β-Catenin Expression Negatively Correlates with WIF1 and Predicts Poor Clinical Outcomes in Patients with Cervical Cancer

Aberrant activation of the canonical Wnt pathway plays a significant role in cervical cancer (CC). However, limited data show the correlation between the cancer clinicopathological characteristics and the key molecules such as β-catenin and Wnt inhibitory factor 1 (WIF1). In this study, β-catenin and WIF1 expression were analyzed by immunohistochemistry for 196 patients with CC, 39 with cervical intraepithelial neoplasia (CIN), and 41 with normal cervical epithelium (NCE). Significant overexpression of β-catenin was detected in CC (67.9%) when compared to CIN (43.6%) or NCE (34.1%), p < 0.01, while low WIF1 expression was detected in CC (24.0%) when compared to CIN (59.0%) or NCE (58.5%), p < 0.001. Negative correlation was shown between β-catenin and WIF1 expression (r = −0.637, p < 0.001). In addition, multivariate analysis revealed that both lymph node metastasis and β-catenin expression were the independent prognostic factors not only for disease-free survival (HR = 5.029, p < 0.001; HR = 2.588, p = 0.035, resp.), but also for overall survival (HR = 5.058, p < 0.001; HR = 2.873, p = 0.031, resp.). Our findings indicate that, besides lymph node metastasis, β-catenin expression may also be a poor prognostic factor for CC while WIF1 could be a potential drug target for treatment of advanced CC.


Introduction
Cervical cancer (CC) is the fourth most common cancer in women. According to existing data, there were 528,000 registered new cases and 266,000 deaths only in year 2012 [1]. Most patients with early CC have good prognosis. By contrast, patients with a later cancer stage or metastatic CC have poor survival rate because of less effective treatments available [2]. Therefore, additional studies on late cancer development and prognosis methods are necessary.
Aberrant activation of the Wingless-type (Wnt)/ -catenin (canonical Wnt pathway) is a very common pathway in human CC [3]. Recent molecular testing has demonstrated that the CC biological behavior may arise as a multistep gene process. Specifically, infection with human papillomavirus (HPV) could be "the first hit" [4], while the dysregulation of canonical Wnt pathway may be required as "the second hit" in cervical oncogenesis [3,5,6]. However, the mechanism involving Wnt pathway in CC is still not well understood and requires additional studies. In the canonical Wnt pathway, the binding of Wnts to a heterodimeric receptor complex stabilizes the -catenin expression and leads to the activation of -catenin target genes inside the nucleus [7,8]. Furthermore, Wnt inhibitory factor 1 (WIF1) is an upstream secreted Wnt antagonist, first identified as highly conserved gene in the human retina [9]. WIF1 main function is to bind the extracellular Wnt ligands [10], disturbing the Wnt interaction with the receptors [7] and 2 BioMed Research International consequently leading to -catenin degradation, therefore inhibiting the canonical Wnt pathway. Currently, there are dozens of studies regarding the correlation with -catenin and various types of cancer [7,11]. In most of these cases, elevated levels of -catenin have been strongly correlated with poor cancer prognosis. However, so far, there have been limited reports focusing on the association between -catenin, WIF1, and the clinicopathological characteristics in CC.
In the present study, we investigated the association between the canonical Wnt pathway ( -catenin as the hallmark [7]), WIF1, and clinicopathological features of 196 patients with CC and analyzed their prognostic value in CC.

Patients and Tissue Samples. Between December 2002
and October 2007, 196 patients from the Department of Gynecologic Oncology (Sun Yat-Sen Memorial Hospital, Canton, China), diagnosed with CC (stages IA-IIB) and submitted to curative surgical resection, were recruited for this study. Cases were restricted to those who did not receive any tumor related treatment before surgery. The clinicopathological characteristics were summarized in Table 3. Furthermore, an additional 39 cervical intraepithelial neoplasia (CIN) and 41 normal cervical epithelium (NCE) samples were obtained from women undergoing hysterectomy for noncancerous diagnosis. Histological diagnosis and tumor stage and grade were determined according to the World Health Organization (WHO) and the International Federation of Gynecology and Obstetrics (FIGO) staging systems [12,13]. All specimens were anonymously coded in accordance with local ethical guidelines (as stipulated by the Declaration of Helsinki). All study protocols were approved by the University Ethics Committee.

