MET-Driven Resistance to Dual EGFR and BRAF Blockade May Be Overcome by Switching from EGFR to MET Inhibition in BRAF-Mutated Colorectal Cancer

Anticancer targeted therapies are hardly ever curative. Even in the setting of precision medicine, targeted agents fail to achieve prolonged responses in the vast majority of cases. Molecular characterization of tumor samples taken at relapse can inform about the mechanisms underlying acquired drug resistance and drive subsequent lines of therapy, with the ultimate goal of prolonging disease control.

Combinations of anti-EGFR monoclonal antibodies and BRAF inhibitors (BRAFi), with or without other agents such as MEK or PI3K inhibitors, are emerging as an effective strategy to target BRAF ^V600E-mutated metastatic colorectal cancer (1–6). Therefore, BRAF mutations are evolving from a poor prognostic biomarker and a negative predictive factor for single-agent anti-EGFRs or BRAFi (7, 8) to a positive selection marker for BRAFi-based dual or triple combinations. Nevertheless, the long-term efficacy of these targeted combinations is limited by the relatively rapid emergence of drug resistance (1, 2). Scant data are available about the molecular mechanisms underlying acquired resistance to targeted agent combinations in colorectal cancer. A previous study including 3 patients reported that acquired resistance to BRAFi in combination with either EGFR or MEK blockade in colorectal cancer cell lines and/or clinical samples is caused by the emergence of MAPK pathway molecular alterations, including amplification of KRAS, mutated BRAF ^V600E, or mutations in KRAS or MAP2K1 (9). However, the identification of the above-mentioned molecular mechanisms has had no or very limited impact on the clinical management of individual BRAF-mutant patients. In this work, we present how defining the spectrum of resistance mechanisms is critical to developing optimal subsequent treatment strategies aimed at prolonging survival. By genotyping a tissue biopsy, we identified the emergence of MET gene amplification as a mechanism of acquired drug resistance in a patient with BRAF-mutated metastatic colorectal cancer initially treated with the combination of panitumumab and vemurafenib. We describe how the combined data obtained by tumor molecular profiling and functional experiments in cell lines assisted therapeutic decision making and offered the opportunity to deliver precision oncology.

The patient clinical history is summarized in Fig. 1. Approximately 2 years after multimodality treatment for BRAF^V600E-mutated, microsatellite-stable mucinous rectal cancer, a 48-year-old man was diagnosed with pelvic relapse and liver metastases. First-line therapy with FOLFOXIRI obtained a disease stabilization lasting 9 months, indicating that intensive triplet treatment may achieve best outcomes in the subset of patients with poor-prognosis BRAF-mutant metastatic colorectal cancer. Following liver disease progression, the patient was given off-label treatment with panitumumab and vemurafenib according to a previously published phase II study (5). Within days of treatment start, the patient symptoms improved and, after 2 months, a CT scan showed a partial response according to RECIST version 1.1 criteria (Fig. 2). Unfortunately, after 4 months, disease progression occurred (Fig. 2). At this time, a liver tumor biopsy was obtained in order to investigate molecular mechanisms of acquired drug resistance. Molecular analyses were carried out in parallel both in pretreatment primary tumor tissue and in posttreatment liver biopsy. Amplicon-based next-generation sequencing (NGS) of selected exons in 110 genes (Supplementary Table S1) did not identify nucleotide variants previously implicated in resistance to EGFR monoclonal antibodies or BRAF targeted therapies. This analysis ruled out secondary mutations in likely candidate resistance genes including EGFR, KRAS, NRAS, or MAP2K1.

