Article Text

Download PDFPDF

EpCAM is predominantly expressed in high grade and advanced stage urothelial carcinoma of the bladder
  1. A Brunner1,
  2. M Prelog2,
  3. I Verdorfer1,
  4. A Tzankov3,
  5. G Mikuz1,
  6. C Ensinger1
  1. 1
    Institute of Pathology, Medical University of Innsbruck, Austria
  2. 2
    Department of Pediatrics, Medical University of Innsbruck, Austria
  3. 3
    Institute of Pathology, University of Basel, Switzerland
  1. Dr Andrea Brunner, Institute of Pathology, Medical University of Innsbruck, Muellerstrasse 44, Innsbruck, Austria; andrea.brunner{at}


Background: EpCAM is an adhesion molecule of the basolateral membranes in a variety of epithelial cells. Over-expression has been detected in many epithelial tumours and has been associated with high stage, high grade and a worse survival in some tumour types.

Aims: To assess the prognostic value of EpCAM in urothelial carcinoma of the bladder.

Methods: EpCAM expression was analysed by immunohistochemistry using a monoclonal antibody (clone VU-1D9) on a tissue microarray comprising 99 urothelial carcinomas of the bladder diagnosed between 1994 and 1997.

Results: A significant relationship between high grade, advanced stage, and EpCAM expression was found, and expression of EpCAM was associated with a worse overall survival when compared to EpCAM negative tumours (p = 0.033). Multivariate analysis showed that EpCAM expression was not an independent prognostic factor for overall survival in urothelial carcinoma of the bladder.

Conclusion: EpCAM expression is associated with advanced stage, high grade and poor overall survival in urothelial carcinoma of the bladder, but lacks an independent prognostic significance. The strong association with high grade tumours suggests a possible role during tumour progression and makes EpCAM a potential target for antibody mediated therapy.

View Full Text

Statistics from

EpCAM is a type 1 transmembrane glycoprotein expressed in developing tissues as well as at the basolateral membrane of most adult epithelial tissues, except for squamous epithelium.1 2 Functionally, EpCAM mediates calcium-dependent homotypic cell–cell interactions, interacting with E-cadherin dependent cell adhesion.13 Recently, EpCAM has been shown to be directly involved in intracelluar signal transduction and growth regulation inducing c-myc and cyclins A and E in epithelial cells and fibroblasts.4 Due to its widespread expression in epithelial cancers, EpCAM has also become a potential target for immunotherapy as well as for gene therapy.510

De novo expression of EpCAM was described in cervical intraepithelial neoplasia, associated with increasing grade of dysplasia.11 In a variety of epithelial tumours, including cancer of the thyroid, breast, tongue, lung, gastrointestinal tract, prostate and bladder, EpCAM expression was reported.7 1223 In certain tumour types, over-expression of EpCAM was associated with dedifferentiation, advanced stage, invasive behaviour and metastasis and was linked to a worse overall survival in breast, ovarian, cholangiocellular, pancreas and oesophageal carcinoma.7 1317 21 24

In normal urothelium, expression of EpCAM is weak and restricted to the superficial cells.22 23 In urothelial carcinoma of the bladder, EpCAM expression was reported to be more frequently found in intermediate and high grade tumours.22 23 25 To date, the prognostic value of EpCAM in urothelial bladder cancer has not been investigated.

To assess the prognostic value of EpCAM in low and high grade urothelial carcinoma of the bladder, we performed an immunohistochemical study using a tissue microarray.



Our retrospective study included 99 cases of urothelial carcinoma of the bladder, diagnosed between January 1994 and December 1997 at the Institute of Pathology, Medical University of Innsbruck; only specimens of primary diagnoses obtained by transurethral resection were included. Disease recurrence and progression were defined as previously reported.26

Tissue microarray construction

Samples were brought into a tissue microarray format as previously described.26 27 Briefly, three tissue cores (0.6 mm) were obtained; in small specimens, 1–3 cores were obtained, based on tissue amount, including the superficial as well as the invasive portion of the tumour and the adjacent tumour stroma.


Immunohistochemistry was performed by an automated immunostainer (Nexes, Ventana, Tucson, Arizona, USA), applying the streptavidin–biotin peroxidase technique with diaminobenzidine as chromogen. The monoclonal antibody ESA (clone VU-1D9, NovoCastra, Medac, Hamburg, Germany) was used at a dilution of 1:50. Antigen retrieval was performed with pronase 1 (Ventana) for 4 min (37°).

