Activation of MET pathway predicts poor outcome to cetuximab in patients with recurrent or metastatic head and neck cancer

The incidence of head and neck cancer is increasing worldwide, and has recently become the sixth most common malignancy [1]. Malignancies of the head and neck are associated with tobacco use, alcohol consumption, and Epstein–Barr virus (EBV) and/or papillomavirus (HPV) infections [2]. Advances in diagnosis, prevention, and management of advanced cases have been made in recent years, and while long-term survival rates have improved [1], they remain some of the lowest among major cancer types worldwide. Head and neck squamous cell carcinoma (HNSCC) is a particularly prevalent type of head and neck cancer, constituting 90 % of all head and neck cancers. Survival rates are low due to late diagnosis at advanced stages, the failure of treatment [3], and the development of secondary malignant tumors. These problems underscore the importance of improving strategies of primary chemotherapy and chemoprevention of HNSCC.

Traditionally, the concurrent use of surgery, radiation, and/or multiagent chemotherapy for the management of patients with late-stage, locoregionally advanced unresectable disease has been the standard for treatment of HNSCC [4]. Alternatively, focus has shifted in recent years towards biological therapies [5]. These include drugs that target growth factors and their receptors, signal transduction, cell cycle control, protein degradation, hypoxia, and angiogenesis. Epidermal growth factor receptor (EGFR) is a receptor tyrosine-kinase that is overexpressed in 90 % of HNSCC tumors and is involved in tumor growth, invasion, metastasis, and angiogenesis [6]. It is an early marker of carcinogenesis in HNSCC, and has been associated with a poor outcome [7]. This makes it a reasonable target for specific biological drugs [8], from antibodies to small-molecule inhibitors. Cetuximab is a chimeric IgG1 monoclonal antibody that inhibits ligand binding to the EGFR extracellular domain [9] (hence interfering with receptor activation) and enhances the activity of chemotherapeutic agents [10]. Numerous clinical trials with cetuximab have improved the treatment of recurrent/metastatic HNSCC both as first-line therapy and following failure of platinum-based chemotherapy [11, 12]. Since its approval for HNSCC in 2006, the clinical data produced suggest that cetuximab plays an important role in the locoregional treatment of these pathologies [13]. A second anti-EGFR strategy targets the intracellular domain of the receptor with low-molecular-weight tyrosine kinase inhibitors (erlotinib, gefitinib) and influences downstream signaling processes [14]. Erlotinib has FDA/EMEA approval for locally advanced or metastatic NSCLC as well as advanced or metastatic pancreatic carcinoma, and has also been studied in HNSCC. The response rate to EGFR-targeted therapies is smaller than expected, due to primary resistance [15] and to the development of acquired resistance [16]. In recurrent/metastatic HNSCC, monoclonal antibody therapies in clinical trials have demonstrated superior increases in OS and PFS than tyrosine kinase inhibitors [17, 18]. Therefore, it is important to gain knowledge of the molecular mechanisms of drug resistance since the identification of the tumors that rely on EGFR signaling for their growth is critical for the optimal selection of patients for therapy.

One molecule that has been shown to be involved in resistance to EGFR inhibitors in different tumor types is MET, the receptor tyrosine-kinase for hepatocyte growth factor (HGF) [19]. MET can activate many of the same downstream signaling pathways as EGFR, such as ERK1/2 and PI3K/AKT; additionally, it promotes tumor growth by affecting proliferation, anti-apoptosis, invasion, and angiogenesis in several tumor types [20]. In HNSCC, MET is expressed on epithelial cells and is activated by HGF through a paracrine mechanism. HGF is synthesized by stromal fibroblasts as an inactive precursor, and requires activation to generate responses via MET stimulation in the target cells [21]. Reports of lung, colorectal carcinoma, and glioblastoma have shown that MET is activated in the presence of EGFR blockade as a compensatory pathway, resulting in a mechanism of acquired resistance to EGFR inhibitors [22, 23]. MET and HGF are both consistently overexpressed in HNSCC [24], and such overexpression correlates with an aggressive disease and poor prognosis [25, 26]. Following several reports that established the HGF/MET pathway as an important driving force in HNSCC metastasis, and after some studies correlated its expression with the clinicopathological parameters and the survival of HNSCC patients [27], several clinical trials (http://​clinicaltrials.​gov) were conducted with HGF antagonists (rilotumumab, ficlatuzumab) and MET inhibitors (foretinib, crizotinib) in order to determine whether the inhibition of the HGF/MET pathway may be of therapeutic benefit in HNSCC patients [28].

