Co-occurring Alterations in the RAS–MAPK Pathway Limit Response to MET Inhibitor Treatment in MET Exon 14 Skipping Mutation-Positive Lung Cancer

Somatic MET mutations leading to splicing-mediated loss of exon 14 and subsequent MET overexpression are an emerging therapeutic target present in 2% to 4% of lung adenocarcinomas (1, 2). MET tyrosine kinase inhibitor (TKI) treatment was associated with improved overall survival in a retrospective study of patients with METex14-mutated NSCLC (3). In ongoing prospective studies, response rates of 32% to the multikinase inhibitor crizotinib, 42% to the MET TKI tepotinib, and up to 71% for treatment-naïve patients to the MET TKI capmatinib have been reported (4–6). Crizotinib has recently received FDA breakthrough designation for use in the treatment of METex14-mutated NSCLC.

Although MET remains an attractive therapeutic target, both primary and acquired resistance limit the long-term survival of patients with METex14-mutated NSCLC. Second-site MET mutations and downstream signaling reactivation via acquired KRAS amplification have been reported at acquired resistance to MET TKI therapy, and may inform treatment decisions (7–10). By contrast, the mechanisms mediating both primary MET TKI resistance and tumor cell persistence during initial MET TKI treatment remain largely undefined, as does the full landscape of alterations promoting acquired resistance.

Recent studies show that advanced-stage NSCLCs often harbor multiple oncogenic alterations, which may impact response to targeted therapies (11–13). A prior 28-patient cohort describing co-occurring genomic alterations as measured by tissue biopsy demonstrated variability in frequency of co-occurring alterations between METex14-mutated NSCLC and NSCLC driven by other genomic alterations (14). Analysis of cell-free circulating tumor DNA (cfDNA) utilizing next-generation sequencing (NGS) provides another avenue to describe the genomic landscape within a cancer patient and offers the potential to capture genomic changes reflecting heterogeneity across distinct metastatic tumor sites (11, 15). A more detailed understanding of the mutational landscape that coexists with oncogenic MET alterations in NSCLC may facilitate an improved understanding of the determinants of MET TKI response and identify rational polytherapy strategies to improve clinical outcomes.

Here, we describe the spectrum of co-occurring genomic alterations observed within the cfDNA of patients with METex14-mutated, advanced-stage NSCLC and identify prominent co-alteration of RAS pathway genes as a contributing factor to disease progression.

This study and waiver of written consent were approved by the Institutional Review Board (IRB) at the University of California, San Francisco. The conduct of this research was in accordance with the U.S. Common Rule. The cfDNA analysis included 332 consecutive samples from 289 patients with advanced (stage IIIB/IV) NSCLC with a METex14 mutation obtained between October 2015 and March 2018, and 1653 consecutively tested samples from 1,489 patients with EGFR-mutated NSCLC (L858R and del19) obtained between April 2016 and May 2017, as well as previously published (11) cohorts of EGFR wild type and TKI-naïve EGFR-mutated NSCLC (Supplementary Table S1).

Ba/F3 cells were purchased from ATCC (ATCC HB-283) and maintained in culture for a total of approximately 2 to 3 months in DMEM supplemented with 1 ng/mL IL3 (Peprotech). Mycoplasma testing was not performed. Retrovirus was generated using TransIT-LT1 transfection reagent (Mirus). Cells were infected with filtered retrovirus, expressing either the ORF control mCherry, human wildtype MET, or human MET^ex14 in a pBABE-puro vector backbone as described previously (9) and selected in puromycin (2 μg/mL). Expression was confirmed by immunoblotting. KRAS-overexpressing cells were obtained by retroviral infection with a pBABE-hygro KRAS4B construct and selected in hygromycin (800 μg/mL). Knockdown of NF1 was achieved by lentiviral transduction of the following sequence: 5′-CCGGGCCAACCTTAACCTCTCTAATCTCGAGATTAGAGAGGTTAAGGTTGGCTTTTTG-3′, expressed from a pLKO.1-hygro plasmid backbone (Addgene #24150). Three days after lentiviral transduction, cells were selected via treatment with hygromycin B and knockdown was confirmed by immunoblotting. All drugs were purchased from Selleck Chemicals.

