Multiplex RNA-based detection of clinically relevant MET alterations in advanced non-small cell lung cancer Cristina Aguado1, Cristina Teixido2,3 , Ruth Roma n1, Roxana Reyes4, Ana Gime nez-Capita  n1 , 3,4 5 Elba Marin , Carlos Cabrera , Nuria Vin ~ olas , Sergi Castillo , Silvia Mun 4 6 ~ oz , Ainara Arcocha4, 6 Laura Lo  pez-Vilaro , Ivana Sullivan , Erika Aldeguer , Sonia Rodrıguez , Irene Moya5, 7 8 1 1 5,9 Santiago Viteri , Andre s Felipe Cardona10,11 , Ramon Palmero12, Cristina Sainz13,14, Miguel Mesa-Guzma n , Maria D. Lozano14,16,17, Andre 15 s Aguilar-Herna ndez5, Alejandro Martınez- Bueno , Marıa Gonza 5 lez-Cao , Elena Gonzalvo , William P. J. Leenders18, Rafael Rosell5,19,20, 5 2 13,14,16,17 Luis M. Montuenga , Aleix Prat3,4, Miguel A. Molina-Vila1 and Noemi Reguart3,4 1 Laboratory of Oncology, Pangaea Oncology, Quiro n Dexeus University Hospital, Barcelona, Spain 2 Thoracic Oncology Unit, Department of Pathology, Hospital Clınic, Barcelona, Spain 3 Translational Genomics and Targeted Therapeutics in Solid Tumors, Institut d’Investigacions Biome diques August Pi i Sunyer (IDIBAPS), Barcelona, Spain 4 Thoracic Oncology Unit, Department of Medical Oncology, Hospital Clınic, Barcelona, Spain 5 Dr Rosell Oncology Institute, Dexeus University Hospital Quiron Salud Group, Barcelona, Spain 6 Division of Medical Oncology, Hospital General de Granollers, Barcelona, Spain 7 Department of Pathology, Hospital de la Santa Creu i Sant Pau, Barcelona, Spain 8 Division of Medical Oncology, Hospital de la Santa Creu i Sant Pau, Barcelona, Spain 9 Dr Rosell Oncology Institute, Teknon Medical Center, Quiron Salud Group, Barcelona, Spain 10 Foundation for Clinical and Applied Cancer Research-FICMAC, Bogot a, Colombia 11 Clinical and Translational Oncology Group, Institute of Oncology, Clınica del Country, Bogot a, Colombia 12 Division of Medical Oncology, Catalan Institute of Oncology, L’Hospitalet, Barcelona, Spain 13 Center for Applied Medical Research (CIMA), University of Navarra, Spain 14 CIBERONC, Madrid, Spain 15 Department of Thoracic Surgery, University of Navarra, Pamplona, Spain 16 IDISNA, Pamplona, Spain 17 Department of Pathology, Anatomy and Physiology, School of Medicine, University of Navarra, Pamplona, Spain 18 Department of Biochemistry, Radboud Institute for Molecular Life Sciences, Nijmegen, The Netherlands  en Cie 19 Institut d’Investigacio ncies de la Salut Germans Trias i Pujol, Badalona, Spain 20 Universitat Autonoma de Barcelona, Bellaterra, Spain Keywords MET inhibitors have shown activity in non-small-cell lung cancer patients amplification; expression; lung cancer; MET; (NSCLC) with MET amplification and exon 14 skipping (METDex14). RNA; skipping However, patient stratification is imperfect, and thus, response rates have varied widely. Here, we studied MET alterations in 474 advanced NSCLC Correspondence N. Reguart, Thoracic Oncology Unit, patients by nCounter, an RNA-based technique, together with next-genera- Department of Medical Oncology, Hospital tion sequencing (NGS), fluorescence in situ hybridization (FISH), immuno- Clınic Barcelona Translational Genomics and histochemistry (IHC), and reverse transcriptase polymerase chain reaction Targeted Therapeutics in Solid Tumors, (RT–PCR), exploring correlation with clinical benefit. Of the 474 samples IDIBAPS, Villarroel 170, 08036 Barcelona, analyzed, 422 (89%) yielded valid results by nCounter, which identified 13 Spain patients (3%) with METDex14 and 15 patients (3.5%) with very-high Tel: +34 93 227 54 02 MET mRNA expression. These two subgroups were mutually exclusive, E-mail: nreguart@clinic.cat displayed distinct phenotypes and did not generally coexist with other dri- Cristina Aguado, Cristina Teixido and Ruth vers. For METDex14, 3/8 (37.5%) samples positive by nCounter tested Rom an share first authorship negative by NGS. Regarding patients with very-high MET mRNA, 92% Miguel A. Molina-Vila and Noemi Reguart had MET amplification by FISH and/or NGS. However, FISH failed to share last authorship identify three patients (30%) with very-high MET RNA expression, among which one received MET tyrosine kinase inhibitor treatment deriving 350 Molecular Oncology 15 (2021) 350–363 ª 2020 The Authors. Published by FEBS Press and John Wiley & Sons Ltd. This is an open access article under the terms of the Creative Commons Attribution License, which permits use, distribution and reproduction in any medium, provided the original work is properly cited. C. Aguado et al. MET in advanced non-small cell lung cancer (Received 9 September 2020, revised 23 clinical benefit. Our results indicate that quantitative mRNA-based tech- October 2020, accepted 20 November niques can improve the selection of patients for MET-targeted therapies. 2020, available online 7 December 2020) doi:10.1002/1878-0261.12861 yet defined. The controversies around MET testing 1. Introduction [25] have been complicated by the small number of Aberrant activation of the mesenchymal–epithelial comprehensive studies on MET status in advanced transition (MET) gene has recently emerged as an NSCLC and the fact that most reports evaluating the actionable target, particularly in non-small-cell lung performance of different techniques have focused on a cancer (NSCLC) [1,2]. Multiple molecular mechanisms single MET alteration. including amplification, point mutations, alternative The NanoString nCounterTM Analysis System is a splicing, and protein overexpression [3,4] can lead to high-throughput, quantitative transcript-based hybri- abnormal MET activation, which increases cell prolif- dization technology that allows for the simultaneous eration, survival, invasion, and metastasis. MET analysis of the expression of hundreds of target genes amplification has been described in 1–6% of newly [26] and can be easily incorporated in the routine diagnosed NSCLC tumors and constitutes a frequent molecular testing workflow of tumor samples [27]. mechanism of acquired resistance in EGFR-mutant Although nCounter has been used to determine some (EGFR-mut) NSCLC patients treated with tyrosine MET alterations in particular types of tumors [28–30], kinase inhibitors (TKI) [5]. MET exon 14 alterations it has never been employed for MET testing in in donor and acceptor splicing sites—including point NSCLC. In the previous studies, we demonstrated that mutations, indels, and whole-exon deletions—lead to this methodology can identify relevant gene rearrange- the exclusion (skipping) of MET exon 14 at the RNA ments in advanced NSCLC [27,31]. Here, we aimed to level (METΔex14), which has been described in 3–4% determine whether nCounter could improve the char- of patients with advanced NSCLC [6–8]. acterization of clinically relevant MET alterations. To Amplification of the MET gene in NSCLC and muta- this