Abstract
MET or hepatocyte growth factor (HGF) receptor pathway signaling mediates wound healing and hepatic regeneration, with pivotal roles in embryonic, neuronal, and muscle development. However, dysregulation of MET signaling mediates proliferation, apoptosis, and migration and is implicated in a number of malignancies. In non–small cell lung cancer (NSCLC), aberrant MET signaling can occur through a number of mechanisms that collectively represent a significant proportion of patients. These include MET or HGF protein overexpression, MET gene amplification, MET gene mutation or fusion/rearrangement, or aberrations in downstream signaling or regulatory components. Responses to MET tyrosine kinase inhibitors have been documented in clinical trials in patients with MET-amplified or MET-overexpressing NSCLC, and case studies or case series have shown that MET mutation/deletion is a biomarker that is also predictive of response to these agents. However, other recent clinical data have highlighted an urgent need to elucidate optimal biomarkers based on genetic and/or protein diagnostics to correctly identify patients most likely to benefit in ongoing clinical trials of an array of MET-targeted therapies of differing class. The latest advances in the development of MET biomarkers in NSCLC have been reviewed, toward establishing appropriate MET biomarker selection based on a scientific rationale. Mol Cancer Ther; 16(4); 555–65. ©2017 AACR.
Introduction
The MET proto-oncogene was originally identified as a fusion partner with the translocated promoter region of the TPR gene in a chemically transformed osteosarcoma-derived cell line (1). The MET protein encoded by this proto-oncogene was later found to be a transmembrane receptor tyrosine kinase (RTK) activated by an endogenous ligand, scatter factor, or hepatocyte growth factor (HGF; refs. 2–5). Binding of HGF to MET [or HGF receptor (HGFR)] results in receptor dimerization and phosphorylation of tyrosine residues, ultimately leading to the phosphorylation of intracellular docking sites where adaptor proteins bind to activate downstream signaling (4, 6, 7). Activated signaling pathways include mitogen-activated protein kinase (MAPK), phosphoinositide 3-kinase (PI3K)/AKT (protein kinase B), signal transducer and activator of transcription proteins, and nuclear factor-κB (8–10). In normal physiology, these signaling pathways promote activation of cytoplasmic and nuclear processes, which lead to a variety of cellular functions, including proliferation and protection from apoptosis (8–10). The MET pathway also mediates functions such as wound healing and hepatic regeneration, and has pivotal roles in normal liver development (11), embryonic placental development, and the formation of muscle and neurons (12–15).
HFG/MET in Lung Cancer
Dysregulation of MET signaling–mediated proliferation, apoptosis, and migration through overexpression of MET and amplification or mutation of the MET gene has been widely demonstrated in oncogenic processes across multiple tumor types and has been reviewed elsewhere (10, 16–18). Moreover, it is notable that all three of these mechanisms of MET/MET dysregulation have been documented in non–small cell lung cancer (NSCLC; refs. 19–22). Early studies established that MET can be overexpressed or activated [as measured by phosphorylation of the catalytic domain as well as juxtamembrane (JM) domain], or the gene mutated (in the semaphorin or JM domains) and/or amplified in lung cancer. For instance, studies on small cell lung cancer (SCLC) cell lines established the multipurpose nature of MET/HGF pathway activation during tumor progression and invasion, which occurs via dysregulation of diverse biological functions such as proliferation and differentiation, transcriptional control, cell-cycle G1/S checkpoint, cytoskeletal functions, survival, motility, and apoptosis (23). Both epidermal growth factor receptor (EGFR) and MET are widely expressed on cancer cells, and both RTKs are implicated in these diverse signaling processes. Indeed, synergistic effects of epidermal growth factor (EGF) and HGF on proliferation, and additive effects on motility, were noted in preclinical studies in NSCLC cells. For example, increased membrane ruffling to form a motile cell surface was observed when cells were stimulated with HGF and EGF independently, and when these growth factors were combined, an additive effect was observed (24). These preclinical studies suggested that the combination of inhibitors for MET and EGFR RTKs could potentially produce synergistic antitumor effects (24). Indeed, a synergistic effect on inhibition of cell proliferation and apoptosis was seen when a novel first-generation MET inhibitor, SU11274, was combined with EGFR inhibitors such as AG1478 or gefitinib (24). The synergism and cross-talk of EGFR and MET pathways in NSCLC, and the potential of simultaneous inhibition, were thus recognized in these studies.
