Tel.: +46-(0) 90-7858584, Fax +46(0)90-7852031
Cancer Cell Biology
Increased epidermal growth factor receptor expression at the invasive margin is a negative prognostic factor in colorectal cancer
Article first published online: 15 JUL 2010
Copyright © 2010 UICC
International Journal of Cancer
Volume 128, Issue 9, pages 2031–2037, 1 May 2011
How to Cite
Ljuslinder, I., Melin, B., Henriksson, M. L., Öberg, Å. and Palmqvist, R. (2011), Increased epidermal growth factor receptor expression at the invasive margin is a negative prognostic factor in colorectal cancer. Int. J. Cancer, 128: 2031–2037. doi: 10.1002/ijc.25559
- Issue published online: 8 MAR 2011
- Article first published online: 15 JUL 2010
- Accepted manuscript online: 15 JUL 2010 12:00AM EST
- Manuscript Accepted: 5 JUL 2010
- Manuscript Received: 7 JUN 2010
- The Swedish Cancer Society
- The Swedish Research Council and the Northern Sweden Cancer Research Foundation
- invasive margin;
- colorectal cancer;
The receptor tyrosine kinase epidermal growth factor receptor (EGFR) is often expressed in solid malignant tumours, and the expression has been correlated to disease progression. Multiple new agents targeted against the EGFR have been developed during the last decade, but treatment selecting criteria are still not clear. This immunohistochemical study includes 386 colorectal cancer patients and focuses on EGFR expression variations within the tumour, comparing central parts to the invasive margin. Positive immunostaining for EGFR was evident in the central part in 176/386 (46%) of analyzed primary tumours. The invasive margin was positive in 222/386 (58%). A similar expression in both the central part and the invasive front was evident in 286/386 (74%). An increased score at the invasive margin compared to central parts (EGFRi) was evident in 97/386 (25%) of the tumours. Moreover, the results show a significant survival disadvantage for the EGFRi group, both in potentially curatively resected colon cancer patients (n = 170, p = 0.01) and in potentially curatively resected colon and rectal cancer patients combined (n = 273, p = 0.013). Multivariate survival analysis adjusted for age, gender, bowel localisation, grade, stage and tumour type showed an increased risk of cancer death for EGFRi tumours (HR, 1.53; 95% CI, 1.04–2.23; p = 0.029). A significant correlation between EGFR expression at the invasive margin and the presence of budding was seen (p = 0.0001). This investigation of a large patient material implies that EGFR immunohistochemical analysis still has a role in risk evaluation of colorectal cancer patients.
Colorectal cancer is the second most common cause of cancer death in the western world. The prognosis is correlated to tumour stage International Union Against Cancer (UICC)/Tumour Node Metastasis (TNM) at time of primary tumour surgery. The five-year survival rate of Stage II patients ranges from 60 to 80% compared to 30 to 60% in Stage III, and there is indeed a need of additional prognostic markers besides tumour stage.1
The invasive tumour front is an area of specific prognostic interest in colorectal cancer. Already in 1987, an invasive growth pattern of the tumour front was shown to be a negative prognostic factor in comparison to an expanding growth pattern.2 The tumour cells at the invasive front leads the tumour invasion through nonneoplastic stromal tissue, and some of these invading cells have been suggested to be migrating cancer stem cells.3 The development from highly differentiated epithelial mucosal cells to invasive phenotypes includes the process called epithelial-mesenchymal transition (EMT).4 During EMT, the molecular changes leading to loss of cell-cell interactions and cell adhesion results in dissemination.5 The phenomenon of small clusters of malignant cells at the invasive margin, called budding, has been shown to represent a negative prognostic factor6–9 A new staging system has been proposed, where the presence of budding is included in the risk analysis.10 However, the biological mechanisms included in local invasion and budding are still poorly understood.
