Association between polymorphisms in microRNAs and ischemic stroke in an Asian population: evidence based on 6,083 cases and 7,248 controls.

BACKGROUND: Polymorphisms in miR-146a (rs2910164), miR-196a2 (rs11614913), miR-149 (rs2292832) and miR-499 (rs3746444) have been associated with ischemic stroke (IS), but studies have given inconsistent results.
METHODS: This meta-analysis investigated the possible association between IS risk and the four polymorphisms. A total of 14 case-control studies from Asian populations involving 6,083 cases and 7,248 controls for the four polymorphisms were included.
RESULTS: Results showed that the GG genotype of miR-146a (rs2910164) may be associated with increased IS risk according to the recessive model (OR=1.20, 95% CI=1.02-1.42, P=0.03). Similarly, the CC genotype of miR-149 (rs2292832) may be associated with increased IS risk according to the recessive model (OR=1.28, 95% CI=1.08-1.52, P=0.005) and the homozygous model (OR=1.31, 95% CI=1.09-1.58, P=0.004). In contrast, miR-196a2 (rs11614913) and miR-499 (rs3746444) polymorphisms did not show significant association with IS risk in any of the five genetic models.
CONCLUSION: These results indicate that the GG genotype of miR-146a (rs2910164) and CC genotype of miR-149 (rs2292832) may confer increased susceptibility to IS, while miR-196a2 (rs11614913) and miR-499 (rs3746444) polymorphisms may not be associated with IS risk in Asian populations. These conclusions should be verified in large and well-designed studies.

Introduction

Stroke is a significant worldwide problem. An estimated 80% of the patients survive for at least 1 year after stroke, yet >70% have enduring disabilities.1,2 Ischemic stroke (IS) and intracerebral hemorrhage account for ~80%–85% and 15%–20% of all stroke cases, respectively.3 IS is a complex syndrome whose pathological development involves multiple components, which include environmental and genetic factors.4 Established environmental risk factors include age, sex, body mass index, hypertension, diabetes mellitus, smoking, and hyperlipidemia. However, recent studies suggested that genetics may contribute more than environment to IS, considering that a number of single-gene disorders are related to IS.5–8 Nevertheless, the factors defining genetic susceptibility to IS remain unclear.

MicroRNAs (miRNAs) represent a group of short non-coding RNA molecules, 18–25 nucleotides in length. Bioinformatics data indicate that a single miRNA can bind to as many as 200 gene targets, and miRNAs may regulate the expression of approximately one-third of protein-coding mRNAs. A single-nucleotide polymorphism (SNP) in miRNA may create a mismatch, leading to gene expression disorder and diseases.9 Evidence has indicated that miRNAs regulate various IS-related biological processes, such as atherosclerosis, hypertension, and plaque rupture.10 In fact, altered miRNA expression has been observed in IS in preclinical animal models and patients, suggesting a potential role in predicting the diagnosis and prognosis of IS.11,12

More specifically, the literature suggests an association between IS and polymorphisms in miR-146a (rs2910164), miR-196a2 (rs11614913), miR-149 (rs2292832), and miR-499 (rs3746444).13–26 However, these associations are controversial because individual studies relied on relatively small samples. Therefore, to obtain a more comprehensive understanding of the available evidence, we conducted this meta-analysis of 14 case–control studies to evaluate the possible association between IS risk and miR-146a (rs2910164), miR-196a2 (rs11614913), miR-149 (rs2292832), and miR-499 (rs3746444) in Asian populations.

Materials and methods

Search strategy

All clinical and experimental case–control studies of miRNA polymorphisms and IS risk published through February 1, 2018 were identified through systematic searches in PubMed, EMBASE, Google Scholar, and the Chinese National Knowledge Infrastructure (CNKI) databases using English and Chinese. The search terms used were as follows: microRNA; miRNA; these two terms in combination with polymorphism, polymorphisms, SNP, variant, variants, variation, genotype, genetic, or mutation; and all the above-mentioned terms in combination with stroke or ischemic stroke. Reference lists in identified articles and reviews were also searched manually to identify additional eligible studies.

Inclusion criteria

To be included in our review and meta-analysis, studies had to 1) have a case–control design for assessing the association of IS risk with miR-146a (rs2910164), miR-196a2 (rs11614913), miR-149 (rs2292832), and miR-499 (rs3746444); 2) be accessible as a full-text article and report sufficient data for estimating ORs with 95% CIs; 3) report genotype frequencies; and 4) involve humans rather than animal models.

