Abstract Background and Aims: Ovarian cancer is the most common cause of death from gynaecological cancer worldwide (1). Often called the “silent killer”, the high mortality rate is associated with the late stage in which the disease is diagnosed (2). Unlike other solid cancers, ovarian cancer cells rarely spread via the vasculature, instead shed from the primary tumor and disseminate within the peritoneal fluid or ascites (2). The formation of spheroids within the non-adherent peritoneal environment is a barrier to effective treatment, due to the enhanced ability of spheroids to survive chemotherapies and seed distal metastases (3-5). Invading ovarian cancer cells interact with mesothelial cells lining the peritoneal surface, attach to, and invade the underlying basement membrane to establish secondary lesions (6). The molecular events at the interface between invading ovarian cancer and peritoneal cells at the onset of invasion have been poorly studied, existing methods capture events after invasion is well-underway. Therefore, the regulatory signals which govern the initiation of a metastatic lesion are poorly understood. We aimed: i) to develop an innovative method to co-culture ascites-derived primary ovarian cancer cells to measure invasion through peritoneal mesothelial cell layers in real time (7); ii) identify the unique protein expression profile at the ovarian cancer spheroid-peritoneal interface during early invasion using proteomics. Methods: Real Time Cell Analysis (RTCA) technology (xCELLigence, ACEA) was adapted to establish a three-dimensional co-culture model of the tumor microenvironment of the peritoneum (7). RTCA wells were coated with Matrigel (basement membrane matrix) and a confluent monolayer of human mesothelial cells. Multicellular spheroids were generated from primary ovarian cancer lines (high-grade serous carcinoma, leiomysarcoma, adenocarcinoma, carcinosarcoma) or benign ascites derived cells cultured under non-adherent conditions. Spheroids were added to the RTCA plate and real time measurements were conducted to determine the precise period when invasion commenced. Parallel cultures were histologically-preserved at invasion initiation and processed for assessment by Matrix-Assisted Laser Desorption/Ionization (MALDI) imaging. Results: We assessed invasion in a panel of ovarian cancer samples, benign ascites cells and the mesothelial cell line. Continuous real time measurements revealed that individual ovarian cancer samples exhibited differences in the timing of the onset of invasion. High-grade serous carcinoma samples initiated invasion rapidly after addition, with some samples detected in under 30 min, a time period not captured by traditional Transwell assays. The benign and mesothelial cells themselves did not invade. MALDI imaging allowed for the identification of peptides expressed at the tumor-mesothelial interface during early invasion. These molecules represent novel molecules involved in the initiation of the metastatic process and are the subject of further characterisation. Conclusion: This method represents a high-throughput quantitative analysis of ovarian cancer spheroid invasion of mesothelial and ECM barriers. The study has provided an unprecedented insight into the molecular mechanisms and molecules involved in the initiation of ovarian cancer invasion of the mesothelium, identifying new therapeutic targets.