Epithelial ovarian cancer (EOC) is still considered the most lethal gynecological malignancy and improved early detection of ovarian cancer is crucial to improving patient prognoses. To address this need, we tested whether candidate EOC biomarkers can be identified using three-dimensional (3D) in vitro models. We quantified changes in the abundance of secreted proteins in a 3D genetic model of early-stage EOC, generated by expressing CMYC and KRAS G12V in TERT-immortalized normal ovarian epithelial cells. Cellular proteins were labeled in live cells using stable isotopic amino acid analogues, and secreted proteins identified and quantified using liquid chromatography-tandem mass spectrometry. Thirty-seven and 55 proteins were differentially expressed by CMYC and CMYC1KRAS G12V expressing cells respectively (p < 0.05; >2-fold). We evaluated expression of the top candidate biomarkers in~210 primary EOCs: CHI3L1 and FKBP4 are both expressed by >96% of primary EOCs, and FASN and API5 are expressed by 86 and 75% of cases. High expression of CHI3L1 and FKBP4 was associated with worse patient survival (p 5 0.042 and p 5 0.002, respectively). Expression of LGALS3BP was positively associated with recurrence (p 5 0.0001) and suboptimal debulking (p 5 0.018) suggesting that these proteins may be novel prognostic biomarkers. Furthermore, within early stage tumours (I/II), high expression of API5, CHI3L1 and FASN was associated with high tumour grade (p 5 3 3 10 24 , p 5 0.016, p 5 0.010, respectively). We show in vitro cell biology models of early-stage cancer development can be used to identify novel candidate biomarkers for disease, and report the identification of proteins that represent novel potential candidate diagnostic and prognostic biomarkers for this highly lethal disease.Despite recent advances in surgery and chemotherapy, epithelial ovarian cancer (EOC) remains the most lethal gynecologic malignancy, mainly due to the absence of specific symptoms and a lack of effective screening tools. The majority of patients present with advanced stage disease where the 5-year survival rate is only 27%. For advanced stage tumours, recurrence rates are over 60%, and 20% of these patients respond poorly to platinum-based chemotherapy. Therefore, detecting patients with early EOC continues to be an urgent clinical need and it remains a clinical priority to detect highgrade serous EOC early during disease development.Given the difficulties accessing the ovary for biopsy and the rapid rate at which the most aggressive EOC subtypes progress, the identification of clinical biomarkers detectable in the blood would represent a significant advance for the identification of patients with EOC. Currently the most widely used ovarian cancer biomarker is serum CA125, and this marker is particularly good at detecting disease recurrence. A second marker, HE4, was approved in 2009 for monitoring ovarian cancer progression, and can be elevated