Biomarker development in MET-targeted therapy

Activation of the MET receptor tyrosine kinase by its ligand, hepatocyte growth factor (HGF), has been implicated in a variety of cellular processes, including cell proliferation, survival, migration, motility and invasion, all of which may be enhanced in human cancers. Aberrantly activated MET/HGF signaling correlates with tumorigenesis and metastasis, and is regarded as a robust target for the development of novel anti-cancer treatments. Various clinical trials were conducted to evaluate the safety and efficacy of selective HGF/MET inhibitors in cancer patients. There is currently no optimal or standardized method for accurate and reliable assessment of MET levels, or other biomarkers that are predictive of the patient response to MET-targeted therapeutics. In this review, we discuss the importance of accurate HGF/MET signal detection as a predictive biomarker to guide patient selection for clinical trials of MET-targeted therapies in human cancers.


INtrODUctION
Receptor tyrosine kinases (RTKs) are involved in many vital processes including mammalian development, cell function and tissue homeostasis. The RTK MET and its ligand hepatocyte growth factor (HGF) activate multiple signaling pathways mediating embryogenesis, tissue regeneration and wound repair under normal physiological conditions [1]. However, an aberrant MET/ HGF axis results in cell migration and survival and promotes tumor development and progression [2][3][4].
The MET proto-oncogene, located on chromosome 7 (7q21-31), is widely expressed in the epithelial cells of organs such as the liver, lung, gastrointestinal tract, and kidney during embryogenesis and adulthood [5]. HGF binding to the extracellular domain of MET results in its homodimerization and autophosphorylation at multiple tyrosine residues, including Y1234 and Y1235 in the kinase domain and subsequently Y1349 and Y1356 in the carboxy-terminal tail [6,7]. Tyrosine phosphorylation of MET results in its activation and recruitment of signaling effectors, including adaptor proteins Grb2, Gab-1, Src and SHC, and subsequent phosphorylation of downstream transducers such as PI3K, ERK, PLC-γ, FAK and STAT3 [8]. HGF-induced MET activation can trigger cell proliferation, survival, motility, invasion, angiogenesis and branching morphogenesis [9][10][11].
Gab-1 binding to MET through GRB2 primarily leads to activation of the phosphoinositide 3-kinase (PI3K)/Akt and mitogen-activated protein kinase (MAPK)/ERK pathways. In the MAPK cascade, MET activation stimulates SHC and GRB2 recruitment to Gab-1, leading to MAPK activation [12], along with recruitment of SHP2 to Gab-1 that can link MET signaling to the MAPK cascade and extend the duration of MAPK phosphorylation [13]. The ERK-MAPK pathway is responsible for cell proliferation, cell-cycle progression and cell motility. The p85 subunit of PI3K can bind to MET directly and signal through the AKT pathway, stimulating cell survival [14]. Additional pathways responsible for cell migration and invasion response to MET signaling including RAS, CRK, and focal adhesion kinase (FAK) signaling [15]. STAT3 activation and nuclear translocation following MET binding is associated with tissue invasion [16]. MET ubiquitination by E3-ubiquitin ligase c-Cbl leads to MET degradation and is critical for controlled regulation of MET activity [17].
Interactive crosstalk between MET and other cell membrane proteins has been heavily investigated due to the development of drug resistance in targeted therapies. For example, the adhesion protein CD44 variant (CD44v6) is thought to promote downstream activation of the Ras pathway by complex formation between GRB2 and ezrin, radixin and moesin [18]. MET can bind integrin α6β4 in the presence of HGF, resulting in integrin phosphorylation and cell invasion. MET can be transactivated by directly Review www.impactjournals.com/oncotarget binding epidermal growth factor receptor (EGFR) in the absence of HGF and presence of EGF or TGF-α [19]. Interactions between MET and other RTKs, including RON, PDGFR, Axl, HER2, ERBB3 and VEGFR family members, have been observed in a variety of cancers [20,21].

