Identifying a wide range of actionable variants using capture-based ultra-deep targeted sequencing in treatment-naive patients with primary lung adenocarcinoma.

Precision medicine requires accurate multi-gene clinical diagnostics. In current clinical practice, the minimum confidence threshold for variant calling of targeted next-generation sequencing (NGS) on surgical specimens is set to 2%-5%. However, few studies have been conducted to identify a wide range of actionable variants using capture-based ultra-deep targeted sequencing, which has limit of detection (LOD) of 1%. The AmoyDx® Essential NGS panel for capture-based ultra-deep targeted sequencing (dual-indexed sequencing adapters with UMIs) was performed on 372 surgical specimens obtained from treatment-naive patients with primary lung adenocarcinoma, to detect actionable somatic driver mutations associated with each patient. Single-nucleotide variants, insertion/deletion events, and rearrangements were reported. Amplification-refractory mutation system (ARMS) assay and fluorescence in situ hybridization (FISH) were performed for the validation of hotspot mutations in EGFR and ALK, ROS1, and RET fusions. Potentially actionable variants were identified in 80.5% (352/437) of the nonsynonymous variants that were able to be sequenced, and were most commonly found in EGFR mutations (59.7%, 261/437), followed by KRAS mutations (5.5%, 24/437), PIK3CA mutations (3.7%, 16/437), ALK rearrangements (3.4%, 15/437), BRAF mutations (2.7%, 12/437), ERBB2 mutations (2.5%, 11/437), and RET rearrangements (2.3%, 10/437). A total of 7.2% (28/372) of the samples had multiple actionable mutations. Among the 93 triple-negative cases, which did not harbor mutations in EGFR, KRAS, or BRAF, gene fusions were detected in 26 cases (28%). Of the 328 samples, concordance of EGFR between the ARMS assay and NGS was observed in 318 samples (97.0%), and among 32 samples, concordance between ARMS/FISH test and NGS for ALK/ROS1/RET fusion genes was observed in 30 samples (93.8%). Here, we demonstrated that the capture-based ultra-deep targeted sequencing method, which has a LOD of 1% to profile a wide range of actionable variants in surgical specimens of treatment-naive lung adenocarcinoma patients, highlights the need for treatment-naive patients to undergo genomic profiling.