Current classification of colorectal cancer (CRC) is based on TNM staging which does not account for considerable within stage variability in clinical outcome. Currently, there is no immune-based classification of colon cancer although data suggest that intratumoral T-cell densities can provide prognostic information that is more robust than that has been provided by genomic or transcriptomic profiling. A subset of CRCs show microsatellite instability (MSI) that is a predictive biomarker for immune checkpoint inhibitors whose indication in CRC is limited to metastatic tumors. In a prior study, we compared the density and localization of CD3+ and cytotoxic CD8+ T-cell densities in non-metastatic colon cancers with deficient (dMMR) vs proficient (pMMR) DNA mismatch repair (MMR). CD3+ at the tumor invasive margin (IM) was most strongly associated with patient overall survival, and dichotomization of CD3+ prognostically stratified both dMMR and pMMR tumors. The ability to stratify dMMR tumors was a novel observation and revealed a poorer prognosis dMMR group. By multivariable (MV) analysis, CD3+ density and N (nodal) stage contributed the most to patient survival. In separate studies, expression of FOXP3+ Tregs in CRC has paradoxically been associated with more favorable survival. Interestingly, we found that colon cancers with low FOXP3+ and low CD8+ are a very poor prognostic subset. Immunoscore (IS; HalioDx) is a validated assay that examines CD3+ and CD8+ using digital pathology with conversion of densities into percentile scores using a software algorithm. In patients with stage III colon cancers from a completed adjuvant trial, recurrence rates were inversely related to IS score (0-4), and the dichotomized IS was associated with significantly better survival across all disease subgroups (Forest plot). Stage III colon cancers are categorized into low risk (T1-3N1) and high risk (T4 and/or N2) groups for guiding the duration of adjuvant chemotherapy (3 vs 6 months). IS was shown to prognostically stratify patients within both risk groups and was only a significant variable in low risk tumors by MV analysis. These data suggest the potential of IS to further risk stratify patients and to potentially inform treatment decisions. In patients with metastatic CRC, analysis of IS was shown to predict tumor metastatic burden which had prognostic value.

Data also suggest that potential for IS to provide predictive information for chemotherapy (fluoropyrimidine + oxaliplatin) response in CRC, yet further studies are needed. A potential mechanism is immunogenic cell death induced by some anti-cancer drugs, including oxaliplatin. Regarding immune checkpoint inhibitors (ICPs), only 40-50% of patients with metastatic CRC respond to such therapy, and small studies in suggest that TIL densities may be a factor in observed variability in response to ICP inhibitors. To further explore this issue, we are examining tumor immunophenotypes in relationship to response and outcome in CRC patients treated with anti-PD-1 antibodies. We aim to develop a MV model, and to validate it in an independent dataset. Important and relevant issues include the fact that only a fraction of mutations result in neoantigens recognized by T-cells, the potential for poor presentation of neoantigens by MHC1, as well as activating or suppressive cytokines.

Citation Format: Frank A. Sinicrope. Analysis of immune contexture to refine prognosis in patients with colon cancer: Implications for immunotherapy [abstract]. In: Proceedings of the American Association for Cancer Research Annual Meeting 2021; 2021 Apr 10-15 and May 17-21. Philadelphia (PA): AACR; Cancer Res 2021;81(13_Suppl):Abstract nr SY17-01.