Frequency of mesenchymal-epithelial transition factor gene (MET) and the catalytic subunit of phosphoinositide-3-kinase (PIK3CA) copy number elevation and correlation with outcome in patients with early stage breast cancer
The current study was conducted to determine the frequency and association between recurrence-free survival (RFS) and MET and catalytic subunit of phosphoinositide-3-kinase (PIK3CA) copy number elevations in patients with early stage breast cancer.
Tumor DNA was extracted from 971 formalin-fixed, paraffin-embedded early breast cancers for molecular inversion probes arrays. Data were segmented using the single-nucleotide polymorphism (SNP)-FASST2 segmentation algorithm. Copy number gains were called when the copy number of each segment was greater than 2.3 or 1.7, respectively. RFS was estimated by the Kaplan-Meier method. Cox proportional hazards models were fit to determine independent associations between copy number and RFS.
Of the 971 tumors studied, 82 (8.44%) and 134 (13.8%) had an elevation of the MET or PIK3CA copy number, respectively, and 25.6% of tumors with a MET copy number elevation had a PIK3CA copy number elevation. Patients with either a MET or PI3KCA high copy number tended to have poorer prognostic features (larger tumor size, higher tumor grade, and hormone receptor negativity). Both MET and PIK3CA high copy numbers were more likely to occur in patients with triple receptor-negative disease (P = .019 and P < .001, respectively). At a median follow-up of 7.4 years, there were 252 cases of disease recurrence. The 5-year RFS rates were 63.5% and 83.1% for MET high copy number and MET normal/low copy number, respectively (P = .06) and 73.1%, and 82.3% for PIK3CA high copy number and PIK3CA normal/low copy number, respectively (P = .15). A high copy number for either gene was not found to be an independent predictor of RFS.
The hepatocyte growth factor (HGF) and its receptor, the transmembrane tyrosine kinase MET interaction, leads to the formation of phosphorylated MET, which activates downstream effectors such as phosphoinositide-3-kinase (PI3K)/AKT, phospholipase C γ, RAS-mitogen-activated protein kinase (MAPK), c-Src, and signal transducer and activator of transcription (STATs) to promote cell proliferation, survival motility, and invasion as well as morphogenic changes that in normal cells stimulate tissue repair and regeneration but also are coopted during tumor growth.1-8 MET overexpression, with or without gene amplification, has been reported in a variety of human cancers including breast, lung, and gastrointestinal malignancies.9-11 Furthermore, high levels of HGF and/or MET have been correlated with poor prognosis in several tumor types, including breast, ovarian, cervical, gastric, head and neck, and non-small cell lung cancer.10, 12-15
The PI3K pathway plays a key role in cell growth, protein translation, autophagy, metabolism, and cell survival.16-18 Thus, tight regulation of the PI3K pathway is paramount to ensure that cellular inputs are integrated for appropriate cellular outcomes. The PI3K pathway is downstream of most growth factor tyrosine kinase receptors (TKRs) including MET, epidermal growth factor receptor (EGFR), human epidermal growth factor 2 (HER2), and insulin-like growth factor receptor (IGFR) that have been implicated in breast cancer. PI3K signaling is deregulated through a variety of mechanisms, including overexpression or activation of TKRs, activating mutations, and gene amplification of catalytic subunit of phosphoinositide-3-kinase (PIK3CA) and AKT isoforms, as well as loss of the negative regulators phosphatase and tensin homolog (PTEN) and inositol polyphosphate 4-phosphatase type II (INPP4B).19, 20 Other forms of deregulation and aberrations of this pathway have been implicated not only in breast cancer development and progression, but also in resistance to targeted therapies directed toward TKRs and hormone receptors.20-23 As a result, multiple drugs targeting the PI3K pathway are in early clinical trials as monotherapies or combination therapies in patients with breast cancer.18, 24
In patients with breast cancer, reported aberrations in the PI3K pathway have been restricted for the most part to the detection of activating mutations and the loss of suppressors, and data in MET receptor aberrations have been limited in the disease. The purpose of the current study was to determine the frequency and association between recurrence-free survival (RFS) and MET and PIK3CA copy number elevations and their interaction in a large cohort of patients with early stage breast cancer.
MATERIALS AND METHODS
Patients and Tumor Samples
Adequate tumor DNA from formalin-fixed, paraffin-embedded (FFPE) tissue blocks, clinical history, and follow-up data from 1003 patients diagnosed with early breast cancer between 1985 and 1999 were identified from the Early Stage Breast Cancer Repository at The University of Texas MD Anderson Cancer Center. Clinical information (including the patient's age, race/ethnicity, stage of disease, tumor size, lymph node status, nuclear grade, and hormone receptor status) and primary treatment (including surgery, radiotherapy, chemotherapy, and endocrine therapy) was extracted from the medical records.