Because amplicon-based NGS approaches do not allow accurate assessment of gene copy-number changes, we performed immunohistochemical and in situ hybridization (ISH) analyses of HER2 and MET, as the activation of these genes had previously been associated with acquired resistance to EGFR monoclonal antibodies in colorectal cancer (10, 11). No changes in HER2 expression or gene copy-number status were observed (Supplementary Table S2). However, a marked difference was seen in MET expression (Fig. 2). Analysis of the archival rectal sample showed heterogeneous MET immunostaining, with a strong membrane signal in about 50% of neoplastic cells (H-score: 150), whereas the remaining cell population showed a weak to moderate membranous staining. Heterogeneity was also observed in MET gene copy number. By fluorescence in situ hybridization (FISH) and silver-enhanced in situ hybridization (SISH) analyses, we estimated that the signals corresponding to the MET and CEP7 (chromosome seven α-centromeric) probes ranged from 1 to 8 (mean values of 3.52 and 2.89 for MET and CEP7, respectively), with a gene-to-chromosome ratio of 1.2, indicating chromosome 7 polysomy. However, in a small fraction of cells within the pretreatment sample, the gene-to-chromosome ratio was ≥6, indicating subclonal MET gene amplification (Fig. 2; Supplementary Table S2). In the postprogression liver biopsy, MET was strongly expressed at the membrane level in about 90% of neoplastic cells (H-score: 270). In this sample, MET copy number varied from 1 to 30 (mean, 8.5), whereas CEP7 ranged from 1 to 16 (mean, 3.24) with an overall gene-to-chromosome ratio of 2.62. At least 75% of cells in the liver biopsy were found to carry MET gene amplification (Fig. 2; Supplementary Table S2; defined by a gene-to-chromosome ratio ≥6), indicating that dual EGFR–BRAF blockade had positively selected for MET-amplified clones.

Because MDM2 gene amplification has frequently been found to co-occur with MET genetic alteration in lung cancer (12), we assessed MDM2 gene copy number by ISH. This analysis suggested that dual BRAF and EGFR treatment had selected for MDM2-coamplified clones (Supplementary Fig. S1A and S1B).

In order to investigate the prevalence of MET amplification in the context of BRAF-mutant colorectal cancer, we performed MET ISH analysis on archival tumor tissue of 17 BRAF^V600E advanced colorectal cancer samples. Three of 17 tumors (18%) from chemonaïve patients displayed MET-amplified subclones (Supplementary Fig. S2A–S2D).

In order to assess whether MET overexpression alone is causally responsible for resistance to combinatorial vemurafenib and panitumumab treatment, we conducted in vitro forward genetic experiments. We took advantage of the BRAF-mutant colorectal cancer cell line WiDr, which we had previously shown to be refractory to either single-agent BRAF or EGFR inhibitors, but sensitive to their combination (13). Ectopic overexpression of MET in WiDr parental cells was able to confer resistance to combined vemurafenib and panitumumab by activating ERK signaling (Fig. 3A) and sustaining proliferation (Fig. 3B). Importantly, addition of the dual ALK–MET inhibitor crizotinib was able to restore sensitivity to combined vemurafenib and panitumumab in MET transduced cells (Fig. 3B).

In vitro modeling of acquired resistance in cancer cell lines has proven effective in identifying resistance mechanisms that occur clinically (14, 15). From a comprehensive effort aimed at defining the molecular landscape of resistance to targeted therapy combinations in BRAF-mutant colorectal cancer cell lines (16), we detected the emergence of MET increased gene copy number and overexpression (Fig. 3C) in a drug-resistant WiDr subline (WiDr res) obtained by prolonged exposure of parental cells with cetuximab in association with the BRAFi encorafenib and the selective PI3Kα inhibitor alpelisib (Fig. 3D), a triplet regimen that is being tested in clinical trials (1). Functional experiments indicated that ectopic expression of MET in WiDr parental cells is able to confer resistance not only to combined BRAF and EGFR blockade, but also to the triplet regimen that includes PI3K inhibition (Fig. 3D). WiDr-resistant derivatives exhibited basal levels of activated RAS-GTP protein (Supplementary Fig. S3) and constitutive phosphorylation of ERK despite combinatorial treatment with encorafenib + cetuximab + alpelisib (Fig. 3E). WiDr-resistant cells displaying MET amplification were clearly refractory to EGFR, BRAF, and MEK targeted therapies, either as single agents or in combination (Fig. 3F). Importantly, crizotinib alone showed no effects on the proliferation of WiDr-resistant cells, whereas its combination with BRAF kinase inhibition by vemurafenib led to a marked decrease of cell viability over prolonged times (Fig. 3F).

