Article Text

other Versions

The role of biological markers of epithelial to mesenchymal transition in oesophageal adenocarcinoma, an immunohistochemical study
  1. M J D Prins1,
  2. J P Ruurda1,
  3. M P Lolkema3,
  4. R Sitarz4,
  5. F J W Kate ten2,
  6. R van Hillegersberg1
  1. 1Department of Surgery, University Medical Center Utrecht, Utrecht, The Netherlands
  2. 2Department of Pathology, University Medical Center Utrecht, Utrecht, The Netherlands
  3. 3Department of Medical Oncology, Erasmus Medical Center, Rotterdam, The Netherlands
  4. 4Department of Surgery, Medical University of Lublin, Lublin, Poland
  1. Correspondence to Richard van Hillegersberg, University Medical Center Utrecht, Department of Surgery, P.O. Box 85500, 3508 GA, Utrecht, The Netherlands; R.vanHillegersberg{at}umcutrecht.nl

Abstract

Background E-cadherin, β-catenin, epidermal growth factor receptor (EGFR), neuronal cadherin (N-cadherin) and Cyclin D1 are involved in epithelial to mesenchymal transition (EMT). However, the prognostic significance of EMT markers in oesophageal adenocarcinoma (OAC) is unknown. Aim of this study was to evaluate the prognostic value of, and the association between different EMT markers in OAC.

Methods Tumour cores of 154 patients with OAC were included in a tissue microarray. Scoring criteria was based on immunohistochemical staining intensity.

Results EMT-associated markers were expressed in OAC: reduced membranous E-cadherin and β-catenin were seen in 11.4% and 51.7%, nuclear β-catenin in 19.1% and EGFR and Cyclin D1 overexpression in 56.5% and 27.4% of tumours. Mesenchymal marker N-cadherin was not expressed in OAC. A positive correlation was seen between membranous β-catenin and E-cadherin expression (R=0.209, p=0.001) and between EGFR and Cyclin D1 (R=0.257, p=0.002). In univariate analysis, EGFR overexpression and membranous β-catenin staining were significantly associated with a poor survival (HR 2.145; 95% CI 1.429 to 3.218, p<0.001 and HR 1.665; 95% CI 1.114 to 2.488; p=0.013). However, Cyclin D1 (HR 1.092; 95% CI 0.702 to 1.698; p=0.697), nuclear β-catenin (HR 1.322; 95% CI 0.799 to 2.189; p=0.277) and E-cadherin (HR 1.012; 95% CI 0.554 to 1.851; p=0.968) were not associated with survival. In multivariate analysis, EGFR overexpression was an independent prognostic factor for poor survival (HR 1.678; 95% CI 1.055 to 2.668; p=0.029) together with T stage (HR 2.759; 95% CI 1.356 to 5.576; p=0.005).

Conclusions This study supports the presence of EMT in OAC. Moreover, EGFR overexpression was independently associated with a poor survival.

  • CANCER
  • IMMUNOHISTOCHEMISTRY
  • GASTROINTESTINAL DISEASE
  • ADHESION
  • GI NEOPLASMS

Statistics from Altmetric.com

Request Permissions

If you wish to reuse any or all of this article please use the link below which will take you to the Copyright Clearance Center’s RightsLink service. You will be able to get a quick price and instant permission to reuse the content in many different ways.

Introduction

Radical surgical resection of the oesophagus offers the best treatment option for patients with oesophageal cancer.1 Addition of neoadjuvant chemo(radio)therapy has further improved survival rates.2 Patients might further benefit from addition of other types of therapy, for example molecular therapy that specifically targets biological markers that have a prominent role in cell cycle control progression, lymph node metastases and tumour progression.

