Analysis of consensus molecular subtypes (CMS) classification in the TRICOLORE trial: A randomized phase III trial of S-1 and irinotecan (IRI) plus bevacizumab (Bmab) versus mFOLFOX6 or CapeOX plus Bmab as first-line treatment for metastatic colorectal cancer (mCRC).

Abstract

169 Background: The TRICOLORE trial previously demonstrated that S-1+IRI+Bmab (Arm B) was non-inferior to mFOLFOX6/CapeOX+Bmab (Arm A) in terms of progression-free survival (PFS) as first-line treatment for mCRC, irrespective of RAS status. CMS was reported to be a prognostic factor for mCRC. Previously, it was reported that the CMS could be a predictive factor for the efficacy of oxaliplatin (OX) and IRI (Okita A, et al. Oncotarget. 2018). In this exploratory study of TRICOLORE trial, we examined if the CMS could be a predictive factor of OX and IRI regimen. Methods: We collected formalin-fixed paraffin-embedded (FFPE) primary tumor tissues and performed gene expression analysis by microarray. The subtype of CMS was determined by the CMS classifier (Guinney et al. Nature Med. 2015). PFS and overall survival (OS) were compared between arm A and B in each subtypes of CMS. Results: Total of 308 FFPE samples from 487 cases who were enrolled in the TRICOLORE study were collected. Number of patients classified into CMS1 to CMS4 were 47, 72, 99 and 90, respectively. Similar to the previous reports, the proportion of RAS mutant cases was highest in CMS3 and the most of BRAF mutant cases were classified into CMS1. The median PFS (mPFS) of arm A and B were almost same in CMS2 (16.1m vs. 17.6m, HR = 1.10, p = 0.79) and CMS3 (10.6 vs. 11.4, HR = 0.95, p = 0.81). In CMS1, mPFS of arm A was worse than that of arm B, although not statistically significant (7.4m vs. 13.2m, HR = 0.73, p = 0.35). In CMS4, mPFS of arm B was better than that of arm A, although not statistically significant again (9.7m vs. 14.1m, HR = 0.72, p = 0.15). The median OS (mOS) of both arm in CMS2-CMS4 and arm B in CMS1 were good and were 29.0m or more. On the other hand, mOS of arm A in CMS1 (18.1m) was worse than the others. Conclusions: From these results, it was not concluded that CMS was a predictor of mFOLFOX6/CapeOX+Bmab and S-1+IRI+Bmab. However, mPFS and mOS of mFOLFOX6/CapeOX+Bmab in CMS1 were poor unlike other CMS subtypes, and these were consistent with a previous report. Whereas S-1+IRI+Bmab was effective in all subtypes of CMS. Clinical trial information: UMIN000007834.