Role of CD10 Immunohistochemical Expression in Predicting Aggressive Behavior of Phylloides Tumors

Background: Phylloides tumors are rare breast neoplasms with a variable clinical course depending on the tumor category. Along with histologic features, the role of immunohistochemical staining has been studied in predicting their behavior. Objectives: Our aim was to evaluate the role of CD 10 immunohistochemical staining in predicting survival, recurrence and metastasis in phylloides tumor. We also evaluated correlations of other clinicopathological features with overall and disease-free survival. Materials and Methods: CD10 expression was studied in 82 phylloides tumors divided into recurrent/metastatic and non-recurrent/non-metastatic cohorts. The Chi-square test was applied to determine the significance of differences in CD10 expression between outcome cohorts. Uni and multivariate survival analyses were also performed using log-rank test and Cox regression hazard models. Results: All 3 metastatic cases, 5 out of 6 (83.3%) recurrent cases and 37out of 73 (50.7%) non-recurrent and non-metastatic cases expressed significant (2+ or 3+) staining for CD10. This expression significantly varied between outcome cohorts (p<0.03). Tumor category and histological features including mitotic count and necrosis correlated significantly with recurrence and metastasis. A significant decrease in overall and disease free survival was seen with CD10 positivity, malignant category, increased mitoses and necrosis. Neither CD10 expression nor any other clinicopathologic feature proved to be an independent prognostic indicator in multivariate analysis. Conclusions: CD10 immunohistochemical staining can be used as a predictive tool for phylloides tumor but this expression should be interpreted in conjunction with tumor category.


Introduction
Breast cancer is the most common malignancy in females globally and the incidence is higher in Pakistan as compared to neighboring countries and rest of the world (Moore et al., 2009;Shaukat et al., 2013;Asif et al., 2014). Phylloides tumor (PT) is a fibroepithelial neoplasm which characteristically exhibit proliferation of stromal component accompanied by compression of breast ducts and impart typical leaf life appearance (Noguchi et al., 1993;Tan et al., 2013). PT accounts for less than 1% of all breast neoplasm and like others, inherit the recurrent and metastatic potential (Rowell et al., 1993;Barth, 1999;Asoglu et al., 2004;Khurshid et al., 2006;Khurshid et al., 2013). Prediction of possible behavior and outcome of the disease helps oncologist to choose the most suitable treatment modality. Complete surgical resection with safe margins is the mainstay of treatment for all PTs and malignant PTs may require frequent follow-up visits and additional radiotherapy and/ or chemotherapy (Burton et al., 1989, Hawkins et al., 1992Barth, 1999;Khosravi-Shah, 2011).
On the basis of a constellation of histological features,

Role of CD10 Immunohistochemical Expression in Predicting Aggressive Behavior of Phylloides Tumors
Muhammad Usman Tariq, Saroona Haroon*, Naila Kayani these tumors are classified into benign, borderline and malignant categories (Tan et al., 2013). Malignant PTs have the highest recurrence (36-65%) and metastatic frequency (35%) and benign PTs have the lowest recurrence (8-21%) and metastatic (7%) frequency (Rowell et al., 1993;Asoglu et al., 2004). The histological features used to characterize PTs do not hold an individual predictive value. Few researchers have evaluated the role of immunohistochemistry (IHC) in predicting the recurrence and metastasis of PTs. Immunohistochemical markers such as Ki-67, p53, CD 31, CD34, CD 117, vimentin, actin, VEGF and EGFR have been evaluated for their possible predictive but none has yet demonstrated a significant role (Millar et al., 1999;Chen et al., 2000;Ortega et al., 2001;Tse et al., 2001;2003;. CD 10 is a matrix metalloprotease which plays an important role in stromal differentiation and tumor invasion. It has an established diagnostic role in a variety of tumors especially follicular lymphomas, Burkitt's lymphomas and endometrial stromal tumors (Stein et al., 1984;Gregory et al., 1987;Chu et al., 2001). Expression of CD10 has been demonstrated in a number of other non-hematopoietic neoplasms as well as in normal myoepithelial cells of breast (Chu et al., 2000;Moritani et al., 2002). CD10 has also been evaluated in different categories PTs and increased immunohistochemical expression in observed with increasing tumor category (Tse et al., 2005;Tsai et al., 2006;Al-Masri et al., 2012;Hussin et al., 2013). The role of CD10 has been evaluated in predicting possible outcome of PTs and no study has described any correlation of recurrence with CD10 immunohistochemical expresssion. Two studies have described significantly increased expression in metastatic cases (Tsai et al., 2006;Al-Masri et al., 2012). As the studies conducted so far are few in number, therefore, validation with further studies on a larger number of cases is required.
The aim our study was to evaluate the role of CD 10 immunohistochemical stain and clinicopathological features in predicting the survival, recurrence and metastasis in phylloides tumors.