Glycogen adhesion protein E-cadherin, β-catenin, mesenchymal marker neuronal cadherin (N-cadherin), epidermal growth factor receptor (EGFR) and cell cycle regulator Cyclin D1 are involved in epithelial to mesenchymal transition (EMT), a biological process in which epithelial cells lose their intercellular and intracellular adhesions and change into a mesenchymal phenotype.3 ,4 EMT plays a role in physiological processes such as embryogenesis and tissue repair and in tumour dissemination.3 ,5 In cancer, EMT is considered a critical process during invasion, dissemination and non-response to chemotherapy. Downregulation of E-cadherin and to a lesser extent β-catenin are believed to be key players in EMT.3 ,5 Loss of membranous E-cadherin leads to disruption of the cellular adhesions and in increased Cyclin D1 gene promoter activity, further increasing tumour invasiveness.6 Furthermore, the E-cadherin/β-catenin complex and EGFR are both localised in the cellular adherens junctions resulting in cross-links between EGFR and the catenin complex signalling.7 ,8

We hypothesised that EGFR and other EMT-associated markers are expressed in a substantial part of oesophageal adenocarcinomas (OACs).

The aim of the study was to evaluate the prognostic value of, and correlations between E-cadherin, β-catenin, EGFR, N-cadherin and Cyclin D1 expression in a homogenous group of patients with OAC.

Materials and methods

Study population

All patients who underwent oesophagectomy for cancer between August 1988 and November 2009 at the University Medical Center Utrecht were collected in a database. Patients with histologically proven OAC were included. Patients with neoadjuvant treatment were excluded. Positive resection margins (R1) according to the College of American Pathologists criteria and with pathological T4 disease were also excluded from the study. These factors were considered of strong prognostic value, irrespective of tumour marker expression, and therefore were excluded from further analysis.

All tumour resection specimens were reviewed by a pathologist (FJWKt). Tumours were staged according to the TNM staging system (7th edition).9 Follow-up data were collected with the use of chart examination, general practitioners archives and the Dutch Cancer Registry. Patients who died from postoperative complications, defined as death within 30 days of oesophagectomy or death during hospitalisation, were excluded from analysis.

Primary outcome was the percentage of molecular marker expression. Secondary outcomes were the correlation between tumour marker expression and overall survival (OS; defined as the time between surgery and death) and cancer-specific survival (CSS; defined as the time between surgery and cancer related death). The study was carried out in accordance with the ethical guidelines of our institution concerning informed consent about the use of patient's materials after surgical procedures.10

Tissue microarray construction

Formalin-fixed, paraffin-embedded tumour blocks were used for construction of a tissue microarray (TMA) as described previously.11

Immunohistochemistry

From the TMA, 4 μm sectioned slides were deparaffinised in xylene (15 min) and dehydrated in serial ethanol dilutions (15 min). Between all steps, tumour slides were rinsed with phosphate-buffered saline pH 7.4. The endogenous peroxide activity was blocked by hydrogen peroxidase (0.3%) methanol solution for 20 min. Antigen retrieval was achieved by protein kinase (2 drops) for 5 min. To prevent background staining, slides were blocked using protein block (2 drops, Novolink) for 5 min. Subsequently, the slides were incubated with primary EGFR antibody (EGFR Zymed, clone 31G7, 1:30) overnight (4°C). Incubation with the secondary antibody (Post Primary block, Novocastra) for 30 min was followed by incubation of Novolink polymer (Novocastra) for 30 min Slides were developed with 3,3′-diaminobenzidine (DAB) (Novolink chromogen) for 5 min, followed by haematoxylin counterstaining. Peroxidase activity was visualised by incubating the slides for 5 min with 3,3′-diaminobenzide and the sections were counterstained with haematoxylin (10 s).

E-cadherin, β-catenin, N-cadherin and Cyclin D1 staining was performed using an automated staining machine (Bond System, Leica Microsystems GmbH, Wetzlar, Germany). For E-cadherin staining automated steps include antigen retrieval (EDTA solution for 20 min) and incubation with primary antibody directed against E-cadherin (Novocastra, clone 36B5 138532 (1:40)). For Cyclin D1 antigen retrieval was achieved using EDTA solution for 20 min followed by incubation with the primary antibody against Cyclin D1 (Neomarkers SP4, 9104S111OH (1:10)).

For β-catenin staining a mouse monoclonal anti-β-catenin (Novocastra, clone 17C2 6016184 (1:20)) was used and for N-cadherin a mouse monoclonal anti-N-cadherin (Sigma, clone GC4 082M4785V (1:80)) was used. In both stainings, antibody retrieval method was performed with EDTA.

