+ All Categories
Home > Documents > Ethnicity modifies the association between functional microRNA polymorphisms and breast cancer risk:...

Ethnicity modifies the association between functional microRNA polymorphisms and breast cancer risk:...

Date post: 23-Dec-2016
Category:
Upload: bei
View: 217 times
Download: 0 times
Share this document with a friend
15

Click here to load reader

Transcript
Page 1: Ethnicity modifies the association between functional microRNA polymorphisms and breast cancer risk: a HuGE meta-analysis

RESEARCH ARTICLE

Qiao-Hui Chen & Qing-Bing Wang & Bei Zhang

Received: 29 May 2013 /Accepted: 5 August 2013 /Published online: 29 August 2013# International Society of Oncology and BioMarkers (ISOBM) 2013

Abstract Common functional polymorphisms in the promot-er region of microRNAs (miRNAs), based onmultiple lines ofevidence, might participate in transcriptional regulation andother biological processes, which interact to increase the riskof developing breast cancer. Since 2005, many studies haveinvestigated the association between breast cancer risk andcommon single nucleotide polymorphisms (SNPs) inmiRNAs. However, the findings of several meta-analysesare inconclusive or ambiguous. The aim of this Human Ge-nome Epidemiology meta-analysis was to determine moreprecisely the relationship between common miRNA polymor-phisms and breast cancer risk. Twelve case–control studieswith a total of 7,170 breast cancer patients and 8,783 healthycontrols were included. Eight SNPs in miRNA genes wereexamined. When all eligible studies were pooled in the meta-analysis, the miR-196a-2 rs11614913*T, miR-499rs3746444*T, and miR-605 rs2043556*A alleles predicted adecreased risk of breast cancer among Asians, while notCaucasians. In addition, themiR-27a rs895919*C allele mightbe a protective factor for breast cancer among Caucasians.However, for the miR-146a rs2910164 (G>C), miR-149rs2292832 (G>T), miR-373 rs12983273 (C>T), and miR-423 rs6505162 (C>A) polymorphisms, we failed to find anysignificant association with the risk of breast cancer in anygenetic model. In conclusion, the current meta-analysis sup-ports that the miR-196a-2 rs11614913*T, miR-499rs3746444*T, miR-605 rs2043556*A, and miR-27a

rs895919*C alleles might be protective factors for breastcancer.

Keywords Breast cancer .MicroRNA . Single nucleotidepolymorphism .Meta-analysis

Introduction

Breast cancer is by far the most common cancer amongfemales worldwide and is the leading cause of cancer-relatedmortality, responsible for almost 14 % of all cancer deaths [1].Currently, multiple etiologies have been indicated in the path-ogenesis of breast cancer. These include extrinsic factors, suchas higher levels of alcohol consumption, less physical activity,and delays in childbearing, and intrinsic factors, such asfamily history, menopausal state, and genetic mutations [2].Interventions based on extrinsic environmental factors havenot shownmuch direct influence on the development of breastcancer, but may result in synergistic interactions with intrinsicgenetic factors in triggering the disease [3]. Therefore, geneticsusceptibility to breast cancer has received enormous attentionin the last decade. For example, BRCA1 and BRCA2 are thetwo most widely studied genes implicated in the disease [4].Based on some investigations, BRCA1 and BRCA2 muta-tions account for approximately 5–10 % of all breast cancercases and 20 % of familial clustering of breast cancer [5, 6].Recently, in mutational screening of patients, a wide range ofnovel genes, such as CHEK2, ATM, BRIP1, and PALB2,have been identified to be functionally associated withBRCA1/2 [7].

MicroRNAs (miRNAs), a large class of small noncodingRNAs, play a critical role inmodulating gene expression at theposttranscriptional level. They participate in the regulation ofmany cellular functions, such as developmental patterning,cell differentiation, proliferation, apoptosis, genomerearrangements, and transcriptional regulation [8–10].

Q.<H. Chen :Q.<B. WangDepartment of Radiology, Minhang District Central Hospital,Xinsong Road No.170, Minhang District, Shanghai 201199, China

B. Zhang (*)Department of Radiology, Ruijin Hospital, Shanghai Jiao TongUniversity School of Medicine, Ruijin 2 Road No.197,Shanghai 200025, Chinae-mail: [email protected]

Tumor Biol. (2014) 35:529–543DOI 10.1007/s13277-013-1074-7

Ethnicity modifies the association between functionalmicroRNA polymorphisms and breast cancer risk: a HuGEmeta-analysis

Page 2: Ethnicity modifies the association between functional microRNA polymorphisms and breast cancer risk: a HuGE meta-analysis

miRNAs are generally defined as a discrete class of smallnonprotein-coding RNAs (ncRNAs) of 19–24 nucleotides inlength, and to date, more than 1,424 miRNAs have beenidentified in genome-wide association studies [11]. In 2005,Iorio and colleagues made the exciting discovery that humanserum/plasma contains a large amount of stable miRNAs andthat the aberrant expression patterns of serum/plasmamiRNAs have great potential as breast cancer biomarkers[12]. Later, other studies demonstrated that polymorphismsin pre-miRNAs influence the expression of their matureforms, and that they regulate the binding of some nuclearfactors involved in miRNA processing [13, 14]. Further workhas shown that many additional miRNAs are expressed dif-ferently depending on breast cancer phenotype, and that theirexpression is related to disease classification, diagnosis, andprognosis [15–19]. Therefore, it is hypothesized that polymor-phisms in pre-miRNAs affect the processing or expression ofmiRNAs, causing their up- and downregulation and thus theirfunctions as tumor suppressors or oncogenes in human carci-nogenesis [20–22].

Although several meta-analyses have recently beenconducted to explore the association between miRNAs andcancer risk, results are conflicting. For example, six previousmeta-analyses found no significant association between themiR-146a rs2910164 polymorphism and breast cancer risk inall comparison models tested [23–28]. However, a recentupdated meta-analysis demonstrated that the association be-tween the miR-146a rs2910164 polymorphism and breastcancer was modified by ethnicity [29]. Furthermore, threemeta-analyses support the view that the miR-499 rs3746444(T>C) allele may increase breast cancer susceptibility[30–32], while a meta-analysis published in 2012 does not[33]. Subsequently, four additional studies have been com-pleted since the appearance of these contradictory results.Therefore, to comprehensively examine the association be-tween miRNA SNPs and breast cancer risk across differentethnic populations, we performed a carefully designed meta-analysis that includes all the eligible case–control studiespublished to date.

Materials and methods

Literature search

Relevant papers published before the end of April 2013 wereidentified through a search of the PubMed, Embase, Web ofScience, Cochrane Library and CBM databases using thefollowing keywords and MeSH terms: (“polymorphism” or“SNP” or “gene mutation” or “genetic variants”) and (“breastneoplasms” or “breast carcinoma” or “breast cancer” or“breast tumor” or “breast carcinogenesis”) and (“microRNAs”or “miRNAs” or “primary microRNA” or “primary miRNA”

or “pri-miRNA” or “pre-miRNA”). We included all case–control studies and cohort studies of the association ofmiRNAs and breast cancer risk consist of genotyping datafor at least one of eight polymorphisms: miR-196a-2rs11614913 (C>T), miR-146a rs2910164 (G>C), miR-27ars895919 (T>C), miR-499 rs3746444 (T>C), miR-149rs2292832 (G>T), miR-373 rs12983273 (C>T), miR-605rs2043556 (A>G), and miR-423 rs6505162 (C>A). The ref-erences cited by eligible articles and textbooks were reviewedto find additional sources.

Inclusion and exclusion criteria

Studies included in our meta-analysis had to meet the follow-ing criteria: (1) a case–control or cohort design evaluated atleast one of eight miRNA polymorphisms and breast cancerrisk; (2) all patients diagnosed with breast cancer were throughpathological examinations and controls were confirmed to becancer-free; (3) inclusion of sufficient data on the size of thesample, odds ratio (OR), and 95 % confidence interval (CI).There was no language restriction. Studies were excludedwhen they represented duplicates of previous publications,or were meta-analyses, letters, reviews, or editorial articles.Additionally, when case–control data was included inmultiplestudies using the same case series, either the study with thelargest sample size or most recently published was selected.Any disagreements about study inclusion were resolvedthrough discussions and subsequent consensus.

