Genomic characteristics of hepatocellular carcinoma patients with response to sorafenib

Abstract

Background and aims: Sorafenib is a multiple receptor tyrosine kinase inhibitor which is the standard systemic therapy for advanced hepatocellular carcinoma (HCC). However, the objective response rate is low only reaching 10% and since there are other new 1st line treatment options such as lenvatinib and immune checkpoint inhibitors, biomarkers that may predict patients who will respond well to sorafenib is required. We implemented RNA sequencing (RNAseq) in HCC tumors to identify potential biomarkers that would predict response to sorafenib and uncover underlying biological features associated with better response.

Method: A total of 33 patients who had undergone liver resection prior to sorafenib treatment were enrolled. Matched tumor/ surrounding tumor tissues were obtained and RNA-seq was performed with the NextSeq500. Cluster analysis was performed and gene signature associated with sorafenib response was identified. The gene signature was validated in independent Sorafenib as Adjuvant Treatment in the Prevention Of Recurrence of Hepatocellular Carcinoma (STORM) cohort. Gene network analysis by Ingenuity Pathway Analysis (IPA) was performed to uncover activated pathways and key upstream regulators associated with response to sorafenib. The composition of infiltrated immune cells in tumors was also investigated by using the CIBERSORTx algorithm.

Results: The mean age was 58 ± 11 years with male predominance (81.8%), median child pugh score was 5 (range, 5–8) and 57.6% of the patients switched to second-line chemotherapy mostly due to HCC progression. The best response among 33 patients was complete response (CR) observed in 1 patient, partial response (PR) in two patients, stable disease (SD) in 12 patients while 18 patients showed disease progression. Gene signature (721 genes) associated with disease control (SD, PR, CR vs. no response) was derived using cluster analysis and was named as Korea University Sorafenib Response (KUSOR) gene signature. When applied on STORM cohort, KUSOR gene signature was able to predict patients who do not recur on adjuvant setting of sorafenib treatment after HCC resection or ablation with sensitivity of 91% and specificity of 74%. Gene network analysis by IPA revealed that patients who showed disease control were characterized by IL-6 and IL-1ß activation. In contrast, MYC was more activated in HCC tumors showing no benefit of the treatment, suggesting that MYC may trigger resistance of HCC cells to sorafenib. In addition, regulatory T cells (Treg cells) and M2 macrophage fractions were significantly higher in poor response group while the fraction of activated NK cells and CD4 cells were substantially higher in disease control group.

Conclusion: Our study reveals that KUSOR gene signature was able to identify patients who would show disease control when treated with sorafenib. MYC promotes hepatocarcinogenesis in chronic liver disease and overexpression is associated with poor response. Furthermore, it can also be inferred that poor response to sorafenib could be related immune evasion through overexpression of Treg cells and M2 macrophages in tumor microenvironment. Our study is in accord with previous studies where patients with high MYC activation showed poor response to sorafenib indicating that combination therapy such as immune checkpoint blockade should be recommended for these patients.