Immunohistochemical scoring

Scoring was performed by FJWKt. The scoring of EGFR expression ranged from 0 (negative), 1+ (weak), 2+ (moderate) and 3+ (strong membranous staining). Tumour cores staining 0 or 1+ were considered as low and cores with a score of 2+ or 3+ as high expression.

Scoring of E-cadherin expression was based on intensity of membranous staining that ranged from 3+ (strong membranous), 2+ (moderate), 1+ (weak) and negative (no membranous E-cadherin expression).12 Tumour cores that showed ≤2+ expression were considered as reduced E-cadherin membranous expression. Scoring of β-catenin expression was based on intensity of membranous staining that ranged from 3+ (strong membranous), 2+ (moderate), 1+ (weak) and negative (no membranous β-catenin). Cores were considered as normal β-catenin expression (ie, normal) when ≥2+ staining was seen.

N-catenin expression was not seen in the OAC TMA. The positive control, a cardiomyocyte, and perivascular structures present in the TMA, stained positive for N-cadherin.

Scoring of Cyclin D1 was based on the proportion of positive staining tumour cells. This scoring system included 1+ (<10%), 2+ (10–30%) and 3+ staining (≥30% of tumour cells stained positive). Tumours were considered as positive with a score of 2+ and 3+.

Statistical analysis

Association between clinical parameters such as gender and age and pathological parameters were evaluated using cross tabulation (Pearson's χ2 test). The Kaplan-Meier function (logrank test) was used to compare the OS among patients with low versus high EGFR, E-cadherin, β-catenin and Cyclin D1 expression. The Pearson's correlation test was used to evaluate the correlations between the different biological markers using continuous variables.

The following parameters were evaluated in univariate analysis: T-stage (T1 or T2 vs T3), lymph node metastases (yes vs no), differentiation grade (good and moderate vs poor), EGFR (high vs low), Cyclin D1 (high vs low), E-cadherin (normal vs downregulation), membranous β-catenin (normal vs downregulation), nuclear β-catenin staining (yes vs no), lymph node ratio (ie, the number of positive nodes divided by the total number of resected nodes; ≤25% vs >25%), vasoinvasion (yes vs no), perineural growth (yes vs no), median age (<64 years vs ≥64 years), gender and perinodal extension (yes vs no). Variables that were significant in univariate analysis were included in multivariate analysis (Cox proportional hazards regression analysis) in which the enter method was applied. A p value of <0.05 was considered statistically significant. All analyses were performed using standard statistical software (SPSS V.20.0; SPSS, Chicago, Illinois, USA).

Results

Between 1988 and 2009, 290 patients underwent oesophageal resection for OAC. Patients with neoadjuvant treatment (n=49), with tumour-positive resection margins (R1; n=26), with T4 disease (n=7), or when clinical and/or pathological data was incomplete or missing (n=54) were excluded from further analyses. From this cohort, 154 patients met the inclusion criteria (table 1).

Table 1

Baseline characteristics (n=154)

Median follow-up was 27.7 (range 2.7–286.4) months. Median follow-up of the survivors was 132.6 (range 46.1–286.4) months. Recurrence of disease was reported in 88 (61.5%) patients and 5-year OS was 38%. Five-year CSS was 41%.

β-catenin

Tumour cores of 143 (92.9%) patients were available for evaluation of β-catenin staining (figure 3A, B). Downregulation of membranous β-catenin was seen in 51.7% (figure 3B) and nuclear β-catenin expression in 19.1% of tumours (figure 3B). Normal membranous β-catenin expression was more frequently seen in tumours with a good to moderate grade of tumour differentiation (p<0.001) and in early tumours (T1/T2) (p=0.015), and was inversely associated with vasoinvasion (p=0.005), lymph node metastasis (p=0.005) and recurrence of disease (p=0.007). Nuclear β-catenin expression was more frequently seen in patients with early disease (T1/T2) (p=0.10) and in patients without perineural tumour growth (p=0.007) and without lymph node metastasis (p=0.048) (see online supplementary table S1).

