MRI-Based Radiomics for Predicting Treatment Response to TACE Combined with Molecular-Targeted Agents and Immune Checkpoint Inhibitors in Unresectable Hepatocellular Carcinoma

doi:10.3877/cma.j.issn.2095-5782.2026.01.011

Abstract

Abstract:

Objective

To develop and validate a model based on contrast-enhanced magnetic resonance imaging (MRI) radiomics features to evaluate the tumor response of patients with unresectable hepatocellular carcinoma (uHCC) receiving transarterial chemoembolization (TACE) combined with molecular-targeted agents (MTAs) and immune checkpoint inhibitors (ICIs).

Methods

A total of 93 uHCC patients who received TACE-MTA-ICI triple therapy at the First Affiliated Hospital with Nanjing Medical University between June 2019 and January 2024 were retrospectively included. Radiomics features were extracted from MRI images, and multiple machine learning algorithms were employed to construct prediction models, followed by performance evaluation.

Results

The radiomics model based on the Random Forest algorithm demonstrated superior performance. A nomogram model, further integrated with clinical features, achieving areas under the curve (AUC) of 0.933 and 0.890 in the training and testing cohort, respectively.

Conclusion

The combined model based on MRI radiomics and clinical features can effectively predict tumor response to TACE-MTA-ICI triple therapy in uHCC patients.

Key words: Transarterial chemoembolization, Molecular-targeted agents, Immune checkpoint inhibitors, Radiomics, Hepatocellular carcinoma

Xiao Shen, Haitao Yan, Jinxing Zhang, Qingquan Zu, Sheng Liu, Haibin Shi. MRI-Based Radiomics for Predicting Treatment Response to TACE Combined with Molecular-Targeted Agents and Immune Checkpoint Inhibitors in Unresectable Hepatocellular Carcinoma[J]. Chinese Journal of Interventional Radiology(Electronic Edition), 2026, 14(01): 72-80.

/ / Recommend

Add to citation manager EndNote|Ris|BibTeX

URL: https://zhjrfsxdzzz.cma-cmc.com.cn/EN/10.3877/cma.j.issn.2095-5782.2026.01.011

https://zhjrfsxdzzz.cma-cmc.com.cn/EN/Y2026/V14/I01/72

Figures/Tables 9

特征	总计(n=93)	训练集(n=65)	测试集(n=28)	P值
年龄（岁， $x ¯$ ±s）	58.09±10.21	58.58±9.95	56.93±10.87	0.476
性别[例(%)]				0.895
男性	82(88.17)	58(89.23)	24(85.71)
女性	11(11.83)	7(10.77)	4(14.29)
BCLC分期[例(%)]				0.308
0期	49(52.69)	37(56.92)	12(42.86)
1期	44(47.31)	28(43.08)	16(57.14)
病因[例(%)]				0.603
病毒性肝炎	86(92.47)	59(90.77)	27(96.43)
其他	7(7.53)	6(9.23)	1(3.57)
ECOG评分[例(%)]				0.553
0	59(63.44)	43(66.15)	16(57.14)
1	34(36.56)	22(33.85)	12(42.86)
Child-Pugh评分[例(%)]				0.913
5	67(72.04)	46(70.77)	21(75.00)
6	22(23.66)	16(24.62)	6(21.43)
7	4(4.30)	3(4.62)	1(3.57)
ALT (U/L， $x ¯$ ±s)	43.63±23.54	43.48±24.04	44.00±22.76	0.851
AST (U/L， $x ¯$ ±s)	64.88±46.29	64.13±50.20	66.61±36.37	0.395
AFP[例(%)]				0.909
≤400 ng/mL	49(52.69)	35(53.85)	14(50.00)
>400 ng/mL	44(47.31)	30(46.15)	14(50.00)
白蛋白(g/L， $x ¯$ ±s)	36.42±4.22	36.33±4.44	36.63±3.70	0.755
总胆红素(μmol/L， $x ¯$ ±s)	17.93±8.45	17.34±7.61	19.32±10.15	0.498
NLR ( $x ¯$ ±s)	3.24±1.63	3.31±1.77	3.07±1.23	0.769
肿瘤数目[例(%)]				0.119
单发	19(20.43)	10(15.38)	9(32.14)
多发	74(79.57)	55(84.62)	19(67.86)
肿瘤最大径（cm， $x ¯$ ±s）	9.05±3.88	9.08±3.94	8.96±3.78	0.887
血管侵犯[例(%)]				0.835
无	53(56.99)	38(58.46)	15(53.57)
有	40(43.01)	27(41.54)	13(46.43)
肝外转移[例(%)]				0.662
无	74(79.57)	53(81.54)	21(75.00)
有	19(20.43)	12(18.46)	7(25.00)