Data extraction

Two authors (DHZ and CBL) independently extracted the following data from the included studies: first author’s family name, year of publication, ethnicity, testing methods, control source, age, sex, P-value for Hardy–Weinberg equilibrium (HWE) in controls, numbers and genotypes of cases and controls, and frequencies of genotypes in cases and controls. Discrepancies were resolved by consensus. Only those studies that met the predetermined inclusion criteria were included.

Assessment of methodological quality

To assess the quality of the studies included in this analysis, the Newcastle–Ottawa scale was used by two independent assessors (JRW and LC).27 For the Newcastle–Ottawa scale, a full score is nine stars; a score range of 5–9 stars is considered to indicate generally high methodological quality, whereas a range of 0–4 stars is considered to indicate poor quality.28 The quality of all the included studies is summarized in Table 1. Any disagreements about Newcastle–Ottawa scores were resolved by other authors following a comprehensive reassessment. Only high-quality studies were included in our meta-analysis.

Table 1

Methodological quality of the studies included in the final analysis based on the Newcastle–Ottawa scale for assessing the quality of case–control studies

Study	Selection (score)				Comparability (score)	Exposure (score)			Total scoreb
	Adequate definition of patient cases	Representativeness of patient cases	Selection of controls	Definition of controls	Control for important factor or additional factor	Ascertainment of exposure (blinding)	Same method of ascertainment for participants	Non-response ratea
Sun13	1	1	0	1	2	0	1	1	7
Li14	1	1	0	1	0	0	1	1	5
He and Han15	1	1	0	1	2	0	1	1	7
Jeon et al16	1	1	0	1	2	0	1	1	7
Hu et al17	1	1	0	1	2	0	1	1	7
Liu et al18	1	1	0	1	1	0	1	1	6
Zhu et al19	1	1	0	1	2	0	1	1	7
Huang et al20	1	1	0	1	2	0	1	1	7
Zhong et al21	1	1	0	1	2	0	1	1	7
Qu et al22	1	1	0	1	0	0	1	1	5
Lyu et al23	1	1	0	1	2	0	1	1	7
Zhu24	1	1	0	1	2	0	1	1	7
Luo et al25	1	1	0	1	2	0	1	1	7
Zhu et al26	1	1	0	1	2	0	1	0	6

Notes:

When there was no significant difference in the response rate between both groups based on a chi-squared test (P>0.05), one point was awarded.

Total score was calculated by adding up the points awarded in each item.

Statistical analyses

The unadjusted OR with 95% CI was used to assess the strength of the association of IS risk with miR-146a (rs2910164), miR-196a2 (rs11614913), miR-149 (rs2292832), and miR-499 (rs3746444) based on genotype frequencies in cases and controls. The significance of pooled ORs was determined using the Z-test, with P<0.05 defined as the significance threshold. Meta-analysis was conducted using a fixed-effect model when P>0.10 for the Q-test, indicating the lack of heterogeneity among studies; otherwise, a random-effect model was used. All these statistical tests were performed using Review Manager 5.2 (Cochrane Collaboration, Oxford, England).

Publication bias was assessed using Begg’s funnel plot and Egger’s weighted regression, with P<0.05 considered statistically significant. Begg’s funnel plots and Egger’s weighted regression were calculated using Stata 12.0 (StataCorp LP, College Station, TX, USA).

Results

Description of studies

Figure 1 is a flow diagram illustrating the process of searching for and selecting studies. A total of 184 potentially relevant publications up to February 1, 2018 were systematically identified through searches of the PubMed, EMBASE, Google Scholar, and CNKI databases in English and Chinese. Of these, we excluded 161 studies during initial screening based on review of the titles and abstracts. During analysis of the full text of the remaining articles, two studies were excluded for not being case–control studies, three studies were excluded because they did not report precise genotypes, and two articles were excluded because they investigated polymorphisms of miRNAs other than miR-146a (rs2910164), miR-196a2 (rs11614913), miR-149 (rs2292832), or miR-499 (rs3746444). A further two studies were excluded because they were not written in English or Chinese.

Figure 1

Flowchart of study selection.

