Acquired MET Y1248H and D1246N Mutations Mediate Resistance to MET Inhibitors in Non–Small Cell Lung Cancer

The paradigm for the pharmacologic management of non–small cell lung cancer (NSCLC) has revolutionized by the development of EGFR inhibitors. Concomitant with the exciting advancements in this field is the emergence of drug resistance (1–3), with the substitution of methionine for threonine at position 790 (T790M) being the most commonly acquired resistance mutation, accounting for approximately 50% of resistant cases. It can be successfully treated with third-generation tyrosine kinase inhibitors (TKI; ref. 2). MET amplification, responsible for 20% of EGFR-TKI–induced resistance, is a second well-established resistance mechanism known to bypass EGFR inhibition (4). Numerous studies have conferred susceptibility of MET mutations and focal amplification to targeted MET inhibitors (5, 6). MET is a receptor tyrosine kinase, activated by the binding of HGF, resulting in the phosphorylation of HER3 and subsequent activation of PI3K pathway, promoting cell proliferation, survival, mobility, migration, and epithelial to mesenchymal transition (7, 8). Gain-of-function alterations in MET, achieved through receptor overexpression, amplification, mutations, or alternative spicing, have been observed in many cancers, including NSCLC (9, 10).

MET has long been a candidate target for therapeutic development. A number of MET inhibitors, currently in clinical trials, have shown promising outcomes in NSCLC (11). Crizotinib, approved by the FDA for patients with NSCLC harboring anaplastic lymphoma kinase (ALK) rearrangements, was originally developed as a MET inhibitor (8, 12). Numerous studies have reported that patients harboring mutations in the MET exon 14 splice sites derived clinical benefits from crizotinib (5, 6, 13). The National Comprehensive Cancer Network guidelines recommend crizotinib to NSCLC patients harboring high-level MET amplification or MET exon 14 skipping mutation (5, 6, 14, 15). INC280, developed for tumors bearing MET overexpression or amplification, competes for the ATP-binding site in the tyrosine kinase domain, thus preventing activation of c-MET and the subsequent recruitment of downstream effectors (16, 17). In the ongoing Phase I INC280 clinical trial at our hospital, it induces durable response in EGFR-mutant lung cancer patients with acquired resistance to EGFR-TKIs, with preliminary 31% (28/90 patients) unconfirmed responsive rate (RR; ref. 18). Both crizotinib and INC280 are type I MET inhibitors, which preferentially bind the active conformation of MET. In contrast, type II MET inhibitors, such as cabozantinib (XL184), bind the inactive conformation (11). Although drug resistance associated with MET-TKIs has also been reported, there are limited studies investigating the underlying mechanisms. A recent study reported an acquired mutation MET D1228N (D1246N) in the kinase domain at time of progression of a patient exposed to crizotinib (19). Other mutations, including Y1230 (Y1248) occurring at the MET activation loop, have also been shown to have inhibitory effects against MET inhibitors in vitro (20, 21).

In this study, we conduct a cohort of 12 advanced NSCLC patients who developed resistance to gefitinib and were subsequently switched to a combinatorial therapy consisting of gefitinib and a MET-TKI. We provide insight into the diversity of mechanisms underlying MET inhibitors induced resistance by identifying two mutually exclusive mechanisms: EGFR amplification and MET mutations Y1248H and D1246N.

Twelve NSCLC patients who harbored MET overexpression and developed resistance to MET inhibitor underwent tumor biopsy between April 2012 and January 2016. MET expression was detected by immunohistochemistry (IHC; antibody SP44). A minimum of 50% tumor cells with moderate or strong staining was defined as MET positive. Copy numbers were detected by FISH (By Cappuzzo scoring system & MET/CEP7 ratio). A minimum of 5 copies were required to be identified as positive or a MET/CEP7 ratio ≥1.8 was defined as c-MET amplification (low: 1.8–2.2, intermediate: 2.2–5, high: ≥5). Met probes were purchased from KREATECHTM. Patients were either treated in a clinical trial investigating the combinatorial effect of gefitinib and INC280 (n = 11) or with gefitinib and crizotinib (n = 1). Standard histopathology was performed to confirm the diagnosis of malignancy and histologic subtype. Tumor biopsies were obtained under an Institutional Review Board–approved protocol. All patients provided written informed consent.