Therapy combining various targeted agents or other standard therapies with MET inhibitors has also been explored in other preclinical studies, including synergistic effects of when MET and EGFR inhibitors were combined in NSCLC cell lines, and head and neck squamous cell carcinoma (SCCHN) cells (24, 25). Combined inhibition of MET and mammalian target of rapamycin resulted in cooperative inhibition of cell growth in TPR-MET transformed cells expressing a constitutively active variant of MET, and in MET-expressing NSCLC cells (26). MET inhibition is also synergistic with cisplatin in SCCHN cancer cell lines (25) and also appears to be synergistic with radiation in cell lines in some studies, although crizotinib (Pfizer), an inhibitor of MET, anaplastic lymphoma kinase (ALK), and ROS1 kinases, failed to enhance the effects of radiation in SCCHN xenograft models (27). Furthermore, MET inhibition was synergistic with topoisomerase-I inhibition in SCLC cell lines, with a significant positive correlation observed between MET gene copy number (GCN) and topoisomerase-I nuclear expression (28).
Role of MET and HGF in Induction of Drug Resistance
Several RTKs often expressed on cancer cells activate signaling pathways that converge on common downstream effectors. This “overlap” may, in part, be implicated in resistance to RTK-based treatment, which is commonly observed in cancer patients. Resistance to targeted agents may be mediated by the stroma, and preclinical investigations of growth-factor–driven resistance have shown that increasing levels of ligands with a similar signaling output, such as PI3K or MAPK, may confer innate or acquired resistance to inhibitors of an oncogenic kinase (29, 30).
Loss of drug sensitivity in tumor cells through exposure to RTK ligands in the tumor microenvironment was demonstrated in BRAF-mutant melanoma cells, with HGF conferring resistance to the BRAF inhibitor PLX4032 (vemurafenib; ref. 30). Stromal secretion of HGF resulted in activation of MET, thereby reactivating the MAPK and PI3K/AKT pathways (30). Consequently, in cell models at least, it is feasible that dual inhibition of RAF and MET can reverse drug resistance (30). These data highlight the redundancy of RTK-transduced signaling in cancer cells and the wide-ranging effects of RTK ligands that lead to innate and acquired resistance, which may potentially be overcome through combinations of targeted agents (29, 30).
MET amplification in NSCLC is implicated in acquired resistance to EGFR inhibitors and has been reported in approximately one-fifth of cases with EGFR inhibitor resistance (31–34). This provides further therapeutic rationale for combinations of MET inhibitors with EGFR inhibitors to treat selected patients with NSCLC.
MET Biomarkers in NSCLC
The varied mechanisms of MET activation in lung cancer, including overexpression of MET and/or its ligand, HGF, and genetic alterations to MET (e.g., mutations, amplification, translocation, or dysregulated transcription), and impaired degradation of MET, provide an array of potential biomarkers (Table 1). The challenge now faced is to identify which of these biomarkers holds the most promise to select appropriate patients for MET-targeted treatment with the array of agents currently in development.