In the end, 14 studies13–26 were included in this meta-analysis based on our search strategy and inclusion criteria. Their characteristics are summarized in Table 2. Of these, 13 studies13,14,16–26 (Table 3) involving 5,726 cases and 7,175 controls evaluated the association between miR-146a (rs2910164) polymorphism and IS risk. Seven studies16,18–20, 24–26 (Table 3) involving 3,090 cases and 3,047 controls evaluated the association between miR-196a2 (rs11614913) polymorphism and IS risk. Six studies15–17,24–26 (Table 3) involving 2,448 cases and 2,322 controls evaluated miR-149 (rs2292832) polymorphism and IS risk. The remaining seven studies16,18,20,23–26 (Table 3) involving 3,082 cases and 3,044 controls evaluated miR-499 (rs3746444) polymorphism and IS risk. The distribution of genotypes in controls was consistent with HWE (P>0.05) in all but three studies.14,20,22 The overall quality of the included studies was adequate, and the mean Newcastle–Ottawa score for the included studies was 6.57 (Table 1).

Table 2

Characteristics of the studies included in the meta-analysis

Study	Year	Ethnicity	Country	Testing method	Control source	Age (years, mean ±SD)		Male, n (%)		SNP
						Cases	Controls	Cases	Controls
Sun13	2011	Asian	China	PCR-RFLP	Hospital-based healthy volunteers	63±12	62±13	236 (61.9)	347 (53.4)	miR-146a
Li14	2010	Asian	China	PCR-RFLP	Hospital-based healthy volunteers	64±11	45±12	188 (67.2)	579 (57.3)	miR-146a
He and Han15	2013	Asian	China	PCR-RFLP	Hospital-based healthy volunteers	65.7±11.5	66.3±10.2	205 (55.0)	193 (51.7)	miR-149
Jeon et al16	2013	Asian	South Korea	TaqMan	Hospital-based healthy volunteers	64.16±11.90	63.14±10.19	336 (49.6)	244 (44.1)	miR-146a miR-149 (rs2292832); and miR-499 (rs3746444)
Hu et al17	2014	Asian	China	PCR-RFLP	Hospital-based healthy volunteers	64±11.7	63±10.5	94 (48.0)	95 (46.3)	miR-146a (rs2910164) and miR-149 (rs2292832)
Liu et al18	2014	Asian	China	PCR-RFLP	Hospital-based healthy volunteers	67.52±10.29	66.34±11.07	227 (58.06)	180 (60.81)	miR-146a (rs2910164); miR-196a2 (rs11614913); and miR-499 (rs3746444)
Huang et al20	2015	Asian	China	TaqMan	Hospital-based healthy volunteers	63 (54–70)a	61 (54–68)a	327 (61.6)	327 (61.6)	miR-146a (rs2910164); miR-196a2 (rs11614913); and miR-499 (rs3746444)
Zhong et al21	2016	Asian	China	PCR	Hospital-based healthy volunteers	62.6±8.63	61.1±9.58	177 (59.6)	170 (56.7)	miR-146a (rs2910164)
Qu et al22	2016	Asian	China	PCR-LDR	Hospital-based healthy volunteers	61.30±9.40	59.50±8.50	718 (63.0)	903 (57.0)	miR-146a (rs2910164)
Lyu et al23	2016	Asian	China	TaqMan	Hospital-based healthy volunteers	58±11.9	58±11.9	210 (55.6)	210 (55.6)	miR-146a (rs2910164) and miR-499 (rs3746444)
Zhu24	2016	Asian	China	PCR-RFLP	Hospital-based healthy volunteers	63.74±4.49	63.31±4.84	215 (54.3)	202 (53.4)	miR-146a (rs2910164); miR-196a2 (rs11614913); miR-149 (rs2292832); and miR-499 (rs3746444)
Luo et al25	2017	Asian	China	PCR	Hospital-based healthy volunteers	67.70±12.33	60.17±10.32	196 (65.8)	181 (59.8)	miR-146a (rs2910164); miR-196a2 (rs11614913); miR-149 (rs2292832); and miR-499 (rs3746444)
Zhu et al26	2017	Asian	China	TaqMan	Hospital-based healthy volunteers	61.0±10.2	59.7±9.9	321 (62.9)	311 (59.4)	miR-146a (rs2910164); miR-196a2 (rs11614913); miR-149 (rs2292832); and miR-499 (rs3746444)

Note:

These data are expressed as median (25th, 75th quartiles).

Abbreviations: LDR, ligase detection reaction; PCR, polymerase chain reaction; RFLP, restriction fragment length polymorphism; SNP, single-nucleotide polymorphism.