Data extraction

Using a standardized form, data from included studies wereindependently extracted independently by two authors (Q.Chen and Q. Wang). We extracted the following informationfrom each manuscript: the first author’s surname, year ofpublication, country of origin, language of publication, eth-nicities of subjects, diagnosis, sample size, detected sample,genotyping method, genotyping information, and evidence ofHardy–Weinberg equilibrium (HWE) in controls. In cases ofconflicting evaluations, disagreements on inconsistent datafrom the eligible studies were resolved through discussionsand careful reexaminations of the full text by the authors.

Quality assessment of included studies

Two authors independently assessed the quality of papersaccording to a modified STROBE quality score system [34].Forty assessment items matching the quality appraisal wereused, with scores ranging from 0 to 40. Scores of 0–20, 20–30,and 30–40 were defined as low, moderate, and high quality,respectively. Disagreements on STROBE scores of the includ-ed studies were resolved through a comprehensivereassessment by the authors.

530 Tumor Biol. (2014) 35:529–543

Page 3: Ethnicity modifies the association between functional microRNA polymorphisms and breast cancer risk: a HuGE meta-analysis

Statistical analysis

Crude ORs together with their corresponding 95 % CIs werecalculated to assess the strength of association betweenmiRNA polymorphisms and breast cancer risk under fivegenetic models: allele, dominant, recessive, homozygous,and heterozygous models. The deviation of frequencies fromthose expected under HWE was assessed by Chi-squaredgoodness of fit tests in controls. To explore potential sourcesof heterogeneity, subgroup analyses were performed based onethnicity. The statistical significance of the pooled OR wasassessed with a Z test. Between-study variation and heteroge-neity were estimated using Cochran’s Q-statistic, with P <0.05 as a cutoff for statistically significant heterogeneity [35].

We also quantified the effect of heterogeneity with the I2

test (ranges from 0 to 100 %), which represents the proportionof interstudy variability that can be attributed to heterogeneityrather than chance [36]. The random-effect model(DerSimonian–Laird method) was conducted when Q test issignificant with P <0.05 or I2>50 % which indicates theexistence of heterogeneity among studies, otherwise the ran-dom effects model (DerSimonian–Laird method) was appliedfor meta-analysis. To ensure the reliability of results, sensitiv-ity analysis was performed by omitting individual studies inturn. Begger’s funnel plots were used to detect publicationbias. In addition, Egger’s linear regression test, which

measures funnel plot asymmetry via a natural logarithm scaleof OR, was also used to evaluate publication bias [37]. All Pvalues were two sided. Analyses were conducted with STATAVersion 12.0 software (Stata Corp, College Station, TX).

Results

The characteristics of included studies

Through searching electronic databases, using different com-binations of key terms, a total of 259 potentially relevantstudies were initially identified. Of these studies, 12 case–control studies that met the inclusion criteria are shown inTable 1 [38–49]. In total, 7,170 cases and 8,783 controls wereinvolved in the pooled analyses. The details of the selectionprocess are presented in a flow chart in Fig. 1. The publicationyear of selected studies ranged from 2009 to 2012. Theethnicities studied were Caucasian, Asian, and other mixedpopulations. Almost all of the cases were histologically con-firmed. The detected samples used for examination of miRNASNPs were extracted from blood in all included studies.Methods used for genotyping in the studies included DNAsequencing, TaqMan SNP genotyping assays, MassArraymultiplexing, and polymerase chain reaction-restriction frag-ment length polymorphism detection (PCR-RFLP). HWE

259 of records identified through electronic databases searching

(April 12th, 2013)

140 of records were excluded, due to: (n=31) Obviously irrelevant studies (n=107) Letters, reviews, meta-analysis (n=2) Not human studies

Abstract retrieved for further evaluation

(n=119)

74 of records were excluded, due to: (n=2) Not case-control study (n=33) Not explore breast cancer (n=39) Not explore microRNA genes

Full-text retrieved for detail evaluation

(n=45)

33 of records were excluded, due to: (n=1) Duplicate publications (n=18) Not explore polymorphism (n=14) Not provide sufficient data for further analysis

12 of independent studies finally included in this meta-analysis

(n=12)

Fig. 1 Flow diagram of studyselection and specific reasons forexclusion from the present meta-analysis

Tumor Biol. (2014) 35:529–543 531

Page 4: Ethnicity modifies the association between functional microRNA polymorphisms and breast cancer risk: a HuGE meta-analysis

Table 1 Characteristics of included studies in this meta-analysis

First author[Ref.]

Year Country Language Ethnicity Diagnosis Number Sample Genotypemethod

miRNAgene

SNPs Qualityscores

Case Control

Hoffman et al.[38]

2009 USA English Caucasian BC 441 479 Blood MassArray miR-196a-2

rs11614913(C>T)

27

Hu et al. [39] 2009 China English Asian BC 1,009 1,093 Blood PCR-RFLP miR-146a

rs2910164(G>C)

25

miR-149 rs2292832(G>T)

miR-196a-2

rs11614913(C>T)

miR-499 rs3746444(T>C)

Catucci et al.[40]

2010 Italy English Caucasian FBC 1,894 2,760 Blood TaqMan miR-146a

rs2910164(G>C)

28

miR-196a-2

rs11614913(C>T)

miR-499 rs3746444(T>C)

Pastrello et al.[41]

2010 Italy English Caucasian FBC 88 155 Blood DNAsequenc-ing

miR-146a

rs2910164(G>C)

27

Yang et al.[42]

2010 Germany English Caucasian FBC 1,217 1,422 Blood DNAsequenc-ing

miR-373 rs12983273(C>T)

26

miR-27a rs895919(A>G)

Jedlinski et al.[43]

2011 Australia English Caucasian BC 193 190 Blood PCR-RFLP miR-196a-2

rs11614913(C>T)

27

Zhang et al.[44]

2011 China Chinese Asian TNBC 384 192 Blood MassArray miR-373 rs12983273(C>T)

27

miR-605 rs2043556(A>G)

miR-146a

rs2910164(G>C)

miR-423 rs6505162(C>A)

miR-27a rs895819(T>C)

miR-196a-2

rs11614913(C>T)

miR-149 rs2292832(G>T)

Alshatwi et al.[45]

2012 SaudiArabia

English Asian BC 100 89 Blood TaqMan miR-499 rs3746444(T>C)

26

miR-146a

rs2910164(G>C)

miR-196a-2

rs11614913(C>T)

Catucci et al.[46]

2012 Italy English Caucasian FBC 1,025 1,593 Blood TaqMan miR-27a rs895819(T>C)

Linhares et al.[47]

2012 Brazil English Mixed BC 388 388 Blood TaqMan miR-196a-2

rs11614913(C>T)

Smith et al.[48]

2012 Australia English Caucasian BC 179 174 Blood TaqMan miR-423 rs6505162(C>A)

27

Zhang et al.[49]

2012 China English Asian BC 252 248 Blood PCR-RFLP miR-605 rs2043556(A>G)

26

miR-149 rs2292832(G>T)

miR-27a

532 Tumor Biol. (2014) 35:529–543

Page 5: Ethnicity modifies the association between functional microRNA polymorphisms and breast cancer risk: a HuGE meta-analysis

tests were performed in all but three included studies [44, 45,47]. Based on the modified STROBE quality score system, thescores of all included studies were moderately high (higherthan 20 points).

Association between miRNA SNPs and breast cancer risk

Eight miRNA SNPs were addressed: miR-196a-2 rs11614913(C>T), miR-146a rs2910164 (G>C), miR-27a rs895919(T>C), miR-499 rs3746444 (T>C), miR-149 rs2292832(G>T), miR-373 rs12983273 (C>T), miR-605 rs2043556(A>G), and miR-423 rs6505162 (C>A). A summary of themeta-analysis findings on the association of these polymor-phisms with breast cancer risk is shown in Table 2.

