ARID1A mutation to predict disease progression during first-line chemotherapy in biliary tract cancer patients.

4105 Background: Biliary tract cancer (BTC) is a retractable disease showing a dismal prognosis with therapeutic resistance. There are clinical unmet needs on predicting therapeutic response and precise strategy for the patient classification according to clinically relevant tumor biology in the patients with BTC. We aimed to identify clinically detectable genomic alteration predicting therapeutic response after first-line chemotherapy in BTC using real-world data. Methods: A comprehensive genomic analysis of multi-institutional cohorts of BTC cases was performed using next-generation sequencing (NGS) with targeted DNA panel and patients’ clinicopathologic data. Results: A total of 200 BTC patients with NGS panel tests from three cancer centers were included in this study. The genomic alteration of TP53 (55.5%), KRAS (23%), ARID1A (10%), and ERBB2 amplification (10%) were the most frequent alteration events in the BTC. Pathologically-proved BTC including extrahepatic (n = 52), ampulla of Vater (n = 4), gallbladder (n = 56), intrahepatic (n = 88) cancers showed a distinct pattern of genomic alterations in terms of ARID1A for extrahepatic BTC and ERBB2 amplification, RB1, ARID2 for GB cancer, and KRAS, IDH1, PBRM1, BAP1 for intrahepatic BTC respectively (chi-square test, P < 0.05). The oncologic outcomes for progression-free and overall survival were significantly stratified according to the best response after the first-line chemotherapy (log-rank test, P < 0.001). The logistic regression test revealed that ARID1A, BRCA2, and STK11 could significantly predict disease progression during first-line chemotherapy. ARID1A, especially, was the only independent predictive marker in the multivariate regression model in total BTC (OR 3.91, 95%CI 1.25-11.66, P = 0.015) and extrahepatic BTC (OR 5.71, 95%CI 1.23-28.98, P = 0.027). The predictive performance of significant genomic alteration was enhanced with the tumor marker CA19-9 (DeLong’s test, Z = 1.933, P = 0.053, AUC 0.73, 95%CI 0.623-0.837). Conclusions: Clinically available NGS test showed distinct genomic alterations in terms of different deterioration patterns for oncogenic molecular pathways according to the anatomic locations of BTC. Integrative analysis using the data for genomic alteration and therapeutic response for the first-line chemotherapy uncover that the patients with ARID1A mutation show a significant disease progression rate during initial treatment for BTC, especially in the extrahepatic BTC. Prospective translational studies revealing underlying biology and precision strategy should be followed to improve the therapeutic response of BTC.