随机生存森林模型预测肝动脉化疗栓塞治疗肝细胞癌患者的预后

【摘要】【目的】应用随机生存森林算法研究影响肝动脉化疗栓塞（transcatheter arterial chemoembolization，TACE）治疗不能切除肝细胞癌（hepatocellular carcinoma，HCC）患者的预后因素，并构建预后模型。【方法】回顾性地纳入2014年1月至2017年12月期间就诊于本中心的、且一线治疗为TACE的HCC患者，并按照7:3比例划分为训练集和验证集。根据患者的临床数据、实验室指标及随访生存数据，建立Cox比例风险模型和基于机器学习算法的随机生存森林模型，并评估两种模型的预测能力。本研究统计方法由R 4.1.2软件完成。【结果】共有636例符合入排标准的患者纳入本研究，训练集有445例患者，验证集有191例患者。肿瘤负荷（肿瘤数目（个）+肿瘤最大径（cm））、年龄、基线G-谷氨酰转肽酶水平、基线甲胎蛋白水平和白蛋白-胆红素分级是影响TACE治疗不能切除HCC患者的独立预后因素。Cox回归模型的训练集1年、3年、5年的受试者操作特征曲线下面积（area under the receiver operating characteristic curve，AUC）为0.782、0.796和0.791，验证集中为0.750、0.766和0.766。随机生存森林模型训练集1年、3年和5年AUC为0.896、0.894和0.875，验证集中为0.743、0.763和0.770。随机生存森林模型能将患者区分为预后好组和预后差组，二者生存期有显著差异。决策曲线示随机生存森林模型的净获益要优于Cox比例风险模型。【结论】随机生存森林模型是预测TACE治疗HCC患者预后的可靠工具。【关键词】随机生存森林，Cox比例风险回归，肝动脉化疗栓塞，肝细胞癌，预后模型

Random survival forest model predicts the prognosis of patients with hepatocellular carcinoma treated by transcatheter arterial chemoembolization

AbstractsObjective: A random survival forest algorithm was applied to explore the significant factors and develop the prognosis model for patients with hepatocellular carcinoma (HCC) after transcatheter arterial chemoembolization (TACE). Methods: Unresectable HCC patients who underwent first-line TACE treatment in our center from January 2014 to December 2017 were retrospectively enrolled and been divided into training set and validation set with a 7:3 ratio. Based on the clinical data, laboratory indicators and follow-up survival of patients, the Cox proportional-hazards regression model and the random survival forest model based on machine learning algorithm was developed, and the predictive ability of the two models was evaluated. All statistical methods were completed by R 4.1.2 software. Results: A total of 636 eligible patients were enrolled in the study, including 445 patients in the training set and 191 patients in the validation set. The tumor burden (number of tumors + maximum tumor diameter (cm)), age, baseline gamma-glutamyl transpeptidase (GGT) level, baseline alpha-fetoprotein (AFP) level and albumin-bilirubin grade (ALBI) were independent factors affecting the prognosis of HCC patients treated with TACE. In the Cox model, the 1-year, 3-year and 5-year AUC of the training set was 0.782, 0.796 and 0.791, respectively, and the 1-year, 3-year and 5-year AUC of the validation set was 0.750, 0.766 and 0.766, respectively. The 1-year, 3-year and 5-year AUC of the training set in the random survival forest model was 0.896, 0.894 and 0.875, respectively, and the 1-year, 3-year and 5-year AUC of validation set was 0.743, 0.763 and 0.770, respectively. Random survival forest model could distinguish patients into good prognosis group and poor prognosis group, and the overall survival of these two groups is significantly different. The decision curve analysis showed that the net benefit of the random survival forest model is better than that of the Cox model. Conclusions: The random survival forest model is a reliable tool for predicting the prognosis of HCC patients treated with TACE.Key words: random survival forest, Cox proportional-hazards regression, transcatheter arterial chemoembolization, hepatocellular carcinoma, prognostic model