The last decade has resulted in substantial development in treating colorectal cancer patients with an increasing number of new targeted therapies currently under use and additional ones are likely to be introduced. One of the first targeted treatments in clinical use was Cetuximab, a recombinant murine-murine chimeric antibody (Ab) targeting the epidermal growth factor receptor (EGFR) receptor.11 The EGFR (ERBB1) is a member of the type 1 receptor tyrosine kinase family, comprising of four (ERBB1, ERBB2, ERBB3 and ERBB4) transmembrane receptors with significant importance in several human cancers.12 The receptors initiate intracellular signalling, resulting in proliferation and survival through the Ras/Raf/MEK/ERK or PI3/PTEN/AKT pathways. Activation of these pathways leads to further progression of colon cancer.13 The EGFR has been reported to be expressed in 16–97% of colorectal cancers, and the positivity has been correlated to advanced tumour stage, but the impact on prognosis remains not fully clarified.14–16 Previously, immunohistochemical analysis of EGFR expression was used to select patients suitable for treatment with EGFR-targeted antibodies, but recent studies have failed to detect any correlation between EGFR expression by immunohistochemistry (IHC) and response to Cetuximab treatment.17 Instead, the presence of v-Ki-ras2 Kirsten rat sarcoma viral oncogene homolog (KRAS) mutation has been shown to correlate to nonresponsiveness to Cetuximab treatment, and today, mutation analysis is most often performed before anti-EGFR treatment is initiated.18
In our study, we have investigated EGFR expression with special attention to the invasive margin in 386 colorectal cancers by IHC and related the findings to clinico-pathological features. The results show a worse outcome in patients with a higher EGFR expression at the invasive front compared to central tumour areas.
Material and Methods
We studied tumour specimens from 464 patients diagnosed with colorectal cancer, and the tissue samples were collected during primary tumour surgical resection during the period 1995–2003 at the Department of Surgery, Umeå University Hospital, Sweden. In total, 57 specimens were excluded because of lack of internal positive control (see below), and 21 because of the lack of adequate tissue available (i.e. both central areas and the invasive margin were not represented within the tissue block) and repeated tissue loss during the antigen retrieval step, thus leaving 386 patients in the study. Complete survival data was missing in additionally 25 patients. Out of these 386 patients, 63 patients received preoperative radiotherapy (5 × 5 Gy), 35 received long-term radiotherapy (25 × 2 Gy) and 94 patients received postoperative adjuvant chemotherapy. No patient received anti-EGFR treatment.
All stained sections were first reviewed by an experienced pathologist (R.P.), who performed all histopathological classifications including stage, grade and tumour type (mucinous or nonmucinous). Clinical data were obtained by reviewing the patient records (Å.Ö.), and survival data were collected during spring 2005. The study was approved by the local ethics committee. Table 1 illustrates the clinical characteristics of all the cases.
Colorectal cancer (CRC) specimens were fixed in formalin and paraffin embedded using routine procedures. For our study, 4-μm sections were cut, dewaxed and rehydrated and placed into a semiautomated immunostaining machine (Ventana ES, Ventana Medical Systems, Tucson, AZ). In the subsequent steps, the slides were subjected to protease treatment (Protease 1), primary EGFR antibody incubation (mouse monoclonal E30, DAKO, Denmark, dilution 1:50), secondary antibody incubation, followed by amplification and then visualisation using 3,3-diaminobenzidine (DAB) as a chromogen. Hematoxyline counterstaining was used together including a bluing step. All reagents in the staining procedure besides the primary antibody were from Ventana Medical Systems. In each staining run, an external positive control section consisting of normal skin was included. Because internal positive control perineurium was used19 and sections lacking positive staining in perineural structures were excluded from further analyses.
Analysis of the EGFR
EGFR was stained mainly at the cell membrane. Immunohistochemical staining was evaluated by one of the authors (I.L.) together with an experienced pathologist (R.P.), and then repeated by I.L., and discordant cases were reviewed in a joint evaluation. An internal positive control was mandatory for inclusion as described above. The Hercept scale was used to interpret the staining in tumour cells14, 20: Score 0: no staining or staining in <10%, Score 1+: weak incomplete staining of >10%, Score 2+: moderate and complete staining of >10% and score 3+: strong and complete staining of >10%. The EGFR variable was further reclassified into EGFRi (EGFRi, increased; tumours scoring higher at the invasive margin compared to central parts) vs. non-increased EGFRn (EGFRn, nonincreased; tumours with equal or lower score at the invasive margin compared to central parts).
The invasive margin was defined as the five most distant cell layers apart from the central parts of the tumour. The presence of small cancer cell clusters (less than five cells) at the invasive margin was scored as budding9 (Fig. 1).
Microsatellite instability screening status
Immunohistochemical analyses were performed using standard procedures, including antigen retrieval treatment in EDTA (pH 8.0) and the use of a semiautomatic staining machine (Ventana ES; Ventana, Tucson, AZ). Primary monoclonal MLH1 (clone G168-15, dilution 1:50; BD Biosciences Pharmingen, Belgium), MSH2 (clone FE11, 1:50; Oncogene Research Products, San Diego, CA), MSH6 (clone 44, 1:50; BD Biosciences Pharmingen) and PMS2 (A16-4, 1:25; BD Biosciences, Belgium) antibodies were used. Tissue samples with tumour cells lacking nuclear staining for at least one of the mismatch repair proteins were considered to have a positive MSI screening status. Cases without internal positive control staining, such as lymphocytes, were considered uninformative.
