Fig. 4

a-b The immune checkpoint molecules HAVCR2, IDO1, IDO2, ADORA2A, LAG3, TIGIT, VISTA, VTCN1, PD-1, PDCD1LG2 (PD-L2), CD274 (PD-L1) and CTLA-4 are significantly higher in MSI colorectal adenocarcinomas, compared to their MSS counterparts. c-d The total count of classically defined neoepitopes (CDNs) and alternatively defined neoepitopes (ADNs) per individual COAD tumor was significantly correlated with the levels of cytolytic activity (CYT). e-f The number of MHC class I and II CDN and ADN was significantly higher in CYT-high COAD tumors. g-h High levels of cytolytic activity (CYT-high) were significantly correlated with MHC-I CDNs (p = 0.047, Pearson’s rho = 0.2) and with MHC-II ADNs (p = 6.9e-05, Pearson’s rho = 0.39) in COAD tumors. i-j Pairwise correlation between the cytolytic index (log-average of GZMA and PRF1) versus individual genes of immune suppression index (log-average of VTCN1, VISTA, TIGIT, PD1, LAG3, ADORA2A, IDO1/2, CTLA-4, PDCD1LG2 (PD-L2), CD274 (PD-L1), and HAVCR2) in COAD and READ, respectively. The Pearson’s rho (R) and statistical significance (p-value) are indicated in each graph. Loess regression (blue line) was used to diminish the noise of the variables during correlation analysis. k Kaplan-Meier curves depict the overall survival of COAD and READ patients after synergistic analysis for CTLA-4 and PD-L1 in these tumors. Both individual and simultaneous high levels of CTLA-4 and PD-L1 had a positive (though not statistically significant) effect on the patients' overall survival. On the reverse, simultaneous low expression of both genes led to a significant shift towards negative effect versus all other patients