Because intensity varies between cases due to different tissue preservation, the staining intensity was not considered. Cytoplasmic precipitation was regarded as unspecific and only cells with distinct membranous reactivity were evaluated. EpCAM expression was calculated as the relative proportion (percentage) of EpCAM positive tumour cells for each case.


For statistical analysis, the Statistical Package of Social Sciences (SPSS V.12.0 for Windows) was used. The χ2 test was used to test the relationship between EpCAM expression and clinical and pathological parameters. Analysis of variance (ANOVA) was applied to study differences of means between groups. For determination of optimal cut-off values of continuous variables, receiver operating characteristic (ROC) curves, plotting sensitivity versus 1 − specificity, were used. Considering the diagnostic performance, an area under the ROC (AUROC) closer to 1 or to 0 indicates greater discriminatory power, whereas an AUROC of 0.5 denotes no diagnostic potential. The optimal cut-off point was calculated using Youden’s index (Y), denoting Y = sensitivity+specificity−1, since this method can be applied when there is no particular requirement on sensitivity and/or specificity.28 Recurrence-free survival (RFS) and overall survival were analysed by the Kaplan–Meier method, applying the cut-off values calculated by the ROC/Y and compared by the log rank test, except for continuous variables, where a Cox regression model was applied. Multivariate analysis was performed to identify independent prognostic markers for overall survival using a Cox multistep regression model. p<0.05 was considered significant.



The clinical and pathological characteristics of the 99 patients have been reported elsewhere.26 Briefly, there were 27 pTa, 33 pT1, 34 pT2-4 tumours and five carcinomata in situ (CIS), all reclassified according to the current WHO classification.29 Fifty-two patients had recurrent disease, with 17 progressions at relapse and tumour associated CIS in 12 (12%) patients. Cystectomy was performed in 17 patients, 8 immediately and 9 later on. Within the mean follow-up of 60 months (range 0–142 months) there were 39 deaths as a result of the disease; 40 patients died from other causes.

Tissue microarray quality

Three tissue cores/case could be arrayed in 90 cases, whereas 2 and 1 cores could be obtained in 6 and 3 cases, respectively, resulting in a total number of 285 tissue microarray cores. After immunohistochemistry, 231 cores (81%) and 94 cases (95%) could be evaluated.

EpCAM expression and correlations

Positive EpCAM staining with a distinct membranous pattern was found in 51 (54%) specimens; 43 (46%) cases remained negative (fig 1). There was a significant correlation between EpCAM expression and tumour grade (pχ2 = 0.004), stage (pχ2 = 0.014), as well as with expression of tetranectin in tumour cells (pχ2 = 0.002). High grade tumours expressed EpCAM in a mean proportion of 22 (32.8)% of tumour cells, while low grade tumours expressed EpCAM in a mean proportion of 1.4 (2.3)% of tumour cells (pANOVA = 0.004). No significant correlation was seen with age, associated CIS, progressive disease and development of recurrence (table 1).

Figure 1 High grade urothelial cancer shows a strong homogeneous (A) (100×) and distinct membranous expression of EpCAM (B) (600×), while low grade tumours reveal only a weak, membranous staining (C) (400×) or completely lack EpCAM (D) (400×).
Table 1 Correlation of EpCAM expression with clinical and pathological parameters (n = 94)

Survival analysis

Overall survival was significantly influenced by age, grade, stage and number of recurrences, but not by tumour associated CIS and progressive disease.26 ROC showed borderline discriminatory power considering overall survival for the expression of EpCAM (AUROC = 0.604, p = 0.121). The optimal cut-off value for EpCAM expression considering overall survival determined by ROC/Y was at 2.5% positivity with a specificity of 62% and sensitivity of 60% (fig 2). Analysed by the Kaplan–Meier method and applying cut-off values suggested by ROC/Y, patients with EpCAM expression in >2.5% of tumour cells was associated with worse overall survival (41 mortalities in 51 patients, mean overall survival 53 months, median overall survival 35 months) compared to patients with EpCAM expression in <2.5% of tumour cells (27 mortalities in 43 patients, mean overall survival 77 months, median 72 months; p = 0.033) (fig 3). Multivariate analysis considering grade, stage, age, progressive disease and EpCAM expression revealed that only grade, age and number of recurrences but not stage and EpCAM expression were independent prognostic factors considering overall survival (table 2).

