Fax: (011) 81-3-5800-0281
Prognostic value of CD44 isoform expression in thymic epithelial neoplasms
Article first published online: 13 APR 2005
Copyright © 2005 American Cancer Society
Volume 103, Issue 10, pages 2015–2022, 15 May 2005
How to Cite
Sonobe, S., Miyamoto, H., Nobukawa, B., Izumi, H., Futagawa, T., Ishikawa, N., Yamazaki, A., Uekusa, T., Abe, H. and Suda, K. (2005), Prognostic value of CD44 isoform expression in thymic epithelial neoplasms. Cancer, 103: 2015–2022. doi: 10.1002/cncr.21046
- Issue published online: 28 APR 2005
- Article first published online: 13 APR 2005
- Manuscript Revised: 23 DEC 2005
- Manuscript Received: 30 SEP 2005
- Manuscript Accepted: 12 JAN 2005
- thymic epithelial tumor;
- World Health Organization histologic classification;
- Masaoka staging system
Many histologic classifications of thymic epithelial tumors have been reported to date, but to the authors' knowledge, none of them closely reflect the clinical behavior or prognosis of the tumor. Therefore, it is necessary to establish a biologic marker for thymic epithelial tumors. Variants of CD44 may be important in promoting tumor progression and metastasis. Accordingly, the expression of CD44 isoforms in thymic epithelial neoplasms was investigated using immunohistochemistry to assess their possible value as prognostic indicators.
Expression of CD44v6 in thymic epithelial tumors was investigated with immunohistochemistry using consecutive surgical specimens resected from 108 patients between 1983 and 2002 at Juntendo University Hospital in Tokyo, Japan.
Among the 108 thymic epithelial tumors, 70 were negative for CD44v6, 20 were weakly positive, and 18 were strongly positive. The status of CD44v6 expression (negative vs. weakly plus strongly positive) was found to be correlated with the tumor stage according to the Masaoka staging system (noninvasive vs. invasive tumors) (P = 0.0214). When patients with tumors that were negative and weakly positive for CD44v6 expression were combined, the 5-year, 10-year, and 15-year recurrence-free survival rates were 98.2%, 95.9%, and 86.1%, respectively, whereas the corresponding rates for patients with strongly positive tumors were 73.5%, 73.5%, and 55.1%, respectively. Therefore, these two groups demonstrated a significant difference with regard to recurrence-free survival (P = 0.0172).
CD44v6 expression in thymic epithelial neoplasms demonstrated a significant difference based on the World Health Organization classification, the Masaoka stage (invasive vs. noninvasive tumors), and recurrence, if an appropriate cutoff value was chosen in each case. This suggests that CD44v6 can be used as a marker that reflects the stage of thymic tumors. Cancer 2005. © 2005 American Cancer Society.
Thymic epithelial neoplasms are defined as tumors that arise from epithelial cells of the thymus.1 To our knowledge, although there are many histologic classifications of thymoma, none provide a useful indication of the clinical behavior or prognosis of these tumors. In contrast, the Masaoka staging system2 is useful because it is related to the prognosis. However, it is still necessary to establish a tumor marker that reflects both clinical behavior and prognosis.
CD44 is a surface molecule that mediates cell-to-cell adhesion as well as adhesion between cells and the interstitial matrix.3 This molecule has diverse functions, being involved in the activation, differentiation, migration, and proliferation of cells. CD44 is widely expressed on leukocytes, erythrocytes, fibroblasts, vascular endothelial cells, and epithelial cells, as well as in the central nervous system.3 There are various isoforms of CD44 derived from alternative splicing of 10 exons, and isoform expression depends on the cell type and extent of activation. Recent studies of CD44 have shown that its isoforms are expressed by various tumor cells and in various tissues, and also are involved in tumor metastasis and invasion. The aim of the current study was to analyze the expression of CD44 isoforms in thymic epithelial neoplasms by immunohistochemistry and to evaluate their possible role as prognostic factors for these tumors.
MATERIALS AND METHODS
The current study included 108 consecutive patients who underwent surgery for thymic epithelial neoplasms at Juntendo University Hospital in Tokyo, Japan between October 1983 and December 2002. All patients underwent surgical resection and there were no unresectable cases reported. There were 62 men and 46 women with a mean age of 52.4 years (range, 19–78 years).
