Prospective organoid drug profiling and clinical response correlation for patients with primary or recurrent ovarian carcinoma (OC) in the PROSPERITY study

We evaluated drug sensitivity profiling of primary and recurrent OC patient-derived organoids (PDOs) from 42 patients prospectively enrolled to the PR o filing O varian cancer S to improve PER sonal I zed T herap Y (PROSPERITY) study . A total of 50 biopsy samples from 36 participants were received including paired pre- and post- chemo samples for 8 and paired samples from 2-3 different metastatic sites for 6 participants. Short-term PDOs were generated by SEngine Precision Medicine (TEMPUS) and assessed for drug response to a panel of 47 chemotherapeutic or targeted drugs. A filter-based algorithm was applied to each concentration-response curve to generate a sensitivity numerical score (SPM score) ranking drug responses from 15 to 1. Additional metrics were employed to group and assign categories: SPM PDO scores of 15-14 were categorized as exceptional responses, 13-12 as good, 11-9 as moderate to low, and below 9 as no response. Tumor origin and driver mutations were confirmed using whole-exome sequencing. Patient clinical response was assessed post-PARIS test using radiologic response, chemotherapy response score, and KELIM CA-125 kinetics and categorized as concordant or non-concordant relative to the PDO prediction. Drug profiling was successful in 43/50 (86%) samples with a median turnaround time of 20 days for pre-treatment samples and 26 days for post-chemo samples. Tested samples included 16 primary OC collected before chemotherapy, 20 collected after 3-4 cycles of neoadjuvant chemotherapy, and 7 recurrent OCs. 7/21(33%) post-chemo samples failed quality control (QC) metrics compared to 2/17 (11.7%) pre-chemo. Of 9 technically and clinically evaluable primary OC obtained pre-chemo, clinical response to carboplatin-based therapy was concordant with PDO prediction in 7 (77.8%). Of the 9 post-chemo evaluable cases, 6 (66.6%) were clinically concordant with PDO prediction. Of the 6 evaluable participants with recurrent OC, none were treated with a PARIS-guided therapy, preventing assessment of clinical concordance. In both the primary and recurrent OCs, unique PDO drug sensitivity profiles were identified in all profiled samples that would not have been predicted by standard molecular profiling, suggesting actionability. In 5 cases tested at two different metastatic sites, SPM scores were highly concordant (Spearman ranks, 0.48-0.77), one pair in the sixth case failed QC and was non-evaluable. In conclusion, drug profiling of both primary and metastatic OC derived organoids is highly feasible and shows promise for clinically relevant drug response predictions with high concordance in samples from more than one metastatic site. Coverage of off-label drugs with predicted exceptional and good responses has been a major barrier to clinical utilization of PDO profiling in the recurrent setting.