The impact of PROgnostic SIgnatures in stage II colon cancer patients enrolled in the phase III clinical trial TOSCA: the PROSIT study

Background: Adjuvant chemotherapy (ACT) indication in stage II colorectal cancer (CRC) remains challenging, as clinical, pathological, and ctDNA inadequately predict benefit. Immune infiltration correlates with outcome, yet transcriptional signatures have not been systematically evaluated in prospective trial cohorts. We aimed to identify a prognostic transcriptional signature in high-risk stage II CRC patients treated with ACT within the TOSCA trial. Methods: From the TOSCA trial, 124 high-risk stage II patients were selected using a matched case-control design, with matching based on center, pT stage, ACT regimen, and treatment duration. Whole-transcriptome sequencing was performed on FFPE samples. MSI status was inferred bioinformatically. Prognostic associations were assessed using log-rank tests and multivariable Cox models adjusting for MSI status and tumor sidedness. Results: Seventy-two samples met RNA quality criteria. Median follow-up was 70.9 months; 40% were right-sided, 60% received FOLFOX, 54% 3-month ACT, and 19% were MSI-high. Transcriptomic analyses highlighted immune-regulatory and TGF-β/stromal programs associated with outcome. A biologically defined Immune Exclusion (IE) phenotype, characterized by low ICR and high Hallmark TGF-β signaling, was observed in approximately 28% of patients and was associated with shorter DFS (HR=2.22, p = 0.036). This association persisted after adjustment for MSI status and tumor sidedness (HR=2.47, p = 0.018). A reduced seven-gene IE signature (rIE) was subsequently derived and was also associated with shorter DFS after multivariable adjustment (HR=2.83, p = 0.012). Conclusion: The IE phenotype and its reduced rIE signature were associated with shorter DFS in high-risk stage II colon cancer and represent exploratory candidate prognostic biomarkers requiring validation in independent cohorts. Statement of translational relevance: Identifying stage II colorectal cancer (CRC) patients who will relapse despite standard adjuvant chemotherapy (ACT) remains an unmet clinical need. Current clinicopathological and ctDNA-based approaches do not fully resolve risk stratification in this setting. In this study, whole-transcriptome profiling of high-risk stage II CRC from the prospective TOSCA trial identified an Immune Exclusion (IE) signature associated with significantly shorter disease-free survival, which remained associated with DFS after adjustment for MSI status and tumor sidedness. These findings provide a biological framework for further validation in independent cohorts and for future hypothesis-generating studies aimed at refining risk stratification and exploring alternative therapeutic approaches.