RNA-based fusion testing in gastrointestinal (GI) carcinomas: A single-institution experience.

Abstract

3061 Background: Molecular profiling with next-generation sequencing is critical to provide appropriate clinical care in most GI cancers. However, outside of biliary tract cancers (BTCs), the utility of routine comprehensive fusion testing in GI carcinomas is not well established. We evaluated fusion testing of GI carcinomas in routine clinical care at Massachusetts General Hospital Cancer Center (MGHCC). Methods: Patients with confirmed GI carcinomas who underwent RNA-based tissue fusion testing at MGHCC between Jan 2016 and April 2023 were included. Samples with tumor cellularity <20% or not meeting quality controls were excluded. Fusions of known clinical relevance were evaluated for molecular and histopathologic associations. Results: 2300 GI cancers with at least 1 fusion assay were included (see Table). As expected, BTCs had the highest incidence of fusions (13.7%), 80% of which were FGFR2. Fusions were detected in 3.9-7.7% of colorectal (CRC), pancreatic ductal (PDAC), esophagogastric (EGC), and small bowel (SBC) carcinomas. Fusions of BRAF, FGFR1-3, NTRK1/3, RSPO2/3, and NRG1 were rare (excepting FGFR2 in BTCs) but observed in 3 or more cancer types. Consistent with prior studies, the 3 NTRK1 fusions observed in CRC were associated with MLH1 loss and 15 of 21 (71%) PDAC tumors with fusions were KRAS wild-type (WT). RSPO2/3fusions were the second most frequent after FGFR1-3. RSPO3fusions were predominantly seen in CRC and SBC and were associated with poorly differentiated histology (43%), microsatellite stability (100%), APC WT (92.6%), and KRAS (44.4%), BRAF V600E (40%), and BRCA1/2 mutations (29.6%). Of 5 RSPO2fusions, 4 were in EGC and 100% had poorly differentiated histology and TP53 mutations. No fusions were detected in 47 GI neuroendocrine tumors. Conclusions: Although individually rare, in aggregate RNA-based testing identifies clinically relevant fusions in a small but significant fraction of BTC, CRC, PDAC, EGC, and SBC. Molecular selection may identify subpopulations in which fusion testing is higher yield. CRC and SBC harboring RSPO3 fusions have a distinct molecular and pathologic phenotype, including a potentially novel association with BRCA1/2 mutations. Several detected fusions had novel partners that were nonetheless predicted actionable, highlighting the utility of partner-agnostic fusion testing in this population. [Table: see text]