Molecular Biology Reports

, Volume 40, Issue 4, pp 3003–3013

Effects of common polymorphisms rs2910164 in miR-146a and rs3746444 in miR-499 on cancer susceptibility: a meta-analysis

Authors

  • Zhihua Yin
    • Department of Epidemiology, School of Public HealthChina Medical University
    • Key Laboratory of Cancer Etiology and Prevention (China Medical University), Liaoning Provincial Education Department
  • Lei Yan
    • Department of Medical InformaticsChina Medical University
  • Zhigang Cui
    • China Medical University
  • Xuelian Li
    • Department of Epidemiology, School of Public HealthChina Medical University
    • Key Laboratory of Cancer Etiology and Prevention (China Medical University), Liaoning Provincial Education Department
  • Yangwu Ren
    • Department of Epidemiology, School of Public HealthChina Medical University
    • Key Laboratory of Cancer Etiology and Prevention (China Medical University), Liaoning Provincial Education Department
    • Department of Epidemiology, School of Public HealthChina Medical University
    • Key Laboratory of Cancer Etiology and Prevention (China Medical University), Liaoning Provincial Education Department
    • Department of Epidemiology, Key Laboratory of Cancer Etiology and InterventionChina Medical University, University of Liaoning Province
Article

DOI: 10.1007/s11033-012-2372-7

Cite this article as:
Yin, Z., Yan, L., Cui, Z. et al. Mol Biol Rep (2013) 40: 3003. doi:10.1007/s11033-012-2372-7
  • 644 Views

Abstract

MicroRNAs (miRNAs) are a class of new non-coding RNA, which may play a more important role in the pathogenesis of human cancers. Rs2910164 in miR-146a and rs3746444 in miR-499 are shown to be associated with increased/decreased cancer risk. We performed a meta-analysis to systematically summarize the possible association. We retrieved the relevant articles from PubMed databases. Studies were selected using specific inclusion and exclusion criteria. ORs and 95% CIs were calculated to access the strength of association between microRNA polymorphism and cancer risk. All analyses were performed using the Stata software. Twenty-nine studies were included in this meta-analysis. There were not significant associations between miR-146a rs2910164 and miR-499 rs3746444 polymorphisms with overall cancer risk. In the subgroup analysis by ethnicity, significantly affected cancer risks were found among Asians for both rs2910164 (GC vs. GG: OR = 0.89, 95% CI = 0.82–0.96; CC vs. GG: OR = 0.80, 95% CI = 0.66–0.97; GC + CC vs. GG: OR = 0.86, 95% CI = 0.76–0.97; C vs. G: OR = 0.91, 95% CI = 0.82–1.00) and rs3746444 (GG + AG vs. AA: OR = 1.21, 95% CI = 1.00–1.46). In the tumor type subgroup analysis, rs2910164 C allele decreased the risk of hepatocellular carcinoma (C vs. G: OR = 0.89, 95% CI = 0.80–1.00) and cervical squamous cell carcinoma (C vs. G: OR = 0.72, 95% CI = 0.62–0.84). The rs2910164 in miR-146a and the rs3746444 in miR-499 are likely to be associated with cancer risk.

Keywords

Lung cancerMicroRNASingle nucleotide polymorphismSusceptibilityMeta-analysis

Background

MicroRNAs (miRNAs) are a class of new non-coding RNA, which extensively exist in plant, eelworm and human cell. miRNA regulate expression by binding to cis-regulatory regions of 3′-UTR regions of genes, restrain the translation of genes at post-transcription level [1]. MiRNAs are one of the several classes of small RNA guides that provide sequence specificity to RNA silencing pathways [2]. Through imperfect pairing with target mRNAs of protein-coding genes and transcriptional or posttranscriptional regulation of their expression, miRNAs are involved in crucial biological processes, including development, differentiation, apoptosis, proliferation and various diseases including cancer [3]. More than 50 % of miRNA genes are located in cancer-associated genomic regions or in fragile sites, suggesting that miRNAs may play a more important role in the pathogenesis of a limited range of human cancers than previously thought [4].

Single nucleotide polymorphisms (SNPs) occurring in the miRNA gene region may affect the property of miRNAs and the genetic variation may contribute to cancer risk by altering miRNA expression and/or maturation [5, 6]. Jazdzewski et al. [7] reported that the rs2910164 SNP in pre-miR-146a could reduce mature mir146a expression and affect target mRNA binding. Rs2910164 SNP in miR-146a and/or rs3746444 SNP in miR-499 are associated with an increased risk of cervical squamous cell cancer (CSCC) [8]. Rs2910164 SNP was also found to be associated with susceptibility to papillary thyroid carcinoma (PTC) [7]. But other studies of associations between the miR-146a rs2910164 variant C genotypes and the risk of breast cancer [9], bladder cancer [10], renal cancer [11] failed to show any overall association in Chinese or Caucasian populations. Tian et al. [12] reported that no significant association was observed between these two SNPs and lung cancer. The population-based case–control study of gallbladder cancer (GBC) in North India for genetic polymorphism in miR-146a(rs2910164) and miR-499 (rs3746444) genes suggested an increasing trend of association of GBC risk albeit the associations were not significant [13].