end, we screened a large cohort of NSCLC tions leading to METΔex14 were first reported in 2005 patients and compared the nCounter results with those and 2006, respectively [9,10], while two seminal works obtained by standard techniques. Our results indicate published in 2015 identified METΔex14 as a potential that multiplex, RNA-based techniques such as nCoun- therapeutic target in advanced NSCLC [11,12]. Since ter have the potential to become the technology of then, several trials have evaluated the efficacy of MET choice to select patients for MET-targeted therapies. inhibitors in patients with METΔex14 and MET ampli- fication [13–17], with response rates varying widely 2. Materials and methods across the different studies. Capmatinib is the first MET inhibitor that has gained recent Food and Drug Admin- istration approval for the treatment of advanced 2.1. Patients, samples, and cell lines NSCLC with METΔex14 [18], and novel mechanisms of resistance have meanwhile started to emerge [19]. A total of 474 formalin-fixed, paraffin-embedded The most frequent technologies used to assess MET (FFPE) tumor samples from patients with NSCLC were gene copy number variations in the clinical setting are tested to identify MET alterations. Samples were col- fluorescence in situ hybridization (FISH) and next-gen- lected from 10 participating hospitals (Supporting Infor- eration sequencing (NGS), whereas for METΔex14 mation) with prior full informed patient consent and detection, both NGS and reverse transcription poly- approval from the corresponding ethical committees. merase chain reaction (RT–PCR) are commonly used All advanced NSCLC patients arriving to our institu- [6,7,13,14,20–24]. However, the optimal method(s) and tions and having biopsies available with sufficient tumor the more adequate thresholds for stratification are not tissue were offered to participate in the study, which Abbreviations FISH, fluorescence in situ hybridization; HS, histoscore; IHC, immunohistochemistry; nC, nCounter; NGS, next-generation sequencing; NSCLC, non-small-cell lung cancer; RT–PCR, reverse transcription polymerase chain reaction. Molecular Oncology 15 (2021) 350–363 ª 2020 The Authors. Published by FEBS Press and John Wiley & Sons Ltd. 351 MET in advanced non-small cell lung cancer C. Aguado et al. was conducted in accordance with the principles of the frequently altered in lung cancer, including DNA alter- Declaration of Helsinki. FFPE slides (4 µm) were ations at exons 13–15 of MET and the surrounding obtained by standard procedures and stained with intronic regions. The GeneRead panel can also detect hematoxylin and eosin. A pathologist determined the amplifications in five genes. For the GeneRead panel, tumor area and evaluated the percentage of tumor infil- up to 40 ng of purified DNA was used as a template. tration. RNA was extracted with a high purity FFPE Clonal amplification was performed on 625 pg of RNA isolation kit (Roche, Mannheim, Germany), while pooled libraries, and, following bead enrichment, the the GeneRead DNA FFPE Kit or the QIAamp DNA GeneReader instrument was used for sequencing. FFPE Tissue Kit (Qiagen, Hilden, Germany) was used RNA-NGS was performed with the GeneRead for DNA extraction from FFPE samples, according to QIAact Lung RNA Fusion UMI Panel (Qiagen) the manufacturer’s instructions. DNA and RNA con- according to the manufacturer’s instructions. This panel centrations were measured by Qubit (Thermo Fisher targets a fusion-specific and splicing variants (Table S2), Scientific, Waltham, MA, USA). Three cell lines including METDex14. The GeneRead QIAact Lung (Hs746T, PC9, and E98) were used for validation pur- RNA Fusion UMI Panel is designed to enrich selected poses. The Hs746T cell line, harboring METDex14, was fusion targets starting with 100 ng of total RNA. After purchased from the American Type Culture Collection. target enrichment and library preparation, clonal ampli- E98 is a patient-derived astrocytoma cell line with fication was performed using 625 pg of pooled libraries, amplification of MET gene. EGFR-mut PC9 cells were and, following bead enrichment, the GeneReader instru- obtained from F. Hoffman-La Roche Ltd (Basel, ment was used for sequencing. Switzerland) with the authorization of Dr. Mayumi Qiagen Clinical Insight Analyze (QCI-A) software was Ono (Kyushu University, Fukuoka, Japan). All cell employed to align the read data and call sequence vari- lines were cultured in RPMI medium with 10% fetal ants, which were imported into the Qiagen Clinical bovine serum under standard conditions and counted Insight Interpret (QCI-I) web interface for data inter- after trypsinization. Pellets from a minimum of five T- pretation and generation of final custom report. In the 75 flasks were used to generate FFPE blocks. case of the OST panel, 10 ng of purified DNA was used as a template. Libraries were pooled at 20 pM and, fol- lowing ion spheres, sequenced using the Ion Personal 2.2. FISH and immunohistochemistry (IHC) Genome Machine (Thermo Fisher Scientific). The Ion FISH for MET was performed with the ZytoLight Reporter Server (Thermo Fisher Scientific) was used to SPEC MET/centromere 7 (MET/CEP7) Dual Color align the read data to the human reference genome and Probe (ZytoVision, Bremerhaven, Germany) according call sequence variants. The MET gene copy numbers by to manufacturer’s instructions. Three positivity criteria NGS were assessed using the QCI-A and QCI-I soft- for MET amplification were used as follows: (a) a ratio ware. Copy numbers provided by QCI-I were selected (r) MET/CEP7 ≥ 2; (b) gene copy number (GCN) per according to the tumor infiltration of the sample, as cell ≥ 6; (c) ≥ 5 copies in ≥ 50% of cells; (d) or ≥ 15 assessed by an expert pathologist. The GCN ≥ 6 was copies in > 10% tumor cells. These three criteria have chosen as a threshold based on the previous experience been employed for patient stratification in clinical trials of our laboratory. During the validation of the NGS of anti-MET therapies (Table S1). Immunostaining was panel prior to its implementation in the clinical setting, performed with MET SP44 clone (Roche) on a Bench- this threshold had been found to show the highest corre- Mark ULTRA automated tissue staining system (Ven- lation with FISH MET/CEP7 > 2. tana Medical Systems, Tucson, AZ, USA). Two different cutoff points for IHC positivity were consid- 2.4. RT–PCR analysis for METDex14 transcripts ered: (i) membrane intense staining (3+) in ≥ 50% of the tumor cells and (ii) histoscore (HS) ≥ 220. RNA was converted to cDNA using M-MLV retro- transcriptase (Thermo Fisher Scientific) and oligo-dT primers, and METDex14 was amplified using HotStart 2.3. NGS sample preparation, sequencing run, Taq polymerase (Qiagen) in a 20 µL reaction and visu- and data processing