CD24 Expression Is a New Prognostic Marker in Breast Cancer

Breast cancer is the most common malignant tumor of females in the western world. A rising incidence of this disease was observed in Germany in the last 20 years, and 46,000 new cases were diagnosed in 1998. In the United States alone, 203,500 new cases were expected for 2002 (1, 2). The incidence of breast cancer remains high, and the clinical courses are highly variable. It is of general importance to predict the biology of the tumor and, thus, the course of the disease in the individual patient to ensure adequate therapy and patient surveillance. Conventional prognostic and predictive markers for breast cancer are nodal status, tumor grade, tumor size, and tumor type (3, 4). Additionally, molecular markers are being sought and established to allow for a refined classification of prognosis, especially in patient subgroups whose outcome can only insufficiently be predicted by conventional parameters. Among others, candidate genes of current interest are telomerase, kallikrein 5, urokinase plasminogen activator and its inhibitor (PAI-1), tissue inhibitor of metalloproteinases 1, Ep-Cam, c-erbB2, and osteopontin (5, 6, 7, 8, 9, 10). A novel prognostic marker gene we identified recently in epithelial ovarian cancer, non-small cell lung cancer, and prostate cancer is CD24 (11, 12, 13).

CD24 is a small, heavily glycosylated mucin-like glycosylphosphatidyl-inositol-linked cell surface protein that is expressed in a wide variety of human malignancies, e.g., B cell lymphoma, renal cell carcinoma, small cell and non-small cell lung carcinoma, nasopharyngeal carcinoma, hepatocellular carcinoma, bladder carcinoma, glioma, epithelial ovarian cancer, and breast cancer (12, 13, 14, 15, 16, 17, 18, 19, 20, 21, 22, 23, 24, 25, 26, 27, 28, 29, 30, 31, 32, 33, 34). Functionally, CD24 expression might enhance the metastatic potential of tumor cells, because CD24 has been identified as an alternative ligand of P-selectin, an adhesion receptor on activated endothelial cells and platelets (35, 36, 37, 38, 39). Fogel et al. (31) described CD24 expression in breast cancer in a frozen section-based pilot study with 29 cases that did not allow for a detailed analysis. We aimed to investigate the expression of CD24 in a larger tumor collection of breast cancer and to evaluate its prognostic significance.

The A125 adenocarcinoma cell line and CD24-transfected A125 cells have been described recently (37). The monoclonal CD24 antibody (Ab-2, clone 24C02/SN3b) was purchased from Neomarkers (Fremont, CA). Staining of the cells with this antibody and phycoerythrin-conjugated goat antibodies to mouse immunoglobulins (SERVA, Heidelberg, Germany) has been performed as described (39) and was analyzed with a FACScan (Becton Dickinson, Heidelberg, Germany).

Our study included 201 patients with primary breast cancer who were diagnosed at the Institute of Pathology, Charité University Hospital, between 1991 and 1997. Patient age at the time of diagnosis ranged from 30 to 87 with a median of 59 years. Follow-up data including overall survival, and disease recurrence or progression times were available for all of the cases. The average observation time for overall survival was 65 months for patients still alive at the time of analysis, and ranged from 5 to 129 months. Thirty-four patients (16.9%) died during follow-up, and 68 patients (33.8%) experienced disease progression defined by either metastatic or local recurrent disease. Adjuvant therapy was administered as follows: 20 patients did not receive any adjuvant therapy, 29 patients received radiotherapy only, 26 patients had CTX² only, 10 patients had radio- and CTX, 49 patients were treated with TAM exclusively, 16 patients had TAM after CTX, 36 patients had TAM after radiotherapy, and 3 patients had a combination of radio- and CTX plus TAM. For 12 patients no data on adjuvant therapy was accessible. For statistical analysis of the impact of therapy, we arranged the patients into two groups. The first group received either no or local therapy/radiotherapy (49 cases), or systemic therapy excluding TAM (36 cases). The second group had received TAM with or without an additional systemic or local therapy (104 cases). The selection of cases for this study was based on availability of tissue, and these were not stratified for any known preoperative or pathological prognostic factors. Cases with systemic disease (M1) at the time of diagnosis were excluded. Tumor histology was determined according to the criteria of the WHO. The stage of tumors was assessed according to Unio Internationale Contra Cancrum (40). Tumors were graded according to Bloom and Richardson in the modification of Elston and Ellis (41). Data regarding the estrogen receptor status and the expression of c-erbB2 were taken from the archival pathology reports. Estrogen receptor positivity was defined as an immune reactive score >3. Overexpression of c-erbB2 was defined according to the clinical trial assay (2+, 3+) as recommended in the hercep-test (DAKO). The clinicopathological characteristics of the tumor collection are described in Table 1.

