实体瘤功能性药物敏感性检测的真实世界一致性分析及胰腺癌前瞻性观察性病例系列

Background: Fluorouracil- and Gemcitabine-based combinations are the preferred early treatments of pancreatic ductal adenocarcinoma (PDAC). However, treatment choice is empirical due to the lack of reliable predictive biomarkers. We report an observational case series using Optim.AI™, a functional drug sensitivity testing (DST) platform, to identify the activity toward standard-of-care regimes and assess concordance with clinical response. Methods: Tumor samples from PDAC patients underwent ex vivo analysis using Optim.AI™ platform. Tumor cells were isolated and tested with a customized panel of chemotherapeutic and targeted agents, including standard-of-care treatment. Optim.AI™ assesses all possible combinations and ranks them by sensitivity for individual patients. Clinical responses were obtained from chart review, and concordance with ex vivo findings was assessed. To further evaluate performance across tumor types, 76 solid cancer cases were evaluated for Optim.AI™ functional testing were also analyzed. Retrospective concordance was determined by evaluating tumor cell viability responses generated by Optim.AI™ against the treatments previously administered to the corresponding patients. Results: Optim.AI™ reports were successfully generated for 95% of the samples with sufficient yield for testing. Pan-cancer retrospective analysis also demonstrated 91% concordance between Optim.AI™-reported sensitivities and prior clinical responses across the 76 solid tumor cases evaluated. Six PDAC patients underwent functional DST, with Optim.AI™ reports generated within a median turnaround of six days from sample collection. Substantial inter-patient variability in ex vivo drug sensitivity toward the standard-of-care was observed. Across six patients, there were seven assessable clinical outcomes corresponding to the combinations tested ex vivo. Optim.AI™ correctly predicted resistance to the prior or ongoing therapy in six out of seven instances (NCV > 0.5 defined as ex vivo resistance), resulting in a predictive accuracy of 85.7%. Furthermore, Optim.AI™ identified a combination regimen that was more sensitive than the treatments previously administered, underscoring its potential value in guiding more effective therapeutic options for refractory cases. Conclusions: This study highlights the feasibility and potential clinical relevance of Optim.AI™ in PDAC. Optim.AI™ revealed individualized drug response profiles, including cases in which standard treatments may be resistant, achieving a predictive accuracy of 85%. These findings are consistent with results from the pan-cancer retrospective concordance analysis.