Abstract
Amplified and/or mutated MET can act as both a primary oncogenic driver and as a promoter of tyrosine kinase inhibitor (TKI) resistance in non–small cell lung cancer (NSCLC). However, the landscape of MET-specific targeting agents remains underdeveloped, and understanding of mechanisms of resistance to MET TKIs is limited. Here, we present a case of a patient with lung adenocarcinoma harboring both a mutation in EGFR and an amplification of MET, who after progression on erlotinib responded dramatically to combined MET and EGFR inhibition with savolitinib and osimertinib. When resistance developed to this combination, a new MET kinase domain mutation, D1228V, was detected. Our in vitro findings demonstrate that METD1228V induces resistance to type I MET TKIs through impaired drug binding, while sensitivity to type II MET TKIs is maintained. Based on these findings, the patient was treated with erlotinib combined with cabozantinib, a type II MET inhibitor, and exhibited a response.
Significance: With several structurally distinct MET inhibitors undergoing development for treatment of NSCLC, it is critical to identify mechanism-based therapies for drug resistance. We demonstrate that an acquired METD1228V mutation mediates resistance to type I, but not type II, MET inhibitors, having therapeutic implications for the clinical use of sequential MET inhibitors. Cancer Discov; 6(12); 1334–41. ©2016 AACR.
See related commentary by Trusolino, p. 1306.
This article is highlighted in the In This Issue feature, p. 1293
Introduction
The majority of patients with advanced non–small cell lung cancer (NSCLC) harboring mutations in the EGFR gene can initially be successfully treated with EGFR tyrosine kinase inhibitors (TKI); however, the emergence of resistance to these agents is inevitable in most patients (1). Several mechanisms of resistance have now been clinically recognized, the most common being the EGFRT790M mutation, which can be successfully treated with third-generation EGFR TKIs like osimertinib (2). MET amplification, which can occur as an acquired or de novo resistance mechanism, is a second well-established resistance mechanism known to bypass EGFR inhibition and impart resistance to EGFR TKIs, and as such requires a therapeutic strategy directed at both EGFR and MET (3). Indeed, patients with such a signature of TKI resistance have benefited from various off-label combinations of EGFR TKIs and the MET inhibitor crizotinib (4).
The MET oncogene has long been a candidate for drug development in NSCLC, but with modest success. Notably, multiple phase III trials of agents aiming to target MET signaling have recently failed (5–7). However, the identification of oncogenic MET alterations in subsets of lung cancer has again led to increased interest in targeting this oncoprotein. Three scenarios now exist in which oncogenic MET alterations have been described as potentially targetable—de novo amplification of MET (8), exon 14 skipping mutations in MET (9), and acquired amplification of MET after resistance develops to EGFR TKI (3).
Multiple MET TKIs are in development for advanced NSCLC, but much of the clinical data on drug activity are preliminary. Like many other TKIs, MET inhibitors fall into two functionally distinct classes: type I inhibitors which preferentially bind the active conformation of MET, such as crizotinib, and type II inhibitors which bind the inactive conformation, such as cabozantinib (10). Type II inhibitors often inhibit a broader array of kinase targets, potentially resulting in more off-target side effects. Although the availability of multiple MET inhibitors lends itself to sequential targeted therapies, specific molecular factors affecting drug sensitivity and resistance have not been well characterized.
Here, we describe a patient with metastatic NSCLC with MET-mediated resistance to EGFR TKI who responded to treatment with a type I MET inhibitor, savolitinib (11), given in combination with a third-generation EGFR inhibitor, osimertinib (2). The patient then developed acquired resistance mediated by a novel MET kinase domain mutation.
Results
Case Report
The patient was a 30-year-old female never-smoker who presented with dyspnea and was found to have a large left upper lobe mass and mediastinal adenopathy. Bronchoscopic biopsy demonstrated lung adenocarcinoma positive for an EGFR exon 19 deletion mutation. Due to the extent of locally advanced disease and the presence of a potentially targetable genotype, she was initiated on induction afatinib prior to definitive therapy. Surprisingly, scans after 4 weeks demonstrated mild growth of the lung mass. She was switched to induction chemotherapy with cisplatin and pemetrexed and had a radiographic response, which was then followed by concurrent chemoradiation. Unfortunately, she developed progression as she was completing radiotherapy, with a growing left cervical lymph node. This was biopsied and demonstrated recurrent lung adenocarcinoma. Imaging further demonstrated new bone metastases. She received a brief course of erlotinib, but her disease continued to progress (Fig. 1A).