Immunohistochemistry (IHC) and Samples Evaluation.
According to the previously described method [14,15], IHC analysis was performed using anti--catenin antibody (CST, #9562, Boston, MA, America) and anti-WIF1 antibody-N-terminal (Abcam, ab71204, Cambridge, MA, America). Human breast tissues were used as positive controls; negative controls were obtained by replacing the primary antibodies with phosphate buffered saline. The staining of -catenin was evaluated according to Maruyama's method [16]. If more than 10% of cancer cells were positively stained for cytoplasm and/or nuclei, the cells were regarded as -cateninpositive expression. By contrast, membrane staining was only classified as -catenin-negative expression; for WIF1 protein expression, nuclear staining was considered to be negative [17]. Finally, if more than 10% of cancer cells were positively stained for WIF1 in cytoplasmic and/or on cell membranes, the positive IHC results were recorded.

Statistical Analysis.
All the statistical analysis was conducted using SPSS software, version 13.0 (SPSS Inc., Chicago, IL, USA). Associations between the clinicopathological characteristics and the pattern of WIF1 and -catenin expression were examined using Pearson's 2 test. Survival rates were calculated using the Kaplan-Meier method and compared using the log-rank test. Univariate and multivariate survival analyses were performed using the Cox regression model for DFS and OS. A forward stepwise procedure was used to identify independent variables in the multivariate analysis.

IHC of -Catenin and WIF1 in Representative Cases
with CC. Images of WIF1 and -catenin staining for representative CC cases were shown in Figure 2, which indicated that, with the decreased expression of WIF1, clearly increased expression of -catenin was observed. Specifically,

CC
-Catenin the expression of -catenin was usually absent in patients who had the most intense cytoplasmic WIF1 staining while the staining of -catenin was intense in those who had no cytoplasmic expression of WIF1.

Special WIF1 Staining Pattern of CC Cells with Mitotic
Figures. WIF1 was usually found to be moderate-to-strong nuclear staining but without cytoplasmic staining in this study. However, as shown in Figure 3, another pattern of WIF1 staining was observed in mitotic figures in some CC samples, which were characterized by blue-staining (hematoxylin stain) nuclei and brown-staining (anti-WIF1 stain) cytoplasm and were registered for the first time. The individual mitotic figures could present pleomorphic appearances, such as

The Expression of WIF1 and -Catenin and Their
Association with the Clinicopathological Features of CC. As shown in Table 3, -catenin-positive expression was associated with a higher rate of lymphovascular space invasion ( = 0.017). Furthermore, WIF1 positive staining was associated with less   cervical stromal invasion ( = 0.002) and a lower rate of lymphovascular space invasion ( = 0.035).  Figure 4(b)) and OS (93.6% versus 74.5%, = 0.007; Figure 4(d)) than those with negative WIF1 results.