Reported incidence and functional consequences of MET biomarkers in lung cancer
MET dysregulation . | Functional consequences . | Biomarker . | Reported incidence in samples from lung cancer patients . | Reference . |
---|---|---|---|---|
Gene overexpression | Reduces or removes the requirement for ligand activation, leading to sustained or altered signaling properties of the receptor | MET/p-MET expression by IHC | NSCLC | |
37% IHC ≥2+ | (36) | |||
61% IHC ≥2+ | (37) | |||
ADC | ||||
35% | (76) | |||
67% IHC ≥2+ | (37) | |||
72% | (77) | |||
SCC | ||||
38% | (77) | |||
p-MET in NSCLC | ||||
67% | (35, 37) | |||
HGF expression | Ligand-induced activation could cause sustained or altered signaling | Circulating plasma HGF | Elevated in SCLC | (40) |
Gene mutation | MET mutation can lead to reduced degradation, with consequent overexpression and sustained or altered signaling | MET exon 14 skipping mutation | ADC | |
3% | (43–46) | |||
4% | (43, 44, 46, 78) | |||
SCC | ||||
2% | (45) | |||
Other lung cancer subtypes | ||||
2% | (43, 44, 46) | |||
1%–8% | (45) | |||
Gene amplification | Can lead to overexpression and reduce or remove the requirement for ligand activation, leading to sustained or altered signaling properties of the MET receptor | MET GCN | Newly diagnosed ADC | |
MET/CEP7 ratio | 2% | (46) | ||
4% | (21, 56) | |||
5% | (55) | |||
EGFR TKI-resistant ADC | ||||
5% | (57, 58) | |||
17% | (33) | |||
21% | (31) | |||
22% | (34) | |||
Gene rearrangement | May reduce or remove the requirement for ligand activation, leading to sustained or altered signaling properties of the MET receptor | MET rearrangement | Identified in an ADC patient | (67) |
Downstream MET signaling alteration | Decreases RTK turnover to perpetuate oncogenic activation of MET | CBL mutation or LOH | Detected in NSCLC patients | (69) |
MET dysregulation . | Functional consequences . | Biomarker . | Reported incidence in samples from lung cancer patients . | Reference . |
---|---|---|---|---|
Gene overexpression | Reduces or removes the requirement for ligand activation, leading to sustained or altered signaling properties of the receptor | MET/p-MET expression by IHC | NSCLC | |
37% IHC ≥2+ | (36) | |||
61% IHC ≥2+ | (37) | |||
ADC | ||||
35% | (76) | |||
67% IHC ≥2+ | (37) | |||
72% | (77) | |||
SCC | ||||
38% | (77) | |||
p-MET in NSCLC | ||||
67% | (35, 37) | |||
HGF expression | Ligand-induced activation could cause sustained or altered signaling | Circulating plasma HGF | Elevated in SCLC | (40) |
Gene mutation | MET mutation can lead to reduced degradation, with consequent overexpression and sustained or altered signaling | MET exon 14 skipping mutation | ADC | |
3% | (43–46) | |||
4% | (43, 44, 46, 78) | |||
SCC | ||||
2% | (45) | |||
Other lung cancer subtypes | ||||
2% | (43, 44, 46) | |||
1%–8% | (45) | |||
Gene amplification | Can lead to overexpression and reduce or remove the requirement for ligand activation, leading to sustained or altered signaling properties of the MET receptor | MET GCN | Newly diagnosed ADC | |
MET/CEP7 ratio | 2% | (46) | ||
4% | (21, 56) | |||
5% | (55) | |||
EGFR TKI-resistant ADC | ||||
5% | (57, 58) | |||
17% | (33) | |||
21% | (31) | |||
22% | (34) | |||
Gene rearrangement | May reduce or remove the requirement for ligand activation, leading to sustained or altered signaling properties of the MET receptor | MET rearrangement | Identified in an ADC patient | (67) |
Downstream MET signaling alteration | Decreases RTK turnover to perpetuate oncogenic activation of MET | CBL mutation or LOH | Detected in NSCLC patients | (69) |
Abbreviations: ADC, adenocarcinoma; CEP, chromosome enumeration probe; SCC, squamous cell carcinoma.
Expression of MET/p-MET and HGF proteins
Human tissue microarray studies reveal that while HGF is widely expressed in both tumor and nonmalignant tissue, MET is differentially expressed in solid tumors (35). Positive staining for MET and HGF, which is thought to have a progenitor role, was observed in the bronchioalevolar junctions of lung tumors (35). Overexpression of MET occurs with a high frequency (35%–72%) in NSCLC tumors (Table 1). For example, in a recent study of more than 200 NSCLC samples, 37% were scored as immunohistochemistry (IHC) ≥2+ for MET expression (36). In another study, MET was detected in eight of nine NSCLC cell lines and in all of 23 NSCLC tumor samples examined (37). Furthermore, 61% of tumor tissues strongly expressed MET, with high MET expression being confirmed as particularly common in adenocarcinoma (67%). It is noteworthy that increased levels of circulating MET mRNA, which were 1.4–8 times above normal concentrations in 68% of patients with overexpression of MET in their tumors, have been found to correlate with early disease recurrence in NSCLC patients (38).
