Abstract
Sustained type-I and type-II interferon (IFN) signaling can drive multiple mechanisms of resistance to immune checkpoint blockade (ICB). Here, we used single-cell RNA sequencing data to characterize the effects of IFNs in the tumor-immune microenvironment (TME) of renal cell carcinoma (RCC) and then examined how IFN-driven cellular phenotypes modulate ICB efficacy. Using mixed-effects models, we inferred the IFN inducibility of putative IFN-stimulated genes (ISGs) within cell types. Genes encoding inhibitory ligands and immune checkpoints were strongly expressed and IFN inducible in macrophages but less so in RCC tumor cells. In orthogonal clinical trial cohorts, a signature of myeloid IFNγ signaling, but not tumor IFNγ signaling, predicted primary resistance to first-line ICB plus anti-VEGF therapy. Functionally, IFNγ-conditioned macrophages inhibited T cell killing of RCC tumor cells in vitro. Our inferential modeling approach offers a framework for biomarker discovery through deconvolution of cytokine signaling effects in the TME and points to myeloid cells as mediators of tumor-extrinsic, IFN-driven resistance to immunotherapy in RCC.