Abstract 2778: Comparative multiplex digital phenotyping of the tumor microenvironment across multiple tumor types using a well characterized multi-tumor tissue microarray

Abstract

Abstract The tumor microenvironment (TME) represents a complex interaction of immune and host cells and can be described broadly as ‘inflamed', ‘immune excluded' and ‘desert'; categories that may serve as predictive parameters for response to different immunotherapies. Multiplexed fluorescence and associated digital imaging platforms are important tools for defining the spatial components of the TME and can accelerate development of novel immunotherapy strategies. To investigate the TME across multiple tumor types simultaneously we have utilized a multi-tumor tissue microarray (TMA) covering 11 different tumor types comprising a total of 144 cases, each in duplicate with cores (1mm) taken from invasive margin (IM) and tumor center (TC). The TMA was stained using the UltiMapper I/O Immuno8 panel, which includes markers for CD3, CD4, CD8, FOXP3, CD68, PD-1, PD-L1, and a pan-CK/SOX10 cocktail as a tumor indicator. Stained TMAs were scanned on a fluorescence whole slide scanner (20X magnification) and marker images aligned using the UltiStacker software using the nuclear counterstain images as references from two rounds of imaging, to provide accurate marker colocalization data. Digital image analysis was performed using Visiopharm software to generate cell phenotype data, including localization to tumor and stroma, for each of the 288 cores. Comparative analysis of all tumor types regardless of core location (IM and TC combined) revealed gastric cancer, CRC, TNBC and NSCLC to be most highly infiltrated with CD3+ T cells, and ER+ and Her2+ breast cancers least infiltrated. The contribution of cytotoxic T cells (CD3+ CD8+) to the TME was greater for gastric cancer (approx. 50%) compared with NSCLC (approx. 20%) and the proportion of cytotoxic T cells activated/exhausted as defined by expression of PD-1 ranged from approx. 25% (eg. TNBC, CRC) to 1% (eg. ER+ BC). Spatial localization analyses of cytotoxic T cells revealed pancreatic and hepatic cancers as the most ‘excluded' as evidenced by low CD3+/CD8+ T cell numbers in tumor (CK+) compared with stromal regions (CK-). Interestingly, cytotoxic T cells with an exhausted phenotype (CD3+ CD8+ PD-1+) were preferentially infiltrated within CK+ tumor areas compared with stroma (CK-) for inflamed tumors such as gastric, TNBC and CRC. These data highlight the benefits of combining a fast, multiplexed staining protocol, with a detailed digital analysis of immune phenotypes spatially within well characterized tumor samples to better understand the TME (eg. inflamed, excluded) and identify tumor types more likely to benefit from different immunotherapeutic strategies. Citation Format: Marie Cumberbatch, Douglas Wood, Christopher Womack, Milan Bhagat, Mael Manesse, Alison Bigley, Lorcan Sherry. Comparative multiplex digital phenotyping of the tumor microenvironment across multiple tumor types using a well characterized multi-tumor tissue microarray [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 2778.