Development and clinical validation of news transcriptomic tools for predicting the response to individual drug of the mfolfirinox regimen in patients with pancreatic ductal adenocarcinoma

FRAUNHOFFER NICOLAS A.; TEYSSEDOU, CARLOS; BIGONNET, MARTIN; PESSAUX, PATRICK; IOVANNA, JUAN; DUSETTI, NELSON

Chicago

Congreso; American Society of Clinical Oncology; 2023

American Society of Clinical Oncology

Background: Current therapeutic options for patients with a pancreatic ductal adenocarcinoma (PDAC) implicate monotherapy (gemcitabine) or combined therapies (modified FOLFIRINOX). mFFX treatment is accompanied by a high toxicity and therefore, administration of mFFX is conditioned by the performance of the patients, rather by rational criteria. In this work, we developed transcriptomic signatures for each drug of the mFFX regimen and validate their clinical interest. Methods: To extract biologically relevant signatures for 5FU, oxaliplatin and irinotecan, we integrated transcriptomic data from patient-derived primary cell cultures (PDC), patients-derived organoids (PDO) and patient-derived xenografts (PDX) with their corresponding chemo-response profiles to capture the biological components responsible for the response to each drug. The genes displaying the highest levels of contribution defined the signatures 5FUCore, OxaCore and IriCore. We further validated the signatures in a pooled cohort of 167 patients with advanced and metastatic PDAC (94 patients from the COMPASS cohort and 73 from the Angers-Strasbourg cohort). Results: All three signatures captured high responder patients for overall survival (OS) and progression-free survival (PFS) in the mFFX arm exclusively. We then studied the response of patients to 0, 1, 2 and 3 drugs and we identified a positive correlation between the number of drugs predicted as sensitive and the OS and PFS, and with the objective response rate. We next evaluated the level of association between the PurIST classifier and our signatures. We found an association between both OxaCore and IriCore and the classical subtype, but none with 5FUCore. However, we observed in a multivariate Cox regression that our model based on independent signatures displayed higher predictive value for the OS and PFS. Conclusions: We developed three novel transcriptome-based signatures which define sensitivity for each mFFX components that can be used to rationalize the administration of the mFFX regimen and could help to avoid unnecessary toxic effects.