SOX1 and PAX1 Are Hypermethylated in Cervical Adenocarcinoma and Associated with Better Prognosis

Background The increased risk and poor survival outcome of cervical adenocarcinoma (CAC) demand for effective early diagnostic biomarkers that can predict the disease progression and outcome. The purpose of this study was to investigate the value of methylation status of SOX1 and PAX1 in the detection and prognosis of CAC. Methods We performed a quantitative methylation-specific polymerase chain reaction in 205 cervical paraffin-embedded specimens (175 CACs, 30 noncancer cervical tissues). Overall and progression-free survival (OS and PFS, respectively) rates were calculated and compared using the Kaplan-Meier method. The prognostic value of SOX1m and PAX1m on CAC patients was assessed by the Cox regression model. A mathematical formula combining SOX1m, PAX1m, and age was constructed for survival prediction. Results The methylation status of SOX1 and PAX1 was higher in CAC tissues than in noncancer cervical tissues. In addition, SOX1m-positive CAC patients showed a higher 5-year OS rate than SOX1m-negative patients. In CAC patients with smaller tumor size (<4 cm), the PAX1m-positive group showed a higher 5-year PFS rate than the PAX1m-negative group. In the algorithm combining SOX1m, PAX1m, and age, the low-risk group showed a better 5-year OS and PFS rate than the high-risk group. Conclusion SOX1 and PAX1 methylation levels are higher in CAC than in normal cervical tissues and are potential biomarkers for monitoring CAC prognosis.


Introduction
Cervical cancer is the second commonest tumor in women worldwide [1]. Moreover, it is a major cause of cancerrelated death among women in developing countries [2]. Cervical adenocarcinoma (CAC) ranks second after cervical squamous cell carcinoma (SCC) as the most common pathological type. Although the use of human papillomavirus (HPV) vaccines and effective Pap smear screening have significantly decreased the incidence rate of cervical carcinoma in most regions, the percentage of CAC has been increasing, especially in younger women, accounting for 20-25% of all cervical carcinoma in some developed countries [2,3]. Moreover, the high propensity of CAC for ovarian metastases always leads to fertility destruction in young women [4][5][6][7][8][9][10]. Several studies have shown that at the same stage, CAC has a worse prognosis than SCC, mainly because of its higher rate of metastase [6] and lower sensitivity to radiotherapy and chemotherapy [11]. The increased frequency and poor survival outcome raise the need for useful biomarkers that can predict the progression and prognosis of CAC.
Epigenetic alterations include heritable DNA methylation and histone protein modifications that do not affect the DNA transcriptional sequence [12][13][14][15]. DNA methylation is an epigenetic alteration, which always occurs in the early stage of carcinogenesis, leading to lessen even lost expression of the methylated gene [16]. Expression of SOX1 correlated with early embryogenesis, central nervous system development, and neural stem cell maintenance [17]. Nonetheless, hypermethylation in the promoter region and/or somatic mutations in the so-called tumor suppressor genes might cause silencing or inhibition of SOX1, which in turn may result in cancer cell proliferation and migration and finally progression of cervical carcinogenesis [18]. Hypermethylated SOX1 (SOX1 m ) has been associated with several cancer types, including hepatocellular cancer, lung cancer, urothelial bladder cancer, endometrial cancer, and SCC [19][20][21][22]. PAX1 expression is correlated with embryogenesis, particularly with the development of the thymus, parathyroid glands, and skeleton [23][24][25]. Hypermethylated PAX1 (PAX1 m ) has been found in ovarian cancer, oral cancer, and SCC [26][27][28][29]. Our previous study confirmed SOX1 and PAX1 methylation as promising screening biomarkers in cervical neoplasia, mainly in high-grade squamous intraepithelial lesions and SCC, because of its remarkable discriminating ability between normal tissues and high-grade cervical lesions [30][31][32][33]. However, CAC and SCC are different with respect to tumor development, progression, and molecular pathology. It remains unknown whether the methylation status of SOX1 and PAX1 is different between CAC and SCC. In addition, the methylation level and prognostic value of SOX1 and PAX1 for CAC are unclear. In this study, we investigated the methylation levels of SOX1 and PAX1 differ in CAC and evaluated the potential value of SOX1 and PAX1 gene methylation for prognosis in CAC.

DNA Preparation, Bisulfite Conversion, and Quantitative
Methylation-Specific Polymerase Chain Reaction (qMSP). An ISO17025-certified laboratory (iStat Biomedical Co., Ltd., New Taipei City, Taiwan) carried out total methylation tests. They first deparaffinized paraffin-embedded cervical tissues and then extracted genomic DNA (gDNA) samples and bisulfite converted by using an Epigene™ nucleic acid extraction kit and an Epigene™ bisulfite conversion kit (iStat Biomedical Co., Ltd., New Taipei City, Taiwan). Quantitative methylation-specific PCR (qMSP) was then used for analyzing the methylation level of SOX1 and PAX1 by the TaqMan Probe system in a Light Cycler LC480 system (Roche Applied Science, Penzberg, Germany). Our previous study described specific primers and probes for qMSP [31,34,35]. The registered A375 and CaSki two cancer cell lines were treated as methylation and nonmethylation controls to ensure the quality of the bisulfite conversion and qMSP processes. The DNA methylation level was assessed as the methylation index (M-index) using the formula [36]: 10,000 × 2 ðCpvalueofgene−Cpvalueof COL2AÞ . SOX1 and PAX1 (positive) were deemed to be hypermethylated (positive) if the delta Cp was smaller than 11 and 9, respectively.

