A novel immune-based score and its immune contexture exploration in colorectal cancer.

Abstract

234 Background: The majority of currently available biomarkers in colorectal cancer (CRC) perform poorly with respect to prognostication and prediction of response. Immunoscore (IS) has been proposed as a signature of host immune cancer response reflecting a favorable prognosis. Next, through AtezoTRIBE, ‘Immunoscore immune checkpoint’ (IS-IC) was introduced, indicating its potential predictive value for immune checkpoint blockade (ICB) in pMMR metastatic CRC. IS-IC is computed through quantitative and spatial variables related to density and proximity of CD8 cells and programmed cell death-ligand 1 (PD-L1) cells in the tumor core (TC). There are little confirmatory studies. In this work we test the application of IS-IC through multiplex immunofluorescence (mIF) on our own cohort of patients. Next we explore immune contexture differences through subgroup analysis. Methods: 15 prospectively collected, treatment naïve, primary CRC samples (9 dMMR/MSI-H, 6 pMMR/MSS) where analyzed with mIF. Cell density, proximity between cells, cell clustering and spatial distribution of selected mIF biomarkers (CD3, CD20, CD11c, CD163, CD15, CD4, CD8, Foxp3, CD56, Granzyme B, PD-L1, programmed cell death protein 1 (PD-1), pan-cytokeratin (CK) and Ki67) were analyzed both at the invasive margin (IM) and TC using digital pathology. All slides subjected to digital image analysis met predefined quality control criteria. Results: Serial mIF showed PD-L1 was enriched in stromal cells (CK - ) within IM ( p =0.0148) and confirmed its co-localization with macrophages (CD163 + ) ( p =0.0092). Hence, both TC and IM were included to quantify the densities of CD8, PD-L1 and the proximity between them, as a suggested novel IS-IC TC+IM . During exploration of the immune context we found that both IS-IC and IS-IC TC+IM high groups exhibited significantly increased infiltration of PD-1 + cells ( p =0.0074 and p =0.0058, respectively) and CD8 + PD-1 + cells ( p =0.0101 and p =0.0062, respectively), compared to the corresponding low groups. Whereas only the IS-IC TC+IM high group showed increased natural killer (NK) cells (CD56 + ) ( p =0.0192), found mainly at IM ( p =0.0167) in contrast to TC ( p =0.0824). The novel IS-IC TC+IM scoring system captures the immunosuppressive and immune responsive interaction among PD-1 + cells, PD-L1 + cells, NK cells and CD8 + cells in TC and IM from the cell numbers and cell-cell proximity. A validation cohort is underway. Conclusions: In this study we confirm the ease of replicating IS-IC and the presence of ICB sensitive stroma in the high score group. This study further explored the immune context in both TC and IM, emphasizing the importance of regional differences. Additionally it was found that NK cells were only increased at IM. Altogether we suggests adding dual quantification in TC and IM as a novel IS-IC TC+IM , which may be a more sufficient reflection of the tumor micro environment.