For miR-196a2 rs11614913 (C>T), no significant associa-tion was found (allele model: OR=0.95 (95 % CI, 0.89–1.00),P=0.055; dominant model: OR=1.00 (95 % CI, 0.85–1.18),P=0.994; recessive model: OR=0.89 (95 % CI, 0.74–1.07),P=0.206; homozygous model: OR=0.90 (95 % CI, 0.69–1.17), P=0.429; heterozygous model: OR=0.89 (95 % CI,0.78–1.01), P=0.038). However, in the ethnicity subgroupanalysis, a statistically significantly decrease in breast cancerrisk was found among Asians (allele model: OR=0.89 (95 %CI, 0.80–0.98), P=0.016; recessive model: OR=0.81 (95 %CI, 0.69–0.95), P=0.012, P=0.206; homozygous model:OR=0.76 (95 % CI, 0.62–0.94), P=0.010; heterozygousmodel: OR=0.83 (95 % CI, 0.70–0.99), P=0.038). No sig-nificant correlations were observed in any genetic model forCaucasians (Fig. 2). We also performed subgroup analysesbased on genotype methods. The results indicated that miR-196a2 was associated with decreased risk of breast cancer inthe PCR-RFLP subgroups. However, there were no associa-tions between miR-196a2 and decreased risk of breast cancerin the MassARRAYand TaqMan subgroups.

For miR-146a rs2910164 (G>C), no genetic model showeda significant correlation among all studies (allele model: OR=1.01 (95 % CI, 0.94–1.09), P=0.768; dominant model: OR=0.99 (95 % CI, 0.90-1.10), P=0.968; recessive model: OR=1.04 (95 % CI, 0.91–1.20), P=0.546; homozygous model:OR=1.07 (95 % CI, 0.90–1.27), P=0.449; heterozygousmodel: OR=1.05 (95 % CI, 0.91–1.22), P=0.544). In the

analysis stratified by ethnicity, there was still no significantassociation between miR-146a rs2910164 (G>C) and breastcancer for Asians or Caucasians (Fig. 3). Additionally, furtheranalysis on genotype methods was performed, and we foundthat miR-146a was still no significantly associated with de-creased risk of breast cancer in DNA sequencing,MassARRAY, PCR-RFLP, and TaqMan subgroups.

For miR-27a rs895919 (T>C), an analysis stratified byethnicity and genotype methods showed an association withincreased breast cancer risk among Caucasians and TaqMansubgroup in two genetic models (allele model: OR=0.89(95 % CI, 0.82–0.97), P=0.008; dominant model: OR=0.84(95 % CI, 0.75–0.94), P=0.002). Furthermore, the meta-analysis also showed that miR-499 rs3746444 (T>C) waslinked to increased breast cancer risk in two genetic models(allele model: OR=1.10 (95 % CI, 1.01–1.20), P=0.036;dominant model: OR=1.13 (95 % CI, 1.01–1.26), P =0.028). When stratified by ethnicity, miR-499 rs3746444*Cremained in association with increased breast cancer risk inAsians, under the same two genetic models (allele model:OR=1.26 (95 % CI, 1.08–1.47), P=0.004; dominant model:OR=1.31 (95 % CI, 1.09–1.57), P=0.003) (Fig. 4). Addition-ally, when stratified analysis by genotype methods wasperformed, a higher prevalence of the variant allele in miR-499 was observed only among PCR-RFLP subgroup.

For miR-605 rs2043556 (A>G), significantly increasedbreast cancer risk was observed in the recessive and hetero-zygous models (recessive model: OR=2.20 (95 % CI, 1.23–3.96), P=0.008; heterozygous model: OR=2.31 (95 % CI,1.26–4.23), P=0.007). In the analysis stratified by genotypemethods, there was still no significant association betweenmiR-605 and breast cancer in MassARRAY and PCR-RFLPsubgroups. For miR-149 rs2292832 (G>T), miR-373rs12983273 (C>T), and miR-423 rs6505162 (C>A), therewas no significant association with the risk of breast cancerin any of the genetic models (all P >0.05) (Fig. 5).

Sensitivity analysis

Sensitivity analyses assessed the influence of each individualstudy on the pooled ORs by omitting individual studies in

Table 1 (continued)

First author[Ref.]

Year Country Language Ethnicity Diagnosis Number Sample Genotypemethod

miRNAgene

SNPs Qualityscores

Case Control

rs895819(T>C)

miR-196a-2

rs11614913(C>T)

BC breast cancer, FBC familial breast cancer, TNBC triple-negative breast cancer, PCR polymerase chain reaction, RFLP restriction fragment lengthpolymorphism, SNPs single nucleotide polymorphisms

Tumor Biol. (2014) 35:529–543 533

Page 6: Ethnicity modifies the association between functional microRNA polymorphisms and breast cancer risk: a HuGE meta-analysis

Tab

le2

Meta-analysisof

theassociationbetweenthemicroRNAgene

polymorphismsandbreastcancer

risk

microRNA

gene

No.of

studies

2allelevs

1allele(allelemodel)

1/2+2/2vs

1/1(dom

inant

model)

2/2vs

1/1+1/2(recessive

model)

2/2vs

1/1(hom

ozygousmodel)

2/2vs

1/2(heterozygousmodel)

OR

95%

CI

PPh

OR

95%

CI

PPh

OR

95%

CI

PPh

OR

95%

CI

PPh

OR

95%

CI

PPh

miR-196a-2(C

>T)

Overall

80.95

0.89–1.00

0.055a

0.002

1.00

0.85–1.18

0.994a

0.009

0.89

0.74–1.07

0.206a

0.040

0.90

0.69–1.17

0.429a

0.002

0.89

0.78–1.01

0.038

0.325

Asian

0.89

0.80–0.98

0.016

a0.279

0.97

0.74–1.26

0.798a

0.088

0.81

0.69–0.95

0.012

a0.822

0.76

0.62–0.94

0.010

a0.608

0.83

0.70–0.99

0.038

0.782

Caucasian

0.97

0.90–1.04

0.335a

0.003

0.99

0.77–1.26

0.916a

0.019

0.89

0.64–1.24

0.505a

0.014

0.90

0.58–1.40

0.642a

0.002

0.89

0.69–1.16

0.386

0.110

MassA

RRAY

0.85

0.64–1.12

0.224a

0.078

0.90

0.60–1.35

0.609a

0.089

0.67

0.41–1.12

0.127a

0.073

0.68

0.32–1.43

0.309a

0.022

0.86

0.72–1.04

0.112

0.721

PCR-RFLP

0.86

0.77–0.95

0.004

a0.774

0.82

0.69–0.97

0.018

a0.900

0.82

0.69–0.97

0.023

a0.646

0.74

0.60–0.92

0.006

a0.668

0.67

0.50–0.91

0.009

0.239

TaqM

an1.19

0.95–1.51

0.136a

0.019

1.34

0.94–1.91

0.107a

0.017

1.10

0.84–1.44

0.489a

0.198

1.29

0.81–2.05

0.284a

0.037

0.99

0.84–1.16

0.889

0.537

miR-146a(G

>C)

Overall

51.01

0.94–1.09

0.768

0.968

0.99

0.90–1.10

0.968

0.966

1.04

0.91–1.20

0.546

0.596

1.07

0.90–1.27

0.449

0.711

1.05

0.91–1.22

0.494

0.544

Asian

0.99

0.89–1.10

0.824

0.922

0.99

0.82–1.21

0.982

0.794

0.97

0.83–1.15

0.749

0.914

0.97

0.77–1.21

0.760

0.885

1.27

0.97–1.67

0.779

0.869

Caucasian

1.03

0.93–1.15

0.532

0.845

0.99

0.88–1.13

0.973

0.741

1.25

0.96–1.62

0.096

0.765

1.23

0.94–1.61

0.127

0.820

0.98

0.82–1.16

0.081

0.690

DNA

sequencing

0.99

0.64–1.54

0.965

–0.91

0.54–1.56

0.742

–1.50

0.44–5.05

0.516

–1.42

0.41–4.86

0.581

–1.64

0.46–5.81

0.444

MassA

RRAY

0.95

0.74–1.22

0.704

–0.89

0.57–1.39

0.598

–0.98

0.68–1.41

0.900

–0.89

0.54–1.48

0.659

–1.01

0.68–1.50

0.954

PCR-RFLP

0.99

0.88–1.12

0.896

–0.99

1.01–0.80

0.943

–0.98

0.81–1.17

0.802

–0.99

0.77–1.28

0.948

–0.97

0.80–1.18

0.774

TaqM

an1.04

0.94–1.15

0.492

0.945

1.01

0.89–1.15

0.887

0.688

1.22

0.94–1.59

0.139

0.501

1.21

0.92–1.58

0.170

0.564

1.24

0.94–1.63

0.129

0.447

miR-27a

(T>C)

