Abstract
Background/Aim: Previous association studies have linked the functional miR-146a SNP rs2910164 with risk of digestive system cancer; however, the results of these studies are inconclusive and inconsistent. The objective of the following study is to provide an up-to-date and comprehensive meta-analysis of the association of miR-146a rs2910164 and digestive system cancer risk. Patients and Methods: We searched the PUBMED/MEDLINE and Cochrane/EBM databases. The following inclusion criteria were used for the study selection: i) Case–control studies; ii) studies with reported allelic frequency/genotype data; and iii) studies with reported association with risk of a digestive system cancer. The following exclusion criteria were used: Review article, meta-analysis, case report and case series; studies evaluating relationships of miR-146a rs2910164 with outcomes other than cancer incidence, such as cancer morbidity or mortality. Study quality was assessed using the Newcastle-Ottawa Scale and publication bias assessed graphically and numerically using Begg's funnel plot and Egger's regression test and rank test. Outcome measure was the pooled odds ratios under the allelic frequency, dominant, recessive, and over-dominant models. Subgroup analysis was conducted for country of study origin. Results: No association of miR-146a rs2910164 with overall risk of digestive system cancer was identified. However, subgroup analysis showed an association with overall risk in the European population in the over-dominant model. Furthermore, miR-146a rs2910164 was associated with reduced risk of gastric cancer in the dominant model (pooled odds ratio=0.91, 95% confidence intervaI=0.83-0.99), increased risk of colorectal cancer under the recessive model and reduced risk of colorectal cancer under the over-dominant model in the European population. No significant within-study or publication biases were detected. Conclusion: miR-146a rs2910164 was not associated with overall risk of digestive system neoplasms. However, the GG genotype and the CC genotype may be linked to higher risk of gastric cancer and colorectal cancer, respectively; and the CG genotype may be protective against digestive system neoplasms overall in the European population, especially for colorectal cancer.
- miR-146a
- rs2910164
- single nucleotide polymorphism
- esophageal cancer
- gastric cancer
- colorectal cancer
- hepatocellular carcinoma
- cholangiocarcinoma
- gallbladder cancer
- pancreatic cancer
Genome-wide association studies have identified microRNAs as key genes involved in carcinogenesis (1). MicroRNAs have pleiotropic properties and are involved in wide-ranging cellular processes, including the regulation of cell differentiation, proliferation, apoptosis, and inflammation. Single nucleotide polymorphisms (SNPs) that alter the function of microRNA genes may confer risk or protect against specific cancer types (1). Identification of such functional SNPs may provide important insight into understanding the pathogenesis of specific cancers and provide novel avenues for drug and biomarker development.
MicroRNA-146a (miR-146a) functions as part of a negative feedback loop in inflammation signaling by targeting downstream mediators such as interferon-associated kinase 1 (IRAK1) and tumor-necrosis factor alpha-associated factor 6 (TRAF6) (2). Three functional SNPs in miR-146a have so far been discovered: rs2910164 G>C is located in the pre-microRNA sequence of miR-146a and is associated with reduced levels of the mature microRNA (3-5); rs57095329 is found in the promoter of the gene and has also been found to be associated with lower levels of miR-146a products due to inhibited transcription and translation; and the intergenic SNP rs2431697 is also associated with reduced function of miR-146. These three SNPs have been extensively studied to reveal the involvement of miR-146a in various disease processes, including chronic autoimmune and inflammatory disorders, as well as cancer (6-8).
To our knowledge, numerous case–control studies and six meta-analyses have been published on the association between miR-146a rs2910164 and digestive system neoplasms, with inconsistent results. Two of these meta-analyses reported no association between miR-146a rs2910164 and overall risk of digestive neoplasms. An association was also not found when stratification by cancer type was performed in those two studies (9, 10). On the other hand, two other studies reported increased overall risk, while another study reported reduced overall risk of digestive neoplasms (11-13). The outcomes of these three studies remained significant for each cancer type even after stratification. Yet another study reported no association with overall risk but increased risk of colorectal cancer under the dominant model (14). The latter four studies with significant associations reported that the relationship remained significant after subgroup analysis in the Asian but not the Caucasian population. One common drawback of the above studies is their failure to analyze the data under each of the allelic frequency (G vs. C), dominant (GG vs. GC+CC), recessive (GG+GC vs. CC), and over-dominant models (CG vs. GG+CC). Therefore, our aim in this study was to provide a comprehensive up-to-date systematic review and meta-analysis of the association between miR-146a rs2910164 and cancer of the digestive system.
Patients and Methods
Protocol and registration. The following systematic review and meta-analysis was not registered.
Search strategy. The search included studies published from inception to July 29, 2019 in PUBMED/MEDLINE and the Cochrane Database/EBM. The following key words were used for the search: miR-146a, rs2910164, single nucleotide polymorphism, esophageal cancer, gastric cancer, colorectal cancer, hepatocellular carcinoma, cholangiocarcinoma, and pancreatic cancer, single nucleotide polymorphism(s). Additional efforts to identify articles included hand searching of journals and checking reference lists.
