Dissecting the immune landscape of cancer to understand immunotherapy response

The emergence of immune checkpoint blockade underscores the inextricable link between the immune system and cancer. However, its efficacy is limited to certain cancer types, and only a minority of patients experience durable responses, highlighting the complex and tightly regulated interaction between cancer and the immune system. Furthermore, the immune system and cancer do not interact in isolation; they coexist within the complex cellular ecosystem of the tumor microenvironment. To understand the mechanisms driving response to immune checkpoint blockade, it is crucial to consider the dynamic relationships between the various cell types within the tumor microenvironment. This thesis begins by analyzing bulk RNA-seq data from metastatic melanoma patients treated with first-line immune checkpoint blockade. To overcome the inherent limitations of bulk RNA-seq, a deconvolution algorithm is employed to identify cancer cell-specific expression changes associated with durable responses. The cancer cells in durable responders downregulate interferon gamma compared to patients that develop acquired resistance, but maintain higher levels of MHC class II activity. Additionally, acquired resistance is linked to an elevated tumor mutational burden, which can be attributed to a concurrent increase in intra-tumor heterogeneity. Next, this thesis explores single-cell RNA-seq data from metastatic melanoma patients treated with first-line immune checkpoint blockade. To overcome the challenges of sparse and shallow sequencing inherent to single-cell RNA-seq, transcription factordirected coexpression networks (regulons) are inferred and used to characterize immune cell states associated with treatment response. These regulons demonstrate utility in deconvolving bulk RNA-seq data and in characterizing treatment response across four independent cohorts. Finally, this thesis analyzes matched immunohistochemistry of tumor-infiltrating lymphocytes and global proteomics of whole tumor sections to identify mechanisms of disease progression and immune escape in high-grade serous ovarian cancer. During disease progression, tumor sites with increased immune infiltration exploit SNX8 as a
molecular switch for non-canonical interferon-gamma signaling, which recruits regulatory T cells through CXCL9/10, leading to immune suppression. The exclusion of CD4+ T cells from the tumor epithelium shows a stronger correlation with disease progression compared to the exclusion of CD8+ T cells, which are predominately exhausted. Keratin-expressing cancer cells retain a function from their epithelial origin that promotes Treg infiltration via MHC class II, leading to an inflammatory phenotype characterized by limited interferon-gamma production and non-clonally expanded T cells. Overall, this thesis systematically characterizes the tumor microenvironment in metastatic melanoma and high-grade serous ovarian cancer. It identifies mechanisms that are associated with an initial and durable response to immune checkpoint blockade in melanoma, and highlights mechanisms that can bet targeted to improve immunotherapy response rates in ovarian cancer.