即使经过根治性切除，肝内胆管癌（iCCA）患者的预后仍然令人失望，因为术后复发率极高。本次回顾性多中心队列研究招募了来自三个独立机构的280例根治性肝切除术后的iCCA患者。 iCCA 极早期复发 (VER) 定义为 6 个月内出现复发。来自对比增强 CT (CECT) 的 3D 肿瘤感兴趣区域 (ROI) 用于放射组学分析。 VER 的独立临床预测因子是组织学分期、AJCC 分期和 CA199 水平。我们实施了 K 均值聚类算法来研究 iCCA 的新型基于放射组学的亚型。 VER 预测采用了六种类型的机器学习 (ML) 算法，包括逻辑、随机森林 (RF)、神经网络、贝叶斯、支持向量机 (SVM) 和极限梯度提升 (XGBoost)。此外，还开发了六种临床 ML (CML) 模型和六种放射组学临床 ML (RCML) 模型来预测 VER。预测性能通过训练队列中的 10 倍交叉验证进行内部验证，并在外部验证队列中进一步评估。大约 30% 的 iCCA 患者经历了 VER，结果极其令人沮丧（风险比 (HR) = 5.77，95%置信区间 (CI) = 3.73-8.93，P < 0.001）。根据放射组学特征确定了两种不同的 iCCA 亚型，亚型 2 的 VER 比例较高（47.62% Vs 25.53%），且生存时间显着短于亚型 1。CML 和 RCML 模型的平均 AUC 值为 0.744 ± 0.018 、训练队列中的 0.900±0.014，以及外部验证队列中的 0.769±0.065 和 0.929±0.027。确定了两种基于放射组学的 iCCA 亚型，并开发了 6 个 RCML 模型来预测 iCCA 的 VER，这可以用作指导临床实践中个体化管理的有效工具。版权所有 © 2023 作者。由爱思唯尔有限公司出版。保留所有权利。

Even after curative resection, the prognosis for patients with intrahepatic cholangiocarcinoma (iCCA) remains disappointing due to the extremely high incidence of postoperative recurrence.A total of 280 iCCA patients following curative hepatectomy from three independent institutions were recruited to establish the retrospective multicenter cohort study. The very early recurrence (VER) of iCCA was defined as the appearance of recurrence within 6 months. The 3D tumor region of interest (ROI) derived from contrast-enhanced CT (CECT) was used for radiomics analysis. The independent clinical predictors for VER were histological stage, AJCC stage, and CA199 levels. We implemented K-means clustering algorithm to investigate novel radiomics-based subtypes of iCCA. Six types of machine learning (ML) algorithms were performed for VER prediction, including logistic, random forest (RF), neural network, bayes, support vector machine (SVM), and eXtreme Gradient Boosting (XGBoost). Additionally, six clinical ML (CML) models and six radiomics-clinical ML (RCML) models were developed to predict VER. Predictive performance was internally validated by 10-fold cross-validation in the training cohort, and further evaluated in the external validation cohort.Approximately 30 % of patients with iCCA experienced VER with extremely discouraging outcome (Hazard ratio (HR) = 5.77, 95 % Confidence Interval (CI) = 3.73-8.93, P < 0.001). Two distinct iCCA subtypes based on radiomics features were identified, and subtype 2 harbored a higher proportion of VER (47.62 % Vs 25.53 %) and significant shorter survival time than subtype 1. The average AUC values of the CML and RCML models were 0.744 ± 0.018, and 0.900 ± 0.014 in the training cohort, and 0.769 ± 0.065 and 0.929 ± 0.027 in the external validation cohort, respectively.Two radiomics-based iCCA subtypes were identified, and six RCML models were developed to predict VER of iCCA, which can be used as valid tools to guide individualized management in clinical practice.Copyright © 2023 The Authors. Published by Elsevier Ltd.. All rights reserved.