Myeloid-T Cell Interplay and Cell State Transitions Associated with Checkpoint Inhibitor Response in Melanoma

Background: The treatment of melanoma, the deadliest form of skin cancer, has greatly benefited from immunotherapy. However, many patients do not show a durable response, which is only partially explained by known resistance mechanisms. Methods: We performed single-cell RNA sequencing of tumor immune infiltrates and matched peripheral blood mononuclear cells of 22 checkpoint inhibitor (CPI)-naive stage III-IV metastatic melanoma patients. After sample collection, the same patients received CPI treatment, and their response was assessed. Findings: CPI responders showed high levels of classical monocytes in peripheral blood, which preferentially transitioned toward CXCL9 expressing macrophages in tumors. Trajectories of tumor-infiltrating CD8 + T cells diverged at the level of effector memory/stem-like T cells, with non-responder cells progressing into a state characterized by cellular stress and apoptosis-related gene expression. Consistently, predicted non-responder-enriched myeloid-T/natural killer cell interactions were primarily immunosuppressive, while responder-enriched interactions were supportive of T cell priming and effector function. Conclusions: Our study illustrates that the tumor immune microenvironment prior to CPI treatment can be indicative of response. In perspective, modulating the myeloid and/or effector cell compartment by altering the described cell interactions and transitions could improve immunotherapy response. Funding: This research was funded by Roche Pharma Research and Early Development.