Figure 2 ROC curve for determination of the optimal cut-off value for EpCAM expression in urothelial carcinoma of the bladder.
Figure 3 Prognostic significance of EpCAM expression in patients with bladder cancer. EpCAM positive (>2.5%) vs EpCAM negative cases compared by the log rank test (n = number of events/number of cases).
Table 2 Multivariate analysis for the independent prognostic value of factors influencing overall survival


We have reported on the prognostic value of EpCAM in urothelial carcinoma of the bladder. EpCAM expression has been more frequently detected in intermediate grade urothelial carcinoma with a heterogeneous expression pattern and in high grade tumours, revealing strong and homogeneous expression.22 23 25 Indeed, we showed a significant correlation of EpCAM expression with tumour grade. This finding is further in line with the known association of EpCAM over-expression with other poorly differentiated and aggressive tumour types.7 14 16 24 Furthermore, we showed a significant association of EpCAM in urothelial carcinoma with a worse overall survival. However, results concerning the prognostic significance of EpCAM expression are conflicting. High levels of EpCAM expression have been shown to be an independent prognostic factor associated with worse overall survival in breast and ovarian cancer, oesophageal carcinoma and carcinoma of the gall bladder, whereas in other tumours no such association or even a better overall survival in EpCAM positive tumours was reported.14 15 17 18 20 21 24 3032

The different results obtained in various studies might reflect the complex function of EpCAM in different tumour types, which is probably related to the physiological function of EpCAM in the underlying epithelium and the associated molecular pathways. Furthermore, it has been suggested that EpCAM might be only temporarily expressed during tumour progression, varying between specific tumour types.33 Finally, some bias might be due to difficulties and inconsistencies by the determination of cut-off levels, making study results frequently incomparable. Our study is the first in which determination of cut-off levels for EpCAM expression has rationally been addressed applying ROC curves.28

The exact functions of EpCAM are still under investigation. In vitro studies in cell lines from breast and tongue cancer have shown that blocking of EpCAM by short interfering RNAs results in a decrease of proliferation, migration and invasion, suggesting a direct oncogenic role for EpCAM.6 7 EpCAM was found to directly up-regulate the oncogene c-myc.4 EpCAM gene silencing leads to decrease of cytoplasmic β-catenin, which is necessary for wnt-signalling.6 7 The wnt-pathway is known to be involved in cancer development, activating several target genes such as c-myc, vascular endothelial growth factor and cyclooxygenase 2.6 7 In urothelial carcinoma of the bladder EpCAM expression is related to an aggressive phenotype, therefore we assume a direct oncogenic role.

The association with intratumoural production of proteins involved in proteolysis of the extracellular matrix, such as tetranectin, may further strengthen involvement of EpCAM in tumour progression and invasion.26 The association of EpCAM expression in urothelial carcinoma of the bladder with high grade, advanced stage and poor outcome points towards its possible role as a target for adjuvant immunotherapy in progressive urothelial carcinoma. Importantly, an EpCAM-specific immunotoxin consisting of a humanised anti-EpCAM antibody fused to a subunit of the bacterial Pseudomonas exotoxin is being adapted for the local treatment of urothelial carcinoma.8

In conclusion, we have shown that in urothelial carcinoma of the bladder, EpCAM expression accompanies advanced stage, high grade and is associated with poor overall survival. Although EpCAM does not have an independent prognostic value, increased expression in high grade tumours suggests a possible role during tumour progression and makes EpCAM a potential target for antibody-mediated treatment in urothelial carcinoma of the bladder.

Take-home messages

  • EpCAM expression is significantly associated with high grade and advanced stage in urothelial carcinoma of the bladder.

  • EpCAM expression in urothelial carcinoma indicates a worse overall survival.

  • Strong EpCAM expression in high grade urothelial carcinoma makes it a potential target for immunotherapy.


We thank H Muehlboeck from the Cancer Registry of Tyrol/Austria, Institute for Epidemiology of TILAK for providing epidemiological data and S Jöbstl for excellent technical assistance.


View Abstract


  • Competing interests: None.

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.