In addition to total levels of the protein, MET activated by ligand to induce phosphorylation of the JM domain can be assayed via phospho-MET (p-MET). Using IHC, specific expression of p-MET has been observed in approximately two-thirds of lung cancer samples and has also been reported to be preferentially expressed at the invasive fronts of NSCLC tumors (35, 37). In a study of the expression and prognostic role of MET, p-MET, and HGF in patients with NSCLC and SCLC (N = 129), high expression of two specific forms of p-MET—cytoplasmic expression of Y1003 and nuclear expression of Y1365—was associated with significantly worse overall survival [OS; P = 0.016; hazard ratio (HR), 1.86; 95% confidence interval: 1.12–3.07; and P = 0.034; HR: 1.70; 95% confidence interval: 1.04–2.78, respectively]. Consequently, specific forms of p-MET may also serve as potential biomarkers in lung cancer (39).
Serum HGF (s-HGF) is also feasible as a biomarker in MET-addicted cancer. Levels of s-HGF were significantly elevated in patients with SCLC compared with healthy individuals (0.40 ± 0.17 vs. 0.26 ± 0.093, P = 0.0083; ref. 40). A high s-HGF level has also been shown to be associated with epithelial-to-mesenchymal transition in patients with SCLC (N = 112; ref. 41). Of these patients with stage IV disease, increased s-HGF levels at response evaluation (P = 0.042) and at progression (P = 0.003) were associated with poor outcome (41).
MET mutations
Mutations in the splice site of MET that result in skipping of exon 14 are important molecular drivers in NSCLC (37, 42). Such mutations have recently been shown to occur in 3% to 4% of NSCLC adenocarcinomas, 2% of squamous cell carcinomas, and 1% to 8% of other subtypes of lung cancer (Table 1; refs. 43–46). Novel JM domain (exon 14/15) mutations in MET were first identified in SCLC, in three of 10 cell lines and in four of 32 SCLC tumor tissue samples examined (42). MET alterations included two different missense mutations in the JM domain (R988C found in NCI-H69 and H249 cell lines; and T1010I in SCLC tumor sample). Also, a semaphorin domain missense mutation (E168D in SCLC tumor sample), two-base-pair insertional mutations [IVS13-(52–53)insCT in SCLC tumor samples] within the pre-JM intron 13, as well as an alternative transcript involving exon 10 (H128 cell line), were identified (42). The two reported JM mutations affected cell proliferation, resulting in small but significant growth factor independence in the IL3-dependent BaF3 cell line, and were found to regulate cell morphology and adhesion, and enhanced tumorigenicity when introduced into an SCLC cell line (42). The JM mutations also altered MET RTK signaling, resulting in preferentially increased constitutive tyrosine phosphorylation of various cellular proteins, with significant implications in cytoskeletal functions and metastatic potential. These novel MET JM gain-of-function somatic mutations associated with a more aggressive phenotype were among those mutations subsequently identified in NSCLC adenocarcinoma tissues (R988C, R988C + T1010I, S1058P, and an alternative exon 14 splice variant product skipping the entire JM domain; ref. 37).
Using NSCLC tissues and cell lines, we (37) and Kong-Beltran and colleagues (22) functionally characterized tumor-specific somatic intronic MET mutations, which immediately flank exon 14 (22). Exon 14 was found to encode the JM domain and Y1003 residue that serves as the binding site for casitas B-lineage lymphoma (CBL), the E3 ubiquitin ligase that controls MET turnover (22). In each case of MET exon 14 skipping, confirmed by reverse transcriptase polymerase chain reaction, the result was a decrease in the ubiquitination of MET and consequent delayed receptor downregulation after stimulation with HGF. Downstream ligand-dependent signaling through MAPK was also prolonged in cells transfected with a MET exon 14 splice variant (22). Furthermore, a xenograft model of Rat1A fibroblasts stably transfected with a MET exon 14 splice variant developed particularly aggressive tumors compared with wild-type MET (22). Overall, the biological effects of MET JM mutations are increased tumorigenicity, reduced adhesion, and disorganized cytoarchitecture compared with wild-type, increased cell survival, motility and migration, increased phosphorylation of focal adhesion proteins, such as paxillin, and HGF independence (23, 37, 47).
Elegant studies have validated the nematode C. elegans as a model for rapid screening of the functional aspects of mutant forms of cancer genes, with METR988C and METT1010I harboring wild-type or frequently seen mutant forms of MET in lung cancer (48). Expression of these mutations in this model led to low fecundity and abnormal vulval development characterized by hyperplasia. In addition, exposure of MET-mutant transgenic worms to nicotine resulted in enhanced abnormal vulval development, fecundity, and locomotion (48). This model also demonstrated colocalization of MET and EGL5 (PAX8 ortholog) proteins in embryos of the organism (49). PAX8 provides signals for growth and motility of NSCLC cells and is required for MET and RON expression; also, it may have therapeutic potential (49).
