Overexpression of CD24: Association With Invasiveness in Urothelial Carcinoma of the Bladder

Bladder carcinoma has one of the highest rates of fatality among urogenital cancers. Most urothelial carcinomas (UCs) of the bladder (75%) are present as superficial tumors at the time of diagnosis. Treatment management for such cancers usually involves transurethral resection (TUR) or a combination of intravesical chemotherapy and immunotherapy. However, patients with bladder cancers that have invaded the muscle layer generally develop a high risk of local invasion and distant metastasis.1 Tumor staging and grading are considered the best prognostic markers, but the development of molecular markers may provide useful information for more effective diagnoses and treatments of bladder cancer.

Human CD24 consists of a small protein core comprised of 30 amino acids. The protein is extensively glycosylated and bound to the membrane via a glycosyl-phosphatidylinositol anchor.2,3 Nearly one half the amino acids in CD24 are serine or threonine residues that are potential sites for O-linked glycans, indicating that CD24 is a mucin-type protein. CD24 was originally described as a B-cell specific marker expressed in the early stages of B-cell development.4 Subsequent studies demonstrated that CD24 is also a marker for renal tubular epithelial cells.5 Several reports have shown that CD24 can be expressed in epithelial tumors from the lung, breast, prostate, and kidney, as well as from ovarian carcinomas.6–10 These reports revealed that CD24 was strongly associated with a more aggressive course of the disease. CD24 is a ligand of P-selectin (CD62P),11 an adhesion receptor on activated endothelial cells and platelets, suggesting that CD24-positive tumor cells might develop into tumor thrombi with metastasizing capacities.12–14 

To our knowledge, the expression of CD24 in bladder UC has not yet been reported. The aim of this study was to evaluate the status of CD24 expression by immunohistochemistry in bladder UC and to investigate the association of CD24 protein expression with clinicopathologic parameters.

A total of 135 patients with bladder UCs were included in this study. Of these, 87 patients underwent TUR and 48 patients had a radical cystectomy at the Samsung Medical Center between 1995 and 2003. Inclusion criteria were histopathologic diagnosis of urothelial tumors, availability of clinical data, and the availability of paraffin-embedded tissue specimens. All patients underwent surgery but did not receive chemotherapy or radiotherapy prior to surgical resection. All of the samples were histologically evaluated as UCs and were graded. They were staged according to the tumor-node-metastasis 1997 classification.15,16 Secondary tumors were excluded from the study, and only patients with UC were used. The cases used in this study included 56 patients with stage pTa, 29 patients with stage pT1, 19 patients with stage pT2, and 31 patients with stage pT3 tumors. The patients with stage pTi tumors were excluded from this study because the sample size was small. The specimens included 22 grade 1, 50 grade 2, and 63 grade 3 tumors. The clinicopathologic characteristics of the 135 bladder cancer patients in the study are summarized in Table 1. Fifteen normal bladder tissue samples from cystectomy specimens resected for nonurologic diseases or nonneoplastic diseases were also included in the study.

For immunohistochemistry, sections were deparaffinized in xylene and rehydrated through a graded series of ethanol solutions. Endogenous peroxidase activity was quenched by incubation in 3% hydrogen peroxide in distilled water for 10 minutes. After washing with distilled water, antigen retrieval was achieved by microwaving in 10 mM citrate buffer, pH 6.0, for 5 minutes, 3 times. To reduce nonspecific staining, each section was treated with 4% bovine serum albumin and 10 mg/L dextran (Sigma Chemical Company, St Louis, Mo) in phosphate-buffered saline with 0.1% Tween 20 (PBST) for 30 minutes. Sections were incubated with an anti-CD24 monoclonal antibody (Ab-2, clone 24C02, Neomarkers, Fremont, Calif) diluted 1:100 in PBST containing 15 mg/L goat globulin (Sigma) at room temperature for 40 minutes. Sections were incubated with an anti-mouse polymer kit (DakoCytomation, Carpinteria, Calif) for 30 minutes at room temperature. For a chromogen, 3,3′-diaminobenzidine (Dako) was used. Sections were counterstained with Meyer hematoxylin. Ovarian serous adenocarcinoma tissue sections were used as a positive control.