Molecular Inversion Probes and Copy Number
Tumor DNA was extracted from FFPE tissues and processed for copy number analyses. Briefly, 5 to 10 macrodissected tumor sections (measuring 5 μm each) containing > 80% tumor cells per protocol were pooled and treated 3 times with proteinase K in ATL tissue lysis buffer (Qiagen, Valencia, Calif). After lysis, samples were applied to uncoated Argylla Particles (Argylla Technologies, Tucson, Ariz) and processed according to the manufacturer's recommendations (http://www.argylla.com). For 129 cases, DNA from non–tumor-bearing lymph nodes, stored as FFPE, was isolated as an internal germline reference for the population. Tumor and normal DNA at a concentration of 10 ng/μL was shipped to the Affymetrix MIP laboratory for copy number measurement. The laboratory was blinded to all sample and subject information including the identity of duplicates. Data from the MIP high-density arrays were deposited at the National Center for Biotechnology Information (GSE31424).
Nexus Copy Number v5.1 (BioDiscovery, El Segundo, Calif) was used to process the MIP data from these patient samples. Nexus Copy Number segmented the data using the single-nucleotide polymorphism (SNP)-FASST2 segmentation algorithm, and called copy number gains or losses when the estimated copy number of each segment was greater to/less than 2.3 or 1.7 for MET and PIK3CA, respectively. Each sample has, in general, a different set of segments. Common segments were derived to perform analyses, with a size of 77,487 from the union of all segment breakpoints for 971 samples. To reduce the dimensionality of the data, similar procedures were followed for the comparative genomic hybridization (CGH) regions. We clustered consecutive segments if no 2 segments within the cluster had different gain/loss calls for at least 97% of the samples. This simple criterion yielded 3378 segments with common breakpoints across all 971 samples.
Patient characteristics were tabulated and described by their medians and ranges by copy number (high vs normal/low) with a chi-square test or Wilcoxon rank sum test as appropriate. RFS was measured from the date of diagnosis to the date of first local/distant metastasis or last follow-up. Patients who died before experiencing disease recurrence were considered to be censored at their date of death in the analysis. Survival outcomes were estimated according to the Kaplan-Meier product limit method. Using log-rank statistics, groups were compared between those with a high copy number and those with a normal/low copy number for MET, PI3KCA, and their coamplifications as well as other important clinical variables.
Three multivariate Cox proportional hazard models were developed. The first model incorporated MET and PIK3CA copy numbers and their interaction. The second and third models incorporated either MET copy number or PIK3CA copy number and other prognostic clinical variables. Models were based on a backward selection procedure in which all variables of interest were first included in a full model for screening and only variables with a P < .1 were retained.
P values < .05 were considered to be statistically significant. Analysis was performed using R 2.11.2 statistical software (R Development Core Team, Vienna, Austria).
Table 1 illustrates the patient and tumor characteristics as well as the therapy received by the MET and PIK3CA copy number groups. A total of 82 tumors (8.44%) were found to have an elevated MET copy number and 134 (13.8%) had an elevated PIK3CA copy number; 25.6% of tumors with an elevated MET copy number also had an elevated PIK3CA copy number, and 15.7% of tumors with elevated PIK3CA copy numbers also had elevated MET copy numbers (Fig. 1). Patients with tumors harboring either a MET or PI3KCA high copy number tended to have more aggressive prognostic features, including larger tumor size, higher tumor grade, and negative hormone receptor status. There were no significant differences noted with regard to receipt of adjuvant chemotherapy or radiotherapy. However, the majority of patients with normal/low copy numbers in the either MET or PIK3CA groups received adjuvant endocrine therapy (P = .003 and P < .000, respectively). Elevated MET or PIK3CA copy numbers were more likely to occur in patients with triple receptor-negative disease (P = .019 and P < .001, respectively).
Table 1. Patient and Tumor Characteristics
Abbreviations: HER2, human epidermal growth factor receptor 2, PI3KCA, catalytic subunit of phosphoinositide-3-kinase.
Age at diagnosis, y
Tumor size, cm
Lymph node status
Breast cancer subtype
Triple receptor negative
High copy no.
Normal/low copy no.
High copy no.
Normal/low copy no.