Ten milliliter of whole blood was collected in K3EDTA-containing tubes (Cell-Free DNA BCT) and centrifuged at 2,000 g for 10 minutes at +4°C within 72 hours after collection. Supernatant was transferred to a fresh 15 mL centrifuge tube without disturbing the buffy coat layer and subjected to an additional centrifugation for 10 minutes at 16,000 g at +4°C. The supernatant was again transferred to a new tube. The plasma was stored at −80°C until further analysis. Circulating cell-free DNA was extracted from plasma samples using the QIAamp Circulating Nucleic Acid Kit (Qiagen) according to the manufacturer's instructions. Quantification of cell-free DNA (cfDNA) was performed using the Qubit 2.0 Fluorometer with the dsDNA HS assay kits (Life Technologies).

DNA was extracted using the QIAamp DNA FFPE tissue Kit (Qiagen) according to the manufacturer's instructions. DNA concentration was measured using Qubit dsDNA assay.

Genetic profiles of all tissue samples were assessed by performing capture-based targeted deep sequencing using the OncoScreen panel (Burning Rock Biotech Ltd.), covering 2.02 MB of human genomic regions, including all exons and critical introns of 295 genes. DNA quality and size were assessed by high sensitivity DNA assay using a bioanalyzer. cfDNA samples were profiled using LungPlasma panel covering selected exons and introns of 168 genes, spanning 160 KB of human genomic regions. All indexed samples were sequenced on a NextSeq 500 (Illumina, Inc.) with pair-end reads.

The sequencing data in the FASTQ format were mapped to the human genome (hg19) using BWA aligner 0.7.10. Local alignment optimization, variant calling, and annotation were performed using GATK 3.2, MuTect, and VarScan, respectively. DNA translocation analysis was performed using both Tophat2 and Factera 1.4.3. Gene-level copy-number variation was assessed using a t statistic after normalizing reads depth at each region by total reads number and region size, and correcting GC-bias using a LOESS algorithm.

Genomic DNA from each sample was used for sequence analysis of MET exon 19. These exons were amplified by PCR, and the resulting PCR products were purified and labeled for sequencing using the BigDye 3.1 Kit (Applied Biosystems) according to the manufacturer's protocols. Samples were ran on an ABI 3100 Genetic Analyzer (Applied Biosystems).

Droplet Digital PCR (ddPCR) was performed on a Bio-Rad QX100 ddPCR instrument. MET D1246N (G>A) primers and probes were purchased from Life Technologies. Sequences are available upon request. The following PCR conditions were used: 1 cycle of 95°C × 10 minutes, 40 cycles of 95°C × 15 seconds, and 55°C × 1 minute, followed by 4°C hold. MET mutation–specific signals are generated in the VIC channel, whereas the MET wild-type (WT) signals are in the FAM channel. The quantification of the target molecule was presented as number of total copies (mutant plus WT) per sample in each reaction. Fractional abundance (F.A.) is calculated as follow: F.A. % = (N_mut/(N_mut + N_wt) × 100), where N_mut is the number of mutant events and N_wt is the number of WT events per reaction. ddPCR analysis of normal control DNA (genomic DNA) and no DNA template controls were always included.

NIH3T3 cell line was obtained from the American Type Culture Collection (ATCC). Cells were grown in DMEM (GIBCO-Invitrogen) supplemented with 10% NCBS and maintained at 37°C in a humidified atmosphere at 5% CO₂. pBabe puro c-MET WT was a gift from Joan Brugge (Addgene plasmid #17493). The MET sequence was cloned into the pMT143 lentiviral vector and was driven by a CMV7 promoter (Sunbio, plasmid#PSE2209). Two point mutation vectors (D1226N,Y1228H) were constructed based on PSE2209. All vectors were verified by DNA sequencing. Lentivirus particles were packaged and purified by Sunbio lentiviral system. NIH3T3 cells infected with lentivirus containing MET-WT and MET-Mutants were selected with 2 μg/mL puromycin for 5 days to obtain stable expression of indicated protein and kept in 1 μg/mL puromycin. All cell lines tested Mycoplasma negative (Mycoplasma Detection Kit; Cat. SE000-8010; Sesh-Biotech) within 6 months of performing the experiments. Cell line authentication was performed.

