A transcriptomic signature to predict immune checkpoint inhibition efficacy in proficient mismatch repair metastatic colorectal cancer: Translational analysis of the AVETRIC and TRIPLETE studies.

Abstract

3570 Background: ICIs provide limited benefit in proficient mismatch repair (pMMR) metastatic colorectal cancer (mCRC), highlighting the need for predictive biomarkers. We aimed to uncover transcriptomic profiles predictive of ICI efficacy, using tumor specimens from two matched cohorts of pMMR, RAS/BRAF wild-type (wt) mCRC patients (pts) enrolled in two academic clinical trials investigating first-line FOLFOXIRI plus an anti-EGFR agent ± ICI. Methods: The phase II AVETRIC study (NCT04513951) provided the ICI-treated cohort (mFOLFOXIRI/anti-EGFR plus avelumab), while the experimental arm of the phase III TRIPLETE study (NCT03231722) served as the ICI-untreated control cohort (mFOLFOXIRI/anti-EGFR). Bulk RNA-seq analysis was performed using the Signomax platform on naive tumor specimens of 39 and 32 pts from the ICI and control cohorts, respectively. Firstly, gene set variation (GSV) analyses compared extended PFS (e-PFS) versus limited PFS (l-PFS) groups (using median PFS of the control cohort as a cut-off) within each cohort, leading to a focus on immune- and EGFR-related pathways. Secondly, using the median immune- and EGFR-related pathway scores in the ICI cohort as a cut-off, samples were classified combining both pathway activities into an immune high -EGFR high group versus a pooled group comprising all the other cases (immune-EGFR others ). Associations with PFS and OS were assessed by treatment cohort. Results: Thirty-four (87%) (23 e-PFS, 11 l-PFS) and 26 (81%) (13 e-PFS, 13 l-PFS) samples were successfully analyzed in the ICI and control cohorts, respectively. In the ICI cohort, l-PFS samples showed lower activity of both immune- and EGFR-related pathways than e-PFS ones, while no significant differences in the activation of the same pathways were observed between PFS groups in the control cohort. Additionally, 29 pts were classified as immune high -EGFR high (13 [45%] and 16 [55%] in the ICI and control cohorts, respectively) and 31 pts as immune-EGFR others (21 [68%] and 10 [32%] in the ICI and control cohorts, respectively). Immune high -EGFR high pts reported longer PFS (HR: 0.57, 95% CI 0.33-0.98, P =0.04) and OS (HR: 0.66, 95% CI 0.36-1.24, P =0.20) than other pts. Borderline significant interaction between treatment and immune/EGFR signature was found in PFS ( P =0.145) with ICI benefit restricted to pts bearing the immune high -EGFR high signature (HR PFS: 0.56, 95% CI 0.25-1.24, versus 1.25, 95% CI 0.60-2.64 in the others). Conclusions: Exploratory transcriptomic profiling may identify a subset of pMMR RAS/BRAF wt mCRC with concurrent immune activation and EGFR dependency that could benefit from combining ICI to an intensified anti-EGFR-based first-line treatment.