Anthracycline and taxanebased
At a median follow-up of 7.5 years (range, 0 years-21.1 years), there were 252 recurrences reported. Table 2 summarizes the 5-year RFS by MET and PIK3CA copy number and by other patient and tumor characteristics. The 5-year RFS rate was 63.5%, and 83.1% for the MET high copy number and MET normal/low copy number groups, respectively (P = .06), and were 73.1% and 82.3% for the groups with a PIK3CA high copy number and PIK3CA normal/low copy number, respectively (P = .15) (Figs. 2A and 2B). To evaluate the interaction of coordinate gene copy elevations in MET and PIK3CA, patients were classified into 4 groups: normal/low copy numbers for both PIK3CA and MET, high copy numbers for both PIK3CA and MET, MET high copy number, and PIK3CA high copy number. No statistically significant difference in the 5-year RFS estimates was found (P = .137) (Fig. 2C).
Table 2. Five-Year RFS Estimates by Copy Number and Patient and Tumor Characteristics
The Kaplan-Meier survival curves by MET and PIK3CA gene copy number and breast cancer subtype are presented in Figure 3. When examining MET copy numbers, breast cancer patients with hormone receptor-positive status and a high MET copy number were found to have a significantly lower 5-year RFS rate compared with patients with hormone receptor-positive and normal/low MET copy number breast cancer (76.4% vs 85.4%; P = .034). There was a trend toward a worse 5-year RFS rate in patients with HER2-positive and high MET copy number breast cancer compared with patients with HER2-positive and normal/low MET copy number breast cancer (64.3% vs 77.2%; P = .061). No difference was noted in patients with triple receptor-negative disease (P = .80). When examining PIK3CA copy number, there were no differences noted with regard to 5-year RFS estimates by breast cancer subtype. Exploratory survival analysis to evaluate the interaction of a high gene copy number for both MET and PIK3CA by breast cancer subtype demonstrated no statistically significant difference in the 5-year RFS estimates (data not shown).
Table 3 summarizes the MET and PIK3CA copy numbers and their interaction in multivariate models. The results were consistent with univariate analysis of RFS. Overall, patients with tumors harboring a high MET copy number tended to be at higher risk of developing disease recurrence compared with patients with tumors with a normal/low MET copy number (hazard ratio [HR],1.53; 95% confidence interval [95% CI], 0.98-2.38 [P = .06]). A high PIK3CA copy number was not found to be an independent predictor risk for disease recurrence (HR,1.3; 95% CI, 0.91-1.86 [P = .147]), nor was the interaction of both MET and PIK3CA high copy numbers (HR, 0.7; 95% CI, 0.28-1.77)[P = .458]). When examining patients with hormone receptor-positive breast cancer, patients with tumors with a high MET copy number were more likely to develop disease recurrences (HR, 1.86; 95% CI, 1.07-3.25 [P = .029]). In multivariate models including patient and tumor characteristics, MET or PIK3CA high copy numbers were not found to be independent predictors of RFS after adjustment for age, stage of disease, lymph node status, tumor size, and breast cancer subtype (HR, 1.21; 95% CI, 0.8-1.82 [P = .357]; and HR, 1.29; 95% CI, 0.85-1.94 [P = .229]).
Table 3. Multivariate Cox Proportional Hazards Model Including MET andPIK3CA Copy Numbers and Their Interaction
The purpose of the current study was to determine the frequency of MET and PIK3CA copy numbers in patients with breast cancer and their associations with patient outcome. We found that 82 (8.44%) and 134 (13.8%) tumors had an elevated MET or PIK3CA copy number, respectively, and 25.6% of tumors had high copy numbers for both PIK3CA and MET and that high copy numbers of MET or PIK3CA were associated with poorer prognostic features and triple receptor-negative disease. Patients with tumors harboring an elevated MET copy number tended to have worse 5-year RFS (P = .06). A high copy number for either gene was not found to be an independent predictor of RFS.