Overall

40.92

0.85–0.99

0.023

0.444

0.96

0.79–1.18

0.706a

0.031

0.90

0.76–1.06

0.210

0.309

0.88

0.74–1.05

0.154

0.865

0.89

0.65–1.22

0.462

0.046

Asian

1.04

0.86–1.25

0.706

0.491

1.25

0.96–1.63

0.098a

0.525

0.75

0.52–1.07

0.115

0.173

0.97

0.64–1.46

0.880

0.509

0.69

0.37–1.28

0.237

0.133

Caucasian

0.89

0.82–0.97

0.008

0.717

0.84

0.75–0.94

0.002

a0.319

0.94

0.79–1.13

0.538

0.493

0.86

0.71–1.05

0.133

0.804

1.04

0.86–1.27

0.674

0.316

MassA

RRAY

1.12

0.84–1.47

0.446

–1.16

0.82–1.65

0.410a

–1.09

0.56–2.11

0.798

–1.16

0.59–2.28

0.669

–1.00

0.50–2.00

1.000

PCR-RFLP

0.98

0.76–1.26

0.855

–1.38

0.92–2.05

0.116a

–0.63

0.40–1.00

0.059

–0.87

0.52–1.47

0.598

–0.53

0.33–0.84

0.007

TaqM

an0.89

0.82–0.97

0.008

0.717

0.84

0.75–0.94

0.002

a0.319

0.94

0.79–1.13

0.538

0.493

0.86

0.71–1.05

0.133

0.804

1.04

0.86–1.27

0.674

0.316

miR-499

(T>C)

Overall

31.10

1.01–1.20

0.036

0.214

1.13

1.01–1.26

0.028

0.102

1.07

0.71–1.60

0.758a

0.059

1.17

0.92–1.48

0.215

0.189

0.95

0.60–1.51

0.839a

0.033

Asian

1.26

1.08–1.47

0.004

0.760

1.31

1.09–1.57

0.003

0.182

0.97

0.29–3.18

0.957a

0.019

1.50

0.98–2.30

0.061

0.157

0.75

0.18–3.08

0.690a

0.008

Caucasian

1.03

0.92–1.15

0.628

0.924

1.04

0.91–1.18

0.569

0.536

1.01

0.74–1.38

0.945a

0.289

1.03

0.77–1.38

0.844

0.381

0.99

0.67–1.48

0.988a

0.200

PCR-RFLP

1.27

1.08–1.50

0.005

–1.26

1.04–1.52

0.019

–1.67

1.04–2.69

0.034

a–

1.75

1.08–2.83

0.022

–1.46

0.88–2.41

0.139a

TaqM

an1.04

0.93–1.16

0.493

0.794

1.11

0.89–1.39

0.335

0.113

0.92

0.62–1.36

0.673a

0.194

1.01

0.76–1.34

0.955

0.605

0.81

0.47–1.41

0.458a

0.053

miR-149

(G>T)

30.94

0.84–1.05

0.254

0.347

0.92

0.80–1.07

0.279

0.259

0.92

0.73–1.16

0.478

0.584

0.89

0.70–1.13

0.344

0.537

0.95

0.74–1.22

0.693a

0.546

MassA

RRAY

0.89

0.67–1.16

0.387

–0.94

0.66–1.34

0.739

–0.72

0.43–1.23

0.231

–0.73

0.42–1.27

0.266

–0.71

0.40–1.27

0.247a

PCR-RFLP

0.95

0.84–1.07

0.382

0.167

0.92

0.79–1.08

0.301

0.101

0.97

0.75–1.26

0.827

0.737

0.93

0.71–1.22

0.605

0.418

1.01

0.77–1.33

0.925a

0.872

miR-373

(C>T)

20.98

0.81–1.18

0.794

0.651

1.00

0.81–1.23

0.999

0.639

0.78

0.43–1.41

0.405

0.963

0.78

0.43–1.42

0.411

0.978

0.76

0.41–1.42

0.395

0.866

534 Tumor Biol. (2014) 35:529–543

Page 7: Ethnicity modifies the association between functional microRNA polymorphisms and breast cancer risk: a HuGE meta-analysis

turn. These analyses suggested that no individual study sig-nificantly affected the pooled ORs for miR-196a-2rs11614913, miR-146a rs2910164 and miR-27a rs895919under the allele model (Fig. 6), meaning the results are statis-tically accurate.

Publication bias

Begger’s funnel plot and Egger’s linear regression testassessed publication bias of the included studies. The shapesof the funnel plots for miR-196a-2 rs11614913, miR-146ars2910164, and miR-27a rs895919 under the allele modeldid not reveal any obvious asymmetry (Fig. 7). Egger’s testalso showed no statistically significant evidence of pub-lication bias for any of the genetic models (miR-196a-2:t =0.20, P=0.850; miR-146a: t =0.22, P=0.843; miR-27a:t =3.93, P=0.059).

Discussion

miRNAs are a family of endogenous small noncoding RNAsthat participate in the regulation of diverse biological func-tions, such as cell proliferation, differentiation, apoptosis,genome rearrangements, and transcriptional regulation [9,50]. Many studies have highlighted the role that microRNAshave in physiological processes, and how their deregulationcan lead to cancer [51]. It is therefore not surprising that asingle deregulated miRNA can induce global changes incellular physiology [52]. Unfortunately, to date, the cause ofmiRNA dysregulation in the pathogenesis of breast cancer isstill not fully elucidated. Nevertheless, some studies suggestthat this dysregulation might be partly due to genetic poly-morphisms in important mammary miRNAs [3, 53, 54]. In-terestingly, while specific miRNAs are often over-expressedin cancer cells, most miRNAs are actually downregulated intumors [55]. Global miRNA repression enhances cellulartransformation and tumorigenesis in both in vitro andin vivo models, indicating that miRNA loss-of-function canhave pro-tumorigenic effects [56].

To date, the correlation of several miRNAs with geneticsusceptibility to breast cancer have been studied. However,the results of these studies differ from each other [45, 49].Besides, several meta-analyses have been published, but noclear consensus has been reached [28, 29, 32, 33]. Given thecontroversial results in previous meta-analyses, we took amore comprehensive approach. With a larger sample andsubgroup analysis of all eligible case–control studies, weaimed to evaluate more reliably the relationship betweenmiRNA SNPs and breast cancer.

This meta-analysis involved 7,170 breast cancer cases and8,783 healthy controls in 12 case–control studies. The pooledORs of eight studies, with the overall analysis model, suggestT

able2

(contin

ued)

microRNA

gene

No.of

studies

2allelevs

1allele(allelemodel)

1/2+2/2vs

1/1(dom

inant

model)

2/2vs

1/1+1/2(recessive

model)

2/2vs

1/1(hom

ozygousmodel)

2/2vs

1/2(heterozygousmodel)

OR

95%

CI

PPh

OR

95%

CI

PPh

OR

95%

CI

PPh

OR

95%

CI

PPh

OR

95%

CI

PPh

miR-605

(A>G)

21.15

0.93–1.42

0.199

0.908

1.04

0.79–1.37

0.773

0.655

2.20

1.23–3.96

0.008

0.534

2.13

1.17–3.88

0.013

0.667

2.31

1.26–4.23

0.007

0.420

miR-423

(C>A)

21.12

0.91–1.38

0.301

0.093

1.16

0.86–1.56

0.322

0.015

1.13

0.77–1.67

0.535

0.696

1.52

0.93–2.48

0.097

0.144

0.99

0.66–1.50

0.969

0.866

1allele

wild

allele,2allele

mutantallele,1/1

wild

homozygote,1/2heterozygote,2

/2mutanth

omozygote,PhPvalueof

heterogeneity

test

aEstim

ates

forrandom

effectsmodel

Boldentriesmarkedsignificantresults(P

<0.05)

Tumor Biol. (2014) 35:529–543 535

Page 8: Ethnicity modifies the association between functional microRNA polymorphisms and breast cancer risk: a HuGE meta-analysis

Overall (I2 = 67.6%, P = 0.002)

Zhang et al (2011)

Subtotal (I2 = 21.9%, P = 0.279)

Caucasian

Subtotal (I2 = 78.6%, P = 0.003)

Alshatwi et al (2012)

Catucci et al (2010)

AsianHu et al (2009)

Linhares et al (2012)Jedlinski et al (2011)

