Abstract
Expounding the heterogeneity for ovarian cancer (OC) with the cognition in developmental biology might be helpful to search for robust prognostic markers and effective treatments. In the present study, we employed single-cell RNA-seq with ovarian cancers, normal ovary, and embryo tissue to explore their heterogeneity. Then the differentiation process of clusters was explored; the pivotal cluster and markers were identified. Furthermore, the consensus clustering algorithm was used to explore the different clinical phenotypes in OC. At last, a prognostic model was construct and used to assess the prognosis for OCs. As a result, eight diverse clusters were identified, and the similarity existed in some clusters between embryo and tumours based on their gene expression. Meaningfully, a subtype of malignant epithelial cluster, PEG10+ EME, was associated with poor survival and was an intermediate stage of embryo to tumour. PEG10 was a CSC marker and might influence CSC self-renewal and promote cisplatin resistance via NOTCH pathway. Utilising specific gene profiles of PEG10+ EME based on public data sets, four phenotypes with different survival and clinical response to anti-PD-1/PD-L1 immunotherapy were identified. These insights allowed for the investigation of single-cell transcriptome of OCs and embryo, which advanced our current understanding of OC pathogenesis and resulted in promising therapeutic strategies.