Up to now, the role of genetic variants in miRNAs on cancer susceptibility is mostly unknown. There are some meta-analysis reports on the risk of overall cancer with miR-146a rs2910164 [1417] and only one on miR-499 rs3746444 [18], in which the most recent update time was July 2011. However the relationship between these two SNPs and cancer risk is the hot topic, many studies have been published since then [1927]. In addition, these reported results were contradictory and inconclusive. So we perform an updated meta-analysis on all available case–control studies to access the overall cancer risk with both rs2910164 in miR-146a and rs3746444 in miR-499.

Materials and methods

Data sources

We retrieved the articles using the following terms “miRNA or miR-146a or miR-499” and “cancer” and “polymorphism” from PubMed. (Last search was updated on Aug 2012). We evaluated potentially relevant publications by examining their titles and abstracts and all studies matching the eligible criteria were retrieved.

Study selection and data extraction

Eligible studies were selected according to the following explicit inclusion criteria: (a) evaluation of the rs2910164 and/or rs3746444 and cancer risks, (b) using the methodology of a case–control study. (c) There was sufficient published data for the computation of odds ratios (ORs) with 95 % confidence intervals (95% CIs).

Duplicate and obviously unrelated articles were eliminated by a single reviewer (Z.Y.). Abstracts of the remaining articles were examined independently by three reviewers (Z.Y., Z.C., and L.Y.) to determine whether the full-text article should be obtained. The following information was sought from each publication: first author’s name, publication date, country origin, ethnicity, cancer type, control characteristics, total number of cases and controls, and numbers of cases and controls with miR-146a G/C, miR-499 A/G genotypes, respectively. For study including subjects of different countries of origin group, we extracted data separately.

Statistical methods

We first assessed Hardy–Weinberg equilibrium (HWE) for each study using χ2-test in control groups. ORs corresponding to 95% CIs were calculated to access the strength of association between microRNA SNPs and cancer risks. Pooled ORs were obtained from combination of single study by heterozygote comparison (GC vs. GG for rs2910164; AG vs. AA for rs3746444), homozygote comparison (CC vs. GG for rs2910164; GG vs. AA for rs3746444), dominant model (CC + GC vs. GG for rs2910164; GG + AG vs. AA for rs3746444), recessive model (CC vs. GC + GG for rs2910164; GG vs. AG + AA for rs3746444) and allelic model (C vs. G for rs2910164; G vs. A for rs3746444) respectively. For each genetic comparison model, subgroup analysis according to ethnicity was investigated to estimate ethnic-specific ORs for Asian and Caucasian. Meanwhile stratified analyses by tumor type or control characteristics were also applied for each genetic comparison model.

We investigated the between-study heterogeneity by the Cochran’s Q test and quantified by I2 (a significance level of P < 0.10 and/or I2 ≥ 50 %). To obtain summary statistics for ORs of microRNA polymorphism and cancer risk, we performed initial analyses with a fixed-effect model and confirmatory analyses with a random-effect model if there was significant heterogeneity.

The effect of publication bias was examined by inverted funnel plots, the Egger’s test and the Begg’s test. The significance of the intercept was determined by the t test as suggested by Egger’s test. All of P values were two-sided and all analyses were performed using the Stata software version 11.0 (Stata Corp, College Station, TX).

Results

According to these criteria, a total of 45 articles were eligible. Six studies of meta-analysis and two studies of review were excluded. Eight studies were excluded because of no cancer risk and data missing. Finally 29 articles were included and used in quantitative synthesis for systematic review [713, 1940]. These studies included 42 data sets of two SNPs. Twenty-nine data sets were about miR-146a(rs2910164) SNP, including 15,027 cases and 21,081 controls. There were 6,801 cases and 8,147 controls in 13 data sets for miR-499(rs3746444) SNP. Flow chart of the study selection process was shown in Fig. 1. The characteristics of selected studies are summarized in Table 1. Of the 29 studies, sample sizes ranged from 200 to 4005. There were 12 studies of Europeans and 17 studies of Asians. Almost all of the cases were histologically confirmed. Controls were mainly frequency matched by gender and age, of which 18 were population-based and 11 were hospital-based. The distribution of genotypes in the controls was mostly in HWE, except for two data sets of rs2910164 and three data sets of rs3746444. Studies with the controls not in HWE were subjected to a sensitivity analysis.
https://static-content.springer.com/image/art%3A10.1007%2Fs11033-012-2372-7/MediaObjects/11033_2012_2372_Fig1_HTML.gif
Fig. 1

Flow chart of the study selection process

Table 1

Characteristics of all studies in meta-analysis

Author [Ref]

Country

Ethnicity

Cancer

SNP

Genotyping method

Study design

No.