GAPDH was used as a loading control. The Akt, phospho-Akt(S473), Erk1/2, phospho-Erk1/2(T202/Y204), MET(D1C2,25H2), phospho-MET (Y1234/Y1235), S6 ribosomal protein (54D2), phosphor-S6 ribosomal protein (ser235/236), and phosphor-p70 S6 Kinase(Thr421/Ser424) antibodies were purchased from Cell Signaling Technology. All antibodies were used at a 1:1,000 dilution, except for GAPHD antibody, which was used at 1:10,000 dilution.

NIH3T3 cells (1,000) were plated in a 384-well plate and treated with indicated concentrations of INC280, crizotinib, or 0.1% DMSO for 72 hours. Cell survival was measured by MTS assay (G3580; Promega) following the manufacturer's instructions. Relative cell survival rate was normalized to the DMSO-treated group. Each data point represents the average of biological duplicates.

To establish tumor xenografts, MET-mutants cells (1 × 10⁶) were subcutaneously inoculated in the right hind of flank of female nu/nu athymic nude mice (6–8 weeks old) individually. After tumors reached a mean volume of 150 mm³, mice were randomized (n = 5 per group) and orally administered INC280 (30 mg/kg twice daily), crizotinib (100 mg/kg/d), XL184 (20, 60 mg/kg/d), or vehicle for 14 days. Tumor size was measured 2 to 3 times weekly by caliper. Tumor volume was calculated using the following formula: Tumor volume = (length × width²) × 0.5. Mice were sacrificed when tumor volume reached 2,000 mm³ according to the principle of animal welfare. All experiments were performed following protocols approved by the Institutional Animal Care and Use Committee.

Due to the rapid progression of the tumors in vehicle, INC280, or crizotinib-treated groups, mice reached the maximum tumor volume prior to the scheduled treatment endpoints. Therefore, we had to terminate drug administration on the indicated days, according to the principle of animal welfare. So the values of percentage of tumor growth were calculated on the indicated days by comparing tumor volumes in drug-treated and vehicle-treated groups. * indicates P < 0.05 (Mann–Whitney U).

To elucidate potential mechanisms of resistance associated with MET-TKIs, we conducted a cohort of 12 patients with advanced NSCLC who showed MET overexpression after the development of resistance to gefinitib and were subsequently switched to a combinatorial regimen consisting of gefinitib and a type I MET-TKI. We performed capture-based targeted sequencing on serial tumor biopsies and blood samples using OncoScreen and LungPlasma panels, respectively, to detect and quantify genetic alterations. The OncoScreen panel, spanning 2.024 MB of human genome, consists of all exons and critical introns of 295 genes; LungPlasma panel, spanning 160 KB of human genome, consists of critical exons and introns of 168 genes.

At baseline, all patients harbored somatic EGFR-activating mutations confirmed by both ARMS-PCR and next-generation sequencing (NGS): 7 patients carried EGFR exon 19 deletions and the remaining 5 patients harbored EGFR exon 21 L858R substitutions. They were initially treated with gefitinib, an EGFR-TKI. After the emergence of drug resistance, they were subsequently switched to a combinatorial regimen consisting of gefitinib and a type I MET-TKI based on the detection of MET overexpression in tissue biopsy samples by IHC. One patient was treated with gefitinib and crizotinib; the remaining 11 patients were treated with gefitinib and INC280. Our NGS results revealed acquired MET amplification after developing resistance to gefitinib in half of the patients (6/12). Based on our NGS results, among 7 patients achieved partial response (PR), 5 of them harbored MET amplification prior to MET-TKI treatment; in contrast, among 5 patients achieved stable disease (SD) or progressive disease (PD), only 1 patient harbored MET amplification, suggesting the existence of other resistance mechanisms in patients achieved SD and PD despite the fact that they showed MET overexpression by IHC. Furthermore, this result suggested that utilizing NGS to interrogate MET amplification is a better predictive method. Interestingly, most patients achieved PR (5/7) harbored EGFR L858R at baseline; on the other hand, all SD and PD patients harbored EGFR exon 19 deletions at baseline, suggesting patients with baseline EGFR L858R mutations are more likely to respond to the combinatorial treatment (P = 0.028, Fisher exact test; Fig. 1).