Transformation assay was performed by removing IL3 through centrifugation and adding 50 ng/mL human HGF (Peprotech 100–39 H). For proliferation assays cells were seeded in 96-well plates at 5,000 cells/well and the following day were exposed to crizotinib (Selleck Chemicals, #S1068) at 0 to 10 μmol/L and/or trametinib (Selleck Chemicals, #S2673) at 0.01 μmol/L. After 72 hours of drug exposure, CellTiter-Glo (Promega) reagent was added and luminescence was measured on a Spectramax spectrophotometer (Molecular Devices) according the manufacturer's instructions. Data are presented as percentage of viable cells compared with control cells (vehicle treatment).

Cells were washed in PBS and lysed with 25 mmol/L Tris-HCL (pH 7.6), 150 mmol/L NaCl, 1% NP-40, 1% sodium deoxycholate, 0.1% SDS supplemented with Halt Protease Inhibitor Cocktail (Thermo Fisher Scientific), and Halt Phosphatase Inhibitor Cocktail (Thermo Fisher Scientific). Lysates were separated in a 4% to 15% SDS-PAGE gel and transferred onto a nitrocellulose membrane (Bio-Rad). Membranes were blocked with 5% FBS in Tris-buffered saline (TBS) containing 0.1% Tween and incubated with the appropriate antibodies. Detection was performed via ECL Prime (Amersham Biosciences). Antibodies against the following were obtained from Cell Signaling Technology and were used at a dilution of 1:1,000: MET (#3148), p-MET Y1349 (#3121), pMEK S217/221 (#9121), ERK1/2 (#3493), p-ERK1/2 T202/Y204 (#9106), HSP90 (#4874), PARP (#9546), NF1 (#14623), horseradish peroxidase (HRP)-conjugated anti-mouse (#7076), and HRP-conjugated anti-rabbit (#7074). The following antibody was obtained from EMD Millipore: RAS (05-516, 1:2,000 dilution). Detection was performed via ECL Prime (Amersham Biosciences).

Samples were shipped to a Clinical Laboratory Improvement Act (CLIA)-certified, College of American Pathologists–accredited laboratory (Guardant Health). cfDNA was extracted from whole blood collected in 10-mL Streck tubes. After double centrifugation, 5 to 30 ng of cfDNA was isolated for digital sequencing of either a 70- or 73-gene panel (Supplementary Table S2) as described previously (16). Only those genes in common to these 2 panels were included in subsequent analysis. Nonsynonymous mutations were further processed with the R statistical computing program (version 3.3). Variants with unknown or neutral predicted functional significance were filtered prior to analysis as described previously (11) to include those with known impact on gene function within the Cosmic (17), GENIE (18, 19), and ClinVar (20) databases. Those remaining with unknown functional impact after review of these databases were included if with predicted functional impact utilizing the Mutation Assessor (version 3) algorithm (21). Mutations previously reported as associated with clonal hematopoiesis were also excluded (22). Assignment as clonal or subclonal was performed by normalized mutational allele frequency to percentage detected using a cut off of 0.2 as described previously (11). Residue numbering was standardized to MET UniProtKB-P08581.

Tumor sample NGS was performed in CLIA-approved laboratories. The Foundation One and Foundation ACT assays are commercially available assays, which were used in the clinical standard-of-care setting. The UCSF500 assay sequences 479 cancer-associated genes to target 200× coverage, utilizing sequencing of a PBMC sample, target 100× coverage, as a control (23). The University of Florida GatorSeq NGS assay, utilized for both tumor tissue and PBMC analysis, performs sequencing of 76 cancer-associated genes with target 500× coverage (Supplementary Table S2). Germline mutations were subtracted utilizing sequencing of buccal swab samples to a target depth of 100×.

