不可切除胆道癌致梗阻性黄疸患者行介入术后的生存预测模型建立

目的探讨因不可切除胆道癌致阻黄患者行介入治疗后预后生存的影响因素，并构建列线图个体化预测患者总生存期（OS）。 方法回顾性分析我院2015年6月至2021年7月收治的261例胆道癌行经皮介入降黄患者的资料。总体划分为训练组（n = 188）及验证组（n = 73）。单因素和多因素Cox回归分析筛选患者OS的独立影响因素，并以此构建列线图，个体化预测不可切除胆道癌伴有阻黄患者介入治疗术后生存情况。绘制AUC曲线及一致性曲线评估模型预测性能。通过K-M曲线评估分类预测因子的危险分层能力。 结果在261例患者中，训练组中位OS为244 d（IQR：213，275）。验证组中位OS为236 d（IQR：186，285）。单因素及多因素回归分析显示，总胆红素、总胆固醇、血红蛋白、血清钠、糖类抗原199水平和肿瘤亚型为预测不可切除胆道癌伴阻黄患者3、6个月和1年的OS的独立相关因素，并建立诺模图。ROC曲线下面积显示，该模型在训练组（0.817、0.825和0.796）和验证组（0.921、0.880和0.904）对3、6个月和1年的OS预测具有良好的分辨能力，可以较准确地预测患者的总体生存情况。 结论本模型能够较好预测不可切除胆道癌伴阻黄患者3、6个月、1年的生存机会，为临床治疗策略选择提供一定的帮助。

Establishment of a survival prediction model for patients with obstructive jaundice caused by unresectable biliary tract cancer undergoing interventional therapy

ObjectiveTo explore the factors affecting the prognosis and survival of patients with unresectable biliary tract cancer caalignant biliary obstruction after interventional therapy, and to construct an individualized nomogram to predict the overall survival time (overall survival, OS). MethodsFrom June 2015 to July 2021, a total of 261 patients with obstructive jaundice caused by unresectable biliary tract cancer treated with interventional decompression in our center were retrospectively analyzed. The patients were divided into two groups: a training group (n = 188) and a validation group (n = 73). The univariate and multivariate Cox proportional hazard regression analyses were used to determine the independent factors related to OS, and an individual nomogram was constructed to visualize the new model and predict 3-month, 6-month, and 1-year survival. The area under the curves (AUC) and the calibration curves were drawn to evaluate predictive performance in total patients . The risk stratification ability of each categorical predictors was evaluated by Kaplan-Meier curves. ResultsFor the enrolled patients, the median OS was 244 days (IQR: 213, 275) in the training group and 236 days (IQR: 186, 285) in the validation group. The univariate and multivariate regression analyses showed that the independent factors were total bilirubin, total cholesterol, hemoglobin, serum sodium, carbohydrate antigen-9 and subtypes of tumor. A nomogram was successfully established to predict 3-months, 6-month and 1-year survival probability. The AUCs showed that the new model had good discrimination for 3-month, 6-month and 1-year OS prediction in the training group (0.817, 0.825 and 0.796, respectively) and the calibration group (0.921, 0.880 and 0.904, respectively). ConclusionsThe new model could effectively predict 3 months, 6 months and 1 year OS for patients with unresectable biliary tract cancer caused obstructive jaundice undergoing interventional decompression, which may help for clinical decision making.