Subtotal Linhares et al (2012)

Hoffman et al (2009)

Non−Caucasian

Zhang et al (2012)

100.00

1.64

1.361.36

10.11

4.24

5.35

35.22

63.42

44.02

23.99

5.493.80

0.94 (0.89, 1.00)

0.99 (0.77, 1.27)

0.88 (0.80, 0.98)

0.96 (0.90, 1.04)

1.23 (0.81, 1.88)

0.97 (0.89, 1.06)

0.85 (0.75, 0.96)

1.31 (1.05, 1.65)0.94 (0.70, 1.27)

1.55 (1.00, 2.41)1.55 (1.00, 2.41)

0.74 (0.61, 0.90)

0.81 (0.61, 1.09)

10.415 1 2.41

Study ID OR (95% CI) Weight %a

Study ID OR (95% CI) Weight %c

NOTE: Random effects analysis

Overall (I2 = 50.5%, P = 0.040)

Non−Caucasia

Asian

Subtotal

Subtotal (I2 = 71.6%, P = 0.014)Linhares et al (2012)

Hu et al (2009)Zhang et al (2011)

Linhares et al (2012)

Zhang et al (2012)

Catucci et al (2010)

Alshatwi et al (2012)

Hoffman et al (2009)

Jedlinski et al (2011)

Caucasian

Subtotal (I2 = 0.0%, P = 0.822)

100.00

6.02

54.3413.10

21.3712.58

6.02

4.60

22.03

1.09

11.10

8.10

39.65

0.89 (0.74, 1.07)

1.61 (0.83, 3.12)

0.89 (0.64, 1.24)1.28 (0.89, 1.85

0.82 (0.68, 0.98)0.86 (0.59, 1.27)

1.61 (0.83, 3.12)

0.62 (0.28, 1.35

0.95 (0.80, 1.13)

0.49 (0.09, 2.74)

0.51 (0.34, 0.79)

0.97 (0.56, 1.66)

0.81 (0.69, 0.95

)

))

10.0877 1 11.4

Study ID OR (95% CI) Weight %b

NOTE: Random effects analysis

Overall (I2= 66.8%, P = 0.002)

Subtotal (I2 = 0.0%, P = 0.608)

Subtotal

Zhang et al (2011)

CaucasianHoffman et al (2009)

Jedlinski et al (2011)Catucci et al (2010)

Non−Caucasian

Hu et al (2009)

Zhang et al (2012)

Linhares et al (2012)

Asian

Linhares et al (2012)

Alshatwi et al (2012)

Subtotal (I2 = 80.4%, P = 0.002)

100.00

39.12

6.2

12.38

12.77

9.8618.75

17.62

7.12

13.24

6.2

2.01

54.6

0.90 (0.69, 1.17)

0.76 (0.62, 0.94)

2.30 (0.96, 5.50)

1.00 (0.62, 1.60)

0.47 (0.30, 0.73)

0.91 (0.50, 1.660.94 (0.78, 1.14)

0.73 (0.57, 0.93)

0.58 (0.26, 1.29)

1.61 (1.05, 2.48)

2.30 (0.96, 5.50)

0.66 (0.11, 3.79)

0.90 (0.58, 1.40)

)

10.114 1 8.79

NOTE: Random effects analysis

Overall (I2 = 13.1%, P = 0.325)

Hoffman et al (2009)

Non−Caucasian

Alshatwi et al (2012)

Linhares et al (2012)

Zhang et al (2011)

Subtotal

Subtotal (I2 = 0.0%, P = 0.782)

AsianHu et al (2009)

Zhang et al (2012)

Subtotal (I2 = 50.3%, P = 0.110)Linhares et al (2012)

Catucci et al (2010)Jedlinski et al (2011)

Caucasian

100.00

0.58

3.35

9.17

3.35

41.48

29.15

2.58

55.189.77

32.334.97

8.098.09

0.88 (0.78, 1.01)

0.56 (0.36, 0.86)

0.40 (0.07, 2.26)

1.35 (0.66, 2.74)

0.80 (0.53, 1.21

1.35 (0.66, 2.74)

0.83 (0.70, 0.99)

0.86 (0.70, 1.05)

0.68 (0.30, 1.52)

0.89 (0.68, 1.16)1.10 (0.74, 1.63)

0.96 (0.80, 1.15)1.02 (0.57, 1.81)

10.0698 1 14.3

Study ID OR (95% CI) Weight %d

Fig. 2 Forest plot of ORs for theassociation of miR-196a-2rs11614913with breast cancer risk isillustrated in subgroup analysis byethnicity (a allelemodel;b recessivemodel; c homozygous model; dheterozygous model)

536 Tumor Biol. (2014) 35:529–543

Page 9: Ethnicity modifies the association between functional microRNA polymorphisms and breast cancer risk: a HuGE meta-analysis

that there is no significant association between the poly-morphism miR-196a-2 rs11614913 and breast cancer, afinding inconsistent with previously reported results. Inthe ethnicity subgroup analysis, Asian individuals carryingthe miR-196a-2*T allele had a decreased breast cancer risk,whereas there were no significant correlations in any ge-netic model in Caucasians. Different conclusions by previ-ous meta-analyses may be due to the limited number ofstudies used in these analyses and the relatively smallnumber of patients suitable for inclusion (eight in the pres-ent study versus three in Wang et al., four in Qiu et al., threein Chu et al., three in Xu et al., four in Wang et al., three in

Tian et al., and three in Gao et al.) [24, 28, 57–61]. Never-theless, because of the inconsistency, additional adequatelypowered studies based on larger populations are warrantedto make definitive conclusions.

For miR-146a rs2910164, we did not detect any significantassociation with breast cancer risk, consistent with previouslyreports. In our ethnicity subgroup analysis, nonsignificantrisks were not observed in Asians nor Caucasians. This con-clusion conflicts with a prior meta-analysis [29], which sug-gested that the CC homozygote of rs2910164might contributeto breast cancer susceptibility in Europeans. This differencemay be a result of the limited statistical power of Lian and

Overall (I2 = 0.0%, P = 0.968)

Hu et al (2009)

Pastrello et al (2010)

Zhang et al (2011)

Catucci et al (2010)

Alshatwi et al (2012)

Subtotal (I2 = 0.0%, P = 0.922)

Caucasian

Asian

100.00

36.78

2.85

9.16

51.34

48.48

2.72

48.66

1.01 (0.94, 1.09)

0.99 (0.88, 1.12)

0.99 (0.64, 1.54)

0.95 (0.74, 1.22)

1.03 (0.93, 1.15)

1.04 (0.93, 1.15)

1.05 (0.68, 1.64)

0.99 (0.89, 1.10)

10.609 1 1.64

Overall (I2 = 0.0%, P = 0.966)

Hu et al (2009)

Subtotal (I2 = 0.0%, P = 0.741)

Alshatwi et al (2012)

Catucci et al (2010)

Caucasian

Subtotal (I2 = 0.0%, P = 0.794)

Zhang et al (2011)

Pastrello et al (2010)

Asian

1.00 (0.90, 1.11)

1.01 (0.80, 1.27)

1.00 (0.88, 1.13)

1.13 (0.65, 1.96)

1.00 (0.88, 1.14)

1.00 (0.82, 1.21)

0.89 (0.57, 1.39)

0.91 (0.54, 1.56)

100.00

21.01

69.52

3.45

65.37

4.16

30.48

6.02

)

10.509 1 1.96

Overall (I2 = 0.0%, P = 0.596)

Zhang et al (2011)

Asian

Alshatwi et al (2012)

Pastrello et al (2010)

Subtotal (I2 = 0.0%, P = 0.765)

Hu et al (2009)

CaucasianCatucci et al (2010)

Subtotal (I2 = 0.0%, P = 0.914)

1.04 (0.91, 1.20)

0.98 (0.68, 1.41)

0.66 (0.11, 4.04)

1.50 (0.44, 5.05)

1.25 (0.96, 1.62)

0.98 (0.81, 1.17)

1.24 (0.95, 1.61)

0.97 (0.83, 1.15)

14.49

0.74

1.04

25.42

100.00

59.34

24.39

74.58

10.108 1 9.27

%thgieW)IC%59(RODIydutS a

)IC%59(RODIydutS Weight %b

)IC%59(RODIydutS Weight %c

Fig. 3 Forest plot of ORs for theassociation of miR-146ars2910164 with breast cancer riskis illustrated in subgroup analysisby ethnicity under theheterozygous model (a allelemodel; b dominant model; crecessive model)

Tumor Biol. (2014) 35:529–543 537

Page 10: Ethnicity modifies the association between functional microRNA polymorphisms and breast cancer risk: a HuGE meta-analysis

colleagues’meta-analysis, because of inclusion of two studieslacking genotype data that should have been excluded [38,62]. Hoffman and colleagues evaluated the association be-tween miR-196a-2 rs11614913 and breast cancer, rather thanthe miR-146a rs2910164 polymorphism in consideration hereand hence should have been excluded [38]. The study designreported by Lian et al. to analyze the data published by Garciaand colleagues does appear to be case only rather than case–control [62]. In the study byGarcia and colleagues, the controlgroup was comprised of nonaffected subjects with BRCA1/2mutations, and the case group of affected patients withBRCA1/2 mutations. The study by Garcia and colleaguesincluded in the meta-analysis by Lian et al., the authorsselected as controls nonaffected patients with BRCA1/2 mu-tation, and as cases, affected breast cancer patients withBRCA1/2 mutations, did not met the inclusion criteria.

