Ovarian cancer is the fifth most common cause of cancer-related deaths in women and the most mortal cancer type among all the gynecological cancers. Approximately 90% of the ovarian malignancies are epithelial in origin, also known as epithelial ovarian cancer (EOC). Due to the disease’s asymptomatic in early stages, more than 70% of the cases are diagnosed at advanced-stage with peritoneal metastasis. Although patients respond well to standard treatments including cytoreductive surgery and chemotherapy, the recurrence rate of advanced-stage EOC can be as high as around 80%. Consequently, the overall survival of EOC patients is very poor. Therefore, a better understanding of this disease is urgently needed.

Tumor heterogeneity and metastasis are two significant challenges in cancer treatment. The phenomenon of tumor heterogeneity occurs both between different tumors (inter-tumor heterogeneity) and within a single tumor tissue (intra-tumor heterogeneity). Intra-tumor heterogeneity implies the diversity of malignant tumor cells, as well as different types of stromal and immune cells within a tumor. Metastasis is another challenge in fighting with 90% of cancer-related deaths. Epithelial to mesenchymal transition (EMT) is suggested as the primary force for epithelial cells to undergo a series stages of invasion-metastasis cascade. The disseminated tumor cells (DTCs) are considered as metastatic “seeds” and responsible for subsequent metastasis and distant relapse. It has become more and more clear that the interaction between tumor cells and their microenvironment plays a major role in cancer progression and metastasis. Although a lot of bioinformatic computation models have been established to dissect the transcriptions of conventional RNA-seq, this bulk tumor analysis approach is not competent to discover cellular variabilities from the cell population average. Therefore, a single-cell analysis approach is needed to better understand the heterogeneity, metastasis and relapse of EOC.

In this thesis, we performed single-cell transcriptomic analysis on both metastatic “seeds” and tumor tissues collected from clinical for the understanding of heterogeneity and metastasis in EOC. The patient samples contained four types: malignant ascites, primary tumors, peritoneal metastasis tumors and relapse tumors. We first designed a targeted panel of 53 EMT signature genes for high-throughput single-cell qPCR analysis of the disseminated single cells (DSCs) and disseminated tumor cell clusters (DTCCs) in EOC ascites. Through the transcriptional analysis of 120 DSCs and 195 intra-DTCC single cells from 27 DTCCs, we uncovered the intra-cluster heterogeneity of DTCCs. The intra-DTCC heterogeneity analysis showed that cancer-associated fibroblasts (CAFs) induced EMT of tumor cells via TGF-β signaling within the DTCC microenvironment. The activation of EMT program, in particular the upregulation of ZEB2, resulted in the acquisition of additional chemo-resistance and metastasis abilities of tumor cells. We further utilized the single-cell RNA sequencing (scRNA-seq) approach to do the comparison between primary tumors and metastatic tumors. Through the transcriptional analysis of 13,369 cells from eight EOC patients, including four primary tumors, two untreated peritoneal metastasis tumors and two relapse tumors, we identified the relapse-initiation subpopulation of tumor cells in the primary tumors of EOC. The time-resolved analysis for the developmental sequence of tumor cells in metastatic tumors identified the residual DTCs responsible for the replanting. The relapse-initiation DTCs were associated with the stress program in primary tumors and conservatively express CYR61 in different EOC patients. In addition to tumor cells, a subpopulation of CAFs uniquely expressing RGS5 was rated as strongly supportive to tumor metastasis. Our studies reported in this thesis characterized the heterogeneity of EOC at single-cell resolution, providing additional insights into the understanding of metastasis and relapse of this disease.