Study selection. Eligibility assessment was performed independently in a non-blinded standardized manner by two reviewers. Disagreements between reviewers were resolved by consensus. Inclusion criteria included the following: i) Case–control studies; ii) included genetic or allele frequencies; iii) study included risk data for digestive cancer. Exclusion criteria included the following: Review article, meta-analysis, case report and case series. In addition, studies evaluating relationships other than miR-146a rs2910164 and disease susceptibility, such as disease severity, were excluded.
Data collection process and data items. A data extraction form was developed a priori. Two reviewers in tandem conducted data extraction and final review was performed by the third reviewer. When overlapping data were identified, the most recent or comprehensive study was included. Disagreements were resolved by discussion between the reviewers. Information extracted from each study included the following: i) Author; ii) year of publication; iii) ethnicity of study subjects; iv) country of study origin; v) agreement with Hardy–Weinberg equilibrium (HWE); and vi) genetic/allelic frequency distribution of cases and controls.
Risk of bias in individual studies. Risk of bias in individual studies was assessed using the Newcastle-Ottawa Quality Assessment Scale (15).
Summary measures, synthesis of results, and assessment of bias across studies. Pooled odds ratios (ORs) and 95% confidence intervaIs (CIs) were calculated using a meta-analytical method that weighed the logarithm of the OR by the function of its variance for each study. ‘Exposure’ was defined as the C allele of miR-146a rs2910164. The random-effects model was used when heterogeneity across studies was significant, whereas the fixed-effects model was used when it was non-significant. Both per-allele and per-genotype models were used for OR calculations. Study-specific results were assessed using the forest plot in order to validate each study's results prior to calculating summary estimates. Agreement of genotypic frequencies with HWE in the control population in each study is indicated in Table I. Publication bias was assessed using Begg's funnel plot and Egger's test. A p-value less than 0.05 was considered statistically significant. Meta-analysis was performed using the package ‘metafor’ of the R-project.
Additional analyses. Subgroup analysis was conducted based on the country of study origin. Studies were categorized into either East Asian or European or Middle Eastern based on the country location. In this analysis, East Asian countries consisted of China, Japan, Korea, and Taiwan; European countries of Greece, Lithuania, and Latvia; and Middle Eastern countries of Turkey and Iran. One study conducted in multiple various European countries was included in the European subgroup.
Results
Study selection. The initial database search yielded 303 studies in total. After selection of studies based on title and abstract review, 42 remained for full review. Additionally, two studies were included via reference search and three studies were excluded based on the pre-specified criteria. Finally, 41 studies were included in the meta-analysis (Figure 1).
Study characteristics. A total of 41 publications were included in the final analysis. Of these, five were studies of esophageal cancer, 10 of gastric cancer, 10 of colorectal cancer, 13 of hepatocellular carcinoma, and there was one publication each of cholangiocarcinoma, gallbladder cancer, and pancreatic cancer. All of the studies were published between 2010 and 2016, except for two published in 2018 and 2019, respectively. The most common country of study origin was China and the stated ethnicity in these studies were almost always Han Chinese. The other studies were from various Asian and European countries, including Korea, Japan, India, Iran, Greece, Turkey, Lithuania, Latvia, Czech Republic, and Germany. HWE for genotypic frequencies was not reported for five studies; frequencies for all of the other studies, except one, were in HWE.
Results of individual studies. The results of each individual study were validated under the allelic frequency model. miR-146a rs2910164 was associated with reduced risk of esophageal cancer in one study; reduced risk of gastric cancer in two studies and increased risk in one study; increased risk of colorectal cancer in two studies and reduced risk in three studies; and reduced risk of hepatocellular carcinoma in one study. miR-146a rs2910164 SNP was not associated with cholangiocarcinoma, gallbladder cancer, or pancreatic cancer (Table I).
Characteristics of included studies on miR-146a rs2910164 in digestive cancer.
Synthesis of results. Meta-analysis was conducted for esophageal, gastric, colorectal, and hepatocellular carcinoma under the allelic frequency, dominant, recessive, and over-dominant models. miR-146a rs2910164 was not significantly associated with esophageal cancer, colorectal cancer, or hepatocellular carcinoma under any of the tested models but was associated with a reduced risk of gastric cancer (pooled OR=0.91, 95% CI=0.83-0.99) under the dominant model. SNP miR-146a rs2910164 was not associated with cholangiocarcinoma, gallbladder cancer, or pancreatic cancer under any of the tested models (Table II).