Case

Control

Type

 

GG/AA

GC/AG

CC/GG

GG/AA

GC/AG

CC/GG

HWE(P)

Jazdzewski et al. [7]

Finland Poland USA

Caucasian

PTC

rs2910164

TaqMan

PB

608/901

305

287

16

526

320

55

0.50

Horikawa et al. [11]

US

Caucasian

RCC

rs2910164

SNPlex

PB

261/235

144

103

14

126

94

15

0.65

Yang et al. [10]

US

Caucasian

GBC

rs2910164

SNPlex

PB

691/674

414

242

35

385

258

31

0.14

Hu et al. [9]

China

Asian

BC

rs2910164

PCR–RFLP

PB

1,009/1,093

165

515

329

180

551

362

0.22

rs3746444

PCR–RFLP

PB

1,009/1,093

707

258

44

816

248

29

0.06

Xu et al. [33]

China

Asian

HCC

rs2910164

PCR–RFLP

HB

479/504

80

241

158

58

249

197

0.12

Tian et al. [12]

China

Asian

LC

rs2910164

PCR–RFLP

PB

1,058/1,035

360

510

188

364

502

169

0.85

rs3746444

PCR–RFLP

PB

1,058/1,035

781

253

24

755

254

26

0.40

Hoffman et al. [34]

US

Caucasian

BC

rs2910164

Massarray multiplex

PB

439/478

234

176

29

273

178

27

0.77

Srivastava et al. [13]

India

Caucasian

GBC

rs2910164

PCR–RFLP

PB

230/224

129

90

11

138

81

5

0.08

rs3746444

PCR–RFLP

PB

230/230

112

97

21

121

94

15

0.57

Liu et al. [28]

USA

Caucasian

SCCHN

rs2910164

PCR–RFLP

HB

1,109/1,130

630

411

68

655

405

70

0.49

rs3746444

PCR–RFLP

HB

1,109/1,130

745

309

55

710

366

54

0.44

Okubo et al. [29]

Japan

Asian

GC

rs2910164

PCR–RFLP

HB

552/697

73

243

236

121

322

254

0.28

rs3746444

PCR–RFLP

HB

552/697

364

151

37

466

198

33

0.05

Catucci et al. [31]

Italy, Germany

Caucasian

BC

rs2910164

Taqman PCR

PB

1,559/2,147

860

590

109

1186

838

123

0.11

Italy, Germany

Caucasian

BC

rs3746444

Taqman PCR

PB

1,579/2,167

950

545

84

1305

742

120

0.28

Zeng et al. [35]

China

Asian

GC

rs2910164

PCR–RFLP

HB

304/304

62

153

89

53

132

119

0.12

Pastrello et al. [36]

Italian

Caucasian

BC/OC

rs2910164

PCR-direct sequencing

PB

101/155

60

36

5

90

59

6

0.33

Guo et al. [37]

China

Asian

ESCC

rs2910164

SNPshot assay

PB

444/468

234

190

20

206

220

42

0.12

Xu et al. [38]

China

Asian

PC

rs2910164

PCR–RFLP

HB

251/280

68

135

48

54

150

76

0.19

Zhou et al. [8]

China

Asian

CSCC

rs2910164

PCR–RFLP

PB

226/309

43

113

70

34

159

116

0.06

rs3746444

PCR–RFLP

PB

226/309

134

84

8

223

71

15

0.005

George et al. [30]

India

Caucasian

PC

rs2910164

PCR–RFLP

PB

159/230

4

79

76

7

107

116

0.002

rs3746444

PCR–RFLP

PB

159/230

48

98

13

104

92

34

0.07

Mittal et al. [32]

India

Caucasian

BC

rs2910164

PCR–RFLP

PB

212/250

127

79

6

135

108

7

0.007

rs3746444

PCR–RFLP

PB

212/250

95

92

25

121

94

35

0.02

Permuth et al. [39]

US

Caucasian

Glioma

rs2910164

Illumina’s GoldenGate technology

PB

593/614

345

198

50

375

214

25

0.42

Hishida et al. [40]

Japan

Asian

GC

rs2910164

PCR-confronting two-pair primers

HB

583/1,637

82

271

230

229

775

633

0.12

Akkiz et al. [19]

Turkish

Caucasian

HCC

rs3746444

PCR–RFLP

PB

222/222

45

87

90

47

93

82

0.04

Akkiz et al. [20]

Turkish

Caucasian

HCC

rs2910164

PCR–RFLP

PB

222/222

137

75

10

144

67

11

0.38

Yue et al. [21]

China

Asian

CSCC

rs2910164

PCR–RFLP

HB

447/443

118

224

105

87

206

150

0.29

Garcia et al. [22]

French

Caucasian

BC

rs2910164

TaqMan

HB

1,130/596

676

388

66

352

220

24

0.15

Zhou et al. [23]

China

Asian

GC

rs2910164

Taqman PCR

HB

1,686/1895

578

822

286

551

951

393

0.64

Lung et al. [24]

China

Asian

NC

rs2910164

Tm-shift allele-specific genotyping

PB

229/3,776

24

88

117

497

1807

1472

0.12

Zhou et al. [25]

