Liquid-biopsy detection of FGFR2 and other actionable rearrangements in GI malignancies.

Abstract

4085 Background: Actionable rearrangements (RE) represent an emerging therapeutic target for GI malignancies. However, there remains uncertainty whether liquid biopsy (LBx) assays testing circulating tumor DNA (ctDNA) can detect RE with the same fidelity as tissue biopsy. Here we report the performance of ctDNA-based comprehensive genomic profiling (CGP) for the detection of RE leveraging an FDA-approved >300-gene platform, with a focus on data from patients with GI malignancies. Methods: An institutional research database of clinical CGP results for tissue biopsy (TBx, FoundationOneCDx) and LBx (FoundationOneLiquid CDx) from patients with cancers of the GI tract was reviewed. Sensitivity for FGFR2 RE was studied in patients with cholangiocarcinoma (CCA) or carcinoma of unknown primary (CUP) and both TBx and LBx CGP available. Results: Across 7870 GI LBx, 1094 predicted pathogenic RE were detected in 826 cases (10%) including 283 oncogenic kinase RE, 686 inactivating RE, and 125 other gain-of-function RE. FGFR2 was the most frequently rearranged gene, enriched in CCA (4.4%, 37/833) and stomach (3.0%, 11/368). EGFR RE (38) occurred across 6 cancer types and included 8 fusions, 15 c-terminal truncations, 4 intragenic deletions/inversions, and 3 kinase domain duplications. BRAF RE (31) were detected among in colorectal (22) and pancreas (9) cases. LBx detected 44 exon 3 skipping (protein stabilizing) RE in CTNNB1 (beta-catenin) across all samples. Other frequent activating RE were detected in MYC (36), FGFR3, RET, ROS1 (21 each), and ALK (13). Potentially targetable inactivating RE were detected in BRCA1 (28), ARID1A (24), STK11 (22), PTEN (17), TSC2 (14), and BRCA2 (8). Focusing on CCA, FGFR2 RE were more prevalent in cases with elevated (≥1%) ctDNA tumor fraction (TF, 26/342, 7.6%), matching the prevalence in TBx (7.8%, 505/6,492) with a similar spectrum of fusion partners seen for both TBx and LBx (30% BICC1, 70% rare for both). In 13 CCA/CUP patients positive for FGFR2 RE by TBx, LBx detected 12 (92%) including 8 different fusion partners; the false negative case had TF <1%. In 11 CCA LBx samples with FGFR2 resistance mutations detected, LBx CGP successfully detected a driver FGFR2 RE in 10 and driver C382R mutation in 1. Conclusions: Rearrangements are represented across GI malignancies, including many that result in known oncogenic drivers that may be targetable with available therapies. We observe reliable detection of FGFR2 fusions in ctDNA, in contrast to some prior publications. ctDNA represents a pragmatic analyte for detection of rearrangements and other actionable alterations, offering timely result return when tissue is inadequate or unavailable. [Table: see text]