Responses to the MET inhibitors crizotinib and cabozantinib have been documented in case reports of patients with lung adenocarcinoma and MET exon 14 alterations (Table 2; refs. 44, 50–52). In phase I clinical studies of the investigational MET inhibitor capmatinib (INC280, Novartis), two patients with MET-dependent NSCLC and MET exon 14 alterations were identified by comprehensive genomic profiling. In one patient with large-cell carcinoma who was treated for over 5 months, there was a partial response comprising a 53% reduction in tumor, and in the other patient, who had squamous NSCLC that had failed prior therapies and was treated for 13 months, there was a partial response comprising a 61% reduction in tumor (43). MET mutations in the semaphorin domain have been shown to affect ligand binding: MET-N375S, the most frequent mutation of MET, most common among male smokers and squamous cell carcinoma, confers resistance to MET inhibition based on HGF binding, molecular modeling, and apoptotic susceptibility to MET inhibitor studies (53).
Case reports and series of patients with lung adenocarcinomas and MET exon 14 alterations responding to MET inhibitors
Patient no. . | Age and gender . | Smoking status . | MET exon 14 alterations . | MET IHC . | MET amp . | Best response . | Reference . |
---|---|---|---|---|---|---|---|
1 | 86 M | NS | Splice acceptor deletion | 2+ | NA | PR to crizotinib | Jenkins et al., 2015 (50) |
2 | 71 M | ES | D1028H Splice donor mutation | NA | No | PR to crizotinib | Waqar et al., 2015 (51) |
3 | 76 F | ES | D1010H | NA | NA | PR to crizotinib | Mendenhall et al., 2015 (52) |
4 | 80 F | NS | Splice donor mutation | 3+ | Yes | CR (PERCIST) to cabozantinib | Paik et al., 2015 (44) |
5 | 78 M | ES | Splice donor deletion | 3+ | NA | PR to crizotinib (lung) PD to crizotinib (liver) | Paik et al., 2015 (44) |
6 | 65 M | ES | Splice donor mutation | NA | NA | PR to crizotinib | Paik et al., 2015 (44) |
7 | 90 F | NS | Splice donor mutation | NA | NA | PR to crizotinib | Paik et al., 2015 (44) |
8 | 67 F | NS | D1028N Splice donor mutation | NA | NA | PR to crizotinib | Mahjoubi et al., 2016 (79) |
Patient no. . | Age and gender . | Smoking status . | MET exon 14 alterations . | MET IHC . | MET amp . | Best response . | Reference . |
---|---|---|---|---|---|---|---|
1 | 86 M | NS | Splice acceptor deletion | 2+ | NA | PR to crizotinib | Jenkins et al., 2015 (50) |
2 | 71 M | ES | D1028H Splice donor mutation | NA | No | PR to crizotinib | Waqar et al., 2015 (51) |
3 | 76 F | ES | D1010H | NA | NA | PR to crizotinib | Mendenhall et al., 2015 (52) |
4 | 80 F | NS | Splice donor mutation | 3+ | Yes | CR (PERCIST) to cabozantinib | Paik et al., 2015 (44) |
5 | 78 M | ES | Splice donor deletion | 3+ | NA | PR to crizotinib (lung) PD to crizotinib (liver) | Paik et al., 2015 (44) |
6 | 65 M | ES | Splice donor mutation | NA | NA | PR to crizotinib | Paik et al., 2015 (44) |
7 | 90 F | NS | Splice donor mutation | NA | NA | PR to crizotinib | Paik et al., 2015 (44) |
8 | 67 F | NS | D1028N Splice donor mutation | NA | NA | PR to crizotinib | Mahjoubi et al., 2016 (79) |
Abbreviations: CR, complete response; ES, ever-smoker; F, female; M, male; NA, not available; NS, never-smoker; PERCIST, PET Response Criteria in Solid Tumors; PR, partial response.
