Chendong Wang,1,* Bin Leng,1,* Ran You,1,* Zeyu Yu,1 Ya Lu,1 Lingfeng Diao,1 Hao Jiang,1 Yuan Cheng,2 Guowen Yin,1 Qingyu Xu1 

1Department of Interventional Oncology, Jiangsu Cancer Hospital, Jiangsu Institute of Cancer Research, The Affiliated Cancer Hospital of Nanjing Medical University, Nanjing, People’s Republic of China; 2Department of Medical Oncology, Jinling Hospital, Affiliated Hospital of Medical School, Nanjing University, Nanjing, People’s Republic of China

*These authors contributed equally to this work

Correspondence: Yuan Cheng, Department of Medical Oncology, Jinling Hospital, Affiliated Hospital of Medical School, Nanjing University, No. 34 Changfu Street, Nanjing, Jiangsu, 210007, People’s Republic of China, Email chengyuan@csco.ac.cn Qingyu Xu, Department of Interventional Oncology, Jiangsu Cancer Hospital, Jiangsu Institute of Cancer Research, The Affiliated Cancer Hospital of Nanjing Medical University, No. 42 Baiziting, Nanjing, Jiangsu, 210009, People’s Republic of China, Email xqy1984king@njmu.edu.cn

Purpose: The response to transarterial chemoembolization (TACE) varies among individuals with hepatocellular carcinoma (HCC). This study aimed to identify a biomarker for predicting TACE response in HCC patients and to investigate its correlations with the tumor microenvironment and pre-TACE radiomics features.
Patients and Methods: GSE104580 data were obtained from the Gene Expression Omnibus (GEO) database. Differentially expressed gene analysis and machine learning algorithms were used to identify genes for constructing the TACE failure signature (TFS). TFS scores were then calculated for HCC patients in The Cancer Genome Atlas (TCGA) cohort. After obtaining images from The Cancer Imaging Archive (TCIA), tumor labeling and radiomics feature extraction, the Rad-score model was generated. Correlation analysis was performed between the TFS score and the Rad-score. CIBERSORT, ssGSEA and TME analysis were performed to explore differences in the immune landscape among distinct risk groups. The immunotherapy response was compared between different groups.
Results: ADH1C, CXCL11, EMCN, SPARCL1 and LIN28B were selected and incorporated into the TFS, which demonstrated satisfactory performance in predicting TACE response. Patients in the high TFS score group had poorer overall survival (OS) than those in the low TFS score group. The Rad-score model was constructed using six radiomics features, and the Rad-score was significantly correlated with hub gene expression and the TFS score. The high-TFS group was also characterized by an immunosuppressive tumor microenvironment and exhibited unfavorable responses to immunotherapy with PD-1 and CTLA-4 checkpoint inhibitors.
Conclusion: This study established a transcriptomic biomarker for predicting the efficacy of TACE that correlates with radiomics features on pretreatment imaging, tumor immune microenvironment characteristics, and the efficacy of immunotherapy and targeted therapy in HCC patients.

Keywords: hepatocellular carcinoma, transarterial chemoembolization, transcriptomic, radiomics, tumor microenvironment