DAPK1 Promoter Methylation and Cervical Cancer Risk: A Systematic Review and a Meta-Analysis

Objective The Death-Associated Protein Kinase 1 (DAPK1) gene has been frequently investigated in cervical cancer (CC). The aim of the present study was to carry out a systematic review and a meta-analysis in order to evaluate DAPK1 promoter methylation as an epigenetic marker for CC risk. Methods A systematic literature search was carried out. The Cochrane software package Review Manager 5.2 was used. The fixed-effects or random-effects models, according to heterogeneity across studies, were used to calculate odds ratios (ORs) and 95% Confidence Intervals (CIs). Furthermore, subgroup analyses were conducted by histological type, assays used to evaluate DAPK1 promoter methylation, and control sample source. Results A total of 20 papers, published between 2001 and 2014, on 1929 samples, were included in the meta-analysis. DAPK1 promoter methylation was associated with an increased CC risk based on the random effects model (OR: 21.20; 95%CI = 11.14–40.35). Omitting the most heterogeneous study, the between study heterogeneity decreased and the association increased (OR: 24.13; 95% CI = 15.83–36.78). The association was also confirmed in all the subgroups analyses. Conclusions A significant strong association between DAPK1 promoter methylation and CC was shown and confirmed independently by histological tumor type, method used to evaluate methylation and source of control samples. Methylation markers may have value in early detection of CC precursor lesions, provide added reassurances of safety for women who are candidates for less frequent screens, and predict outcomes of women infected with human papilloma virus.

detection of CC precursor lesions, provide added reassurances of safety for women who are candidates for less frequent screens, and predict outcomes of women infected with human papilloma virus. methylation markers may have value in early detection of CC precursor lesions, provide added reassurances of safety for women who are candidates for less frequent screens, and predict outcomes of women infected with HPV [34].
The aim of the present study was to carry out a systematic review and a meta-analysis in order to summarize the current published studies and to evaluate DAPK1 promoter methylation as an epigenetic marker for CC risk.

Search strategy and selection criteria
Firstly, a systematic literature search in the Medline database, using PubMed, was carried out for epidemiological studies, published before July 2014, investigating the association between gene promoter methylation and CC risk. Literature search was conducted independently by two Authors using the keywords "promoter methylation" and "cervical neoplasia". The searches were limited to studies written in English; abstracts and unpublished studies were not included. Moreover, the reference lists from selected articles were checked to search for further relevant studies. The aim of the first selection was to identify studies that investigated the association between promoter methylation of any gene and CC risk; no studies were excluded a priori for weakness of design or data quality. Accordingly, articles were selected only if they satisfied the following criteria: i) case-control or cohort study designs, and ii) studies that assessed the association of gene promoter methylation and CC. Subsequently, since DAPK1 gene has been identified as the most common analyzed and studied gene, a meta-analysis of articles reporting the association between DAPK1 promoter methylation and CC risk was performed. Thus for inclusion in the quantitative analysis, studies had to meet the following criteria: i) studies that assessed the association between DAPK1 methylation and CC and ii) provided data about the frequency of DAPK1 methylation in cancer and in control groups. Furthermore, exclusion criteria were as follows: i) studies that did not use exfoliated cells, cervical biopsies or urines as samples and ii) in which control or cancer groups included individuals with various types of precancerous lesions. The preferred reporting items for systematic reviews and meta-analysis (PRISMA) guidelines were followed [37] (S1 and S2 Files).

Data extraction and quality assessment
Two of the Authors independently reviewed all the eligible studies and abstracted the following information in a standard format: first Author's last name, year of publication, country where the study was performed, sample type, experimental methods to assess DAPK1 methylation and number of cases and controls subjects.