Aberrant MEt activation in cancer
Aberrant HGF/MET axis activation has been implicated in the progression of multiple human tumor types, including liver, lung and gastric carcinomas [22,23], and results in cell survival and migration and tumor development and progression [2][3][4]. HGF/MET signaling pathway activation can occur via MET gene amplification [24], overexpression [25], mutations [26][27][28] or paracrine and autocrine activation of MET by HGF [29], all of which have been observed in multiple human tumor types [22,23].
MET gene amplification resulting in protein overexpression and constitutive activation of the MET receptor has been described in NSCLC, gastric carcinoma and HCC, as well as in preclinical models [24] 'addicted' to the MET signaling pathway. In gastric cancer, MET activation has been attributed to MET gene amplification or overexpression, which reduces apoptosis and promotes tumor cell survival, proliferation, differentiation and migration [34,35].
MET mutations occur only rarely in cancers, but may correlate with tumor development. Constitutively activated MET mutations alter the molecular conformation of the protein structure, either promoting receptor dimerization or modifying catalytic activity [15]. Missense mutations in MET tyrosine kinase domains were recently detected in hereditary papillary renal cell carcinoma (RCC) [26], childhood HCC [27] and other cancers, and these residues were speculated to inhibit MET enzymatic activity. Somatic mutations have been observed in the MET juxtamembrane domain, deleting the exon responsible for E3-ubquitin protein ligase Cbl recruitment and reducing MET degradation [28]. Additional mutations have been identified in the MET sema domain in lung cancer, and are associated with HGF binding and receptor dimerization.

MEt As A PrEDIctIVE cANcEr bIOMArKEr
MET status in patients may serve as a potential biomarker for disease prognosis and a predictor of response to HGF/ MET inhibitors in the clinic. Tables  1, 2 and 3 summarize MET gene and protein expression patterns reported from different platforms in gastric, lung and liver carcinomas, respectively. Different reagents and scoring systems that define clinical MET positivity, and correlations between MET status and patient prognosis or outcome are discussed.

Prevalence of MET gene amplification in cancers
MET gene amplifications that result in protein overexpression and constitutive activation of the MET receptor kinase have been reported in NSCLC, gastric cancers and HCC [24]. Variable MET gene amplification rates were detected depending on the detection method (e.g., fluorescence in situ hybridization [FISH], singlenucleotide polymorphism [SNP] genotyping and quantitative polymerase chain reaction [qPCR]) and the different scoring criteria that define high amplification. In gastric cancer, MET gene amplification prevalence varies from 2 to 23% among studies limited by small sample sizes. In one study, a Southern blot using a [α-32P] dCTP-labeled MET-H probe (Oncor, Inc., Gaithersburg, MD, USA) detected MET amplification in 10% of chemotherapy-naïve primary gastric carcinomas compared with the surrounding normal mucosa [36]. In another study, 21.2% of formalin-fixed, paraffinembedded (FFPE) primary tumor tissues exhibited MET amplification defined as copy number >4 by qPCR [37]. MET amplification was detected by FISH in 2% of patients with gastric cancer; in this case, MET amplification was defined as a MET/CEP7 ratio >2 [38]. Several studies with large cohorts correlated MET gene amplification with advanced tumor stage and poor prognosis in gastric cancer patients undergoing surgery with or without chemotherapy [39][40][41]. By contrast, MET amplification was barely detectable in patients with HCC who underwent resection: 4.5% by SNP genotyping array (Illumina), defined as a copy number ≥3, and 2.3% by FISH (Abbott Molecular), defined as MET/CEP7 ≥2.0 [42,43].
In NSCLC patients with wild-type EGFR, MET amplification rates ranged between 2.4% and 21% depending on the detection approach and study criteria ( Table 2). MET amplification was observed in just 3.9% of TKI-treatment-naïve NSCLC patients as measured by FISH (MET/CEP7 >2), but in 18% of patients when measured by qPCR (MET CN ≥3). MET-FISH positivity or increased MET copy number predicted worse survival [44,45]. In TKI-resistant NSCLC patients, MET is amplified at a higher rate, from 15 to 22%, based on FISH, qPCR and aCGH profiling [46][47][48]. These findings highlight the need for identification of lung cancer patients with MET amplification who will benefit from combination therapy due to primary or acquired resistance, regardless of their EGFR status.