In addition to HGF and MET overexpression, this pathway can also be activated through genetic alterations such as MET-activating mutations that, although rare in all tumor types, are certainly contributing to carcinogenesis. Two somatic activating MET mutations have been identified in HNSCC (Y1248C, and Y1253D), which increase the kinase activity of MET and subsequently lead to tumor proliferation and metastasis [29]. Additionally, evidence suggests that EBV and HPV infections are risk factors for the development of HNSCC. Viral infection has a prognostic impact on HNSCC, and of these, HPV-positive cancers have a more favorable prognosis [30], whereas the HPV-negative group, overwhelmingly made up of tobacco-related cancers, is the highest-risk group and has the worse prognosis [31]. However, few studies have investigated the association of the HGF/MET pathway expression/activation with HPV status [32].

Owing to the above mentioned, MET has been established as a marker of biological significance in cancer. We have investigated the impact on cetuximab sensitivity of HGF and MET overexpression, MET activation, MET gene status, and MET mutations in recurrent/metastatic HNSCC patients. We show that MET and p-MET overexpression are associated with poor outcome in recurrent/metastatic patients. In addition, we find that phosphorylation of MET is an independent prognostic factor in these patients. Taken together, our results support the idea that HGF/MET pathway might act as a resistance mechanism against EGFR inhibition in advanced HNSCC [33]. Consequently, a dual blocking strategy with anti-HGF/MET and -EGFR therapy may be an effective approach that would eventually benefit HNSCC patients who are resistant to other therapies.

We have retrospectively addressed the correlation of HGF/MET pathway overexpression and activation with cetuximab response in samples from patients that later developed recurrent/metastatic HNSCC, and we have concluded that it correlated with worse outcome in patients treated with a cetuximab-based regimen. The point that samples were collected prior to treatment suggests that it may act as a primary resistance mechanism for EGFR inhibitors. Previous studies have already reported that primary resistance can decrease the response rate to EGFR-targeted therapies [15]. An additional activation of the HGF/MET pathway—that may function as a compensatory route—has been reported in some other tumors [19, 38], but the first description of MET overexpression in HNSCC patients [39] was published only recently. In agreement with previous studies, our work demonstrates that MET was expressed in 58 % of HNSCC patients in our total series, and that p-MET was expressed to a lesser extent, in just 30 % of cases, with no significant differences between the test and control groups. Our series also shows a tight correlation between MET gene amplification and MET overexpression in HNSCC patients (Fig. 2c). These findings reveal a direct link between gene amplification, gene high expression, and protein overabundance. Given that several studies indicate that MET activation is responsible for approximately 20 % of resistance to EGFR inhibitors [29, 40], our results suggest a condition of potential primary resistance.

Biomarker levels were determined in samples collected at the time of initial diagnosis but not at the time of recurrence/metastasis diagnosis, in order to avoid levels alterations due to additional chemotherapy treatments. Although expression levels might fluctuate from diagnosis to recurrence/metastasis, biomarkers at the time of diagnosis are characteristic of every individual, and their profiles possess prognostic value to evaluate the progression of the disease.