Statistical Analysis
All data were analyzed using the Review Manager 5.2 software provided by the Cochrane Collaboration (http://ims.cochrane.org/revman).
Forest plots were generated to illustrate the study-specific effect sizes along with a 95% CI. The fixed-effects or random-effects models, according to heterogeneity across studies, were used to calculate the ORs and 95% CIs in order to assess the association between DAPK1 promoter methylation and CC risk. Where a value of zero in the number of promoter methylation events caused problems with computation of the ORs for individual studies, the Review Manager 5.2 software provided to add a value of 0.5 to all cells of the related crosstab [38].
Heterogeneity across studies, was measured using the Q-test based on the χ2 statistic, considering significant statistical heterogeneity as p <0.1. As Cochran's test only indicates the presence of heterogeneity and not its magnitude, we also reported the I 2 statistic, which estimates the percentage of outcome variability that can be attributed to heterogeneity across studies. An I 2 value of 0% denotes no observed heterogeneity, whereas, 25% is ''low", 50% is ''moderate" and 75% is ''high" heterogeneity [39]. We also estimated the between-study variance using tau-squared (t) statistic [40].
Furthermore, subgroup analyses were conducted by histological type (SCC and AC), by assays used to evaluate DAPK1 promoter methylation (Methylation Specific PCR-MSP and real-time quantitative MSP-qMSP), and by control sample source (normal cervical tissues-NT and benign cervical tissues-BCT). A sensitivity analysis was performed to find relatively poor-quality studies by the omission of a single study at a time and to see whether a particular omission could affect the overall OR value and the heterogeneity across studies.
To determine the presence of publication bias, the symmetry of the funnel plots in which ORs were plotted against their corresponding standard errors were assessed.

Search results and data characteristics
The detailed steps of the systematic review and meta-analysis process are given as a PRISMA flow chart (Fig 1). A total of 519 articles were retrieved from the database. After exclusion of studies that not met the inclusion criteria, DAPK1 resulted the most common analyzed gene.   [60][61][62]. All studies evaluated DAPK1 promoter methylation in SCC and 12 studies (44.4%) also in AC. Regarding the method of promoter methylation evaluation, the "gold standard method", used in most studies (67%), was MSP, followed by qMSP (26%), sequencing (3.5%) and Methylation specific-multiplex ligation-dependent probe amplification (MS-MLPA) (3.5%).

Meta-analysis
Of the 27 selected articles, 4 studies conducted without a control group, 2 studies which included precancerous lesions in control or in case groups and 1 study which reported inadequate data, were excluded from the meta-analysis. Thus, 20 studies (74%) evaluating DAPK1 promoter methylation both in tumor and in healthy control samples were included in the present meta-analysis. Overall, the studies reported results obtained from 1929 samples: 1092 from cancer patients and 837 from controls. Regarding the source of control samples, 10 studies evaluated DAPK1 promoter methylation in BCT from patients having gynaecological diseases such as uterine myoma, adenomyoma, and uterine prolapse, 8 studies in NT from healthy people and 2 studies in normal cervical tissues adjacent to the tumor. DAPK1 promoter methylation was associated with an increased CC risk with a pooled OR of 19.97 (95% CI = 13.57-29.38) based on the fixed effects model. However, due to the significant heterogeneity (I 2 = 49%; p = 0.007), a pooled OR of 21.20 (95%CI = 11.14-40.35), based on the random effects model, was obtained (Fig 2). Subgroup analyses were performed by histological types, methods for methylation analysis and sources of control samples. The association between DAPK1 promoter methylation and CC was confirmed in each subgroup (S1 Table).
In addition, the sensitivity analysis found the study by Yang et al. (2010) [54], as the relatively poor-quality study. When this study [54] was omitted, the between study heterogeneity decreased to I 2 = 39% (p = 0.04), and the association between DAPK1 promoter methylation and CC risk increased (OR: 24.13; 95% CI = 15.83-36.78) (S1 Fig) Subgroup analyses omitting the heterogeneous study [54] were performed. Subgroup analysis by histological types showed that the heterogeneity totally disappeared in AC subgroup (I 2 = 0%; p = 0.93) and the association was confirmed both in SCC (OR = 33.84; 95% CI = 15.61-73.37; based on the random effects model) ( Fig 3A) and AC (OR = 21.89; 95% CI = 8.64-55.48; based on the fixed effects model) (Fig 3B) subgroups. Furthermore, subgroup analysis based on assays methods used to evaluate DAPK1 promoter methylation was performed including the two common techniques, MSP and qMSP. The ORs were 23.45 (95% CI = 10.56-52.09), based on the random effects model, in MSP subgroup, and 34.25 (95% CI = 12.34-95.04), based on the fixed effects model, in qMSP subgroup, while the I 2 were 51% and 0%, respectively (Fig 4A  and 4B). The subgroup analysis by source of control sample, and particularly between NT and BCT, reported that the ORs were 16.99 for NT (95% CI = 9.09-31.76) and 22.00 for BCT (95% CI = 11.95-40.51), respectively. Heterogeneity in NT and BCT subgroups were low with I 2 = 48% and I 2 = 14%, respectively (Fig 5).
The funnel plot of the pooled analysis (Fig 6), which is quite symmetric, suggests no significant bias among the included studies, however the shapes of the subgroups analyses (S2, S3 and S4 Figs) indicate small to moderate asymmetry, therefore publication bias cannot be completely excluded as a factor of influence on the present meta-analysis.