According to Masaoka et al.,2 thymomas are classified as follows: Stage I: macroscopically completely encapsulated, and microscopically no capsular invasion; Stage IIa: macroscopic invasion into surrounding fatty tissue, or mediastinal pleural tissue; Stage IIb: microscopic invasion into capsule; Stage III: macroscopic invasion into neighboring organs (pericardium, great vessels, or lung); Stage IVa: pleural or pericardial dissemination; and Stage IVb: lymphogenous or hematogenous metastasis. The disease was classified as Stage I in 56 patients, Stage II in 34 patients, Stage III in 15 patients, and Stage IV in 3 patients.
According to the World Health Organization (WHO) histopathologic classification,4 thymomas are classified into two types depending on tumor cell morphology. Tumors comprised of spindle/oval cells are defined as type A, whereas those comprised of dendritic or plump/epitheliod cells are classified as type B. Type AB tumors are comprised of both cell types. Type B tumors are further classified into type B1, type B2, and type B3 depending on the extent of lymphocytic infiltration and tumor cell morphology. According to this classification, thymic carcinoma is considered a type C tumor. Of the 108 patients, the tumor was type A in 10 patients, type AB in 16 patients, type B1 in 20 patients, type B2 in 23 patients, type B3 in 35 patients, and type C in 4 patients (Table 1).
The 108 patients underwent annual evaluation by mail, telephone, or interview at the outpatient clinic of the study institution and were followed for an average of 84.7 months. In principle, all patients were followed by clinical interview and chest X-ray every 3 months and a chest computed tomography (CT) scan was obtained every year during the initial 5 years, and by clinical interview and chest X-ray every 6 months and a chest CT scan every 2 years over the subsequent 5 years, respectively. All patients with a tumor that was greater than Stage III also were followed by abdominal CT scan, bone scintigraphy, and brain magnetic resonance imaging (MRI) every 1–2 years.
Recurrence was diagnosed using radiographic imaging and/or pleural effusion cytology. Nine patients developed a recurrence, including pleural dissemination in seven patients, mediastinal lymph node recurrence in one patient, and metastasis to the liver and bone in one patient. Of the nine patients who developed a recurrence, five patients died. Only those deaths resulting from disease recurrence were included in the statistical analysis.
The surgically resected tumor specimens were fixed in 4% buffered formaldehyde and routinely processed for embedding in paraffin. Neutral-buffered formaldehyde (pH of 7.0) was obtained by adding dilute phosphate buffer. Sections were cut at a thickness of 4 μm, deparaffinized, and stained with hematoxylin and eosin (H & E). Other paraffin sections of the tumor samples were mounted on superfrost slides, deparaffinized, and subjected to the avidin-biotin complex technique (Novostain Super ABC Kit (universal), Novocastra™ Laboratories, Ltd., Newcastle-upon-Tyne, UK). Endogenous peroxidase activity was inhibited by preincubation with hydrogen peroxide, whereas nonspecific binding was blocked by incubation for 30 minutes with 3% bovine serum albumin/phosphate-buffered saline. CD44v6 was detected with a mouse monoclonal CD44v6 antibody (NCL-CD44v6, clone VFF-7; Novocastra Laboratories, Ltd.) at a dilution of 1:100. Sections were incubated with the primary antibody for 60 minutes at 37 °C. Immunoreactivity was visualized with 3,3′-diaminobenzidine and sections were counterstained with hematoxylin. Nonneoplastic epithelium in the resected specimens was used as the positive control, whereas negative control sections were created by omission of the primary antibody. All slides were evaluated without knowledge of the clinical outcome.
The surgical specimens demonstrated three patterns of staining for CD44v6. If no tumor cells were stained, the specimen was defined as negative. If only scattered tumor cells were stained or strongly stained tumor cells were localized to part of the lesion, it was defined as weakly positive. When cells were strongly and widely stained, the lesion was defined as strongly positive. If not only tumor cells but nontumor cells as well were strongly and widely stained, the lesion was defined as negative because of cross-reactive overstaining (Figs. 1–5).