China

Asian

HCC

rs2910164

PCR–RFLP

PB

186/483

33

86

67

71

254

158

0.06

rs3746444

PCR–RFLP

PB

186/483

141

41

4

371

100

12

0.1

Xiang et al. [26]

China

Asian

HCC

rs2910164

PCR–RFLP

HB

100/100

27

45

28

21

46

33

0.51

rs3746444

PCR–RFLP

HB

100/100

36

40

24

54

36

10

0.28

Kim et al. [27]

Korean

Asian

HCC

rs2910164

PCR–RFLP

PB

159/201

57

88

14

74

103

24

0.19

rs3746444

PCR–RFLP

PB

159/201

109

47

3

120

74

7

0.28

BC breast cancer, CSCC cervical squamous cell carcinoma, ESCC esophageal squamous cell carcinoma, GBC gallbladder cancer, GC gastric cancer, HCC hepatocellular carcinoma, LC lung cancer, NC nasopharyngeal carcinoma, OC ovarian cancer, PC prostate cancer, PTC papillary thyroid carcinoma, RCC renal cell carcinoma, SCCHN cell carcinoma of head and neck, HWE Hardy–Weinberg equilibrium

miR-146a(rs2910164) SNP

The C allele frequency of the miR-146a polymorphism (rs2910164) among the controls across different ethnicities ranged from 0.20 to 0.74. The C allele frequencies across different ethnicities were also observed. The average C allele frequencies in Asian and Caucasians populations were 55.8 and 23.8 %, respectively. The overall ORs with its 95% CIs didn’t show statistically association between rs2910164 polymorphism and cancer risk (GC vs. GG: OR = 0.96, 95% CI = 0.89–1.03, P = 0.008 for heterogeneity, I2 = 42.8 %; CC vs. GG: OR = 0.92, 95% CI = 0.78–1.07, P < 0.001 for heterogeneity, I2 = 69.1 %; CC + GC vs. GG: OR = 0.95, 95% CI = 0.88–1.04, P < 0.001 for heterogeneity, I2 = 56.6 %; C vs. G: OR = 0.97, 95% CI = 0.91–1.03, P < 0.001 for heterogeneity, I2 = 69.1 %) (Table 2). Through stratified analyses, the heterogeneity of the subgroup significantly reduced. Table 3 showed the results of stratified analyses for miR-146a rs2910164 polymorphism. In the subgroup analysis by ethnicity, significantly decreased cancer risks were found among Asians (GC vs. GG: OR = 0.89, 95% CI = 0.82–0.96, P = 0.008; CC vs. GG: OR = 0.80, 95% CI = 0.66–0.97, P = 0.022; dominant model: OR = 0.86, 95% CI = 0.76–0.97, P = 0.013; allele contrast: OR = 0.91, 95% CI = 0.82–1.00, P = 0.039). No significantly evaluated risk was found among Caucasians in any of the genetic models. In the tumor type subgroup analysis, the ORs of some models with cervical squamous cell carcinoma (CSCC) and hepatocellular carcinoma (HCC) are statistically significant. For HCC, the significant results were observed in homozygote comparison and allelic model (CC vs. GG: OR = 0.71, 95% CI = 0.55–0.92, P = 0.001). Individuals with GC or CC genotype were in decreased CSCC risk than those with GG genotype (ORs were 0.72 and 0.50, 95% CIs were 0.55–0.95 and 0.37–0.68, P values were 0.042 and < 0.001, respectively), and the results were significant in recessive model and allelic model (ORs were 0.65 and 0.72, 95% CIs were 0.52–0.82 and 0.62–0.84, P values were all less than 0.001). In the subgroup analysis by control characteristics, the ORs were significant in the heterozygote comparison, dominant and allelic model for the hospital-based control (GC vs. GG: OR = 0.90, 95% CI = 0.83–0.98, P = 0.039; CC + GC vs. GG: OR = 0.87, 95% CI = 0.77–0.99, P = 0.035; C vs. G: OR = 0.89, 95% CI = 0.80–0.99, P = 0.037) (Table 3).
Table 2

Association between miR-146a(rs2910164) and miR-499(rs3746444) with cancer risk

 

Data set number

Fixed effect

Random effect

Phet

I2 (%)

rs2910164

 GC vs. GG

29

0.97 (0.92,1.02)

0.96 (0.89,1.03)

0.008

42.8

 CC vs. GG

29

0.91 (0.85,0.99)

0.92 (0.78,1.07)

0.000

69.1

 CC + GC vs. GG

29

0.97 (0.92,1.01)

0.95 (0.88,1.04)

0.000

56.6

 CC vs. GC + GG

29

0.96 (0.90,1.02)

0.95 (0.84,1.08)

0.000

69.2

 C vs. G

29

0.97 (0.94,1.00)

0.97 (0.91,1.03)

0.000

69.1

rs3746444

 AG vs. AA

13

1.04 (0.97,1.12)

1.12 (0.96,1.30)

0.000

69.9

 GG vs. AA

13

1.12 (0.96,1.29)

1.13 (0.93,1.38)