alized in agarose gels. Primers used were located in DNA NGS was performed with the GeneRead exons 13 and 15, sequences were as follows: forward QIAact Lung DNA UMI Panel (Qiagen) or Oncomi- (exon 13) 50 -TTTTCCTGTGGCTGAAAAAGA-30 neTM Solid Tumour (OST) DNA Panel (Thermo and reverse (exon 15) 50 -GGGGACATGTCTGTCA- Fisher Scientific), according to the manufacturer’s GAGG-30 . Amplification generated a 246-bp band for instructions (Table S2). Both panels target genes wild-type (wt) MET RNA and a 106-bp band for 352 Molecular Oncology 15 (2021) 350–363 ª 2020 The Authors. Published by FEBS Press and John Wiley & Sons Ltd. C. Aguado et al. MET in advanced non-small cell lung cancer METDex14. Positive samples were confirmed by bidi- cells, we established that 1 mm2 of a 4-µm section rectional Sanger sequencing of RT–PCR products, with a minimum of 10% of Hs746T cells was using the big-dye 3.1 sequencing kit (Applied Biosys- required for METDex14 detection, whereas 30% of tems, Waltham, MA, USA). E98 cells were needed to detect high levels of MET mRNA expression by nCounter. In similar experi- ments, RT–PCR detected METDex14 mRNA in mix- 2.5. nCounter analysis tures with 0.1% mutant tumor cells. Finally, we Total RNA was hybridized with a custom-designed performed repeatability studies using FFPE tumors mixture of biotinylated capture tags and fluorescently from advanced NSCLC patients. In the case of labeled reporter probes (Elements Chemistry) that METDex14 detection by nCounter, four positive and included, among others, probes for MET-wt and 20 negative samples were analyzed in two indepen- METDex14 target sequences. Detailed sequence infor- dent experiments showing a 100% concordance. mation for the MET gene target regions is provided in Regarding MET mRNA expression, 29 samples run Table S3. The codeset also included probes for house- in two independent experiments revealed concordant keeping genes (actin beta, ACTB; proteasome 26S sub- results in the classification for 28 of them (96.6%, unit ATPase 4, PSMC4 and mitochondrial ribosomal CI = 82.8–99.4). protein L19, MRPL19), positive and negative controls. All processes of hybridization, capture, cleanup, and 3. Results digital data acquisition were performed with nCounter Prep Station and Digital Analyzer (NanoString Technologies, Seatle, WA, USA) according to the 3.1. Clinical samples manufacturer’s instructions. Reporter counts were col- lected with the NSOLVER analysis software version 2.6. A total of 474 FFPE NSCLC tumor samples were pro- Samples were considered not evaluable if the geometri- filed using nCounter. Among them, 52 (11%) had geo- cal mean (geomean) of counts corresponding to the mean of housekeeping gene counts below 100 and housekeeping genes was lower than 100. Counts from were excluded from the study. The remaining 422 MET probes were normalized in two steps, as evaluable samples, corresponding to 405 patients, were described [31], and subjected to a logarithmic transfor- mostly stage IIIB/IV adenocarcinomas (Table S4, mation to obtain the so-called log-MET expression Table 1). Four additional techniques were used in dif- values. Two cutoff values were used for log-MET ferent subsets of samples to validate the MET status results; (a) the mean plus standard deviation (SD) was determined by nCounter. Copy number alterations used to define cases with moderately (mod.)-elevated were analyzed by FISH and/or NGS, METDex14 tran- MET mRNA, (b) the mean plus two SD for very-high scripts by RT–PCR or DNA-based NGS, and protein MET mRNA levels. Regarding METDex14 testing, expression by IHC (Fig. 1). log-ratios were obtained dividing the normalized counts of the METDex14 probe by the normalized 3.2. Detection of METΔex14 mRNA by nCounter counts for the MET-wt probe. The cutoff for MET- in clinical samples Dex14 positivity was established as the average log-ra- tio of the sample cohort plus 2 SD. Samples with no For each of the 422 samples evaluable, we calculated counts for the METDex14 probe were directly consid- the log-ratio of the normalized nCounter counts corre- ered negative. sponding to the METDex14 vs. the MET-wt probes. A sample was considered positive if the log-ratio was above a threshold value, established as the mean plus 2.6. Validation of nCounter for detection of MET two times the SD of all samples analyzed (Fig. 2A). alterations The METDex14/ MET-wt log-ratios in our cohort Using frozen pellets from Hs746T or E98 cells, we showed a bimodal distribution with the cutoff value found that 5000 cells and 25 ng of RNA were suffi- separating the two populations (Fig. 2B). A total of 13 cient for successful detection of spliced transcripts or METDex14-positive patients (3%) were identified, with very-high MET mRNA levels, respectively. In con- a majority of nonsmoking females and a median age trast, pellets containing 500 000 of PC9 cells (MET- of 70 years (Table 1). Of them, five were treated with wt) tested negative. Next, using FFPE blocks pre- MET-TKIs and showed partial responses (N = 4) or pared by spiking different numbers of MET-depen- stabilization of the disease (N = 1) by RECIST criteria dent cells (Hs746T and E98) in a suspension of PC9 (Fig. 2C). Molecular Oncology 15 (2021) 350–363 ª 2020 The Authors. Published by FEBS Press and John Wiley & Sons Ltd. 353 MET in advanced non-small cell lung cancer C. Aguado et al. Table 1. Characteristics of all patients with valid results, patients (log-MET). According to our classification algorithm positive for METDex14 by nCounter, and patients with very-high (see Section 2), 15/422 samples (3.5%) presented very- and moderately elevated MET mRNA levels, also by nCounter, N high MET mRNA levels (Fig. 3) and 36/422 samples (%). (8.5%) mod-elevated MET mRNA levels by nCounter MET MET mRNA (Fig. 4). In contrast to the log-ratio, the distribution All mRNA mod- of the log-MET values was unimodal, although a Kol- patients METDex14 very-high elevated mogorov–Smirnov test revealed a significant deviation Characteristics N = 405 N = 13 N = 15 N = 36 from normality (P = 0.001; Fig. 3A,B). Most of the 15 Gender patients with very-high MET mRNA levels were Male 248 4 (30.8) 10 (66.7) 19 (52.8) males, former or current smokers with a median age (61.2) of 58 years (Table 1). They invariably showed very Female 146 9 (69.2) 5 (33.3) 16 (44.4) strong IHC staining (≥ 220), tested negative for MET- (36.1) Dex14 by nCounter and 92% (11/12) had MET ampli- Unknown 11 (2.7) 0 (0) 0 (0) 1 (2.8) Age at diagnosis fication by FISH or NGS (Fig. 3C). Finally, among Median 63 70 58 64 the 9 baseline patients with very-high MET, no other Range 31–89 57–84 55–61 31–84 driver was detected in seven (77.8%). Five patients Smoking status with very-high MET mRNA were treated with MET- Never 77 5 (38.5) 1 (6.7) 15 (41.7) TKIs, all