Although all enrolled patients tested negative for EGFR T790M mutation by ARMS-PCR prior to the combinatorial treatment, NGS still revealed 4 patients, who harbored T790M mutation. Three of them (P06, P09, and P11) belonged to the SD or PD group, and 1 patient (P05) belonged to the PR group. Prior to the combinatorial treatment, 6 of 7 patients were T790M negative in the PR group; in contrast, in SD and PD group, only 1 of 4 patients was T790M negative (Fisher exact test: P = 0.088, OR = 12.2), suggesting T790M is a possible concurrent resistance mechanism. The Student t test shows a marginal difference between patients harboring T790M mutation and those without it prior to the combinatorial treatment, as demonstrated in the Kaplan–Meier curves (Supplementary Fig. S1), although it did not reach statistical significance. The allele frequency (AF) of T790M has increased during the course of combinatorial treatment for a majority of patients who carried T790M prior to the combinatorial treatment except for 1 patient (P11). In addition, 2 patients (P01 and P10) developed de novo T790M mutation after the combinatorial treatment (Fig. 1). Collectively, our result indicates that among patients who presented MET amplification as a resistance mechanism to gefitinib, it is not unlikely to have EGFR T790M as a concurrent resistance mechanism. Patients harboring both mutations are less likely to respond to the combinatorial treatment.

At time of progressive disease, our NGS results revealed two newly acquired mutations at the kinase domain of MET, Y1248H and D1246N, from 2 patients whose best response was PR. Neither mutation was reported in treatment-naïve NSCLC patients from interrogating the Cancer Genome Atlas (TCGA) database. Both patients showed MET amplification after gefitinib treatment (Fig. 2F, G, M, and N). Patient 02, a 32-year-old female, whose cancer had progressed on multiple prior lines of treatments, demonstrated PR after 2 months of combinatorial treatment consisting of gefitinib and crizotinib (Fig. 2D); however, her CT scans revealed systemic progression after 8.8 months of treatment (Fig. 2E). NGS analysis of serial tissue biopsy and blood samples revealed two acquired in trans MET mutations, Y1248H and D1246N, at time of disease progression (Fig. 2A). MET Y1248H was detected in both tumor biopsy (AF = 20.8%) and blood sample (AF = 1.36%) taken after exposure to crizotinib; D1246N (AF = 1.51%) was only detected in the blood sample, indicating additional heterogeneity captured by liquid biopsy. Clonal progression analysis revealed the existence of EGFR L858R prior to the treatment of gefinitib followed by the subsequent development of MET amplification. After switching to the combinatorial treatment consisting of gefitinib and crizotinib, the patient acquired de novo MET Y1248H, D1246N, and EGFR T790M in addition to existing MET amplification and EGFR L858R (Supplementary Fig. S2A).

Patient 04, a 63-year-old female, demonstrated PR after 1 month of combinatorial treatment consisting of gefitinib and INC280 (Fig. 2K). She also developed systemic progression after 7.2 months of treatment (Fig. 2L). NGS analysis revealed a MET mutation D1246N occurring in exon 19, present in both tissue biopsy (AF = 5.21%) and blood samples (AF = 4.81% and 2.66%) taken at time of disease progression, but not before MET-TKI treatment (Fig. 2E). MET Y1248H from tumor biopsy of P02 and D1246N from both tissue and plasma samples of P04 after progression were successfully validated by Sanger sequencing and ddPCR, respectively (Fig. 2B and I). Clonal evolution for this patient is relatively simple, revealing the existence of EGFR 19del prior to gefitinib followed by acquired MET amplification. The combinatorial treatment induced de novo MET D1246N (Supplementary Fig. S2B).

In addition to MET mutations, acquired EGFR copy-number gain was also observed in 4 of 7 patients whose best response was PR. Patients with SD or PD as their best response did not harbor EGFR amplification (Fig 1). The other 2 patients whose best response was PR acquired secondary MET mutations, suggesting a mutually exclusive relationship between MET mutations and EGFR amplification. Unfortunately, we were unable to collect tissue biopsy from P12; therefore, EGFR copy-number difference cannot be calculated for this patient (Fig. 1).

