基于Lasso-Logistic回归构建胫骨平台骨折术后发生创伤性关节炎的风险预警模型  

Construction of a risk warning model for post-operative traumatic arthritis of tibial plateau fractures based on Lasso-Logistic regression

作　　者：仓挺松 曹峰 吴骏 任鹏鹏 陈阳 CANG Tingsong;CAO Feng;WU Jun;REN Pengpeng;CHEN Yang(Department 1 of Bone Injury,Suzhou Hospital of Integrated Traditional Chinese and Western Medicine,Suzhou 215101,China)

机构地区：[1]苏州市中西医结合医院骨伤一科,江苏苏州215101

出　　处：《安徽医学》2025年第4期431-438,共8页Anhui Medical Journal

基　　金：苏州市科技发展计划基金项目(编号:SYSD2020235)。

摘　　要：目的分析胫骨平台骨折术后发生创伤性关节炎(TA)的危险因素,并构建其风险预警模型。方法选取2016年3月至2021年5月苏州市中西医结合医院行胫骨平台骨折手术治疗的286例患者为研究对象,根据术后是否发生TA分为TA组(n=73)、非TA组(n=213)。收集两组患者临床资料,经Lasso-Logistic回归分析患者术后发生TA的影响因素,根据影响因素构建患者术后发生TA的Nomogram风险预警模型,并对构建的Nomogram风险预警模型进行验证。结果两组患者年龄、骨质疏松、Schatzker分型、半月板损伤、损伤方式、受伤至手术时间、体力劳动、术后Rasmussen评分、白介素(IL-6)、基质金属蛋白酶-13(MMP-13)、肿瘤坏死因子-α(TNF-α)、骨形态发生蛋白(BMP)-2、BMP-9水平比较,差异有统计学意义(P<0.05);将单因素分析结果中13个有统计学意义的因素纳入Lasso回归,筛选出最佳9个预测变量为骨质疏松、Schatzker分型、损伤方式、术后Rasmussen评分、血清MMP-13、BMP-2、BMP-9、IL-6、TNF-α水平;骨质疏松、Schatzker分型、损伤方式、术后Rasmussen评分、血清MMP-13、IL-6、TNF-α水平是患者术后发生TA的危险因素,血清BMP-2、BMP-9水平是患者术后发生TA的保护因素(P<0.05);根据logistic回归获得的相关因素构建患者术后发生TA的Nomogram风险预警模型,并进行内部验证,决策曲线和受试者工作特征(ROC)曲线显示该模型具有较好的临床净收益和预测效能,校准曲线显示模型与实际观测结果一致性较好,临床影响曲线(CIC)结果显示在阈值概率范围内TA高风险的人数与实际情况具有较高的符合率。结论胫骨平台骨折术后发生TA的因素包括骨质疏松、Schatzker分型、损伤方式、术后Rasmussen评分、血清MMP-13、BMP-2、BMP-9、IL-6、TNF-α水平,基于以上因素构建的风险预警模型具有良好预测效能和临床适用度。Objective To analyze the factors contributing to the development of traumatic arthritis(TA)after surgery for tibial plateau fractures and construct a risk warning model.Methods A total of 286 patients who underwent surgical treatment for tibial plateau fracture in our hospital from March 2016 to May 2021 were selected and divided into TA group(n=73)and non-TA group(n=213)according to whether TA occurred after surgery.The clinical data of the two groups of patients were collected.The factors affecting the occurrence of TA in patients after surgery were analyzed using Lasso-Logistic regression.Based on these factors,a Nomogram risk warning model for the occurrence of TA in patients after surgery was constructed and validated.Results There were significant differences in age,osteoporosis,Schatzker classification,meniscus injury,injury mode,time from injury to surgery,physical labor,postoperative Rasmussen score,interleukin(IL-6),matrix metalloproteinase-13(MMP-13),tumor necrosis factor-α(TNF-α),bone morphogenetic protein(BMP)-2,and BMP-9 levels between the two groups of patients(P<0.05).Thirteen significant factors from the single factor were included in the Lasso regression,and the best nine predictive variables were selected as osteoporosis,Schatzker classification,injury method,postoperative Rasmussen score,serum MMP-13,BMP-2,BMP-9,IL-6,and TNF-αlevels.Osteoporosis,Schatzker classification,injury mode,postoperative Rasmussen score,serum levels of MMP-13,IL-6,and TNF-αwere risk factors for the development of TA in patients after surgery,while serum levels of BMP-2 and BMP-9 were protective factors for the development of TA in patients after surgery(P<0.05).A Nomogram risk warning model for patients with postoperative TA was constructed based on the relevant factors obtained from logistic regression,and internal validation was performed.The decision curve and receiver operating characteristic(ROC)curve showed that the model had good clinical net benefit and predictive performance.The calibration curve showed th

关 键 词：胫骨平台骨折术 创伤性关节炎 因素 风险 预警模型 

分 类 号：R687.3[医药卫生—骨科学]