Formalin-fixed, paraffin embedded tissue was freshly cut (4 μm). The sections were mounted on superfrost slides (Menzel Gläser, Braunschweig, Germany), dewaxed with xylene, and gradually hydrated. Antigen retrieval was achieved by pressure cooking in 0.01 M citrate buffer for 5 min. The primary CD24-antibody (Ab-2, clone 24C02; Neomarkers) was diluted 1:100 using a background reducing dilution buffer (DAKO, Hamburg, Germany). No other blocking agents were used. The primary antibody was incubated at room temperature for 2 h. As a negative control, four slides were processed without primary antibody. Detection took place by the conventional labeled-streptavidin-biotin (DAKO) method with alkaline phosphatase as the reporting enzyme according to the manufacturer’s instructions. Fast-Red (Sigma-Aldrich, Munich, Germany) served as chromogen. Afterward, the slides were briefly counterstained with hematoxylin and aquaeously mounted.

The immunostainings were examined independently by a doctoral candidate (E. M.) and two clinical pathologists (G. K., H. G.), who were blinded to patient outcome. We aimed to keep our scoring system of the CD24 staining as simple as possible to minimize interobserver variability and to enhance the reproducibility of our findings in future studies. We evaluated the membranous and the cytoplasmic staining intensity of CD24 separately. Negative cases had to show definitely no CD24 immunoreactivity in any part of the tumor. All of the other cases, beginning with a weak but unequivocal staining of tumor cells, were defined as positive. The preliminary analysis of membranous CD24 staining did not reveal any significant associations with prognostic or other tumor parameters. Therefore, we concentrated on the qualities of total CD24 and cytoplasmic CD24 for statistical analysis. All of the cases with different scorings were discussed at a multiheaded microscope until consensus was reached.

The data were compiled with the software package SPSS, version 10.0. Fisher’s exact and χ² tests were used to assess the statistical significance of the correlation between expression of CD24 and clinicopathological parameters. Univariate survival analysis was performed according to Kaplan-Meier, and differences in survival curves were assessed with the log rank test. Multivariate survival analysis was performed on all of the parameters that were found to be significant on univariate analysis using the Cox regression model. Ps <0.05 were considered significant. All of the statistics were accredited by the head biostatistician of the Tumor Centre, Charité University Hospital (J. B.).

We examined the reactivity of two monoclonal antibodies to CD24 (ML5 and Ab-2) using CD24-transfected cell lines. As shown in Fig. 1, both antibodies demonstrated a stronger staining in the CD24-transfected cell line (Fig. 1,B), whereas the nontransfected cell line showed no significant signal (Fig. 1 A).

In normal breast tissue, CD24 was rarely expressed, only focally an apically accentuated membranous staining of ductal epithelium was noted (Fig. 1,C). Dilated ducts virtually always displayed a strong apical immunoreactivity (Fig. 1,D). This apical staining pattern was also observed in a fraction of carcinomas. In intraductal and invasive carcinomas an additional cytoplasmic staining quality was observed, which was often membranously accentuated at the circumference of the cell, or showed a mosaic pattern (Fig. 1, E–G). In invasive carcinomas, we observed membranous CD24 expression in 55.7% and a cytoplasmic CD24 expression in 76.1%. Both staining qualities showed a rate of concordance of 62.7%. In total, 84.6% of tumors expressed CD24.

