1389 Phenoplex™ spatial analysis of whole-slide colon adenocarcinoma imaged with Orion™ 13-plex one-round staining and imaging

<h3>Background</h3> The ability to image tissue microenvironments (TME) at high-plex over an entire slide without requiring multiple staining steps allows for unprecedented insights into tissue architecture and the molecular mechanisms of immune and disease processes. Here we investigate a whole slide tissue section of high-grade colon adenocarcinoma, a hot tumor that contains prominent clusters of PD-L1 expression scattered throughout, using single-step high-plex staining and imaging at single-cell resolution followed by the analysis of single-cell phenotypes, tissue segmentation and spatial proximity and nearest neighbor analysis. <h3>Methods</h3> A standard pathology slide was stained in one round with a 13-plex immune-oncology panel then a 21.7 x 18.5 mm area imaged in one round with the Orion™ spatial biology platform (RareCyte). The panel consisted of Hoechst/nucleus, autofluorescence, Ki-67, CD3e, CD4, CD163, CD8a, PD-L1, PD-1, FOXP3, CD68, CD20, Pan-CK, and Vimentin. Analysis was performed on the resulting OME.TIFF file with Phenoplex™ (Visiopharm) using the following pipeline: tissue segmentation into epithelial, necrotic and stromal regions of interest; cell segmentation using the Visiopharm Blueprint APP for DAPI nuclear detection; and phenotyping and spatial biodistribution analysis using the Guided Highplex Workflow. <h3>Results</h3> Data revealed intricate architecture and cellular transitions as tumor cells express varying levels of PAN-CK, with a subset proliferating, and PD-L1 clusters within the lamina propria. Leukocyte clusters are seen interacting with PD-L1+ macrophages and, at higher zoom level, the proximity can be measured of infiltrating T and B cells near the tumor cells. A variety of spatial distance-metric analyses were performed. As an example, neighbor analysis within the stromal ROI showed that PD-L1+ macrophages had a different neighborhood of immune cells within their vicinity compared to PD-L1- macrophages. <h3>Conclusions</h3> These data highlight the importance of sufficient plex, resolution and whole slide context to derive reliable spatial biomarkers of potential prognostic value. The ability to generate such rich data in just hours across a whole section with multiplexed biomarker quantitation makes this approach suited to multi-patient clinical studies. Deep-learning based image analysis enables deeper insights into the TME of cancer samples and enables researchers to robustly and accurately quantify and classify tissue samples to understand biology of disease.