Characterization of antibody-drug conjugate target antigen expression in patients with biliary tract cancer.

e15042 Background: Data on antibody–drug conjugates (ADC)s target expression prevalence, intertumoral heterogeneity, concordance with gene drivers, and effect on clinical outcome remain limited for patients (pts) biliary tract cancers (BTC). Methods: Under IRB approved retrospective biospecimens protocols, resected primary specimens and matched metastatic sites from pts with BTC were assembled into tissue microarrays and tested for CLDN18.2, c-MET, Nectin-4, TROP2, and HER2 expression by immunohistochemistry (IHC). A subset of cases underwent targeted next-generation sequencing for genomic characterization using MSK-IMPACT (NCT01775072). Protein positivity was determined using predefined thresholds (IHC 1+ ≥10% for CLDN18.2, c-MET, Nectin-4, TROP2; and HER2 by ASCO-GEA scoring criteria), and H-scores (except for HER2). Fisher exact test was used to examine associations between ADC target expression and anatomic site and between, genomic alterations. Univariate Cox regression method was used to correlate target expression with overall (OS) and recurrence-free (RFS) survivals. Cohen’s kappa(κ) was used to determine concordance on target expression (positive vs negative) between paired samples. Results: Sixty-five pts with resected biliary tract cancer with 18 paired metastatic sites were identified—median age 72 years, 55% male, 43% extrahepatic cholangiocarcinoma (eCCA), 40% intrahepatic cholangiocarcinoma (iCCA), and 17% gallbladder cancer (GBC). All evaluated target antigens were overly expressed; percent positivity and H-score > = 200 in descending order of frequency were: TROP2 (83%, 26%), c-MET (75%, 26%), Nectin-4 (66% ;35%), and CLDN18.2 (46%; 7.7%). HER2 overexpression occurred in 3.1% of tumors. Fifty-four (83%) pts had more than 1 target antigen expressed, 19 (29%) had at least 2 antigens expressed. Target percent positivity in TROP2, c-MET, and CLDN18.2 had higher expression in GBC and/or eCCA anatomic sites. Agreement among paired primary and metastatic samples ranged from 43% to 75% with the highest observed for TROP2 (71%; κ not available due to near-uniform positivity) and HER2 (75%; κ = 0.29). Overall concordance was low for c-MET, CLDN18.2, and Nectin-4. In 33 pts with genomics, frequently altered genes included TP53 (36%), SMAD4 (27%), ELF3 (21%). ERRB2 amplification was observed in 1 (3%) pt and 2 pts were HER2 positive; MET amplification was not observed, and 7 pts were c-MET H-score ≥ 200. Target antigen expression did not associate with underlying genomics, RFS, or OS. Conclusions: Acknowledging limitations of this retrospective analysis, BTC displays frequent but heterogeneous expression of multiple ADC targets. These findings suggest inherent complexity of target protein quantification, target threshold determination, and target sampling discordance while providing a biologic rationale for ongoing biomarker-driven ADC trials for pts with BTC.