We did not find any significant association of overall CD24 staining intensity with patient age, tumor type, tumor size (pT), tumor grading, estrogen receptor, or c-erbB2 expression. Only a positive nodal status (P = 0.018) was linked to CD24 positivity (Table 2). Comparable associations were found for cytoplasmic CD24 staining. No correlations between membranous CD24 immunoreactivity and these clinicopathological parameters could be demonstrated.

In univariate survival analyses, cumulative survival curves were calculated according to the Kaplan-Meier method (Table 3). Differences in survival were assessed with the log-rank test. We analyzed the impact of all of the parameters on overall survival and disease-free survival. The conventional prognostic markers tumor grade, size/pT, nodal status, and disease stage according to Unio Internationale Contra Cancrum reached significance for overall and disease-free survival. Cytoplasmic CD24 positivity was associated with shorter disease-free survival (P = 0.0008) and overall survival (P = 0.031), whereas the influence of total CD24 expression was only significant for shorter disease-free survival (P = 0.004). Membranous CD24 remained insignificant in either analysis. Fig. 2 illustrates the impact of CD24 expression on survival times.

To assess whether CD24 is a true prognostic marker or a marker of sensitivity to adjuvant therapy, we repeated the aforementioned analyses in the groups of patients who had received either CTX only or TAM with or without CTX. Kaplan-Meier analysis revealed that in both therapy subgroups, total CD24-positive tumors were associated with shorter disease-free survival times, reaching significance in the CTX group (P = 0.0298) and a trend (P = 0.0639) in the TAM group. For cytoplasmic CD24, only in the TAM group could a significantly shorter disease-free survival time be demonstrated (CTX: P = 0.129; TAM: P = 0.0055). In the accordingly stratified analysis of overall survival times, no significant differences became apparent. In patients who had not received any systemic therapy (n = 49), no impact of CD24 status on either disease-free survival or overall survival was found, which might reflect the few cases.

Significantly shorter disease-free survival times were also found in CD24-positive cases in the subgroups of estrogen-positive tumors (P = 0.0145) and tumors without c-erbB2 overexpression (P = 0.0189).

The analysis was based on the Cox regression model to test the influence of each parameter on overall and disease-free survival. In the Cox regression we included CD24 expression (negative versus positive), tumor grade (G1, G2, and G3), nodal status (negative versus positive), and pT-stage (pT1, pT2, pT3, and pT4). We tested the impact of total CD24 (Table 4) and cytoplasmic CD24 (Table 5) on overall and disease-free survival.

Disease-free survival time was significantly dependent on nodal status, tumor grading, and CD24 expression for cytoplasmic and total CD24 positivity alike. The number of events (deaths) was insufficient for a valid Cox regression analysis of overall survival. Although we observed relative risk values similar to those of the disease-free survival analysis, these did not reach significance with the exception of pT stage.

In a preliminary multivariate analysis we have tried to evaluate the prognostic significance of CD24 in different adjuvant therapy subgroups. In this stratified repetition of this multivariate analysis, CD24 showed only a trend (P = 0.066) for disease-free survival and lost its significance for overall survival (P = 0.171). This might be based on the few cases in either group and requires a larger study.

CD24 has been described in B-cell development and B-cell neoplasia, in the developing pancreas and brain, and in regenerating muscle, keratinocytes, renal tubules, and a large variety of malignant tumors (12, 13, 18, 19, 20, 21, 22, 23, 24, 25, 26, 27, 28, 29, 30, 32, 33, 34). In this immunohistochemistry-based study we describe the expression of CD24 protein in breast cancer. Basically, we can confirm the findings of Fogel et al. (31), who were the first to describe CD24 expression of breast cancer in an immunohistochemistry study based on frozen sections. They reported an apical membranous CD24 immunoreactivity in benign ducts and an additional cytoplasmic staining in invasive carcinomas, and described CD24 expression as a marker of breast cancer. We used a commercially available monoclonal CD24 antibody that is applicable to paraffinized tumor tissue and, thus, made a comprehensive analysis of clinically characterized archive material possible. The specificity of the antibody was additionally ascertained by cytofluorographic analysis using a described CD24-transfected cell line (31).

