Figures
Abstract
Object
To combine the data from previously conducted studies about the associations between miR-608 rs4919510 polymorphism (C>G) and breast cancer risks.
Methods
According to the Preferred Reporting Items for Systematic Reviews and Meta-Analyses (PRISMA) guidelines, we conducted a systematic review of the related literatures searched from PubMed, Embase, Cochrane Library, Web of Science, and China National Knowledge Internet (CNKI) (time: ~ December 2016). Using DerSimonian-Laird random-effects models [Pooling Model: Mantel Haenszel (MH)], odd ratios (ORs) with 95% confidence intervals (95% CIs) were estimated in the allele model, homozygote model, heterozygote model, dominant model and recessive model. Heterogeneity was analyzed using Labbr plots and I2 statistic. Publication bias was analyzed using contour-enhanced funnel plots.
Results
We included 5 eligible studies with 7948 patients. The ORs and their 95% CIs in the 5 genetic models mentioned above were 1.009 (95% CI: 0.922, 1.104; p = 0.847), 1.098 (95% CI: 0.954, 1.264; p = 0.194), 1.076 (95% CI: 0.956, 1.211; p = 0.227), 1.043 (95% CI: 0.880, 1.236; p = 0.628), 1.007 (95% CI: 0.906, 1.118; p = 0.899), respectively.
Citation: Wang J, Kong X, Xing Z, Wang X, Zhai J, Fang Y, et al. (2017) A meta-analysis: Is there any association between MiR-608 rs4919510 polymorphism and breast cancer risks? PLoS ONE 12(8): e0183012. https://doi.org/10.1371/journal.pone.0183012
Editor: Graham R. Wallace, University of Birmingham, UNITED KINGDOM
Received: August 10, 2016; Accepted: July 14, 2017; Published: August 22, 2017
Copyright: © 2017 Wang et al. This is an open access article distributed under the terms of the Creative Commons Attribution License, which permits unrestricted use, distribution, and reproduction in any medium, provided the original author and source are credited.
Data Availability: All relevant data are within the paper and its Supporting Information files.
Funding: This research was supported by Beijing Municipal Natural Science Foundation (Y.F., no. 7162163), Medical and health science and technology innovation project of Chinese Academy of Medical Sciences (J.W., 2016-12M-1-007), and National Natural Science Foundation of China (J.W., General Program:81372829). The funders had no role in study design, data collection and analysis, decision to publish, or preparation of the manuscript.
Competing interests: The authors have declared that no competing interests exist.
Introduction
As one of the malignancies with the highest incidence and mortality rates globally, breast cancer accounts for more than a million cases every year[1, 2]. In North America, the incidence of breast cancer is the highest among women, and its mortality ranks the second in all cancer deaths in female[1]. Although causes of breast cancer are not yet completely understood, genetic factors are considered to play pivotal roles in the pathogenesis of this malignancy[3].
MicroRNAs, as gene expression regulators, regulate a variety of biochemical processes, such as apoptosis, proliferation, metabolism, cellular differentiation, and cancer development[3–6]. It has been proposed that rs4919510 C>G variant in miR-608 can alter its binding to target genes. MiR-608 expectedly targets growth hormone receptor (GHR), interleukin-1 alpha (IL1A), insulin receptor (INSR), and TP53[7–9]. Several studies examined the impacts of miR-608 rs4919510 C>G on breast cancers risks, but the results were inconsistent[8, 10–13]. In Huang's report, the results showed that the single nucleotide polymorphism (SNP) could alter the secondary structure of primary miR-608[11]. A single nucleotide variation located at introns has also been experimentally shown to change DNA and RNA secondary structures, and consequently associate with gene expressions and diseases. A single study might not be able to conclusively confirm the correlations, especially if the study is of small-sample-size. In 2013, Hu et al. conducted a meta-analysis regarding 8 precursor-miRNA SNPs (including miR-608 rs4919510) in 8 common cancers (including breast cancer), and did not find significant associations between miR-608 and cancers[14]. However, in their study, the authors generally analyzed all caner types together instead of specifically discussing breast cancer. Besides, the studies included in their study were limited and several more recent studies have been finished up to now. To further elucidate the exact effects of miR-608 rs4919510 polymorphism on breast cancer risk, we accumulate data from different case control studies and perform this meta-analysis to make an evaluation.