Isolated genomic DNA extracted from FFPE was amplified using ddPCR Supermix for Probes (Bio-Rad) using KRAS and MET assays (PrimePCR ddPCR Mutation Assay, Bio-Rad, and custom-designed). DNA template (8 μL) was added to 10 μL of ddPCR Supermix for Probes (Bio-Rad) and 2 μL of the primer/probe mixture. This 20-μL reaction mix was added to a DG8 cartridge together with 70 μL of Droplet Generation Oil for Probes (Bio-Rad) and used for droplet generation. Droplets were then transferred to a 96-well plate (Eppendorf) and then thermal cycled with the following conditions: 5 minutes at 95°C, 40 cycles of 94°C for 30 seconds, 55°C for 1 minute followed by 98°C for 10 minutes (ramp rate 2°C/second). Droplets were analyzed with the QX200 Droplet Reader (Bio-Rad) for fluorescent measurement of FAM and HEX probes. Gating was performed based on positive and negative controls, and mutant populations were identified. The ddPCR data were analyzed with QuantaSoft analysis software (Bio-Rad) to obtain fractional abundance of the mutant DNA alleles in the wild-type (WT)/normal background. The quantification of the target molecule was presented as number of total copies (mutant plus WT) per sample in each reaction. Fractional abundance is calculated as follows: F.A. (%) = (N_mut/(N_mut + N_wt)) × 100), where N_mut is number of mutant events and N_wt is number of WT events per reaction. Multiple replicates were performed for each sample. ddPCR analysis of normal control genomic DNA (gene fragment obtained from IDT) and no DNA template (water) controls was performed in parallel with all samples, including multiple replicates as contamination-free controls.

Pairwise sample group comparisons for cfDNA analysis were carried out using a 2-tailed Fisher exact t test, with Benjamini–Hochberg correction for multiple comparisons, using a FDR of less than 0.2. For cell viability curves, comparisons were performed using the 2-sided student t test, with significance threshold of P value < 0.05.

We analyzed a cohort of 289 patients with advanced-stage NSCLC who had a METex14 mutation identified upon cfDNA analysis of 68 cancer-relevant genes using a clinically validated assay (Guardant360). This is the largest reported cohort to-date describing the genomic landscape of advanced-stage METex14-mutated NSCLC. We evaluated the frequency with which METex14 mutations co-occur with other cancer-associated mutations (11, 16). To focus our analysis on co-occurring mutations with potential functional impact, synonymous mutations and those with predicted neutral or unknown functional impact were excluded, as described previously (11), as were mutations previously associated with clonal hematopoiesis (Supplementary Fig. S1; ref. 22). 86.5% of samples contained co-occurring genomic alterations, with a mean of 2.74 alterations per sample (range 0–22), in addition to the METex14 mutation. We use the term genomic alterations here to include both gene mutations/rearrangements and copy number gain as detected by cfDNA. The most commonly altered genes, co-occurring with the METex14 mutation in at least 10% of patients, were TP53 (49.5% of patients), EGFR (16.3%), NF1 (neurofibromatosis-1; 15.6%), BRAF (10.7%), and CDK4 (10.4%). Additional MET gene alterations were also common in this patient population, in which 9.3% of patients had co-occurring MET copy number gain and 12.1% had a co-occurring MET mutation (Fig. 1A).

Seventeen of 34 second-site MET mutations were located in the tyrosine kinase domain. All identified MET second-site mutations, regardless of predicted functional impact, were included in this analysis. Of these, 14 (G1163R, L1195V/F, F1200I, D1228H/N/Y, and Y1230H/S) were located at residues previously associated with MET TKI resistance (7–9, 24–26). Nine of these occurred in patients with known prior MET TKI exposure and 5 in patients with unknown prior treatment history. The remaining 3 mutations (H1094Y, R1336W, and I1084L) occurred in patients without available prior treatment history. Although the MET H1094Y mutation is known to lead to MET activation (27), the effects of the other 2 MET mutations remain uncharacterized.