Four studies have examined the association between themiR-27a rs895919 polymorphism and risk of breast cancer.The present meta-analysis shows that the miR-27a

rs895919*C variant is associated with a decreased risk ofbreast cancer, consistent with a prior meta-analysis [63]. Insubgroup analysis based on ethnicity, there is a statisticallysignificant association between this SNP and breast cancerrisk in Caucasians. This finding is biologically plausible, asrecent studies have demonstrated that miR-27a, an onco-miR,exhibits oncogenic activity by regulating target genes.Downregulation of miR-27a, by causing the upregulation ofits targets, may thus contribute to decreased cancer risk[64–66].

Only three studies have looked at the association betweenmiR-499 rs3746444 and breast cancer risk. The present meta-analysis of these found a significantly decreased risk under theallele and dominant genetic models, which is consistent withpreviously reported results. When stratified by ethnicity, thissequence variant remained associated with decreased risk inAsians but not Europeans. One possible interpretation for thisresult may be that ethnic differences in genetic backgroundshave an etiological role in breast cancer. Alternatively,

Overall (I2 = 0.0%, P = 0.444)

Catucci et al (2012)

Zhang et al (2011)

Zhang et al (2012)

Asian

Subtotal (I2 = 0.0%, P = 0.491)

Subtotal (I2 = 0.0%, P = 0.717)

Yang et al (2010)

Caucasian

0.91 (0.85, 0.99)

0.91 (0.80, 1.02)

1.11 (0.84, 1.47)

0.98 (0.76, 1.26)

1.04 (0.86, 1.25)

0.89 (0.82, 0.97)

0.88 (0.78, 0.99)

100.00

39.49

6.91

9.05

15.96

84.04

44.56

1 74.11876.0

NOTE: Random effects analysis

Overall (I2 = 66.2%, P = 0.031)

Caucasian

Yang et al (2010)

Asian

Catucci et al (2012)

Subtotal (I2 = 0.0%, P = 0.319)

Zhang et al (2011)

Zhang et al (2012)

Subtotal (I2 = 0.0%, P = 0.525)

0.96 (0.79, 1.18)

0.79 (0.68, 0.93)

0.89 (0.76, 1.04)

0.84 (0.75, 0.94)

1.16 (0.82, 1.65)

1.38 (0.92, 2.05)

1.25 (0.96, 1.63)

100.00

33.14

32.96

66.1

18.23

15.6

33.90

1 50.21884.0

%thgieW)IC%59(RODIydutS a

%thgieW)IC%59(RODIydutS b

Fig. 4 Forest plot of ORs for theassociation of miR-27a rs895919with breast cancer risk isillustrated in subgroup analysis byethnicity (a allele model; bdominant model)

538 Tumor Biol. (2014) 35:529–543

Page 11: Ethnicity modifies the association between functional microRNA polymorphisms and breast cancer risk: a HuGE meta-analysis

because small sample sizes of case patients and control sub-jects can increase false-positive report probability and there-fore affect statistical power [67], this result might due tochance. Therefore, further studies are required to investigatethe correlation between the miR-499 rs3746444 polymor-phism and the risk for breast cancer in Caucasians.

In addition, this study reveals that the CC genotype of miR-605 rs2043556 might increase breast cancer susceptibility inAsians. All participants in the present analysis of miR-605rs2043556 are Asian, so additional, large case–control studiesare necessary to validate this finding, especially in non-Asians. Finally, we also analyzed the miR-149 rs2292832,miR-373 rs12983273, and miR-423 rs6505162 polymor-phisms in breast cancer susceptibility. There were no signifi-cant associations between these polymorphisms and risk inany genetic model.

Several specific limitations complicate the interpretationour meta-analysis. First, the case subjects were simply definedas breast cancer patients, with both familial breast cancer and

triple-negative breast cancer patients enrolled in some of thestudies. Second, the genotype distribution of three studies didnot conform to HWE expectations, which may distort theresults. Third, there were an insufficient number of eligiblestudies for several SNPs, including miR-605 rs2043556, miR-149 rs2292832, miR-373 rs12983273, and miR-423rs6505162, which limited further stratified analysis based onethnicity. Finally, this meta-analysis was based on unadjustedORs estimates. A lack of available information prevented amore precise evaluation with adjusted ORs by other covariatessuch as age, gender, smoking status, menopausal status, ge-notype methods, or other factors. Despite these limitations,however, our meta-analysis still has several strengths thatmerit attention. This analysis includes the largest number ofcases and controls from each included study reported to date,which significantly increases statistical power. Moreover, ourresults are in relatively good agreement with those observed inthe previous largest study, and the results remain valid inalmost every subgroup analysis.

Overall (I2 = 33.0%, P = 0.214)

Hu et al (2009)

Subtotal (I2 = 0.0%, P = 0.760)

Catucci et al (2010)

Asian

Catucci et al (2010)

Alshatwi et al (2012)

Caucasian

Subtotal (I2 = 0.0%, P = 0.924)

1.10 (1.01, 1.20)

1.27 (1.08, 1.50)

1.26 (1.08, 1.47)

1.03 (0.89, 1.20)

1.02 (0.86, 1.21)

1.19 (0.79, 1.78)

1.03 (0.92, 1.15)

100.00

26.82

31.52

38.64

29.83

4.70

68.48

10.561 1 1.78

Overall (I2 = 51.7%, P = 0.102)

Catucci et al (2010)

Asian

Catucci et al (2010)

Subtotal (I2 = 0.0%, P = 0.536)

Alshatwi et al (2012)

Subtotal (I2 = 43.8%, P = 0.182)

Hu et al (2009)

Caucasian

1.13 (1.01, 1.26)

1.08 (0.90, 1.30)

0.99 (0.82, 1.21)

1.04 (0.91, 1.19)

1.91 (1.07, 3.41)

1.31 (1.09, 1.57)

1.26 (1.04, 1.52)

100.00

35.70

31.82

67.52

2.64

32.48

29.84

10.293 1 3.41

%thgieW)IC%59(RODIydutS a

)IC%59(RODIydutS Weight %b

Fig. 5 Forest plot of ORs for theassociation of miR-499rs3746444 with breast cancer riskis illustrated in subgroup analysisby ethnicity (a allele model; bdominant model)

Tumor Biol. (2014) 35:529–543 539

Page 12: Ethnicity modifies the association between functional microRNA polymorphisms and breast cancer risk: a HuGE meta-analysis

0.83 0.97 0.86 1.10 1.16

Hoffman et al (2009)

Hu et al (2009)

Catucci et al (2010)

Jedlinski et al (2011)

Zhang et al (2011)

Alshatwi et al (2012)

Linhares et al (2012)

Linhares et al (2012)

Zhang et al (2012)

Lower CI Limit Estimate Upper CI Limit

10.198.0 21.190.149.0

Hu et al (2009)

Catucci et al (2010)

Pastrello et al (2010)

Zhang et al (2011)

Alshatwi et al (2012)

Lower CI Limit Estimate Upper CI Limit

0.83 0.91 0.85 0.99 1.05

Yang et al (2010)

Zhang et al (2011)

Catucci et al (2012)

