Meta-analysis of rs2910164 in miR-146a and risk of colorectal cancer
Methods
Search strategy and study selection
Systematic literature search in PubMed database updated in March 25, 2014 was performed using the following search strategy: (microRNA-146a OR miR-146a OR pre-miR-146a OR rs2910164) AND (gene OR polymorphism OR variation OR allele) AND (colorectal cancer OR colorectal neoplasm) without any restriction on language. All potentially relevant publications were retrieved, and the reference lists as well as previously published review articles and meta-analyses were checked to identify additional eligible studies. The inclusion criteria were: (i) unrelated case-control study design, (ii) genotype distribution of the controls were in Hardy-Weinberg equilibrium (HWE), and (iii) reporting odds ratios (ORs) with 95% confidence intervals (CIs) or sufficient data to calculate.
Data extraction and quality assessment
Data extraction was completed independently by two investigators (Mao and Li) to include the following items: study design, publication year, country, ethnicity, characteristics of controls and matching criteria, genotyping methods, total number of cases and controls, and allele frequencies in cases and controls. Quality assessment was done independently by two reviewers (Jing and Cai) using a set of structure criteria modified from previous studies [1, 2]. The total score ranges from 0 to 12, with a higher score indicating higher quality. Discrepancies were resolved by consensus and discussion.
Data synthesis and statistical analysis
Crude ORs and corresponding 95% CIs were used to assess the association between rs2910164 polymorphism and colorectal cancer risk. Chi-square-based Q-test and the I2 index were used to assess the heterogeneity across different studies. A random-effect model was used when a notable heterogeneity was observed (Q-test P0.1 and/or I250%); otherwise the fixed-effect model was applied [3]. Publication bias was determined by Begg’s funnel plot and Harbord’s test [4]. Statistical analyses were performed using STATA, version 11.0 (STATA Corp, College Station, Texas).
Results
A total of six eligible studies involving 2720 cases and 3371 controls were included in the current meta-analysis. Characteristics of these studies as well as genotype distributions in cases and controls were summarized in Supplementary Table 1. Of the six studies, five were conducted in Asians and one in Caucasians. G allele frequency of controls was the MAF in four studies, and C allele frequency was the MAF in two studies. Overall, no significant association between rs2910164 polymorphism and risk of colorectal cancer was observed in any genotype comparison under the random-effect model (the recessive model, OR=0.98, 95% CI: 0.77~1.25, Supplementary Figure 1; the allele model: OR=1.02, 95% CI: 0.86~1.21, Supplementary Figure 2A; the dominant model, OR= 1.05, 95% CI: 0.74~1.49, Supplementary Figure 2B; the additive model, ORGG vs. CC=1.05, 95% CI: 0.74~1.50, Supplementary Figure 2C, ORGC vs. CC=1.03, 95% CI: 0.71~1.50, Supplementary Figure 2D). Significant evidence of between-study heterogeneities was observed in all genetic models (P for heterogeneity <0.1 or I2>50%). After stratification by MAF in controls, heterogeneities decreased in the subgroup of G>C (P for heterogeneity >0.1). However, using meta-regression analysis, we did not find differences in MAF (P=0.203), ethnicity (P=0.833), or quality score (P=0.203) contribute significantly to the observed heterogeneity. No significant evidence of publication bias was found in any genetic model using Harbord’s test (P>0.05). Begg’s funnel plot did not reveal any remarkable asymmetry in the distribution of scatter points under any genetic model (e.g., the recessive model, P=0.707, Supplementary Figure 3).
