Predictive 'biomarker piggybacking': an examination of reflexive pan-cancer screening with pan-TRK immunohistochemistry.

Abstract

Tropomyosin receptor kinase (TRK)-targeted therapies represent an important therapeutic option for patients with advanced solid tumours harbouring neurotrophin receptor kinase (NTRK) gene fusions. However, NTRK fusions are rare in common adult carcinomas, and systematic approaches to screening for these alterations are lacking. Pan-TRK immunohistochemistry (IHC) has been proposed as one method to screen for NTRK fusion-positive tumours. Reflexive testing strategies have been endorsed for several IHC-based biomarkers, and thus offer a convenient and low-cost entry point to incorporate pan-TRK screening. In this study, 447 consecutive cases of adult solid tumours undergoing mismatch repair (MMR), human epidermal growth factor receptor 2 (HER2) and/or programmed cell death ligand 1 (PD-L1) testing were prospectively stained with pan-TRK IHC. Four cases (0.9%) were pan-TRK positive, including three (1.3% of 223) colonic adenocarcinomas, two of which were MMR-deficient and one (1.4% of 71) was gastroesophageal carcinoma. None of 108 non-small cell lung carcinomas showed pan-TRK expression. NTRK gene fusion was confirmed by DNA sequencing in one MMR-deficient colonic adenocarcinoma. In one MMR-deficient tumour, an alternative mitogen-activated protein kinase (MAPK) driver was identified. In the oesophageal (squamous cell) carcinoma, RNA sequencing identified relative NTRK2 transcript overexpression in the absence of a fusion. In one MMR-proficient colonic adenocarcinoma, no MAPK drivers were identified; therefore, a falsely negative sequencing result was favored. None of the patients met clinical criteria for TRK-targeted therapy. The clinical impact of pan-TRK IHC 'piggybacking' on existing reflexive testing strategies in surgical pathology appears negligible. Carcinomas may rarely show high-level pan-TRK expression in the absence of an underlying NTRK fusion event.