Zhang et al (2012)

Lower CI Limit Estimate Upper CI Limit

a

b

c

Fig. 6 Sensitivity analyses of thesummary odds ratio coefficientsof a miR-196a-2 rs11614913, bmiR-146a rs2910164, and c miR-27a rs895919 are illustrated underthe allele model. Results werecomputed by omitting each studyin turn. The two ends of the dottedlines represent the 95 % CI

540 Tumor Biol. (2014) 35:529–543

Page 13: Ethnicity modifies the association between functional microRNA polymorphisms and breast cancer risk: a HuGE meta-analysis

In conclusion, our meta-analysis provides evidence that themiR-196a-2 rs11614913*T, miR-499 rs3746444*T, and miR-605 rs2043556*A alleles might decrease the risk of breast

cancer in Asians. In addition, the miR-27a rs895919*C allelemight be a protective factor for breast cancer in Caucasians.However, no significant associations with breast cancer risk

0 0.5 1

−2

−1

0

1

2

0 0.5 1

−2

−1

0

1

2

0 0.05 0.1 0.15 0.2

−0.5

0

0.5

a

b

c

Log[O

R]

Log[O

R]

Log[O

R]

SELog[OR]

SELog[OR]

SELog[OR]

Fig. 7 Begger’s funnel plot ofpublication bias in selection ofstudies on a miR-196a-2rs11614913, b miR-146ars2910164, and c miR-27ars895919, under the allele model.Each point represents a separatestudy by the indicated association.Log[OR] natural logarithm ofOR. Horizontal line, meanmagnitude of the effect

Tumor Biol. (2014) 35:529–543 541

Page 14: Ethnicity modifies the association between functional microRNA polymorphisms and breast cancer risk: a HuGE meta-analysis

were found for miR-146a rs2910164, miR-149 rs2292832,miR-373 rs12983273, and miR-423 rs6505162 in any geneticmodel. Because of the study’s limitations, however, detailedstudies are warranted to confirm these findings.

Conflict of interest None.

References

1. Jemal A, Bray F, Center MM, Ferlay J, Ward E, et al. Global cancerstatistics. CA Cancer J Clin. 2011;61:69–90.

2. McCormack VA, Boffetta P. Today’s lifestyles, tomorrow’s cancers:trends in lifestyle risk factors for cancer in low- and middle-incomecountries. Ann Oncol. 2011;22:2349–57.

3. Stuckey A. Breast cancer: epidemiology and risk factors. Clin ObstetGynecol. 2011;54:96–102.

4. Antoniou AC, Easton DF. Models of genetic susceptibility to breastcancer. Oncogene. 2006;25:5898–905.

5. Campeau PM, Foulkes WD, Tischkowitz MD. Hereditary breastcancer: new genetic developments, new therapeutic avenues. HumGenet. 2008;124:31–42.

6. Evans JP, Skrzynia C, Susswein L, HarlanM. Genetics and the youngwoman with breast cancer. Breast Dis. 2005;23:17–29.

7. Balmana J, Diez O, Rubio IT, Cardoso F. BRCA in breast cancer:ESMO clinical practice guidelines. Ann Oncol. 2011;22 Suppl6:vi31–4.

8. Bartel DP.MicroRNAs: genomics, biogenesis, mechanism, and func-tion. Cell. 2004;116:281–97.

9. He L, Hannon GJ. MicroRNAs: small RNAs with a big role in generegulation. Nat Rev Genet. 2004;5:522–31.

10. FilipowiczW, Bhattacharyya SN, Sonenberg N.Mechanisms of post-transcriptional regulation by microRNAs: are the answers in sight?Nat Rev Genet. 2008;9:102–14.

11. Esteller M. Non-coding RNAs in human disease. Nat Rev Genet.2011;12:861–74.

12. Iorio MV, Ferracin M, Liu CG, Veronese A, Spizzo R, et al.MicroRNA gene expression deregulation in human breast cancer.Cancer Res. 2005;65:7065–70.

13. Duan R, Pak C, Jin P. Single nucleotide polymorphism associatedwith mature mir-125a alters the processing of pri-mirna. Hum MolGenet. 2007;16:1124–31.

14. Jazdzewski K, Murray EL, Franssila K, Jarzab B, Schoenberg DR,et al. Common SNP in pre-miR-146a decreases mature miR expres-sion and predisposes to papillary thyroid carcinoma. Proc Natl AcadSci U S A. 2008;105:7269–74.

15. Chen J, Tian W, Cai H, He H, Deng Y. (2012) Down-regulation ofmicroRNA-200c is associated with drug resistance in human breastcancer. Med Oncol 29:2527–2534

16. Toyama T, Kondo N, Endo Y, Sugiura H, Yoshimoto N, et al. Highexpression of microRNA-210 is an independent factor indicating apoor prognosis in Japanese triple-negative breast cancer patients. JpnJ Clin Oncol. 2012;42:256–63.

17. Wang C, Gao C, Zhuang JL, Ding C, Wang Y (2012) A combinedapproach identifies three mRNAs that are down-regulated bymicroRNA-29b and promote invasion ability in the breast cancer cellline MCF-7. J Cancer Res Clin Oncol 138:2127–2136

18. Zhao FL, Hu GD, Wang XF, Zhang XH, Zhang YK, et al. Serumoverexpression of microRNA-10b in patients with bone metastaticprimary breast cancer. J Int Med Res. 2012;40:859–66.

19. Zhou X, Marian C, Makambi KH, Kosti O, Kallakury BV, et al.MicroRNA-9 as potential biomarker for breast cancer local recur-rence and tumor estrogen receptor status. PloS One. 2012;7:e39011.

20. Chendrimada TP, Gregory RI, Kumaraswamy E, Norman J, CoochN, et al. TRBP recruits the Dicer complex to Ago2 for microRNAprocessing and gene silencing. Nature. 2005;436:740–4.

21. Lu J, Getz G, Miska EA, Alvarez-Saavedra E, Lamb J, et al.MicroRNA expression profiles classify human cancers. Nature.2005;435:834–8.

22. Esquela-Kerscher A, Slack FJ. OncomiRs—microRNAs with a rolein cancer. Nat Rev Cancer. 2006;6:259–69.

23. Qiu LX, He J, Wang MY, Zhang RX, Shi TY, et al. The associationbetween common genetic variant of microRNA-146a and cancersusceptibility. Cytokine. 2011;56:695–8.

24. Xu W, Xu J, Liu S, Chen B, Wang X, et al. Effects of commonpolymorphisms rs11614913 in miR-196a2 and rs2910164 in miR-146a on cancer susceptibility: a meta-analysis. PloS One.2011;6:e20471.

25. Wang AX, Xu B, Tong N, Chen SQ, Yang Y, et al. Meta-analysisconfirms that a common G/C variant in the pre-miR-146a genecontributes to cancer susceptibility and that ethnicity, gender andsmoking status are risk factors. Gen Mol Res: GMR.2012;11:3051–62.

26. Wang F, Sun G, Zou Y, Fan L, Song B. Lack of association of miR-146a rs2910164 polymorphism with gastrointestinal cancers: evi-dence from 10206 subjects. PloS One. 2012;7:e39623.

27. Wang J, Bi J, Liu X, Li K, Di J, et al. Has-miR-146a polymorphism(rs2910164) and cancer risk: a meta-analysis of 19 case–controlstudies. Mol Biol Rep. 2012;39:4571–9.

28. Wang J, Wang Q, Liu H, Shao N, Tan B, et al. The association ofmiR-146a rs2910164 and miR-196a2 rs11614913 polymorphismswith cancer risk: a meta-analysis of 32 studies. Mutagenesis.2012;27:779–88.

29. Lian H, Wang L, Zhang J. Increased risk of breast cancer associatedwith cc genotype of has-miR-146a rs2910164 polymorphism inEuropeans. PloS One. 2012;7:e31615.

30. Wang F, Sun G, Zou Y, Li Y, Hao L, et al. Association of microRNA-499 rs3746444 polymorphism with cancer risk: evidence from 7188cases and 8548 controls. PloS One. 2012;7:e45042.

31. Wang L, Qian S, Zhi H, Zhang Y, Wang B, et al. The associationbetween Hsa-miR-499t>c polymorphism and cancer risk: a meta-analysis. Gene. 2012;508:9–14.