0.129

31.7

 GG + AG vs. AA

13

1.05 (0.98,1.13)

1.12 (0.98,1.29)

0.001

68.2

 GG vs. AG + AA

13

1.08 (0.94,1.24)

1.08 (0.89,1.32)

0.070

39.5

 G vs. A

13

1.05 (0.99,1.11)

1.09 (0.98,1.20)

0.001

62.8

Table 3

Pooled ORs and 95% CIs for mir-146a polymorphism of stratified meta-analysis

Subgroup

Genotype

No. of studies

Test of association

Test of heterogeneity

OR (95% CI)

Z

P value

Model

χ2

P value

I2 (%)

Asian

GC vs. GG

17

0.89 (0.82,0.96)

2.67

0.008

F

19.20

0.258

16.7

CC vs. GG

17

0.80 (0.66,0.97)

2.29

0.022

R

58.85

0.000

72.8

CC + GC vs. GG

17

0.86 (0.76,0.97)

2.50

0.013

R

35.46

0.003

54.9

CC vs. GC + GG

17

0.88 (0.76,1.02)

1.66

0.097

R

61.18

0.000

73.8

C vs. G

17

0.91 (0.82,1.00)

2.06

0.039

R

67.50

0.000

76.3

Caucasian

GC vs. GG

11

1.05 (0.95,1.17)

0.97

0.334

R

19.94

0.030

49.9

CC vs. GG

11

1.15 (0.92,1.45)

1.25

0.213

R

18.53

0.047

46.0

CC + GC vs. GG

11

1.06 (0.99,1.13)

1.53

0.127

F

13.04

0.222

23.3

CC vs. GC + GG

11

1.13 (0.88,1.45)

0.97

0.333

R

23.30

0.01.

57.1

C vs. G

11

1.06 (1.00,1.12)

2.04

0.041

F

7.66

0.662

0.0

PC

GC vs. GG

2

0.76 (0.51,1.14)

1.34

0.182

F

0.76

0.384

0.0

CC vs. GG

2

0.56 (0.35,0.90)

1.37

0.172

F

1.42

0.234

29.5

CC + GC vs. GG

2

0.69 (0.47,1.01)

1.93

0.054

F

0.91

0.340

0.0

CC vs. GC + GG

2

0.76 (0.57,1.01)

1.59

0.111

F

1.41

0.235

29.1

C vs. G

2

0.80 (0.66,0.97)

1.60

0.109

F

1.64

0.200

39.1

GC

GC vs. GG

4

0.91 (0.81,1.02)

0.43

0.667

F

5.54

0.136

45.8

CC vs. GG

4

0.92 (0.63,1.34)

0.46

0.648

R

18.90

0.000

84.1

CC + GC vs. GG

4

0.96 (0.74,1.24)

0.31

0.753

R

11.14

0.011

73.1

CC vs. GC + GG

4

0.92 (0.70,1.21)

0.61

0.543

R

18.18

0.000

83.5

C vs. G

4

0.95 (0.78,1.16)

0.48

0.633

R

22.05

0.000

86.4

HCC

GC vs. GG

5

0.89 (0.73,1.09)

1.00

0.319

F

5.11

0.276

21.8

CC vs. GG

5

0.71 (0.55,0.92)

2.56

0.011

F

2.42

0.659

0.0

CC + GC vs. GG

5

0.86 (0.71,1.04)

1.26

0.207

F

5.58

0.233

28.3

CC vs. GC + GG

5

0.86 (0.72,1.04)

1.54

0.125

F

3.81

0.432

0.0

C vs. G

5

0.89 (0.80,1.00)*

1.61

0.108

F

4.58

0.333

12.7

CSCC

GC vs. GG

2

0.72 (0.55,0.95)*

2.04

0.042

F

1.30

0.254

23.1

CC vs. GG

2

0.50 (0.37,0.68)*

4.39

0.000

F

0.06

0.814

0.0

CC + GC vs. GG

2

0.62 (0.54,1.14)

3.41

0.001

F

0.76

0.382

0.0

CC vs. GC + GG

2

0.65 (0.52,0.82)*

3.64

0.000

F

0.84

0.359

0.0

C vs. G

2

0.72 (0.62,0.84)*

4.33

0.000

F

0.07

0.796

0.0

BC

GC vs. GG

5

0.97 (0.89,1.07)

0.55

0.580

F

3.29

0.510

0.0

CC vs. GG

5

1.14 (0.97,1.34)

1.54

0.123

F

2.49

0.647

0.0

CC + GC vs. GG

5

1.00 (0.91,1.10)

0.01

0.990

F

3.07

0.546

0.0

CC vs. GC + GG

5

1.09 (0.95,1.25)

1.24

0.216

F

3.92

0.417

0.0

C vs. G

5

1.02 (0.96,1.09)

0.62

0.533

F

2.66

0.616

0.0

GBC

GC vs. GG

2

0.94 (0.89,1.14)

0.15

0.877

F

1.86

0.173

46.2

CC vs. GG

2

1.22 (0.78,1.92)

