Ep-CAM expression in squamous cell carcinoma of the esophagus: a potential therapeutic target and prognostic marker

The tumor tissue was snap-frozen in liquid nitrogen immediately after removal and stored at -80°C until use. The Ep-CAM antigen was detected with the monoclonal antibody Ber-EP4 (IgG1, Dako, Hamburg, Germany), which can be used on snap frozen material as described previously[24]. Briefly, 5 μm cryostat sections were cut from each tumor and were transferred onto glass slides pretreated with 3-triethoxysilyl-propylamin (Merck, Darmstadt, Germany). One section was stained with conventional hematoxylin an eosin (H & E) staining and the following section was stained using the alkaline phosphatase-antialkaline phosphatase (APAAP) technique[24]. Sections of normal colonic mucosa and MCF-7 cells served as positive staining controls, and isotype-matched, irrelevant murine monoclonal antibodies served as negative controls (MOPC 21 for IgG1; Sigma, Deisenhofen, Germany). For the evaluation of the Ep-CAM staining we used standardized criteria to evaluate membranous staining specified by Dako for interpretation of the HercepTest for p185^HER2 expression. The Dako scoring system has a scale from 0 - 3+: 0, no staining or 10% or less of the tumor cells show any level of positive staining; 1+, a faint membrane staining is detected in more than 10% of the tumor cells or the cells are only stained in part of their membrane; 2+, weak to moderate staining of the entire membrane is observed in more than 10% of the tumor cells; 3+, strong staining of the entire membrane in more than 10% of the tumor cells. Heterogeneous staining was defined as different staining intensity in more than 25% of Ep-CAM-positive tumor cells. The inter-observer reproducibility in our study was 93%. The slides with discrepant assessments were reevaluated, and a consensus was reached in all cases.

To test the correlation between the clinico-pathological data and the level of Ep-CAM expression we used the Fisher's exact test and whenever appropriate the χ²-test. All of the variables were dichotomised. For analysis of follow-up data, life table curves were calculated using the Kaplan-Meier method, and survival distributions were compared using the log-rank test. The primary end points were disease-specific survival or relapse-free survival, as measured from the date of surgery to the time of the last follow-up or cancer-related death or tumor relapse, respectively. Data of patients who were still alive and without evidence of tumor relapse at the end of the study were censored. For multivariate analysis, a parametric survival regression model based on a Weibull distribution was applied. Model selection was carried out using the Bayesian Information Criterion (BIC) [25, 26]. This information criterion was adopted to search for a parsimonious model that fit the data [27]. The lower the BIC the better the model explains the data. A search in the model space was done with a stepwise algorithm. The range of the models examined in the stepwise search was from the simplest model with a constant term, to the most complex model with all covariates included. The investigated distributions were loglogistic, lognormal, logistic, exponential, Gaussian and Weibull. Both, the exponential and the Weibull distributions fitted the data best. Because the exponential distribution is a particular case of the Weibull when the scale parameter equals one, we decided to use the Weibull distribution. Statistical computations were done with SPSS and the statistical package R [28]. The level of significance was set at p < 0.05.

In a first step, we investigated the Ep-CAM expression in 10 normal, non-pathologic squamous epithelia of the esophagus. As previously reported [13], the non-pathologic esophageal squamous epithelium was negative for Ep-CAM expression. In contrast four different levels of Ep-CAM expression (Figure 1), which were classified according to the Dako scoring system, were observed in the 70 esophageal SCC. We found 3+ Ep-CAM expression in 29/70 of the investigated tumors (41%). Eight of the 70 samples were scored as 2+ (11%) and their membranous Ep-CAM staining was less intense, when compared to the 3+ samples. A weak Ep-CAM staining scored as 1+ was observed in 18/70 (26%) tumors and an absent expression of Ep-CAM was detected in 15/70 tumors (21%). Neither in the complete patient collective (n = 70) nor in the group with clinical follow up (n = 53, Table 1 and 2, respectively) any statistical significant correlation between the level of Ep-CAM expression and the clinico-pathological parameters pT-category, pN-category and grading were observed. It is important to note that a heterogeneous Ep-CAM expression was seen at all expression levels and no correlation was observed between Ep-CAM heterogeneity within a tumor and its Ep-CAM expression score (Chi-square test, p = 0.134). Also, a specific staining pattern, e.g. intense staining at the invasive front, was not detected.

In the 53 patients with follow-up data and a median observation time of 15.0 months (range: 2 – 95 months) we found, that the general neo-expression of Ep-CAM (1+ - 3+) did not correlate with the patient's prognosis. The median cumulative disease-specific survival of the patients was 17.0 months for Ep-CAM 1+ - 3+ positive tumors and 24.0 months for Ep-CAM negative tumors (log rank, p = 0.3491; relapse-free survival: Ep-CAM 1+ - 3+ 16.0 months vs. Ep-CAM negative 43.0 months; log rank, p = 0.5993). Interestingly, tumors exhibiting a strong Ep-CAM expression (3+) conferred a significantly decreased median relapse-free survival interval. The median cumulative relapse-free survival interval of patients with Ep-CAM 3+ expression was 9.0 months, compared to 43.0 months for patients with 0 - 2+ Ep-CAM expression (log rank, p = 0.0001) (Figure 2A). The median disease-specific survival was 11.0 months for patients with Ep-CAM 3+ expressing tumors compared to > 24.0 months for those with 0-2+ Ep-CAM staining (log rank, p = 0.0003, Figure 2B). As expected, the pN-category was also of prognostic significance in our investigated series with a median disease-specific survival of 13.0 months for patients with positive nodal status compared to 75.0 months for patients without regional lymph node involvement (log rank, p = 0.0170). The prognostic value of the pT-category for disease-specific survival was not statistically significant (pT1–2 vs. pT 3–4, 16 months vs. 28 months, respectively; log rank, p = 0.1178). Tumor grading was also of no prognostic significance.

For multivariate regression analysis we included Ep-CAM expression (Ep-CAM 3+ vs. Ep-CAM 0-2+), sex, age, pN, pT and grading. We found an independent prognostic influence for Ep-CAM 3+ on relapse-free survival and disease-specific survival (p = 0.002 and p = 0.005, respectively) (Table 3). We also tested the possible interaction between the pN category and Ep-CAM expression within a multivariate model and confirmed the indepence of both, Ep-CAM and pN-category as independent prognostic factors (data not shown).