Abstract
Purpose: CD44 is a multifunctional cell surface receptor with many known splice variants, some of which have been reported to play a role in tumor progression. The purpose of this study was to evaluate the prognostic significance of CD44 isoforms in early-stage, lymph node–negative invasive breast carcinoma.
Experimental Design: Immunohistochemical staining for CD44 isoforms was done on archival paraffin tissue sections of invasive breast carcinoma from a cohort of lymph node–negative patients who received no adjuvant tamoxifen or chemotherapy and who had a mean clinical follow-up period of 15 years. Immunohistochemical staining was done with antibodies to CD44s, the standard isoform of CD44, and to isoforms containing variant exon 6 (CD44v6); levels of staining were correlated with clinical outcome data.
Results: There was a trend towards increased disease-free survival for patients whose tumors had high anti-CD44s positivity (P = 0.05), and a significant association was observed between anti-CD44s positivity and disease-related survival (P = 0.04). Expression of CD44v6 isoforms did not correlate with clinical outcome.
Conclusion: CD44 expression, as assessed by immunohistochemical staining with anti-CD44s, may be a favorable prognostic factor in patients with node-negative invasive breast carcinoma.
- CD44s
- CD44v6
- breast carcinoma
- prognostic marker
CD44 glycoproteins comprise a family of cell surface receptors that recognize hyaluronan, a component of the extracellular matrix, as their principal ligand (1–6). The standard isoform, CD44s, is expressed in a wide range of normal tissues (7). There are multiple splice variants of CD44 and these variants generally have more restricted expression (8). Up-regulation of one or more CD44 variants has been postulated to play a role in the progression of some tumors (1, 4, 9). CD44 isoforms containing variant exon 6 (CD44v6) are the variants that have been studied most extensively. CD44v6 isoforms have been shown to mediate metastasis of pancreatic and breast carcinoma cell lines (9), and their expression has been associated with adverse clinical outcome in non-Hodgkin's lymphoma (10) and colorectal carcinoma (11).
In breast tissue, both CD44s and CD44v6 isoforms are highly expressed in the myoepithelial cell layer of ducts and lobules, but variant isoforms are absent from the luminal epithelium (12–14). As most invasive breast carcinomas have a phenotype more similar to luminal than myoepithelial cells (15), high expression of CD44v6 in invasive breast carcinoma would indicate overexpression of this cell surface receptor compared with the normal epithelial counterpart of these tumors.
The prognostic significance of CD44 expression in breast cancer is unclear, in part because most previous studies have included both node-positive and node-negative patients who received different treatments (13, 14, 16–24). In this study, we examined the immunohistochemical expression of CD44 isoforms using antibodies directed against CD44s and CD44v6 in a cohort of lymph node–negative patients with invasive breast carcinoma who received no adjuvant tamoxifen or chemotherapy and who had long clinical follow-up, and we correlated expression levels of these markers with disease-free and disease-related survival.
Materials and Methods
This study was approved by the M.D. Anderson Institutional Review Board as Protocol LAB02-426. Tissue specimens from patients with lymph node–negative invasive breast carcinoma who had surgery done at the M.D. Anderson Cancer Center between July of 1978 and October of 1995 were included in the study if the patients received no adjuvant tamoxifen or chemotherapy and if archival tissue blocks and clinical follow-up information were available. Clinical data were obtained from a database in the M.D. Anderson Cancer Center Department of Breast Medical Oncology. Formalin-fixed, paraffin-embedded tissue blocks were identified in the M.D. Anderson Cancer Center surgical pathology archives for 230 patients meeting the study criteria.
Immunohistochemical staining was done with monoclonal antibodies against CD44s (clone SFF-2, Chemicon International, Temecula, CA) and CD44v6 (clone 2F10, R&D Systems, Minneapolis, MN) following heat-induced antigen retrieval using a standard immunoperoxidase assay with diaminobenzidine chromogen. The commercial antibody to CD44v6 was prepared by the manufacturer using the entire recombinant human CD44s protein as the immunogen. The commercial antibody to CD44v6 was prepared by the manufacturer using a human chimeric fusion protein CD44v3-10-Fc as the immunogen, and specificity for isoforms containing variant exon 6 was determined by fluorescence-activated cell sorting analysis on a panel of COS cells transfected with various CD44 variant isoforms (7). Immunohistochemical staining was also done using monoclonal antibodies against estrogen receptor (ER, clone 6F11, Novocastra, Newcastle upon Tyne, United Kingdom), progesterone receptor (clone 1A6, Neomarkers, Fremont, CA), and HER-2/neu (clone e2-4001, Neomarkers). Immunohistochemical staining using anti-CD44s and anti-CD44v6 was scored as 3+ (high) if at least 10% of cells had strong cytoplasmic or membranous staining. Staining was scored as 2+ (moderate) if at least 10% of cells had moderate cytoplasmic or membranous staining. Cases without moderate or strong staining in at least 10% of cells were scored as 1+ (low). Staining for HER-2/neu was interpreted as positive for overexpression if at least 10% of tumor cells had strong complete membranous staining, and tumors with at least 10% nuclear staining for ER were interpreted as positive.
