Association between Gene Promoter Methylation and Cervical Cancer Development: Global Distribution and A Meta-analysis

Cervical cancer is the second most frequent gynecologic malignancy in the world, ranking the fourth most frequently diagnosed cancer and the fourth leading cause of cancer-related deaths (1). According to GLOBOCAN, it accounted for approximately 570,000 new cases and more than 311,000 deaths in 2018, worldwide (1). Cervical cancer arises from the normal epithelium by distinct morphologic changes consisting of two major histotypes: squamous cell carcinoma and adenocarcinoma, accounting for 85%–90% and 10%–25% of cases, respectively (2). The progression to carcinoma is well documented and goes through well-defined preinvasive lesions (3). Cervical cancer epidemiologic and molecular studies have shown that the persistence of human papillomavirus (HPV) is the main etiologic factor for cervical carcinogenesis, but not exclusive (4). Additional genetic and epigenetic alterations are required for invasive cancer development from precancerous lesions (5). It is widely reported that epigenetic modifications, involving DNA methylation, histone modifications, nucleosome positioning, and noncoding RNAs, are associated with cancer development and progression (6, 7). Moreover, DNA methylation, considered as an important epigenetic mechanism for gene silencing, tends to accumulate with disease severity (8).

In the human genome, 60% of all promoters are associated with CpG islands and approximately 90% are unmethylated in the untransformed cells (9, 10). Conversely, in cancer, the methylation of CpG regions of many genes promoters is higher (11). Significantly, many of the inactivated genes are tumor suppressor genes and the inhibition of these genes expression by methylation is involved in cancer initiation, development, and progression (12).

Hypermethylation of the CpG islands of gene promoter is an important epigenetic mechanism for gene silencing, which may confer tumor cells of growth advantage (13). This epigenetic event leads to inhibition or inactivation of many cell signaling pathways. Indeed, hypermethylation of promoters may lead to the silencing of genes affecting important cellular signaling pathways, including apoptosis, cell cycle, DNA repair, cell adhesion, and detoxification (14). There is ample evidence indicating that hypermethylation of promoter regions of certain genes, such as DAPK1 (15), cyclins A1 (16), and FHIT (17), which are involved in cell cycle control and regulation, leads to apoptosis inhibition. Also, methylation of promoters of genes involved in cell adhesion, such as CDH1 and CADM1, leads to downregulation, thus leading to an increased risk of cervical cancer (18). Likewise, this methylation also silences other genes, such as RAR-β2, RASSF1, and APC, involved in cell signaling (19–21).

Scientific evidence has clearly reported that epigenetic alterations linked to hypermethylation affect the CpG islands on the human genome and these alterations are not randomly distributed during carcinogenesis, but they affect specific genes involved in specific signaling pathways (22). Therefore, assessment of the methylation status of such genes is of a great interest to set up molecular signatures that could be used for diagnosis, prognosis, and/or as therapeutic targets.

Thus, this study was planned to conduct a meta-analysis study to evaluate the hypermethylation status of some genes involved in cervical cancer development, to assess the correlation between genes hypermethylation and populations' origins and ethnicity, and to identify genes with the major impact on the onset of cervical cancer.

Systematic search strategies were conducted via a range of online literature databases prior to February 1, 2019, including PubMed, EMBASE, EBSCO, Web of Science, and Cochrane Library. We systematically reviewed the studies of promoter methylation of 41 genes (DcR1, DcR2, hTERT, P73, P16, PTEN, CDH1, APC, MGMT, Hmlh1, RASSF1, DAPK, FHIT, HIC1, RAR-β, TIMP2, TIMP3, CAV1, ESR1, miR-124, miR-34B, miR‐203, ER1, ZNF582, ADCYAP1, SFRP1, SFRP2, SYK, CADM1, PRDM14, P53, CASP8, CCNA1, DOC2B, MAL, SOX1, PAX1, LMX1A, NKX6-1, WT1, and ONECUT1) in cervical cancer.

The keywords combinations of Medical Subject Headings terms with all known genes name were (“gene name 1” or “gene name 2” or … or “gene name n”) and (“methylation” or “DNA methylation” or “promoter methylation”) and (“cervical cancer” or “cervical carcinoma” or “cancer of uterine cervix”).

