时序影像组学预测肝癌免疫联合治疗疗效

Temporal radiomics predicts response to immuno-combination therapy in hepatocellular carcinoma

摘要

目的：构建融合多期相增强影像、肿瘤内异质性（imaging，intratumoral heterogeneity，ITH）、时间序列影像组学（time-series radiomics，TSR）与Delta影像组学（delta-radiomics，DR）的预测模型，评估其在肝细胞癌（hepatocellular carcinoma，HCC）免疫联合治疗疗效判定中的价值，并以该模型为核心建立临床决策支持系统（clinical decision support system，CDSS）。方法：回顾性收集2021年1月—2024年12月在南京医科大学第一附属医院接受免疫联合治疗的62例HCC患者资料。围绕动脉期、门静脉期与延迟期增强影像提取传统影像组学、ITH、TSR、DR以及临床检验指标密度（clinical test index density，CTId）特征，经由三阶段筛选确定最优特征集，并借助多种机器学习算法训练融合预测模型；在此基础上搭建CDSS以考察其实际应用效能。结果：融合模型对治疗进展预测的验证集曲线下面积（area under the curve，AUC）达到0.821（95%CI：0.629~0.987），显著优于单期相影像组学模型（0.706~0.738）、ITH单一模型（0.752）及传统临床模型（0.685），DeLong检验均具统计学意义（P<0.05）；对≥Ⅱ级、≥Ⅲ级并发症的预测，AUC分别为0.803与0.845。影像组学风险分层在无进展生存期（HR=4.36，95%CI：1.94~9.80， P<0.001）与总生存期（HR=4.23，95%CI：1.78~10.08， P=0.001）层面均构成独立预后因子；在Ⅰ~Ⅱ期早期亚组内PFS判别效能保持稳定（χ²=14.60，P<0.001），总生存期因事件数偏少未达统计学意义（P=0.341）。所搭建CDSS完成单例分层用时≤60 s，与人工分析一致性达100%。结论：融合多期相与纵向动态特征的影像组学模型能够较为准确地刻画HCC免疫联合治疗的疗效及并发症风险，依托该模型搭建的CDSS可为个体化临床决策提供可量化的支持工具。

Abstract

Objective：To develop a combined prediction model that integrates multi-phase contrast-enhanced imaging，intratumoral heterogeneity（ITH），time-series radiomics（TSR）and delta-radiomics（DR）features for assessing the therapeutic response of hepatocellular carcinoma（HCC）to immuno-combination therapy，and to build a clinical decision support system（CDSS）upon it. Methods：Sixty-two HCC patients who received immuno-combination therapy at the First Affiliated Hospital of Nanjing Medical University between January 2021 and December 2024 were retrospectively enrolled. Traditional radiomics，ITH，TSR，DR，and clinical test index density（CTId）features were extracted from arterial，portal-venous and delayed-phase images. A three-stage feature selection pipeline was employed to identify the optimal feature set，and multiple machine-learning classifiers were trained to construct the combined model，based on which a CDSS was subsequently developed and evaluated. Results：The combined model yielded a validation AUC of 0.821（95%CI：0.629-0.987）for predicting disease progression，significantly outperforming single-phase radiomics models（0.706-0.738），the ITH-only model（0.752）and the clinical model（0.685），with all differences statistically significant via Delong test（all P<0.05）. The AUC values for predicting grade ≥Ⅱ and ≥Ⅲ complications reached 0.803 and 0.845，respectively. Radiomics-based risk stratification independently predicted both progression-free survival（HR=4.36，95%CI：1.94-9.80， P<0.001）and overall survival（HR=4.23，95%CI：1.78-10.08， P=0.001）. In the early-stage（TNM Ⅰ-Ⅱ）subgroup，PFS stratification remained robust（χ ²=14.60，P<0.001），whereas the OS difference did not reach statistical significance（P=0.341），likely owing to the limited subgroup sample size. The established CDSS completed stratification for a single case within 60 seconds，achieving 100% consistency with manual analysis. Conclusion：The combined radiomics model integrating multi-phase and longitudinal dynamics accurately characterises therapeutic response and complication risk in HCC immuno-combination therapy，and the accompanying CDSS offers a quantitative decision-support tool for individualised clinical management.

关键词

肝细胞癌 / 增强磁共振成像 / 影像组学 / 时间序列 / 临床决策支持系统

Key words

hepatocellular carcinoma / contrast-enhanced MRI / radiomics / time-series / clinical decision support system

参考文献

[1] BRAY F, LAVERSANNE M, SUNG H, et al. Global cancer statistics 2022：GLOBOCAN estimates of incidence and mortality worldwide for 36 cancers in 185 countries[J]. CA Cancer J Clin, 2024, 74(3):229-263

[2] 国家卫生健康委员会医政医管局. 原发性肝癌诊疗指南（2024年版）[J]. 临床肝胆病杂志, 2024, 40(5):893-918

[3] National Health Commission of the People’s Republic of China. Guidelines for diagnosis and treatment of primary liver cancer(2024 edition)[J]. Journal of Clinical Hepatology, 2024, 40(5):893-918

[4] FINN R S, QIN S, IKEDA M, et al. Atezolizumab plus bevacizumab in unresectable hepatocellular carcinoma[J]. N Engl J Med, 2020, 382(20):1894-1905

