Identification of a novel Gene Signature and potential mechanisms related to Targeted Drug Resistance in Hepatocellular Carcinoma Based on integrating bioinformatics and machine learning

Abstract

Abstract Ajuvant therapy with molecularly targeted drugs has become the effective treatment for advanced hepatocellular carcinoma (HCC). While Hypoxia often induces changes in the tumor immune microenvironment and affects the progression of targeted drug resistance, there is a critically unmet need for effective identification of drug resistance progression to reverse targeted drug resistance. Herein, we identified 64 sorafenib-resistance genes for hierarchical clustering of 374 HCC patients in the TCGA database. The functional enrichment between low (LR-group) and high (HR-group) resistance groups was explored through GO, KEGG, GSVA, ssGSEA, CIBERSORT, XCELL and three hypoxia scoring formula. It was found that the upregulated epithelial-mesenchymal transition (EMT), higher hypoxic scores and lower CD8 + T cell infiltration in HR-group. we further identified that HR-group had higher CD8 + T cell exhaustion, and the immune checkpoints of CD8 + T cell involved in tumor antigen recognition disorders significantly increased. Furthermore, form hypoxia-related resistance gene signature (HDRGs)(including 9 key genes),we derive a risk score: the score correlates strongly with hypoxia, targeted drug resistance, CD8 + T cell infiltration and exhaustion and is accurately verified in TCGA, ICGC and GAO’ HCC Cohort. Additionaly, experimental verification showed that ADM were upregulated under hypoxia, so knockdown of ADM can inhibit EMT under hypoxia and increase the sensitivity of Lenvatinib. Collectively, this study reveals that hypoxia-induced dysfunction of CD8 + T cells causes drug resistance, which can be effectively predicted by our HDRGs, and broadly leveraging this risk score to provide guidance for tumor targeting and combination immunotherapy.