The co-occurrence of other established oncogenic driver alterations (KRAS, EGFR, ALK, ROS1, BRAF) was uncommon except for the presence of activating KRAS mutations in 5.2% of patients (G12C/D/S/V 3.5%, G13C 1%, Q22K 0.3%, and Q61H 0.3%), canonical EGFR-activating mutations in 3.5% (del19 3.1%, L858R 0.7%, T790M 2%), and an ALK gene rearrangement in 0.7%. In addition, a HER2 exon 20 insertion was detected in 1 patient. In a patient with known clinical outcomes data, with both EGFR del19/EGFR T790M mutations and a METex14 mutation, there was partial response (RECIST 1.1) to treatment with the EGFR TKI osimertinib, which lasted 13.8 months. The remaining alterations detected in KRAS, EGFR, ALK, ROS1, and BRAF reflected mutations which are not canonical-driver alterations as listed above and/or reflected copy number gain. Although in the overall dataset the METex14 mutation detected was predominantly clonal (80.6% of samples, defined as normalized MAF >0.2), in samples with a detectable co-occurring oncogenic driver alteration the METex14 alteration was more likely to be subclonal (21.4% of samples clonal, P value < 0.0001). The converse was also true; co-occurring established oncogenic driver mutations that were detected were more likely to be clonal than the other detected co-occurring genomic alterations (68.6% vs. 46.3%, P-value = 0.0144; Supplementary Fig. S2).

To understand how the genomic landscape in advanced-stage METex14-mutated NSCLC compares to NSCLCs with a different targetable oncogenic driver mutation associated with high upfront response rates to TKI therapy, we used identical cfDNA profiling to compare a cohort of patients with known METex14-mutated NSCLC to a previously unpublished, independent cohort of 1653 samples from 1,489 patients with advanced-stage EGFR-mutant (del19, L858R) NSCLC (Supplementary Table S1). This comparison demonstrated differential frequency of co-occurring genomic alterations in 17 genes (Fig. 2A). In the METex14-mutated cohort, co-occurring alterations in NF1, CDK4, STK11, ALK, KRAS, ATM, CDKN2A, NRAS, TSC1, and ESR1 were more commonly identified. In the EGFR-mutated cohort, co-occurring alterations in AR, ERBB2, CCNE1, PIK3CA, BRAF, CTNNB1, and MYC were more commonly identified, independently validating our prior findings identifying these as common co-occurring alterations in EGFR-mutated NSCLC (11). Among the 10 genes with a greater frequency of co-occurring genomic alteration in patients with a METex14 mutation, 3 (NF1, KRAS, and NRAS) are key components of the RAS–MAPK signaling pathway. When compared with a previously published, independent cohort of 918 patients with EGFR wild-type/METex14 negative NSCLC (28), alterations in NF1 remained significantly more common in METex14-mutated NSCLC. Genomic alterations in KRAS were significantly enriched in the EGFR wild-type/METex14 negative cohort compared with either the METex14- or EGFR-mutated cohorts, consistent with prior reported rates of KRAS gene alteration in NSCLC (Fig. 2A; ref. 12).

A propensity towards genomic alterations promoting downstream hyperactivation of the RAS pathway may favor primary resistance to TKI therapy and help explain the comparatively lower TKI response rates in METex14-mutated NSCLC. Examination of the subset of cfDNA samples obtained from patients with METex14-mutated NSCLC prior to known MET TKI treatment (n = 61) demonstrated high rates of genomic alterations capable of promoting RAS–MAPK pathway activation (37.7% of patients, Fig. 2B). Among those patients with a co-occurring RAS–MAPK pathway genomic alteration, 34.8% had more than one simultaneous alterations within this pathway (median 1, mean 1.53, range 1–5 RAS–MAPK genomic alterations). When compared with the subset of cfDNA samples from TKI-naïve patients with EGFR-mutated NSCLC (n = 58) derived from the dataset discussed here and from a previously published patient cohort (11), there remained a trend towards more common detection of RAS–MAPK pathway alterations before treatment in METex14-mutated NSCLC (37.7% vs. 19.0%, P-value 0.0297, q-value = 0.297 with correction for multiple hypothesis testing by Benjamini–Hochberg for FDR <20%) which was not present for other categories of gene alterations (Fig. 2C).