Supplementary table 1. Characteristics of eligible studies in the meta-analysisAuthors / Publication / Country / Ethnicity / Matching / Genotyping / Sample size / Genotype distributions (CC/CG/GG) / Quality / HWE
Year / origin / criteria / method / cases/controls / cases / controls / score
Min et al. [5] / 2012 / Korea / Asian / NA / PCR-RFLP / 446/502 / 151/233/62 / 188/245/69 / 10 / Yes
Hezova et al. [6] / 2012 / Czech Republic / Caucasian / age / TaqMan / 197/212 / 12/70/115 / 9/79/124 / 7.5 / Yes
Lv et al. [7] / 2013 / China / Asian / NA / PCR-RFLP / 331/513 / 47/230/54 / 143/274/96 / 7 / Yes
Ma et al. [8] / 2013 / China / Asian / age/gender / TaqMan / 1147/1203 / 169/534/444 / 192/614/397 / 11 / Yes
Chae et al. [9] / 2013 / Korea / Asian / NA / PCR-RFLP / 399/568 / 156/182/61 / 165/282/121 / 10 / Yes
Hu et al. [10] / 2013 / China / Asian / NA / PCR-RFLP / 200/373 / 84/82/34 / 142/187/44 / 10 / Yes
HWE: Hardy-Weinberg equilibrium; NA: not available; PCR-RFLP: polymerase chain reaction-restriction fragment length polymorphism
Supplementary Fig 1. Forest plot of ORs for the association between miR-146a polymorphism and colorectal cancer risk. rs2910164 was not associated with risk of colorectal cancer under the recessive model (OR=0.98, 95% CI: 0.77~1.25). Significant evidence of between-study heterogeneity was observed (P for heterogeneity =0.008, I2=68.2%). After stratification by MAF in controls, heterogeneities decreased in the subgroup of G>C (P for heterogeneity =0.246, I2=25.6%).
Supplementary Fig 2. Forest plot of ORs for the association between miR-146a polymorphism and colorectal cancer risk. rs2910164 was not associated with risk of colorectal cancer under the allele model (OR=1.02, 95% CI: 0.86~1.21) (A), the dominant model (OR= 1.05, 95% CI: 0.74~1.49) (B) or the additive model (ORGG vs. CC=1.05, 95% CI: 0.74~1.50 (C), ORGC vs. CC=1.03, 95% CI: 0.71~1.50 (D)). Significant evidence of between-study heterogeneity was observed (P for heterogeneity <0.1 or I250%). After stratification by MAF in controls, heterogeneities decreased in the subgroup of G>C.
Supplementary Fig. 3 Begg’s funnel plot under the recessive model (P=0.707). The logarithm values of relative risk were plotted against their standard errors. One circle dot represents one published study.
References
1. Thakkinstian A, McEvoy M, Minelli C, Gibson P, Hancox B, Duffy D et al. Systematic review and meta-analysis of the association between {beta}2-adrenoceptor polymorphisms and asthma: a HuGE review. Am J Epidemiol. 2005;162(3):201-11.
2. Gao LB, Pan XM, Li LJ, Liang WB, Zhu Y, Zhang LS et al. RAD51 135G/C polymorphism and breast cancer risk: a meta-analysis from 21 studies. Breast Cancer Res Treat. 2011;125(3):827-35.
3. Lau J, Ioannidis JP, Schmid CH. Quantitative synthesis in systematic reviews. Ann Intern Med. 1997;127(9):820-6.
4. Harbord RM, Egger M, Sterne JA. A modified test for small-study effects in meta-analyses of controlled trials with binary endpoints. Stat Med. 2006;25(20):3443-57.
5. Min KT, Kim JW, Jeon YJ, Jang MJ, Chong SY, Oh D et al. Association of the miR-146aC>G, 149C>T, 196a2C>T, and 499A>G polymorphisms with colorectal cancer in the Korean population. Mol Carcinog. 2012;51 Suppl 1:E65-73.
6. Hezova R, Kovarikova A, Bienertova-Vasku J, Sachlova M, Redova M, Vasku A et al. Evaluation of SNPs in miR-196-a2, miR-27a and miR-146a as risk factors of colorectal cancer. World J Gastroenterol. 2012;18(22):2827-31.
7. Lv M, Dong W, Li L, Zhang L, Su X, Wang L et al. Association between genetic variants in pre-miRNA and colorectal cancer risk in a Chinese population. J Cancer Res Clin Oncol. 2013;139(8):1405-10.
8. Ma L, Zhu L, Gu D, Chu H, Tong N, Chen J et al. A genetic variant in miR-146a modifies colorectal cancer susceptibility in a Chinese population. Arch Toxicol. 2013;87(5):825-33.
9. Chae YS, Kim JG, Lee SJ, Kang BW, Lee YJ, Park JY et al. A miR-146a polymorphism (rs2910164) predicts risk of and survival from colorectal cancer. Anticancer Res. 2013;33(8):3233-9.
10. Hu X, Li L, Shang M, Zhou J, Song X, Lu X et al. Association between microRNA genetic variants and susceptibility to colorectal cancer in Chinese population. Tumour Biol. 2013.