0.81

0.418

F

1.75

0.185

43.0

CC + GC vs. GG

2

1.02 (0.74,1.42)

0.12

0.901

R

2.42

0.120

58.7

CC vs. GC + GG

2

1.26 (0.81,1.97)

0.92

0.359

F

1.30

0.254

23.0

C vs. G

2

1.06 (0.79,1.41)

0.37

0.708

R

2.63

0.105

62.0

PB

GC vs. GG

17

1.01 (0.91,1.12)

0.11

0.910

R

31.19

0.013

48.7

CC vs. GG

17

1.00 (0.81,1.22)

0.04

0.970

R

42.40

0.000

62.3

CC + GC vs. GG

17

1.02 (0.92,1.12)

0.34

0.737

R

31.10

0.013

48.1

CC vs. GC + GG

17

1.03 (0.86,1.22)

0.28

0.783

R

47.20

0.000

66.1

C vs. G

17

1.02 (0.95,1.10)

0.63

0.532

R

36.34

0.003

56.0

HB

GC vs. GG

11

0.90 (0.83,0.98)

2.06

0.039

F

12.73

0.239

21.5

CC vs. GG

11

0.81 (0.64,1.02)

1.77

0.077

R

38.34

0.000

73.9

CC + GC vs. GG

11

0.87 (0.76,0.99)

2.11

0.035

R

24.45

0.007

59.1

CC vs. GC + GG

11

0.87 (0.73,1.03)

1.62

0.106

R

34.36

0.000

70.9

C vs. G

11

0.89 (0.80,0.99)

2.08

0.037

R

40.79

0.000

75.5

OR odds ratio, vs versus, R random effect model, F fixed effect model, BC breast cancer, CSCC cervical squamous cell carcinoma, GBC gallbladder cancer, GC gastric cancer, HCC hepatocellular carcinoma, PC prostate cancer, PTC papillary thyroid carcinoma, PB population-based, HB hospital-based

Heterogeneity between studies was observed in overall comparisons and also subgroup analyses. We assessed the source of heterogeneity by cancer type, ethnicity and source of control. Cancer type (χ2 = 74.74, df = 13, P < 0.001) and ethnicity (χ2 = 71.82, df = 1, P < 0.001), but not source of controls (χ2 = 2.19, df = 1, P = 0.138) contributed substantially to heterogeneity. Every one single study involved in the meta-analysis was deleted each time to reflect the influence of the individual data set to the pooled ORs. This procedure did not change the pooled ORs supporting the robustness of our findings. Furthermore, when two studies whose genotype frequencies in controls derived from HWE were excluded, the results were in agreement with the findings from foregoing analysis for all populations.

No publication bias was detected by either the inverted funnel plot or Begg’s test. The shapes of the funnel plot for the comparison of the G allelic and the C allelic of rs2910164 SNP seemed approximately symmetrical and P value of the Egger’ test was not statistical significant (t = 1.49, P = 0.164).

miR-499(rs3746444) SNP

The G allele frequency of the miR-499 polymorphism (rs3746444) among the controls across different ethnicities ranged from 0.21 to 0.58. The G allele frequencies across different ethnicities were also observed. The average G allele frequencies in Asian and Caucasians populations were 17.8 and 24.5 %, respectively. Five kinds of genetic models did not produce significant association among all studies. Considering significant heterogeneity in the above overall analysis, stratified analyses were conducted by ethnicity, tumor type and source of control (Table 4). In the stratified analysis by ethnicity, the significant associations were found among Asians in dominant model (GG + AG vs. AA: OR = 1.21, 95% CI = 1.00–1.46, P = 0.048). There was no evidence for the influence of rs3746444 polymorphism on cancer susceptibility of cases for stratified by tumor type or control characteristics.
Table 4

Pooled ORs and 95% CIs for mir-499 polymorphism of stratified meta-analysis

Subgroup

Genotype

No. of studies

Test of association

Test of heterogeneity

OR (95% CI)

Z

P value

Model

χ2

P value

I2 (%)

Asian

AG vs. AA

9

1.22 (1.00,1.51)

1.91

0.056

R

28.22

0.000

71.6

GG vs. AA

9

1.18 (0.86,1.62)

1.02

0.309

R

14.74

0.064

45.7

GG + AG vs. AA

9

1.21 (1.00,1.46)

1.97

0.048

R

25.91

0.001

69.1

GG vs. AG + AA

9

1.05 (0.74,1.48)

0.27

0.787

R

18.27

0.019

56.2

G vs. A

9

1.14 (0.98,1.32)

1.75

0.081

R

22.20

0.005

64.0

Caucasian

AG vs. AA

4

0.94 (0.85,1.05)

0.79

0.427

F

4.52

0.207

34.2

GG vs. AA

4

1.03 (0.84,1.26)

0.28

0.783

F

1.60

0.660

0.0

GG + AG vs. AA

4

1.05 (0.98,1.13)

0.58

0.561

F

4.55

0.208

34.1

GG vs. AG + AA

4

1.06 (0.88,1.28)

