Abstract 1394: RT-PCR and NGS comparison: detection of known EGFR mutations in non-small-cell lung cancer clinical samples with routine laboratory testing

Abstract

Abstract Next-generation sequencing (NGS) promises to deliver clinical mutation analysis for multiple actionable biomarkers at once. However, most targeted therapies are developed with companion diagnostic assays after rigorous analytical and clinical validation. In order to compare the consistency of biomarker results under standard laboratory workflow, we measured the detection of known EGFR mutations in 17 clinical and 2 reference samples using the cobas® EGFR Mutation Test version 1(cobas test), the TruSeq and TruSight panels using the Illumina MiSeq and AmpliSeq panel using the Thermo Fisher PGM instruments. Samples were tested in triplicate at 2 sites for each platform with 2 sites performing cobas testing and sequencing. Archived clinical samples were NSCLC adenocarcinoma with average tumor cell content at 60.3% ± 21.0% and mutant allele frequency of 36.0% ± 17.0% based on pyrosequencing during initial clinical testing. Three samples had scar tissue present and the remainder had little to no necrosis. The average age was 67.6 ± 7.0 years and samples came from 10 women, 6 men and 1 gender not specified. Data were analyzed to assess the inter- and intra- platform discordant rates, the mean error rates relative to the known mutation status, platform invalid rates, average turnaround time and repeat testing rates. All platforms had low discordant and mean error rates when using the 2.5% mutation frequency cutoff and a read depth (RD) limit = 100. One PGM site had a sample invalid rate between 20%-51% for different EGFR amplicons at RD = 100; the upper limit of the invalid rate for PGM sites was >70% with RD = 500. The cobas z 480 platform had a 0-0.32% intra-platform discordant rate and a 0.65-0.93% mean error rate. The MiSeq platform had a 0-0.69% intra-platform discordant rate and a 0-0.69% mean error rate. The PGM platform had a 0.99-3.15% intra-platform discordant rate and a 1.5-4.02% mean error rate. The mean numbers of all non-silent COSMIC mutations detected by PGM and MiSeq at the 2.5% cutoff were 4.36 and 2.32, respectively. However, the mean number of EGFR mutations detected relative to the expected number was 100% for MiSeq and 92.4% and 100% for PGM at 2.5% and 10% frequency cutoffs, respectively. The 3 platforms were assessed for samples that required repeat testing, re-extraction of DNA and the turnaround time (TAT) from DNA isolation to result. In all three cases, the cobas test required the least TAT, fewer repeats and re-extractions relative to PGM and MiSeq. One MiSeq site required the longest turnaround time (14 days) but did not require repeat testing for invalid runs. Re-extraction from backup sections was common for both PGM and MiSeq but not cobas. The cobas test provides accurate, precise, and fast actionable results for NSCLC patients. NGS approaches can result in accurate and precise results when adequate RD is achieved but TAT remains a challenge relative to the cobas test. Citation Format: John F. Palma, Yu Chuan Tai, Wei-min Liu, Steve Anderson, Anup Madan, John W. Longshore, Arundhati Rao. RT-PCR and NGS comparison: detection of known EGFR mutations in non-small-cell lung cancer clinical samples with routine laboratory testing. [abstract]. In: Proceedings of the 107th Annual Meeting of the American Association for Cancer Research; 2016 Apr 16-20; New Orleans, LA. Philadelphia (PA): AACR; Cancer Res 2016;76(14 Suppl):Abstract nr 1394.