Prevalence of MET mutations in cancers
Although MET mutations happen rarely in cancers, they may correlate with tumor development. Protein structure alterations, either via promotion of receptor dimerization or alteration of catalytic activity, can be attributed to increased kinase activity in MET mutants [15]. Missense mutations were detected in the MET juxtamembrane domain in only 1% of patients with primary gastric cancer using methods such as denaturing HPLC (DHPLC) (Transgenomics) or cold single-strand conformation polymorphism (SSCP; Novex); these mutations may contribute to tumorigenesis [49]. In NSCLC, somatic mutations in the MET juxtamembrane domain result in the deletion of exon 14, which is responsible for recruitment of the E3-ubquitin ligase, Cbl. This leads to the accumulation of abnormally spliced, activated MET unregulated by Cbl-induced degradation. This mutation was associated with elevated MET expression in primary tumors, which was detected in approximately 3% of NSCLC patients [28,50]. Recently, polymorphisms in the juxtamembrane (R988C and T1010I) and sema (N375S) domains were detected in 1.7 and 4% of NSCLC patients, respectively, by PCRbased sequencing; however, no associations between MET mutations and clinical and pathological NSCLC features were observed [51,52]. MET mutation was detected only in the kinase domain in 30% of childhood HCC cases by the SSCP method [27].

Prevalence of MET overexpression in cancers
Immunohistochemistry (IHC), reverse transcriptase PCR (RT-PCR), Western blot and enzyme-linked immunosorbent assay (ELISA) analyses have indicated that MET and HGF levels vary in tumors compared with surrounding normal tissues. MET overexpression was detected in vivo in 9.6 to 71% of human gastric carcinomas based on methodology and tissue type (Table 1). Notably, different antibodies that recognize various MET epitopes and domains have shown different membrane and/or cytoplasmic staining intensities by IHC. For example, 46.1% of primary gastric carcinoma patients exhibited cell membrane and cytoplasmic staining in >5% of tumor cells using a C-28 antibody (Santa Cruz Biotechnology) and the Dako ENVISION system [53]. In another study with the same antibody, 63% of patients showed positive MET expression defined as ≥25% of tumor cells with staining intensities of 2+ or 3+ [54]. MET overexpression in gastric cancer ranged from 9.6 to 23.8%, as defined by IHC staining intensities of 2+ or 3+, via an SP44 rabbit monoclonal antibody from Ventana Medical System. MET IHC 3+ expression was associated with a shorter OS and PFS, and MET gene copy number detected by FISH correlated with MET protein expression detected by IHC in gastric cancer patients [39][40][41]55].
The prevalence of HGF or MET in tissues has also been described in NSCLC tumors with wild-type EGFR. Positive intratumoral HGF expression was identified in 57% of 104 specimens using an anti-HGF antibody (sc-7949, Santa Cruz) according to the IHC Allred scoring system, and expression was associated with poor OS [56]. Using the same antibody in a separate study, HGF overexpression, in which ≥50% of tumor cells exhibited positive staining, was identified in 25% of 88 patients [57]. Western blot analysis also showed high HGF expression in NSCLC compared with normal lung tissue [58,59]. HGF overexpression has been implicated in acquired resistance to EGFR inhibitors in patients with EGFR-mutant NSCLC [47]. Notably, the overall rate of MET overexpression measured via IHC varied widely from 13.7 to 61% in published studies based on cohorts of various sizes with different types of diseases and scoring systems ( Table  2). Still, MET FISH-positive and IHC-positive patients reportedly have significantly shorter survival than METnegative patients [57,60].
Similarly, MET overexpression ranged from 20 to 100% in HCCs compared with surrounding normal hepatic tissue (Table 3). Western blot analysis showed MET overexpression in 52% of 62 HCC patients, which correlated with increased intrahepatic metastases and shorter (5-yr) OS [61]. Biopsy IHC analysis of 86 HCC patients revealed 20% with MET overexpression [62], whereas in a separate study employing RT-PCR, most (20/24) patients overexpressed MET [63]. Multiple studies reported that serum HGF concentrations detected via ELISA after hepatectomy were higher as compared to normal tissues. Furthermore, HGF levels were correlated with tumor size, node cirrhosis, tumor recurrence or metastases in HCC patients [64].
Multiple studies have demonstrated that MET/HGF is associated with a poor prognosis. A study of 194 HCC patients showed that those with high MET expression with strong staining patterns (++) had significantly shorter 5-year survival than those with low expression with negative staining patterns (-) [65]. Thus, MET may represent a promising target for cancer therapies. MET and HGF alterations are indeed associated with clinical outcome, metastasis, invasion and disease severity in human cancers, and identification of patients with specific alterations may be critical to predict clinical response to targeted therapies. However, based on reported MET gene amplification and overexpression prevalence (Tables  1-3), discrepancies clearly result from the use of different detection methods, the number of patients enrolled in a given study, the use of different scoring systems and differences in cancer types. Standardized and optimized methods are needed to identify robust biomarkers that may assist in the selection of MET-positive patients for future clinical trials of MET-targeted therapies.