To confirm MET mutations mediated resistance against type I MET-TKIs, we stably expressed mouse MET Y1228H (equivalent to human Y1248H) and mouse MET D1226N (equivalent to human D1246N) individually in NIH3T3 cells (mouse embryonic fibroblasts) and assessed cell proliferation 72 hours after the treatment. Cells were treated with increasing concentration of either INC 280 or crizotinib. Both of them are type I MET-TKI. The growth rate of cells expressing WT MET was significantly inhibited upon either INC280 or crizotinib treatment. In contrast, mutant forms exhibited relatively sustained growth rate upon type I MET-TKI treatment (Fig. 3A and B). Both WT and mutant expressing cells were sensitive to XL184, a type II MET-TKI, exhibiting no difference in growth inhibition (Fig. 3C). To investigate the downstream effects of resistance to MET-TKIs, we examined the protein levels of key downstream effectors of MET and major cross-talk pathways' participants. We collected cell extracts at different inhibitor concentrations and assessed the total as well as the phosphorylated form of MET, AKT, ERK, S6, and S6K. Upon either INC280 or crizotinib treatment, cells expressing WT MET showed a downregulation of the phosphorylated form of the proteins assayed, reflecting its responsiveness to such inhibitor; in contrast, the expression level of total proteins remained constant. On the other hand, cells expressing either form of the mutants exhibited constant protein levels of both the phosphorylated forms and the nascent forms of downstream effector proteins, reflecting their resistance to such inhibitors (Fig. 3D and F). Upon XL184 treatment, both WT and mutant cells showed a downregulation of the phosphorylated forms of major downstream effectors proteins (Fig. 3F). Collectively, our Western blot results are in an agreement with our cell proliferation assays, demonstrating resistance against INC280 and crizotinib but not XL184.

To confirm that both mutations are resistant to type I MET inhibitors but sensitive to MET type II inhibitors, we treated cells expressing either mutant with two additional inhibitors, savolitinib, a type I MET-TKI or BMS777607, a type II MET-TKI. Both our cell proliferation assay and immunoblotting results are consistent with our previous results (Supplementary Fig. S3A–S3D). Cell growth rate was relatively sustained in mutant cell lines after savolitinib treatment; in contrast, upon BMX777607 treatment, the growth rates of both mutant cell lines were inhibited. Taken together, our results demonstrate that both MET mutations are resistant to type I MET-TKIs but remain sensitive to type II MET-TKIs.

Next, we investigated MET mY1228H and mD1226N (equivalent to human Y1248H and D1246N, respectively) mediated resistance in vivo. Cells expressing either mutant were inoculated into female nu/nu mice. When tumor reached 150 m³, mice were randomized (5 per group) and orally administered INC 280, crizotinib, XL 184, or vehicle. Only XL 184 treatment resulted in significant tumor growth inhibition. Upon XL 184 treatment, mice expressing MET mY1228H or mD1226N had an average tumor volume that was 8% or 16% of the vehicle-treated mice, respectively. In contrast, in mice bearing MET mY1228H or mD1226N, treatment with INC280 resulted in an average tumor volume that was 77% or 88% of the original tumor and treatment with crizotinib resulted in an average tumor volume that was 67% or 68% of the original tumor volume (Fig. 3G and H). Mice bearing MET mD1226N showed gradual body weight gain under all treatments. Mice bearing MET mY1228H showed gradual body weight gain after exposure to INC280 and XL184 at 5 mg/kg/d. The body weight of mice remained constant after exposure to crizotinib and XL184 at higher dose: 20 mg/kg/d and 60 mg/kg/d (Supplementary Fig. S4A and S4B) Furthermore, dose-dependent inhibition was observed when treating mice bearing MET mutations with increasing concentrations of XL184, ranging from 0 mg/kg/d to 60 mg/kg/d (nonlinear mixed-effect model, P < 0.0001; Supplementary Fig. S4C and S4D). The inhibition effect of XL184 progressively increases as the dosage increases.

