[Risk factors and construction of a nomogram model for cirrhotic portal vein thrombosis combined with esophagogastric variceal bleeding].

Objective: To investigate the risk factors and construct a nomogram model for predicting the occurrence of cirrhotic portal vein thrombosis in patients combined with esophagogastric variceal bleeding (EVB). Methods: Clinical data on 416 cirrhotic PVT cases was collected from the First Hospital of Lanzhou University between January 2016 and January 2022. A total of 385 cases were included after excluding 31 cases for retrospective analysis. They were divided into an esophagogastric variceal bleeding group and a non-esophagogastric variceal bleeding group based on the clinical diagnosis. The esophagogastric variceal group was then further divided into an EVB group and a non-bleeding group. All patients underwent gastroscopy, serology, and imaging examinations. The risk factors of PVT combined with EVB were identified by univariate analysis using SPSS 26. The prediction model of cirrhotic PVT in patients combined with EVB was constructed by R 4.0.4. The prediction efficiency and clinical benefits of the model were evaluated by the C-index, area under the receiver operating characteristic curve, calibration plots, and decision curve. The measurement data were examined by a t-test or Mann-Whitney U test. The counting data were tested using the χ(2) test or the Fisher exact probability method. Results: There were statistically significant differences in the etiology, Child-Pugh grade,erythrocyte count, hematocrit, globulin, and serum lipids between the esophageal and non-esophageal varices groups (P < 0.05). There were statistically significant differences in etiology, erythrocyte count, hemoglobin, hematocrit, neutrophil percentage, total protein, globulin, albumin/globulin, urea, high-density lipoprotein cholesterol, calcium, and neutrophil lymphocyte ratio (NLR) between the EVB and non-bleeding groups (P < 0.05). Multivariate logistic regression analysis showed that etiology (OR = 3.287, 95% CI: 1.497 ~ 7.214), hematocrit (OR = 0.897, 95% CI: 0.853 ~ 0.943), and high-density lipoprotein cholesterol (OR = 0.229, 95% CI: 0.071 ~ 0.737) were independent risk factors for cirrhotic PVT patients combined with EVB. The constructed normogram model predicted the probability of bleeding in patients. The nomogram model had shown good consistency and differentiation (AUC = 0.820, 95% CI: 0.707 ~ 0.843), as verified by 10-fold cross-validation (C-index = 0.799) and the Hosmer-Lemeshow goodness of fit test (P = 0.915). The calibration plot and the decision curve suggested that the prediction model had good stability and clinical practicability. Conclusion: The risk factors for EVB occurrence include etiology, erythrocyte, hemoglobin, hematocrit, percentage of neutrophils, total protein, globulin, albumin/globulin, urea, high-density lipoprotein cholesterol, calcium, and NLR in patients with cirrhotic liver. The constructed prediction model has good predictive value, and it can provide a reference for medical personnel to screen patients with high bleeding risk for targeted treatment.目的： 探讨肝硬化患者门静脉血栓（PVT）合并食管胃底静脉曲张破裂出血（EVB）的危险因素，构建预测肝硬化PVT患者发生EVB的列线图模型。 方法： 收集兰州大学第一医院2016年1月至2022年1月的416例肝硬化PVT患者临床资料，排除31例后，共纳入385例进行回顾性分析。根据临床诊断分为食管胃底静脉曲张组与非食管胃底静脉曲张组；食管胃底静脉曲张组再分为EVB组与非出血组。所有患者行胃镜、血清学及影像学检查，通过SPSS 26进行单因素分析，得出PVT合并EVB的危险因素；通过R 4.0.4建立肝硬化患者PVT合并EVB的预测模型，并采用一致性指数（C-index）、受试者操作特征曲线下面积、校准图、决策曲线评估模型的预测效能及临床效益。计量资料采用t检验或Mann-Whitney U检验；计数资料采用χ(2)检验或Fisher精确概率法检验。 结果： 食管胃底静脉曲张组与非食管胃底静脉曲张组中患者的病因、Child-Pugh分级、红细胞计数、红细胞压积、球蛋白、血脂指标的差异具有统计学意义（P值均< 0.05）。EVB组与非出血组中患者的病因、红细胞计数、血红蛋白、红细胞压积、中性粒细胞百分比、总蛋白、球蛋白、白蛋白/球蛋白比值、尿素、高密度脂蛋白胆固醇、钙、中性粒细胞淋巴细胞比值（NLR）的差异具有统计学意义（P值均< 0.05）。多因素logistic回归分析显示病因（乙型病毒性肝炎与其他的OR = 3.287，95% CI：1.497~7.214）、红细胞压积（OR = 0.897，95% CI：0.853~0.943）、高密度脂蛋白胆固醇（OR = 0.229，95% CI：0.071~0.737）是肝硬化PVT患者EVB的独立危险因素，并基于该模型构建列线图，用于预测患者的出血概率。列线图表现出良好的一致性和区分度（曲线下面积为0.820，95% CI：0.707~0.843），并经10折交叉验证进行内部验证（C-index = 0.799）和Hosmer-Lemeshow拟合优度检验（P = 0.915），校准曲线提示预测模型具有良好稳定性，决策曲线提示具有临床实用性。 结论： 肝硬化PVT患者发生EVB的危险因素有病因、红细胞计数、血红蛋白、红细胞压积、中性粒细胞百分比、总蛋白、球蛋白、白蛋白/球蛋白比值、尿素、高密度脂蛋白胆固醇、钙、NLR，建立的预测模型具有良好的预测价值，可为医务人员筛查高出血风险患者及针对性治疗提供参考依据。.