e20618 Background: Non-synonymous SNPs (nsSNPs) in RTKs can alter kinase activity and are not exclusive to the tyrosine kinase domain (TKD). In NSCLC, EGFR lesions were previously identified using TKD-limited tests; however, next-generation sequencing (NGS) enables the entire protein sequence of many RTKs to be interrogated. Methods: We analyzed all nsSNPs in 28 RTKs in lung cancer pts who received tumor profiling with Caris NGS from 2013-2015. Mutations were classified by location including the TKD, extracellular domain (ECD), transmembrane domain (TM), juxtamembrane domain (JM), and carboxy-terminal (CT) regions. nsSNPs underwent in silico analysis using PolyPhen-2 (Harvard) to predict pathogenicity. Results: 167 pts (156 NSCLC, 11 SC) were identified with a median age 65 (range 26-85); 51% male; 65% white, 31% black; 77% ≥20 pack-years (py), 11% non-smokers; 52% samples tested were metastases. NSCLC pts were 63% adenocarcinoma, 22%, squamous, 8% large-cell; 81% stage IV, 14% III; 17 were EGFR+, 6 BRAF+, 3 HER2+, 3 ROS1 rearranged and 1 MET exon 14. A total 300 nsSNPs (286 NSCLC, 14 SC) were found in 28 RTKs, excluding EGFR. 123/156 NSCLC pts (79%) and 9/11 SC (82%) had ≥1 RTK lesion with median 2 (range 0-8); 143/300 (48%) nsSNPs were predicted-damaging (pnsSNP) by in silico and 89 pts (53%) had ≥1 pnsSNP (median 1; range 0-5). 28/28 RTKs had ≥3 mutations, with median 11 (range 3-23), and 26/28 contained ≥1 pnsSNP (median 5; range 0-14). RTKs in NSCLC with the most frequent nsSNPs were EPHA3 (14/23 variants were pnsSNP), EPHA5 (11/17), EPHB1 (10/11), RET (9/11), ERBB4 (8/12), ALK (7/16), NTRK3 (7/15), ROS1 (6/22) and FLT1 (6/15). 6/14 lesions in SC pts were pnsSNPs in ERBB3, ERBB4, FGFR1, FLT1, RET and ROS1. nsSNPs were found along RTKs: 57% were ECD (72/172 pnsSNP), 26% TKD (47/77), 10% CT (14/29), 6% JM (8/18) and 1% TM (2/4). 6/6 SC pnsSNPs were ECD. 67% BRAF+ and ROS1-rearranged, 59% EGFR+, 33% HER2+ and 0/1 MET exon 14 pts had ≥1 pnsSNP. Conclusions: Nearly 80% NSCLC and SC pts had ≥1 nsSNP in 28 RTKs, excluding EGFR, with 48% pnsSNPs by in silico analysis. As > 70% nsSNPs were extra-TKD lesions, further characterization is needed to identify kinase-effecting variants and their potential clinical significance.
Identification of key candidate tumor biomarkers in non‑small‑cell lung cancer by in silico analysis