Next, we performed molecular studies to investigate the conformational changes induced by two mutations: Y1248H and D1246N. WT MET allows INC280 and crizotinib to be bound properly, occupying the ATP-binding site and making contacts with neighboring proteins through hydrogen bonds and extensive hydrphobic interations. Mutant structures underwent distortion, thus abrogating the binding of INC280 and crizotinib and subsequently leading to drug resisitance (Fig. 4A–G). In contrast, the binding of XL184 does not involve the ATP-bidning site; therefore, both mutations failed to render inhibitory effect (Fig. 4C and F). The results from our ligand-binding studies agreed with results from both cell proliferation assays and electrophoresis.

The realization that drug resistance inevitably arises has spurred great interest in elucidating resistance mechanisms. To the best of our knowledge, this is the first study interrogating resistant mutations induced by INC280 in patients. In this study, we investigated resistance to MET-TKIs and identified two acquired MET mutations Y1248H and D1246N. Neither was reported in treatment-naïve NSCLC patients from interrogating TCGA database. Subsequently, we confirmed their inhibitory activities both in vitro and in silico. Treatment of NIH3T3 cells stably expressing two mutants individually with type I inhibitors: INC280 or crizotinib but not type II inhibitor XL184 resulted in the maintenance of downstream PI3K-AKT and MEK-ERK signaling pathways, indicating type II inhibitor such as XL184 might lead to a more durable response. This finding is in an agreement with a previous study, which found MET D1228N rendered resistance to NVP-BVU972, a type I MET inhibitor but not to AMG 458, a type II MET inhibitor (21). Furthermore, the sequential use of Met inhibitors has been reported. A patient was switched to a treatment consisting of an EFGR inhibitor and a type II MET inhibitor, cabozantinib, after developing an acquired mutation, MET D1228V, and demonstrated a dramatic response (22). We observed that even in WT MET, the type I inhibitors appear to be less effective than type II inhibitor (Fig. 3A to C). This can be potentially attributed to the differential binding affinities. Predicted binding affinity data revealed that cabozantinib is 10-fold more potent than the other two inhibitors. Furthermore, cabozantinib (MW: 501.513, Vol: 1515.029) is bigger than INC280 (MW: 412.425, Vol: 1273.715) and crizotinib (MW: 450.342, Vol: 1294.587), resulting in extending to and occupying the hydrophobic pocket next to the c alpha-Helix. Recently, two studies have reported two mutations at this residue: Heist and colleagues reported MET D1228N as an acquired resistance to crizotinib, and Bahcall and colleagues reported MET D1228V as an acquired resistance to type I MET inhibitors, suggesting this resistance mechanism will likely to impose a clinical problem (19, 22). A more comprehensive analysis of serial plasma and tumor biopsy samples is needed to elucidate the incidence rate of MET mutations Y1248H and D1246N. Efforts are needed to develop next-generation MET inhibitors.

Our study highlights a relatively underappreciated concept: heterogeneity associated with resistance mechanisms, and reveals the importance of recognizing such heterogeneity. Previous studies show that MET amplification often occurs independent of T790M mutation (2). Our study revealed 4 patients, among whom 3 of them belonged to the SD/PD group, harboring T790M mutation in conjunction with MET amplification after exposure to gefitinib, highlighting heterogeneity in resistance mechanisms. We observed that patients with multiple resistance mechanisms (T790M and MET amplification) showed inferior responses and shorter progression-free survival when treated with the combinatorial therapy. In these cases, INC280 may effectively suppress the growth of subclones harboring MET amplification, but a population of T790M-positive subclones may have a growth advantage, resulting in the progression of disease. Furthermore, T790M status evolves during the course of combinatorial treatment, necessitating continuous monitoring. Among the 6 patients who did not harbor T790M mutation prior to MET-TKI, 4 of them acquired such mutation during the course of combinatorial treatment. The AF of T790M has increased during the course of combinatorial treatment for a majority of patients who carried T790M prior to MET-TKI. Understanding the mechanisms of resistance facilitates categorizing patients for targeted therapy to maximize clinical benefits. A combinatorial treatment consists of MET-TKI, and AZD9291 may be more beneficial to patients harboring both T790M and MET amplification.