BIOMARKER VALIDATION IN MET INHIBITOR CLINICAL TRIALS
Currently, there is no consensus on the optimal platforms to explore the relationship between MET status and drug efficacies in clinical trials. FISH and IHC assays, both traditional and commercialized approaches, require advanced technical skills and experienced experts or pathologists to analyze results. A validated biomarker detection strategy should be developed to identify a predictive or prognostic factor for these therapies. Here, we summarize the correlation between biomarker validation or clinical inclusion criteria and the therapeutic efficacy of selective MET/HGF kinase inhibitors or antibodies in clinical trials of multiple solid tumors, including GC, NSCLC and HCC.

MET as a therapeutic anti-cancer target
MET signaling dysregulation in cancer is associated with poor patient outcome. In gastric cancer, MET overexpression or amplification is correlated with tumor stage, metastasis, and shorter overall survival (OS) and progression-free survival (PFS MET/HGF has been regarded as a promising therapeutic target in cancer treatment, whose gene or protein status may be indicative of patient response to MET-targeted drugs. Numerous preclinical and clinical studies have demonstrated that antibody or small-molecule inhibitors targeting MET or HGF are effective anti-cancer therapies [5,22,67]. The intensity score: 0=no appreciable staining in the tumour cells, 1=faint/barely perceptible partial membrane staining, 2=weak to moderate staining of the entire membrane, and 3=strong staining of the entire membrane. The fraction score: 0=less than 5%, 1=from 5% to 25%, 2=from 26% to 50%, 3=from 51% to 75%, and 4=more than 75%. The total score was calculated by multiplying the intensity score and the fraction score, producing a total range of 0-12. For statistical analyses, scores of 0-3 were considered negative, and scores of 4-12 were considered positive   The extent of positive staining was scored as follows: 0, ≤10%; 1, >10-25%; 2, >25-50%; 3, >50-75%; and 4, >75%. The intensity was scored as follows: 0, negative; 1+, weak; 2+, moderate; and 3+, strong. The final score was obtained by multiplying the extent scores and intensity scores, producing a range from 0 to 12. Scores 9-12 were defined as strong staining pattern (++), scores 0-4 were defined as negative expression (−), and scores 6-8 were defined as intermediate staining pattern (+) [