特征	总计(n=93)	训练集(n=65)	测试集(n=28)	P值
年龄（岁， $x ¯$ ±s）	58.09±10.21	58.58±9.95	56.93±10.87	0.476
性别[例(%)]				0.895
男性	82(88.17)	58(89.23)	24(85.71)
女性	11(11.83)	7(10.77)	4(14.29)
BCLC分期[例(%)]				0.308
0期	49(52.69)	37(56.92)	12(42.86)
1期	44(47.31)	28(43.08)	16(57.14)
病因[例(%)]				0.603
病毒性肝炎	86(92.47)	59(90.77)	27(96.43)
其他	7(7.53)	6(9.23)	1(3.57)
ECOG评分[例(%)]				0.553
0	59(63.44)	43(66.15)	16(57.14)
1	34(36.56)	22(33.85)	12(42.86)
Child-Pugh评分[例(%)]				0.913
5	67(72.04)	46(70.77)	21(75.00)
6	22(23.66)	16(24.62)	6(21.43)
7	4(4.30)	3(4.62)	1(3.57)
ALT (U/L， $x ¯$ ±s)	43.63±23.54	43.48±24.04	44.00±22.76	0.851
AST (U/L， $x ¯$ ±s)	64.88±46.29	64.13±50.20	66.61±36.37	0.395
AFP[例(%)]				0.909
≤400 ng/mL	49(52.69)	35(53.85)	14(50.00)
>400 ng/mL	44(47.31)	30(46.15)	14(50.00)
白蛋白(g/L， $x ¯$ ±s)	36.42±4.22	36.33±4.44	36.63±3.70	0.755
总胆红素(μmol/L， $x ¯$ ±s)	17.93±8.45	17.34±7.61	19.32±10.15	0.498
NLR ( $x ¯$ ±s)	3.24±1.63	3.31±1.77	3.07±1.23	0.769
肿瘤数目[例(%)]				0.119
单发	19(20.43)	10(15.38)	9(32.14)
多发	74(79.57)	55(84.62)	19(67.86)
肿瘤最大径（cm， $x ¯$ ±s）	9.05±3.88	9.08±3.94	8.96±3.78	0.887
血管侵犯[例(%)]				0.835
无	53(56.99)	38(58.46)	15(53.57)
有	40(43.01)	27(41.54)	13(46.43)
肝外转移[例(%)]				0.662
无	74(79.57)	53(81.54)	21(75.00)
有	19(20.43)	12(18.46)	7(25.00)

特征	单因素分析		多因素分析
特征	OR(95% CI)	P值	OR(95% CI)	P值
年龄	1.00(0.99~1.00)	0.196
性别	0.75(0.21~2.64)	0.706
BCLC分期	1.55(0.82~2.92)	0.261
病因	0.20(0.03~1.21)	0.142
ECOG评分	1.20(0.59~2.43)	0.670
Child-Pugh评分	0.94(0.87~1.02)	0.190
ALT	1.00(0.99~1.01)	0.692
AST	1.00(0.99~1.00)	0.774
AFP	1.14(0.63~2.09)	0.715
白蛋白	0.99(0.98~1.00)	0.160
总胆红素	0.99(0.97~1.01)	0.445
NLR	0.96(0.86~1.08)	0.575
肿瘤数目	0.82(0.66~1.03)	0.149
肿瘤最大径	0.98(0.94~1.02)	0.457
血管侵犯	2.38(1.19~4.75)	0.040	2.38 (1.19~4.75)	0.040
肝外转移	2.00(0.73~5.47)	0.258

