Biomarkers Associated With Response to Regorafenib in Patients With Hepatocellular Carcinoma

Background & AimsIn a phase 3 trial (RESORCE), regorafenib increased overall survival compared with placebo in patients with hepatocellular carcinoma (HCC) previously treated with sorafenib. In an exploratory study, we analyzed plasma and tumor samples from study participants to identify genetic, microRNA (miRNA), and protein biomarkers associated with response to regorafenib.MethodsWe obtained archived tumor tissues and baseline plasma samples from patients with HCC given regorafenib in the RESORCE trial. Baseline plasma samples from 499 patients were analyzed for expression of 294 proteins (DiscoveryMAP) and plasma samples from 349 patients were analyzed for levels of 750 miRNAs (miRCURY miRNA PCR). Tumor tissues from 7 responders and 10 patients who did not respond (progressors) were analyzed by next-generation sequencing (FoundationOne). Forty-six tumor tissues were analyzed for expression patterns of 770 genes involved in oncogenic and inflammatory pathways (PanCancer Immune Profiling). Associations between plasma levels of proteins and miRNAs and response to treatment (overall survival and time to progression) were evaluated using a Cox proportional hazards model.ResultsDecreased baseline plasma concentrations of 5 of 266 evaluable proteins (angiopoietin 1, cystatin B, the latency-associated peptide of transforming growth factor beta 1, oxidized low-density lipoprotein receptor 1, and C-C motif chemokine ligand 3; adjusted P ≤ .05) were significantly associated with increased overall survival time after regorafenib treatment. Levels of these 5 proteins, which have roles in inflammation and/or HCC pathogenesis, were not associated with survival independently of treatment. Only 20 of 499 patients had high levels and a reduced survival time. Plasma levels of α-fetoprotein and c-MET were associated with poor outcome (overall survival) independently of regorafenib treatment only. We identified 9 plasma miRNAs (MIR30A, MIR122, MIR125B, MIR200A, MIR374B, MIR15B, MIR107, MIR320, and MIR645) whose levels significantly associated with overall survival time with regorafenib (adjusted P ≤ .05). Functional analyses of these miRNAs indicated that their expression level associated with increased overall survival of patients with tumors of the Hoshida S3 subtype. Next-generation sequencing analyses of tumor tissues revealed 49 variants in 27 oncogenes or tumor suppressor genes. Mutations in CTNNB1 were detected in 3 of 10 progressors and VEGFA amplification in 1 of 7 responders.ConclusionWe identified expression patterns of plasma proteins and miRNAs that associated with increased overall survival times of patients with HCC following treatment with regorafenib in the RESORCE trial. Levels of these circulating biomarkers and genetic features of tumors might be used to identify patients with HCC most likely to respond to regorafenib. ClinicalTrials.gov number NCT01774344. NCBI GEO accession numbers: mRNA data (NanoString): GSE119220; miRNA data (Exiqon): GSE119221 In a phase 3 trial (RESORCE), regorafenib increased overall survival compared with placebo in patients with hepatocellular carcinoma (HCC) previously treated with sorafenib. In an exploratory study, we analyzed plasma and tumor samples from study participants to identify genetic, microRNA (miRNA), and protein biomarkers associated with response to regorafenib. We obtained archived tumor tissues and baseline plasma samples from patients with HCC given regorafenib in the RESORCE trial. Baseline plasma samples from 499 patients were analyzed for expression of 294 proteins (DiscoveryMAP) and plasma samples from 349 patients were analyzed for levels of 750 miRNAs (miRCURY miRNA PCR). Tumor tissues from 7 responders and 10 patients who did not respond (progressors) were analyzed by next-generation sequencing (FoundationOne). Forty-six tumor tissues were analyzed for expression patterns of 770 genes involved in oncogenic and inflammatory pathways (PanCancer Immune Profiling). Associations between plasma levels of proteins and miRNAs and response to treatment (overall survival and time to progression) were evaluated using a Cox proportional hazards model. Decreased baseline plasma concentrations of 5 of 266 evaluable proteins (angiopoietin 1, cystatin B, the latency-associated peptide of transforming growth factor beta 1, oxidized low-density lipoprotein receptor 1, and C-C motif chemokine ligand 3; adjusted P ≤ .05) were significantly associated with increased overall survival time after regorafenib treatment. Levels of these 5 proteins, which have roles in inflammation and/or HCC pathogenesis, were not associated with survival independently of treatment. Only 20 of 499 patients had high levels and a reduced survival time. Plasma levels of α-fetoprotein and c-MET were associated with poor outcome (overall survival) independently of regorafenib treatment only. We identified 9 plasma miRNAs (MIR30A, MIR122, MIR125B, MIR200A, MIR374B, MIR15B, MIR107, MIR320, and MIR645) whose levels significantly associated with overall survival time with regorafenib (adjusted P ≤ .05). Functional analyses of these miRNAs indicated that their expression level associated with increased overall survival of patients with tumors of the Hoshida S3 subtype. Next-generation sequencing analyses of tumor tissues revealed 49 variants in 27 oncogenes or tumor suppressor genes. Mutations in CTNNB1 were detected in 3 of 10 progressors and VEGFA amplification in 1 of 7 responders. We identified expression patterns of plasma proteins and miRNAs that associated with increased overall survival times of patients with HCC following treatment with regorafenib in the RESORCE trial. Levels of these circulating biomarkers and genetic features of tumors might be used to identify patients with HCC most likely to respond to regorafenib. ClinicalTrials.gov number NCT01774344. NCBI GEO accession numbers: mRNA data (NanoString): GSE119220; miRNA data (Exiqon): GSE119221