Pharmacogenomic Blood-Based Assay To Predict Chemotherapy Response And Survival In Pancreatic Cancer

738 Background: Pancreatic adenocarcinoma (PDAC) is for most patients a refractory disease. Modern cytotoxic chemotherapeutics (C) are not yet optimal for inducing responses and extending life. We are developing a blood-based pharmacogenomic (PGx) assay profiling circulating tumor and invasive cells (CTICs) to predict sensitivity and resistance to C. Methods: The PGx assay was studied in two cohorts of patients (pts) presenting for frontline C therapy of metastatic PDAC. Cohort 1 is from an ongoing prospective study (planned n=80); pts are enrolled prior to receiving either FOLFIRINOX or gemcitabine (Gem)/nab-paclitaxel (Nab-P). Cohort 2 (n=50) consists of pts enrolled prior to receiving bespoke combinations of C agents informed by the assay. 6 mL of peripheral blood was collected from pts at baseline and while on C therapy. CTICs were isolated by previously described collagen invasion assay, total RNA was extracted and gene-expression analysis was performed. PGx models for seven C agents used in PDAC were applied, and correlated to treatment received. Pts were classified as sensitive if C received were predicted to be effective and resistant if not. Objective endpoints were PFS and OS. Results: Cohort 1 patients who received sensitive first-line C combinations experienced significantly longer time to progression (TTP) v resistant (Table). In Cohort 2, the PGx assay was predictive of TTP and OS when used across multiple lines of therapy, with a two-year survival of 38%. Greater OS was observed in Cohort 2 pts receiving heterogenous C regimens more highly correlated to those predicted by the PGx assay. Conclusions: The PGx assay has promising predictive performance in both standard and personalized C regimens. A prospective, directed trial comparing these approaches is warranted. Clinical trial information: NCT03033927. [Table: see text]