Methods
Publication search and selection criteria
Two authors (XK and JW) independently searched the database of Embase, PubMed, Cochrane Library, Chinese National Knowledge Infrastructure (CNKI), and Web of Knowledge (time: ~ December 2016). Search terms used separately or in combination were: “breast carcinoma or breast cancer” and “rs4919510 or miR-608” and “mutation or variant or polymorphism”. Detailed searching strategies with the start and end date of searches were listed in S1 Table. For example, for PubMed database, the search strategy “breast[Title/Abstract] AND ((MiR-608[Title/Abstract] OR MicroRNA-608[Title/Abstract]) OR rs4919510[Title/Abstract])” was adopted and 7 articles were obtained. We reviewed related references to find out other potentially eligible studies. The exclusion criteria and inclusion criteria are listed in Table 1.
Data extraction
According to the inclusion criteria set in Table 1, 2 independent authors (JW and XK) reviewed and extracted the needed data and information from the included articles. The following data were extracted: author name, publication year, country, ethnicity or race (Asian, Caucasian or others), genotyping methods, total number of controls and cases, number of controls and cases with rs4919510 polymorphism, number of cases and controls with C/C, C/G, and G/G genotypes, control source (hospital-based or population-based), and P value for Hardy-Weinberg equilibrium (HWE).
Methodological quality assessment
According to the methodological quality assessment scale (see Table 2), which was adjusted from a previous publication by Guo et al. in PLos One in 2012[15], two authors (XK and JW) independently estimated the quality of the included studies. Disagreement would be solved by discussion. In this methodological quality assessment scale, 5 items, including quality control of genotyping methods, source of controls, sample size, cases representativeness and HWE were carefully checked. The quality scores range from 0 to 10. The higher the score is, the higher the quality of the study.
Statistical analysis
Our study was based on the PRISMA checklists (S2 Table) and the meta-analysis-on-genetic-association-studies-form (S3 Table)[16]. HWE in each study was assessed, followed by the calculations of ORs with 95% CIs to reflect the correlation strength between rs4919510 polymorphism and the risk of breast cancer. The pooled ORs were calculated and used for comparisons respectively in allele model (G vs. C), homozygote model (GG vs. CC), heterozygote model (CG vs. CC), dominant model (CG+GG vs. CC), and recessive model (GG vs. CC+CG). The Labbe plot, Cochran's Q-test, and I2 statistic (Table 3) were used to access the heterogeneities [17]. Since fixed effect models might underperform in the presence of any heterogeneity[18], while DerSimonian-Laird random-effects models are more conservative and able to provide better estimates with wider confidence intervals, we adopted the latter [Pooling Model: Mantel Haenszel (MH)] for all the analyses of all the 5 genetic models[18]. To estimate the stabilities of the pooled results, probabilistic sensitivity analyses of meta-analysis (explanation in Table 3) were made[19]. By contour-enhanced funnel plots (explanation in Table 3), we accessed possible publication biases.
P < 0.05 reflected statistical significance. The statistical analyses were made by Stata 13.0 (StataCorp LP, College Station, TX, USA) software. The Stata commands is metan.
Results
Search results and characteristics of the studies
According to PRISMA statement, a study selection flowchart was reported in Fig 1.
From: Moher D, Liberati A, Tetzlaff J, Altman DG, The PRISMA Group (2009). Preferred Reporting Items for Systematic Review s and Meta-Analyses: The PRISMA Statement. PLOS Med 6(6):e1000097. Doi:10.1371/journal.pmed1000097. For more information, visit www.prisma-statement.org.
A total of 35 studies were identified: 7 in Pubmed, 7 in Embase, 0 in Cochrane Library, 14 in Web of Science and 7 in CNKI (S1 Table). Finally, a total of 5 articles involving 7948 patients were included[8, 10, 11, 13, 20]. Two studies were on the basis of Caucasian backgrounds and were done in Iran (352 cases) and Chile (1247 cases)[20]. Three studies were on the basis of Asian backgrounds and were done in China (6349 cases in total)[10, 11, 13]. Four studies were written in English[8, 10, 11, 20] and 1 was in Chinese[13]. Breast cancers were all confirmed by histopathologic examinations. In all included studies, genotype distributions of rs4919510 (C > G) in the controls were consistent with HWE. A variety of genotyping methods were applied including SNPstream[11, 13], PCR-RFLP[8], TaqMan Genotyping Assay[20] and Sequenom MassARRAY RS100[10]. Genomic miRNA was isolated from blood samples in all included studies. Controls were matched in terms of age. Four studies were population-based[8, 11, 13, 20] and 1 was hospital-based[10]. Excluded studies and the rational for the exclusion were listed in Table 4. The characteristics including the basic information of the literatures, the original data, P for HWE, and the methodological quality assessment results of the included literatures were shown in Table 5.
