332 Novel TGF-β signatures in metastatic colorectal cancer patients treated with vactosertib in combination with pembrolizumab

Background Dual inhibition of transforming growth factor beta (TGF-β) signaling and PD-1 is a promising strategy to reverse immunosuppressive tumor microenvironment and poor responses to immunotherapy. Based on preliminary clinical data with vactosertib, a highly selective and potent inhibitor of TGF-β receptor type 1, in combination with pembrolizumab, this study aimed to explore a biomarker with predictive value for this regimen in metastatic microsatellite stable (MSS) colorectal cancer (CRC). Methods Tumor biopsy samples were obtained from 24 CRC patients at baseline and cycle 2 in the ongoing MP-VAC-204 study and analyzed by RNAseq and DNAseq. Consensus molecular subtype (CMS), TGF-β responsive gene signatures, IFN-γ signatures, and tumor mutation burden (TMB) were analyzed. Clinically benefited patients were defined by those who achieved objective response assessed by RECIST v1.1/iRECIST or progression free survival more than 24 weeks. Vactosertib responsive gene signature (VRGS) that showed significantly different expression among previously identified TGF-β responsive gene signature and IFN-γ signature in responders than in non-responders was identified and VRGS score was calculated by a mean value of VRGS filtered-in gene expressions divided by 6 house-keeping gene expressions. Results As of July 1, 2020, of the total evaluable 24 patients, 71% were CMS4 subtype and 33% were with high TMB (≥10 mut/Mb). Clinical benefit rate was 33.3% including 3 PR and 1 iPR patients. No significant associations in response rate were observed with CMS subtypes or TMB status. VRGS score was significantly enriched in responders than in non-responders (P value = 0.006; AUC = 0.836). A preliminary cut-off value of 2.179 resulted in 94% specificity and 75% sensitivity with 85.7% patients correctly classifying as a responder. After treatment of vactosertib plus pembrolizumab, TGF-β-related VRGS was significantly decreased and the extent of decrease was greater in responders, compared to non-responders. Ethics Approval The study was approved by Ethics Board of Asan Medical Center, Yonsei University College of Medicine, Samsung Medical Center, and Seoul National University Bundang Hospital with approval number 2018-1215, 4-2018-0728, SMC 2018-07-146-006, and B-1808/487-003, respectively. Conclusions Development of VRGS as a predictive biomarker for this combination treatment with vactosertib and pembrolizumab is ongoing and its potential clinical utility for patient selection will be explored. Trial Registration NCT03724851