In the cases of cystectomy specimens, staining in areas of UCs and normal urothelium were evaluated separately. The intensity of CD24 staining was semiquantitatively scored as 0 (negative), 1 (weak), 2 (moderate), or 3 (strong). Cases in which the percentage of positive cells is less than 10% with weak intensity were regarded as negative (score 0), and this was the most frequent pattern identified in normal urothelium. In sections containing more than 10% of the area staining positive, the sections were scored by intensity (score 1, 2, and 3). Each lesion was examined and scored separately by 2 pathologists (Y.L.C. and S.H.K.). Cases with discrepant scores were discussed until a uniform decision was reached. The final evaluation assigned samples to 2 groups: group I (low-level expression) for sections with a score of 0 or 1, and group II (high-level expression) for sections with a score of 2 or 3.

Statistical analyses were performed using the Pearson χ² test and Student t test to evaluate the correlation between CD24 expression and clinicopathologic parameters. P values < .05 were considered significant. All statistical analyses were performed using SPSS v11.5 software (SPSS, Chicago, Ill).

The age of patients in this study ranged from 26 years to 87 years (mean, 62.0 years). There were 117 men and 18 women, giving a male-female ratio of 6.5:1. Fifteen cases of normal bladder epithelium (n = 15) demonstrated weak immunoreactivity for CD24 with well-confined localization in the apical cytoplasm or apical membranes (Figure 1, A through G, and Table 1). Even though not all epithelial cells expressed CD24, positive staining in a distinct apical localization was present in some areas of normal bladder epithelium. In cancer specimens, the adjacent nonneoplastic urothelium showed a pattern similar to normal urothelium from nonneoplastic cases (Figure 2, A through F).

Of the 135 primary bladder UCs, a high level of CD24 expression was detected in 83 cases (61.5%) (Table 2): 44 (50.6%) of the 87 patients with TUR and 39 (81.3%) of the 48 patients with radical cystectomy showed high-level expression of CD24. Of the stage pTa papillary UCs, 28.6% displayed increases in CD24 expression in tumor cells, and most staining showed an apical localization (Figure 3, A through C). In contrast, 96.6%, 78.9%, and 77.4% of stage pT1, pT2, and pT3 tumors, respectively, demonstrated high-level expression of CD24 (Figure 4, A through D). In this study, the CD24 expression was significantly higher in invasive UCs than in noninvasive UCs (P < .001). Among tumor specimens expressing high levels of CD24, 45.5%, 48.0%, and 77.8% were grade 1, grade 2, and grade 3, respectively. The frequency of CD24 overexpression was significantly greater in high-grade tumors (P = .001) (Table 2).

Furthermore, the most distinctive feature of CD24 expression was that the invasive cancer cells lost the apical polarity of CD24 localization, which was found in normal tissue and in low-grade tumors. Even in the areas where the stromal reaction was negligible or absent, the invasive cancer cells showed a nonpolarized cytoplasmic staining of CD24.

Among TUR specimens (n = 87), 25.9% (14/54), 95.4% (21/22), and 81.8% (9/11) of stage pTa, pT1, and pT2 tumors, respectively, overexpressed CD24 and a significant association between CD24 overexpression and invasiveness was found (P < .001, Table 3). Among TUR specimens, 45.5% (10/22), 39.0% (16/41), and 75.0% (18/24) of grade 1, 2, and 3 tumors, respectively, overexpressed CD24. Overexpression of CD24 was observed significantly more frequently in grade 3 tumors compared with grade 1 or grade 2 tumors (P = .02). Among pTa lesions, there was a tendency toward increased expression of CD24 in higher grade tumors (P = .08). Among cystectomy specimens (n = 48), no significant association between CD24 overexpression and lymphovascular invasion or lymph node metastasis was found (Table 2).

In this study, we investigated the pattern of CD24 expression by immunohistochemistry in bladder UCs from TUR and cystectomy specimens. CD24 was localized in the apical cytoplasm in normal bladder epithelium at very low levels. In superficial noninvasive tumors, CD24 expression increased but the apical localization was maintained. However, some tumors showed loss of polarity in the staining, with CD24 expression detected diffusely throughout the cytoplasm. The frequency of CD24 overexpression was significantly increased in invasive carcinomas compared with noninvasive carcinomas (84.8% vs 28.6%). In 1 case from a 56-year-old man whose original report indicated no stromal invasion, an increased nonpolarized cytoplasmic staining for CD24 was detected and prompted an exhaustive search for stromal invasion. Multifocal stromal invasion was identified on re-evaluation of the specimen. Invasive cancer cells lost the apical polarity of CD24 staining and showed a nonpolarized cytoplasmic staining of CD24.