Cross-tabulations were tested using χ2-tests, or Fisher's exact tests when expected frequencies in any cells were <5. To test the linear association between two ordinal scale variables, the exact linear-by-linear association test was performed. The Kaplan-Meier survival analysis was used to estimate the cancer-specific survival, and the comparisons between groups were performed with the log-rank test. The Cox proportional hazard model was used for multivariate survival analyses. Cancer-specific events were defined as death with known disseminated or recurrent disease, and cases were censored at the end of follow-up or at time of death by other causes. Statistical analyses were performed using SPSS statistical software version 15.0. p-values <0.05 were considered statistically significant.
Positive immunostaining for EGFR was evident in the central part in 176/386 (46%) of analyzed primary tumours, and the invasive margin was positive in 222/386 (58%). A similar expression in both the central part and the invasive front was evident in 286/386 (74%) tumours, and an increased expression was evident in 97/386 (25%) tumours (Fig. 1, Table 2).
EGFR expression in relation to clinicopathological parameters
Table 1 shows the correlation between EGFR expression at the invasive margin and clinicopathological variables. When including all tumours, no correlation was seen to gender, age, stage, localisation or tumour type. On the other hand, there was a significant correlation between the EGFR expression at the invasive margin and tumour grade (p = 0.015) and the presence of budding (p = 0.0001), respectively.
Microsatellite instability screening status
Microsatellite instability screening status was not significantly related to EGFR expression in tumour centre (p = 0.42), neither to EGFR at the invasive margin (p = 0.40)(Table 2) nor to EGFRn/i (p = 0.13).
Kaplan-Meier survival curves in relation to EGFR expression at the invasive front are presented in Figure 2 for the full study group (361 cases with complete survival set of data) and in Figure 3 for the subgroup of colon cancers treated with potentially curative surgery (n = 170). No significant difference between the four EGFR scoring groups was observed (p = 0.070 and p = 0.26, respectively), with exception for 2+ at the invasive margin in the full study group, which had a poorer survival than the reference group (0), p = 0.012.3
The cases were regrouped into EGFRi (i = increased, tumours scoring higher in the invasive margin than in the central part) vs. EGFRn (n = non-increased; tumours with the same or higher score in the central part compared to the invasive margin). According to EGFRi, Kaplan-Meier survival curves for the full study group are shown in Figure 4, where a significant poorer survival was observed for EGFRivs. EGFRn (Univariate Cox regression analysis HR, 1.67; 95% CI, 1.16–2.40; p = 0.006; 5-years survival 52.1% vs. 61.4%). When analyzing the subgroup of potentially curatively resected colorectal cancer patients with available survival data (n = 273), the analysis showed a significant survival disadvantage for the EGFRi group, (univariate HR, 2.05 95% CI, 1.15–3.66; p = 0.013; 5-years survival 74.9% vs. 79.8%). The poorer prognosis was also evident when analysing the subgroup of potentially curatively resected colon cancer patients (HR, 2.57; 95% CI, 1.22–5.44; p = 0.014; 5-years survival 72.3% vs. 83.9%; Fig. 5).
Multivariate survival analysis
A multivariate analysis was performed to further investigate the importance of increasing EGFR expression in comparison to other prognostic parameters. Multivariate analysis was performed including gender, age, localisation, grade, stage and tumour type, and a lower survival rate for EGFRi tumours was evident with a relative risk of 1.53 (p = 0.029) (Table 3). When adding the variable of budding to the same multivariate model, the risk estimate for EGFRi was attenuated and lost statistical significance (HR, 1.18, 95% CI, 0.76–1.82; p = 0.46).
Our study of EGFR expression in 386 colorectal cancer patients shows that patients with increased EGFR expression at the invasive margin have a significantly worse prognosis. Moreover, an increased EGFR expression at the invasive margin was significantly related to the presence of tumour budding.
EGFR is upregulated in many types of cancers, including colorectal cancer, and the receptor has attracted a lot of focus during the development of new therapeutic targets.11 Increased levels of EGFR seems to be an important factor driving the aggressive behaviour of cancer cells.12 However, even though many studies have shown EGFR expression in colorectal cancer, the impact on prognosis has so far not been fully investigated.15 To the best of our knowledge, our study is the first one to correlate increased EGFR expression at the invasive margin to a poor prognosis in both uni- and multivariate survival analyses including all tumour stages, and our findings have implications for further studies.