Materials and Methods
The study was approved by institutional "Ethical Review Committee". We retrieved 82 cases of Phylloides Tumor from the surgical pathology database of Section of Histopathology, Aga Khan University Hospital for cases reported between January 2006 and March 2014 through "Integrated Laboratory Management System (ILMS)" software. We included the excisional biopsy, wide local excision, mastectomy and modified radical mastectomy (MRM) specimen. Trucut biopsies, incisional biopsies and blocks received (from outside) for second opinion were not included. Moreover, specimen with clear margins i.e presence of normal breast tissue around the entire periphery of tumor, were included. Verbal informed consent and follow up information regarding recurrence and metastasis was obtained from the patients via telephonic conversation on their contact numbers mentioned at the requisition slips. Pathology reports and slides of the cases were reviewed and data regarding the patient's age, and pathological features such as tumor size, tumor borders, resection margin status, stromal cellularity, stromal overgrowth, nuclear atypia, necrosis, heterologous element, mitotic counts and distance from resection margin was obtained. These cases were divided into three categories including benign, borderline, and malignant according to WHO criteria (Tan et al., 2013).
Representative block of the tumor with maximum cellularity and internal control of myoepithelial cells was selected for prospective staining with CD10 immunohistochemical staining. In recurrent cases, blocks of initial tumor were selected. Immunohistochemical staining was performed on the selected slides (as per kit manufactuer's instructions) by a technologist, utilizing commercially available monoclonal (ready to use) CD10 antibody (code 56C6, Dako) on automated immunostainer. Immunostaining was then be assessed by at least two pathologists. The percentage of stromal cells staining positive was scored from 0% to 100%. Staining intensity was scored as 0, 1+, 2+ and 3+ (no staining, weak, moderate and strong staining, respectively). IHC was considered positive for CD10 if more than 20% stromal cells exhibit moderate (+2) to strong (+3) expression (Tsai WC et al., 2006).

Statistical analysis
Pearson Chi-Square test was applied to examine the correlation of CD10 expression and clinicopathological features with recurrence and metastasis. Disease-free survival (DFS) and overall survival (OS) periods were calculated from the dates of pathologic diagnosis to the dates of recurrence or metastasis and death, respectively. Univariate survival curves were plotted using the Kaplan-Meier method, and statistical differences were determined by using the log-rank test. Multivariate analysis was performed using the stepwise backward LR Cox regression hazards model. A p value of less than .05 was considered significant.
When separately analyzed, rate of recurrence and metastasis increased with tumor category but statistical significance was not observed. However, when collectively analyzed as a cohort, combined recurrence and metastatic rate increased significantly with tumor category (p=0.021). Death rate also correlated significantly with tumor category (p=0.008). OS and DFS insignificantly decreased with tumor category.
Similarly, when CD10 expression was separately analyzed in cases with recurrence and metastasis statistical difference was not observed. However, when collective cohort of recurrent and metastatic cases was analyzed, statistical significance was observed (p=0.03) ( Table  1). We also correlated clinicpathologic and histological features with recurrence and metastasis and found positive correlation of mitotic count and necrosis with recurrence and metastasis.
Tumor categories were also showed an association with a combined increase in recurrence and metastatic rate (p=0.021). Among histological features, mitotic activity and necrosis demonstrated positive correlation with recurrence and metastasis (p=0.008 and p=0.016 respectively). Stromal atypia, stromal cellularity, tumors margins, sarcomatous component, patient's age, tumor size and clear (<1cm) margins were not significantly correlated with these adverse events (Table 2).
Overall survival (OS) and DFS (DFS) significantly varied with significant CD10 staining (p=0.04 & p=0.018), malignant tumor category (p=0.009 & p=0.02),   (Figures 2 and 3). Cases with safe margins of ≥1 cm did not have significantly increased recurrence free survival in comparison to the cases with clear but <1cm surgical margin (p=0.52). The differences in overall follow up durations and disease free durations of these factors are summarized in Table  3. When multivariate analysis was done using Cox regression hazards model neither CD10 expression nor any of the clinicopathologic features showed significance independently (not shown in table).