All the methodologies and procedures adopted in this work followed the guidelines of the Preferred Reporting Items for Systematic Reviews and Meta-Analyses (PRISMA) statement, which was published previously (23). The PRISMA flow diagram was used to include or exclude any publication found in the literature concerning the selected keywords (Fig. 1). Accordingly, studies were included in our meta-analysis if they met the following selection criteria: (i) studies assessing the association of the selected gene methylation with cervical cancer, (ii) detailed information about the frequency of the selected gene methylation for both the cancer group and the normal control group was reported, (iii) the methylation status was assessed on human samples, and (iv) only the most recent study with the most informative results was included in case of multiple publications for the same group and using duplicated sample data.

Available data were extracted for the selected studies by two authors independently. For each gene, all information regarding author's name, publication year, patients' ethnicities, methods used in the methylation assessment, and the number of participants in the case and control groups was extracted (Supplementary Tables S1–S16). In case of disagreements, differences were discussed by the two authors. All the selected studies used in this meta-analysis are cited in the “References” section in the Supplementary Table S17.

The strength of association is represented as an overall OR with corresponding 95% confidence interval (CI). The heterogeneity of all eligible studies was quantified using the I² statistic and χ² tests, with corresponding P value. A Mantel–Haenszel model was applied for the meta-analysis. When there was a source of heterogeneity in the meta-analysis, a sensitivity analysis was performed to assess the stability of the results by omitting a single study in the meta-analysis iteration to determine the effect of the individual data on the overall pooled OR. The stability of our results was also tested by using one of the two models, DerSimonian–Laird or Mantel–Haenszel (24). Publication bias was quantitatively estimated with Begg linear regression test. Data were analyzed by the Meta-Essentials tool (version 1.4). Graphs were generated using GraphPad Prism 8.0 (GraphPad Software, Inc.).

A total of 194 studies were included in this meta-analysis to assess the methylation status of selected genes. In this meta-analysis, the interest was given to 14 genes known for their implication in affecting important cellular signaling pathways, including apoptosis, cell cycle, DNA repair, cell adhesion, and detoxification (14). Accordingly, apoptosis, considered as an important life process, is blocked through the hypermethylation of the promoters regions of some genes involved in cell-cycle control and regulation, including the DAPK1 (15), A1 cyclins (16), and the FHIT (17). Other genes have been reported to be likely involved in the development of cervical cancer when they are methylated. Among them, genes involved in cell adhesion, such as CDH1 and CADM1, are downregulated in cervical cancer cases following methylation of promoters (18). In addition, the RAR-β2, RASSF1, and APC, reported to play an important role in cell signaling, are affected during carcinogenesis by a genetic silencing event following methylation (19–21).

Likewise, the triggering of cervical cancer is also influenced by the methylation of the promoters of hypermethylated in cancer 1 gene (HIC1), T-lymphocyte maturation associated gene (MAL), p16, and paired box 1 (PAX1), which contribute mainly as zinc-finger transcription factor (25), in apical transport of membrane (26), cell-cycle regulation (27), and embryogenesis (28), respectively. Cervical carcinogenesis was also linked to the hypermethylation of O(6)-methylguanine-DNA methyltransferase (MGMT) and human mutL homolog 1 (hMLH1), which participate in DNA repair functions (27). The exclusion of other genes from the meta-analysis was considered as an essential criterion because of the lack and scarcity of studies on them.

The distribution of selected studies according to their geographic area is reported in Fig. 2. Results clearly showed that most studies were conducted on Asian populations, with a total of 96 studies (49.48%), and Caucasian populations, with 83 studies (42.78%). However, only 15 studies were conducted on African populations (7.73%).

Regarding the method of promoter methylation assessment, methylation-specific PCR (MSP), considered as the “gold standard method,” was used in most studies (67.01%), whereas 16.75% of studies used the quantitative methylation-specific PCR (qMSP). However, BSA and PCR were rarely used by researchers.

