Putative stem cell markers in cervical squamous cell carcinoma are correlated with poor clinical outcome

Background The aim of this study was to elucidate the value of putative cancer stem cell markers Musashi-1, ALDH1, Sox2, and CD49f in predicting the prognosis in cervical squamous cell carcinoma (CSCC). Methods Real-time PCR and immunohistochemistry staining was performed to examine Musashi-1, ALDH1, Sox2, and CD49f expression in archived specimens of CSCC patients with postoperative chemotherapy. Kaplan–Meier analysis and Cox proportional hazards model were used to assess the prognostic impact of CSC markers for overall survival (OS) and recurrent-free survival (RFS). Results The Real-time PCR data showed that the expression of all markers were increased in CSCC tissues compared with in paired normal cervical tissues (P < 0.05). The IHC result showed that high expression of Msi1, ALDH1, Sox2, and CD49f was found in 25.7 %, 43.0 %, 62.0 % and 29.0 % CSCC samples, respectively. Moreover, high expression of Msi1 (P = 0.033 and P = 0.003, respectively), ALDH1 (P = 0.015 and P = 0.002, respectively), and Sox2 (P = 0.005 and P = 0.003, respectively), and low expression of CD49f (P = 0.027 and P = 0.025, respectively) were correlated with poor OS and PFS in CSCC patients. Interestingly, tumors with Msi1high/CD49flow expression had the poorest prognosis according to Msi1/CD49f stratification. In multivariate Cox regression analysis, Sox2 expression (P = 0.047 and P = 0.018, respectively), ALDH1 expression (P = 0.013 and P = 0.003, respectively), and CD49f expression (P = 0.008 and P = 0.003, respectively) were independent prognostic markers for both OS and RFS. Conclusions Our results suggest that cancer stem cell markers are linked with poor prognosis of CSCC patients. Electronic supplementary material The online version of this article (doi:10.1186/s12885-015-1826-4) contains supplementary material, which is available to authorized users.


Background
Cervical cancer is the second most common gynecologic cancer in developing countries, with over 500,000 new cases and 274,000 deaths annually [1]. Despite advances in surgical capabilities and chemotherapy strategies, a substantial proportion of patients still die from recurrent or chemoresistant disease. Approximately 30-50 % of advanced-stage patients will develop recurrent disease [2]. Even in patients with early-stage disease, 10-20 % will recur locoregionally or have distant metastases following treatment. The reported median overall survival time of patients who recur after radical surgery or radiotherapy are between 7 and 12 months [3]. Adjuvant platinumbased drugs are standard chemotherapy treatment for cervical cancer. Meanwhile, the duration for response to chemotherapy in patients with recurrent diseases remains disappointing.
Currently, there is a lack of biomarkers for predicting the response to chemotherapy in cervical cancer patients.
Cancer stem cells (CSCs) are defined as a small population of cells within a tumor that can self-renew and drive tumorigenesis. Increasing evidence has suggested that CSCs are naturally resistant to chemotherapeutic agents and might be responsible for tumor recurrence following chemotherapy [4]. The potential mechanisms of CSC resistance to chemotherapy include slow cell cycle kinetics, overexpression of DNA repair proteins and multidrug resistance transporters, and protection by hypoxic niches [5].
It is postulated that CSCs may originate from normal stem cells and express the same cell markers as normal stem cells. Such a model has been proposed for cervical CSC markers as well [6]. To date, several potential cervical epithelial stem cell markers including Musashi-1 (Msi1), ALDH1, SOX2, and CD49f have been used to identify cervical CSCs (CCSCs). Musashi-1 is a RNA-binding protein to regulate the proliferation of multipotential stem/progenitor cells and the proliferative activity of tumor cells [7]. This protein may also have roles in cervical carcinogenesis [8]. ALDH1 is a polymorphic enzyme involved in oxidation of aldehydes to carboxylic acids. It has been reported to enhance the self-renewal and differentiation potentials of cervical cancer cells [9]. Sox2 is essential for the formation of many different tissues and organs during embryonic development. Sox2-positive population of cervical cancer cells show characteristics of tumor-initiating cells. Its expression increases the expression of CSC markers, the potential to form tumor spheres, and the tumor initiating capacity of cervical cancer cells [10]. CD49f gene encodes an integral cell-surface protein that has been proposed to play a role in cell adhesion as well as in cellsurface mediated signaling. CD49f has been used as an epithelial stem cell marker and was recently found highly expressed on cervical cancer-initiating cells [11].
As CSCs are thought to be responsible for tumor chemoresistance and recurrence, the evaluation of CSC markers expression in cervical cancer could highlights the mechanisms underlying cervical cancer progression and recurrence. Moreover, to determine the association of CSC markers expression to cervical cancer chemoresponsiveness could help in the development of targeted agents to treat chemoresistant disease. The aims of this work were therefore to evaluate the expression pattern of Msi1, ALDH1, Sox2, and CD49f in cervical cancer tissues and to determine their significance in predicting the prognosis in cervical cancer patients.