We found that 25% of the primary tumours studied had an increased EGFR expression at the invasive margin compared to the tumour centre. An increase of EGFR expression at the invasive margin has been previously observed in smaller studies only including Stage IV tumours.19, 21, 22 The same was also noticed but not reported in a recent study of Stage IV patients from our group.23 The largest study of the studies mentioned above investigated 102 Stage IV colorectal cancer patients, showing a shorter survival in patients with high EGFR levels at the invasive margin.19 Buckley and Kakar21 studied different antibodies for colorectal EGFR analysis in 65 Stage IV tumours and also briefly mentions a higher EGFR expression level in the advancing edge. Nevertheless, our study represents a larger and unselected patient material strengthening and generalising the previous observations.
In the last decades, the biological processes at the invasive margin of colorectal cancer tumours have been the scope of many studies.24–26 In colorectal cancer, EMT can often be observed at the invasive margin. During EMT, invasive colorectal cancers form tumour buddings, small clusters of malignant cells close to but separated from the invasive front. Budding without respect to EGFR status is well studied in colorectal cancer, and several studies have suggested the use of budding as a prognostic factor in the clinical setting.10, 27 Our results, where the increased EGFR expression at the invasive margin and the presence of buddings correlated to clinical outcome, indicate a mechanism used by the tumour cells to invade the surrounding tissue.
In the EMT process, the cell-cell adhesion molecule E-cadherin is downregulated, and the tumour cells becomes more migratory.5 EMT is induced by. for example, epidermal growth factor (EGF), transforming growth factor β (TGFβ), hepatocyte growth factor (HGF), insulin growth factor and metalloproteinases (MMPs).28 EGF signalling is known to induce expression of the transcription factors Snail, Slug and Twist that suppresses transcription from the E-cadherin promoter, and there are reports showing E-cadherin downregulation at the invasive front.29–31 In addition, the ligand EGF has also been shown to stimulate the formation of invadopodia,32 suggesting that EGFR signalling at the invasive margin is important both by mediating proteolysis of the extracellular matrix and migration and invasion of the cancer cells.
The new EGFR-targeted therapies provided in combination with traditional chemotherapy improves outcome in the high-risk population, but the methods to select patients for the treatments are still not well defined. Immunohistochemical EGFR expression was previously mandatory to select patients for Cetuximab treatment, but subsequent studies have shown that IHC, when using >1% as cut-off level for EGFR positivity, does not correlate to therapeutic response to therapy with Cetuximab.17 Selection of the patients is at present performed by PCR to exclude cases with KRAS mutations. Nevertheless, even though these criteria are met, only a minority of patients respond to the treatments, leaving room for speculations of other involved mechanisms.33 Interestingly, epithelial cells that has undergone EMT is much less sensitive to treatment with EGFR inhibitors than other epithelial tumour cells,34 and there are data suggesting that it is not the dosage of the EGFR but rather the activation of EGFR signalling that mediates sensitivity to anti-EGFR treatment.35 In our study, no patients had undergone EGFR-targeted therapy, but it would be interesting to see whether the EGFR expression at the invasive front correlates to response to anti-EGFR treatment.
As outlined in Material and Methods section, 12.3% (57/464) of the cases were excluded due to lack of positive staining in perineurial structures, which was used as an internal positive control as suggested by Goldstein and Armin.19 The reason for loss of staining is not fully understood, but the tissue fixation process is important. The use of internal positive controls and exclusion of negative cases is a significant strength of our study and makes the immunohistochemical-based results more reliable. Moreover, the excluded cases seem to be randomly distributed in the patient material, and they have not significantly different distribution in terms of stage and grade compared to the cases included.
In conclusion, the results of our study show that an increased EGFR expression at the invasive margin, especially in tumour budding, indicates a tumour subtype with a more aggressive behaviour. The results stress the need for further studies of EGFR expression in CRC which maybe in the context of selecting patients for anti-EGFR treatments.
The authors thank Björn Tavelin for help with the statistical analysis and Kerstin Näslund for skilful technical assistance.
- 11Targeted therapy in advanced colon cancer: the role of new therapies. Ann Oncol 2004; 15 ( suppl 4): iv55–62., , , , , .
- 19Epidermal growth factor receptor immunohistochemical reactivity in patients with American Joint Committee on Cancer Stage IV colon adenocarcinoma: implications for a standardized scoring system. Cancer 2001; 92: 1331–46., .