Discussion
Dysregulation of Wnt pathway is involved in different diseases, including cancer. -Catenin, the key factor of canonical Wnt pathway, conducts Wnt signals to the nucleus and upregulates oncogenes during tumorigenesis [7,8]. It has been demonstrated that the expression of -catenin can be upregulated in various cancers, including CC [3,5,18]. Consistent with previous reports [19][20][21], -catenin accumulation inside the cytoplasm was found to be significantly increased in CC when compared with CIN and NCE in this study. In addition, by Cox regression analysis, it was shown thatcatenin-positive expression was significantly correlated with poor prognosis in 5-year DFS and 5-year OS. In the present study, a negative correlation between cytoplasmic/nuclearcatenin accumulation and cytoplasmic WIF1 immunostaining (positive expression) was found. In addition, decreased WIF1 expression in CC was consistent with some previous studies which were based on human tumor study, such as gastrointestinal tract, kidney, glioblastoma, osteosarcoma, lung, pituitary, bladder, and oral cavity [17,[22][23][24][25][26][27][28]. Therefore, we hypothesize that the canonical Wnt pathway was activated, whereas the Wnt antagonist WIF1 was inhibited, by the multistep gene process in CC.
Our results have suggested that WIF1 expression was negative in 94.7% of CC samples with -catenin-positive expression ( Table 2), while the results in Figure 2 support the association between downregulation of WIF1 and upregulation of -catenin expression in CC. This may indicate that inactivation of WIF1 and accumulation of stabilized -catenin are a gradual process during tumorigenesis and progression of CC. Our study showed that the positive staining of WIF1 was significantly reduced in patients with >1/2 cervical stromal invasion and lymphovascular space invasion, while positive staining of -catenin was associated with lymphovascular space invasion. Furthermore, though no statistical significance was registered, the expression of WIF1 was much lower in the patients with surgical margin involvement (14.3%) and lymph node metastasis (14.8%) than in those with no surgical margin involvement (24.3%) or no lymph node metastasis (27.5%; Table 3). These results stand in support of the idea that WIF1 downregulation is an early event [6] and can potentially inhibit the early progression stage of CC by antagonizing canonical Wnt pathway.
Recurrence in many advanced cancers has been associated with chemoresistance. The related mechanisms include tumor angiogenesis, maintenance of resistant cancer stem cells, dysregulation of cell cycle, and defects in apoptosis, which are all at least partly regulated by the canonical Wnt pathway [29]. The expressions of components related to this pathway are frequently altered. Increased expression of Wnt ligands was reported in breast cancer [30] and dishevelled in cervical cancer [31], while decreased expression of dickkopf-1 was reported in pancreatic cancer [32], secreted frizzledrelated protein 1 (SFRP1) in lung cancer [33], and WIF1 in cervical cancer [6]. The potential therapeutic targets related to these compounds in cancer have been investigated through successful preclinical studies [34]. Therefore, just as in some other carcinomas [22,[35][36][37], WIF1 is a potent drug target in CC treatment.
Although the cytoplasmic WIF1 staining pattern was consistent with previous studies [6,14,15], the moderateto-strong brown nuclear staining of WIF1 was common in our study. In some CC cases, mitotic figures were cytoplasmpositive and nuclei-negative staining (Figure 3), which made them easily identifiable compared to nonmitotic cancer cells.
These results indicate that the changes of the localization of WIF1 expression may be related to the uncontrolled CC cell division. However, at present, the reason of opposite cytoplasmic/nuclear WIF1 staining patterns between mitotic figures and nonmitotic CC cells is unclear. Lack of study on the mechanisms and functions of the translocation of WIF1 between mitotic figures and nonmitotic CC cells is a limitation of our study. Further investigations for the reason are needed.

Conclusions
Our results have demonstrated the upregulation of -catenin and downregulation of WIF1 in CC samples compared to CIN and NCE. Along with clinicopathological characteristics, such as lymph node metastasis and cervical stromal invasion, increased -catenin expression has been shown to be a poor prognostic factor for CC, indicating the aberrant activation of canonical Wnt pathway. Moreover, the WIF1 staining pattern in mitotic figures (IHC, cytoplasm-positive and nucleus-negative) was opposite to most other CC cells in this study, and the reason needs to be explored in the future.

Ethical Approval
This study was approved by the Ethics Committee of Sun Yat-Sen Memorial Hospital (Permit no. 2014-63) and all aspects of the study comply with the Declaration of Helsinki.

Consent
Each patient (or her next of kin in case the patient is dead) who was enrolled in this study has signed informed consent to participate in this research.