分组	模型名称	AUC (95% CI)	准确率	敏感度	特异度
训练集	逻辑回归	0.940 (0.887~0.992)	0.877	1.000	0.789
	朴素贝叶斯	0.863 (0.771~0.954)	0.831	0.741	0.895
	支持向量机	0.923 (0.855~0.991)	0.877	0.815	0.921
	K近邻算法	0.852 (0.762~0.942)	0.800	0.667	0.895
	随机森林	0.904 (0.829~0.979)	0.877	0.741	0.974
	多层感知机	0.900 (0.826~0.973)	0.846	0.815	0.868
测试集	逻辑回归	0.781 (0.592~0.970)	0.750	0.909	0.647
	朴素贝叶斯	0.829 (0.679~0.979)	0.750	1.000	0.588
	支持向量机	0.765 (0.579~0.950)	0.786	0.455	1.000
	K近邻算法	0.767 (0.582~0.953)	0.714	0.727	0.706
	随机森林	0.864 (0.728~0.999)	0.786	0.818	0.765
	多层感知机	0.829 (0.677~0.980)	0.786	1.000	0.647

分组	模型名称	AUC (95%CI)	准确率	敏感度	特异度
训练集	临床模型	0.747 (0.637~0.856)	0.754	0.704	0.789
	影像组学	0.904 (0.829~0.979)	0.877	0.741	0.974
	联合模型	0.933 (0.869~0.997)	0.877	0.889	0.868
测试集	临床模型	0.567 (0.371~0.763)	0.571	0.545	0.588
	影像组学	0.864 (0.728~0.999)	0.786	0.818	0.765
	联合模型	0.890 (0.738~1.000)	0.893	0.909	0.882