of them showed partial responses by (19.0) RECIST criteria. Interestingly, one was negative by Former 134 2 (15.4) 6 (40.0) 7 (19.4) FISH (Fig. 3C). (33.1) Current 115 0 (0) 6 (40.0) 9 (25.0) Regarding the 36 cases with mod-elevated MET (28.4) mRNA levels, we found comparable numbers of males Unknown 79 6 (46.1) 2 (13.3) 5 (13.9) and females, smokers, and never smokers with median (19.5) age of 64 years (Table 1). The majority (19/27) of Sample collection time baseline samples with mod-elevated MET and geno- No data 62 3 (23.1) 2 (13.3) 4 (11.1) typing data available harbored concurrent genetic (15.3) alterations, being mutations in KRAS proto-oncogene Baseline 294 10 (76.9) 9 (60.0) 27 (75.0) (72.6) GTPase (KRAS) and epidermal growth factor receptor Baseline and 7 (1.7) 0 0 0 (EGFR), fusions in ALK receptor tyrosine kinase progression (ALK), and METDex14 the most prevalent (N = 4 Progression 42 0 4 (27.8) 5 (13.9) each; Fig. 4). only (10.4) 3.4. Comparison of METDex14 nCounter results with RT–PCR and DNA-based NGS in clinical Eight of the 13 METDex14-positive cases could be samples submitted to DNA-based NGS. Mutations affecting The concordance of nCounter with RT–PCR and exon 14 splicing sites were detected in five of them DNA-based NGS for the detection of METDex14 is (62.5%), mainly deletions in the acceptor splice-site shown in Table S6, together with the corresponding region (Table S5). Regarding other alterations, two values of specificity, sensitivity, and Cohen’s kappa. patients showed PIK3CA mutations while concomitant We observed a substantial agreement when comparing MET amplification by GCN (6–10) was identified in nCounter vs. DNA-based NGS, with a 98.5% concor- only one sample (20%). However, this case was nega- dance rate (CI = 95.6–99.5, Cohen’s kappa 0.76) and tive (< 2) by MET/CEP7 ratio. Sufficient tissue to per- only three discordant samples, all of them positive by form MET IHC was available for four of the 13 nCounter and negative by DNA-based NGS. Regard- patients; MET staining was intermediate in three cases ing nCounter and RT–PCR, there was a fair agree- and strong in one (Fig. 2C). ment (90.2%, CI = 83.3–94.4, Cohen’s kappa 0.65) and all the discordant cases (N = 11) were negative by 3.3. Quantification of MET mRNA expression nCounter and positive by RT–PCR. Six of those dis- levels by nCounter in clinical samples cordant samples had been analyzed by DNA-based NGS; mutations associated with METDex14 were not The raw counts of the MET probes for each sample detected in any case (Fig. S1). Five of them with were transformed into logarithmic normalized data remaining material were further investigated by an 354 Molecular Oncology 15 (2021) 350–363 ª 2020 The Authors. Published by FEBS Press and John Wiley & Sons Ltd. C. Aguado et al. MET in advanced non-small cell lung cancer Fig. 1. Flowchart of the study. FISH, fluorescence in situ hybridization; IHC, immunohistochemistry; NGS, next-generation sequencing; NSCLC, non-small-cell lung cancer; RT–PCR, reverse transcription polymerase chain reaction. orthogonal RNA-based NGS and all tested negative elevated cutoff was selected, the agreement was only for METDex14 skipping transcripts. Finally, we sys- fair with any of the three FISH amplification criteria, tematically sequenced the 106 bp cDNA band, corre- with concordance rates of 50%-70% and Cohen’s sponding to the METDex14 mRNA, obtained in RT– kappa 0.185–0.410 (Table S8). PCR-positive samples. No differences were observed DNA-based NGS with the GeneRead platform, between concordant and discordant samples, being the which can detect amplifications in several genes, was base sequence of the exon 13–exon 15 junction identi- performed in 80 samples with nCounter data. MET cal in all cases to the sequence described in the litera- amplification by DNA-based NGS showed an almost ture [9]. perfect agreement with very-high MET mRNA levels, with a Cohen’s kappa of 0.886 and a 97.5% concor- dance rate (Table S7). The only two discordant sam- 3.5. Comparison of MET expression levels by ples were MET amplified by NGS but did not show nCounter with MET amplification by FISH and very-high MET mRNA expression levels by nCounter. NGS in clinical samples In contrast, if mod-elevated MET mRNA levels were A total of 40 samples had evaluable data by FISH and employed, the agreement with MET amplification by nCounter. Three different criteria for MET positivity NGS was significantly worse (Cohen’s kappa 0.494) were used for FISH evaluation (see Section 2). A mod- and the 14 discordant cases had moderately elevated erate to substantial agreement was observed if the MET mRNA but did not show copy number gains nCounter very-high cutoff was employed, with the (Table S8). highest agreement for the FISH ratio MET/CEP7 ≥ 2 (concordance rate 92.5%, Cohen’s kappa 0.778; 3.6. Comparison of MET expression levels by Table S7). The only three discordant samples were nCounter and IHC in clinical samples positive by nCounter and negative by FISH. Two of them had remaining material available and were sub- Ninety-one samples were used to perform a compara- mitted to RNA-based NGS, testing negative for tive study of IHC vs. nCounter for the quantification known MET gene fusions. If the nCounter mod- of MET expression levels. Among these 91 samples, 34 Molecular Oncology 15 (2021) 350–363 ª 2020 The Authors. Published by FEBS Press and John Wiley & Sons Ltd. 355 MET in advanced non-small cell lung cancer C. Aguado et al. A C B Fig. 2. Detection of METDex14 by nCounter. (A) METDex14/MET-wt normalized counts obtained by nCounter, expressed as log-ratios. Only samples with detectable counts for METDex14 are plotted. The line indicates the cutoff for positivity (mean + 2SD). (B) Plot showing the bimodal distribution of METDex14/MET-wt nCounter log-ratios in the cohort. The line indicates the cutoff for positivity (mean + 2SD). (C) Heatmap displaying METDex14-positive samples by nCounter (N = 13) and correlative results for RT–PCR, mRNA expression (nC), MET mutations (NGS), copy number alterations (FISH or NGS), immunohistochemistry (IHC), and other co-occurring driver alterations. Patient numbers are shown in the top row. FISH, fluorescence in situ hybridization; HS, histoscore; IHC, immunohistochemistry; Mod, moderately; N/A nonavailable data; nC, nCounter; NGS, next-generation sequencing; PR, partial response; RT–PCR, reverse transcription polymerase chain reaction; SD, stable disease. [Correction added on 9 December 2020, after first online publication: Missing details in the figure were amended.] showed very strong MET by IHC (HS ≥ 220), 19 and KRAS mutations or ALK rearrangements. Next, intermediate (HS 100–220), 21 weak (HS 50–99), and we compared MET mRNA expression in samples at 17 were very weak-negative (HS < 