肝硬化合并肠系膜上静脉栓塞患者2年预后预测模型的构建与验证

目的 探讨肝硬化合并肠系膜上静脉栓塞(superior mesenteric venous thrombosis, SMVT)长期预后的影响因素，构建2年预后预测模型并进行预测效果评价。方法 采集2016年12月至2020年3月确诊的肝硬化合并SMVT患者254人的临床信息，以2019年3月30日为时间节点分为建模组(170例)和验证组(84例)。建模组依据随访2年时的结局分为生存组(148例)和死亡组(22例),运用多因素Logistic回归分析患者预后影响因素，建立模型并绘制列线图；采用受试者工作特征(receiver operating characteristic, ROC)曲线、校准曲线和Hosmer-Lemeshow拟合优度检验评价预测模型的准确性、区分度和校准度；决策曲线分析(decision curve analysis, DCA)评估预测模型的临床有效性。结果 年龄(HR=4.625,95%CI:1.493～14.322,P=0.008)、粒细胞/淋巴细胞比值(neutrophils/lymphocyte ratio, NLR)(HR=4.650,95%CI:1.34...

Establishment and validation of 2-year predictive scoring model for patients with liver cirrhosis complicated with superior mesenteric venous thrombosis

Objective To investigate the influencing factors of long-term prognosis of liver cirrhosis complicated with superior mesenteric venous thrombosis (SMVT). Methods The clinical information of 254 patients with liver cirrhosis complicated with SMVT diagnosed from December 2016 to March 2020 was collected. Taking March 30, 2019 as the time point, the patients were divided into modeling group (170 cases) and validation group (84 cases). The modeling group was divided into survival group (148 cases) and death group (22 cases) according to the outcome of follow-up for 2 years. The influencing factors of prognosis were analyzed by Logistic regression and the model was established. The receiver operating characteristic (ROC)curve,calibration curves, and Hosmer-Lemeshow test were used to validate and evaluate the discrimination and calibration of the model, while the decision curve analysis was used to evaluate its clinical validity. Results Age(HR=4.625,95% CI:1.493－14.322,P=0.008), neutrophils/lymphocyte ratio (NLR) (HR=4.650,95% CI:1.343－16.099,P=0.015) were independent prognostic factors in patients with liver cirrhosis complicated with SMVT. Calculation formula of prediction model is PI=－4.753+1.531×age+1.537×NLR. The area under the curve(AUC) of the predictive scoring model was 0.838(95%CI:0.751－0.925), the sensitivity was 86.4%, the specificity was 62.8%; The AUC of internal validation was 0.806, the sensitivity was 81.8%, and the specificity was 79.4%, Kappa=0.408.The Hosmer-Lemeshow test showed perfect fitting degree (modeling:P=0.494;validation:P=0.128). Decision curve analysis showed a risk threshold of 10% and demonstrated a clinically effective predictive model. Conclusion Age and NLR are independent factors affecting the 2-year prognosis of patients with liver cirrhosis complicated with SMVT. The model established in this study has certain application value in predicting the 2-year prognosis of such patients.