32. Wang Y, Yang B, Ren X. Hsa-miR-499 polymorphism (rs3746444)and cancer risk: a meta-analysis of 17 case–control studies. Gene.2012;509:267–72.

33. Qiu MT, Hu JW, Ding XX, Yang X, Zhang Z, et al. Hsa-miR-499rs3746444 polymorphism contributes to cancer risk: a meta-analysisof 12 studies. PloS One. 2012;7:e50887.

34. da Costa BR, Cevallos M, Altman DG, Rutjes AW, Egger M. Usesand misuses of the strobe statement: bibliographic study. BMJ Open.2011;1:e000048.

35. Higgins JP, Thompson SG. Quantifying heterogeneity in a meta-analysis. Stat Med. 2002;21:1539–58.

36. Zintzaras E, Ioannidis JP. Heterogeneity testing in meta-analysis ofgenome searches. Genet Epidemiol. 2005;28:123–37.

37. Peters JL, Sutton AJ, Jones DR, Abrams KR, Rushton L. Compari-son of two methods to detect publication bias in meta-analysis.JAMA : J Am Med Assoc. 2006;295:676–80.

38. HoffmanAE, Zheng T,Yi C, Leaderer D,Weidhaas J, et al.MicroRNAmiR-196a-2 and breast cancer: a genetic and epigenetic associationstudy and functional analysis. Cancer Res. 2009;69:5970–7.

39. Hu Z, Liang J, Wang Z, Tian T, Zhou X, et al. Common geneticvariants in pre-microRNAs were associated with increased risk ofbreast cancer in Chinese women. Hum Mutat. 2009;30:79–84.

40. Catucci I, Yang R, Verderio P, Pizzamiglio S, Heesen L, et al.Evaluation of SNPs in miR-146a, miR196a2 and miR-499 as low-

542 Tumor Biol. (2014) 35:529–543

Page 15: Ethnicity modifies the association between functional microRNA polymorphisms and breast cancer risk: a HuGE meta-analysis

penetrance alleles in German and Italian familial breast cancer cases.Hum Mutat. 2010;31:E1052–7.

41. Pastrello C, Polesel J, Della Puppa L, Viel A, Maestro R. Associationbetween Hsa-miR-146a genotype and tumor age-of-onset in BRCA1/BRCA2-negative familial breast and ovarian cancer patients. Carci-nogenesis. 2010;31:2124–6.

42. Yang R, Schlehe B, Hemminki K, Sutter C, Bugert P, et al. A geneticvariant in the pre-miR-27a oncogene is associated with a reducedfamilial breast cancer risk. Breast Cancer Res Treat. 2010;121:693–702.

43. Jedlinski DJ, Gabrovska PN, Weinstein SR, Smith RA, Griffiths LR.Single nucleotide polymorphism in Hsa-miR-196a-2 and breast can-cer risk: a case control study. Twin ResHumGenet Off J Int Soc TwinStud. 2011;14:417–21.

44. Zhang P (2011) Polymorphisms of microRNA and ESR1 genes andtheir association with triple negative breast cancer risk and prognosis.Ph.D. thesis, Beijing Union Medical College

45. Alshatwi AA, Shafi G, Hasan TN, Syed NA, Al-Hazzani AA, et al.Differential expression profile and genetic variants of microRNAssequences in breast cancer patients. PloS One. 2012;7:e30049.

46. Catucci I, Verderio P, Pizzamiglio S, Bernard L, Dall’olio V, et al. TheSNP rs895819 in miR-27a is not associated with familial breastcancer risk in Italians. Breast Cancer Res Treat. 2012;133:805–7.

47. Linhares JJ, Azevedo Jr M, Siufi AA, de Carvalho CV,Wolgien MD,et al. Evaluation of single nucleotide polymorphisms in microRNAs(Hsa-miR-196a2 rs11614913 C/T) fromBrazilian women with breastcancer. BMC Med Genet. 2012;13:119.

48. Smith RA, Jedlinski DJ, Gabrovska PN,Weinstein SR, Haupt L, et al.A genetic variant located in miR-423 is associated with reducedbreast cancer risk. Cancer Genomics Proteomics. 2012;9:115–8.

49. Zhang M, Jin M, Yu Y, Zhang S, Wu Y, et al. Associations of miRNApolymorphisms and female physiological characteristics with breastcancer risk in Chinese population. Eur J Cancer Care. 2012;21:274–80.

50. Chen T, Li Z, Yan J, Yang X, Salminen W. MicroRNA expressionprofiles distinguish the carcinogenic effects of riddelliine in rat liver.Mutagenesis. 2012;27:59–66.

51. Ryan BM, Robles AI, Harris CC. Genetic variation in microRNAnetworks: the implications for cancer research. Nat Rev Cancer.2010;10:389–402.

52. Landi D, Gemignani F, Landi S. Role of variations withinmicroRNA-binding sites in cancer. Mutagenesis. 2012;27:205–10.

53. Negrini M, Calin GA. Breast cancer metastasis: a microRNA story.Breast Cancer Res. 2008;10:203.

54. Teraoka SN, Bernstein JL, Reiner AS, Haile RW, Bernstein L, et al.Single nucleotide polymorphisms associated with risk for

contralateral breast cancer in the Women’s Environment, Cancer,and Radiation Epidemiology (WECARE) Study. Breast CancerRes. 2011;13:R114.

55. Gaur A, Jewell DA, Liang Y, Ridzon D, Moore JH, et al. Character-ization of microRNA expression levels and their biological correlatesin human cancer cell lines. Cancer Res. 2007;67:2456–68.

56. Kumar MS, Lu J, Mercer KL, Golub TR, Jacks T. ImpairedmicroRNA processing enhances cellular transformation and tumori-genesis. Nat Genet. 2007;39:673–7.

57. Chu H, Wang M, Shi D, Ma L, Zhang Z, et al. Hsa-miR-196a2rs11614913 polymorphism contributes to cancer susceptibility: evi-dence from 15 case–control studies. PloS One. 2011;6:e18108.

58. Gao LB, Bai P, Pan XM, Jia J, Li LJ, et al. The association betweentwo polymorphisms in pre-miRNAs and breast cancer risk: a meta-analysis. Breast Cancer Res Treat. 2011;125:571–4.

59. Qiu LX, Wang Y, Xia ZG, Xi B, Mao C, et al. Mir-196a2c allele is alow-penetrant risk factor for cancer development. Cytokine.2011;56:589–92.

60. Tian T, Xu Y, Dai J, Wu J, Shen H, et al. Functional polymorphismsin two pre-microRNAs and cancer risk: a meta-analysis. Int J MolEpidemiol Genet. 2010;1:358–66.

61. Wang F, Ma YL, Zhang P, Yang JJ, Chen HQ, et al. A genetic variantin microRNA-196a2 is associated with increased cancer risk: a meta-analysis. Mol Biol Rep. 2012;39:269–75.

62. Garcia AI, Cox DG, Barjhoux L, Verny-Pierre C, Barnes D, et al(2011) The rs2910164:G>c snp in themir146a gene is not associatedwith breast cancer risk in brca1 and brca2 mutation carriers. HumanMutat 32:1004–1007

63. Zhong S, Chen Z, Xu J, Li W, Zhao J (2013) Pre-mir-27a rs895819polymorphism and cancer risk: a meta-analysis. Mol Biol Rep40:3181–3186

64. Mertens-Talcott SU, Chintharlapalli S, Li X, Safe S. The oncogenicmicroRNA-27a targets genes that regulate specificity protein tran-scription factors and the G2-M checkpoint in MDA-MB-231 breastcancer cells. Cancer Res. 2007;67:11001–11.

65. Guttilla IK, White BA. Coordinate regulation of foxo1 by miR-27a,miR-96, and miR-182 in breast cancer cells. J Biol Chem.2009;284:23204–16.

66. Ma Y, Yu S, Zhao W, Lu Z, Chen J. MiR-27a regulates the growth,colony formation and migration of pancreatic cancer cells bytargeting sprouty2. Cancer Lett. 2010;298:150–8.

67. Wacholder S, Chanock S, Garcia-Closas M, El Ghormli L, RothmanN. Assessing the probability that a positive report is false: an ap-proach for molecular epidemiology studies. J Natl Cancer Inst.2004;96:434–42.

Tumor Biol. (2014) 35:529–543 543


Recommended