50). In general, a baseline vs. progression. Overall, there were no signifi- fair agreement was observed between IHC and nCoun- cant differences in log-MET values between the 306 ter expression levels, with IHC positivity observed basal and the 52 progression samples included in the among cases with both mod-elevated and very-high study. When classified by drivers, we did observe a sig- mRNA levels (Tables S7 and S8). nificant increase in MET mRNA levels in tumors at progression vs. baseline only in EGFR-mu tant sam- ples. However, a subpopulation of samples at progres- 3.7. Expression levels of MET mRNA in clinical sion with high MET mRNA expression was apparent samples with driver alterations not only in EGFR mutant, but also in ALK-positive Finally, we compared MET mRNA levels by nCounter rebiopsies. in patients harboring different drivers (Fig. S2). Base- line MET expression by nCounter was found to be 4. Discussion significantly increased in samples with MET amplifica- tion, METDex14, and BRAF mutations (P < 0.05 in a In this study, we comprehensively characterized the Mann–Whitney U test), but not in cases with EGFR biologically relevant MET alterations—amplification 356 Molecular Oncology 15 (2021) 350–363 ª 2020 The Authors. Published by FEBS Press and John Wiley & Sons Ltd. C. Aguado et al. MET in advanced non-small cell lung cancer A C B Fig. 3. Quantification of MET mRNA expression levels by nCounter. (A) MET normalized counts obtained by nCounter, expressed as log- MET. The cutoff values for moderately elevated MET mRNA expression (mean + SD) and very-high MET mRNA expression (mean + 2SD) are indicated with lines. (B) Plot showing the unimodal distribution of log-MET in the sample cohort. (C) Heatmap displaying samples with very-high levels of MET mRNA expression by nCounter (N = 15) and corresponding results for METDex14 (nC), RT–PCR, MET mutations (NGS), copy number alterations (FISH or NGS), immunohistochemistry (IHC), and other co-occurring driver alterations. Patient numbers are shown in the top row. FISH, fluorescence in situ hybridization; GCN, gene copy number; HS, histoscore; IHC, immunohistochemistry; Mod, moderately; N/A nonavailable data; nC, nCounter; NGS, next-generation sequencing; PR, partial response; RT–PCR, reverse transcription polymerase chain reaction; SD, stable disease and exon 14 skipping—in a large cohort of 474 negative. In addition, most of them were found to har- NSCLC samples using of a RNA-based technology bor known drivers, particularly KRAS. Although RT– (nCounter; Fig. 5) and we systematically compared the PCR is used in some laboratories to identify results with other currently available methods for METΔex14 transcripts, these results suggest that it MET testing. We also propose an algorithm for the might not be the most adequate technique for this pur- selection of patients to be considered for MET-TKI pose. Low level of METDex14 transcripts can emerge treatment, based on the RNA-based results. by splicing ‘mistakes’ in the cell without translating in Our prevalence of METDex14 by nCounter (3%) is any genomic alteration with oncogenic relevance [34– in line with the literature [6,7], particularly when other 36]. These splicing ‘mistakes’ would test positive by RNA-based techniques such as RNA sequencing or mRNA amplification (RT–PCR), but not by nCoun- quantitative RT–PCR were used [32,33]. In our series, ter, DNA-, or RNA-based NGS. all cases positive for METDex14 transcripts by nCoun- DNA-based NGS using commercially available pan- ter (N = 13) were also detected by RT–PCR; however, els is also employed to determine METΔex14. How- a large subset of samples positive by RT–PCR ever, this technique has been reported to detect only (N = 11) tested negative by nCounter. Five of these 63% of literature-described splicing mutations associ- discordant samples could be submitted to an orthogo- ated with METDex14 [37], and reanalysis of 232 pan- nal RNA-based NGS, which confirmed them as negative samples by DNA-based NGS using RNA Molecular Oncology 15 (2021) 350–363 ª 2020 The Authors. Published by FEBS Press and John Wiley & Sons Ltd. 357 MET in advanced non-small cell lung cancer C. Aguado et al. Fig. 4. Heatmap displaying samples with mod-elevated levels of MET mRNA expression by nCounter (n = 36). Corresponding results for METDex14 (nC) RT–PCR, MET mutations (NGS), copy number alterations (FISH or NGS), immunohistochemistry (IHC), and other co- occurring driver alterations are also shown. Patient numbers are shown in the top row. FISH, fluorescence in situ hybridization; GCN, gene copy number; HS, histoscore; IHC, immunohistochemistry; Mod, moderate; N/A nonavailable data; nC, nCounter; NGS, next-generation sequencing; RT–PCR, reverse transcription polymerase chain reaction. sequencing revealed six (2.5%) METDex14-positive a prospective trial [15–17] indicate that it does affect cases previously missed [33]. In our cohort, among the response to MET-TKIs, endorsing METΔex14 as a eight tumors with METDex14 by nCounter that were truly separate driver in NSCLC. Indeed, all submitted to DNA-based NGS, three (37.5%) tested METΔex14 patients in our cohort treated with MET- negative for mutations associated with METDex14 TKIs derived clinical benefit, regardless of MET splicing sites. One of these discordant patients was amplification status (Patients 2, 3, 4, 6, 7; Fig. 2C). treated with a MET-TKI and derived clinical benefit In addition to the presence of METΔex14 tran- (Patient 7, Fig. 2C). It is probable that at least some scripts, we analyzed wt MET mRNA and could iden- of these cases involved large genomic deletions that tify different clusters of patients according to MET are difficult to detect on DNA-based NGS assays [32]. expression levels. The group with very-high MET Therefore, our results confirm that DNA-based tech- mRNA (3.5%) comprised a majority of males, former, niques might underestimate the detection of MET- or current smokers with a median age of 58 years. Dex14 alterations. Interestingly, very-high MET mRNA expression clo- Consistent with other studies [6,7,13–15,38], most of sely correlated with MET gene amplification by NGS the MET Δex14 patients detected by nCounter were or FISH and was mutually exclusive with METDex14. elderly females, with a considerable proportion of Therefore, it is not surprising that the phenotype of never smokers (38%; Table 1). In addition, as previ- the patients with very-high MET mRNA agrees with ously reported, METΔex14 rarely coexisted with other the characteristics previously reported in patients with drivers and was associated with moderate but not high high MET amplification (defined as FISH gene copy MET expression by IHC [6,38]. Finally, we only number, GCN ≥ 6 or ≥ 10) [16,17,39,40]. Further- detected MET amplification