Abstract 2082: Spatial deconvolution of non-small-cell lung cancer tissues prior to immune checkpoint inhibitor therapy

Abstract Introduction. Immune checkpoint immunotherapy (ICI) has shaped disease management for patients with non-small cell lung cancer (NSCLC). The blockade of the interaction between PD-1 and PD-L1 elicits pro-immunogenic tumour clearance in a subset of patients, however current methods to stratify patients likely to benefit from ICIs is lacking. This study leverages full tissue sections from first-line nivolumab treatment in stage IV or recurrent NSCLC to understand the properties of tissues from patients that benefit from ICI compared to those with refractory disease. Methods. Pre-treatment tissues from patients that received first-line nivolumab in advanced stage setting were processed by high-plex immunofluorescence on the Akoya PhenoCycler platform. Patient response to ICI was grouped by RECIST 1.1 category (CR=1, PR=4, PD=4, SD=1) Supervised clustering was performed to phenotype broad cell lineages, and gaussian mixture models were fit to assign positive expression of functional markers. Spatial K-means clustering on tumour or non-tumour cells allowed tumour, non-tumour, and tumour-interface regions to be defined. Additionally, contours of tumour density were mapped as discrete regions. Cell frequencies and cell-cell interactions were considered in each of these spatial compartments to delineate the features that aligned with clinical outcomes. Results. The staining panel allowed for deep phenotyping of 3.3x106 tumour cells and 3.4x106 non-tumour cells. Of immune cell lineages, we identify the most abundant macrophage populations express discrete mixtures of LAMP-1, HLA-DR, PD-L1, MMP9 and VISTA. CD8 T cells most commonly express granzyme-B and PD-1, with discrete populations of LAMP-1+/HLA-DR+, while CD4 T cells and Tregs express discrete mixtures of PD-1 and ICOS. Interestingly, both CD8 and CD4 T cells display activated CD38+/IFNg+ populations. Spatially, we observe that higher vasculature as well as macrophages positive for LAMP-1 and HLA-DR in both tumour and tumour-interface regions are associated with refractory disease. Additionally, activated CD38+ CD4 cells are enriched in the stroma of ICI responsive patients. Higher levels of tumour cells positive for both PD-L1 and FOXP3 were also associated with poorer ICI response. Conclusion. Our study showcases deep cellular phenotyping of large tissue specimens and utilises unbiased and automated methods to identify tissue regions for which to measure cellular features. We identify a number of properties that define the characteristics of NSCLC tissue in relation to therapy outcomes. Citation Format: James Monkman, Chin Wee Tan, Rafael Tubelleza, Ken OByrne, Arutha Kulasinghe. Spatial deconvolution of non-small-cell lung cancer tissues prior to immune checkpoint inhibitor therapy [abstract]. In: Proceedings of the American Association for Cancer Research Annual Meeting 2025; Part 1 (Regular Abstracts); 2025 Apr 25-30; Chicago, IL. Philadelphia (PA): AACR; Cancer Res 2025;85(8_Suppl_1):Abstract nr 2082.