开发并验证预测肝门部胆管癌根治性切除术后生存时间的模型

Development and Validation of a Model to Predict Overall Survival After Curative Resection of Perihilar Cholangiocarcinoma

目的开发并验证预测肝门部胆管癌(pCCA)根治性切除术后生存时间(OS)的模型。方法回顾性分析我院2009年12月至2018年6月接受根治性切除术患者的临床资料共310例。按照7∶3的比例,随机分为建模组(n=217)和验证组(n=93)。应用建模组数据,通过多因素Cox回归模型确定与pCCA根治目的切除术后OS独立相关的因素,并通过R中“rms”包开发模型。应用验证组数据,通过R中“rms”包验证模型。通过一致性检验和校准曲线来评价模型的准确性。比较模型与第八版美国癌症联合委员会(AJCC)肿瘤分期预测pCCA根治性切除术后OS的能力。结果研究队列的中位OS为23.0个月(95%CI 20.5~25.5),1年-、3年-、5年-OS率分别为74.9%、33.7%、22.8%。多因素分析结果提示:CA19-9>150 U/L、肿瘤浸润长度>3 cm、大血管侵犯、微血管侵犯、分化程度、淋巴结转移是影响pCCA患者根治性切除术后OS的独立危险因素。模型在建模组和验证组的C-指数分别为0.721(95%CI 0.678~0.764)和0.704(95%CI 0.689~0.719)。模型预测pCCA根治性切术后OS的能力均显著优于AJCC分期。结论开发并验证一个预测pCCA根治性切除术后OS的模型。该模型能够为患者和临床医生提供预测的预后信息,指导辅助治疗的临床决策。

Objective The purpose of this study was to develop and validate a model for predicting overall survival(OS)after radical resection of perihilar cholangiocarcinoma(pCCA).Methods This study was a retrospective analysis.The clinical data of 310 patients who underwent radical resection in the The First Affiliated Hospital of Army Medical University(Chongqing Southwest Hospital)from December 2009 to June 2018 were selected.According to the ratio of 7∶3,they were randomly divided into a modeling group(n=217)and a validation group(n=93).Using the modeling group data,the factors independently associated with OS after pCCA radical resection were determined by multivariate Cox regression model,and the model was developed by the"rms"package in R.The validation group data was applied to validate the model through the"rms"package in R.The accuracy of the model was evaluated by consistency test and calibration curve.The ability of the model to predict OS after radical pCCA resection was compared with the eighth edition of American Joint Committee on Cancer(AJCC)tumor staging.Results The median OS of the study cohort was 23.0 months(95%CI 20.5~25.5),and the 1-,3-,and 5-year-OS rates were 74.9%,33.7%,and 22.8%,respectively.Multivariate analysis showed that CA19-9>150 U/L,tumor invasion length>3 cm,macrovascular invasion,microvascular invasion,degree of differentiation,and lymph node metastasis were independent risk factors for OS after radical resection in pCCA patients.The C-index of the model in the modeling and validation groups was 0.721(95%confidence interval 0.678~0.764)and 0.704(95%confidence interval 0.689-0.719),respectively.The ability of the models to predict OS after radical resection of pCCA was significantly better than that of AJCC staging.Conclusions A model for predicting OS after radical resection of pCCA is developed and validated.The model can provide patients and clinicians with predicted prognostic information to guide clinical decision-making for adjuvant therapy.