Larger clinical studies focusing on patients with MET mutations, particularly exon 14 alterations, are now required to prospectively obtain response rates associated with MET inhibitors in this patient population. These studies will also need to evaluate any association of MET inhibitor efficacy with known disease driving mutations such as KRAS, EGFR, BRAF, or ALK (54). Nonetheless, because the potentially actionable genetic alterations within exon 14 are diverse, in-depth molecular profiling of all lung cancer patients, irrespective of additional disease driving mutations, is recommended (54).
MET amplification
In NSCLC, amplification of MET typically occurs in about 2% to 5% of newly diagnosed adenocarcinomas (Table 1; refs. 21, 46, 55, 56). Interestingly, a much greater incidence of MET amplification, ranging from 5% to 22%, has been reported in patients with NSCLC following treatment with erlotinib/gefitinib (Table 1; refs. 31, 33, 34, 57, 58). Amplification of MET (and overexpression of the protein) is also a common event in brain metastases of NSCLC (59). Furthermore, fluorescence in situ hybridization (FISH)–positive MET status predicts worse survival in patients with advanced NSCLC (56, 60). An analysis of OS based on MET FISH status-derived GCN revealed that increased GCN is an independent negative prognostic factor in surgically resected NSCLC, with OS of 25.8 months for patients with MET ≥5 copies/cell compared with 47.5 months for patients with MET <5 copies/cell (P = 0.0045; ref. 21). These data support further investigation of anti-MET therapeutic strategies in appropriately selected patients (21). The question remains as to how biomarkers should be best utilized for patient selection.
While preclinical studies indicated that agents targeting MET are effective in the presence of high levels of MET gene amplification (61, 62), there is currently no clear consensus on how best to determine MET gene amplification in the clinical setting. In a phase I study of capmatinib, preliminary antitumor activity was seen in patients with EGFR-wild-type NSCLC and a high level of MET amplification (MET GCN ≥6; ref. 63), while a study of capmatinib plus gefitinib in patients with EGFR-mutant, MET-positive NSCLC reported an overall response rate of 50% in patients with MET GCN ≥6 (64). Although, based on preliminary data, MET GCN appears to be a good predictive biomarker, the FISH MET/chromosome enumeration probe 7 (CEP7) ratio is also a relatively simple primary measure of amplification. In a study of crizotinib in MET-amplified NSCLC, as determined by MET/CEP7 ratio [≥1.8 to ≤2.2 (low), >2.2 to <5 (intermediate) and ≥5 (high)], antitumor activity was seen in patients with intermediate and high ratios, with a high response to therapy only observed in individuals in the gene ratio ≥5 category (65). One possible drawback of using the MET/CEP7 gene ratio is that this technique may not identify all amplified patients due to the unique pathophysiology of NSCLC. In some cases, amplicons occur that include the centromere control protein and the MET gene or the centromere protein but not the MET gene; in the latter case, the ratio may be falsely lowered (66).
MET rearrangement
Compared with mutations and amplification of MET, gene rearrangements are less well documented. However, the kinase fusion KIF5B–MET has been identified in a case of lung adenocarcinoma, and it is feasible that this translocation event could potentially account for a significant portion of MET-driven oncogenesis (67). This fusion-driven activation of MET is most likely due to constitutive dimerization and is likely to be an actionable target for drug-induced inhibition, similar to other fusions in lung cancer such as those involving ALK, ROS1, and RET (67).
MET processing: degradation/transcription
CBL is a mammalian gene encoding an E3 ubiquitin ligase and adaptor protein involved in cell signaling and protein ubiquitination (68). CBL thus has an important role in RTK downregulation and degradation (68). Somatic mutations [or loss of heterozygosity (LOH)] in CBL have been detected in NSCLC patients, and expression of these mutations in cell lines was found to result in increased proliferation and cell motility (69). CBL LOH is associated with either MET or EGFR mutations and may contribute to their oncogenic potential (69). As already described, it is noteworthy that the JM domain of MET is involved in the binding and E3 activity of CBL, and MET JM mutations (e.g., Y1003) therefore affect CBL binding and decrease RTK turnover to perpetuate oncogenic activation of MET (22, 70).