Discussion
Tumor suppressor genes belonging to different pathways, as cell adhesion, DNA repair, cell cycle checkpoint control and nuclear receptors, have been found to be hypermethylated in CIN and CC [41,55,60].
A previous review [63] summarized the results of 51 published studies on methylation analysis performed in cervical tissues and cells and proposed that the combination of DAPK1, CADM1, and RARB genes would appear the most promising methylated gene panel to obtain an appropriate performance for CC screening.
The recent meta-analysis by Xiong et al. [64], including 15 studies, suggested a strong association between DAPK1 promoter methylation and CC (pooled OR = 19.66; 95%CI = 8.72-44.31) indicating that DAPK1 promoter methylation may be a biomarker during cervical carcinogenesis.
Our study reports results of a more comprehensive meta-analysis and, taking into account that promoter methylation could be a tissue-specific event [65,66], provides a subgroup analysis by histological tumor type. The present meta-analysis concerned 20 unique articles and, on a total of 1092 from cancer patients and 837 control samples, reports a significant pooled OR of 21.20. Because of the moderate heterogeneity between studies, a sensitivity analysis and subgroup analyses by histological tumor types, sources of control samples and assays used to evaluate DAPK1 promoter methylation were performed. Interestingly, removing the most heterogeneous study [54], the association between DAPK1 promoter methylation and CC risk increased (OR: 24.13) and was confirmed in SCC and AC subgroups with a heterogeneity between study of I 2 = 48% and I 2 = 0%, respectively.
The gold standard method of promoter methylation evaluation was MSP, in which PCR products are run on a gel, and the results are reported as methylated or unmethylated at the target DNA sequence. Consequently, this method does not allow the identification of partial levels of methylation, a feature which is extremely relevant both biologically and clinically. Thus, qMSP has been developed in recent years to overcome this limitation of conventional MSP. In fact, qMSP is reported to be more specific and more sensitive than conventional MSP and allows for high-throughput analysis, making it more suitable as a screening tool [67][68][69]. In the present meta-analysis, considering these two detection methods, both subgroups reported a significant association between DAPK1 promoter methylation and CC. Although heterogeneity between studies stood moderately high in MSP subgroup (I 2 = 51%), the heterogeneity in qMSP subgroup decreased to I 2 = 0%.
Finally, the subgroup analysis by source of control sample revealed a significant association in both subgroups; the heterogeneity in NT and BCT subgroups was moderately low (I 2 = 48% and I 2 = 14%, respectively).
The present study has some limitations. The number of studies included in the meta-analysis is modest (n = 20). Moreover, since all studies included had a case-control design, it is not possible to clarify if DAPK1 promoter methylation is an early cancer-causing aberration or an effect of carcinogenesis. Accordingly, the potential of DNA methylation measurements requires validation in retrospective studies, but ultimately in large prospective clinical studies [70].
In addition, although sensitivity analysis and subgroup analyses were performed, the pooled estimates should be interpreted with caution, due to the moderate heterogeneity across studies. Finally, the small to moderate asymmetry in the funnel plots, suggests that publication bias cannot be completely excluded.
The usefulness of DAPK1 tumour suppressor gene hypermethylation as an epigenetic marker is under intense investigation in many different cancers, including CC and its precursor lesions and the present meta-analysis provides scientific evidences to this debate, showing a significant strong association between DAPK1 promoter methylation and CC. This result was confirmed independently by histological tumor type, method used to evaluate methylation and source of control samples.  Table. Subgroups analyses based on histological cancer type, method and source of control sample. SCC: Squamous Cell Carcinoma; AC: Adenocarcinoma; MSP: Methylation-