Risk of bias across studies. No significant publication biases were detected per Begg's funnel plots, rank tests and Egger's regression tests (overall: rank test p=0.5540, regression test p=0.5291; esophageal cancer: rank test p=0.8167, regression test pp=0.4874; gastric cancer: rank test p=0.6007, regression test p=0.7701; colorectal cancer: rank test p=0.7275, regression test p=0.6050; hepatocellular carcinoma: rank test p>0.99, regression test p=0.7207).
Flowchart of study selection.
Additional analysis. In the European population, miR-146a rs2910164 was associated with increased risk of colorectal cancer under the recessive model (pooled OR=1.44, 95% CI=1.01-2.05) and with reduced risk under the over-dominant model (pooled OR=0.7, [95% CI=0.62-0.98). Otherwise, subgroup analyses did not find any association between miR-146a rs2910164 and any of the cancer types for the assessed subgroup populations (Table III).
Risk of bias within studies. Individual studies were assessed for bias using the Newcastle-Ottawa Scale for case–control studies (Table IV). One study scored 9 points, eight studies 8 points, and the rest scored 7 points out of a total 10 points.
Discussion
Our meta-analysis found no association between miR-146a rs2910164 and overall risk of digestive system neoplasms. However, our results show that harboring the CC or CG genotypes may be protective against gastric cancer and that harboring the CC genotype may increase the risk of colorectal cancer, whereas having the CG genotype may protect against colorectal cancer in the European subgroup. Furthermore, the CG genotype was protective for overall risk of digestive system neoplasms in the European subgroup. The disparate results between our study and previous studies may be attributable to the characteristics of the included studies. Further validation in larger studies controlled for ethnicity may be required to assess whether the associations found in our analysis are truly meaningful.
The reason that the over-dominant model (CG vs. GG + CC) was studied was because previous studies have shown that a heterozygous genotype produces a phenotype different from that of the homozygous genotypes by producing two separate gene products. For a microRNA transcript, a heterozygous genotype can produce two separate transcripts with potentially different target affinities. This has been shown in a previous study where the CG genotype was associated with risk of papillary thyroid cancer (3).
Subgroup analysis by ethnicity is important in analyzing miR-146a rs2910164 since evidence from previous studies showed significantly different allelic frequencies depending on ethnicity (16, 17). In contrast to previous meta-analyses, our subgroup analysis was conducted by country of study origin and not by ethnicity for the following two reasons: i) A number of studies did not report the ethnicities of the sample population; ii) delineation into subgroups by geographical location is much clearer compared to delineation by ethnicity. By using country of study rather than ethnicity itself, we may have introduced certain biases, including overlooking the fact that certain countries may have ethnic heterogeneity so country of study may be a poor indicator of ethnicity. Furthermore, due to our subgrouping based on geographical location of study, we divided our groups into three: East Asian, European, and Middle Eastern. This should be taken into account, especially for the European countries, which tend to be more ethnically heterogeneous than countries such as China, Japan, and Korea.
The strength of this study is the following: the meta-analysis was conducted under all genotypic and allelic models to uncover all possible relationships between miR-146a rs2910164 and risk of digestive system neoplasm; and we conducted the most comprehensive analysis to date on SNP miR-146a rs2910164 and digestive system neoplasms having included two studies that had not been included in the previous meta-analyses. The notable weaknesses of our study are the following: i) The number of studies for the European and Middle Eastern subgroup populations are small when compared to the East Asian population, which means the pooled outcome may be strongly influenced by the results of a single study and is more susceptible to bias. ii) Studies which failed to report their congruence or which were incongruent with the HWE were included in our meta-analysis. The inclusion of these studies may be a source of bias and this should be noted. iii) Stratification by country of origin is an imperfect method for attempting to capture the differential genotypic and allelic distributions among different ethnicities. However, as more than a few studies did not report the ethnicities of their sample population, it was necessary that a separate marker be used as a surrogate for ethnicity.
In conclusion, there was no association between miR-146a rs2910164 and overall risk of digestive system neoplasm. However, the GG genotype and the CC genotype may increase the risk of gastric cancer and colorectal cancer, respectively; and the CG genotype may be protective against digestive system neoplasms overall, especially colorectal cancer in the European population.
Meta-analysis of included studies on miR-146a rs2910164 in digestive cancer.
Subgroup analysis by country of study.
Newcastle-Ottawa Scores for included studies on miR-146a rs2910164 in digestive cancer.
Footnotes
Authors' Contributions
Robin Park contributed via development of hypothesis, research design, data analysis, article writing, and proofreading. Laercio Lopes contributed via data analysis and proofreading. Anwaar Saeed contributed via research design, data analysis, article writing, and proofreading.
Conflicts of Interest
The Authors have no conflicts of interest to disclose.
- Received January 15, 2020.
- Revision received January 20, 2020.
- Accepted January 24, 2020.
- Copyright© 2020, International Institute of Anticancer Research (Dr. George J. Delinasios), All rights reserved
References
In this issue
Jump to section
Related Articles
Cited By...
- No citing articles found.