Multiplexed imaging mass cytometry reveals distinct tumor-immune microenvironments linked to immunotherapy responses in melanoma

Background Single-cell technologies have enabled extensive analysis of complex immune composition, phenotype and interactions within tumor, which is crucial in understanding the mechanisms behind cancer progression and treatment resistance. Unfortunately, knowledge on cell phenotypes and their spatial interactions has only had limited impact on the pathological stratification of patients in the clinic so far. We explore the relationship between different tumor environments (TMEs) and response to immunotherapy by deciphering the composition and spatial relationships of different cell types. Methods Here we used imaging mass cytometry to simultaneously quantify 35 proteins in a spatially resolved manner on tumor tissues from 26 melanoma patients receiving antiprogrammed cell death-1 (anti-PD-1) therapy. Using unsupervised clustering, we profiled 662,266 single cells to identify lymphocytes, myeloid derived monocytes, stromal and tumor cells, and characterized TME of different melanomas. Results Combined single-cell and spatial analysis reveals highly dynamic TMEs that are characterized with variable tumor and immune cell phenotypes and their spatial organizations in melanomas, and many of these multicellular features are associated with response to anti-PD-1 therapy. We further identify six distinct TME archetypes based on their multicellular compositions, and find that patients with different TME archetypes responded differently to anti-PD-1 therapy. Finally, we find that classifying patients based on the gene expression signature derived from TME archetypes predicts anti-PD-1 therapy response across multiple validation cohorts. Conclusions Our results demonstrate the utility of multiplex proteomic imaging technologies in studying complex molecular events in a spatially resolved manner for the development of new strategies for patient stratification and treatment outcome prediction. Plain language summary Immunotherapies help the immune system to fight cancer. However, they only benefit a subset of melanoma patients, and currently no single marker is sufficient to determine which patients will respond to these treatments. Here, we use imaging mass cytometry, a technique to measure the levels of multiple markers in individual cells, to analyze tumor tissue from melanoma patients receiving immunotherapy. By determining the different cell types present and the spatial relationships between them, we identify six distinct melanoma cellular environments that are associated with different clinical responses to immunotherapy. Our results demonstrate how complex information about the spatial relationships of cell types can be integrated to help to identify patients that might benefit from immunotherapy.