by FISH or NGS in one more, all the five very-high MET mRNA cases treated (14%) of the METΔex14-positive cases, compared to with MET inhibitors achieved partial responses; the the 8–30% reported in the literature using different majority of case baseline (7/9) did not harbor any detection techniques and thresholds [6,7,13,15,17]. The detectable driver. In contrast, the cohort of patients significance of MET amplification in the context of with mod-elevated MET mRNA expression did not METΔex14 alterations is unclear, but recent data from correlate with any specific phenotype, being distributed 358 Molecular Oncology 15 (2021) 350–363 ª 2020 The Authors. Published by FEBS Press and John Wiley & Sons Ltd. C. Aguado et al. MET in advanced non-small cell lung cancer Fig. 5. Graphical summary of the results obtained with nCounter. The numbers and percentages of evaluable samples with different MET mRNA expression are represented (N = 422). FISH, fluorescence in situ hybridization; METΔex14, MET exon 14 skipping mutation; Mod, moderately; TKI, tyrosine kinase inhibitor. independently of gender and smoking habits, and consequence of the different thresholds, methodologies, known drivers were detected in most cases. Although and scoring systems (MET/CEP7 ratios or GCN) that the biological oncogenic significance of MET amplifi- have been used as eligibility criteria. cation in NSCLC has been controversial [3,41], our Our study can shed some light on this relevant issue, results endorse the concept of MET being a true driver and the characterization of MET expression levels in in patients with very-high expression mRNA levels. In NSCLC here presented provides a unique opportunity contrast, we believe that MET is unlikely to play a to advance in the understanding of the processes clinically relevant role in tumors with mod-elevated underlying MET biology. For instance, our RNA- expression levels. based assessment allowed the identification of amplifi- Several trials have recently evaluated the efficacy of cation-negative tumors with very-high MET expression MET-TKIs in METΔex14-positive NSCLC. Overall levels and also the relatively infrequent cases with gene response rates of 32% have been reported for crizotinib, amplification that did not express MET mRNA. a nonselective inhibitor, compared to 46–68% and 40– Remarkably, one of the patients with very-high MET 55% in first or subsequent lines for the selective inhibi- mRNA levels but negative by FISH (using the two tors capmatinib, tepotinib, and savolitinib [11,13,14]. In usual cutoff values) was treated with a MET inhibitor contrast, the results obtained so far in trials enrolling achieving a partial response. These observations sug- MET-amplified patients are significantly worse, with gest that FISH positivity might be capturing a hetero- overall response rate (ORR) 20–47% geneous group regarding dependence on aberrant [16,17,20,39,40,42]. It is unclear whether the variety of MET signaling and that RNA-based techniques could responses observed in trials of MET-TKIs may partly improve the performance of FISH for patient selec- underlie a heterogeneous disease population with dis- tion. This, in turn, may help to explain the inferior tinct sensitivities to MET-TKIs and/or be a response rates observed with MET-TKIs in FISH- Molecular Oncology 15 (2021) 350–363 ª 2020 The Authors. Published by FEBS Press and John Wiley & Sons Ltd. 359 MET in advanced non-small cell lung cancer C. Aguado et al. amplified tumors. In order to improve outcomes to Conflict of interest MET inhibition in this setting, it would be important to further investigate the efficacy of MET-TKI in CT has received fees for consultancy/advisory roles those tumors with gene amplification that do not from Pfizer, Novartis, MSD, Roche, AstraZeneca, and express MET mRNA, and the real clinical value of Takeda, and research funding from Pfizer and Novar- high expression levels in amplification-negative tumors tis. CC has received travel grants from MSD, Pierre- where MET inhibition therapy could be considered. Fabre Oncology, fees for consultancy/advisory from Regarding METDex14, recent studies have suggested Boehringer Ingelheim, Roche, Pfizer, and Boehringer that MET protein expression is required for clinical ben- Ingelheim. LLV has received fees for consultancy/advi- efit from MET-TKIs [43]. Interestingly, the only MET- sory roles from AstraZeneca and Roche. SV has Dex14 patient in our cohort who exhibited stable disease received fees for consultancy/advisory roles from Bris- to MET-targeted therapy had low levels of mRNA by tol-Myers Squibb, Roche, MSD, Abbvie, Ose Pharma, nCounter, whereas all treated patients with mod-ele- and Merck. AAH has received fees for consultancy/ad- vated levels had partial responses. Whether mRNA visory roles from Bristol-Myers Squibb, Roche, MSD, expression levels can assist to predict outcome in and Lilly. MGC has received fees for consultancy/ad- patients with METDex14 is currently unknown but also visory roles from Bristol-Myers Squibb, Roche, Astra- merits further investigation. One of the limitations of Zeneca, and Pierre-Fabre Oncology, and research our study derives from the fact that MET is known to funding from AstraZeneca. AP has received fees for be expressed not only in tumor cells but also in normal consultancy/advisory roles from Pfizer, Lully, Nanos- epithelial [44], dendritic, and other immune cells [45]. In tring Technologies, Amgen, Oncolytics Biotech, Dai- consequence, particularly in cases with important ichi Sankyo, PUMA, Bristol-Myers Squibb, Novartis, stroma and/or inflammatory component, the mRNA and Daiichi Sankyo, and research funding from Pfizer, expression levels obtained by nCounter might reflect the Amgen, Roche, Novartis, and Daiichi Sankyo. NR level of MET expression in the whole tumor rather than has received fees for consultancy/advisory roles from only in cancer cells. However, the good agreement MSD, BMS and Pfizer, and research funding from Pfi- observed between MET mRNA levels by nCounter and zer, Novartis, and Ministry of Health, Instituto de MET IHC staining in tumor cells indicated that, in most Salud Carlos III, Spain. CA, RR, RRe, AGC, EM, samples, the contribution of noncancer components to NV, SC, SM, AA, IS, EA, SR, APR, IM, AFC, RP, the nCounter results was not significant. CS, MMG, MDL, AMB, EG, WPJL, RR, LMM, and MAMV have declared no conflicts of interest. 