NSCLC tumors were collected from diagnostic samples, which were forwarded to the N.N. Petrov Institute of Oncology (St.-Petersburg, Russia) between 2020–2022. The study included formalin-fixed paraffin-embedded (FFPE) specimens obtained from non-smokers with mutation-negative NSCLC, of whom 76 patients were younger than 50 years and 13 subjects were non-smoking women aged 58–79 years. The proportion of tumor cells in the analyzed FFPE sections was above 70%.

RNA was extracted from manually dissected tumor cells with the PureLink FFPE Total RNA Isolation Kit (Thermo Fisher Scientific) according to the manufacturer’s protocols. RNA was purified with DNase I (Thermo Fisher Scientific) and quantified with the Qubit RNA BR Assay Kit (Thermo Fisher Scientific). RNA samples were subjected to library preparation with the KAPA RNA HyperPrep kit (Roche Sequencing Solutions, Inc.). Target enrichment was performed using the HyperCap Target Enrichment kit (Roche Sequencing Solutions, Inc.) and custom probe panels covering coding sequences of 650 kinase-related genes in accordance with the manufacturer’s guidelines. Kinase genes were divided into two groups according to the basic level of expression in lung tissues [see The HUMAN PROTEIN ATLAS resource (https://www.proteinatlas.org)]: a panel of 207 genes with a high level of expression and a panel of 443 genes with low and medium levels of expression (Tables S1 and S2). The original panels were developed by the HyperDesign Tool (Roche Sequencing Solutions, Inc.) with the specified stringency (maximum close matches: 5–10; overhang: 30 bp). Sequencing was performed on the NextSeq 2000 System (Illumina, USA) in a paired-end mode for 150 cycles in both orientations.

NGS (next generation sequencing) data were analyzed by the STAR-Fusion bioinformatic pipeline (V.1.4.0) for the identification of gene fusions and the MuTect2 tool (GATK 4.3.0.0) for the analysis of somatic mutations [13, 14]. All alterations were manually curated using the IGV (https://igv.org) and the Golden Helix GenomeBrowse tool (https://www.goldenhelix.com/products/GenomeBrowse).

The analysis of gene expression was performed for BAM-files after STAR alignment with the RSEM instrument [15]. The threshold for overexpression was a 100-fold elevation for 207 genes with a high level of expression and a 250-fold increase for low/medium expressors, with 75% quartile taken as a base-line.

The analysis of frequencies of newly identified fusions was performed by variant-specific PCR. Newly identified mutations were analyzed by allele-specific PCR. Primers and probes are presented in Tables S3 and S4. PCR reactions were performed on the CFX-96 Real-Time PCR Detection System (Bio-Rad, USA). PCR mix contained 1 μL of cDNA sample, 1× GeneAmp PCR Buffer I (Applied Biosystems, USA), 250 μM of each dNTP, 200 nM of each primer and probe, 2.5 mM MgCl₂, and 1 U of TaqM polymerase (AlkorBio, Russia) in a total volume of 20 μL. PCR reactions were initiated by the enzyme activation (95˚C, 10 min.) and included 38 cycles (95˚C for 15 s followed by 58˚C for 1 min.). Gene fusions were analyzed in 1,059 NSCLC samples negative for common druggable mutations. The analysis of point mutations included 551 NSCLCs; this group of tumors was intentionally composed of both NSCLCs lacking alterations in EGFR, ERBB2 (HER2), ALK, ROS1, RET, NTRK1/2/3, MET, KRAS, NRAS or BRAF genes (n = 367) and carcinomas carrying activating mutations in EGFR or KRAS genes (n = 184). Clinical characteristics of the patients are given in Table 1.

We initially considered 2,994 NSCLC patients who underwent molecular testing for common EGFR, ERBB2 (HER2), ALK, ROS1, RET, NTRK1/2/3, MET, KRAS, NRAS, and BRAF gene alterations from August 2020 to May 2022. Smoking status was available for 2,034/2,994 (67.9%) of these cases. Presence of driver mutation in the above genes was detected in 369/1,017 (36.3%) smokers vs. 595/1,017 (58.5%) non-smokers (p < 0.0001), 776/1,953 (39.7%) males vs. 642/1,041 (61.7%) females (p < 0.0001), and 1,237/2,650 (46.7%) patients aged above 50 years vs. 181/344 (52.6%) subjects of 50 years or younger (p = 0.0383). The frequency of druggable events in female non-smokers approached 421/623 (67.6%). We further selected for the study 89 non-smoking mutation-negative patients, who were represented by 76 subjects younger than 50 years (24 females and 52 males) and 13 women aged 58–79 years. These tumors were analyzed for alterations in 650 kinase genes.

RNA sequencing revealed 93 chimeric transcripts in 89 tumors, with 0–6 fusions per tumor sample. There were 31 translocations involving two chromosomes each and 62 intrachromosomal rearrangements (17 inversions, 27 deletions, and 18 duplications). Thirty-two rearrangements were in-frame, i.e., they were able to produce a potentially functional transcript, while the remaining 61 fusions were out-of-frame. Nine rearrangements occurred more than once in this data set, however, they all were out-of-frame. Tyrosine kinase domain was fully preserved in 9 out of 32 in-frame rearrangements (Table 2), and a part of the kinase portion of the gene was retained in 8 out of these 32 tumors (Table S5). These 17 rearrangements were confirmed by variant-specific PCR and further analyzed in 1,059 NSCLCs, which lacked common alterations in EGFR, ERBB2 (HER2), ALK, ROS1, RET, NTRK1/2/3, MET, KRAS, NRAS, and BRAF genes. These efforts led to the identification of two additional tumors with ADK::KAT6B rearrangement and one carcinoma carrying RPS6KB1::VMP1 fusion. Both these fusions retain only a relatively small portion of the kinase domain of the involved gene, so they are unlikely to render a direct kinase-activating effect. Furthermore, the oncogenic role of RPS6KB1::VMP1 fusion is believed to be attributed not to RPS6KB1 kinase but to the altered function of VMP1 protein. In addition, the recently reported CLIP1::LTK [16] gene fusion was analyzed in 2,754 NSCLCs, which were negative for all known actionable mutations, however, no new instances of this translocation have been observed.

RNA sequencing revealed 601 non-synonymous somatic mutations. We further considered all mutations, which were located within the kinase domain of the involved genes and had combined annotation dependent depletion (CADD) score above 25 (Table 3). None of these 53 substitutions occurred more than once in 89 NGS-analyzed NSCLCs (Table S5). We further tested all these mutations in 367 NSCLCs lacking driver mutations and 184 carcinomas carrying activating mutations in EGFR or KRAS genes. No new instances of these mutations were detected.

Instances of kinase overexpression were observed in 20/89 (22.5%) NSCLCs; a single kinase gene was activated in 14 tumors, while six NSCLCs demonstrated overexpression of more than one kinase (Table S5). PIK3R1 transcription was significantly elevated in 3/89 (3.4%) tumors; CMPK2, MAP2K6, WNK2, and GRK1 kinases were overexpressed in 2 tumors each; single instances of gene overexpression were observed for 18 kinases (Figure 1). Overall, 7/89 (7.9%) NSCLCs had overexpressed kinases amenable to therapy with clinically available compounds (Table 4).