Structural Alterations of MET Trigger Response to MET Kinase Inhibition in Lung Adenocarcinoma Patients

This article is featured in Highlights of This Issue, p. 1241

Over the past years, we have witnessed a dramatic shift in the clinical routine of lung adenocarcinoma patients, driven by the identification of oncogenically activated and therapeutically actionable targets. Next to oncogenic mutations, structural rearrangements of receptor kinases involving ALK, ROS1, or RET represent an ever increasing pool of druggable targets in lung cancer patients (1–6). Effective inhibition of these oncogenic drivers frequently results in dramatic clinical responses (1, 4–6). Most common oncogenic alterations of MET kinase involve MET exon 14 skipping mutations, but the therapeutic relevance of more complex structural rearrangements remains largely unknown (2, 7–9). Here, we report the discovery of structural alterations of MET that may predict response to targeted MET inhibition in lung cancer patients.

All 3 patients consented to testing, therapy, and registration of data for future publication. Genetic testing was conducted in concordance with local ethical guidelines and reviewed by the Institutional Ethics Committee. The patient material was sequenced using hybrid capture–based next-generation sequencing (NEO, New Oncology GmbH).

Genomic DNA was extracted from formalin-fixed paraffin-embedded (FFPE) material, sheared (Covaris), and subjected to hybrid capture–based next-generation sequencing to detect point mutations, small insertions and deletions, copy number alterations. and rearrangement/gene fusions in a single assay (NEO New Oncology GmbH). In brief, after shearing, adapters were ligated and individual genomic regions of interest were enriched using complementary bait sequences (hybrid capture procedure). The selected baits ensure optimal coverage of all relevant genomic regions. After enrichment, targeted fragments were amplified (clonal amplification) and sequenced in parallel at high sequencing depth. Computational analysis was performed using NEO New Oncology's proprietary computational biology analysis pipeline to detect relevant genomic alterations in a quantitative manner.

Whole blood (18 mL) was collected in Streck tubes (Cell-Free DNA BCT, Streck, Ref. #218997) and cell-free DNA (cfDNA) was extracted using Qiagen's QIAamp Circulating Nucleic Acid Kit (QIAGEN, cat. no. #19419). Fragmented DNA was subjected to hybrid capture–based next-generation sequencing to detect point mutations, small insertions and deletions, copy number alterations, and genomic translocations in a single assay (NEO New Oncology GmbH). In brief, after DNA extraction, adapters were ligated and individual genomic regions of interest were enriched using complementary bait sequences (hybrid capture procedure). The selected baits ensure optimal coverage of all relevant genomic regions. After enrichment, targeted fragments were amplified (clonal amplification) and sequenced in parallel at ultrahigh sequencing depth. Computational analysis was performed using NEO New Oncology's proprietary computational biology analysis pipeline to remove sequencing artifacts and detect relevant genomic alterations in a quantitative manner.

Dideoxy sequencing was performed using manufacturer's standard protocols at the Cologne Center for Genomics or at GATC GmbH. Alignments and verification of sequences were performed using Geneious R8.

For KIF5B- and STARD3NL-MET cases, IHC was performed by using the Ventana CONFIRM SP44 rabbit mAb on the Ventana Benchmark XT staining platform according to the the manufacturer's instructions (Roche Diagnostics/Ventana Medical Systems). In brief, antigene retrieval was done with buffer “CC1” (high pH), detection was done with the OptiView kit that uses DAB as chromogen, and counterstaining was done with hematoxylin. Phospho-MET was stained using D26 rabbit mAb (Cell Signaling Technology). For the MET kinase domain duplication case, IHC was performed using the following antibodies: p40 (Biocare, BC28, 1:100), TTF1 (Dako, 8G7G3-1,1:800), MET (SP44, Ventana, ready to use), and ALK (Monosan, 5A4, 1:10).

ALK break-apart FISH was performed using the Vysis ALK Break Apart FISH Probe Kit (Abbott, #06N38) according to the manufacturer's instructions.

