Aberrant Methylation of MEG3 Functions as a Potential Plasma-Based Biomarker for Cervical Cancer

Methylation alterations of specific genes have recently been identified as diagnostic biomarkers for human cancers. Although MEG3 has been proved to be a tumor suppressor in cervical cancer according to our previous study, the diagnostic value of MEG3 methylation in plasma is still unknown. Therefore, the aim of this study is to identify a novel epigenetic biomarker for cervical cancer. In the current study, the level of MEG3 methylation was evaluated using methylation-specific polymerase chain reaction. The results showed that the level of MEG3 methylation was significantly higher in cervical cancer tissues and patients’ plasmas than those in adjacent normal tissues and plasmas of healthy participants respectively. Moreover, the accuracy was good enough for MEG3 methylation in plasma to discriminate CIN III patients from healthy participants. In addition, MEG3 methylation in plasma also has high discriminating power to predict HR-HPV infection and lymph node metastasis. Furthermore, hypermethylation of MEG3 in plasma was associated with worse recurrence-free and overall survival in cervical cancer patients. In conclusions, MEG3 methylation in plasma can serve as a diagnostic and prognostic biomarker for cervical cancer, providing useful information for clinical management.

Cervical cancer is the leading cancer which originated from the female reproductive system 1 . In 2011, new cases of cervical cancer were estimated to be about 490,000 worldwide, and deaths caused by this disease would be more than 275,000 which most occurred in developing countries 2 . Despite pap smear screening and other diagnostic techniques are used widely, the overall survival of cervical cancer patients remains poor 3 . Therefore, further understanding the mechanisms of tumorigenesis and development which contribute to prevention and cure of cervical cancer is urgent.
Long non-coding RNA (lncRNA) is RNA molecule that is more than 200 nt in length with limited or no protein-coding capacity 4 . LncRNAs are known to participate in a wide range of biological processes, including the modulation of apoptosis, proliferation, invasion and so on [5][6][7] . Human maternally expressed gene 3 (MEG3) is a maternally expressed imprinted gene that encodes a length of 1.6 kb nucleotides which suggests to function as a non-coding RNA 8 . Recent studies have demonstrated that MEG3 is abnormal expressed in various human cancers, such as retinoblastoma 9 , colorectal cancer 10 , and hepatocellular carcinoma 11 . More important, our previous study has proved that MEG3 was down-regulated and affected cell proliferation and apoptosis in cervical cancer 12 . However, little is known about the diagnostic value of MEG3, especially MEG3 methylation status, as a plasma-based biomarker for cervical cancer patients.
In present study, we mainly focused on the diagnostic value of MEG3 methylation in plasma base on our previous study. We first confirmed the reliability of MEG3 methylation in plasma of cervical cancer patients. Then we found that MEG3 methylation in plasma was an effective biomarker for the diagnosis of CIN III (cervical intraepithelial neoplasia), HR-HPV (High risk-Human papillomavirus) infection and lymph node metastasis.  correlation in the same patient (P < 0.001, R = 0.909, Fig. 2a). More important, we observed a decrease in methylation level of MEG3 promoter after operation in plasmas of cervical cancer patients (P < 0.001, Fig. 2b,c).
The relevance of MEG3 methylation and clinical pathological characteristics of cervical cancer patients. There were no significant differences in plasma MEG3 methylation between CIN I-II and their matched healthy controls (Fig. 3a). But the level of MEG3 methylation was increased in plasmas of CIN III patients compared with that of healthy controls (P < 0.001, Fig. 3a). As shown in Fig. 3b, the methylation level of MEG3 in the plasmas was found to higher in HR-HPV positive patients compared with that in negative patients (P < 0.001). Moreover, the methylation level of MEG3 in the plasmas of cervical cancer patients with lymph node metastasis was significantly higher compared with those without metastasis (P < 0.001, Fig. 3c). In addition, Chi-square test also showed that hypermethylation of MEG3 promoter in plasma was a risk factor not only for CIN III and cervical cancer, but also HPV infection and lymph node metastasis (Tables 1 and 2). However, no significant associations were found between MEG3 methylation and age, menopause, histology, depth of invasion, differentiation, tumor size or Lymphatic vascular space invasion (LVSI) (  4a).
In test set, AUCs were 0.788, 0.730 and 0.804 according to MEG3 methylation of plasma for the diagnosis of CIN III, HPV infection and lymph node metastasis respectively. When using best point of training set, the sensitivities were 84.2%, 78.1% and 82.4%, meanwhile specificities were 52.6%, 70.0% and 72.0% for CIN III, HPV infection and lymph node metastasis, respectively (Fig. 4b).  Moreover, we also calculated the sensitivity and specificity using formula method. The sensitivity was 81.3% and specificity was 63.5% for prediction of lymph node metastasis, while sensitivity and specificity was 63.1% and 84.2% for diagnosis of HPV infection (Table 3, n = 168). Then we evaluated the sensitivity and false negative rate of fluid based thin-layer cytological test (TCT) and MEG3 methylation for CIN III. The date showed that the sensitivity of TCT and MEG3 methylation was 87.5% and 79.2%, while the false negative rate of TCT and MEG3 methylation was 12.5% and 20.8% respectively ( Table 4, n = 76). When combined with TCT and MEG3 methylation, the sensitivity and false negative rate was 93.1% and 6.9% respectively ( Table 5, n = 76).
Association of MEG3 methylation with prognosis of cervical cancer. With regard to combined level of MEG3 methylation in plasma and prognosis, we divided the cervical cancer patients into 2 groups: group 1, methylated group (n = 90); group 2, partially methylated (M and U) and unmethylated group (n = 78). The Kaplan-Meier plot showed that patients with high level of MEG3 methylation had poorer recurrence-free survival (RFS) than those with low level of MEG3 methylation (P = 0.0004, Fig. 5a). On the other hand, the aberrant hypermethylation of MEG3 in plasma was also significantly associated with poorer overall survival (P = 0.0013, Fig. 5b).
Furthermore, univariate analysis of RFS revealed that the level of MEG3 methylation in plasma was a prognostic indicator for cervical cancer patients (Table 5). More important, Multivariate survival analysis further demonstrated that hypermethylation of MEG3 in plasma was an independent prognostic marker indicating poorer RFS for cervical cancer patients (Table 5).