The distribution of methylation rate in genes promoters in cancer cases and control samples is reported in Fig. 3. Accordingly, the methylation status of the promoter region of PAX1 gene took interest exclusively in Asian population and was not reported in other regions. In contrast, the population of this region did not present methylation of hMLH1 gene promoter, which was methylated only on Caucasian and African populations. Quantitively, high methylation levels were observed for PAX1, CCNA1, and CADM1 genes and were reported in Asian and Caucasian populations, which also presented significative high methylation of MAL. DAPK1 gene exhibited a similar methylation rate between Asian, Caucasian, and African populations, but overall, the methylation status of selected genes was variable according to the study population. Of particular interest, the methylation status was mostly higher in cancer cases as compared with control samples, excepted for HIC1 and APC genes, assessed in African population, where cancer cases and normal samples shared the same methylation status (Supplementary Table S18).

A total of 21 studies, including 13 on Asian, six on Caucasian, and two on African populations, involving 1,141 cancer cases and 1,158 controls, were included in this analysis to assess the effect of p16 promoter hypermethylation on cervical cancer development. These studies were mostly conducted by using MSP techniques, except two that were done with nested methylation-specific polymerase chain reaction and pyrosequencing. There was a significant association between p16 promoter hypermethylation and increased cervical cancer risk, with an OR of 12.24 (95% CI, 6.38–23.46; Supplementary Fig. S1).

The study of PAX1 promoter gene was quantitatively synthesized across eight studies, including 271 cervical cancer cases and 355 controls. The result indicated that the frequency of methylated PAX1 in cervical cancer samples was significantly higher than that in controls (OR, 157.99; 95% CI, 24.83–1,005.47; Supplementary Fig. S2). A statistically significant heterogeneity across the included studies was observed (I² = 65.40%; P < 0.001). Thus, to confirm the existence of heterogeneity among all relevant studies, a meta-regression was performed showing that no single factor was responsible for the heterogeneity.

Overall, 15 studies, including 971 patients with cervical cancer and 1,126 controls, were performed to evaluate the relationship between RAR-β promoter methylation and cervical cancer development. These studies were conducted on Caucasian (7/15), Asian (6/15), and African (2/15) populations. In these studies, MSP technique was used in most studies (80%), qMSP, pyrosequencing, and MethyLight were used in one study each. Data analysis showed that RAR-β promoter methylation was associated with an increased cervical cancer risk, with a pooled OR of 5.36 (95% CI, 2.71–10.58), with a statistically significant heterogeneity across the included studies (I² = 73.63%; P < 0.000; Supplementary Fig. S3).

In this analysis, RASSF1 methylation status was quantitatively synthesized across 27 studies (19 on Asian population, seven on Caucasians, and one from Africa) including 1,775 cervical cancer samples and 997 controls. Random effects model was applied to calculate the association between RASSF1 gene promoter methylation and cervical cancer risk. Data analysis indicated that the frequency of methylated RASSF1 in cervical cancer samples was significantly higher than that in controls (OR, 7.23; 95% CI, 3.73–13.82). Moreover, a statistically significant heterogeneity across the included studies was reported (I² = 64.35%; P < 0.000; Supplementary Fig. S4).

Overall, only seven studies were reported to have evaluated the relationship between APC promoter methylation status and cervical cancer risk, including 441 patients with cervical cancer and 260 controls. Studies were mainly done on Asian populations; MSP was used in four studies and qMSP in three studies. These studies highlighted a nonsignificant association between APC promoter hypermethylation and cervical cancer risk, with an OR of 2.07 (95% CI, 0.84–5.14). Moreover, there was statistically significant heterogeneity across the included studies (I² = 46.29%; Supplementary Fig. S5).

A total of nine selected studies, including 339 cervical cancer specimens and 285 controls, were analyzed to evaluate the association between CADM1 promoter methylation and cervical cancer development. On the basis of the fixed effects model, the results showed that CADM1 promoter methylation was associated with an increased cervical cancer risk with a pooled OR of 21.73 (95% CI, 5.70–82.89; Supplementary Fig. S6). There was statistically significant heterogeneity across the included studies (I² = 51.01%).