5. Conclusions Author contributions We have comprehensively characterized MET in a large cohort of advanced NSCLC and have validated the use CA, CT, RR, RRe, AGC, EM, MAMV, and NR con- of nCounter to identify METDex14 in this malignancy. ceived and designed the study. CA, CT, RR, RRe, Our work also provides useful insights into the biology AGC, EM, SM, AA, EA, SR, MAMV, and NR con- of MET as a driver in NSCLC, supporting MET very- ducted the experiments, acquired, and analyzed data. high mRNA expression as a surrogate of amplification, CC, NV, SC, SM, LLV, IS, IM, SV, AFC, RP, CS, and suggesting the relevance of MET mRNA levels in MMG, MDL, AAH, AMB, MGC, EG, WPJL, RRo, patients responding to MET-TKIs. Our results support LMM, and AP provided clinical samples and collected the use of mRNA-based techniques for multiplex, accu- clinical information. NR, CT, CA, and MAMV wrote rate, and reliable assessment of MET alterations in the manuscript. All authors reviewed and/or revised order to select patients for MET-targeted therapies. the manuscript. Acknowledgements Peer Review This study has received research funding from Pfizer, The peer review history for this article is available at Novartis, and the Ministry of Health, Instituto de https://publons.com/publon/10.1002/1878-0261.12861. Salud Carlos III, Spain (FIS PI16/00890), to NR; and from the European Union Eurostars program (EUS- Data accessibility 20159616). Additionally, the participation of CIMA has been founded by LMM, FIS PI19/00098, and All data generated or analyzed during this study are AECC grant (GCB14-2170). included in this manuscript and its supplementary 360 Molecular Oncology 15 (2021) 350–363 ª 2020 The Authors. Published by FEBS Press and John Wiley & Sons Ltd. C. Aguado et al. MET in advanced non-small cell lung cancer information files or are available from the authors 12 Frampton GM, Ali SM, Rosenzweig M, Chmielecki J, upon reasonable request. Lu X, Bauer TM, Akimov M, Bufill JA, Lee C, Jentz D et al. (2015) Activation of MET via diverse exon 14 splicing alterations occurs in multiple tumor types and References confers clinical sensitivity to MET inhibitors. Cancer 1 Guo R, Luo J, Chang J, Rekhtman N, Arcila M & Discov 5, 850–859. Drilon A (2020) MET-dependent solid tumours - 13 Paik PK, Felip E, Veillon R, Sakai H, Cortot AB, molecular diagnosis and targeted therapy. Nat Rev Clin Garassino MC, Mazieres J, Viteri S, Senellart H, Van Oncol 17, 569–587. Meerbeeck J et al. (2020) Tepotinib in non-small-cell 2 Drilon F, Cappuzzo SI & Ou DR (2017) Camidge, lung cancer with MET exon 14 skipping mutations. N targeting MET in lung cancer: will expectations finally Engl J Med 383, 931–943. be MET? J Thorac Oncol 12, 15–26. 14 Drilon JW, Clark J, Weiss SI, Ou DR, Camidge BJ, 3 Duplaquet L, Kherrouche Z, Baldacci S, Jamme P, Solomon GA, Otterson LC, Villaruz GJ, Riely RS, Cortot AB, Copin MC & Tulasne D (2018) The Heist MM et al. (2020) Paik, antitumor activity of multiple paths towards MET receptor addiction in crizotinib in lung cancers harboring a MET exon 14 cancer. Oncogene 37, 3200–3215. alteration. Nat Med 26, 47–51. 4 Recondo G, Che J, Janne PA & Awad MM (2020) 15 Wolf J, Seto T, Han J-Y, Reguart N, Garon EB, Groen Targeting MET dysregulation in cancer. Cancer Discov HJM, Tan DS-W, Hida T, Jonge MJD, Orlov SV et al. 10, 922–934. (2019) Capmatinib (INC280) in METDex14-mutated 5 Wang Q, Yang S, Wang K & Sun SY (2019) MET advanced non-small cell lung cancer (NSCLC): efficacy inhibitors for targeted therapy of EGFR TKI-resistant data from the phase II GEOMETRY mono-1 study. J lung cancer. J Hematol Oncol 12, 63. Clin Oncol 37, 9004. 6 Awad MM, Oxnard GR, Jackman DM, Savukoski DO, 16 Wolf J, Overbeck TR, Han J-Y, Hochmair M, Marinis Hall D, Shivdasani P, Heng JC, Dahlberg SE, Janne FD, Ohashi K, Smit EF, Power D, Garon EB, Groen PA, Verma S et al. (2016) MET exon 14 mutations in HJM et al. (2020) Capmatinib in patients with high- non-small-cell lung cancer are associated with advanced level MET-amplified advanced non–small cell lung age and stage-dependent MET genomic amplification cancer (NSCLC): results from the phase 2 and c-Met overexpression. J Clin Oncol 34, 721–730. GEOMETRY mono-1 study. J Clin Oncol 38, 9509. 7 Schrock AB, Frampton GM, Suh J, Chalmers ZR, 17 Wolf J, Seto T, Han JY, Reguart N, Garon EB, Groen Rosenzweig M, Erlich RL, Halmos B, Goldman J, HJM, Tan DSW, Hida T, de Jonge M, Orlov SV et al. Forde P, Leuenberger K et al. (2016) Characterization (2020) Investigators, capmatinib in MET exon 14- of 298 patients with lung cancer harboring MET exon mutated or MET-amplified non-small-cell lung cancer. 14 skipping alterations. J Thorac Oncol 11, 1493–1502. N Engl J Med 383, 944–957. 8 Tong JH, Yeung SF, Chan AW, Chung LY, Chau SL, 18 Dhillon S (2020) Capmatinib: first approval. Drugs 80, Lung RW, Tong CY, Chow C, Tin EK, Yu YH et al. 1125–1131. (2016) MET amplification and exon 14 splice site 19 Recondo G, Bahcall M, Spurr LF, Che J, Ricciuti B, mutation define unique molecular subgroups of non- Leonardi GC, Lo YC, Li YY, Lamberti G, Nguyen T small cell lung carcinoma with poor prognosis. Clin et al. (2020) Molecular mechanisms of acquired Cancer Res 22, 3048–3056. resistance to MET tyrosine kinase inhibitors in patients 9 Kong-Beltran M, Seshagiri S, Zha J, Zhu W, Bhawe K, with MET exon 14-mutant NSCLC. Clin Cancer Res Mendoza N, Holcomb T, Pujara K, Stinson J, Fu L 26, 2615–2625. et al. (2006) Somatic mutations lead to an oncogenic 20 Sequist LV, Han JY, Ahn MJ, Cho BC, Yu H, Kim deletion of met in lung cancer. Cancer Res 66, 283–289. SW, Yang JC, Lee JS, Su WC, Kowalski D et al. 10 Zhao X, Weir BA, LaFramboise T, Lin M, Beroukhim (2020) Osimertinib plus savolitinib in patients with R, Garraway L, Beheshti J, Lee JC, Naoki K, Richards EGFR mutation-positive, MET-amplified, non-small- WG et al. (2005) Homozygous deletions and cell lung cancer after progression on EGFR tyrosine chromosome amplifications in human lung carcinomas kinase inhibitors: interim results from a multicentre, revealed by single nucleotide polymorphism array open-label, phase 1b study. Lancet Oncol 21, 373– analysis. Cancer Res 65, 5561–5570. 386. 11 Paik PK, Drilon A, Fan PD, Yu H, Rekhtman N, 21 Wang Y, Li L, Han R, Jiao L, Zheng J & He Y (2018) Ginsberg MS, Borsu L, Schultz N, Berger MF, Rudin Clinical analysis by next-generation sequencing for CM et al. (2015) Response to MET inhibitors in NSCLC patients with MET amplification resistant to patients with stage IV lung adenocarcinomas harboring osimertinib. Lung Cancer 118, 105–110. MET mutations causing exon 14 skipping. Cancer 22 Bubendorf L, Dafni U, Schobel M, Finn SP, Tischler Discov 5, 842–849. V, Sejda A, Marchetti A, Thunnissen E, Verbeken EK, Molecular Oncology 15 (2021) 350–363 ª 2020 The Authors. Published by FEBS Press and John Wiley & Sons Ltd. 361 MET in advanced non-small cell lung cancer C. Aguado et al. Warth A et al. (2017) Prevalence and clinical 33 Benayed R, Offin M, Mullaney K, Sukhadia P, Rios K, association of MET gene overexpression and Desmeules P, Ptashkin R, Won H, Chang J, Halpenny amplification in patients with NSCLC: results from the D et al. (2019) High yield of RNA sequencing for European Thoracic Oncology Platform (ETOP) targetable kinase fusions in lung adenocarcinomas with Lungscape project. Lung Cancer 111, 143–149. no mitogenic driver alteration detected by DNA 23 Noonan SA, Berry L, Lu X, Gao D, Baron AE, sequencing and low tumor mutation burden. Clin Chesnut P, Sheren J, Aisner DL, Merrick D, Doebele Cancer Res 25, 4712–4722. RC et al. (2016) Identifying the appropriate FISH 34 Pineda JMB & Bradley RK (2018) Most human introns criteria for defining MET copy number-driven lung are recognized via multiple and tissue-specific adenocarcinoma through oncogene overlap analysis. J branchpoints. Genes Dev 32, 577–591. Thorac Oncol 11, 1293–1304. 