Patients and tissue specimens
A total of 179 cervical cancer patients treated between January 2001 and December 2008 were included in the study. The material was retrieved from archival paraffinembedded surgical samples at Sun Yat-Sen University Cancer Center. In addition, 75 pairs of snap-frozen cervical cancer and normal cervical samples from above patients were collected for Real-time PCR. None of the patients had received chemotherapy or radiotherapy before surgery. After surgery, the patients were treated with adjuvant chemotherapy according to the national guidelines. Platinum-based chemotherapy was initiated within two weeks after surgery and then repeated for four cycles at three-week intervals. In all cases, the diagnoses and grading were peer-reviewed according to the principles laid down in the latest International Federation of Gynecology and Obstetrics criteria [12]. Prior written consent was obtained from all patients and this study was approved by the Research Ethics Committee of Sun Yat-Sen University Cancer Center.

Immunohistochemical (IHC) staining
Immunostaining was performed on paraffin-embedded 4 μm sections and mounted on poly-L-lysine-coated slides. The sections were baked at 65°C for 30 minutes, then deparaffinized in xylene and rehydrated. Antigen retrieval was performed by submerging the sections into a 10 μmol/L citrate buffer solution (pH 6.0) for 10 minutes in a microwave oven. The slides were then treated with 3 % hydrogen peroxide in methanol to quench the endogenous peroxidase activity, followed by incubation with 1 % fish skin gelatin to block the nonspecific staining. Tissue sections were incubated overnight with monoclonal rabbit antibody against Msi1 (Abcam, Cambridge, USA; 1:200), monoclonal rabbit antibody against ALDH1 (Abcam, Cambridge, USA; 1:200), monoclonal mouse antibody against Sox2 (Abcam, Cambridge, USA; 1:200), and monoclonal rabbit antibody against CD49f (Abcam, Cambridge, USA; 1:200). After washing, the sections were incubated with prediluted secondary antibody (Abcam, Cambridge, USA), followed by further incubation with 3,3-diaminobenzidine tetrahydrochloride (DAB). Finally, the slides were counterstained with hematoxylin and mounted in an aqueous mounting medium.
For negative controls, primary antibodies were replaced with normal serum.

IHC score
Immunostaining was separately evaluated by two independent pathologists who were blinded as to the patients. Expression of the four CCSC markers was analyzed by an individual labeling score considering percent and staining intensity of positive cells. Intensity of stained cells was graded semi-quantitatively into four levels: 0 (no staining); 1 (weak staining = light yellow); 2 (moderate staining = yellow brown) and 3 (strong staining = brown); and the percentage was scored as: 0, negative; 1, 10 % or less; 2, 11 % to 50 %; 3, 51 % to 80 %; or 4, 80 % or more positive cells. Intensity and fraction of positive cell scores were multiplied for each marker and thus the scoring system was defined as low expression for scores of 0-3, and as high expression for scores of 4-12.

Statistical analysis
All statistical analyses were carried out using SPSS (version 16.0, SPSS Inc, Chicago, USA) statistical software. The overall survival (OS) and recurrence-free survival (RFS) were calculated as the time from the date of primary surgery to the date of first death or recurrence. Survival curves were plotted using the method of Kaplan-Meier, and the log-rank test was used to determine statistical differences between life tables. The correlations between clinicopathologic characteristics and recurrence were analyzed using the χ 2 test. Univariate and multivariate analysis were applied to assess the independent predictive significance of variables on RFS. P < 0.05 in all cases was considered statistically significant.

Clinical results and tumor recurrent
Patients' clinicopathological characteristics were shown in Table 1. In total, 179 cervical cancer patients with up to stage IIB were investigated. The median follow-up period of the patients was 48.8 months (range 1.6-60.0 months). After postoperative chemotherapy, 22 patients had recurrent disease.