Statistical Analyses.
The cut-off values for SOX1 m and PAX1 m were generated from 205 clinical samples (175 CACs and 30 noncancer cervical tissues). Receiver operating characteristic (ROC) curves were performed, and the area under the ROC curve (AUC) was calculated for the detection of the CAC.
All statistical analyses were performed using GraphPad Prism® 7.00 (GraphPad, La Jolla, CA, USA) and SPSS Statistics 24 (SPSS, Inc., Chicago, IL, USA). The correlation between SOX1 m or PAX1 m and each clinicopathological characteristic of the CAC patients was evaluated by the Mann-Whitney and Dunnett's tests. Kaplan-Meier method was used to describe the progression-free survival and overall survival (PFS and OS). The PFS was judged from treatment to the date of the first relapse at any site or death including all causes, and OS was calculated from treatment to death covering any cause. Hazard ratio (HRs) was calculated with multivariate Cox regression analysis.

A Mathematical Algorithm
Combining SOX1 m , PAX1 m , and Age for CAC Prognosis Prediction. To investigate the effectiveness of the combination of the methylation statuses of these two genes and the clinicopathological factors to

BioMed Research International
and PAX1 (515:70 ± 56:30 and 28:19 ± 9:19, respectively, p < 0:05) was significantly higher in CAC tissues than in noncancer tissues (Figure 1(a)). By ROC analysis, the positive cut-off value for SOX1 m was ΔCp ≤ 11, with a high AUC level of 88.42%, a sensitivity of 87.22%, and specificity of 56.67%. The positive cut-off value for PAX1 m was ΔCp ≤ 9, with a high AUC level of 70.80%, sensitivity of 44.30%, and specificity of 100% (Figure 1(b)), which suggested that SOX1 m and PAX1 m may be detection biomarkers for CAC.
The SOX1 and PAX1 methylation statuses showed no significant difference based on age, FIGO stage, tumor size, depth of invasion, lymph node metastasis, and histologic grade in CAC patients (Table 2).

Hypermethylated SOX1 and PAX1 Are Associated with
Better Survival in CAC Patients. Further studies were conducted to investigate whether the methylation of SOX1 and PAX1 was correlated with the prognosis in CAC patients. The SOX1 m -positive group showed a higher 5-year OS rate of 93.35% than the SOX1 m -negative group, which showed a 5-year OS rate of 68.29% (p = 0:048) (Figure 2(a)). While no remarkable finding was obtained in the analysis of the 5-year PFS rate. For PAX1, there was no evident difference in 5-year OS rate or 5-year PFS rate between these two groups (data not shown). However, in CAC patients with smaller tumor size (<4 cm), the PAX1 m -positive group had a higher 5-year PFS rate than the PAX1 m -negative group (100% vs. 82.4%, p = 0:044) (Figure 2(b)).

Discussion
It is well known that the prognosis of CAC is worse than cervical SCC, even for early-stage patients, especially in developing countries [37]. Gene methylation is an epigenetic modification, which has tumor-suppressive or tumorigenic two opposite effects, possibly playing key roles in carcinogenesis and cancer progression. Before this study, the methylation status and prognostic value of SOX1 m and PAX1 m for CAC were unclear. In this study, we observed higher methylation levels of SOX1 and PAX1 in CAC than in noncancer cervical tissues. For the detection of CAC, SOX1 m showed 87.22% sensitivity and 56.67% specificity, while PAX1 m showed 44.30% sensitivity and 100% specificity. Moreover, several studies have demonstrated that PAX1 methylation increases following increased disease grade: PAX1 methylation in SCC > high − grade squamous intraepithelial lesion ðHSILÞ > low − grade squamous intraepithelial lesion ðLSILÞ > normal tissue [38,39]. These suggest that hypermethylation of SOX1 and PAX1 might play an important role in the diagnosis and cancer progression of CAC.
The current study also showed, for the first time, that CAC patients who are SOX1 m -positive show a better

BioMed Research International
has several limitations. First, our patients all came from a single medical center and the sample size was comparatively small. Second, the detailed molecular relationship between SOX1 m and PAX1 m in CAC has not been explored, which deserves further investigation.
Despite the poor prognosis of CAC, there is still no valid prognostic risk model. By far, age is a strong risk factor for cancer [40,41]. The current paper showed corresponding directed changes in DNA methylation with age, which is characterized by hypermethylation of targets of polycomb group proteins (PCGTs) that are crucial in embryonic stem cell lineage differentiation [42]. In the algorithm combining SOX1 m , PAX1 m , and age, the low-risk group, composed of high methylation level of SOX1 and PAX1 and younger age, showed a significantly higher 5-year OS rate and 5-year PFS rate than the high-risk group. These results suggested that the algorithm has the potential for a 5-year CAC prognosis.
In our previous study, negative gene methylation correlated with high protein expression, which increased the resistance of cervical cancer cells to radiation and chemotherapy [43]. In further studies, it will be essential to analyze the correlation between SOX1 and PAX1 methylation status and sensitivity of the cervical cancer cell to radiotherapy and chemotherapy.
In conclusion, this study suggests that SOX1 and PAX1 methylation levels are higher in CAC than in cervical SCC, and SOX1 m and PAX1 m are potential biomarkers for monitoring CAC prognosis.

CAC:
Cervical adenocarcinoma SCC: Cervical squamous cell carcinoma SOX1 m : Methylation status of SOX1 PAX1 m : Methylation status of PAX1 FIGO: International Federation of Gynecology and Obstetrics PFS: Progression-free survival OS: Overall survival.

Data Availability
The data used to support the findings of this study are included in the article.