Meta-analysis results
The main results including heterogeneity tests, effect models adopted accordingly, and the pooled OR with 95% CI and P value of this meta-analysis were shown in Table 6. The Labbe plots for allele model, heterozygote model and dominant model were shown in Fig 2-A, 2B and 2C. For overall studies, there were no statistically correlations between miR-608 rs4919510 polymorphism and decreased or increased breast cancer risks in all the 5 models (allele model: OR 1.009, 95% CI 0.922, 1.104; p = 0.847; Fig 3-A; homozygote model: OR 1.098, 95% CI 0.954, 1.264; p = 0.194; Fig 3-B; heterozygote model: OR 1.076, 95% CI 0.956, 1.211; p = 0.227; Fig 3-C; dominant model: OR 1.043, 95% CI 0.880, 1.236; p = 0.628; Fig 3-D; recessive model: OR 1.007, 95% CI 0.906, 1.118; p = 0.899; Fig 3-E).
Labbe plots in allele model (A), heterozygote model (B), and dominant model (C). Sensitivity analysis in allele model (D), heterozygote model (E), and dominant model (F). Contour-enhanced funnel plots in allele model (G), heterozygote model (H), and dominant model (I).
A: allele model, random effect model; B: homozygote model, random effect model; C: heterozygote model, random effect model; D: dominant model, random effect model; E: recessive model, random effect model.
Sensitivity analysis and publication bias
Sensitivity analysis demonstrated that the pooled ORs were not affected by deleting every single study (Fig 2-G, 2H, 2I, 2J and 2K). The contour-enhanced funnel plots revealed that the studies had missing areas of high statistical significance (in the right-hand side of the plot), indicating no publication bias in this study (Fig 2-L, 2M, 2N, 2O and 2P).
Discussion
Recently, gene polymorphisms which may contribute to the tumorigenesis of breast cancer have attracted more and more scholars’ attention[21]. Some genes or RNA polymorphisms have already been proposed to increase the susceptibility of breast cancers[22]. The number of studies related to breast cancer-related polymorphisms show a general tendency to increase yearly. A timeline of the literatures was shown as Fig 4, which was generated through the following website: http://www.gopubmed.com.
Fig 4 was generated through GoPubMed (website: http://www.gopubmed.com). GoPubMed is a knowledge-based search engine for biomedical texts. The technologies used in GoPubMed are generic and can in general be applied to any kind of texts and any kind of knowledge bases. The system was developed at the Technische Universität Dresden by Michael Schroeder and his team at Transinsight. Creation steps for this timeline: import search items to the Search Box at the home page, then click “Statistics” and download related statistical charts including the timeline and map.
Recently, rs4919510 polymorphism in miR-608 has been reported to predict clinical outcomes for cancer patients in different cancer types. Hashemi et al. evaluated the impact of miR-608 rs4919510 C>G variant on the breast cancer risk[8]. They found that GC genotype decreased breast cancer risks significantly (OR = 0.49, 95% CI 0.28, 0.88; p = 0.018) compared to CC genotype. Furthermore, the G allele decreased the breast cancer risk (OR = 0.53, 95% CI 0.30, 0.92; p = 0.024)[8]. In Huang et al.’s study, miR-608 rs4919510 also affect HER2-positive breast cancer risks and tumor proliferations[11]. However, in Dai et al.’s study, for miR-608 rs4919510, no significant correlations were detected in the genetic comparison models[10]. In 2013, Hu et al. conducted a meta-analysis regarding 8 precursor-miRNA SNPs (including miR-608 rs4919510) in 8 common cancers (including breast cancer), and did not find significant associations between miR-608 and cancers[14]. However, in their study, the authors generally analyzed all caner types together instead of specifically discussing the breast cancer. Besides, the articles included in their study are limited and several more recent studies have been finished up to now.