Discussion
Alternation of methylation pattern in cancer cell genomes is event taking place during the early stages of carcinogenesis and leading to profound modifications in gene expression 13 . Aberrant DNA methylation, which is the most common epigenetic change in human malignancies, may be used as a diagnosis, surveillance and prognostic biomarker according to recent studies [14][15][16] . Although plasma DNA methylation has been proved to be a potential biomarker for some diseases, the study forced on DNA methylation of cancer patients' plasma is still limited 17,18 .
The promoter region of MEG3 is rich in CpG dinucleotides. There are two differentially methylated regions (DMRs) located upstream of the MEG3 gene, IG-DMR and MEG3-DMR, which control the expression of MEG3 19 . It has been well proved that the loss of MEG3 expression in several types of human cancers was at least in part the result of hypermethylation in the MEG3 promoter region 20,21 .
In our previous study 12 , we demonstrated that MEG3 was down-regulated in cervical cancer tissues. Moreover, MEG3 expression was negatively related with HR-HPV infection, FIGO stages and lymphatic metastasis. At last, we revealed that MEG3 functioned as a tumor suppressor by regulating miR-21-5p. However, the diagnostic value of plasma MEG3 methylation for cervical cancer patients is still unknown.
Based on others' and our previous study, present study was designed to fill in this research gap. We first provided evidences to support that MEG3 promoter was hypermethylation not only in cervical cancer tissues, but also in plasmas of cervical cancer patients. We also observed that the level of MEG3 methylation in plasma was lower than that in cervical cancer tissues. DNA in plasma contains a larger background of DNA from normal cells may be one of the primary causes of it. Furthermore, we confirmed that the level of MEG3 methylation was  significantly decreased in postoperative plasma, implying that MEG3 hypermethylation in plasma might be a result of the tumor tissue. More important, we found that the level of MEG3 methylation in plasma was positive related to that in tissue of same patient, indicating that the detection of MEG3 methylation level in plasma could in stead of that in tissue which made it more convenient and noninvasive. In summary, we provided strong evidences supporting that plasma MEG3 methylation could be a candidate of biomarker for cervical cancer. Next, we focused on the diagnostic power of plasma MEG3 methylation. We first confirmed that hypermethylation of MEG3 in plasma was a risk factor for not only CIN III, but also HR-HPV infection and lymph node metastasis. That was consistent with the results of our previous study in tissues. Base on that, we further evaluated the diagnostic power of plasma MEG3 methylation for these relative factors. To make the results more credible, we divided patients and their match healthy participants into training set and test set. ROC curves indicated that plasma MEG3 methylation had high discriminating power to predict CIN III, HPV infection and lymph    node metastasis in training set. Then we tested the optimal cutoff points in test set, showing that the sensitivity and specificity were still acceptable which made our results more credible. More important, we proved that the combination of TCT and MEG3 methylation was a much more powerful discrimination tool in predicting CIN III than TCT alone. For CIN III is a premalignant lesion of cervical cancer, our result implied that plasma MEG3 methylation might have great value in early diagnosis of cervical cancer. In addition, lymph node metastasis is a key factor in treatment decisions, suggesting that plasma MEG3 methylation might use for screening high-risk patients before therapy. Moreover, traditional tumor markers in plasma were almost the products of oncogene. The study focused on tumor suppressor gene as a plasma-based biomarker was still rare. In present study, we opened up a new perspective for exploration of plasma-based biomarker by analysis of DNA methylation of tumor suppressor gene. For plasma MEG3 methylation was closely linked to lymph node metastasis of cervical cancer, we speculated that it might be associated with prognosis too. As anticipated, there was a trend toward the poorer RFS and overall survival in the patients with high level of MEG3 methylation in plasma. More important, Cox multivariate analysis suggested that plasma MEG3 methylation represented an independent prognostic factor for RFS. Thus, plasma MEG3 methylation may be useful as a prognostic marker for the prediction of relapse and survival in cervical cancer patients.