Response and resistance to savolitinib and osimertinib in a patient with EGFR-mutant NSCLC harboring MET amplification. A, Treatment timeline. B, Despite being refractory to prior EGFR inhibitors, a dramatic response to the combination was seen after 4 weeks. C, After 36 weeks on therapy, an isolated lung nodule appeared and was resected; however, further progression was seen. D, Serial plasma genotyping demonstrates response of the known EGFR exon 19 deletion followed by progression with development of an acquired D1228V mutation, in the absence of EGFRT790M.
Targeted next-generation sequencing (NGS) of the patient's biopsy at recurrence confirmed the presence of the known EGFR exon 19 deletion and additionally identified high-level MET amplification, potentially explaining her drug resistance (Supplementary Fig. S1A); no other resistance mechanisms, including EGFRT790M, were detected. The patient then enrolled in a phase I clinical trial (NCT02143466) combining the MET TKI savolitinib (800 mg once daily) with the mutant-selective EGFR TKI osimertinib (80 mg once daily). She experienced a dramatic clinical response, with near-complete resolution of progressive cervical lymphadenopathy after 4 weeks of treatment (Fig. 1B). However, after 36 weeks on therapy, the patient developed a new growing pulmonary nodule consistent with objective disease progression (Fig. 1C). The lung nodule was removed surgically in an attempt at achieving local control of her progressive disease, but the patient had further disease progression in the lung and was removed from study therapy.
Tumor Sequencing Identifies a Secondary MET Kinase Domain Mutation
In search of genomic changes mediating resistance to combined EGFR and MET inhibition, we performed targeted NGS of the biopsy obtained following the development of resistance to savolitinib and osimertinib. Tumor NGS confirmed the high-level MET amplification and the EGFR exon 19 deletion, which had both been present on prior tumor NGS (Supplementary Fig. S1A and S1B). In addition, tumor NGS found 8 variants present at >5% allelic fraction (AF), 6 of which were also detected in a biopsy specimen from before treatment with savolitinib and osimertinib (Supplementary Table S1). Of the two acquired variants detected, one affected the MET kinase domain, D1228V (3683A>T), and was selected as a candidate variant for explaining drug resistance (Supplementary Fig. S1C).
Serial Plasma Genotyping Detects an Acquired MET Mutation as Resistance Develops
We performed serial genotyping of plasma cell-free DNA (cfDNA) collected during the treatment period using droplet digital PCR (ddPCR). Plasma ddPCR detected high levels of the EGFR exon 19 deletion prior to therapy, and rapid reduction of these levels after therapy was started, with levels becoming undetectable after 12 weeks of treatment (Fig. 1D). At the time of disease progression (36 weeks), we could again detect the EGFR exon 19 deletion. We developed a new ddPCR assay for METD1228V and found that this mutation emerged concurrently with the progression of the EGFR exon 19 deletion in plasma. Levels of mutant EGFR and MET in plasma continued to increase as the patient progressed further on postprogression therapy (Fig. 1C and D).
Protein Modeling Predicts Differences in Drug Binding between Type I and Type II MET Inhibitors
We developed models of the wild-type (WT) and mutant MET protein to study drug binding. Savolitinib, just as with other type I MET TKIs, binds to the MET kinase in its active conformation. Our modeling demonstrates that the binding with wild-type MET relies upon a key π-π interaction with Tyr1230, present in the MET kinase activation loop (Fig. 2A). MET Asp1228 and Lys1110 residues are critical for maintaining the activation loop in a position conducive to savolitinib binding through their strong electrostatic interactions. The presence of the D1228V mutation interrupts the salt bridge connecting Asp1228 and Lys1110, which leads to the repositioning of the activation loop and loss of the π-π interaction between MET and savolitinib (Fig. 2B), predicting impaired drug binding. We subsequently modeled the binding of the MET kinase to cabozantinib, a type II inhibitor, which binds MET in its inactive conformation. The binding to wild-type MET does not rely on the availability of Tyr1230 (Fig. 2C), and drug binding is unaffected by the D1228V mutation (Fig. 2D).
