基于MRI影像组学及临床特征预测肿块型肝内胆管细胞癌的术后生存时间

目的 探讨影响肿块型肝内胆管细胞癌（mass-forming intrahepatic cholangiocarcinoma，MFICC）患者手术切除预后的相关因素。方法 回顾性收集2015年10月至2017年5月杭州市第三人民医院行外科手术切除治疗的MF-ICC患者42 例，均经病理学检查证实为MF-ICC。所有患者术前1 个月内行肝脏MRI平扫+增强检查，应用GE-AK软件标注MRI增强平衡期（magnetic resonance-equilibrium phase，MREP）图像上肿瘤病灶并提取9个量化的一阶影像组学特征，纳入临床特征10个指标。电话随访并记录患者术后生存时间，行Cox回归模型分析。结果 MRI影像组学特征及临床特征与MF-ICC患者术后生存具有明显相关性，大血管侵犯（HR 1.532，95%CI 2.245～9.328，P=0.025）、淋巴结肿大（HR 1.273，95%CI 1.352～5.463，P=0.042）和均匀度（HR 2.475，95%CI 3.226～12.001，P=0.002）是影响MF-ICC患者术后生存的不良预后因素。结论 MRI影像组学特征（均匀度）和临床特征（大血管侵犯和淋巴结肿大）是与MF-ICC患者术后总生存时间相关的独立危险因素，提示影像组学特征（均匀度）是具有临床应用前景的生物学特征。

Predicting postoperative survival time of mass-forming intrahepatic cholangiocarcinoma based on MRI radiomics and clinical features

Objective To explore related factors that affect the prognosis of patients with mass-forming intrahepatic cholangiocarcinoma (MF-ICC) after surgical resection. Methods A retrospective analysis wasconducted in 42 patients with MF-ICC who underwent surgical resection between Oct. 2015 to May 2017 in Third Hangzhou People’s Hospital. All MF-ICC were confirmed by pathology. The included patients underwent liver MRI plain scan and enhanced examination within 1 month before surgery, GE-AK software was used to annotate tumor lesions on MR-EP images and extract 9 first-order quantitative radiomics features, and 10 clinical characteristics were included. Telephone follow-up was conducted and the survival time of patients after operation was recorded, and Cox regression model analysis was performed. Results Radiomics and clinical features were significantly correlated with postoperative survival of patients with MF-ICC. Macrovascular invasion (HR 1.532, 95%CI 2.245~9.328, P=0.025), lymphadenopathy (HR 1.273, 95%CI 1.352~5.463, P=0.042) and uniformity (HR 2.475, 95%CI 3.226~12.001, P=0.002) were the poor prognostic factors for MF-ICC aftersurgery. Conclusion Clinical characteristics and radiomics characteristics are independent factors related to the overall survival time of patients with MF-ICC. Radiomics features (uniformity) may be promising biological feature for clinical applications.