Overall, seven studies have reported results on CCNA1 methylation status and cervical cancer risk. These studies were conducted on 565 samples, including 287 cervical cancer specimens and 278 controls. The heterogeneity of the selected studies was evaluated using the Mantel–Haenszel model (I² = 43.30%; χ² = 10.58; P < 0.000) and was statistically significant. On the basis of the fixed effects model, the results showed that CCNA1 promoter methylation was not significantly associated with cervical cancer status (OR, 71.10; 95% CI, 18.66–270.96; Supplementary Fig. S7).

The association of CDH1 methylation status with cervical cancer risk was performed in 21 studies involving 917 cervical cancer samples and 907 controls. Statistical analysis showed a significant heterogeneity across the included studies (I² = 66.51%; P < 0.000). Results indicated that the frequency of methylated CDH1 in cervical cancer samples was significantly higher than that in controls (OR, 8.76; 95% CI, 3.68–20.82; Supplementary Fig. S8).

A total of 1,836 cervical cancer specimens and 1,180 controls from 26 selected articles were used to evaluate the association between DAPK1 methylation status and cervical cancer development. On the basis of the fixed effects model, the results showed that DAPK1 promoter methylation was associated with an increased cervical cancer risk, with a pooled OR of 21.74 (95% CI, 12.29–38.46). Moreover, there was statistically significant heterogeneity across the included studies (I², 86.03%; P < 0.000; Supplementary Fig. S9).

To assess the association between FHIT promoter methylation status and cervical cancer risk, 15 studies were selected, with respect to inclusion and exclusion criteria, involving 980 cervical cancer cases and 679 controls. On the basis of the fixed effects model, results showed that FHIT promoter methylation was significantly associated with an increased cervical cancer risk, with a pooled OR of 7.42 (95% CI, 3.12–17.65; Supplementary Fig. S10). There was statistically significant heterogeneity across the included studies (I², 52.84%; P < 0.000).

In this meta-analysis, only eight studies were selected to assess the association between HIC1 methylation status and cervical cancer risk. These studies were conducted on 404 patients with cervical cancer and 336 controls, and were basically conducted using the MSP approach. Statistical analysis showed a significant heterogeneity across the included studies (I² = 53.78%; P < 0.000). Data analysis showed a great association between HIC1 promoter hypermethylation and increased cervical cancer risk, with an OR of 4.09 (95% CI, 1.60–10.42; Supplementary Fig. S11).

In this meta-analysis, only six studies satisfied the inclusion/exclusion criteria. These studies were conducted on 253 cancer cases and 225 controls and exhibited a good heterogeneity by using the Mantel–Haenszel model (I² = 62.37%; χ² = 13.29; P < 0.000). On the basis of the fixed effects model, results clearly showed that MAL promoter methylation status was not significantly associated with cervical cancer development (OR, 40.95; 95% CI, 5.07–330.94; Supplementary Fig. S12).

The association between MGMT methylation status and cervical cancer risk was evaluated in 1,145 cancer cases and 938 controls, from 15 selected studies published from 2001 to 2015. On the basis of the fixed effects model, results showed that MGMT promoter methylation was highly associated with an increased cervical cancer risk, with a pooled OR of 3.34 (95% CI, 1.58–7.03). Moreover, a statistically significant heterogeneity across the included studies was observed (I² = 68.04%; P < 0.000; Supplementary Fig. S13).

A total of nine studies, involving 441 cancer cases and 361 controls, were included in this meta-analysis. Random effects model was applied to calculate the association of methylated hMLH1 promoter with cervical cancer risk. The results clearly showed that the frequency of methylation of hMLH1 gene promoter in cervical cancer samples was significantly the same as that obtained with controls (OR, 1.19; 95% CI, 0.49–2.88). Moreover, there was lower heterogeneity across the included studies (I² = 4.72%; P < 0.324; Supplementary Fig. S14).

In this meta-analysis, some of the selected articles also included data on precancerous lesions, including low- and high-grade squamous intraepithelial lesions. Therefore, we analyzed the relationships between all promoter genes methylation status and clinicopathologic features [low-grade squamous intraepithelial lesion (LSIL)/high-grade squamous intraepithelial lesion (HSIL)], taking into consideration the geographic distribution and the methylation techniques.