Discussion: Potential of MET as a Biomarker in Lung Cancer
The growing prominence of MET inhibition in lung cancer is reflected in the number of molecular aberrations with oncogenic potential that occur in this disease, and in the number and diversity of MET-targeted agents in clinical development in this indication. These include the monoclonal antibodies emibetuzumab, ficlatuzumab, and rilotuzumab, and tyrosine kinase inhibitors (TKI) such as crizotinib, tepotinib, cabozantinib, and capmatinib (Table 3). Recent negative or disappointing clinical trials results pose the question as to whether the biomarkers and their related cutoff values have been chosen appropriately to select patients for enrollment in all studies to date. For instance, despite positive phase II data (71), the phase III METLung trial (N = 499) of onartuzumab plus erlotinib failed to show clinical benefit compared with placebo plus erlotinib in patients with MET+ NSCLC (Table 4; ref. 72). In this study, patient biomarker–based selection of patients with MET-overexpressing tumors as measured by IHC (MET SP44) was used to determine eligibility. These negative data suggest that IHC may not be sufficiently sensitive as a diagnostic tool for MET positivity; its use as a standard biomarker for overexpression is further compromised by the lack of standardized interpretation or consensus on optimized cutoff values. Moreover, MET protein expression may have a low predictive value as a tool to detect MET activation and may not always reflect tumor dependency on MET signaling (73). Heterogeneity in the expression of MET within a tumor or across metastatic sites may also lead to unreliable results.
Predictive biomarkers evaluated in clinical studies of MET-targeted therapies in patients with NSCLC

Circulating HGF (cHGF) or MET are attractive potential alternative biomarkers for ligand or receptor overexpression, respectively. For example, elevated cHGF, as measured by ELISA, has been used as a pharmacodynamic biomarker of activity with onartuzumab (74). However, in cases of ligand-independent activation of MET, it is feasible that monoclonal antibody therapy, without drug internalization, may be a less effective therapeutic strategy than TKIs that target the receptor protein kinase directly. Recent data suggest that HGF/MET protein expression alone may be an oversimplification of the oncogenic driver status of “MET-positive” NSCLC, where mutations or translocations and amplification reduce the requirement for ligand activation and lead to sustained or altered signaling properties of the receptor. Although IHC data have been shown to correlate with MET amplification (66), clinical study biomarker data (summarized in Table 4) have not confirmed any clear-cut relationships between MET mutation, amplification, and overexpression, when collectively applied as predictive biomarkers for MET-targeted therapy. IHC-based MET expression has not been a successful biomarker approach in clinical studies of monoclonal antibodies, and current clinical and biomarker data suggest that genetic changes in MET, in particular gene amplification, may be the preferred biomarkers for MET TKI therapy (21, 63–65). The data summarized in Table 4 also indicate that biomarkers for MET TKI therapy need to be optimized based on not only MET amplification but also MET mutation or translocation status, which constitutes an additional and numerically significant (>4%) molecular subgroup of NSCLC (46). Mutations or altered expression of signaling proteins downstream of MET signaling, such as CBL mutation, are also emerging biomarkers in NSCLC and extend further the range and diversity of potential MET-related biomarkers in this disease. There is therefore an urgent need to elucidate both optimal biomarkers for MET dysregulation, and their application, based on our growing understanding of this oncogenic driver in NSCLC. To facilitate this goal, the medical oncologist and pathologist now have at their disposal a panel of genetic and protein biomarkers for MET dysregulation that together constitute a significant proportion of lung cancer molecular subgroups. Indeed, current data indicate that panels of MET biomarkers are likely to be necessary in the future, and measurements of potential biomarkers should therefore be included in new clinical trial designs for MET inhibitors to facilitate the robust definition of appropriate therapies for specific MET-dysregulated NSCLC subsets. Furthermore, the recent report of a response to crizotinib in a patient with lung adenocarcinoma with MET copy-number gain but without a high-level MET/CEP7 ratio, MET overexpression, or exon 14 splicing mutation (75) indicates that the list of independent predictive biomarkers for response to MET inhibitors may well be extended further. Importantly, since alterations in MET gene status have been found to occur in both untreated patients and those who have developed resistance to other targeted therapies, new clinical study designs should consider both patient groups. This highlights the future importance of both upfront and resistance-based genetic testing in lung cancer patients, which should include MET as the probable next major biomarker in lung cancer.
Disclosure of Potential Conflicts of Interest
No potential conflicts of interest were disclosed.
Acknowledgments
This manuscript was written by the author with medical editorial assistance provided by Matthew Naylor PhD, funded by Novartis Pharmaceuticals Corporation.
Grant Support
Medical editorial assistance was funded by Novartis Pharmaceuticals Corporation.
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.