Analysis of the contribution of macrophages to the overall tumor PD-L1 microenvironment using a screening multi-tumor tissue microarray and rapid multiplex fluorescence digital phenotyping approach.

Abstract

2620 Background: Programmed death ligand-1 (PD-L1) contributes to immune suppression in the tumor microenvironment (TME) by interacting with programmed cell death-1 (PD-1) on infiltrating T lymphocytes leading to tumor immune escape. Application of omics technologies has shed light on the relevance of the TME for response to immunotherapies and development of novel treatment options. Here we have applied a multi-plex immunofluorescence/multi-tumor tissue microarray (TMA) approach to examine the immunobiology of different TMEs with respect to the contribution by tumor cells and macrophages to overall PD-L1 expression. Methods: A TMA comprising 11 tumor types and a total of 144 different donors, each represented by two cores [1mm; 1 from invasive margin (IM) and 1 from tumor center (TC)], was stained using Ultivue’s Immuno8 FixVUE panel (CD3, CD4, CD8, FOXP3, CD68, PD-1, PD-L1, pan-CK/SOX10). Whole slide images from two rounds of imaging (x20 magnification; four markers in each round) were aligned using Ultivue's UltiStacker software based on the nuclear counterstains from the two imaging rounds, to provide precise marker colocalization data. Cell phenotype data for each core was generated using Visiopharm software. Results: Overall PD-L1 positivity was greatest for NSCLC, SCLC, TNBC and gastric cancer ranging from approximately 600-1000 PD-L1+ cells/mm2/core, compared with CRC, breast, pancreatic, liver, and gastric esophageal junction (GOJ) cancers (approx. 50-300 PD-L1+ cells/mm2/core). Contribution by macrophages to overall PD-L1 expression (dual CD68+/PD-L1+) varied by tumor type representing 25-35% for NSCLC (SCC and ADC) and gastric cancer, whereas a converse pattern was apparent for SCLC, TNBC, breast (ER+ and Her2+) and pancreatic cancers where PD-L1+ macrophages accounted for a large proportion (approx. 60-85%) of overall tumor PD-L1 expression. Interestingly, as a proportion of total macrophage infiltration, approximately 65% of CD68+ macrophages were PD-L1+ for SCLC and TNBC, compared with less than 10% for pancreatic and liver cancer. Spatial analysis revealed PD-L1+ macrophage infiltration to be generally higher for IM versus TC, and the distribution between tumor (CK+) and stroma (CK-) remained similar despite exclusion of CD3+/CD8+ cytotoxic T cells from CK+ tumor regions for pancreatic and liver cancer. Conclusions: Taken together, these data illustrate the benefits of combining multiplexed immunofluorescence staining, with digital analysis of cell phenotypes within well characterized tumor samples to better understand the relative immunobiology of different TMEs. Here we demonstrate that the relative contribution of macrophages to the overall PD-L1 microenvironment varies between tumor types, which may help guide options for successful immunotherapy strategies.