Generation of cell lines, viability assays, and immunoblotting was performed as described previously (5). A detailed description of all additional methods can be found in the Supplementary Material section.

All statistical analyses were performed using Microsoft Excel 2011 or GraphPad Prism 6.0 h for Mac or R (https://www.r-project.org/).

To identify patients with structural MET alterations, we used hybrid capture–based paired-end sequencing (NEOplus, NEO New Oncology) of patients that were negative for any common driver mutation for lung cancer, and we were able to identify 2 patients with lung adenocarcinoma (LADC) (2/337, ∼0.5%) from two different clinical centers. We identified spanning reads involving the MET kinase (3′) and sequences matching to either KIF5B or STARD3NL (5′; Fig. 1A; Supplementary Fig. S1A; ref. 10). Both genes encode domains that may facilitate homodimerization (coiled coil domain KIF5B; MENTAL domain STARD3NL) of the resulting fusion protein (Fig. 1A; refs. 11, 12). We next confirmed the expression of both fusion transcripts by dideoxy sequencing of the RT-PCR products isolated from the FFPE material (Supplementary Fig. S1B). In addition, IHC staining revealed high MET protein expression and was positive for phospho-MET in both samples (Fig. 1B). Although another variant of KIF5B-MET fusion has been observed previously, STARD3NL represents a novel fusion partner for MET (2).

Because our sequencing panel was negative for any other oncogenic driver in both of the patients, the MET kinase fusion seemed to be a potential druggable target. Crizotinib, initially developed as a MET inhibitor, effectively inhibits ALK and MET among several other kinases (4, 13). Therefore, both patients underwent off-label treatment with crizotinib (8, 14).

The first patient, identified with the KIF5B-MET gene fusion, was a 33-year-old female with stage IV LADC who is a former smoker (10 pack-years; Supplementary Fig. S1C). Two weeks after off-label treatment with crizotinib, cough and dyspnea improved remarkably. A subsequent PET/CT showed a response with a decrease in tumor size and FDG uptake (74% SUV_max reduction based on PERCIST; Fig. 1C and D; Supplementary Fig. S1C; ref. 15). The following PET/CT scans confirmed the response (up to SUV_max 77% reduction based on PERCIST) and the treatment is ongoing with minimal side effects and a clinical benefit of currently more than 8 months after initiation (Fig. 1C and D; Supplementary Fig. S1C).

The STARD3NL-MET gene fusion was identified in a 62-year-old female with a stage IV LADC who is a never smoker. After initial diagnosis, off-label treatment with crizotinib was started, leading to a partial response (69% reduction based on RECIST 1.1) of the primary tumor and the metastases (Fig. 1C and D). Thus, our genomic analyses together with clinical observations suggest that MET rearrangements represent a druggable target in lung cancer patients.

To characterize the oncogenicity and druggability of KIF5B-MET and STARD3NL-MET, we stably transduced KIF5B-MET and STARD3NL-MET into NIH-3T3 cells (Fig. 2A). The overexpression resulted in the activation of MET signaling that was robustly inhibited with crizotinib along with partial inhibition of downstream MAPK and PI3K signaling (Fig. 2A). In these cells, stable inhibition of oncogenic MET signaling was obtained during treatment with crizotinib and tepotinib, two structurally independent MET-targeting drugs (Supplementary Fig. S1D). As expected, we observed robust colony formation in soft agar assays only in cells expressing the MET fusions (Fig. 2B). Of note, colony formation was robustly inhibited under crizotinib treatment in a dose-dependent manner, suggesting each MET rearrangement being the oncogenic driver for anchorage-independent growth (Fig. 2B). Furthermore, stable expression of KIF5B-MET or STARD3NL-MET induced IL3-independent growth in Ba/F3 cells, confirming the oncogenic transformation potential of the MET rearrangements. Both gene fusions showed nanomolar sensitivity toward inhibition of MET either with crizotinib or more effectively with tepotinib in viability screening assays (Fig. 2C–E; Supplementary Fig. S1E). Thus, the orthogonal results derived from our functional in vitro experiments clearly demonstrate the therapeutic relevance of MET fusions for lung cancer patients.