We identified 12 patients with cfDNA obtained following treatment with crizotinib and an available matched sample obtained prior to known crizotinib exposure (Fig. 3A and B; Table 1). Details regarding one patient (patient #5) in this dataset have previously been published by other groups (7, 9).

Newly detectable MET second-site mutations (Y1230H/S, D1228H/N, F1200I, L1195V) were detected in 4 of 12 patients following MET TKI treatment (Fig. 3C, Table 1). Although some of these mutations have been reported at acquired MET TKI resistance (7–9, 24–26), the MET F1200I has not yet been described in a patient sample. Identification of specific acquired second-site mutations at resistance to MET TKI therapy may inform treatment decisions. MET TKIs can be classified as type I TKIs (e.g., crizotinib, capmatinib) or type II TKIs (e.g., cabozantinib) based on the kinase domain conformation to which they bind and differ in their activity against second-site MET mutations. For example, the MET Y1230X and D1228X mutations develop at resistance to type I MET TKIs and may predict response to subsequent type II MET TKI treatment (7–9, 24–26). Conversely, the MET L1195 V mutation has been reported at resistance to type II MET TKI treatment (7).

The development of acquired MET F1200 mutations as a resistance mechanism to MET TKIs has previously been predicted in a preclinical drug resistance screen, in which MET F1200 mutations were the dominant resistance mechanism to a type II MET TKI and were observed, though less common, at resistance to a type I MET TKI (29). Molecular modeling studies suggest that MET F1200I alters the conformation of the kinase domain such that it interferes with both the binding of type II MET TKIs within the DFG-out binding pocket, and to a lesser extent, may promote type I MET TKI resistance through disruption of an autoinhibitory MET conformation (Supplementary Fig. S3; ref. 7). Moreover, the F1200 residue is conserved across multiple tyrosine kinases, including ALK, ROS1, NTRK, and ABL, in which mutations at the corresponding residue have been linked to TKI resistance (Supplementary Table S3; refs. 29–35).

In contrast, parallel and downstream pathway alterations with the potential to provide alternative input for RAS–MAPK pathway signaling were newly detectable in 8 of 12 patients following MET TKI treatment as compared with cfDNA samples obtained prior to MET TKI treatment (Fig. 3C; Table 1). In addition to 2 patients with KRAS amplification, which was recently implicated in MET TKI resistance (10), an NF1 frameshift mutation, and copy number gain in EGFR and KIT were identified. In one patient with acquired KRAS amplification, addition of the MEK inhibitor trametinib to crizotinib treatment rapidly decreased detectable circulating tumor cfDNA. In the second patient with acquired KRAS amplification, an activating KRAS G12D mutation and amplification of the MET ligand HGF were also present within a progressing lesion on crizotinib. Further details regarding the 2 patients with acquired KRAS amplification are presented in Supplementary Fig. S4. Potential bypass pathway alterations coexisted with second-site MET alterations in 3 of 4 patients with second-site MET alterations, whereas alterations specifically within RAS-pathway members co-occurred with only 1 of 4 patients with second-site MET alterations.

The clinical data suggested that preexisting or acquired genomic changes leading to RAS–MAPK pathway activation (e.g., in KRAS, NF1) may limit response and induce resistance to MET TKI treatment in METex14-mutated NSCLC. In one patient with KRAS amplification detectable at acquired resistance to crizotinib treatment, combination treatment off-label with both crizotinib (250 mg po BID) and trametinib (2 mg po daily) resulted in rapid loss of both detectable METex14 and KRAS amplification by cfDNA suggestive of treatment response. However, despite molecular evidence of tumor response this combination therapy was poorly tolerated in the context of overall clinical decline, with fatigue, fluid retention, and diarrhea. Despite dose reduction to trametinib 2 mg every other day, the patient expired before radiographic response assessment (Supplementary Fig. S4).