In addition to acquired MET mutations as resistance mechanisms to MET inhibitors, we observed EGFR amplification in 4 patients who achieved PR and had no MET mutation at PD after exposure to INC280 (Fig. 1), suggesting EGFR amplification might also confer INC280 resistance independent of MET mutations. Although EGFR amplification has never been associated with drug resistance, it has been reported to support metastatic progression of prostate cancer (23). More efforts are needed to validate its inhibition against MET-TKIs.

In this study, we observed some discordance between tissue biopsy and plasma samples. MET D1246N was not observed in the tissue biopsy but present in the plasma sample from patient 02. It has been well established that a single biopsy is not adequate in capturing the full scope of resistance mechanisms (24). The majority of cfDNA is released from apoptotic or necrotic tumor cells, thus reflecting the genetic profile of the tumor. The NGS-based cfDNA profiling assay enables us to visualize the entire mutation spectrum, thus paving the way for routine application of liquid biopsies in the clinic.

In summary, we demonstrate that MET mutations Y1248H and D1246N are resistance mechanisms for type I MET-TKIs. NIH3T3 cells expressing either mutation showed resistance to both INC280 and crizotinib but not cabozantinib, indicating the potential of sequential use of MET inhibitors may lead to a more durable response.

C.-r. Xu reports receiving speakers bureau honoraria from AstraZeneca, Eli Lilly, Pfizer, and Roche, and holds ownership interest (including patents) in Illumina. Y.-l. Wu holds ownership interest (including patents) in AstraZeneca, Eli Lilly, and Roche, and is a consultant/advisory board member for AstraZeneca and Roche. No potential conflicts of interest were disclosed by the other authors.

Conception and design: A. Li, J.-j. Yang, X.-c. Zhang, L.-y. Gou, Q. Zhou, Z. Yang, Z.-h. Chen, Y.-l. Wu

Development of methodology: A. Li, X.-c. Zhang, J. Su, L.-y. Gou, S. Chuai, Z. Xie, H. Gao, B. Gan, Z.-h. Chen

Acquisition of data (provided animals, acquired and managed patients, provided facilities, etc.): A. Li, X.-c. Zhang, L.-y. Gou, Q. Zhou, W.-Z. Zhong, Q. Zhang, Zhen Wang, X.-n. Yang, B.-c. Wang, B.-y. Jiang, C.-r. Xu, Y.-l. Wu

Analysis and interpretation of data (e.g., statistical analysis, biostatistics, computational analysis): A. Li, J.-j. Yang, X.-c. Zhang, Z. Zhang, J. Su, L.-y. Gou, Z. Yang, H. Han-Zhang, W.-Z. Zhong, S. Chuai, Q. Zhang, H. Chen, Zheng Wang, C.-r. Xu, Y.-l. Wu

Writing, review, and/or revision of the manuscript: A. Li, J.-j. Yang, X.-c. Zhang, Z. Yang, H. Han-Zhang, S. Chuai, Q. Zhang, Z.-h. Chen, C.-r. Xu, Y.-l. Wu

Administrative, technical, or material support (i.e., reporting or organizing data, constructing databases): A. Li, J.-j. Yang, X.-c. Zhang, Z. Zhang, W.-Z. Zhong, S. Chuai, C.-r. Xu

Study supervision: J.-j. Yang, Z. Yang, H. Gao, Z.-h. Chen, S.-p. Wu, S.-y. Liu, Y.-l. Wu

Other (participated in in vitro assay design): Y. Bai

The authors thank the patients and investigators for their participation in this study and Dr. Xiaotao Wang, former AstraZeneca scientist, who provided the scientific input in virus packaging work.

This study was supported by Guangzhou Science and Technology Bureau (grant no. 2014Y2-00545, Y.-l. Wu; grant no. 2011Y2-00014); Special Fund of Public Interest by National Health and Family Control Committee (grant no. 201402031, Y.-l. Wu); Key Lab System Project of Guangdong Science and Technology Department-Guangdong Provincial Key of Translational Medicine in Lung Cancer (grant no. 2012A06140006, Y.-l. Wu); and National Science Funding of China (grant no. 81472207).

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.