开发用于疗法测试的器官型患者来源卵巢癌模型

Ovarian carcinoma (OVC) is a prevalent malignancy affecting 10.2 per 100,000 women annually. Early detection is challenging, resulting in a five-year survival rate of 50.9% and a death rate of 6 per 100,000 women. To address this, preclinical models that replicate ovarian cancer's histological and molecular characteristics are crucial. This study focused on modeling high-grade serous ovarian carcinoma, an aggressive cancer influenced by genetic and microenvironmental factors that impact tumor progression and therapy response. Here, we developed and applied a comprehensive patient-derived model using five high-grade serous ovarian carcinoma donors. Tumor sections were analyzed for the presence and distribution of stromal and immune cells, showing varying tumor-stroma ratios and immune infiltration levels. OVC cultures incorporating dissociated tumor tissue, cancer-associated fibroblasts (CAFs) and endothelial cells were established using the OrganoPlate Graft UF platform. Over time, these cells formed complex vascularized co-cultures mimicking interactions that resembled the OVC tumor microenvironment. These comprehensive patient-derived models were characterized with immunostainings, showing the presence of tumor cells, CAFs and macrophages. Cytokine and chemokine analysis indicated an immunosuppressive tumor microenvironment. The OVC models demonstrated key tumor microenvironment characteristics, such as the cellular organization of tumor clusters and the release of growth factors, indicating a stable model during the established assay window of 72 hours. Cultures were exposed to conventional chemotherapy. Both carboplatin and paclitaxel elicited varying responses across donors. Interestingly, we could also capture the specific response of the tumor clusters and compare this with the response to the general cell population. This analysis revealed that the treatments were not specifically targeting the tumor clusters, but also the supporting cell types and the vasculature. Compared to conventional 2D cell lines, the 3D model showed enhanced chemoresistance, as expected, and thus might better recapitulate the in vivo tumor behavior. [Iris Schilt and Sander P.M. de Ruiter contributed equally to this work.]