Abstract Ovarian cancers often exhibit high rates of recurrence and poor treatment response. Preclinical models that recapitulate the heterogeneity of human disease are critical to develop new therapeutic approaches. While patient-derived models are a powerful tool for testing various therapeutics, their dependence on immune-compromised mice is severely limiting. Syngeneic mouse models, however, allow for the generation of tumours comprising the full repertoire of non-malignant cell types. Here we have performed a comparative analysis of diverse models of high-grade serous ovarian cancer based on transcriptomic profiling of 22 cell line models, and intrabursal and intraperitoneal tumours from 12 models. Among cell lines, we identify distinct features in signalling activity, such as elevated inflammatory signalling in STOSE and OVE16 models, and MAPK/ERK signalling in ID8 and OVE4 models; metabolic features, such as predicted reduction in glycolysis associated with subsets of engineered ID8 subclones; and relevant functional properties, including differences in EMT activation, PD-L1 and MHC class I expression, and predicted chemosensitivity. Finally, we evaluate variability in properties of the tumour microenvironment among models. We anticipate that this work will serve as a valuable resource, providing new insight to help in the selection of models for specific experimental objectives.