35 Egecioglu DE & Chanfreau G (2011) Proofreading and 24 Schildhaus HU, Schultheis AM, Ruschoff J, Binot E, spellchecking: a two-tier strategy for pre-mRNA Merkelbach-Bruse S, Fassunke J, Schulte W, Ko YD, splicing quality control. RNA 17, 383–389. Schlesinger A, Bos M et al. (2015) MET amplification 36 Kurosaki T & Maquat LE (2016) Nonsense-mediated status in therapy-naive adeno- and squamous cell mRNA decay in humans at a glance. J Cell Sci 129, carcinomas of the lung. Clin Cancer Res 21, 907–915. 461–467. 25 Guo R, Luo J, Chang J, Rekhtman N, Arcila M & 37 Poirot B, Doucet L, Benhenda S, Champ J, Meignin V Drilon A (2020) MET-dependent solid tumours – & Lehmann-Che J (2017) MET exon 14 alterations and molecular diagnosis and targeted therapy. Nat Rev Clin new resistance mutations to tyrosine kinase inhibitors: Oncol 17, 569–587. risk of inadequate detection with current amplicon- 26 Warren S (2018) Simultaneous, multiplexed detection of based NGS panels. J Thorac Oncol 12, 1582–1587. RNA and protein on the NanoString((R)) nCounter 38 Baldacci S, Figeac M, Antoine M, Descarpentries C, ((R)) platform. Methods Mol Biol 1783, 105–120. Kherrouche Z, Jamme P, Copin MC, Tulasne D, Nanni 27 Marin E, Teixido C, Carmona-Rocha E, Reyes R, I, Beau-Faller M et al. (2020) High MET overexpression Arcocha A, Vinolas N, Rodriguez-Mues M, Cabrera C, does not predict the presence of MET exon 14 splice Sanchez M, Vollmer I et al. (2020) Usefulness of two mutations in NSCLC: results From the IFCT independent DNA and RNA tissue-based multiplex PREDICT.amm study. J Thorac Oncol 15, 120–124. assays for the routine care of advanced NSCLC 39 Bang YJ, Su WC, Schuler M, Nam DH, Lim WT, patients. Cancers (Basel) 12, 1124. Bauer TM, Azaro A, Poon RTP, Hong D, Lin CC 28 Park K, Choi MK, Jung HH, Do IG, Lee KH, Ahn T, et al. (2020) Phase 1 study of capmatinib in MET- Kil WH, Kim SW, Lee JE, Nam SJ et al. (2015) positive solid tumor patients: dose escalation and Molecular characterization of patients with pathologic expansion of selected cohorts. Cancer Sci 111, 536– complete response or early failure after neoadjuvant 547. chemotherapy for locally advanced breast cancer using 40 Schuler M, Berardi R, Lim WT, de Jonge M, Bauer next generation sequencing and nCounter assay. TM, Azaro A, Gottfried M, Han JY, Lee DH, Wollner Oncotarget 6, 24499–24510. M et al. (2020) Molecular correlates of response to 29 Kim ST, Do IG, Lee J, Sohn I, Kim KM & Kang WK capmatinib in advanced non-small-cell lung cancer: (2015) The NanoString-based multigene assay as a clinical and biomarker results from a phase I trial. Ann novel platform to screen EGFR, HER2, and MET in Oncol 31, 789–797. patients with advanced gastric cancer. Clin Transl Oncol 41 Castiglione R, Alidousty C, Holz B, Wagener S, Baar 17, 462–468. T, Heydt C, Binot E, Zupp S, Kron A, Wolf J et al. 30 Lee J, Ou SH, Lee JM, Kim HC, Hong M, Kim SY, (2019) Comparison of the genomic background of Jang J, Ahn S, Kang SY, Lee S et al. (2015) MET-altered carcinomas of the lung: biological Gastrointestinal malignancies harbor actionable MET differences and analogies. Mod Pathol 32, 627–638. exon 14 deletions. Oncotarget 6, 28211–28222. 42 Wu YL, Cheng Y, Zhou J, Lu S, Zhang Y, Zhao J, 31 Reguart N, Teixido C, Gimenez-Capitan A, Pare L, Kim DW, Soo RA, Kim SW, Pan H et al. (2020) Galvan P, Viteri S, Rodriguez S, Peg V, Aldeguer E, Tepotinib plus gefitinib in patients with EGFR-mutant Vinolas N et al. (2017) Identification of ALK, ROS1, and non-small-cell lung cancer with MET overexpression or RET fusions by a multiplexed mRNA-based assay in MET amplification and acquired resistance to previous formalin-fixed, paraffin-embedded samples from advanced EGFR inhibitor (INSIGHT study): an open-label, non-small-cell lung cancer patients. Clin Chem 63, 751–760. phase 1b/2, multicentre, randomised trial. Lancet Respir 32 Davies KD, Lomboy A, Lawrence CA, Yourshaw M, Med 8, 1132–1143. Bocsi GT, Camidge DR & Aisner DL (2019) DNA-based 43 Guo R, Offin M, Rose-Brannon A, Chow A & Drilon versus RNA-based detection of MET exon 14 skipping AE (2019) MET inhibitor resistance in patients with events in lung cancer. J Thorac Oncol 14, 737–741. MET exon 14-altered lung cancers. J Clin Oncol 37, 1. 362 Molecular Oncology 15 (2021) 350–363 ª 2020 The Authors. Published by FEBS Press and John Wiley & Sons Ltd. C. Aguado et al. MET in advanced non-small cell lung cancer 44 Di Renzo MF, Narsimhan RP, Olivero M, Bretti S, Table S1. MET amplification criteria used in clinical Giordano S, Medico E, Gaglia P, Zara P & Comoglio trials of MET inhibitors. PM (1991) Expression of the Met/HGF receptor in Table S2. Description of the NGS panels used in the normal and neoplastic human tissues. Oncogene 6, study. 1997–2003. Table S3. Description of the nCounter codeset used in 45 Kurz SM, Diebold SS, Hieronymus T, Gust TC, the study and probes design for MET wild type and Bartunek P, Sachs M, Birchmeier W & Zenke M (2002) METDex14. The impact of c-met/scatter factor receptor on dendritic Table S4. Characteristics of the samples with valid cell migration. Eur J Immunol 32, 1832–1838. results, N (%). Fusions were determined by nCounter, mutations by NGS. Supporting information Table S5. Results for MET exon 14 alteration splice- site regions and mutation types. Additional supporting information may be found Table S6. Concordance of nCounter with RT-PCR online in the Supporting Information section at the end and NGS for METexD14 status. of the article. Table S7. Concordance of the nCounter categorization Fig. S1. Heatmap displaying the molecular and clinical using the cut-off for very high MET levels with IHC, characteristics samples with METDex14 discordant FISH and NGS. results between RT-PCR and nCounter. Table S8. Concordance of the nCounter categorization Fig. S2. MET mRNA expression levels by nCounter in using the cut-off for moderately elevated MET levels samples harboring different driver alterations. with IHC, FISH and NGS. Molecular Oncology 15 (2021) 350–363 ª 2020 The Authors. Published by FEBS Press and John Wiley & Sons Ltd. 363