Furthermore, downregulation of membranous β-catenin was associated with a poor survival (HR 1.665; 95% CI 1.114 to 2.488; p=0.013 and CSS HR 1.847; 95% CI 1.181 to 2.888; p=0.007) but not nuclear β-catenin staining (HR 1.322 95% CI 0.799 to 2.189; p=0.277 and CSS HR 1.216; 95% CI 0.704 to 2.100; p=0.483). Median survival was 41.3 months in patients with normal and 22.9 months in patients with reduced β-catenin expression (p=0.012, log rank, figure 1).

Figure 1

Downregulation of membranous β-catenin was associated with poor overall survival. Median survival in patients with low expression was 22.9 months and in patients with normal expression, it was 41.3 months.

E-cadherin expression

Tumour cores of 149 (96.8%) patients were available for E-cadherin evaluation (figure 3C, D). E-cadherin downregulation was seen in 17 (11.4%) patients (figure 3D). Reduced expression of E-cadherin expression was significantly associated with T-stage (p=0.003, χ2, online supplementary table S1) and was not associated with survival (HR 1.012; 95% CI 0.554 to 1.851; p=0.968 and CSS HR 1.168; 95% CI 0.563 to 2.422; p=0.676).

EGFR expression

Tumour cores of 147 (95.5%) patients were available for immunohistochemical EGFR evaluation. High EGFR expression was seen in 83 (56.5%) patients (figure 3E). High EGFR expression was more frequently seen in tumours with a poor grade of tumour differentiation (p=0.001, χ2, online supplementary table S1), and was positively associated with T-stage (p=0.001), perineural tumour invasion (p=0.042), lymph node metastases (p=0.014) and with recurrence of disease (p=0.006). High EGFR was significantly associated with decreased survival (HR 2.145; 95% CI 1.429 to 3.218; p<0.001, table 2 and with CSS HR 2.166; 95% CI 1.384 to 3.390; p=0.001). Median survival was 59.5 months in patients with low and 20.8 months in patients with high EGFR expression (p<0.001, log rank, figure 2).

Table 2

Univariate and multivariate analysis of associations between clinical and pathological features and overall survival (OS)

Figure 2

Epidermal growth factor receptor (EGFR) overexpression was significantly associated with poor overall survival. Median survival was 59.5 months in patients with low EGFR expression and 20.8 months in patients with high EGFR expression (p<0.001, log rank).

Figure 3

(A) Membranous β-catenin 3+ expression without nuclear accumulation of β-catenin. (B) Negative membranous expression with nuclear accumulation of β-catenin. (C) Normal E-cadherin expression (3+). (D) Absence of E-cadherin. (E) Epidermal growth factor receptor (EGFR) overexpression (3+). (F) Negative EGFR expression. (G) Positive control (cardiomyocyte) for N-cadherin. (H) Negative N-cadherin staining whole tumour slide, taking into account the invasive border of the tumour.

Cyclin D1 expression

In 146 (94.8%) patients, Cyclin D1 was available for immunohistochemical evaluation. Predominantly, a nuclear Cyclin D1 expression pattern was seen. High Cyclin D1 expression was seen in 40 (27.4%) patients (data not shown). Cyclin D1 expression was neither associated with clinical and pathological parameters nor with survival (HR 1.092; 95% CI 0.702 to 1.698; p=0.697 and CSS HR 1.54; 95% CI 0.719 to 1.854; p=0.552).

N-cadherin

N-cadherin staining was not seen in patients with OAC. In addition, also in selected whole tumour slides of patients that developed recurrence of disease (n=3) with regard to the invading edge the tumour, versus patients (n=3) that did not develop recurrence of disease, expression of N-cadherin was not seen (figure 3H). Perineural structures, however stained positive for N-cadherin (data not shown).

Correlations

EGFR expression was significantly associated with Cyclin D1 expression (R=0.257, p=0.002, Pearson's correlation). E-cadherin expression was significantly associated with membranous β-catenin (R=0.209; p=0.001), but not with nuclear β-catenin (R=0.138; p=0.107). Furthermore, E-cadherin downregulation was neither associated with Cyclin D1 (R=0.038; p=0.656) nor with high EGFR expression (R=0.084; p=0.319).