The significant increase in cytoplasmic CD24 protein expression in invasive carcinoma is consistent with the results of a previous study.10 These findings suggest that the polarized apical expression of CD24 is maintained in normal urothelium and noninvasive superficial tumors, but when progression to an invasive condition occurred, the nonpolarized expression of CD24 became prominent (Figure 4, A through D). Cytoplasmic CD24 expression might reflect an overproduction of the protein or a disruption in protein sorting or degradation within the cell. Even though the underlying biologic mechanism remains unclear, immunostaining using the anti-CD24 antibody was relatively easy to analyze and reproducible, as suggested by Kristiansen et al.10 Gromova et al17 reported several genes that were differentially expressed in noninvasive (Ta) and invasive (T2–T4) bladder UCs using an improved differential display procedure. CD24 was one of the differentially expressed genes, which is consistent with our immunohistochemical results. Thus, the extent of CD24 protein expression may be a reliable marker of the invasiveness of tumors.

Maintenance of polarity is a characteristic of highly organized normal cells. The loss of cell polarity is a characteristic feature of neoplasia, especially in cancer cells. Even though the sequence of events leading to loss of polarity and invasiveness is not clear, there was a strong correlation between the loss of apical localization of CD24 and the stromal invasion of tumor cells in our study. A recent article by Tobioka et al18 reported that carcinoembryonic antigen was expressed at the apical cell border in normal colon epithelium and in a diffuse pattern throughout the cell in colon tumors. These data suggest that, during tumor progression, the tumor cells may lose membrane polarity, reflected by the nonpolarized expression of carcinoembryonic antigen. In a recent study, we showed that loss of polarity of CD24 expression in advanced ovarian borderline tumors was correlated with the advanced stage of the tumors.19 Therefore, the nonpolarized cytoplasmic staining of CD24 may be a novel molecular signature of stromal invasion by carcinomas.

In cystectomy specimens, there was no association between high-level expression of CD24 and lymphovascular invasion or lymph node metastasis, which was unexpected. It is possible that CD24-expressing tumor cells can spread more easily because of their capacity either to form thrombi with activated platelets or to adhere to endothelial cells in the bloodstream as demonstrated for CD24-expressing breast cancer cells.20 Kristiansen et al12 reported that CD24 expression was an independent prognostic tumor marker in non–small cell lung cancer even though there was no significant association of CD24 expression with nodal status. These data suggested that lymphatic spread is a biologically different process, although it frequently precedes hematogenous dissemination. This is the most likely explanation for the lack of significant association between CD24 expression and lymph node metastasis in our study.

Several successful antibody-based immunotherapeutics have been developed after more than 2 decades of experience. To ensure specificity, the antibody must react with the target tumor cells but should not cross-react significantly with healthy tissue. Optimally, the antigen should be present on all or nearly all of the malignant cells to allow effective targeting and to prevent a subpopulation of antigen-negative cells from proliferating. Because normal bladder epithelium shows negligible expression of CD24, this protein is an excellent candidate for bladder cancer immunotherapy.

In summary, we found that CD24 is overexpressed in bladder urothelial cancers and that cytoplasmic overexpression is a marker of invasive and high-grade tumors. In superficial bladder tumors, in particular, the loss of apical localization of cytoplasmic staining was correlated with stromal invasion. Although the biologic function of CD24 and significance of the changes in expression and cellular localization is not yet clear, our results demonstrate the diagnostic and therapeutic effectiveness of using CD24 protein expression in evaluating bladder tumors.

We are very grateful to Jung Sun Lee, BS, and Mee Young Sim, BS, for expert technical assistance and to financial support from the Research Institute of Pharmaceutical Science. This work was partly supported by the Therapeutic Antibody Center, Ministry of Health & Welfare, Republic of Korea.