Volitinib
Volitinib is a highly selective small molecule, ATPcompetitive MET kinase inhibitor being investigated as a monotherapy for MET-amplified cancers, such as gastric and lung cancer. Currently, this drug is in clinical development stage, including phase I studies in Australia and China, to test its efficacy against advanced cancers. Other phase I trials seek to test combinations of volitinib and docetaxel in gastric cancer (NCT02252913) and volitinib and gefitinib for EGFR TKI-resistant NSCLC in patients with mutant EGFR (NCT02374645).
In preclinical studies, biomarker analysis has shown that MET-amplified and MET-overexpressing tumor xenograft models are highly responsive to volitinib as a single agent or in combination with other therapies [68][69][70][71]. An in vivo study indicated that volitinib selectively inhibited the growth of MET-driven gastric and lung cancer cell line and primary tumor xenografts [68][69][70][71]. Anti-tumor efficacy in gastric cancer correlated with MET gene amplification/overexpression and high levels of phosphorylated-MET (p-MET). Additionally, the combination of volitinib and docetaxel demonstrated efficacy in a MET-amplified gastric cancer cell line and in primary xenograft models [69]. However, correlations between tumor growth inhibition and MET status were less clear in lung cancer compared with gastric cancer. This may be the result of heterogeneity or variation in p-MET or the activation of compensatory pathways (e.g., EGFR, KRAS) in lung cancer [71]. In particular, the combination of volitinib and taxotere had satisfactory efficacy in tumors less responsive to volitinib monotherapy [71]. The in vivo efficacy of volitinib was tested in a model of EGFR TKI-resistant NSCLC (HCC872C4R) with acquired MET gene amplification. Volitinib with gefitinib produced a synergistic effect compared with volitinib monotherapy, which produced a poor dose response [70]. Therefore, volitinib selectively inhibited tumor growth in a series of human tumor xenograft models with aberrant MET signaling.
MET-FISH was applied to an assessment of MET amplification that involved labeling bacterial artificial chromosome (BAC) DNA with Spectrum Red (ENZO) and CEP7-Spectrum Green probe (Vysis Abbott). MET amplification was defined as a MET/CEP7 ratio ≥2 or cluster signals in >10% of tumor cells. MET-IHC (SP44 Ventana Medical System, Roche) was used to detect MET overexpression, which was defined as a staining intensity of 2+ or 3+ (≥10% of tumor cells with membrane or cytoplasmic staining of moderate or strong intensity). MET-FISH or MET-IHC positivity was considered a predictive marker for response to therapy and determined the patient selection criteria. Volitinib demonstrated robust in vivo anti-tumor effects on predominantly MET-driven gastric and lung cancers in which the MET gene was amplified, while volitinib combined with chemotherapy could produce additional benefits in the treatment of tumors in which MET was a partial driver and where patients exhibited poor responses to volitinib monotherapy [68][69][70][71]. In an early clinical evaluation, a volitinib doseescalation study was performed to evaluate the drug's efficacy against advanced solid tumors, including gastric and lung cancers [72]. In 29 of 35 evaluable patients treated with 100-1000 mg volitinib daily or 300-400 mg twice daily for 21 days, 10% exhibited partial responses (PR), and 59% exhibited stable disease (SD) in at least one post-treatment. Anti-tumor volitinib activities were observed primarily in patients with papillary renal cell carcinoma (PRCC). Furthermore, data suggested that MET gene copy number gain could be a biomarker for therapeutic response. The correlation between MET status and clinical efficacy will be further investigated to improve the patient selection criteria for later-stage clinical trials.