References 22

[1]	Filho AM, Laversanne M, Ferlay J, et al. The GLOBOCAN 2022 cancer estimates: Data sources, methods, and a snapshot of the cancer burden worldwide[J]. Int J Cancer, 2025, 156(7): 1336-1346.
[2]	Llovet JM, Kelley RK, Villanueva A, et al. Hepatocellular carcinoma[J]. Nat Rev Dis Primers, 2021, 7(1): 6.
[3]	European Association for the Study of the Liver. EASL Clinical Practice Guidelines on the management of hepatocellular carcinoma[J]. J hepatol, 2025, 82(2): 315-374.
[4]	Singal AG, Llovet JM, Yarchoan M, et al. AASLD Practice Guidance on prevention, diagnosis, and treatment of hepatocellular carcinoma[J]. Hepatology, 2023, 78(6): 1922-1965.
[5]	中华人民共和国国家卫生健康委员会医政司. 原发性肝癌诊疗指南(2024年版)[J]. 中华肝脏病杂志, 2024, 32(7): 581-630.
[6]	Pinato DJ, Murray SM, Forner A, et al. Trans-arterial chemoembolization as a loco-regional inducer of immunogenic cell death in hepatocellular carcinoma: implications for immunotherapy[J]. J Immunother Cancer, 2021, 9(9): e003311.
[7]	Llovet JM, De Baere T, Kulik L, et al. Locoregional therapies in the era of molecular and immune treatments for hepatocellular carcinoma[J]. Nat Rev Gastroenterol Hepatol, 2021, 18(5): 293-313.
[8]	Sangro B, Kudo M, Erinjeri JP, et al. Durvalumab with or without bevacizumab with transarterial chemoembolisation in hepatocellular carcinoma (EMERALD-1): a multiregional, randomised, double-blind, placebo-controlled, phase 3 study[J]. Lancet, 2025, 405(10474): 216-232.
[9]	Kudo M, Ren Z, Guo Y, et al. Transarterial chemoembolisation combined with lenvatinib plus pembrolizumab versus dual placebo for unresectable, non-metastatic hepatocellular carcinoma (LEAP-012): a multicentre, randomised, double-blind, phase 3 study[J]. Lancet, 2025, 405(10474): 203-215.
[10]	Xu B, Dong SY, Bai XL, et al. Tumor Radiomic Features on Pretreatment MRI to Predict Response to Lenvatinib plus an Anti-PD-1 Antibody in Advanced Hepatocellular Carcinoma: A Multicenter Study[J]. Liver Cancer, 2023, 12(3): 262-276.
[11]	Hua Y, Sun Z, Xiao Y, 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]. J Immunother Cancer, 2024, 12(7): e008953.
[12]	Llovet JM, Lencioni R. mRECIST for HCC: Performance and novel refinements[J]. J Hepatol, 2020, 72(2): 288-306.
[13]	Zhu HD, Li HL, Huang MS, et al. Transarterial chemoembolization with PD-(L)1 inhibitors plus molecular targeted therapies for hepatocellular carcinoma (CHANCE001)[J]. Signal Transduction Targeted Ther, 2023, 8(1): 58.
[14]	Jin ZC, Zhong BY, Chen JJ, et al. Real-world efficacy and safety of TACE plus camrelizumab and apatinib in patients with HCC (CHANCE2211): a propensity score matching study[J]. Eur Radiol, 2023, 33(12): 8669-8681.
[15]	Jin ZC, Chen JJ, Zhu XL, et al. Immune checkpoint inhibitors and anti-vascular endothelial growth factor antibody/tyrosine kinase inhibitors with or without transarterial chemoembolization as first-line treatment for advanced hepatocellular carcinoma (CHANCE2201): a target trial emulation study[J]. E Clinical Medicine, 2024, 72: 102622.
[16]	Reig M, Forner A, Rimola J, et al. BCLC strategy for prognosis prediction and treatment recommendation: The 2022 update[J]. J Hepatol, 2022, 76(3): 681-693.
[17]	Tischfield DJ, Gurevich A, Johnson O, et al. Transarterial Embolization Modulates the Immune Response within Target and Nontarget Hepatocellular Carcinomas in a Rat Model[J]. Radiology, 2022, 303(1): 215-225.
[18]	Fukumura D, Kloepper J, Amoozgar Z, et al. Enhancing cancer immunotherapy using antiangiogenics: opportunities and challenges[J]. Nat Rev Clin Oncol, 2018, 15(5): 325-340.
[19]	Bo Z, Chen B, Zhao Z, et al. Prediction of Response to Lenvatinib Monotherapy for Unresectable Hepatocellular Carcinoma by Machine Learning Radiomics: A Multicenter Cohort Study[J]. Clin Cancer Res, 2023, 29(9): 1730-1740.
[20]	Chen M, Cao J, Hu J, et al. Clinical-Radiomic Analysis for Pretreatment Prediction of Objective Response to First Transarterial Chemoembolization in Hepatocellular Carcinoma[J]. Liver Cancer, 2021, 10(1): 38-51.
[21]	Yuan G, Song Y, Li Q, et al. Development and Validation of a Contrast-Enhanced CT-Based Radiomics Nomogram for Prediction of Therapeutic Efficacy of Anti-PD-1 Antibodies in Advanced HCC Patients[J]. Front Immunol, 2020, 11: 613946.
[22]	Yang C, Yang HC, Luo YG, et al. Predicting Survival Using Whole-Liver MRI Radiomics in Patients with Hepatocellular Carcinoma After TACE Refractoriness[J]. Cardiovasc Inter Rad, 2024, 47(7): 964-977.

Please choose a citation manager

Content to export