For instance, and to have homogeneity between studies, some articles were omitted in the data analysis. This omission was not made in a random or targeted manner to avoid any influence on the results. Indeed, all the studies were compared and only the studies presenting unique data and those that do not correlate with global literature were omitted. In addition, we selected for this omission only studies with conclusions not highlighting the importance of methylation as a cervical cancer marker [e.g., studies showing that methylation is also overexpressed in controls (patients without cervical cancer)]. In addition, the method used was also taken into consideration, if it was classic and used exclusively by one team compared with what was used by other groups, the study may be subject to omission.

Accordingly, the study conducted by Gasperove and colleagues on Caucasian population was omitted (Supplementary Table S17). Indeed, in this study the methylation rate of the CDH1, HIC1, and DAPK1 promoters was abnormally higher in controls as compared with the other studies and showed no significant difference between the different clinicopathologic grades. The study conducted by Pathak and colleagues was also omitted because the CDH1 promoter methylation frequency was significantly higher in LSIL and HSIL than in cancer cases and control tissues (Supplementary Table S18). With the omission of the study conducted by Sun and colleagues (which highlighted the overexpression of methylation of DAPK1 and MGMT in controls), studies exploring the methylation status of DAPK1 and MGMT promoters regions in precancerous and cancerous cervical lesions were highly homogenous (Supplementary Table S18). Moreover, by deleting the Bai and colleagues' study, the results showed that FHIT promoter methylation was significantly correlated with histologic subtype (LSIL and HSIL; Supplementary Table S18).

Deletion of the study by Widschwendter and colleagues made the other studies, evaluating the methylation status of RASSF1 in precancerous lesions, cancer cases, and control tissues, highly homogenous (Supplementary Table S18).

Finally, for MAL gene promoter methylation, the omission of the study by Kim and colleagues (who concluded that methylation of MAL gene promoter was lower in cervical cancer, 4/48), made the studies analyzing LSIL cases highly homogenous (Supplementary Table S18).

Results are summarized in Table 1 and showed that many methylation statuses of promoter genes were correlated with clinicopathologic features. Of particular interest, methylation of p16, CDH1, RASSF1, RAR-β, DAPK1, FHIT, CADM1, MAL, CCNA1, and PAX1 genes promoters was significantly higher in cervical cases than in controls, and total events increased according to the lesion progression. Indeed, for these genes, total methylation events were lower in low grades and significantly higher in advanced stages (Table 1). For hMLH1 and APC genes, no significant difference was observed between cervical cases, including LSIL, HSIL, and cancer cases, with corresponding controls. Methylation events were similar between low grade, high grade, and cancer cases and lower as compared with corresponding controls. Finally, methylation statuses of MGMT and HIC1 genes promoters were slightly higher than corresponding controls, but no significant difference was observed between low grade, high grade, and cancer cases for both of them.

Begg funnel plot with pseudo 95% confidence limits and Egger test were applied to estimate the publication bias of the included literature. Begg and Egger tests revealed an obvious asymmetry in the overall meta-analysis for the most included genes (Supplementary Figs. S15–S17).

Begg funnel plot with pseudo 95% confidence limits and Egger test were also applied to estimate the publication bias with regard to clinicopathologic features (Supplementary Figs. S18–S27).

During cervical carcinogenesis, different genetic and epigenetic alterations occur in both the HPV and the host genomes (29). The main epigenetic alterations are the global DNA hypomethylation and hypermethylation of tumor suppressor genes involved in different cell signaling pathways: DNA repair, cell adhesion, nuclear receptors cell cycle, and checkpoint controls (30, 31). The great interest given to the evaluation of the methylation status of some genes was started with the discovery of demethylating agents able to reactivate the expression of hypermethylated and silenced tumor suppressor genes in tumor tissues (32). Since then, detection of epigenetic changes in cytologic smears had presented a potential interest for assessment of new molecular biomarkers that could be used for early detection, prognosis, and/or as vaccine or therapeutic targets.

In this meta-analysis, we have tried to provide a general and global picture about the methylation profile of genes involved in cervical cancer development and progression. A total of 194 worldwide studies were included in this study and analyzed using a model considering that promoter methylation could be a tissue-specific event, providing a subgroup analysis by histologic tumor type. In this study, the analysis has focused on the methylation status of the promoters of APC, CADM1, CCNA1, CDH1, DAPK, FHIT, HIC1, MAL, MGMT, hMLH1, p16, PAX1, RAR-β, and RASSF1 genes, assessed on 11,201 samples from patients with cancer and 10,184 samples from healthy controls.