Altogether, these results directly underpin a causal relationship between MET amplification and resistance to targeted therapy combinations in BRAF-mutant colorectal cancer and suggest a potential therapeutic strategy to overcome MET-mediated resistance.

The subsequent clinical history of the patient is again summarized in Fig. 1. After 2 cycles of TAS-102, pelvic pain became severe and opioids were insufficient to achieve optimal symptom relief. Disease reassessment with CT and FDG-PET/CT scans confirmed the rapid progression of the disease both loco-regionally (pelvic relapse with penis bulbs involvement and bilateral inguinal lymph nodes) and in the liver (Fig. 4A). Prescription of regorafenib was contraindicated due to urinary fistula. In absence of further treatment options, based on tumor genotyping and preclinical data, and after ethical approval, combined MET and BRAF inhibition was considered as a rational choice. Given the recent phase I data of the crizotinib and vemurafenib combination (17), the patient started this combinatorial regimen in January 2016. One week after treatment start, pelvic pain completely disappeared and the use of opioids was dramatically reduced. After 10-day treatment, a FDG-PET/CT scan showed an early and dramatic metabolic response (Fig. 4B). Serum CA19-9 marker declined by >50% after only 2 weeks, and by almost 90% after 4 weeks (Fig. 1). Treatment is still ongoing without significant toxicity.

In this case study, we found for the first time that MET amplification and protein overexpression were the main molecular drivers of clinical resistance to an anti-EGFR plus BRAFi combination. MET amplification has been previously identified as a mechanism of resistance to targeted agents in other settings, including BRAF-mutated melanoma treated with vemurafenib (10, 18, 19). MET amplification was also found in approximately 5% to 20% of EGFR-mutated non–small cell lung cancers with acquired resistance to EGFR tyrosine kinase inhibitors (20) and in 10% to 20% of RAS/BRAF wild-type colorectal cancer at progression upon anti-EGFR monoclonal antibodies (9). It was previously reported that MET amplification emerges in malignancies with preexisting clones of MET-amplified cells (21), which undergo positive selection during anti–EGFR-based therapy in both patients with non–small cell lung cancer and patients with colorectal cancer (10). The same seems to be true for the patient in our study, as molecular and histologic analyses identified the presence of a subpopulation of MET-amplified cells in the tumor sample before treatment with BRAF targeted therapy. The percentage of cells carrying MET gene copy amplification increased by at least 5-fold in the posttreatment liver biopsy, indicating that dual EGFR–BRAF blockade had positively selected for MET-amplified clones. We argue that the presence of a detectable fraction of preexisting MET-amplified cells has had a negative effect on the duration of response to dual EGFR and BRAF blockade, which was relatively short-lived. It is possible that in this case, combining up-front EGFR, MET, and BRAF inhibition might have led to a more durable response, and we suggest that future clinical studies should further investigate this aspect.

The proportion of BRAF-mutant colorectal tumor samples coharboring MET molecular alterations is unknown, but a large series of 1,115 samples from any tumor types has previously reported that MET amplification was associated with a higher prevalence of BRAF mutations compared with nonamplified tumors (22). In our collection of 17 additional BRAF-mutant colorectal cancer samples from chemonaïve patients, we were able to find MET-amplified subclones in three cases. Although none of the 3 patients actually received BRAF target agents, it is tempting to speculate that subclonal genetic diversity in these tumors could negatively influence the duration of response to BRAF combinatorial therapies due to likely clonal selection of MET amplification. On the other hand, it is possible that MET amplification could also be implicated in de novo resistance to target therapy combinations in BRAF-mutant colorectal cancer. Pretreatment tumor samples from nonresponding patients enrolled in phase I–II studies with BRAFi combination therapies (2–6) will be essential to address this clinically relevant question.