To assess the functional impact of RAS–MAPK pathway hyperactivation on sensitivity to MET TKI treatment, we engineered a new Ba/F3 cell-based system. The IL3-dependent Ba/F3 cell line, while not of epithelial origin, is an established system to assess oncogenic capacity and putative drug resistance mechanisms (36, 37). Stable expression of human METex14 in Ba/F3 cells in the presence of the MET ligand human hepatocyte growth factor (HGF) induced IL3-independent growth (Supplementary Fig. S5A and S5B) and increased downstream Erk phosphorylation compared with expression of wild-type MET (Fig. 4C and F). METex14-mutant expressing cells were sensitive to treatment with crizotinib, as measured by reduced cell growth in standard cell viability assays (Fig. 4A and D).

Overexpression of wild-type KRAS or knockdown of wild-type NF1 in METex14-expressing Ba/F3 cells induced resistance to crizotinib (Fig. 4A and D). Overexpression of wild-type KRAS increased the IC₅₀ to crizotinib from 0.16 to 1.68 μmol/L (P-value < 0.001; Fig. 4A and G) and NF1 downregulation increased the IC₅₀ to crizotinib from 0.16 to 0.75 μmol/L (P-value < 0.001; Fig. 4D and H). Treatment of cells harboring both METex14 and KRAS overexpression with the combination of crizotinib and the MEK inhibitor trametinib in order to block both MET signaling and downstream MAPK pathway signaling restored sensitivity to treatment (crizotinib IC₅₀ of 1.68 μmol/L in the absence of trametinib versus 0.25 μmol/L with trametinib cotreatment, P-value < 0.001; Fig. 4B and G). Similarly, cotreatment of cells harboring both METex14 and NF1 downregulation with both trametinib and crizotinib restored sensitivity to treatment (crizotinib monotherapy IC₅₀ 0.75 μmol/L vs. 0.18 μmol/L upon addition of trametinib, P-value < 0.001; Fig. 4E and H). The selected trametinib dose modestly reduced but did not eliminate cell growth in the absence of crizotinib (Supplementary Fig. S5). Immunoblotting demonstrated sustained Erk phosphorylation despite crizotinib treatment in samples with KRAS overexpression or NF1 downregulation, which was abrogated by the addition of trametinib (Fig. 4C and F). In both genomic contexts, combination treatment was associated with increased levels of cleaved PARP, indicative of apoptosis, which was absent with monotherapy (Fig. 4C and F).

The challenge of therapeutic resistance in patients harboring METex14 mutations is of increasing clinical relevance given the emergence of MET-targeted therapies into the clinic. Although off-label use of MET TKI therapy has demonstrated clinical activity, objective reported response rates of approximately 30% to 70% as reported in early studies (4–6) are generally lower than those seen in response to TKI treatment in NSCLC driven by other canonical oncogenes (EGFR, ALK), where response rates greater than 80% have been reported (38, 39).

RAS–MAPK pathway hyperactivation has an established role in promoting resistance to EGFR, ALK, BRAF, and ROS1 targeted therapies via diverse mechanisms including KRAS amplification and KRAS, BRAF, and NF1 mutations (40–48). Although KRAS amplification has also recently been reported at acquired MET TKI resistance in METex14-mutated NSCLC (10), the broader role of compensatory genomic events providing signaling pathway re-activation remains less well-understood in METex14-mutated NSCLC. Here, we describe frequent co-occurring RAS–MAPK pathway alterations in METex14-mutated NSCLC as compared with EGFR-mutated NSCLC. RAS–MAPK pathway alterations were detected even in TKI-naïve patients; our preclinical and clinical data suggest these co-occurring alterations may promote resistance to MET TKI therapy. More specifically, RAS–MAPK pathway alterations when present as co-occurring genomic events in METex14-mutated NSCLC prior to MET TKI treatment may induce not only primary resistance but also contribute to tumor cell persistence, thus limiting response magnitude and potentially duration of response to initial treatment. This notion is supported by our findings in the Ba/F3 preclinical system we engineered. We also describe the spectrum and relative frequencies of newly detectable genomic alterations following MET TKI treatment as measured by cfDNA, which included alterations within the RAS–MAPK pathway or within RTKs upstream of the RAS–MAPK pathway (49) in two thirds of patients.