Simulations of savolitinib (A and B) and cabozantinib (C and D) binding to the wild-type (A and C) and mutant (B and D) MET kinase. Kinase inhibitors are shown in green, and amino acids are shown in yellow. A, Savolitinib binding to wild-type MET relies upon a π-π interaction with Y1230 in the activation loop (blue line), stabilized by a salt bridge between D1228 and K1110. B, METD1228V results in loss of this salt bridge, resulting in repositioning of Y1230 and loss of π-π stacking (blue line). C, Cabozantinib binds to the inactive conformation of wild-type MET and does not rely upon this π-π interaction with Y1230. D, Cabozantinib binding is not affected by V1228, which is remote from the drug-binding sites.
In Vitro Studies Demonstrate That METD1228V Induces Resistance to Type I but Not Type II MET Inhibitors
To validate our protein modeling studies, we performed a signaling analysis in 293T cells engineered to express either full-length wild-type MET or full-length METD1228V. We found that both type I and II MET inhibitors are able to inhibit autophosphorylation of wild-type MET. However, the METD1228V mutation rendered all type I MET inhibitors ineffective, as evidenced by the failure of savolitinib, crizotinib, and INC280 to inhibit MET phosphorylation (Fig. 3A). In contrast, the type II MET inhibitors cabozantinib, MGCD265, and merestinib (LY2801653) were able to effectively suppress MET phosphorylation in the presence of the D1228V mutation. To further validate our biochemical findings, we introduced the oncogenic TPR–MET fusion (TPR–METWT) or TPR–MET harboring the D1228V mutation (TPR–METD1228V) into Ba/F3 cells or into the EGFR-mutant EGFR TKI-sensitive PC9 cell line, and analyzed the ability of savolitinib alone, cabozantinib alone, or either agent in combination with gefitinib to inhibit viability and/or downstream effectors of the MET and EGFR signaling axes (Fig. 3B–D). The TPR–METWT Ba/F3 cells were highly sensitive to both savolitinib and cabozantinib in the dose-escalated MTS viability assay (Fig. 3B). In contrast, the TPR–METD1228V Ba/F3 cells failed to respond to savolitinib but maintained sensitivity to cabozantinib (Fig. 3B). In addition, the introduction of the wild-type or the D1228V-mutant TPR–MET fusion into PC9 cells (Supplementary Fig. S2) imparted resistance to gefitinib in both cases as expected (Fig. 3C), but the D1228V mutant showed a differential response to the combination of gefitinib and savolitinib, as compared with the wild-type. Although the TPR–METWT PC9 cells responded to gefitinib in combination with either savolitinib or cabozantinib, the D1228V mutant was sensitive only to the gefitinib and cabozantinib combination (Fig. 3C). A Western blot analysis of the EGFR and MET signaling in the TPR–MET bearing PC9 cell derivatives further supported our cell viability studies. In the TPR–METWT PC9 cells, phosphorylation of both EGFR and MET along with the MAPK/ERK pathway were completely inhibited following exposure to either the gefitinib/cabozantinib or gefitinib/savolitinib treatments (Fig. 3D). In contrast, the gefitinib/savolitinib combination failed to inhibit MET phosphorylation in the TPR–METD1228V PC9 cells, consequently resulting in sustained MAPK/ERK signaling. However, the METD1228V mutant retained sensitivity to the gefitinib/cabozantinib combination, as evidenced by the suppression of MET and EGFR phosphorylation as well as inhibition of downstream signaling pathways (Fig. 3D). These cellular and biochemical studies further support our conclusions that Asp1228 is a critical residue for the binding of type I MET inhibitors, and that type II MET inhibitors may successfully address resistance caused by a mutation at this residue.