Our preclinical experiments indicated that MET amplification also emerged as a driving mechanism of resistance in BRAF-mutant colorectal cancer cells selected with cetuximab in combination with the BRAFi encorafenib and the PI3K inhibitor alpelisib. Ectopic overexpression of MET was able to confer resistance not only to panitumumab plus vemurafenib but also to the above clinically relevant triplet which includes PI3K inhibition (1). This strongly suggests that MET activation is able to confer resistance to several BRAFi + EGFRi-based combinations, irrespective of the specific drugs used. Pharmacologic data in colorectal cancer cells indicated that MET inhibition alone was ineffective to overcome resistance and that crizotinib should be combined with BRAF blockade to affect the proliferation of MET-overexpressing cells in both short-term and long-term growth assays. This is consistent with what has been reported very recently in BRAFi-resistant melanoma patient-derived xenograft models, in which only combined vertical blockade of the pathway with MET and BRAFi was shown to induce durable regression in mice (23).

Based on preclinical modeling and functional characterization of MET aberrant signaling as a resistance mechanism to BRAF and EGFR dual blockade, the patient was treated with the combination of crizotinib and vemurafenib. The ALK inhibitor crizotinib is well known to also bind to MET tyrosine kinase domain and has shown clinical activity in MET-amplified cancers (24, 25). Moreover, recent phase I data showed that crizotinib may be safely combined with vemurafenib without occurrence of significant dose-limiting toxicities (17). Imaging revealed that switching from EGFR to MET blockade, while maintaining BRAF inhibition, induced an extremely rapid and dramatic response following the initiation of treatment with crizotinib plus vemurafenib.

In this case, molecular analysis of a single-lesion biopsy did not identify any additional MAPK pathway molecular alterations beyond MET amplification that could eventually lead to treatment failure. Yet we admit that our strategy could have missed other resistance alterations occurring at the subclonal level in the same lesion or as main drivers in other disease sites, because tumor heterogeneity may lead to the simultaneous selection of several inter- or intrametastatic clones carrying different mechanisms of secondary resistance (26). However, metabolic response was observed across all disease sites, with ex juvantibus confirmation that MET might have been driving resistance in a relatively dominant and homogeneous fashion in this individual case of BRAF-mutated colorectal cancer. We acknowledge that early assessment with FDG-PET/CT may not be routinely accepted as an endpoint of treatment activity. However, the role of early metabolic response has gained robust evidence since the introduction of targeted therapies. In the era of precision medicine, the dramatic FDG-PET/CT response, necessarily coupled with the clinical and laboratory data, is indicative of immediate treatment benefit—even if longer follow-up is needed to assess the duration of such benefit and potential emergence of further resistance mechanisms.

Extreme caution should be applied not to overinterpret the implications of this work. Indeed, we are well aware of the potential pitfalls of selecting a targeted therapy strategy based on the molecular profile of a single resistant lesion, as exemplified by recent case reports in patients with colorectal cancer and breast cancer (26, 27). Nevertheless, we believe that this case study may highly affect future research. In fact, specific targeting of MET-driven resistance to dual EGFR and BRAF block may lead to the design of biomarker-driven trials of second-line targeted therapy. Triple EGFR, MET, and BRAF coinhibition may also be investigated as additional strategy to prevent or overcome resistance in molecularly selected patients.

Tumor samples were collected in accordance with an Institutional Review Board–approved protocol, to which the patient provided written informed consent, and all studies were conducted in accordance with the Declaration of Helsinki. Targeted NGS on clinical specimens was performed in the Department of Diagnostic Pathology and Laboratory Medicine at Fondazione IRCCS Istituto Nazionale dei Tumori, Milan, Italy. The patient's insurance company covered the cost of off-label combinatorial therapies (panitumumab + vemurafenib; crizotinib + vemurafenib), to which the patient gave informed consent. CT scans were obtained as part of routine clinical care, whereas FDG-PET/CT scans were performed as part of an ancillary study protocol for patients receiving targeted treatments. The clinical course of the patient is summarized in Fig. 1. 