The Cochran-Armitage trend test was used to test for trends in binomial proportions across levels of a single factor or covariate. The Kruskal-Wallis rank sum test was used to compare groups of independent continuous variables. Disease-free survival was defined as the interval from the date of surgery to the date of disease recurrence or death from breast cancer or to the last follow-up date. Disease-related survival was defined as the interval from the date of surgery to the date of death from breast cancer or to the last follow-up date. Patients who died from causes other than breast cancer were censored when disease-related survival was considered. Survival was estimated using the Kaplan-Meier product limit method. The two-sided log-rank test was used to test the association between particular factors and survival. Multivariate analysis was done using the Cox proportional hazards regression model. Statistical analyses were carried out using SAS 8.02 (SAS, Cary, NC) and S-plus version 6 (Insightful Co., Seattle, WA).
Results
For the 230 patients included in the study, the median patient age was 58 years (range, 28-83 years). Eighty-one percent of patients were postmenopausal. Eighty-two percent were White, 8% were Hispanic, 7% were Black, and 3% were Asian. The surgical procedure for 25% of patients was lumpectomy with axillary dissection and 75% had a modified radical mastectomy. Forty-one percent of patients received adjuvant radiation therapy, including 90% of patients who underwent lumpectomy. None received adjuvant tamoxifen and none received chemotherapy. There were 139 T1, 88 T2, and 3 T3 tumors. Thirty-two tumors were 1.0 cm in size or smaller. The median tumor size was 2.0 cm (range, 0.6-7.0 cm). Histologic subtypes included 91% invasive ductal carcinomas, 6% invasive lobular carcinomas, and 3% other types. Seventy-one percent of tumors were ER positive and 32% had HER-2/neu overexpression. The median clinical follow-up was 15 years.
There was disease recurrence in 63 patients during the 15-year follow-up period. A total of 90 patients died and 39 deaths were attributed to breast cancer. For the entire group, the estimated disease-free survival (survival without tumor recurrence) at 5 and 10 years was 80% and 74%, respectively. The estimated disease-related survival (noncancer-related deaths censored) at 5 and 10 years was 94% and 86%, respectively.
For some small tumors, there was insufficient tumor cellularity on the recut paraffin sections for satisfactory immunohistochemical analysis of one or more markers. Two hundred and thirteen tumors were technically satisfactory for evaluation of anti-CD44s staining. Two hundred and six tumors were satisfactory for evaluation of CD44v6 expression and staining with both antibodies could be evaluated in 202 cases. The luminal cell layer of normal breast epithelium adjacent to tumor consistently stained with anti-CD44s but was negative for CD44v6 expression. The normal myoepithelium consistently stained with both antibodies. Sixty percent of tumors had either moderate or high anti-CD44s positivity and 42% had either moderate or high expression of CD44v6. In most cases with moderate to high anti-CD44s positivity, the staining pattern was cytoplasmic (Fig. 1A), although 31% of tumors with moderate to high anti-CD44s positivity also had distinct membranous staining in at least 10% of the tumor (Fig. 1B). In contrast, staining for CD44v6 had a membranous staining pattern in 94% of the cases with moderate to high expression of CD44v6 (Fig. 1C).
Cytoplasmic (A) and membranous (B) immunohistochemical staining patterns with anti-CD44s, and membranous (C) immunohistochemical staining pattern with anti-CD44v6 on formalin-fixed, paraffin-embedded tissue sections of invasive breast carcinoma (diaminobenzidine chromogen; magnification, ×200).
No clear association between anti-CD44s positivity and CD44v6 expression was evident. A significant association was observed between anti-CD44s positivity and ER expression (P = 0.01, Table 1). Staining with anti-CD44s was not significantly associated with any of the other variables evaluated.
