颅内动脉瘤破裂介入术后并发脑疝的风险预测列线图模型构建

Nomogram model construction for predicting risk of complicated cerebral hernia after interventional surgery for ruptured intracranial aneurysms

目的探讨颅内动脉瘤破裂介入术后并发脑疝的风险因素,并构建预测列线图模型。方法回顾性分析398例颅内动脉瘤破裂患者的临床资料,根据术后脑疝并发情况将其划分为并发组与未并发组,比较两组患者临床资料,采用单因素及多因素Logistic回归分析法(逐步向前法)筛选颅内动脉瘤破裂介入术后并发脑疝的危险因素,并构建风险列线图模型;采用Bootstrap法对列线图模型进行内部验证,绘制受试者工作特征(ROC)曲线评估列线图模型的预测效能。结果398例患者中术后发生脑疝48例,脑疝并发率为12.06%。并发组高血压史、入院时Hunt-Hess分级Ⅳ级~Ⅴ级、入院时CT Fisher分级Ⅲ级~Ⅳ级、入院时世界神经外科联合会(WFNS)分级Ⅲ级~Ⅴ级、手术时间>2 h、脑积水、脑水肿、宽颈动脉瘤及动脉瘤再次破裂占比均高于未并发组(P<0.05或0.01);多因素Logistic回归分析显示,高血压史、入院时Hunt-Hess分级Ⅳ级~Ⅴ级、入院时CT Fisher分级Ⅲ级~Ⅳ级、入院时WFNS分级Ⅲ级~Ⅴ级、手术时间>2 h、宽颈动脉瘤、脑积水、脑水肿、动脉瘤再次破裂均是颅内动脉瘤破裂介入术后并发脑疝的危险因素(P<0.05或0.01);基于上述多因素分析结果构建的列线图模型,一致性指数为0.841,ROC曲线显示该列线图模型预测颅内动脉瘤破裂介入术后并发脑疝风险的曲线下面积为0.836(P<0.01),灵敏度为83.33%,特异度为75.93%。结论颅内动脉瘤破裂介入术后发生脑疝的风险因素众多,以此构建的列线图模型预测效能高,有助于提高临床识别脑疝,为后续治疗奠定基础。

Objective To explore the risk factors of complicated cerebral hernia after interventional surgery for ruptured intracranial aneurysms(RIA),and construct prediction nomogram model.Methods Clinical data of 398 RIA patients were retrospectively analyzed,ones assigned to complication and non-complication group according to whether postoperative complicated cerebral hernia,clinical data compared between two groups,risk factors of postoperative complicated cerebral hernia screened using univariate and multivariate logistic regression analysis,and risk nomogram model was constructed;internal verification of nomogram model was carried out using Bootstrap method and predictive efficacy of nomogram model assessed using receiver operating characteristic(ROC)curve.Results Among 398 patients,48 cases had cerebral hernia after operation and the complication rate of cerebral hernia was 12.06%.The proportions of such indexes were higher in complication than non-complication group as hypertension history,Hunt-Hess grade Ⅳ-Ⅴ,CT Fisher grade Ⅲ-Ⅳ,and World Federation of Neurosurgery(WFNS)grade Ⅲ-Ⅴ at admission,operation time>2 hours,hydrocephalus,cerebral edema,and re-ruptured wide-necked aneurysm and aneurysm(P<0.05 or 0.01);multivariate logistic regression analysis showed that above-mentioned indexes were risk factors of complicated cerebral hernia after interventional surgery for the RIA(P<0.05 or 0.01);consistency index of nomogram model based on analysis results of above-mentioned multivariate analysis was 0.841,area under the ROC 0.836(P<0.01),sensitivity 83.33%,and specificity 75.93%.Conclusion Risk factors of complicated cerebral hernia after interventional surgery for the RIA are numerous,predictive efficacy of nomogram model constructed by this method is higher,and which is conduce to the clinical improvement of identifing cerebral hernia to lay the foundation of subsequent treatment.