Reliable biomarkers are required to predict the response to sorafenib. We investigated genomic variations associated with responsiveness to sorafenib for patients with unresectable hepatocellular carcinoma (HCC). Blood samples from 2 extreme, 2 strong and 3 poor responders to sorafenib were subjected to whole-genome analysis. Then, we validated candidate genomic variations with another 174 HCC patients, and performed in vitro functional analysis and in silico analyses. Genomic data of >96 gigabases/sample was generated at average of ~34X sequencing depth. In total, 1813 genomic variations were matched to sorafenib responses in clinical data; 708 were located within regions for sorafenib-target genes or drug absorption, distribution, metabolism, and excretion (ADME)-related genes. From them, 36 variants were within the coding regions and 6 identified as non-synonymous single-nucleotide variants from 4 ADME-related genes (ABCB1, FMO3, MUSK, and SLC15A2). Validation genotyping confirmed sequencing results and revealed patients genotype for rs2257212 in SLC15A2 showed longer progression-free survival (HR = 2.18). In vitro study displayed different response to sorafenib depending on the genotype of SLC15A2. Structural prediction analysis revealed changes of the phosphorylation levels in protein, potentially affecting sorafenib-associated enzymatic activity. Our finding using extreme responder seems to generate robust biomarker to predict the response of sorafenib treatment for HCC.
Keywords: extraordinary response; genome variations; hepatocellular carcinoma; sorafenib; whole-genome sequencing.