Cohort-level patient-derived organoid assays and prediction of clinical trial outcomes in PDAC.

Abstract

785 Background: Clinical trials are one of the major barriers in contemporary drug development. Improving the number of therapeutic solutions for patients whilst limiting the financial burden of healthcare systems will require dramatically improving clinical trial success rates. Functional assays based on patient-derived organoids (PDO) are a promising new tool for derisking clinical development of new therapies, but their use has been limited due to small cohort sizes and the absence of systematic validation studies. Methods: We have assembled a collection of 70 demographically relevant PDAC PDOs. Each PDO was screened with a 22-drug panel including standard of care, chemotherapies and targeted therapies. Using a best-in-class patient response prediction model we generate cohort scale estimations of overall response rates (ORR) and progression-free survival (PFS). The estimations are compared with data from published clinical trials to determine the validity of the approach. Results: The PDOs exhibit representative treatment histories in terms of number of lines of treatment (mean = 1.60) and treatment types (prior treatment types: folfirinox = 68.6%, gemcitabine = 50%). All the organoids’ mutational profile, past patient treatments and subsequent responses to future lines were analyzed and compared to the distributions observed in the clinic. The cohort-level drug efficacy on our PDO collection matches retrospective clinical trial ORRs reported in the literature. Furthermore, our preclinical assay predicts the relative performance of drugs in different clinical trial arms. Conclusions: We report that large-scale PDO-based assays predict clinical ORR and PFS with best-in-class accuracy. This work shows that well-characterized PDO collections can be used to improve the success rates of clinical studies for new potential treatments in PDAC.