0.58

0.564

F

1.47

0.689

0.0

G vs. A

4

0.98 (0.90,1.06)

0.21

0.833

F

4.46

0.216

32.7

HCC

AG vs. AA

4

1.08 (0.97,1.12)

0.08

0.933

F

5.22

0.156

42.5

GG vs. AA

4

1.14 (0.77,1.69)

0.61

0.540

R

8.28

0.040

63.8

GG + AG vs. AA

4

1.11 (0.94,1.31)

0.42

0.674

R

9.44

0.024

68.2

GG vs. AG + AA

4

1.09 (0.74,1.60)

0.76

0.447

R

6.09

0.107

50.8

G vs. A

4

1.09 (0.92,1.30)

0.60

0.547

R

13.36

0.004

77.5

BC

AG vs. AA

3

0.99 (0.78,1.26)

1.32

0.186

F

2.50

0.286

20.0

GG vs. AA

3

1.27 (0.60,2.69)

0.66

0.512

R

4.84

0.089

58.7

GG + AG vs. AA

3

1.09 (0.73,1.64)

1.27

0.203

F

3.78

0.151

47.1

GG vs. AG + AA

3

1.26 (0.70,2.27)

0.44

0.661

R

4.82

0.090

58.5

G vs. A

3

1.12 (0.77,1.62)

1.02

0.309

R

5.81

0.055

65.6

PB

AG vs. AA

9

1.16 (0.97,1.39)

1.59

0.111

R

27.64

0.001

71.1

GG vs. AA

9

1.07 (0.89,1.28)

0.71

0.480

F

8.08

0.426

0.9

GG + AG vs. AA

9

1.14 (0.97,1.34)

1.59

0.111

R

23.71

0.003

66.3

GG vs. AG + AA

9

1.02 (0.86,1.21)

0.10

0.918

F

11.47

0.177

30.2

G vs. A

9

1.09 (0.97,1.21)

1.45

0.148

R

16.65

0.034

52.0

HB

AG vs. AA

4

1.03 (0.79,1.33)

0.20

0.838

R

8.11

0.044

71.1

GG vs. AA

4

1.33 (0.83,2.14)

1.19

0.233

R

8.96

0.030

66.5

GG + AG vs. AA

4

1.10 (0.82,1.47)

0.65

0.518

R

11.51

0.009

73.9

GG vs. AG + AA

4

1.27 (0.84,1.92)

1.12

0.261

R

7.42

0.060

59.5

G vs. A

4

1.13 (0.87,1.46)

0.90

0.366

R

14.85

0.002

79.8

OR odds ratio, vs versus; R random effect model, F fixed effect model, BC breast cancer, CSCC cervical squamous cell carcinoma, GBC gallbladder cancer, GC gastric cancer, HCC hepatocellular carcinoma, PC prostate cancer, PTC papillary thyroid carcinoma, PB population-based, HB hospital-based

Heterogeneity between studies was also assessed as mentioned above. The source of heterogeneity by cancer type, ethnicity and source of control was assessed. Cancer type (χ2 = 31.00, df = 7, P < 0.001), but not ethnicity (χ2 = 3.76, df = 1, P = 0.052) or source of controls (χ2 = 0.96, df = 1, P = 0.328) contributed substantially to heterogeneity. Every one single study was deleted each time and the pooled ORs did not change. Furthermore, the pooled OR estimates were similar with that of excluded three studies whose genotype frequencies in controls derived from HWE. No publication bias was indicated according to the results of the inverted funnel plot, Begg’s test and Egger’s test (t = −0.23, P = 0.820).

Discussion

It is well known that individual susceptibility plays important role in the development of most cancers. Polymorphisms of genes involved in carcinogenesis may have accounted for the susceptibility. Therefore, genetic susceptibility, especially single nucleotide polymorphism (SNP), to cancer has been a research focus in scientific community. Many studies have been done for figuring out the role of SNPs present in precursor and mature miRNA and their influences on susceptibility and progression of various cancers. Recently, genetic variants of the miR-146a and miR-499 genes in the etiology of several cancers have drawn increasing attention. The studies suggested that two common SNPs, rs2910164 in miR-146a and rs3746444 in miR-499, may alter mature miRNA expression and affect regulation of target mRNAs, further change cancer risk [79, 30]. Growing number of studies have been done to examine the relationship between these two SNPs and the risks of cancer. However, the results are inconclusive. To better understanding of the association between these polymorphisms and cancer risk, a meta-analysis with larger sample and subgroup analysis is necessary. The current study is the largest meta-analysis of the association between miR-146a rs2910164 and miR-499 rs3746444 polymorphisms with the risk of cancer.