NGS analysis was performed as detailed in the Supplementary Methods and in Supplementary Table S2.

IHC was performed on 3-μm formalin-fixed paraffin-embedded (FFPE) tissue sections or on WiDr cytoclots. MET protein expression was detected by using a rabbit monoclonal anti-MET antibody (dilution 1:200; clone SP44; Spring Bioscience), directed against the synthetic peptide derived from C-terminus of human MET displaying membranous and/or cytoplasmic epitope. Other information is provided in the Supplementary Methods.

MET gene status was assessed respectively by both SISH and FISH analyses scored blindly by two observers (A. Gloghini and E. Valtorta). Bright field dual-color SISH analysis was performed on 3-μm FFPE tissue sections using the MET DNP Probe along with the Chromosome 7 DIG Probe (Ventana Medical Systems) on a BenchMark Ultra Platform (Ventana Medical Systems) according to the manufacturer's protocol. Dual-color FISH analysis, both on metaphase chromosomes and interphase nuclei, obtained from WiDr cultured cells and on 3-μm FFPE tissue sections, was performed using D7Z1 (7p11.1-q11.1)/c-MET (7q31.2) probes (Cytocell), respectively, labeled with FITC and Texas Red. Procedure information is described in the Supplementary Methods.

For both SISH and FISH analyses, the probe signals were counted in at least 100 nonoverlapping tumor cell nuclei from each case. Two independent molecular pathologists (A. Gloghini and E. Valtorta) scored the slides in a blinded fashion. MET gene amplification was defined as positive when: (i) MET/CEP7 ratio was ≥ 2 or (ii) average number of MET signals per tumor cell nucleus was > 6.

SISH analyses for MET status were also performed on archival tumor tissue of 17 advanced colorectal tumors carrying mutant BRAF^V600E diagnosed at Fondazione IRCCS Istituto Nazionale dei Tumori from 2012 to 2016.

Similarly, HER2 gene amplification was defined as positive when (i) HER2/CEP17 ratio was ≥ 2 or (ii) average number of HER2 signals per tumor cell nucleus was > 6.

Bright field dual-color SISH analysis was performed for MDM2 using the MDM2 DNP Probe along with Chromosome 12 DIG Probe (Ventana Medical Systems) on a BenchMark Ultra Platform (Ventana Medical Systems) according to the manufacturer's protocol.

WiDr parental cells were obtained from Dr. René Bernards (Amsterdam, the Netherlands) in July 2011. The KRAS mutant SW480 cell line was purchased from the ATCC (LGC Promochem) in January 2011. All cells were cultured at 37°C and 5% CO₂ in RPMI-1640 (Invitrogen), supplemented with 10% FBS, 2 mmol/L l-glutamine, and antibiotics (100 U/mL penicillin and 100 mg/mL streptomycin). The genetic identity of parental cell lines and their resistant derivatives was authenticated by short tandem repeat profiling (Cell ID System; Promega) at 10 different loci (D5S818, D13S317, D7S820, D16S539, D21S11, vWA, TH01, TPOX, CSF1PO, and amelogenin) not fewer than 2 months before drug profiling or biochemical experiments. Cell lines tested negative for Mycoplasma contamination with the VenorGeM Classic Kit (Minerva Biolabs). WiDr cells were made resistant by continuous drug exposure to the combination of cetuximab (5 μg/mL), encorafenib (1 μmol/L), and alpelisib (0.5 μmol/L). A detailed protocol of the drug proliferation assays is reported in Supplementary Methods.

Total cellular proteins were extracted by lysing cells in boiling Laemmli buffer [1% SDS, 50 mmol/L Tris-HCl (pH 7.5), 150 mmol/L NaCl]. The following primary antibodies were used (all from Cell Signaling Technology, except where otherwise indicated): anti-MET (clone D1C2; 1:1,000); anti–phospho MET (Tyr1234/1235, Cat. #3126; 1:1,000); anti–phospho p44/42 ERK (Thr202/Tyr204; 1:1,000); anti-p44/42 ERK (1:1,000); anti–phospho AKT (Ser473; 1:1,000); anti-AKT (1:1,000); HSP90 (Santa Cruz Biotechnology; 1:500).