Correlation between anti-CD44s and anti-CD44v6 positivity and clinicopathologic characteristics
As shown in Table 2, standard pathologic features associated with disease-free survival were tumor size (P = 0.04) and the presence of lymphovascular invasion (P = 0.03). The only standard pathologic feature found associated with disease-related survival in univariate analysis was tumor size (P = 0.04; Table 2). In univariate analysis, there was a trend towards increased disease-free survival for patients whose tumors were strongly positive with the anti-CD44s antibody (P = 0.05; Table 2; Fig. 2A). Disease-free survival at 5 and 10 years was 86% and 83%, respectively, for patients whose tumors had high anti-CD44s positivity compared with 74% and 67%, respectively, for patients whose tumors had low expression. By multivariate analysis (Table 3), the trend towards increased disease-free survival for patients whose tumors had high anti-CD44s positivity was independent of tumor size and lymphovascular invasion, which were found to correlate with disease-free survival in univariate analysis.
Disease-free survival and disease-related survival by clinicopathologic characteristics
Kaplan-Meier disease-free survival (A) and disease-related survival (B) curves for patients with node-negative invasive breast carcinoma in relation to anti-CD44s positivity.
Cox proportional hazards model for disease-free survival and disease-related survival
A significant association was found between anti-CD44s positivity and disease-related survival (P = 0.04; Table 2; Fig. 2B). Both 5- and 10-year disease-related survival were 100% for patients whose tumors had high anti-CD44s positivity, whereas 5- and 10-year disease-related survival were 92% and 81%, respectively, for patients whose tumors had low expression. Anti-CD44s positivity correlated better with disease-related survival than ER status, which had no significant correlation with either disease-related or disease-free survival in this patient cohort (Table 2). Although the P obtained in multivariate analysis (Table 3) might imply that the association between anti-CD44s positivity and disease-related survival is independent of tumor size, the only other variable found to correlate with disease-related survival in univariate analysis, multivariate analysis using the Cox proportional hazards model is probably not meaningful when one of the categories includes only one event (only one death for the high CD44s category).
No significant association was observed between CD44v6 expression and 5- or 10-year disease-free or disease-related survival (Table 2; Fig. 3A and B).
Kaplan-Meier disease-free survival (A) and disease-related survival (B) curves for patients with node-negative invasive breast carcinoma in relation to anti-CD44v6 positivity.
Discussion
CD44 is known to play a role in the migration of some nonneoplastic cells, particularly lymphocytes (25–27). CD44 interacts with mucosal addressin, a protein present on the high endothelial cells of Peyer's patches and lymph nodes, and this interaction facilitates lymphocyte homing to these sites (26). CD44 binding to its principal ligand, hyaluronan, not only mediates cell adhesion to this abundant extracellular matrix glycosaminoglycan but also allows cells to internalize and degrade hyaluronan (28–30). This process may contribute to tissue remodeling and may stimulate cell proliferation and migration (4). It seems that the ability of CD44 to bind hyaluronan is reduced by the presence of its variant exons (31). It has been postulated that up-regulation of one or more CD44 variants may alter the normal regulation of cell proliferation and migration and lead to progression of some types of tumors (4).
The variant isoforms of CD44 are formed by alternatively splicing one or more variant exons between standard exons 5 and 6 (1–3, 6). The variant CD44 proteins contain an additional amino acid sequence in the extracellular domain between the transmembrane region and the hyaluronan-binding region. The variant proteins recognized by monoclonal anti-CD44v6 contain variant exon 6 but may also contain one or more other variant exons. Variant exon 6 promotes metastasis of rat pancreatic and breast carcinoma cells (9), and CD44v6 protein expression is reported to be an adverse prognostic factor in human non-Hodgkin's lymphoma (10) and colorectal carcinoma (11). In breast cancer patients, increased serum-soluble CD44v6 levels have been found to correlate significantly with tumor stage, lymph node status, and the presence of metastasis (32). This suggests that CD44v6 isoforms may play a role in the progression of invasive breast carcinoma.
Previous studies using immunohistochemistry have reported a wide range of anti-CD44s positivity in invasive breast carcinoma, from 26% to 62% (13, 17–19). In our study, 60% of cases had moderate to strong staining with anti-CD44s. Differences in scoring criteria between studies will account for some differences in positivity thresholds, and the use of different antibodies and antigen-retrieval methods will also affect overall rates of positivity. However, tumors from node-positive patients are reported to have decreased staining with anti-CD44s compared with tumors from node-negative patients (19). Because our study included only node-negative patients, we would expect a relatively high rate of anti-CD44s positivity.