Lymph nodes metastasis HPV infection
However, there is still much work should be done. Some techniques, such as methyl-BEAMing and Pyrosequencing, demonstrated higher efficiency and more sensitive than MSP recently 22 . In future studies, we will use these techniques to further confirm our data and make our results more accurate.
In conclusion, plasma MEG3 methylation may be a reliable diagnostic biomarker for the early stage, HR-HPV infection and lymph node metastasis of cervical cancer. Our study also supports its application as a prognostic biomarker for cervical cancer. Therefore, the methylation status of MEG3 in plasma is a potential novel epigenetic biomarker which has great value in the clinical practice of cervical cancer.    The PCR products were separated on a 2.5% agarose gel, stained with ethidium bromide, and visualized under UV illumination. Finally, the gray scanning was carried out by Quantity One software (Bio-Rad, California, US). The primers for the methylated (M) reaction were as follows: 5′-GTT AGT AAT CGG GTT TGT CGG C (sense) and 5′-AAT CAT AAC TCC GAA CAC CCG CG (antisense), with a 160-bp amplification product. The primers for the unmethylated (U) reaction were as follows: 5′-GAG GAT GGT TAG TTA TTG GGG T (sense) and 5′-CCA CCA TAA CCA ACA CCC TAT AAT CAC A (antisense), with a 120-bp amplification product. The percentage of methylation was calculated using following method: gray value of methylation/(gray value of methylation + gray value of unmethylation).

Statistical analysis.
All experiments were repeated independently three times and data was presented as the mean ± standard deviation. Associations between MEG3 methylation and clinical-pathologic characteristics were analyzed by the chi-square test. Receiver operating characteristic (ROC) curves were used to determine the diagnostic value for the patients in this study. Survival probability was analyzed by the Kaplan-Meier methods and evaluated by log-rank test. Cox regression model was also performed to assess the associations of MEG3 methylation with disease relapse or death. Comparisons between groups were evaluated by Student's t test or LSD test using SPSS 11.0 software (SPSS, Chicago, IL, USA). P < 0.05 was regarded as statistically significant.