,136.236),P=0.000]、骨折史[OR=11.546,95%CI(5.420,24.595),P=0.000]、骨水泥椎间隙渗漏[OR=2.817,95%CI(1.109,7.158),P=0.029]是与新发骨折呈正相关的独立危险因素,骨密度T值[OR=0.019,95%CI(0.005,0.076),P=0.000]、受伤时间[OR=0.981,95%CI(0.963,0.999),P=0.041]与术后新发骨折呈负相关;(2)在训练集和测试集中,列线图的曲线下面积值分别为0.927[95%CI(0.864,0.989)]和0.807[95%CI(0.707,0.906)],校准曲线显示预测和实际状态之间的一致性较高,决策曲线下面积分别为0.044和0.062;(3)结果表明,低骨密度、脑血管疾病、骨折史、骨水泥椎间隙渗漏及受伤时间较短是椎体后凸成形治疗后发生新骨折的危险因素。

BACKGROUND:There are many studies on new fractures after percutaneous kyphoplasty,but less is reported on how to visualize the risk for clinical application.OBJECTIVE:To investigate the risk factors for new ve rtebral compression fractures after percutaneous kyphoplasty for osteopo rotic vertebral compression fractures and to establish a nomogram to predict the occurrence of postoperative new fractures.METHODS:Totally 439 patients treated with percutaneous kyphoplasty for osteopo rotic vertebral compression fractures admitted at the Affiliated Hos pital of Xuzhou Medical University from December 2017 to June 2020 were enrolled,including 107 males and 332 females,aged(71.38±9.73) years.All patients were divided into a new fracture group and a non-fra cture group according to the presence of new fractures within 1 year after surgery.Univariate and multiva riate analyses were used to determine the risk factors for the occurrence of new fracture after surge ry.M ultiple interpolation of missing values was performed using the KNN method,and the samples were oversampled and equalized using the SMOTE method.The data we re randomly divided into a training set(75%) and a testing set(25%).A nomogram predicting the risk of new fracture occurrence was created based on the results of multivariate analysis,and its performance was evaluated using receiver operating chara cteristic curve,calibration curve,and decision curve analysis.A web calculator was created to facilitate a more convenient interactive experience for clinicians.RESULTS AND CONCLUSION:Univariate analysis showed significant differences(P<0.05) between the two groups in T value,fracture history,cerebrovascular disease,injury time,and postoperative Cobb angle.M ultiva riate analysis showed that cerebrovascular disease [odds ratio(OR)=35.742,95% confidence interval(CI)(9.377,136.236),P=0.000],history of fra cture [OR=11.546,95% CI(5.420,24.595),P=0.000],cemented interve rtebral leakage [OR=2.817,95% CI(1.109,7.158),P=0.029]were the independent risk factors positively associated with new fractures.T value [OR=0.019,95% CI(0.005,0.076),P=0.000],and time to injury [OR=0.981,95% CI(0.963,0.999),P=0.041] were negatively associated with new fractures postoperatively.The area under the curve values of the nomogram were 0.927 [95% CI(0.864-0.989)] and 0.807 [95% CI(0.707-0.906)] in the training and test sets,respectively,and the calibration curves showed a high agreement between predicted and actual status with an area under the decision curve of 0.044 and 0.062,respectively.To conclude,low bone mineral density,cerebrovascular disease,fracture history,cemented intervertebral leakage,and short duration of injury are risk factors for new fractures after percutaneous kyphoplasty.