UCLA

Gynecologic cancers remain among the most lethal malignancies affecting women, and the most aggressive among them share a common clinical trajectory of initial treatment response followed by recurrence, therapy resistance, and immune escape. High-grade serous ovarian carcinoma and gynecologic carcinosarcoma sit at increasingly aggressive points along this trajectory. Both are treated with the same platinum- and taxane-based chemotherapies, both recur as chemoresistant disease, and both frequently disable the antigen-presentation machinery on which adaptive immunity and immune checkpoint blockade depend. This dissertation applies single-cell profiling, at increasing dimensionality, to expose therapeutic vulnerabilities in these tumors that render them susceptible to innate and innate-like cytotoxic lymphocytes. In Chapter 2, we profile primary high-grade serous ovarian carcinoma samples across disease states, including matched chemonaive and recurrent pairs, by multiparameter flow cytometry. Recurrent tumors are enriched for stem-like populations, and their microenvironment becomes progressively more immunosuppressive. We find that CD1d expression on immunosuppressive tumor-associated macrophages provides a handle through which invariant natural killer T cells can engage both the suppressive microenvironment and the tumor cells themselves. In Chapter 3, we escalate from flow cytometry to transcriptome-wide single-cell analysis of primary gynecologic carcinosarcoma, assembling a comprehensive portrait of tumor and microenvironment. We identify a cancer stem cell-like population that appears to orchestrate an immunosuppressive signaling network, and we identify therapeutic vulnerabilities based on surface expression of the tumor-associated antigen HER2 and the natural killer receptor ligands CD112 and CD155. Using primary patient samples and patient-derived xenograft models, we test NK cell therapies combined with antibody therapy in vitro and in vivo. Together, these studies apply one methodological framework to two diseases, moving from target identification through mechanism to validated therapy. They reveal that aggressive gynecologic tumor cells left unreachable by immunotherapies dependent on antigen presentation are in fact more vulnerable to effector populations that do not rely on antigen recognition at all.