Discussion
CD10 immunohistochemical staining is specific to myoepithelial cells as this staining is neither shared by other stromal components nor by epithelial components (Gillette et al., 1990;Guelstein et al., 1993). Hence, it is efficiently used as internal control. In English literature, so far 4 studies have been conducted to evaluate the role of CD10 in predicting recurrence and/or metastasis of phylloides tumors. In the earliest study, Tse et al. (2005) evaluated 14 recurrent and 2 metastatic cases in a cohort of 181 phylloides tumors. He performed CD10 staining on both initial as well as second (recurrent/metastatic) cases and did not find any association of staining pattern with recurrence or metastasis. Tsai et al. (2006) selected a small cohort of 22 phylloides tumor with 6 recurrent and 3 metastatic cases. He evaluated CD10, ASMA and Vimentin immunohistochemical stains and did not report any association of these with recurrence or metastasis. Al Marsi et al. (2012), in his study of CD10 expression in 46 cases, observed metastasis in 6 cases which positively correlated with increased immunohistochemical staining. Hussin et al. (2013) also failed to prove any association of CD10 immunoexpression with recurrence in their 9 recurrent cases out of total 61 cases. Our findings are also in concordance with the other studies as we do not find association with recurrence (p=0.152) or metastasis (p=0.16) separately. However, when we combine these cases with features of adverse behavior, we find a positive correlation (p=0.03). In our opinion, both recurrence and metastasis are indicators of poor behavior and association of CD10 expression with this cohort should be interpreted as association with poor behavior i.e either recurrence or metastasis. Moreover, death rate which is another indicator of poor behavior, was also associated with increased CD10 expression (p=0.008).
Among various other immunohistochemical markers like p53,CD 31,CD34,CD 117,vimentin,actin, VEGF and EGFR which have been evaluated in phylloides tumor, only  has described a significant association of CD117 staining with recurrence. , in her another study of phylloides tumor, have evaluated the predictive role of histologic parameters in detail and found stromal atypia, stromal cellularity, tumor margins and necrosis to be significant. According to Al Marsi et al. (2012), tumor size and tumor grade was also significantly correlated with metastasis. We also analyzed clinicopathologic and histologic features and observed tumor category, mitotic activity and necrosis to correlate with recurrence and metastasis.
Complete excision of with wide margins is the mainstay of treatment for phylloides tumor. The recurrent rate for negative margins is 10% while it is 18% for excision with positive margins (31). Except for , we did not find any study which mentioned the margin status in recurrent cases at the time of initial excision. As tumor margin status is the sole predictor of recurrence, in other studies, the assessment of other features in recurrent cases could have been affected by this confounding factor. The strength of our study lies in avoiding this confounding factor by selecting cases with clear margin (at least 4mm) of clearance. When analyzed, 5 (10.9%) out of 41 cases with safe margins (>1cm) showed recurrence while only 1 (2.8%) out of 35 cases with clear but <1cm margin showed recurrence. The difference in recurrence rates of these two groups and recurrence free survival did not differed significantly (p=0.223 and p=0.52).
Although CD10 positivity, malignant tumor category, increased mitoses and necrosis showed a significant  reduction in overall survival and disease-free survival but these did not showed independent significance in multivariate analysis.
I n c o n c l u s i o n , w e c o n c l u d e t h a t C D 1 0 immunohistochemical staining can help to predict poor behavior of phylloides tumor in terms of recurrence and metastasis but this expression should be interpreted in conjunction with tumor category and other histological features especially mitotic activity and necrosis. Groups with clear tumor margins of greater than or equal/less than 1 cm do not differ significantly in their recurrence rates. The results of this study will help oncologists in predicting the possible outcome and deciding the treatment plans accordingly such as more frequent follow-up visits, radiotherapy and chemotherapy which might increase the recurrence-free and metastasis-free survival. Tse GM, Tsang AK, Putti TC, et al (2005). Stromal CD10 expression in mammary fibroadenomas and phyllodes tumours. J Clin Pathol, 58, 185-9. Tse GMK, Lui PC, Vong JS, et al (2009