Although our reported dataset is limited by lack of complete clinical outcomes data it highlights the importance of developing future patient cohorts incorporating outcomes data to link the understanding of the genomic landscape to treatment response and prognosis, particularly for those patients with less common or emerging driver mutations. The recently reported association between MET protein expression and response to MET TKI therapy in METex14-mutated NSCLC (50) additionally raises the question of future need for assays incorporating information regarding protein expression to assist with therapeutic decision-making in this patient population and warrants prospective study.

Clinically, although many MET second-site mutations acquired during type I MET TKI (e.g., crizotinib) treatment may be overcome by use of type II MET TKIs (e.g., cabozantinib; refs. 7–9), the genomic alterations favoring RAS–MAPK pathway activation described here will likely require a combination therapy strategy. In our METex14-mutated preclinical model system, RAS–MAPK pathway hyperactivation via KRAS overexpression or NF1 downregulation induced MET TKI resistance that was overcome by the addition of the MEK inhibitor trametinib. Changes in the cfDNA profile suggested early evidence of molecular response to treatment with a crizotinib and trametinib combination therapy in a patient with acquired KRAS amplification, but treatment was poorly tolerated. Future efforts to develop combination therapies against these targets will require attention to agent selection, dosing, and scheduling to achieve both tolerability and efficacy.

This study enhances the understanding of the role of co-occurring genomic alterations in METex14-mutated NSCLC, with implications for the development of personalized therapeutic strategies to enhance the initial response magnitude and duration to MET TKI and delay or overcome acquired resistance. Given the prominence of genomic alterations favoring RAS–MAPK pathway hyperactivation, the addition of a MEK or potentially an ERK inhibitor to MET TKI therapy is a promising combination therapy strategy which warrants further prospective study.

V.M. Raymond is an employee/paid consultant for and holds ownership interest (including patents) in Guardant Health. R.B. Lanman is an employee/paid consultant for Guardant Health, and holds ownership interest (including patents) in Guardant Health, Biolase. Inc. and Forward Medical, Inc. N. Peled is an employee/paid consultant for and reports receiving speakers bureau honoraria from AstraZeneca, Boehringer Ingelheim, Bristol-Myers Squibb, Eli Lilly, Foundation Medicine, Guardant360, MSD, Novartis, NovellusDx, Roche, and Takeda. R.B. Corcoran is an employee/paid consultant for Amgen, Array Biopharma, Astex, Avidity, Bristol-Myers Squibb, Chugai, Fog Pharma, Fount Therapeutics, Genentech, LOXO, N-of-One, Novartis, nRichDx, Revolution Medicines, Roche, Roivant, Shionogi, Spectrum, Symphogen, Taiho and Warp Drive Bio, reports receiving commercial research grants from AstraZeneca and Asana, and holds ownership interest (including patents) in Avidity Biosciences, C4 Therapeutics, Found Therapeutics, nRichDx, and Revolution Medicines. B.C. Bastian is an employee/paid consultant for Lilly Inc. J.S. Fraser is an employee/paid consultant for Relay Therapeutics, and reports receiving speakers bureau honoraria from Novartis. E.A. Collisson is an employee/paid consultant for Guardant Health, reports receiving commercial research grants from Merck Kga, Ignyta, Loxo Bayer, Senti Biosciences, and Ferro, and holds ownership interest (including patents) in Illumina, Guardant Health, Pacific Biosciences, Blood Q, Tatara Thera. C.E. McCoach is an employee/paid consultant for Guardant Health, reports receiving other commercial research support from Novartis and Revolution Medicines, and reports receiving speakers bureau honoraria from Novartis. J. Pacheco is an employee/paid consultant for AstraZeneca, Novartis, Takeda, Gerson Lehrman Group, and Pfizer, reports receiving speakers bureau honoraria from Genentech and Takeda. L. Bazhenova is an employee/paid consultant for AstraZeneca, Takeda, Genentech, BI, Eli Lilly, Bayer, G1 Therapeutics, Abbvie and Pfizer, and reports receiving commercial research grants from Beyondspring Pharmaceutics and holds ownership interest (including patents) in Epic Sciences. T. G. Bivona is an employee/paid consultant for Takeda, Array, Revolution Medicines, Jazz, Novartis, Springworks, Strategia, Omniseq, and reports receiving commercial research grants from Novartis and Revolution Medicines. C.M. Blakely is an employee/paid consultant for Revolution Medicines, reports receiving commercial research grants from Novartis, AstraZeneca, Mirati, Roche, Spectrum and Medimmune. No potential conflicts of interest were disclosed by the other authors.