Our study suggests CD24 protein expression as a new prognostic marker in breast cancer that retained its prognostic impact on disease-free survival even in a multivariate analysis. Because the majority of breast cancers is CD24 positive, this new information is especially valuable for patients with CD24-negative tumors, who appear to carry a very low risk of tumor progression, as shown in Kaplan-Meier curves (Fig. 2). Interestingly, this prognostic impact of CD24 is especially significant in estrogen-positive or c-erbB2-negative (0, 1+) tumors, subgroups supposed to carry a more favorable prognosis under current therapy regimens. Similarly to our recent study on ovarian cancer, the prognostic value of CD24 positivity appeared to be strongest when the analysis concentrated on the cytoplasmic CD24 staining. This is a staining quality we predominantly observed in malignancy and, therefore, might reflect a pathological condition.

A limitation of our study and a general problem of comparable studies are the few patients who had not received any adjuvant therapy. Therefore, we feel it is virtually impossible on the ground of our data to decide whether the prognostic impact of CD24 expression is really restricted to patients with adjuvant therapy. Clearly, a larger study cohort is needed to address the question of whether CD24 is more of a true prognostic marker of patient outcome or a predictive marker of sensitivity to therapy.

We found the immunostainings with the CD24 antibody Ab-2 easy to analyze and reproducible. However, it should be noted that even in experienced hands, the intensity of immunostainings tends to be variable in different laboratories. An elegant way to standardize immunostainings would be to introduce controls, e.g., cell lines with a defined CD24 protein expression status, to accompany each staining set. We are currently screening various cell lines for their CD24 status to establish suitable controls.

How does CD24 positivity of the primary tumor lead to an increased rate of local recurrences? In general, local recurrences result either from residual gross tumor that has been overlooked by the surgeon and the pathologist or from local micrometastases that escaped the scrutiny of both. A correlation of CD24-positive tumors with positive resection margins is rather unlikely; therefore, we hypothesize that local tumor cell spread is the most likely explanation for the shortened disease-free survival time. There is in vitro evidence for a prometastatic role of CD24 in human tumor cells based on its function as a ligand to P-selectin (35, 36, 37, 38, 39). P-selectin is expressed by activated endothelial cells and platelets, and it plays an important role in marginal adhesion and migration of cells. Possibly, CD24 positive tumor cells can spread more easily because of their capacity to either form thrombi with activated platelets or to adhere to endothelia in the bloodstream, which has been demonstrated for CD24-expressing breast cancer cells (38). Moreover, CD24 positivity might induce an increased invasiveness of a tumor as suggested by in vivo data from Senner et al. (29), who reported a locally more aggressive behavior of CD24-positive gliomas in a mouse model. However, this hypothesis is not in concordance with the findings of Schindelmann et al. (42), who screened cell lines of invasive breast cancer and noninvasive breast cells for differential gene expression using cDNA microarrays and reported a significantly down-regulated CD24 expression in invasive tumor cell lines.

CD24 expression has been associated with an adverse prognosis in ovarian cancer, non-small cell lung cancer, and prostate cancer (11, 12, 13). The confirmation of this finding in breast cancer underscores the importance of CD24 in the disease progression of human carcinomas. Additional confirmational studies are currently being conducted to verify our results to establish CD24 as a molecular prognostic marker in breast cancer. This might help the clinician to individualize the surveillance scheme/therapy, for example, favoring a more aggressive regimen in strongly CD24-positive tumors. In addition, transplantation-associated B-cell proliferative syndrome had been treated with i.v. administration of CD21- and CD24-specific antibodies by Fischer et al. (43) and others (44, 45). Possibly, this new therapeutic option might deserve consideration in CD24-positive breast carcinomas as well.

We thank Britta Beyer for excellent technical assistance, Marie-Christin Koll and Jeremy Dunn for comments on the manuscript, and Bernhard Heine and Alfred E. Neumann for discussions.