INc280
INC280 (Novartis) is a potent and highly selective MET kinase inhibitor currently being evaluated in earlystage clinical studies. INC280 inhibited cell proliferation, migration and apoptosis in MET-driven tumor cell lines. INC280 suppressed tumor growth in vivo in a dosedependent manner, and was extremely well tolerated, especially in mouse xenograft models of MET-driven glioblastoma and gastric cancer [73,74].
A phase I clinical study reported stable disease in 24% (8/33) of patients with MET-driven advanced solid tumors, including PRCC, NSCLC, HCC, gastric cancer and others (NCT01324479). INC280 showed preliminary anti-tumor efficacy as a single agent in 50% of patients with EGFR wild-type NSCLC with MET dysregulation, as confirmed by FISH (MET/centromere ratio ≥2.0 or MET gene copy number ≥5) or IHC (MET H-score ≥150 or 50% of tumor cells with a staining intensity of 2+ or 3+) [75].
In MET-positive NSCLC tumors with mutant EGFR that were found resistant to EGFR TKIs, a phase Ib/II study of INC280 in combination with gefitinib reported partial responses in 8/46 (17%) of evaluable patients with high MET status (IHC 3+ and/or MET copy number via FISH ≥5) (NCT01610336). Overall response rates were 40% in patients with MET CN ≥5 and 38% in patients with MET IHC staining intensity of 3+ [76]. Unexpectedly, some NSCLCs with high MET expression (IHC 3+) and amplification exhibited tumor progression following treatment. Whether the platform, reagents and scoring system were appropriate for MET status assessment should be further evaluated in a phase II expansion study.
A phase I study of INC280 plus erlotinib in patients with MET-expressing NSCLC (NCT01911507) is ongoing. In HCC, a phase II clinical trial of INC280 as a first-line treatment is currently recruiting patients with tumors that harbor with activated MET pathways (NCT01737827). rilotumumab Rilotumumab (previously known as AMG 102; Amgen) is a fully human monoclonal HGF antibody that preferentially binds to the mature, active form of the protein [77]. Interaction with rilotumumab prevents HGF from binding to MET, which prevents subsequent signaling [78]. The antibody was tested in preclinical models and in clinical trials in multiple solid tumors, either as a monotherapy or in combination with other chemotherapeutics [79][80][81][82][83]. Based on preclinical data, rilotumumab has shown the anti-tumor activity in vitro and in vivo studies, [79,84]. In a phase I clinical trial, rilotumumab monotherapy was deemed safe and well tolerated in 40 patients with refractory advanced solid tumors. A total of 16/23 (70%) evaluable patients achieved stable disease as a best response with PFS from 7.9 to 40 weeks [81].
In a phase II trial (NTC00719550), rilotumumab in combination with the cytotoxic agents epirubicin, cisplatin or capecitabine (ECX) was evaluated in 121 patients with advanced or metastatic gastric or esophageal junction (G/ EGJ) cancer. Rilotumumab [85].
A phase III study to assess the efficacy and safety of rilotumumab in advanced gastric cancer did not confirm the phase II findings and was terminated due to futility results and casualties with an increased number of deaths as compared to chemotherapy [86]. Rilotumumab arm was not superior to placebo arm for OS (median months, 9.6 vs. 11.5; HR = 1.37) and PFS (median months, 5.7 vs. 5.7; HR = 1.30). OS, PFS and ORR were statistically worse in the rilotumumab arm. Significantly more patients in the placebo arm achieved 12-month OS (49.7% vs. 38.4%; P = 0.053) and ORR (39.2% vs. 30%; P = 0.027) compared to the rilotumumab arm. No benefit from treatment with rilotumumab was observed, including patients with higher MET expression, and a higher incidence of fetal adverse events occurred in the rilotumumab arm [86]. We suggest that defining MET positivity as IHC staining in 25% of tumor cells is not stringent enough to establish true MET overexpression, and could lead to negative results in clinical trials. Additionally, rilotumumab was recently reported to be only a partial, not full, HGF antagonist, resulting in HGF-induced MET phosphorylation. This may at least partially explain the poor responses and deaths in rilotumumab clinical trials, and has considerable implications for the use of this therapeutic antibody [87].