[5] CHENG A L, QIN S, IKEDA M, et al. Updated efficacy and safety data from IMbrave150：atezolizumab plus bevacizumab vs sorafenib for unresectable hepatocellular carcinoma[J]. J Hepatol, 2022, 76(4):862-873

[6] QIN S, CHAN S L, GU S, et al. Camrelizumab plus rivoceranib versus sorafenib as first—line therapy for unresectable hepatocellular carcinoma(CARES—310)：a randomised，open—label，international phase 3 study[J]. Lancet, 2023, 402(10408):1133-1146

[7] LEWIS S, HECTORS S, TAOULI B. Radiomics of hepatocellular carcinoma[J]. Abdom Radiol(NY), 2021, 46(1):111-123

[8] 杨春, 黄鹏, 曾蒙苏. 《原发性肝癌诊疗指南2024版》影像学主要更新及亮点解读[J]. 中国临床医学, 2024, 31(3):345-352

[9] YANG C, HUANG P, ZENG M S. Interpretation of main imaging updates and highlights in guidelines for diagnosis and treatment of primary liver cancer(2024 edition)[J]. Chinese Journal of Clinical Medicine, 2024, 31(3):345-352

[10] WANG Y, WEI R, LIU G, et al. Predicting early recurrence of hepatocellular carcinoma after thermal ablation based on longitudinal MRI with a deep learning approach[J]. Oncologist, 2025, 30(3):oyaf013

[11] HUANG K, GUO L, DAI J, et al. Predicting treatment response and prognosis of immune checkpoint inhibitors—based combination therapy in advanced hepatocellular carcinoma using a longitudinal CT—based radiomics model：a multicenter study[J]. BMC Cancer, 2025, 25(1):523

[12] XU Y, ZHANG M, YANG G, et al. Machine learning based radiomic models outperform clinical biomarkers in predicting outcomes after immunotherapy for hepatocellular carcinoma[J]. J Hepatol, 2025, 83(2):434-446

[13] LI Y, ZHANG Y, FANG Q, et al. Habitat radiomics based on CT images to predict survival and immune status in hepatocellular carcinoma，a multi—cohort validation study[J]. Acad Radiol, 2025, 32(5):2728-2741

[14] ZHANG H, ZHANG Z, GAN K, et al. 18F—FDG PET/CT—based habitat radiomics combining stacking ensemble learning for predicting prognosis in hepatocellular carcinoma：a multi—center study[J]. BMC Cancer, 2024, 24(1):1458

[15] LIU D, LIU F, XIE X, et al. Delta—radiomics analysis based on multi—phase contrast—enhanced MRI for predicting early recurrence in hepatocellular carcinoma after percutaneous thermal ablation[J]. Eur Radiol, 2023, 33(4):2514-2525

[16] LI M, ZHANG R, ZHANG F, et al. Noninvasive prognostic classification of intratumoral heterogeneity in hepatocellular carcinoma with multi—omics insights and therapeutic implications[J]. Sci Adv, 2025, 11(16):eads8323

[17] SONG H, ZHU L, SUN J, et al. Pretreatment CT—based machine learning radiomics model predicts response in unresectable hepatocellular carcinoma treated with lenvatinib plus PD—1 inhibitors and interventional therapy[J]. BMC Cancer, 2024, 24(1):1148

[18] HONG Y M, CHO M, YOON K T, et al. Complications of immunotherapy in advanced hepatocellular carcinoma[J]. Front Oncol, 2023, 13:1267180

[19] XU Y, SUN S, LIU Q, et al. Impact of immune—related adverse events in patients with hepatocellular carcinoma treated with atezolizumab plus bevacizumab[J]. Hepatol Commun, 2024, 8(4):e0428

[20] CLAVIEN P A, BARKUN J, DE OLIVEIRA M L, et al. The Clavien—Dindo classification of surgical complications：five—year experience[J]. Ann Surg, 2009, 250(2):187-196

[21] JUNG J, WOO S, KIM H S, et al. Machine learning—based clinical decision support system for treatment recommendation and overall survival prediction of hepatocellular carcinoma：a multi—center study[J]. NPJ Digit Med, 2024, 7(1):34

[22] WANG J, ZHENG J, MA M, et al. MRI radiomics to predict early treatment response to TACE combined with lenvatinib plus a PD—1 inhibitor for hepatocellular carcinoma with portal vein tumor thrombus[J]. J Hepatocell Carcinoma, 2025, 12:631-644

[23] 包建英, 崔传芳, 林征, 等. 肝细胞肝癌患者首次经动脉化疗栓塞术后中重度腹痛预测模型的构建及验证[J]. 南京医科大学学报（自然科学版）, 2024, 44(5):655-660

[24] BAO J Y, CUI C F, LIN Z, et al. Construction and validation of a prediction model for moderate—to—severe abdominal pain after first transarterial chemoembolization in patients with hepatocellular carcinoma[J]. Journal Nanjing Medical University(Natural Sciences), 2024, 44(5):655-660

[25] 李树芸, 王宗盛, 徐春玲. 增强CT联合肿瘤标志物检测在肝癌诊断及预后中的应用价值[J]. 南京医科大学学报（自然科学版）, 2020, 40(11):1628-1632

[26] LI S Y, WANG Z S, XU C L. Application value of contrast—enhanced CT combined with tumor marker detection in the diagnosis and prognosis of liver cancer[J]. Journal Nanjing Medical University(Natural Sciences), 2020, 40(11):1628-1632

基金资助

江苏省科技厅临床前沿技术(BF2024053)