Scientific evidence has shown that APC, FHIT, CDH1, and HIC1 are associated with Wnt signaling pathway, playing an important role in carcinogenesis, in particular for the development of uterine cancer. In normal cases, APC encodes for a tumor suppressor antagonist protein that interacts with the Wnt signaling pathway by inactivating β-catenin, which is known by its ability to bind to the TCF–LEF complex, inducing the expression of genes involved in carcinogenesis, such as VEGF, c-myc, and matrix metalloproteinase-7 (MMP-7; ref. 33). β-catenin–TCF complex can be also inactivated by E-cadherin or HIC1 protein, which are coded by CDH1 and HIC1 genes, respectively (34, 35). The methylation rate of the APC and HIC1 genes was almost the same when comparing between cancer samples and normal samples in African and Caucasian regions. On the contrary, in Asia, these genes were only hypermethylated in cervical cancer cases. The overmethylation of APC and HIC1 in normal cells was already reported by Reesink-Peters and colleagues and Waki and colleagues, showing that age has an effect on the methylation of certain tumor suppressor genes, such as APC (36, 37).

Moreover, they indicated that this related methylation is not immediately tumorigenic, but gradually participates in the functional inactivation of the gene. Therefore, we can easily assume that detection of APC and HIC1 methylation may help predict future risk of developing cancer in the patient, but will in no way confirm a positive diagnosis. Moreover, taken together, these results suggest that the use of methylation of the promoters of these genes may not be reliable for the diagnosis of cervical cancer, specifically in African and Caucasian patients. In addition, their use as a therapeutic target in these two populations can lead to undesirable effects.

FHIT also binds to β-catenin causing the transcription repression of certain genes, such as cyclin D1 (38). Dysfunction of FHIT induces overexpression of cyclin D1, which in turn leads to abnormalities in the cell cycle and an increased risk of cancer initiation (38). Although the HPV E6 protein induces an increase in the signaling response of the Wnt/β-catenin/TCF pathway (39), methylations occurring in this pathway, especially for APC, FHIT, CDH1, and HIC1, have been reported as essential factors which can influence the development of cervical cancer (17, 40, 41). In fact, cells can escape apoptosis with the overexpression of the Wnt/β-catenin pathway (42). The cells may also benefit from the inactivation of apoptosis when DAPK gene (apoptosis mediator) is methylated and gradually leads to the development of cervical cancer and metastasis (43).

Other genes associated with cell-cycle functions and apoptosis proprieties were also explored, including CCNA1, p16, and RASSF1 encoding for cyclin A1, p16^INK4A, and RASSF1 proteins, respectively. These proteins are essential for cells to regulate cell-cycle progression and to induce apoptosis (44, 45). Hypermethylation of the promoters of these genes is widely reported and occurs in both precancerous lesions and cancers. This epigenetic silencing of CCNA1, p16, and RASSF1 genes is associated with dysfunction of the cell-cycle progression and apoptosis avoidance, thereby promoting cervical cancer (46–50).

RAR-β is a member of the thyroid-steroid hormone receptor superfamily of nuclear transcriptional regulators that binds retinoic acid (the biologically active form of vitamin A) and also mediates cellular signaling during embryonic morphogenesis, cell growth, and differentiation (32, 51). Hypermethylation of RAR-β in many cancer types, including cervical cancer, is widely discussed (52, 53) to limit retinoic acids production, widely known for their involvement in tumor suppressor activity due to their antiproliferative and apoptosis-inducing effects (54).

In this meta-analysis, results clearly showed that methylation of p16, CDH1, RASSF1, RAR-β, DAPK1, FHIT, CADM1, MAL, CCNA1, and PAX1 genes promoters was significantly higher in cervical cases than in controls, and was also present in precancerous lesions. For the others, the methylation status of their promoters regions showed no significant difference between cervical features and was mostly similar to that reported in control specimens.