Deregulation of MET receptor activity occurs in a wide spectrum of human cancers, as a result of germline SNPs, somatic mutations, gene amplification, protein overexpression, and autocrine circuits driven by HGF.24 In patients with breast cancer, reports have been small and were restricted to the assessment of protein overexpression of the MET receptor as well as its ligand HGF in tumor tissues.13-15, 25, 26 The estimated rate of MET protein overexpression has been reported to be 20% to 30%,25, 26 and, as in several other tumor types, the increased expression of MET receptor or its ligand HGF in breast cancer is correlated with increased aggressiveness of disease and an overall poorer prognosis.14, 25, 27 Initial studies assessing immunoreactive (ir)-HGF concentrations in tumor extracts of 258 primary human breast cancers using an enzyme-linked immunoadsorbent assay demonstrated that the ir-HGF level was correlated with large tumor size (P = .05). Patients with a high ir-HGF concentration were found to have significantly shorter RFS (P = .001) and overall survival (P = .001) rates. The ir-HGF level was found to be an independent predictor of RFS (P = .041) and overall survival (P = .036).14 In a smaller cohort of 91 tumors, high MET receptor expression demonstrated by immunofluorescence in patients with positive and negative lymph nodes was correlated with a lower 5-year survival rate (P = .008 and P = .006, respectively).26 A small study of 40 primary breast cancers in which MET and HGF were detected by immunofluorescence and immunohistochemistry indicated that MET levels did not correlate with established poor prognostic factors. However, overexpression of MET was found to correlate with disease progression (P = .037). The authors also demonstrated that when accounting for HER2 status, MET overexpression identified a subset of patients with adverse outcomes independent of HER2 positivity.25 To our knowledge, the current study is the first to report on MET gene copy number, its distribution by tumor subtype, and its correlation with patient outcome. In the current analysis, patients with a high MET copy number had larger tumors that were of higher grade, and negative hormone receptor status. Two interesting findings should be highlighted: the higher percentage of triple receptor-negative disease with this aberration and the prognostic value of copy number in breast cancer, especially in patients with hormone receptor-positive disease. This is important because we continue to require new therapeutic targets that activate signaling in patients with triple receptor-negative breast cancer, and MET receptor inhibition could contribute to cell proliferation, survival, and invasion. Conversely, in our search for mechanisms of therapy resistance to endocrine therapy, MET signaling should be investigated. Finally, we need to define the frequency of MET protein overexpression and its correlation with other aberrations such as gene amplification and activating mutations. Comprehensive work in our institution is currently ongoing to answer this question.
In patients with breast cancer, reported aberrations in the PI3K pathway have been restricted for the most part to the detection of activating mutations and loss of tumor suppressors and the data on PIK3CA gene amplification being limited.21, 28, 29 In a series of 92 primary breast cancers, investigators used quantitative real-time polymerase chain reaction to measure gene copy number and reported that 8.7% (8 of 92 tumors) of the tumors harbored a gain of a PIK3CA gene copy number, suggesting that gene amplification is not the main molecular mechanism in activating the PI3K-driven tumorigenesis pathway in breast cancer.30 In a second cohort of breast cancers, researchers reported that 10 of 161 tumors had PIK3CA gene amplification, and approximately one-half of them also had an activation mutation in the gene, suggesting that an additive effect of point mutation and copy number gain can contribute to oncogenesis.31 In our large cohort of patients with early breast cancer, we also demonstrated that 13.8% of all breast cancers had an elevated PIK3CA copy number. However, when assessing breast cancer subtypes, 28% of triple receptor-negative breast cancers had a high PIK3CA copy number, suggesting that gene amplification in conjunction with loss of PTEN and INPP4B may contribute to PI3K pathway activation in this subtype. However, basal breast cancers have a greater frequency of copy number aberrations and therefore determining whether PIK3CA (or MET) amplification is a tumor “driver” remains to be determined. Previous to our report, in what to our knowledge was the largest tumor set published to date in which PIK3CA amplification was assessed, 292 invasive breast cancers were examined, of which 209 were tested and 28 were found to be amplified (13.4%).32 In this study, the investigators examined other PI3K pathway aberrations and correlated them with breast cancer molecular subtype and outcome. Mutations and copy number gains were almost exclusive events, with only 1 cancer found to encompass both mutation and copy number gain of the PIK3CA gene. Neither mutations nor copy number gains were found to be associated with clinicopathological parameters, breast cancer molecular subtype, or outcome.32
Lastly, we demonstrated that 26% of MET-amplified tumors also have PIK3CA amplification, which is a higher frequency than predicted by chance, thereby indicating that coaberrations in the PI3K and MET pathways occur at a sufficient frequency that could contribute to patient outcomes. Further studies, including a comprehensive analysis of large cohorts of breast cancers (ie, The Cancer Genome Atlas) to determine frequencies of mutations, copy number, and methylation as well as translational changes in PI3K pathway-related genes and MET alone and in combination to determine the frequency of coaberrations in the pathways across multiple modalities, currently are ongoing in our group.
Supported in part by the Kleberg Center for Molecular Markers at The University of Texas MD Anderson Cancer Center, National Cancer Institute (NCI) grant 1K23CA121994-01 (to A.M.G.), NCI Breast Specialized Program for Research Excellence grant P50-CA116199 (to G.B.M. and M.L.B.), NCI through The University of Texas MD Anderson Cancer Center Support grant P30 CA016672, and Susan G. Komen Foundation grants FAS0703849 (to A.M.G. and G.B.M.) and SAC100004 (to A.M.G. and G.R.B.).