With a similar sequencing approach, we sought to identify a potential resistance mechanism in an ALK-rearranged LADC (stage IV, 60, male) that acquired resistance to ceritinib. The patient initially showed a partial response to the selective ALK inhibitor that lasted for 3 months (Fig. 3A). Here, hybrid capture-based sequencing of circulating tumor DNA (ctDNA, NEOliquid) revealed a duplication of the MET kinase domain (MET-KDD) that was confirmed in a FFPE rebiopsy sample but was not present in the pretherapy sample (Fig. 3B; Supplementary Fig. S2A). We validated the expression of the MET-KDD transcript in the post-ceritinib FFPE sample. Sequencing of the cDNA showed a retained short intronic region between the kinase domains that may represent a linker function (Supplementary Fig. S2B).

We speculated that MET-KDD may override the activity of a specific ALK inhibitor, such as ceritinib, and that dual ALK/MET inhibition may overcome this resistance. To test this hypothesis, we transiently overexpressed the wild-type and the rearranged receptor kinase in NIH-3T3 cells (Supplementary Fig. S2C). We observed a protein size shift of the MET-KDD transfected cells compared with cells expressing wild-type MET (Supplementary Fig. S2C). Both cell lines displayed high phospho-MET levels and activation of downstream MAPK signaling, indicating that MET-KDD retains its kinase activity (Supplementary Fig. S2C). To more precisely model the clinical situation in vitro, we stably transduced NIH-3T3 cells with EML4-ALK and transiently transfected the cells with MET-KKD. As expected, only crizotinib but not ceritinib was able to inhibit both ALK- and MET-dependent signaling in cells expressing rearranged ALK and MET-KDD (Fig. 3C).

A subsequent treatment of the patient expressing MET-KDD with crizotinib resulted in a partial response (51% reduction based on RECIST 1.1) that lasted for 3 months (Fig. 3D). Follow-up imaging showed oligo-progression of a thoracic wall lesion. Hybrid capture–based sequencing was performed on ctDNA and on a rebiopsy sample of the oligo-progressive lesion, but did not reveal any additional mutations that may be associated with resistance. Both samples were positive for the ALK fusion and MET-KDD. We also confirmed the presence of the initial ALK rearrangement throughout the treatment lines (Supplementary Fig. S2D). Thus, it is conceivable that the MET-KDD rearrangement identified in the patient is associated with acquired resistance against ceritinib (Fig. 3A and D).

Of note, we were not able to identify a single MET-KDD in a reanalysis of the TCGA dataset (n > 11,000 patients), suggesting that MET-KDD might be specifically associated with acquired ALK inhibitor resistance (Supplementary Fig. S2E). Overall, MET-KDD represents a rare structural alteration that may emerge as a resistance mechanism against ceritinib in ALK-rearranged tumors.

In summary, our results provide evidence that structural rearrangements of MET, such as MET fusions and MET kinase duplications, represent immediate druggable targets in lung cancer patients. Although the MET kinase fusions identified in two LADCs fulfill the criteria of primary oncogenic drivers, the MET-KDD rearrangement may be specifically associated with the ceritinib resistance phenotype in an ALK-rearranged background.

Oncogenic MET alterations are found in several cancers, including lung, glioblastoma, and papillary renal cell carcinomas (8, 14, 16, 17). In lung cancer patients, exon 14 skipping mutations have been identified as recurrent alterations that lead to hyperactivated MET signaling and that trigger response to MET inhibition (3, 8). More recently, MET gene fusions have been shown to represent a potentially druggable target in pediatric glioblastoma (7). Our results provide evidence that these types of MET alterations may be rare but can lead to a therapeutically actionable dependency in lung cancer patients.