In this meta-analysis, the promoters of CADM1, MAL, and PAX1 genes were reported to be highly hypermethylated in low-grade and high-grade lesions, and cancer cases. These genes play an important role during cervical cancer development, they are involved in cell adhesion, apical membrane proteins transport, and cell differentiation, respectively (55–57). Hypermethylation of the promoters regions of these genes is widely reported to induce loss of their functions during cervical carcinogenesis (58–60). On the other hand, the interest was also given to MGMT and hMLH1 genes involved in DNA repair (61, 62). Hypermethylation of the promoters regions of these genes often results in epigenetic silencing which, in turn, suppresses the repair function, leading to an accumulation of genomic abnormalities and thereby, promotion of many kinds of cancers, including cervical cancer (62). In our study, it was shown that MGMT and hMLH1 genes' methylation is rarely studied in the world and showed no significant difference between cases and controls, limiting the interest on these genes as biomarkers for cervical cancer development and progression.

By analyzing all generated data in this meta-analysis, ethnical origin seems to play an important role in determining the methylation profile of investigated genes. Most studies were conducted on Caucasian (43%) and Asian (49%) populations, and limited studies have explored the association between epigenetic alteration and cervical cancer development in African population, which represents the main limitation of this study. Moreover, the methylation of some genes has not been reported in certain regions of the world. For instance, the methylation status of the PAX1 and hMLH1 genes is not explored in African/Caucasian populations and in Asia, respectively. These imply, subject to taking only the data analyzed in this study, the acceptability of the hypothesis that we have issued forward. The impact of ethnicity on epigenetic diversity was already discussed by Ahmad and colleagues, highlighting that epigenetic events are closely related to ethnical/racial differences affecting the overall cancer incidence and mortality. Moreover, it can be assumed that cultural and socio-economic inequalities, notably linked to low income or stress conditions, mainly influence the epigenome while favoring epigenetic modifications in the individual (63).

Of particular interest, some studies done in African populations are conducted on exfoliated cells, suggesting that difference in methylation events could be due to cell sampling rather than a difference between populations.

By omitting the heterogeneous studies, we performed subgroup analysis by histologic types. On the basis of the random effects model, results showed that the heterogeneity totally disappeared in LISL and HISL subgroups (Table 1). Thereby, we found that the hypermethylation rates in LSIL, HSIL, and cervical cancer specimens were gradually increased in some promoter genes, resulting in a growing trend in effects of p16, CDH1, RASSF1, DAPK1, FHIT, CADM1, MAL, CCNA1, and PAX1 hypermethylation on LSIL, HSIL, and cervical cancer development. These results suggest that p16, CDH1, RASSF1, DAPK1, FHIT, CADM1, MAL, CCNA1, and PAX1 promoters hypermethylation may be a promising epigenetic marker for the progression of cervical carcinogenesis. Hence, detecting hypermethylation in these genes' promoters may help clinicians to determine whether patients with cervical neoplasia are in disease regression, persistence, or progression. In this field, evaluation of the hypermethylation of these genes in LSIL cases will be of a great interest for a more effective clinical management of these patients.

This meta-analysis is very informative and showed that some studied genes exhibit a significant association between the methylation of their promoters and the risk of cervical cancer development. However, the main limitation of this investigation is the technique used for methylation status assessment. Indeed, most studies were performed using MSP, considered as the gold standard method (70.30%), in which PCR products are run on a gel and results are reported as methylated or unmethylated at the target DNA sequence. Moreover, this technique lacks to identify partial levels of methylation, which is considered as highly relevant feature both biologically and clinically. To overcome the limitation of this conventional method, qMSP was developed in recent years and was reported to be more specific and more sensitive, and allows for high-throughput analysis, making it more suitable as a screening tool.

In conclusion, this meta-analysis gave evidence that ethnicity plays a crucial role in the distribution of promoters' methylation across human genomes and highlights the association between the methylation status of the promoter region of some genes and promotion of cervical cancer in some cases. However, the technique used in the investigation can also generate questions related to the effectiveness of the detection of this methylation. Other analyses are, therefore, necessary to generate other data, making it possible to validate the highly promoter epigenetic markers for the diagnosis and development of therapeutic targets.

No disclosures were reported.

The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked advertisement in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.