A single study cannot be sufficient enough to confirm the correlation between miR-608 rs4919510 polymorphism and breast cancer risks convincingly, especially for small-sample-size studies. Given this, Pubmed, Embase, Cochrane Library, Web of Science, and CNKI databases were combined to further analyze the associations. The results of our study failed to demonstrate any significant correlation. This analysis is the most updated one to provide an evaluation of the correlations between miR-608 rs4919510 polymorphism and breast cancer risks.
On a contour-enhanced funnel plot, “if the area where studies are perceived to be missing are areas of high statistical significance (the right part of the funnel plot), then publication bias isn’t the cause of funnel asymmetry”[23]. In the present meta-analysis, we found no publication bias.
Several limitations existed in our study: (1) Included studies were relatively insufficient to do subgroup-analyses; (2) The effect of gene-environment interactions and gene-gene interactions was not emphasized; (3) More accurate ORs should be adjusted by patient factors such as gender, age, living styles, medication consumption and other exposure factors; (4) Only published articles were included, the unpublished and ongoing studies could convert our result; (5) When the 95% confidence intervals around I2 are wide, inferences about the heterogeneity extent should be cautious, thus, calculating the confidence intervals for I2 is important for estimating the heterogeneity if the number of included studies are large enough. However, since the present study is only a small meta-analysis, and we used DerSimonian-Laird random-effects models for all the analyses of all the 5 genetic models, we did not calculate the confidence intervals. After all, Cochran Q (i.e. chi-square) is somewhat underpowered to detect heterogeneities, especially for small meta-analyses; thus, we only used the I2 statistic as a rough reflection; (6) Regarding heterogeneity estimates, all these estimates are very likely off, especially for small meta-analyses, and we should be wary about homogeneity assumptions. In this smaller meta-analysis, we failed to identify any heterogeneity, which might exist. In addition, since fixed effect models might underperform in the presence of any heterogeneity, while DerSimonian-Laird random-effects models are more conservative and able to provide better estimates with wider confidence intervals, we adopted the latter for all the analyses of all the 5 genetic models; (7) Publication bias tests and plots only relevant if >10 studies are included otherwise underpowered to detect much and tend to lead to conclusions that are not justified. In the present study, we don’t have enough studies to assess, which is another limitation.
Conclusions
In conclusion, our results suggested that miR-608 rs4919510 polymorphism may not be associated with the susceptibility of breast cancer.
Supporting information
S1 Table. Searching strategies and results for different databases.
https://doi.org/10.1371/journal.pone.0183012.s001
(DOC)
S3 Table. meta-analysis-on-genetic-association-studies-form.
https://doi.org/10.1371/journal.pone.0183012.s003
(DOC)
Acknowledgments
We would like to thank our colleagues at the Department of Breast Surgery, Chinese Academy of Medical Sciences Cancer Hospital.
References
- 1. Siegel RL, Miller KD, Jemal A. Cancer statistics, 2016. CA Cancer J Clin. 2016;66(1):7–30. pmid:26742998.
- 2. Chetlen A, Mack J, Chan T. Breast cancer screening controversies: who, when, why, and how? Clin Imaging. 2016;40(2):279–82. pmid:26093511.
- 3. Chen QH, Wang QB, Zhang B. Ethnicity modifies the association between functional microRNA polymorphisms and breast cancer risk: a HuGE meta-analysis. Tumour Biol. 2014;35(1):529–43. pmid:23982873.
- 4. Sun K, Lai EC. Adult-specific functions of animal microRNAs. Nat Rev Genet. 2013;14(8):535–48. pmid:23817310.
- 5. Krek A, Grun D, Poy MN, Wolf R, Rosenberg L, Epstein EJ, et al. Combinatorial microRNA target predictions. Nat Genet. 2005;37(5):495–500. pmid:15806104.
- 6. Li WJ, Wang Y, Gong Y, Tu C, Feng TB, Qi CJ. MicroRNA-124 rs531564 Polymorphism and Cancer Risk: A Meta-analysis. Asian Pac J Cancer Prev. 2015;16(17):7905–9. pmid:26625819.
- 7. Liu H, Zhou Y, Liu Q, Xiao G, Wang B, Li W, et al. Association of miR-608 rs4919510 polymorphism and cancer risk: a meta-analysis based on 13,664 subjects. Oncotarget. 2016. pmid:27223084.