In 2011, several meta-analyses were conducted to investigate the association between miR-146a rs2910164 polymorphism and overall cancer risk [1417]. At the beginning of 2012, there are meta-analysis reports about rs2910164 polymorphism with susceptibility to hepatocellular carcinoma [41], gastrointestinal cancers [42] and breast cancer [43], and one meta-analysis about miR-499 rs3746444 with overall cancer risk [18]. However, there are many studies about the relationship between these two SNPs and risks of various cancer types published since then [1927]. In addition, these reported results were contradictory and inconclusive. In consistent with those meta-analyses, we did not find statistical evidence for the associations between two SNPs (rs2910164 in miR-146a and rs3746444 in miR-499) and susceptibility of cancer before performing subgroup analysis and sensitivity analysis. However, based on larger sample sizes and increased statistical power, our data indicated these two polymorphisms might be impact factors in Asian individuals rather than in Caucasian individuals. Furthermore, rs2910164 C allele may decrease the risk of hepatocellular carcinoma and cervical squamous cell carcinoma.

The expression of miR-146a from the rs2910164 C allele is lower than that from the G allele in a thyroid cell line and this polymorphism could affect target mRNA binding [7]. The GG genotype of rs2910164 was found to be associated with increased expression level of mature miR-146a and promoted cell proliferation in hepatocellular cancer cells [11] and prostate cancer cells [44]. However, other studies showed the allele C of this polymorphism confer a higher expression level of mature miR-146a [45]. Thus, the results about the association between miR-146a SNP (rs2910164) and mature miR-146a production are controversial in various cancers. There are contentious results of miR-146a expression in various cancer tissues and cell lines. Some studies showed elevated expression in papillary thyroid cancer [46] and cervical cancer [47]. In other studies, reduced or absent expression level of miR-146a was found in prostate cancer [44], gastric cancer [48], pancreatic cancer [49] and breast cancer [50]. Reduction of miR-146a expression resulted in less efficient inhibition of target genes involved in the Toll-like receptor and cytokine signaling pathway (TRAF6, IRAK1), and PTC1, suggesting that increased production of miR-146a may have antitumorgenetic effect for PTC [7]. In this meta-analysis, we found a significant association between miR-146a(rs2910164) with hepatocellular carcinoma, cervical squamous cell carcinoma and prostate cancer, but not with overall cancer risk, indicating that rs2910164 polymorphism may play different roles in various human malignancies.

Considering the number of studies included in this article, we performed the stratified analyses by cancer types for breast cancer, cervical squamous cell carcinoma, GBC, gastric cancer, hepatocellular carcinoma and prostate cancer. Our results suggested that rs2910164 C allele might decrease the risk of hepatocellular carcinoma, cervical squamous cell carcinoma and prostate cancer, but not in other cancer types. The reason may be that the rs2910164 polymorphism may have different effect on carcinogenesis in different organs, reflecting the diversities of the susceptible factors for different tumor types. In addition, the observed different effects could be likely due to chance because studies with small sample size may have insufficient statistical power to detect a slight effect or may have generated a fluctuated risk estimate. So studies with larger sample size in different types of cancer are necessary to fully understand the relationship between the polymorphism and the risk of cancer.

As we know, the incidence of most genetic polymorphisms could vary between different ethnic populations. In this meta-analysis, we found highly significant differences in the prevalence of the rs2910164 C allele among controls between Asian (0.558) and Caucasian (0.238). In the stratified analyses by ethnicity, significantly affected cancer risks were found for both rs2910164 and rs3746444 among Asians but not in Caucasians. It showed that there may be ethnicity difference for association between two SNPs and cancer risks.

Attention must be payed to the relatively large heterogeneity in our results. We assessed the source of heterogeneity for allelic model according to ethnicity, cancer type and source of control. We found the sources of heterogeneity were mainly from ethnicity for rs2910164 and rs3746444 SNP. In the subgroup analyses stratified by tumor site and ethnicity respectively, it can be found that the heterogeneity of the subgroup reduced significantly. Therefore, it can be presumed that the relatively large heterogeneity mainly results from differences of ethnicity and tumor types.

Despite our efforts in performing a comprehensive analysis, some limitations exist in our meta-analysis. First, our analysis used published international studies, which could arose publication bias, although the results for publication bias in our study were not statistically significant. Second, lack of the original data of available studies limited our further evaluation of potential interactions, such as age, gender, family history, environmental factors and lifestyle. Third, there was no study in African population.

In conclusion, our meta-analysis supported that the rs2910164 in miR-146a and the rs3746444 in miR-499 more likely contribute to be associated with cancer risk, especially in the subgroups of Asians. Besides, the C allele of rs2910164 in miR-146a might be associated with protection from hepatocellular carcinoma, cervical squamous cell carcinoma and prostate cancer. Future well-designed and larger population studies are of great value to confirm these findings. Moreover, combination of genetic factors together with environmental exposures should also be considered.

Conclusion

rs2910164 in miR-146a and the rs3746444 in miR-499 might be associated with cancer risk.

Acknowledgments

The authors are most grateful to all the participants in the present study. This study was supported by grant no.81102194 from National Natural Science Foundation of China. The funders had no role in study design, data collection and analysis, decision to publish, or preparation of the manuscript.

Conflict of interest

The authors declared that no competing interests exist.

Copyright information

© Springer Science+Business Media Dordrecht 2013