Patterns of expression
To investigate the potential roles of CSC markers in the development of cervical cancer, we determined the expression of Msi1, ALDH1, Sox2 and CD49f in 179 paraffin-embedded cervical carcinoma samples by immunohistochemistry. Msi1 immunoreactivity was both cytoplasmic and nuclear and high expression was observed in 46 cases (25.6 %). ALDH1 immunoreactivity was cytoplasmic and high expression was observed in 77 cases (43.0 %). Sox2 immunoreactivity was nuclear and high expression was observed in 111 cases (62.0 %). CD49f immunoreactivity was both membranous and cytoplasmic and high expression was observed in 52 cases (29.0 %). The representative immunostaining of Msi1 ( Fig. 1a-b), ALDH1 (Fig. 1c-d), Sox2 (Fig. 1e-f) and CD49f (Fig. 1g-h) were shown in Fig. 1. Moreover, Realtime PCR showed that the expression level of all CSC markers was increased in CSCC compared with that in normal cervical tissues (Fig. 2).

Associations between Msi1, ALDH1, Sox2 and CD49f expression and clinicopathological features
The expression of CCSC markers was evaluated in relation to clinicopathological variables for cervical cancer. Msi1, ALDH1, and Sox2 expression were positively correlated with tumor recurrence (P = 0.005, P = 0.003, and P = 0.003, respectively), while CD49f expression was negatively correlated with tumor recurrence (P = 0.028) ( Table 2). Msi1, ALDH1 and CD49f expression was not associated with any of the clinicopathological variables, while only Sox2 expression was significantly associated with tumor size (P = 0.010). (Additional file 1: Table S1).

Associations between Msi1, ALDH1, Sox2 and CD49f expression and clinical prognosis
To examine the prognostic value of CSCs markers in cervical cancer, we evaluated the correlation between Msi1, ALDH1, Sox2 and CD49f expression and prognosis in CSCC patients. Notably, high expression of Msi1 (P = 0.033 and P = 0.003, Fig. 3a Fig. 4. Tumors with Msi1 high /CD49f low , ALDH1 high /CD49f low , and Sox2 high / CD49f low expression had a poorer prognosis compared with those with Msi1 low /CD49f high (P = 0.002 and P < 0.001, Fig. 4a and b), ALDH1 low /CD49f high (P = 0.008 and P = 0.002, Fig. 4c and d), and Sox2 low / CD49f low expression (P = 0.006 and P = 0.005, Fig. 4e and f ), respectively. Univariate and multivariate Cox proportional hazards regressions analysis were also applied. Univariate analysis showed that FIGO stage, tumor size, and Msi1 expression were significantly correlated with both OS and RFS. The ALDH1 expression, Sox2 expression, and CD49f expression were significantly associated with RFS (data not shown). Multivariate analysis showed that FIGO stage > IB1 (P = 0.002 and P = 0.005, respectively), high Sox2 expression (P = 0.047 and P = 0.018, respectively), high ALDH1 expression (P = 0.013 and P = 0.003, respectively), and low CD49f expression (P = 0.008 and P = 0.003, respectively) were independent prognostic factors for both OS and RFS (Table 3).