Multivariate analyses

In multivariate analysis using the enter method, including clinical and pathological parameters (p<0.10), high EGFR expression was an independent prognostic factor for a poor survival (HR 1.678; 95% CI 1.055 to 2.668; p=0.029) together with T-stage (HR 2.759; 95% CI 1.356 to 5.576; p=0.005). Regarding CSS EGFR overexpression was not independently associated with poor survival (HR 1.468; 95% CI 0.888 to 2.426; p=0.134).

Discussion

EMT is increasingly recognised as a key mechanism in tumour dissemination. Epithelial cancer cells that undergo an EMT, lose their intercellular and intracellular adhesions and change into a mesenchymal phenotype. Identification of cells that underwent an EMT may provide a target to reduce chemo(radio)therapy resistance and recurrence of disease.

Results of this study showed β-catenin downregulation in 51.7%, nuclear accumulation of β-catenin in 19.1%, reduction in E-cadherin in 11.4%, and EGFR and Cyclin D1 overexpression in 56.5% and 27.4%, respectively, of tumours. N-cadherin was not expressed in OAC. Moreover, downregulation of membranous β-catenin and EGFR overexpression were associated with poor survival.

E-Cadherin/β-catenin complex play a major role in epithelial cell-to-cell adhesion providing a link, through α-catenin, between the actin cytoskeleton.13 The interaction between β-catenin and E-cadherin is in agreement with the present study, which showed a positive correlation between expression of E-cadherin and β-catenin.

Downregulation of E-cadherin and to a lesser extent β-catenin are important hallmarks of EMT.3 ,5 The relationship between loss of E-cadherin in cancer cells and passage through EMT is well established.8 ,14 Preclinical experiments showed that silencing of E-cadherin resulted in a morphological shift from an epithelial to a fibroblast phenotype (ie, EMT), as well as a concomitant increase in tumour invasiveness.15 Furthermore, there is increasing evidence that binding of growth factors such as TGF-β and epidermal growth factor (EGF) to receptor tyrosine kinases such as EGFR, induces EMT and that constitutive activation of EGFR stabilises the mesenchymal phenotype.3 ,5 ,8 ,16 Moreover, cancer cells that lose E-cadherin expression are more sensitive for the actions of EGF and TGF-β.17 Loss of E-cadherin is often accompanied with loss of membranous β-catenin and translocation of β-catenin to the nucleus.18 ,19 In the nucleus, β-catenin integrates with the T cell factor/ lymphoid enhancer factor-1 complex and contributes to passage through an EMT by inducing gene transcription of, for example, Cyclin D1.18 ,20

Therefore, results of the present study indicate that a subgroup of cells might have gone through the EMT programme, which is depicted by expression of EMT-associated markers.

The present study has some limitations that should be mentioned. First, a TMA was used to evaluate EMT-associated markers, instead of whole tumour resection specimen. It is believed that when EMT occurs in cancer, it only involves a subset of cancer cells that are mostly located at the invasive front of the primary tumour or tumour-microenvironment interface.21 Although, in some cases the TMA included the invasive border of the tumour, percentage of EMT-associated markers could be higher when whole resection specimens were evaluated. Nevertheless, using the TMA, EGFR overexpression, reduction of membranous β-catenin and E-cadherin, and accumulation of nuclear β-catenin were fairly frequent. Especially the latter is not a frequent finding in all cancer types,3 indicating that accumulation of β-catenin in OAC may have significant importance.

Second, EMT is a dynamic process. Within the tumour, cells may pass through EMT morphogenesis to different phases, with some cells mainly exhibiting an epithelial phenotype whereas others fully transform into mesenchymal cells.5 In addition, EMT is a reversible process, which means that mesenchymal cells can retransform into epithelial derivatives, a process called mesenchymal to epithelial transition.3 ,5 These characteristics of EMT challenge examination of EMT-associated markers in formaldehyde fixed paraffin-embedded tissue.

