Ex vivo tumor testing platform for predicting clinical response to platinum-based therapy in patients with high grade serous ovarian cancer.

Abstract

5563 Background: Precision medicine for cancer patients has brought effective novel therapy options over the past decades. However, treatment for high-grade serous ovarian cancer (HGSOC) is still based on platinum-containing therapy. Around 30% (Morgan 2020) of primary disease patients do not respond to this treatment, and non-response rates in residual disease are higher still. We present an ex vivo 3D tumor testing platform that predicts patient-specific response to standard of care therapy using the CA125 half-life, the primary ovarian cancer marker (Charkhchi, 2020). Methods: Patients with HGSOC, eligible for platinum-based neoadjuvant chemotherapy (NACT), were included in the study between 2019 and 2022 in the Netherlands (IRB P18.032). Clinical data was collected including CA125 levels at baseline and after 2-4 courses of NACT. Tumor clusters enriched from ascites were embedded in hydrogel and exposed to first-line (carboplatin, paclitaxel) and second-line therapies (doxorubicin, gemcitabine, topotecan and olaparib). Screening plates were imaged in a high content screening 3D platform. Morphological features were extracted after image analysis and fitted as dose-response (Hill-)curves. A Bayesian linear regression model was trained on the AUCs of carboplatin and paclitaxel to predict the CA125 half-life. The result was classified according to clinical standards: insensitive (IN, progressive and stable disease), moderate response (MR), or strong response (SR). In addition, for each second-line treatment the top 25% strongest responding samples were classified as sensitive while the bottom 25% were classified as resistant samples. Results: The model was trained using 30 patients. The correlation coefficient between the predicted and actual CA125 half-life is 0.739 (R2 = 0.55). This results in a classification accuracy of 87% (insensitive: 100% (n = 2), MR: 80% (n = 14), SR: 80% (n = 14)). The percentage of samples, per response class, responding to at least one of the alternative chemotherapies is as follows: SR: 58% (8/14), MR: 29% (4/14) and IN 50% (1/2). For olaparib the equivalent results are SR: 36% (5/14), MR: 21% (3/14) and IN: 0% (0/2). When sufficient tumor content is received, the assay has a technical success rate of 89%. Results are delivered in two weeks. Conclusions: The presented platform for ex vivo tumor testing predicts clinical response to platinum-based NACT in HGSOC patients. The same platform measured patient specific patterns of relative sensitivity to other chemotherapies. The platform enables better stratification of responders vs non-responders, and can support informed treatment decisions for first-line and second-line therapies. The value of integration of this predictive tool in the clinical routine will be assessed in an ongoing prospective trial for patients with suboptimal response and recurrent disease. P18.032.Clinical trial information: Clinical trial information: P18.032 .