Resistance to EGFR-inhibition therapies is a growing concern in HNSCC clinical practice, due to both primary resistance and to the development of acquired resistance by many patients that only respond transiently to therapy with EGFR-targeted drugs [41]. The study of resistance to cetuximab therapy in HNSCC closely echoes the strategies used to uncover the mechanisms of resistance to tyrosine kinase inhibitors (gefitinib, erlotinib) in other tumor types [17, 42]. Setting aside the mutations on the EGFR kinase domain (cetuximab targets the extracellular domain of EGFR, and therefore its mechanism of action does not affect the tyrosine kinase domain), MET amplification represents the most obvious focus of research, in terms of prevalence. EGFR shares important downstream signaling targets with MET, another transmembrane tyrosine-kinase receptor, including ERK1/2, PI3K/AKT, STAT3, and PLCγ. The HGF/MET signaling pathway can be activated by MET genomic amplification [19], by overexpression of the ligand HGF [38] or the MET receptor kinase, or by its activating mutations [29, 43]. In all cases, MET activation occurs by phosphorylation of any of several residues in the tyrosine-kinase domain. A sustained activation of common EGFR and MET downstream targets leads to malignant growth [44].

In vitro studies of MET and EGFR have shown that a single amplified receptor tyrosine-kinase can determine growth and survival of different cancer cell lines (lung, gastric). It was found that the amplified MET was constitutively activated, suggesting an oncogene addiction phenomenon that was required for cell survival [45]. Our present results would now demonstrate that this in vitro requirement may correlate with in vivo growth in patients.

Furthermore, our data suggest that, in case of gene amplification, HGF liberation may not be required for MET activation. Conversely, it may be necessary for those cases with no MET amplification. In fact, HGF overexpression did not correlate with MET amplification (P = 0.786), although it was associated with MET phosphorylation (P = 0.001), suggesting a paracrine activation of the receptor (Table 3). It seems that, in these cases, a greater concentration of HGF and/or proximal interaction of tumor and stromal cells are critical for the activation of the MET pathway (although HGF measurements include the stromal pool, and thus they do not necessarily equate with its active form). As has been demonstrated in murine models, while HGF was secreted by HNSCC tumor-derived fibroblasts, but not by HNSCC cells, MET was expressed and functional in HNSCC cells [39]. Addition of HGF induced MET phosphorylation, leading to the activation of AKT and ERK, and tumor proliferation, confirming that HGF acts mainly as a paracrine factor in HNSCC cells.

Regarding the third possible mechanism of MET activation, two somatic constitutively active MET mutations have been identified in lymph node metastases of HNSCC (Y1248C and Y1253D) [46]. Only one Y1248C mutation (2 %) was found in the MET locus (none in Y1253), probably due to the small probability of finding low-prevalence mutations in our small series. Although these mutations are barely detectable in primary tumors from patients, it has been shown that cells carrying the mutations are selected during the metastatic spread [46]. This is in agreement with our hypothesis that MET pathway activity mechanisms preexist in the HNSCC population, and subsequent treatment with anti-EGFR therapy may expose processes of resistance. Intriguingly, we found four hotspots near these two positions, wherein the proportions of non-canonical nucleotide incorporation were significantly elevated. This point merits further research, as it might indicate that these areas in the MET gene tend to accumulate mutations in the tumoral cells, and it might be related to the mechanisms and consequences of the previously described Y1248 and Y1253 mutations in HNSCC.

From a clinical perspective, the most relevant finding of this study was the prognostic role of MET and p-MET expression in HNSCC. In the literature debate surrounds the possible prognostic role of total MET in human cancer, with many studies suggesting a negative prognostic role [20, 47], while others indicate the contrary [40, 48], and some studies even discern no relationship at all. With respect to MET phosphorylation, our data suggest that it may an independent prognostic factor in these patients. Since this is a surrogate marker of receptor activation, our finding is consistent with an adverse role of activated MET receptor in HNSCC, supporting the findings of previous reports that correlate increased MET activation with resistance to cetuximab in both HNSCC cell lines [19] and patient-derived xenografts [49]. In other tumor types similar conclusions about the role of activated MET have been drawn [34].

Since the first reports of EGFR-targeted therapy in HNSCC, it has been known that EGFR expression is needed for cetuximab response, although EGFR expression does not predict response. Our data suggest that combined treatments of a MET inhibitor and cetuximab may be cumulative, and therefore dual blocking of EGFR and HGF/MET pathways could be a reasonable therapeutic option for clinical practice. Among the various options for inhibiting MET, most efforts concern to the use of small molecule inhibitors of MET, or antibody inhibitors of MET or its ligand, HGF [50].