Third, we did not find a (direct) correlation between reduced expression of E-cadherin and β-catenin, and accumulation of nuclear β-catenin. This indicates that other mechanisms such as activation of the Wnt signalling pathway, might have an additional role in inducing translocation of β-catenin to the nucleus. It is known that activation of the Wnt signalling pathway induces EMT as well.3

And fourth, mesenchymal marker N-cadherin was neither expressed in the TMA nor in selected whole resection specimen. N-cadherin is frequently expressed in the nervous system, smooth muscle cells, fibroblasts and endothelial cells.22 In addition, it is established that when cancer cells switch from an epithelial to a mesenchymal phenotype, this transition is marked by a downregulation of E-cadherin and an increase of N-cadherin, called ‘the cadherin switch’.23 The absence of N-cadherin expression in OAC is in contrast with other studies who reported a significant relationship between N-cadherin expression and tumour invasiveness24 and poor prognosis.25 From this study however, it might be concluded that in OAC N-cadherin has a less prominent role in tumour invasiveness.

In the current report, EGFR overexpression was associated with aggressive tumour characteristics. Similar to our study, previous studies showed correlations between EGFR overexpression and T-stage,26 ,27 lymph node metastases,26 ,27 grade of tumour differentiation28 and recurrence of disease,29 confirming the prominent role of EGFR overexpression in oesophageal cancer.

The independent prognostic significance of EGFR overexpression in oesophageal carcinoma has only been reported in a small series of OACs before.30 In the present study, the independent prognostic significance of EGFR was confirmed in a large homogeneous population with adenocarcinoma.

Identification of EMT-associated markers such as reduced expression of E-cadherin and β-catenin, and EGFR overexpression may provide (novel) targeted therapies that can be used to improve patients’ prognoses.

EGFR inhibition is at the forefront of development of molecular targeted therapy. Numerous phase 1–2 studies have been performed with EGFR inhibitors, with response rates varying from 9–72%.31–42 However, results of a recent phase 2/3 study, including 258 patients with non-metastatic oesophageal carcinoma, showed that addition of cetuximab therapy, that is, a monoclonal EGFR antagonist, to definitive chemo(radio)therapy (cisplatin and capecitabine and concurrent radiotherapy) led to a reduced OS compared with patients that received standard therapy (22.1 months vs 25.4 months; HR 1.53; 90% CI 15.1 to 25.5; p=0.035).43 Additionally, a recent study including 12 patients with resectable OAC showed that addition of cetuximab with concurrent radiation to perioperative epirubicin, cisplatin and capecitabine was accompanied with severe adverse events postoperatively without a complete pathological response.44 Therefore, it is not recommended to add cetuximab to chemoradiation therapy.

Strategies to restore expression of β-catenin and E-cadherin are in its infancy, but seem promising. Preclinical trials with migfilin, a compound that promotes degradation of cytosolic β-catenin, showed that transfection resulted in decreased cell motility and tumour invasion in oesophageal squamous cell carcinoma cell lines.45 Moreover, there is evidence that a combination of MS-275, that is, histone deacetylase inhibitor, and gefitinib, an EGFR inhibitor, increased sensitivity to EGFR inhibition and restored E-cadherin levels.46

In summary, markers associated with EMT such as reduced expression of E-cadherin and β-catenin, accumulation of nuclear β-catenin and EGFR overexpression were seen in OAC. β-catenin and EGFR were significantly associated with a poor prognosis, and EGFR was independently associated with a poor survival. The present results support therapeutic targeting of cancer cells that show EMT features by using EGFR inhibitors and possibly also agents that reactivate β-catenin and E-cadherin expression.

Take home messages

  • Epithelial to mesenchymal transition-associated markers could be identified in oesophageal adenocarcinoma using immunohistochemistry.

  • A positive correlation was seen between expression of epidermal growth factor receptor (EGFR) and Cyclin D1 and between E-cadherin and β-catenin.

  • EGFR overexpression was seen in 57% of patients and was independently associated with a poor survival.

Acknowledgments

The authors thank Petra van de Groep from the Department of Pathology of the University Medical Center Utrecht for the EGFR antibody.

References

Footnotes

  • Correction notice This article has been corrected since it was published Online First. The provenance and peer review statement has been amended.

  • Handling editor Runjan Chetty

  • Contributors MJDP wrote the manuscript. JPR, MPL, RS, FJWKt, RH revised the manuscript.

  • Competing interests None.

  • Provenance and peer review Not commissioned; externally peer reviewed.