Purpose: The National Cancer Institute-Molecular Analysis for Therapy Choice (NCI-MATCH) trial was implemented to identify actionable genetic alterations across cancer types and enroll patients accordingly onto treatment arms, irrespective of tumor histology. Using multiplex polymerase chain reaction (PCR) next-generation sequencing, NCI-MATCH genotyped 5,540 patients, discovering gene fusions in 202/5,540 tumors (3.65%). This result, substantially lower than the fusion detection prevalence of 8.5% across all patients with cancer screened at Massachusetts General Hospital's (MGH) clinical laboratories, supported reanalysis of NCI-MATCH samples identified as mutations-of-interest (MOI)-negative. The assay used by NCI-MATCH requires previous knowledge of both fusion genes, cannot detect novel fusions, and may underestimate fusion-positive patients. Anchored multiplex PCR (AMP) technology permits fusion detection with knowledge of just one gene of the fusion partners.
Methods: Using AMP-based kits, we reprocessed 663 MOI-negative samples. 200 ng of RNA per sample were shipped from the Eastern Cooperative Oncology Group-American College of Radiology Imaging Network biorepository to MGH (n = 319) and Yale University (n = 344), processed, and sequenced on the NextSeq550. Reported fusions were manually reviewed, and novel fusions orthogonally verified via reverse-transcription PCR and Sanger sequencing.
Results: AMP identified 148 fusions in 142/663 MOI-negative patients (21% [95% CI, 18 to 25]), of which 28 were covered by the Oncomine Comprehensive Assay (OCA) panel but missed, while 120 were not covered by OCA. Among AMP-identified positive patients, 32 had actionable fusions, 24 contained novel fusions, and six had two fusion events. We identified fusions in 12/34 (35% [95% CI, 20 to 54]) cholangiocarcinomas and 43/109 (39% [95% CI, 30 to 49]) sarcomas.
Conclusion: Technology and awareness of actionable fusions have improved since the NCI-MATCH trial. With AMP-based technology, we identified 142 patients with fusions not detected during NCI-MATCH screening, many potentially actionable. These striking data underscore the need to optimize the fusion-detection capabilities of genotyping assays used in precision medicine.