Histopathologic, molecular, and clinical associations of fusion-positive gastrointestinal carcinomas

Comprehensive genomic profiling (CGP) is critical for optimal clinical care for gastrointestinal (GI) carcinomas. Outside of biliary tract cancers (BTCs), actionable fusions are uncommon and the utility of routine RNA-based fusion testing in GI cancers is not well established. We therefore evaluated routinely conducted RNA-based fusion testing in GI cancers at a single center.

Patients with GI carcinomas who underwent RNA-based tissue fusion testing at Massachusetts General Brigham Cancer Institute were included. Tumors with fusions were characterized for molecular and histopathologic associations. RSPO2/3 fusions in lower GI cancers were further characterized for survival and benefit from targeted therapy.

A total of 2,300 patients with GI carcinomas were included; 139 (6.0%) had a detected fusion. BTCs had the highest fusion incidence (13.7%), mostly FGFR2. Fusions were detected in 3.5-7.7% of colorectal cancer (CRC), pancreatic, esophagogastric adenocarcinoma (EGA), and small bowel 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. RSPO2/3 fusions were the second most frequent overall (1.4%) and most frequent in CRC (2.6%) and small bowel cancers (5.8%). RSPO2/3 fusions were associated with poorly differentiated and mucinous histology, preserved mismatch repair, wild-type (WT) APC, and either KRAS or BRAF V600E mutations, however were not associated with differential survival in CRC patients receiving BRAF targeted therapy. Fifteen of 20 (75%) pancreatic tumors with fusions were KRAS WT; in KRAS-mutated pancreatic cancer, most fusions were of unclear significance. Receipt of fusion targeted therapy was associated with longer survival in fusion positive pancreatic cancers and FGFR2-fusion positive BTCs. 20 fusions detected in EGA were heterogenous, including FGFR, CLDN18, and RSPO2.

RNA-based testing identifies potentially actionable fusions in a clinically significant fraction of GI carcinomas. Patients who receive fusion targeted therapy during their clinical course demonstrate longer survival, supporting CGP including fusion testing for all patients. Lower GI cancers harboring RSPO2/3 fusions have a distinct molecular and pathologic phenotype, but demonstrate few clinical differences. Several detected fusions had novel gene partners, highlighting the utility of partner-agnostic fusion testing.