肝硬化并发门静脉血栓形成患者列线图预测模型分析及外部验证

Nomogram model in predicting portal venous thrombosis in patients with liver cirrhosis

目的探讨建立肝硬化患者门静脉血栓形成(PVT)列线图预测模型及外部验证情况。方法2010年1月~2021年10月我院收治的肝硬化伴或不伴PVT患者,收集一般资料、血清学指标、超声等影像学资料。对建模组数据应用Lasso回归和多因素Logistic回归模型筛选出肝硬化并发PVT的独立危险因素并建立列线图预测模型,应用验证组数据对该模型进行外部验证。结果肝硬化并发PVT患者独立危险因素为脾切除史、门静脉主干内径和脾静脉直径(P<0.05);在验证人群,对列线图预测模型进行外部验证,其ROC曲线下面积为0.751(95%CI:0.674~0.827),预测模型区分度良好,经Hosmer-Lemeshow检验,P值为0.170,提示预测模型的校准度较高。结论我们开发了一种易于使用的肝硬化并发PVT形成的列线图预测模型,可以协助临床医护人员筛选高PVT风险人群,制定及时、个性化的诊治方案,希望最终改善患者预后。

Objective The aim of this study was to establish a nomogram for predicting portal venous thrombosis(PVT)in patients with liver cirrhosis(LC).Method 1000 patients with LC were enrolled in our hospital between January 2010 and October 2021,and the clinical materials,including general data,serological indicators,ultrasonography and CT imaging were collected.A nomogram model was established by Lasso and multivariate Logistic regression analysis.Results The risk factours included splenectomy,widened portal vain diameters and splenic vain diameters in patients with liver cirrhosis complicated by PVT(P<0.05);in the external validation population,the AUC of the nomogram prediction model was 0.751(95%CI:0.6740-0.827)with a good discrimination(P=0.170,by Hosmer-Lemeshow test),suggesting a high reliability of the model.Conclusion We establish an easy-to-use nomogram to predict the PVT formation in patients with cirrhosis.This nomogram might help clinicians screen patients at high risk of PVT,make in timely and individualized clinical decisions to improve outcomes of them.