Onartuzumab
Onartuzumab (MetMAb; Roche/Genentech) is a single-armed humanized modified 5D5 anti-MET antibody that binds to the MET sema domain and effectively prevents HGF from binding to MET [88]. Preclinical studies have demonstrated inhibitory effects on glioblastoma U87 and pancreatic BxPC3 and KP4 tumor xenografts, including reduced cell proliferation and motility [88,89]. In a phase I study, a complete response was observed in a patient with chemotherapy-refractory metastatic gastric cancer with high MET gene polysomy, high MET expression and evidence of autocrine HGF production [90]. In a randomized, double-blind phase II trial in EGFR-unselected NSCLC patients, no increased benefit in PFS (HR = 1.09, p = 0.69) or OS (HR = 0.80, p = 0.34) was noted in patients receiving onartuzumab plus erlotinib vs. placebo plus erlotinib. However, in METpositive (≥50% of tumor cells with a staining intensity of 2+ or 3+) NSCLC patients (52% of 137 individuals), onartuzumab plus erlotinib was associated with improved PFS (1.5 vs. 2.9 months, HR = 0.53, p = 0.04) and OS (3.8 vs. 12.6 months, HR = 0.37, p = 0.002). In contrast, MET-negative patients demonstrated worse OS after treatment with onartuzumab compared with patients treated with erlotinib alone. Despite improved PFS and OS in the MET-positive population, the ORR in this subset was not different from that of the placebo plus erlotinib arm [91], possibly because the inclusion criteria for METpositive patients may not have been properly defined. A randomized phase III trial in patients with MET-positive advanced NSCLC who were to receive the standard chemotherapy has been terminated because treatment with onartuzumab in combination with erlotinib did not demonstrate a significant difference over erlotinib alone [92].

CONCLUSIONS
Because MET is involved in the regulation of tumor cell survival and metastasis, a better understanding of individual patient sensitivities to MET inhibitors can help guide clinical trial design. Therefore, the key to success for MET-targeted therapies in human cancers may lie in MET-driven patient selection, which is determined by many factors.
First, an integrated platform that incorporates accurate, validated methods and reagent kits, such as antibodies and substrates for the characterization of MET alterations, will help to improve MET-driven population selection. For example, a phase III trial of onartuzumab showed a lack of efficacy in patients with advanced NSCLC, who were selected for the trial based on MET overexpression detected via IHC [91]. However, the IHC detection antibody, SP44, binds the cytoplasmic MET domain, whereas onartuzumab binds to the extracellular semaphorin domain. Reported IHC results vary widely because different antibodies that target distinct MET epitopes are used. It is critical to utilize specific antibodies for the detection of MET expression as a predictive biomarker to refine patient selection in clinical trials.
Next, MET-positive patient selection must also take into consideration of tumor stage and type, along with sample storage conditions, and must incorporate stringent clinical inclusion criteria. A single biomarker or a combination of biomarkers may serve as prognostic factors to fully inform patient selection. For example, in the phase III onartuzumab clinical trial, defining IHC MET positivity as 50% of tumor cells with moderate staining intensity might have resulted in the inclusion of patients with false-positive MET-overexpressing tumors.
In a phase II trial, however, IHC analysis correlated better with significant improvements in OS and PFS compared with FISH analysis. In this study, a MET/CEP7 ratio of two or more or a mean of ≥5 MET copies per cell were defined as MET FISH positivity, which may result in the inclusion of patients with no MET-amplified tumors [93]. These criteria could explain why NSCLC patients with MET FISH positivity exhibited negative responses to onartuzumab in a phase III trial [92]. Additionally, in rilotumumab studies, defining MET-positive tumors as those having at least 25% of tumor cells with membrane staining at any intensity may also have resulted in a high rate of false positives. These results together strongly suggest that testing for MET levels should include both MET overexpression assessment by IHC and MET gene amplification analysis by FISH for improved accuracy and reduced false positive detection rates.
Additionally, MET-driven or MET-positive cancer should be determined by the dominant activation of MET and related oncogenic pathways. MET signaling rarely occurs independently. Other oncogenic pathways, such as EGFR and TGF-β, appear to be involved in tumor progression independent of MET signaling [35]. However, EGFR can also bind with MET and PDGFR, thereby constituting a diversified signal transduction network in cancer cells [94]. In many cancer cases, a combined treatment approach that targets both MET and other functionally redundant RTKs should also be considered.
Thus far, no MET inhibitors or antibodies have been approved for clinical use. Evidence is mounting that fighting MET-driven cancers may depend on many elements, such as the specific drug target, disease stage or type, optimal methods for tumor MET status assessment, and, perhaps most importantly, the identification of robust biomarkers as predictors of patient benefit from METtargeting therapeutics.