Association of Consensus Molecular Subtypes and Molecular Markers With Clinical Outcomes in Patients With Metastatic Colorectal Cancer: Biomarker Analyses From LUME-Colon 1

INTRODUCTION: LUME-Colon 1 (NCT02149108) was a global, placebo-controlled phase III study of nintedanib in advanced colorectal cancer (CRC). Pre-specified biomarker analyses investigated the association of CRC consensus molecular subtypes (CMS) and tumor genomic and circulating biomarkers with clinical outcomes. MATERIALS AND METHODS: Archival tumor tissue, cell-free DNA (cfDNA), and plasma samples were collected for genomic, transcriptomic, and proteomic analyses to investigate potential associations between CRC CMS and other biomarkers with nintedanib response and clinical outcomes. RESULTS: Of the 765 treated patients, 735, 245, and 192 patient samples were analyzed in the circulating protein, tumor tissue, and cfDNA datasets, respectively. Patients were classified as CMS1 (1.7%), CMS2 (27.7%), CMS3 (0.9%), CMS4 (51.5%), or unclassified (18.2%). Unclassified/mixed CMS was associated with longer overall survival (OS) with nintedanib vs. CMS2 or CMS4 (interaction P-value = .0086); no association was observed for CMS4. Gene expression-based pathway analysis revealed an association between vascular endothelial growth factor-related signaling and OS for nintedanib (P = .0498). The most frequently detected somatic mutations were APC (72.0% [tumor tissue] vs. 56.8% [cfDNA]), TP53 (47.1% vs. 34.9%), KRAS (40.8% vs. 28.6%), and PIK3CA (16.6% vs. 11.5%); concordance rates were > 80%. Median OS differences were observed for APC and TP53 mutations vs. wild-type in cfDNA, indicating a potential prognostic value. Circulating ANG-2, CA-9, CEACAM1, collagen-IV, IGFBP-1, ICAM-1, IL-8, and uPAR were potentially prognostic for both OS and progression-free survival. CONCLUSION: We demonstrated the feasibility of large-scale biomarker analyses and CMS classification within a global clinical trial, and identified signals suggesting a potential for greater nintedanib treatment response in the unclassified/mixed CMS subgroup, despite these tumors showing heterogeneous patterns of CMS mixtures. Our results revealed a high degree of concordance in somatic mutations between tumor tissue and cfDNA. Associations with prognosis for cfDNA somatic mutations, as well as several protein-based biomarkers, may warrant further investigation in future trials.