Jonathan Chen, MD, PhD, on NSCLC: Predicting Response to Immunotherapy

In our previous study of human MSI and MSS colorectal cancer, we use single cell RNA sequencing to infer a network of CXCR3 ligands and interfering gamma positive T cells. Then using multiplex RNA ISH, we were able to stain for these components and identified a multicellular network comprised of these CXCR3 ligands, CXCL10 and 11 that spatially co-localized with interfering gamma positive cells. We called this multicellular network the Immunity Hub. So our question leading into this study was, are these multicellular hubs, these immunity hubs, important and what is their biology? We built a cohort of 68 pre PD1 immunotherapy human non-small cell lung carcinoma cases, 20 of which went on to respond to immunotherapy. We then used the same multiplex RNA ISH panel to verify the association of immunity hubs with subsequent response to PD1 therapy. We then built a second multiplex antibody panel designed around CD8 T-cell state. We stained a serial section and then co-registered the RNA ISH panel with the antibody panel. We then performed a clustering analysis of the spatial data based upon the phenotypic composition of the immunity hubs. This revealed an especially favorable subclass of immunity hub that we call the Hybrid Hubs, which feature enrichment of CD8, TCF7 double positives stem-like T-cells. These CD8 TCF7 rich hybrid hubs showed variable B-cell content and were not clearly associated with B-cell follicles or germinal centers. The hybrid hubs were found at sites where both CXCL10 11, and CCL19 were found. Hence, the name hybrid hubs. We think hybrid hubs may bring together the CXCR3 network, which is associated with activated T-cells and the CCR7 network, which is more associated with stem-like cells. Additionally, we found hybrid hubs enriched for macrophages that express CXCR3 ligands. These hybrid hubs are morphologically reminiscent of the interfollicular zone of secondary lymphoid organs. We think that hybrid hubs might serve as a reservoir of stem-like antitumor T-cells, and this might partially explain why they're associated with favorable response to PD1 immunotherapy.