Statistical analysis of the association between CD44v6 and clinicopathologic indices was performed using the Fisher exact test. The disease-free survival rate was calculated with the Kaplan–Meier method, with statistical analysis being performed with the log-rank test. A P value < 0.05 was considered to be statistically significant. StatView software (version 5.0 for PC; SAS Institute Inc., Cary, NC) was used for statistical analysis.
Of the 108 patients with thymic epithelial neoplasms studied, CD44v6 expression was negative in 70 patients, weakly positive in 20 patients, and strongly positive in 18 patients. The CD44v6 status (negative, weakly positive, and strongly positive) of the tumors is shown in relation to the WHO classification in Table 1.
When the patients were divided into two groups based on the WHO classification (A + AB + B1 + B2 vs. B3 + C), CD44v6 status (negative vs. weakly plus strongly positive) was found to be correlated with the WHO type, as shown in Table 2 (P < 0.001). The WHO classification excluding type C (A + AB + B1 + B2 vs. B3) also was found to be significantly associated with the CD44v6 expression status (negative vs. weakly plus strongly positive) (P < 0.001). The CD44v6 status (negative plus weakly positive vs. strongly positive) is also presented in correlation with the WHO classification (A + AB + B1 + B2 vs. B3 + C) in Table 3. The CD44v6 status was found to be significantly correlated with the WHO classification (P < 0.001). Even if the patients with type C disease were excluded, there was still a significant correlation found between CD44v6 status and WHO classification (P < 0.001).
|WHO classification||CD44v6(−)||CD44v6(+ and ++)|
|Type B3 and C||10||29|
|WHO classification||CD44v6(− and +)||CD44v6(++)|
|Type B3 and C||22||17|
Tumors from Masaoka Stage II onward were defined as invasive thymoma. When the CD44v6 expression status (negative vs. weakly plus strongly positive) was compared with the Masaoka stage (noninvasive vs. invasive), there was a significant correlation noted between Masaoka stage and CD44v6 status (P = 0.0214) (Table 4). There was no significant correlation noted between Masaoka stage (noninvasive vs. invasive) and the CD44v6 status (negative plus weakly positive vs. strongly positive) (P = 0.0850).
|Masaoka stage||CD44v6(−)||CD44v6(+ and ++)|
There was no significant difference noted between the patients with versus those without recurrence with regard to the CD44v6 expression status (negative vs. weakly plus strongly positive) (P = 0.2733). In the patients whose tumors had negative CD44v6 expression, the 5-year, 10-year, and 15-year recurrence-free survival rates were 97.8%, 95.0%, and 87.7%, respectively, whereas the corresponding rates for the patients with CD44v6 expression that was weakly plus strongly positive were 87.9%, 87.9%, and 62.5%, respectively. The two CD44v6 status groups were not found to be significantly different with regard to recurrence-free survival (P = 0.0680). The CD44v6 expression status (negative plus weakly positive vs. strongly positive) of those patients with and those patients without recurrence is compared in Table 5. There was a significant difference noted between the two groups (P = 0.0406). In the group of patients with CD44v6 negative plus weakly positive expression, the 5-year, 10-year, and 15-year recurrence-free survival rates were 98.2%, 95.9%, and 86.1%, respectively, whereas the corresponding rates for the patients whose tumors demonstrated strongly positive CD44v6 expression were 73.5%, 73.5%, and 55.1%, respectively. Figure 6 presents the recurrence-free survival curves for those patients with CD44v6 negative expression and weakly positive expression versus those patients with CD44v6 strongly positive expression. These groups had significantly different survival rates (P = 0.0172).
|Recurrence status||CD44v6(− and +)||CD44v6(++)|
Various indicators of the prognosis of thymic epithelial neoplasms have been reported to date, but to our knowledge the Masaoka staging system has been widely accepted as the most useful prognostic indicator. The clinical usefulness of this staging system has already been established in the literature.5–14 According to Okumura et al.,10 the 20-year survival rates of patients with Masaoka Stage I, Stage II, Stage III, Stage IVa, and Stage IVb disease were 90%, 90%, 56%, 15%, and 0%, respectively. In the case of patients with Stage I disease, cure can be anticipated if the tumor is treated with radical surgery. For patients with Stage II and more advanced disease, adjuvant therapy is recommended in addition to surgery. Although the Masaoka staging system2 appears to be a useful prognostic indicator, a biologic marker that reflects both clinical behavior and prognosis would be desirable.