- 8. Hashemi M, Sanaei S, Rezaei M, Bahari G, Hashemi SM, Mashhadi MA, et al. miR-608 rs4919510 C>G polymorphism decreased the risk of breast cancer in an Iranian subpopulation. Exp Oncol. 2016;38(1):57–9. pmid:27031722.
- 9. Ryan BM, McClary AC, Valeri N, Robinson D, Paone A, Bowman ED, et al. rs4919510 in hsa-mir-608 is associated with outcome but not risk of colorectal cancer. PLoS One. 2012;7(5):e36306. pmid:22606253.
- 10. Dai ZM, Kang HF, Zhang WG, Li HB, Zhang SQ, Ma XB, et al. The Associations of Single Nucleotide Polymorphisms in miR196a2, miR-499, and miR-608 With Breast Cancer Susceptibility: A STROBE-Compliant Observational Study. Medicine (Baltimore). 2016;95(7):e2826. pmid:26886638.
- 11. Huang AJ, Yu KD, Li J, Fan L, Shao ZM. Polymorphism rs4919510:C>G in mature sequence of human microRNA-608 contributes to the risk of HER2-positive breast cancer but not other subtypes. PLoS One. 2012;7(5):e35252. pmid:22586447.
- 12. Jiao L, Zhang J, Dong Y, Duan B, Yu H, Sheng H, et al. Association between miR-125a rs12976445 and survival in breast cancer patients. Am J Transl Res. 2014;6(6):869–75. pmid:25628797.
- 13. Shao ZM, Huang AJ, Yu KD. MicroRNA-608 Impacts Susceptibility and Progress of HER2-positive Breast Cancer. Doctoral dissertation. 2012.
- 14. Hu Y, Yu CY, Wang JL, Guan J, Chen HY, Fang JY. MicroRNA sequence polymorphisms and the risk of different types of cancer. Sci Rep. 2014;4:3648. pmid:24413317.
- 15. Guo J, Jin M, Zhang M, Chen K. A genetic variant in miR-196a2 increased digestive system cancer risks: a meta-analysis of 15 case-control studies. PLoS One. 2012;7(1):e30585. pmid:22291993.
- 16. Moher D, Liberati A, Tetzlaff J, Altman DG, Group P. Preferred reporting items for systematic reviews and meta-analyses: the PRISMA statement. J Clin Epidemiol. 2009;62(10):1006–12. pmid:19631508.
- 17. Higgins JP, Thompson SG. Quantifying heterogeneity in a meta-analysis. Stat Med. 2002;21(11):1539–58. pmid:12111919.
- 18. DerSimonian R, Laird N. Meta-analysis in clinical trials revisited. Contemp Clin Trials. 2015;45(Pt A):139–45. pmid:26343745.
- 19. Copas J, Shi JQ. Meta-analysis, funnel plots and sensitivity analysis. Biostatistics. 2000;1(3):247–62. pmid:12933507.
- 20. Morales S, Gulppi F, Gonzalez-Hormazabal P, Fernandez-Ramires R, Bravo T, Reyes JM, et al. Association of single nucleotide polymorphisms in Pre-miR-27a, Pre-miR-196a2, Pre-miR-423, miR-608 and Pre-miR-618 with breast cancer susceptibility in a South American population. BMC Genet. 2016;17(1):109. pmid:27421647.
- 21. Truong T, Liquet B, Menegaux F, Plancoulaine S, Laurent-Puig P, Mulot C, et al. Breast cancer risk, nightwork, and circadian clock gene polymorphisms. Endocr Relat Cancer. 2014;21(4):629–38. pmid:24919398.
- 22. Ali AM, AbdulKareem H, Al Anazi M, Reddy Parine N, Shaik JP, Alamri A, et al. Polymorphisms in DNA Repair Gene XRCC3 and Susceptibility to Breast Cancer in Saudi Females. Biomed Res Int. 2016;2016:8721052. pmid:26881229.
- 23. Peters JL, Sutton AJ, Jones DR, Abrams KR, Rushton L. Contour-enhanced meta-analysis funnel plots help distinguish publication bias from other causes of asymmetry. J Clin Epidemiol. 2008;61(10):991–6. pmid:18538991.