Conception and design: J.K. Rotow, R.B. Lanman, N. Peled, F. Fece de la Cruz, R.B. Corcoran, E.A. Collisson, T. Li, T.G. Bivona, C.M. Blakely

Development of methodology: J.K. Rotow, P. Gui, R.B. Lanman, N. Peled, F. Fece de la Cruz, R.B. Corcoran, I. Yeh

Acquisition of data (provided animals, acquired and managed patients, provided facilities, etc.): J.K. Rotow, P. Gui, V.M. Raymond, R.B. Lanman, F.J. Kaye, N. Peled, F. Fece de la Cruz, B. Nadres, R.B. Corcoran, I. Yeh, B.C. Bastian, K. Newsom, V.R. Olivas, E.A. Collisson, C.E. McCoach, D. Ross Camidge, J. Pacheco, L. Bazhenova, T. Li, T.G. Bivona, C.M. Blakely

Analysis and interpretation of data (e.g., statistical analysis, biostatistics, computational analysis): J.K. Rotow, P. Gui, W. Wu, V.M. Raymond, R.B. Lanman, N. Peled, F. Fece de la Cruz, R.B. Corcoran, I. Yeh, B.C. Bastian, P. Starostik, K. Newsom, A.M. Wolff, J.S. Fraser, L. Bazhenova, T. Li, T.G. Bivona

Writing, review, and/or revision of the manuscript: J.K. Rotow, P. Gui, V.M. Raymond, R.B. Lanman, F.J. Kaye, N. Peled, F. Fece de la Cruz, R.B. Corcoran, I. Yeh, K. Newsom, K. Newsom, J.S. Fraser, E.A. Collisson, C.E. McCoach, D. Ross Camidge, J. Pacheco, L. Bazhenova, T. Li, T.G. Bivona, C.M. Blakely

Administrative, technical, or material support (i.e., reporting or organizing data, constructing databases): J.K. Rotow, V.M. Raymond, K. Newsom, C.E. McCoach, T.G. Bivona

Study supervision: J.K. Rotow, T.G. Bivona, C.M. Blakely

The authors acknowledge funding support from NIH/NCI U54CA224068 (to R.B. Corcoran); R01CA227807, R01CA239604, R01CA230263 (to E.A. Collisson); NIH/NCI U01CA217882, NIH/NCI U54CA224081, NIH/NCI R01CA204302, NIH/NCI R01CA211052, NIH/NCI R01CA169338, and the Pew-Stewart Foundations (to T.G. Bivona); and the Damon Runyon Cancer Research Foundation, Doris Duke Charitable Foundation, V Foundation, and American Cancer Society to C.M. Blakely.

The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked advertisement in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.