It should be noted that the commercial antibody to CD44s used in this study was prepared using recombinant human CD44s as the immunogen. Because CD44 variants contain the standard exons of CD44 in addition to one or more alternatively spliced variant exons, the antibody to CD44s would be expected to recognize not only CD44s but also the variant isoforms, whereas the antibody to CD44v6 has been reported to stain only those isoforms with variant exon 6. Although one would expect CD44v6-positive tumors to stain with the anti-CD44s antibody also, some tumors in our study had strong staining for CD44v6 but were negative using the antibody to CD44s. This finding has also been reported by other investigators (19, 34). Glycosylation has been shown to affect the reactivity of some anti-CD44s antibodies (35), and post-translational modifications of some variants may conceal epitopes recognized by antibodies to CD44s (4, 19). As some protein sequences encoded by variant exons are known to interfere with the hyaluronan binding site, these sequences may likewise conceal epitopes recognized by anti-CD44s antibodies (4). Moreover, regulated shedding of a portion of the extracellular domain of CD44s is known to occur, and this may result in the production of some CD44v6 variants that lack particular CD44s epitopes (6, 36).
Most previous studies have reported no independent prognostic significance for immunohistochemical staining with either anti-CD44s or anti-CD44v6 in invasive breast carcinoma, but most of these studies did not control for lymph node status or differences in treatment (13, 16–18, 20–23, 33, 37). The study by Joensuu et al. reported decreased disease-related survival for patients whose tumors stained with anti-CD44s, but when subdivided into node-positive and node-negative cases, significantly decreased survival was observed only in the node-positive cases (33). No independent prognostic significance was maintained in multivariate analysis. Umemura et al. evaluated only node-negative patients and found increased tumor recurrence for patients whose tumors expressed CD44v6 isoforms, but some patients were treated with chemotherapy, and variable treatment was not taken into account in multivariate analysis (20).
The only study with a patient cohort comparable to ours is that of Foekens et al., which included 237 node-negative patients who did not receive hormonal therapy or chemotherapy (14). These authors looked at immunohistochemical expression of CD44v6 isoforms but did not evaluate anti-CD44s positivity. They found no association between CD44v6 protein expression and overall survival but observed increased relapse-free survival for those patients whose tumors expressed CD44v6 isoforms. Margin status and radiation therapy following lumpectomy are known to affect local recurrence rates, and it is unclear whether the association between CD44v6 protein expression and decreased tumor recurrence observed in their study was independent of these important variables.
In our study, we found an association between anti-CD44s positivity and disease-related survival in node-negative patients with invasive breast carcinoma who received no hormonal therapy or chemotherapy. Our results confirm a recent smaller study of CD44s mRNA expression in invasive breast carcinoma by Berner et al., who reported a significant correlation between CD44s mRNA expression by reverse transcription-PCR analysis and both disease-free and overall survival (17). Our results suggest that CD44s protein expression (or expression of a variant recognized with the anti-CD44s antibody) may be a favorable prognostic factor in node-negative invasive breast carcinoma.
We noted a trend towards an inverse relationship between anti-CD44s positivity and HER-2/neu overexpression, and we observed a significant association between anti-CD44s staining and ER positivity. Other studies have found an association between CD44 variant expression in invasive breast carcinoma and ER positivity (18). CD44v6 isoforms seem up-regulated during puberty and down-regulated during lactation and involution, suggesting that CD44 may be regulated in part by steroid hormones (38). Although we observed an association between anti-CD44s positivity and ER expression, staining with anti-CD44s in our node-negative patient cohort correlated better with patient survival than ER status.
It is important to note that our study is limited by the relatively small number of events that occurred during the follow-up period, which was expected for this study on early-stage breast cancer. Although the P for the association between anti-CD44s positivity and disease-related survival is 0.05 (the probability of a false-positive association is 0.05%), the percentage of true positives likely to be detected is decreased in studies with a small number of events. Therefore, larger multicenter trials will be necessary to confirm the findings of this study (39).
Our findings, nevertheless, do lend support to the hypothesis that down-regulation of a CD44 isoform recognized with the anti-CD44s antibody may lead to progression of invasive breast carcinoma. Similar findings have been reported for neuroblastoma and gastrointestinal stromal tumors (40–42). This may be accompanied by up-regulation of one or more CD44 variants or other metastasis-promoting molecules. Although it is unclear at present which variant isoforms of CD44, if any, may be important prognostically in invasive breast carcinoma, we found no prognostic significance for CD44 isoforms containing variant exon 6.
Footnotes
-
Grant support: National Cancer Institute Cancer Center support grant CA16672.
-
The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked advertisement in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.
- Accepted February 11, 2005.
- Received October 27, 2004.
- Revision received January 19, 2005.
References
Volume 11, Issue 9