In 1991, Gunthert et al.15 detected CD44v6, a variant isoform of CD44 with exon V6, in a subclone of rat pancreatic carcinoma cells that had acquired a high metabolic potential. They demonstrated that a cell line with a low metabolic potential was found to demonstrate a high metabolic potential when it overexpressed CD44v6. Subsequently, it was found that expression of this variant isoform of CD44 is associated with the metastatic potential of tumor cells in various organs.
The WHO classification for thymic epithelial neoplasms, published in 1999,4 is based on histopathologic features and its usefulness has been investigated gradually. According to Okumura et al.,16 the 20-year survival rate after surgical resection in patients with type A, type AB, type B1, type B2, and type B3 tumors was 100%, 87%, 91%, 59%, and 36%, respectively. Including deaths without tumor recurrence, patients with type A tumors demonstrated a tendency toward better survival compared with patients with type B3 tumors (P = 0.07).
To our knowledge, the extent of atypia of the epithelial cells in patients with thymoma is not clearly described in the WHO classification, therefore posing a particular problem with regard to spindle-shaped cells that demonstrate atypia. Type A thymomas are comprised mainly of spindle-shaped cells with little atypia and few lymphocytes. Atypia of spindle-shaped cells does occur in type A tumors (although it is uncommon), but it then becomes difficult to determine whether the tumor should be classified as type A or type B3. In the current study, there was no staining for CD44v6 in any of the specimens that were classified as type A according to the WHO classification. In contrast, among the 35 thymomas classified as WHO type B3, CD44v6 expression was negative in 9 patients (25.7%), weakly positive in 12 patients (34.3%), and strongly positive in 14 patients (40.0%). When tumors with weakly positive and strongly positive CD44v6 expression were combined, 74.3% of WHO type B3 tumors were positive for CD44v6. Okumura et al. also stated that there was a significant difference between patients with type B1 and type B3 tumors (P = 0.001), as well as between patients with type B2 and type B3 tumors (P = 0.04), so that the distinction between type B3 tumors and type B1 or type B2 tumors is significant. In the current study, although CD44v6 was rarely found to be expressed positively in spindle-shaped cells, it was positively expressed in dendritic cells. Therefore, this histochemical method may be used to distinguish type A from type B3 tumors and type B3 tumors, although H & E staining is often unable to make this distinction. With regard to the association between CD44v6 expression and invasive growth, CD44v6 expression status (negative vs. weakly plus strongly positive) was found to be correlated with Masaoka stage (noninvasive vs. invasive) (P = 0.0214). However, when the tumors with negative plus weakly positive CD44v6 expression were separated from those with strongly positive expression, there was no correlation noted with the Masaoka stage (noninvasive vs. invasive) (P = 0.0680).
The association between various tumor features and the intensity of CD44v6 expression was not evaluated in the current study. However, strongly positive tumors were found to be significantly more likely to metastasize than tumors with weakly positive or negative expression. This finding suggests that adequate postoperative follow-up is necessary if CD44v6 expression is strongly positive in a tumor. The intensity of CD44v6 expression should be studied further in patients with invasive thymoma and thymic carcinoma.
In the current study, CD44v6 expression demonstrated a significant difference between tumors with regard to WHO type, invasive potential, and risk of recurrence, if an appropriate level of CD44v6 positivity was chosen in each case. Consequently, CD44v6 expression appears to be a useful marker that reflects the disease stage of thymic epithelial neoplasms. In the case of patients with tumors with strongly positive CD44v6 expression, the follow-up interval should be shortened and additional radiographic imaging should be performed.
- 1Tumors of the thymus. Atlas of tumor pathology. 2nd series. Fascicle 13. Washington, DC: Armed Forces Institute of Pathology, 1976., .
- 4Histological typing of tumors of the thymus. International histological classification of tumors. 2nd edition. New York: Springer-Verlag, 1999., .