MET amplifications, on the other hand, are recurrently found in EGFR-mutant patients that acquire resistance against first- and third-generation EGFR inhibitors (18, 19). Two reports suggest that MET signaling could also promote resistance in ALK-rearranged tumors to ALK kinase inhibition (20, 21). Our data suggest that under selective pressure with selective ALK inhibitors, subclones harboring MET kinase duplications may arise and trigger resistance against these drugs. Interestingly, a similar link between the selection of kinase domain duplications and resistance to targeted therapy has been previously described for BRAF in vemurafenib-resistant melanoma (22).

Overall, our functional data provide important insight into the cellular signaling that is engaged through MET kinase fusions and MET kinase domain duplications. The availability of clinically active MET-directed drugs urges for the testing of these rare structural rearrangements within the diagnostic routine and future clinical trials.

R. Büttner is a consultant/advisory board member for Novartis. J. Wolf reports receiving speakers bureau honoraria from and is a consultant/advisory board member for AstraZeneca, Merck, Novartis, and Pfizer. J.M. Heuckmann is an employee of NEO New Oncology and is listed as a co-inventor on a patent, which is owned by Siemens Healthineers, on an algorithm that detects large deletions/insertions and duplications in NGS data. M.L. Sos reports receiving commercial research grants from Novartis. No potential conflicts of interest were disclosed by the other authors.

Conception and design: D. Plenker, R. Riedel, J. Wolf, L. Heukamp, M.L. Sos, J.M. Heuckmann

Development of methodology: A.H. Scheel, J. Brägelmann, T. Persigehl, E. Thunnissen, J. Wolf, L. Heukamp, M.L. Sos

Acquisition of data (provided animals, acquired and managed patients, provided facilities, etc.): D. Plenker, M. Bertrand, A.J. de Langen, R. Riedel, C. Lorenz, A.H. Scheel, J. Müller, J. Daßler-Plenker, C. Kobe, T. Persigehl, A. Kluge, T. Wurdinger, P. Schellen, R. Menon, E. Thunnissen, R. Büttner, F. Griesinger, J. Wolf, L. Heukamp

Analysis and interpretation of data (e.g., statistical analysis, biostatistics, computational analysis): D. Plenker, M. Bertrand, A.J. de Langen, R. Riedel, C. Lorenz, J. Müller, J. Brägelmann, C. Kobe, T. Persigehl, T. Zacherle, R. Menon, E. Thunnissen, F. Griesinger, J. Wolf, L. Heukamp, M.L. Sos, J.M. Heuckmann

Writing, review, and/or revision of the manuscript: D. Plenker, M. Bertrand, A.J. de Langen, R. Riedel, A.H. Scheel, J. Brägelmann, C. Kobe, T. Persigehl, G. Hartmann, E. Thunnissen, R. Büttner, F. Griesinger, J. Wolf, L. Heukamp, M.L. Sos, J.M. Heuckmann

Administrative, technical, or material support (i.e., reporting or organizing data, constructing databases): D. Plenker, R. Riedel, J. Daßler-Plenker, C. Kobe, T. Persigehl, E. Thunnissen, R. Büttner, J. Wolf

Study supervision: D. Plenker, G. Hartmann, J. Wolf, M.L. Sos, J.M. Heuckmann

This work was supported by the German federal state North Rhine Westphalia (NRW) as part of the EFRE initiative (EFRE-0800397 to R. Büttner and M.L. Sos) and by the German Ministry of Science and Education (BMBF) as part of the e:Med program (grant no. 01ZX1303 to J. Wolf and R. Büttner and grant no. 01ZX1406 to M.L. Sos). Additional funding was provided by the Deutsche Krebshilfe as part of the Oncology Centers of Excellence funding program (to R. Büttner, J. Wolf, and G. Hartmann) and by the DFG-funded Cluster of Excellence ImmunoSensation (to G. Hartmann).

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.