Discussion
CSCs are generally thought to arise from normal stem cells that have developed genetic mutations over time. CCSCs could arise from Müllerian duct-derived cervical stem cells located in the basal layer of the ectocervical squamous or endocervical columnar epithelium [13]. These cells have been proposed to drive the resistance to chemotherapy through the expression of drug efflux pumps in normal stem cells from which they were derived [14]. Therefore, the identification of the CSCs subpopulation of tumor cells may offer new directions for the treatment of cervical cancer. In the present work, we provide the first link between the expression of four putative CCSC markers and the clinical outcome in cervical cancer patients.
Previous studies have demonstrated that CSCs are responsible for the high rate of chemoresistance and tumor relapse. The possible mechanisms of chemoresistance include high expression levels of adenosine triphosphatebinding cassette transporters, low self-renewal rate, an active DNA repair capacity, and activated Wnt/β-catenin and Notch signaling [15]. Clinical and experimental studies have provided evidence of the relationship between  different CSCs phenotypes and resistance to chemotherapeutic drugs. Manohar et al. proposed that overexpression of MYCN, the most widely characterized gene associated with poor outcome of neuroblastoma, resulted in increased drug resistance and expression of MRP1, suggesting that MRP1 may be a MYCN target gene involved in the drug-resistance phenotype of neuroblastoma [16]. Wang et al. reported that overexpression of caveolin-1 enhanced chemoresistance of breast CSCs, which could be prevented by downregulation of the β-catenin/ABCG2 pathway [17]. Recent studies also implicated CSCs in cervical cancer chemoresistance [18], while there is lack of data concerning the relevance of the expression of CSC markers with clinical prognosis and chemosensitivity in cervical cancer. Our findings suggest that the expression of CSC markers Msi1, ALDH1, Sox2, and CD49f were increased in CSCC tissues and that their expression are linked to clinical survival in cervical cancer patients undergoing postoperative chemotherapy, suggesting an important role of these markers in predicting clinical prognosis in cervical cancer. Surprisingly, we found that the expression of CSC markers showed no association with age, tumor stage, tumor size, or tumor differentiation, while only Sox2 expression was significantly correlated with tumor size. The results may be due to small group size.
ALDH1 has long been thought to be a CSC marker and has also been described to be associated with tumor progression and recurrence [19]. Many recent studies investigating the ALDH1 gene or protein report its impact on the clinical outcome of patients with malignancies. Dylla et al. reported that the inherent chemotherapeutic  resistance mechanism of colorectal CSCs includes ALDH1 enzymatic activity [20]. Alamgeer et al. described ALDH1 as a useful predictor of chemoresistance in locally advanced breast cancer [21]. Our present investigation found that high ALDH1 expression was an independent predictor of recurrence and overall survival in CSCC patients. Among patients with high ALDH1 expression, 21 % had recurrent cancer, whereas only 6 % patients with low ALDH1 expression did. Our study is supported by Deng et al., who demonstrated that high residual ALDH1 expression after radiochemotherapy significantly predicted tumor metastasis or recurrence [22]. The same finding was also applied for another potential CCSC marker Sox2.
It has recently been linked with enhanced chemoresistance and tumorigenecity of gastric cancer derived cancer stem-like cells [23]. Conversely, suppression of Sox2 can impair the chemosensitivity and stemness of cancer cells [24]. In the present study, we found that up-regulation of Sox2 predict poor overall and recurrence-free survival in CSCC patients. Among patients with high Sox2 expression, 18 % had recurrent cancer, whereas only 3 % patients with low Sox2 expression did, implicating a correlation of  observed a high expression level of Msi1 in these cells [25]. Johannessen et al. demonstrated that the expression of Msi1 was downregulated through inhibition of PI3K/ AKT pathway, which is involved in treatment resistance in glioblastomas [26]. In this study, we document a poorer overall and recurrence-free survival time with high expression of Msi1 in univariate but not multivariable survival time analysis, indicating that the poor prognostic role of Msi1 may be secondary to their association with other established prognostic criteria. Altogether, these studies suggest that increased ALDH1, Sox2, and Msi1 expression are related to the progression and may be related to chemoresistance of CSCC. Interestingly, we found that CD49f expression was increased in CSCC tissues while high expression of CD49f was associated with favourable prognosis of patients. The results suggested that CD49f expression was linked with reduced chemoresistance in cervical cancer. CD49f has been reported to play a critical role in CSC maintenance and in the attachment of tumor cells to laminin [27]. The unexpected finding that CD49f expression was correlated with better clinical outcome suggests that not all CSC markers characterise the chemoresistant cell population and that the prognostic role of CD49f might be tissue specific and diverse in different cancers. Another explanation is that the expression of CD49f may be decreased in the invasive area of cancer tissues, which has already been reported for other cell adhesion molecules, and which has been regarded as a key event in epithelial mesenchymal transition [28].
An important finding emerges from the present investigation is that tumors with high Msi1 and low CD49f expression had the poorest prognosis, whereas tumors with lack of Msi1 and CD49f overexpression had the best prognosis. Double stratification analysis was also performed according to ALDH1/CD49f, and Sox2/CD49f, and the results showed that ALDH1 high /CD49f low and Sox2 high / CD49f low population predict poor recurrent-free prognosis in CSCC patients. The results imply the importance of such a sub-population in predicting the prognosis and chemosensitivity of cervical cancer patients. Targeting this cell population with novel pharmaceutical agents may prove to be effective for the treatment of chemoresistant cervical tumors.

Conclusions
In conclusion, our study provided the first clinical evidence that CSC markers are associated with the clinical prognosis of CSCC patients. High expression of Msi1, ALDH1, and Sox2, and low expression of CD49f predict poor prognosis for cervical cancer patients receiving postoperative chemotherapy. Moreover, double stratification analysis showed that Msi1 high /CD49f low subgroup had the poorest RFS. Our findings may be of great value in the development of personalized therapies for patients with cervical cancer.

Additional file
Additional file 1: Table S1. Correlation of cancer stem cell markers with histopathological variables. (DOCX 15 kb) Submit your next manuscript to BioMed Central and take full advantage of: