A systematic review and meta-analysis approach on diagnostic value of MLH1 promoter methylation for head and neck squamous cell carcinoma.

BACKGROUND: Head and neck squamous cell carcinoma (HNSCC) is the leading histological type among head and neck cancers. Several studies have explored an association between aberrant methylation of MutL homolog-1 (MLH1) promoter and HNSCC risk. We aimed to explore the associations between MLH1 promoter methylation and HNSCC by using a meta-analysis.
METHODS: Systematic literature search was conducted among PubMed, Google Scholar, Web of Science, and China National Knowledge Infrastructure, and Wanfang databases to retrieve relevant articles published up to June 30, 2018. A total of 12 studies were included in this meta-analysis (including 717 HNSCC and 609 controls).
RESULTS: The results demonstrated that MLH1 promoter methylation was notably higher in patients with HNSCC than in controls (odds ratios [ORs] = 2.52, 95% confidence intervals [CIs] = 1.33-4.79). Besides, MLH1 promoter methylation was not associated with tumor stage, lymph node status, smoking behavior, age, clinical stage, gender, and differentiation grade (all P > .05). The pooled sensitivity and specificity rates of MLH1 methylation for HNSCC were 0.23 (95% CI = 0.12-0.38) and 0.95 (95% CI, 0.82-0.99), respectively. The area under the receiver operating characteristic (ROC) curve was presented as 0.64 (95% CI = 0.60-0.68).
CONCLUSION: The results of this meta-analysis suggested that hypermethylation of MLH1 promoter was associated with HNSCC. Methylated MLH1 could be a potential diagnostic biomarker for diagnose of HNSCC.

Introduction

Head and neck squamous cell carcinoma (HNSCC), noted as the sixth most common cancer,[ is the leading histological type among head and neck cancers, accounting for up to 5% among the newly diagnosed malignancies throughout the world.[ Tobacco, alcohol, and human papillomavirus (HPV) infection were identified as risk factors in head and neck tumorigenesis.[ Despite the development in surgical and radiotherapy treatment, the survival rate of HNSCC patients is still relatively low, in which 5-year survival rate of patients was remained less than 50%.[ The survival rate tends to be poorer in advanced stage of cancer due to lack of molecular biomarkers for early diagnosis. Accordingly, it is of great importance to find out early diagnostic biomarkers for HNSCC patients.

Epigenetic regulation of promoter hypermethylation in tumor-suppressor genes (TSGs) has been emerged as an important cause in human carcinogenesis. DNA hypermethylation of various genes were frequently involved in cancer progression and development in HNSCC patients. For example,[ hypermethylation of O6-methylguanine-DNA methyltransferase (MGMT), death-associated protein kinase (DAP-K), and E-cadherin were detected in laryngeal and hypopharyngeal cancer.[DAPK1, leucine-rich PPR-motif containing (LRPPRC), RAS oncogene family member-6 C (RAB6 C), and zinc finger protein 471 (ZNF471) were confirmed in promoter regions of squamous cell carcinoma of the tongue.[ Researchers have revealed that polymorphism of methylenetetrahydrofolate reductase (MTHFR) affected p16 and MGMT methylation frequency in HNSCC.[

The MutL homolog-1 (MLH1), a member of DNA mismatch repair (MMR) gene, plays critical role in different types of cancer. Epigenetic silencing of MLH1 promoter methylation can cause mismatch repair (MMR) deficiency, which may cause insertion or deletion mutations in repeated sequences.[ The MLH1 promoter methylation has been reported as a well-established biomarker in several types of cancer, such as esophageal cancer,[ colorectal cancer,[ non-small cell lung cancer,[ gastric cancer,[ papillary thyroid cancer,[ and bladder cancer.[

To date, several studies have explored an association between aberrant methylation of MLH1 promoter and HNSCC risk. However, using MLH1 promoter methylation in the diagnosis of HNSCC has still remained inconclusive and inconsistent.

The aim of this meta-analysis was to clarify the diagnostic capability of MLH1 methylation status in the assessment of HNSCC patients. We also checked the relationship between MLH1 promoter methylation and the clinicopathological characteristics of HNSCC patients.

Materials and Methods

Literature search and selection criteria

Studies were gathered by thoroughly searching online through PubMed, Google Scholar, Web of Science, and China National Knowledge Infrastructure, and Wanfang databases. The search included all articles published before June 30, 2018. The following search terms and key words were used: “MLH1”, “hMLH1”, “MutL homolog-1”, “methylation epigenetic”, “head and neck cancer (HNSCC)”, “laryngeal squamous cell carcinoma (LSCC)”, “squamous cell carcinoma of the tongue (SCCT)”, “oral squamous cell carcinoma (OSCC)”, and “hypopharyngeal squamous cell carcinoma”. Three reviewers consulted all candidate articles independently. We also scanned review articles to identify eligible studies. The reference lists of the identified articles were manually reviewed to find out relevant articles. Furthermore, title and abstract of each study were scanned to exclude any irrelevant publications. The references of the articles were also screened for additional applicable papers. The investigators reach a consensus in the case of a conflict views by discussion.

The exclusion criteria were as follows:

Irrelevant reviews, letters, personal opinions, book chapters, and meeting abstracts.

Language restrictions.

Full paper copy was not available.

The authors have published articles using the same data.

Newcastle–Ottawa Scale (NOS) criteria were used to assess the quality of studies. Studies that scored seven or more were included.

The literature selection process of the eligible studies is shown in Fig. 1. This study was approved by ethics committee of Lihuili Hospital of Ningbo University.

Figure 1

Flow-chart of study selection.

Data extraction

As mentioned before, 3 reviewers reviewed the eligible articles independently. The following data were extracted as follows: first author's full-name, year of publication, country, methods for detecting the methylation status, hazard ratio (HR) and the corresponding 95% confidence interval (CI), histology of the sample, sample type in case and control group, sample number of the total people (in the case and control groups), clinicopathological characteristics, and the tumor location. Discrepancy in the data was solved and discussed by other two reviewers to reach a mutual agreement.

Statistical analysis

The pooled odds ratios (ORs) with 95% CIs were calculated to assess the association between MLH1 promoter methylation and risk of HNSCC, and the association of MLH1 promoter methylation with other clinical features was investigated as well. Fixed-effects model was adopted when there was no evidence of significant heterogeneity (P > .05 and/or I2 < 50%). On the other hand, a random-effects model was employed to evaluate the potential sources of heterogeneity. Moreover, a meta-regression and subgroup analysis was undertaken to explore the potential sources of heterogeneity. Sensitivity analysis was performed to check the stability of the publications. Subgroup analysis stratified by year of publication, sample size, race, detection method, control type was carried out to detect the potential source of heterogeneity. Begg test and Egger test were performed to evaluate the publication bias of the included studies. A sensitivity analysis was conducted to assess the influence of individual study excluded in the combined OR. Receiver operating characteristic (ROC) curves were plotted to identify diagnose value. Post-test probability was calculated with a presumed pre-test probability of 25%, 50%, and 75%. Statistical analyses were conducted by STATA-12.0 software (Stata Corporation, College Station, TX).

Results

Fig. 1 illustrates the detailed selection process of articles. Here, 208 relevant articles were identified for initial review by literature search. After screening according to the inclusion and exclusion criteria, information obtained from 12 studies was finally included in the meta-analyses. All the eligible articles were published in English. The 12 studies with 717 cases and 609 controls encompassed the years of publication from 2003 to 2015. The detailed characteristics of the included studies are summarized in Table 1. All the sample types were tissue and half of the studies chose heterogeneous tissues as control source. A total of 10 studies conducted methylation-specific polymerase chain reaction (MSP) to assess the gene methylation status, while other 2 studies used restriction enzyme MSP (RE-MSP) to evaluate the MLH1 methylation status.

Table 1

General characteristics of the eligible articles.

An association between MLH1 promoter methylation and HNSCC carcinogenesis

Overall, there were 12 studies with 280 patients, demonstrating MLH1 promoter methylation. There was a significant heterogeneity among the studies (I2 = 60.7%, P = .003). Thus, we used a random-effect model to evaluate the association between MLH1 promoter methylation and HNSCC patients. The methylation status of the MLH1 in HNSCC varied from 4.3% to 76.0% in the literatures. The frequency of MLH1 promoter methylation in HNSCC was 29.3%, which was significantly higher than that in normal controls (OR = 2.521, 95% CI = 1.327–4.788). The pooled ORs and 95% CIs are shown in Fig. 2.

Figure 2

The pooled frequency of MLH1 promoter methylation in HNSCC patients. The estimates for MLH1 methylation frequency were associated with HNSCC in the meta-analysis.

Meta-regression analysis was employed to identify the source of heterogeneity. However, the results suggested that the accuracy required for detecting MLH1 was not significantly affected by the covariates of year of publication, sample size, race, detection method, and control type (see Table 2). On the other hand, we performed a subgroup meta-analysis by published year of studies, sample size, detected method, ethnicity and control source to investigate the potential reasons of heterogeneity. As shown in Table 3, in the control source analyses, the pooled OR for MLH1 methylation in HNSCC compared with heterogeneous controls was 5.349 (95% CI: 0.481–59.496, P = .172), while that was 1.932 (95% CI: 1.363–2.739, P < .001) in the autologous control group. Furthermore, subgroup meta-analysis by studies published after 2010 indicated no significant association between MLH1 methylation and risk of HNSCC with pooled OR = 3.505 (95% CI: 0.811–15.142, P = .093). The results of this subgroup analysis showed that MLH1 hypermethylation was significantly associated with other subgroups (P < .05).

Table 2

Meta-regression analysis with the aid of year of publication, sample size, race, detection method, and control type.

Table 3

Subgroup analyses of MLH1 promoter methylation in HNSCC patients.

The sensitivity analysis was carried out to assess the stability of the conclusions. As illustrated in Fig. 3, the omission of individual studies did not significantly change the pooled OR, which demonstrated the stability of our analyses.

Figure 3

Begg funnel plot and Egger plot of publication bias for MLH1 in HNSCC patients.

Publication bias

We used Begg test and Egger test to assess the publication bias of the 12 studies (Fig. 4). The publication bias was not significantly associated with Begg test (P = .732) and Egger test (P = .254).

Figure 4

Sensitivity analysis of pooled ORs for investigating the association between MLH1 methylation and HNSCC.

Correlation of MLH1 promoter methylation with clinicopathological features of HNSCC

Further analysis of tumor stage, lymph node metastasis, smoking behavior, age, clinical stage, gender, and differentiation grade was conducted. Table 4 demonstrates that no significant association was observed in relation to tumor stage, lymph node metastasis, smoking behavior, age, clinical stage, gender, and differentiation grade in HNSCC patients (OR = 0.484, 95%CI = 0.151–1.556, P = .223; OR = 1.053, 95% CI = 0.587–1.89, P = .863; OR = 1.483, 95% CI = 0.615–3.574, P = .38, OR = 0.471, 95%CI = 0.154–1.434, P = .185; OR = 2.501, 95% CI = 0.368–17.012, P = .349; OR = 0.607, 95% CI = 0.201–1.833, P = .376, OR = 0.90, 95% CI = 0.352–2.299, P = .825; respectively).

Table 4

MLH1 promoter methylation and clinicopathological features of HNSCC patients.

ROC

Subsequently, we estimated the diagnostic values of MLH1 promoter methylation in HNSCC patients. Summary of receiver operation characteristic (SROC) curve was plotted. The summery sensitivity, summery specificity, and area under the curve (AUC) value of MLH1 promoter methylation in HNSCC patients versus healthy individuals were 0.23, 0.95, and 0.64, respectively.

Fagan plot analysis

We performed Fagan plot analysis from 12 case-control studies to evaluate the clinical utility of MLH1 promoter methylation in HNSCC. As illustrated in Fig. 6, when pre-test probabilities of 25%, 50%, and 75% were assumed, the positive post-probabilities were 54%, 81%, and 91%, and the negative post-probabilities were 22%, 45%, and 72%, respectively.

Figure 6

Fagan plot analysis to evaluate the clinical utility of MLH1 promoter methylation in HNSCC.

Discussion

MLH1 was reported as a TSG in various cancers.[ Increasing lines of evidence suggested that MLH1 plays a critical role in genome stability system by correcting replicative DNA polymerase errors or mismatched genes. Inactivation of MLH1 increased microsatellite instability.[ It has been reported by several scholars that aberrant MLH1 methylation appears as a major mechanism in HNSCC.[ Promoter methylation is a well-known epigenetic process that has been implicated in various human cancers, that may affect apoptosis, proliferation, and cell adhesion process.[ It has been reported that MLH1 protein expression was decreased in HNSCC patients compared with normal squamous epithelium due to promoter hypermethylation of MLH1 gene.[ Promoter hypermethylation of MLH1 was found to be associated with cancer regional lymph node invasion, in laryngeal squamous cell carcinoma.[ However, the relationship between MLH1 promoter methylation and HNSCC has still remained inconsistent. The goal of this article was to assess the relationship between the methylation status and HNSCC, in addition to explore the diagnostic value of MLH1 methylation status in HNSCC patients.

After careful screening, the remaining 12 articles were allocated to be analyzed in this study. In our study, 717 cases and 609 controls were ultimately used for pooled analysis. Our results indicated that the frequency of methylation of MLH1 promoter was significantly higher in HNSCC patients than in normal controls, which supported hypermethylation of MLH1 associated with an increased risk of HNSCC. Due to significant heterogeneity among the included studies, we performed meta-regression and subgroup analysis to investigate the potential sources of heterogeneity. Meta-regression analysis revealed year of publication, sample size, race, detection method, and control type contributed to the heterogeneity. However, in subgroup analysis of studies used for autologous controls, a more significant association was found between MLH1 methylation and risk of HNSCC with a lower heterogeneity compared with heterogeneous control studies. On the other hand, the subgroup meta-analysis based on the year of publication revealed that no significant association of MLH1 methylation was available in studies published after 2010, however, those studies published before 2010 presented a significant correlation between hypermethylated MLH1 and HNSCC. We noted that for studies published after 2010, MSP was mainly used to check the methylation frequency. MSP is known as a high sensitive technique for assessment of methylation accompanied with high false positive results because of incomplete bisulfite conversion. In addition, in our subgroup analysis, detection performed by MSP method revealed a higher OR with a higher heterogeneity than other methods. The subgroup analysis by ethnicity showed that Caucasian population had a higher OR than the other population, which suggested that Caucasian population may be more susceptible to MLH1 promoter methylation. Additionally, in subgroup analysis grouped by sample size, studies with sample size less than 60 were significant in terms of heterogeneity, suggesting that the studies with small sample size might be a potential resource for heterogeneity. According to the results of sensitivity analysis, we found our meta-analysis as stable and reliable.

We also found that lymph node metastasis, tumor stage, differentiation grade, and clinical stage were closely associated with the outcomes of patients with HNSCC.[ We sought to determine the associations of methylation with those clinical parameters. However, no significant association was observed between MLH1 promoter methylation and those clinicopathological features, in which it may be related to the small sample size. Therefore, further studies with larger sample size need to be conducted.

It is also essential to know the diagnostic value of the MLH1 methylation status in HNSCC. Hence, in the present study, the ROC curves were plotted, and Fagan plot analysis was performed. The pooled ROC curve (Fig. 5) indicated that MLH1 hypermethylation yielded an AUC value of 0.64 (95% CI: 0.60–0.68) in distinguishing HNSCC from normal control with a sensitivity of 0.23 and a specificity of 0.95. Besides, Fagan plots analysis indicated that the probability of HNSCC diagnosis was significantly elevated by the detection of hypermethylation of MLH1. These results indicated that hypermethylation of MLH1 could be a potential biomarker with low sensitivity for diagnosis of HNSCC. However, we concluded that MLH1 hypermethylation may be a specific method in diagnosis of HNSCC, which could be used as combined with other methods to improve the diagnostic value of a disease.

Figure 5

SROC curve.

This meta-analysis has some limitations that must be taken into consideration. First, the studies were from different countries, and population differences were observed regarding the status of MLH1 methylation. Second, MSP and RE-MSP methods were employed to detect the methylation status in the studies. However, MSP has been commonly considered as non-quantitative method that may increase false positive results.[ Third, due to lack of sufficient information, we did not check the methylation status in other clinical features, such as smoking history, alcohol history, and HPV infection. Lack of original data also limited our further evaluation of clinicopathological features and methylation status of MLH1 promoter.

Conclusions

This research indicated that aberrant methylation of MLH1 promoter was significantly associated with tumor progression in HNSCC patients, and it could be a potential tumor-specific biomarker for diagnose of HNSCC. However, further verification is required by providing large sample studies.

Acknowledgments

Thanks are due to the professional English language copy-editing provided by American Journal Experts Company.

Author contributions

Conceptualization: Zhenhua Wu.

Data curation: Qun Li, Zhisen Shen, Hongxia Deng, Yi Shen, Chongchang Zhou.

Investigation: Qun Li, Hongxia Deng.

Methodology: Zhisen Shen, Yi Shen, Chongchang Zhou.

Project administration: Zhenhua Wu.

Software: Yi Shen.

Writing – original draft: Qun Li.

Writing – review & editing: Qun Li, Jinjiong Hong, Chongchang Zhou.