Lentiviral vector stocks were produced by transient transfection of the pCCL-MET wild-type or lenti-control plasmid (a gift of Elisa Vigna, IRCCS, Candiolo, Turin), the packaging plasmid MDLg/pRRE and pRSV.REV, and the vesicular stomatitis virus (VSV) envelope plasmid pMD2.VSV-G (25, 2.5, 6.25, and 9 μg, respectively, for 15-cm dishes) in HEK293T cells in the presence of 1 mmol/L/L sodium butyrate (Sigma) and 10 ng/mL of doxicicline. Cells were transduced in 6-well plates (3 × 10⁵ per well in 2 mL of medium) in the presence of polybrene (8 μg/mL; Sigma).

F. Pietrantonio reports receiving honoraria from the speakers bureaus of Roche and Amgen. A. Bardelli is a consultant/advisory board member for Biocartis, Horizon Discovery, and Trovagene. No potential conflicts of interest were disclosed by the other authors.

Conception and design: F. Pietrantonio, D. Oddo, M. Di Bartolomeo, F. de Braud, F. Di Nicolantonio

Development of methodology: F. Pietrantonio, D. Oddo, A. Busico, F. Perrone

Acquisition of data (provided animals, acquired and managed patients, provided facilities, etc.): F. Pietrantonio, D. Oddo, A. Gloghini, E. Valtorta, R. Berenato, L. Barault, M. Caporale, F. Morano, A.V. Gualeni, A. Alessi, G. Siravegna, M. Di Bartolomeo

Analysis and interpretation of data (e.g., statistical analysis, biostatistics, computational analysis): F. Pietrantonio, D. Oddo, A. Gloghini, L. Barault, A. Alessi, G. Siravegna, M. Di Bartolomeo, A. Bardelli, F. Di Nicolantonio

Writing, review, and/or revision of the manuscript: F. Pietrantonio, D. Oddo, A. Gloghini, E. Valtorta, R. Berenato, L. Barault, M. Caporale, F. Morano, A.V. Gualeni, G. Siravegna, F. Perrone, M. Di Bartolomeo, A. Bardelli, F. de Braud, F. Di Nicolantonio

Administrative, technical, or material support (i.e., reporting or organizing data, constructing databases): F. Pietrantonio, R. Berenato, M. Caporale, F. Morano, A. Alessi

Study supervision: F. Pietrantonio, A. Bardelli, F. Di Nicolantonio

The authors thank Professor Silvia Giordano from the University of Torino for helpful scientific discussion. They also thank Elisa Vigna and Annalisa Lorenzato for providing lentiviral plasmids and Lara Fontani for help with lentiviral particles preparation. Finally, they thank Chiara Volpi for support in ISH analyses, and Ilaria Bossi for data management.

This work was supported by grants AIRC IG n. 17707 (F. Di Nicolantonio); Fondo per la Ricerca Locale (ex 60%), Università di Torino, 2014 (F. Di Nicolantonio). Partial support was also obtained from AIRC 2010 Special Program Molecular Clinical Oncology 5 per mille, Targeting resistances to molecular therapies in metastatic colorectal carcinomas, Project n. 9970 (A. Bardelli); the Fondazione Piemontese per la Ricerca sul Cancro-ONLUS 5 per mille 2010 e 2011 Ministero della Salute (A. Bardelli, F. Di Nicolantonio); and from the European Community's Seventh Framework Programme under grant agreement no. 602901 MErCuRIC (A. Bardelli, F. Di Nicolantonio). L. Barault was the recipient of a postdoctoral fellowship from the Fondazione Umberto Veronesi in 2015 and of “Assegno di Ricerca” from the University of Torino in 2016.