Table 3

Genotype distributions of miR-146a (rs2910164), miR-196a2 (rs11614913), miR-149 (rs2292832), and miR-499 (rs3746444)

Study	Year	P-value for HWE	Sample size (cases/controls)	No of cases			Allele frequencies of cases, n (%)		No of controls			Allele frequencies of controls, n (%)
miR-146a (rs2910164)				CC	GC	GG	C	G	CC	GC	GG	C	G
Sun13	2011	0.345	358/650	136	161	61	433 (60.5)	283 (39.5)	228	304	118	760 (58.5)	540 (41.5)
Li14	2010	0.009	268/1,010	79	110	79	268 (50.0)	268 (50.0)	345	455	210	1,145 (56.7)	875 (43.3)
Jeon et al16	2013	0.589	678/553	223	327	128	773 (57.0)	583 (43.0)	211	266	76	688 (62.2)	418 (37.8)
Hu et al17	2014	0.193	196/205	75	87	34	237 (60.5)	155 (39.5)	97	82	26	276 (67.3)	134 (32.7)
Liu et al18	2014	0.650	296/391	85	159	52	329 (55.6)	263 (44.4)	116	198	77	430 (55.0)	352 (45.0)
Zhu et al19	2014	0.952	368/381	145	173	50	463 (63.0)	273 (37.0)	132	185	64	449 (80.6)	313 (19.4)
Huang et al20	2015	0.106	531/531	189	261	81	639 (60.2)	423 (39.8)	219	257	55	695 (65.4)	367 (34.6)
Zhong et al21	2016	0.133	297/300	141	128	28	410 (69.0)	184 (31.0)	113	152	35	378 (63.0)	222 (37.0)
Qu et al22	2016	<0.001	1,139/1,585	355	618	166	1,328 (58.3)	950 (41.7)	483	869	233	1,835 (57.9)	1,335 (42.1)
Lyu et al23	2016	0.079	378/378	119	198	61	436 (57.7)	320 (42.3)	153	187	38	493 (65.2)	263 (34.8)
Zhu24	2016	0.521	396/378	131	194	71	456 (57.6)	336 (42.4)	154	179	45	487 (64.4)	269 (35.6)
Luo et al25	2017	0.672	298/303	129	130	39	388 (65.1)	208 (34.9)	119	139	45	377 (62.2)	229 (37.8)
Zhu et al26	2017	0.085	523/510	170	267	86	607 (58.0)	439 (42.0)	204	251	55	659 (64.6)	361 (35.4)
miR-196a2 (rs11614913)				TT	TC	CC	T	C	TT	TC	CC	T	C
Jeon et al16	2013	0.126	678/553	139	352	187	630 (46.5)	726 (53.5)	105	292	156	502 (45.4)	604 (54.6)
Liu et al18	2014	0.060	296/391	51	181	64	283 (47.8)	309 (52.2)	84	214	93	382 (48.8)	400 (51.2)
Zhu et al19	2014	0.384	368/381	71	189	108	331 (45.0)	405 (55.0)	78	198	105	354 (46.5)	408 (53.5)
Huang et al20	2015	0.856	531/531	100	265	166	465 (43.8)	597 (56.2)	112	266	153	490 (46.1)	572 (53.9)
Zhu24	2016	0.354	396/378	112	205	79	429 (54.2)	363 (45.8)	110	196	72	416 (55.0)	340 (45.0)
Luo et al25	2017	0.385	298/303	73	138	87	284 (47.7)	312 (52.3)	75	159	69	309 (51.0)	297 (49.0)
Zhu et al26	2017	0.548	523/510	150	273	100	573 (54.8)	473 (45.2)	146	260	104	552 (54.1)	468 (45.9)
miR-149 (rs2292832)				TT	TC	CC	T	C	TT	TC	CC	T	C
He and Han15	2013	0.303	357/373	138	162	57	438 (66.6)	276 (41.4)	160	175	38	495 (66.4)	251 (33.6)
Jeon et al16	2013	0.921	678/553	299	303	76	901 (66.4)	455 (33.6)	262	238	53	762 (68.9)	344 (31.1)
Hu et al17	2014	0.199	196/205	79	76	41	234 (59.7)	158 (40.3)	80	89	36	249 (60.7)	161 (39.3)
Zhu24	2016	0.720	396/378	165	179	52	509 (64.3)	283 (35.7)	190	158	30	538 (71.2)	218 (28.8)
Luo et al25	2017	0.447	298/303	131	127	40	389 (65.3)	207 (34.7)	121	136	46	378 (62.4)	228 (37.6)
Zhu et al26	2017	0.351	523/510	232	221	70	685 (65.5)	361 (34.5)	240	213	57	693 (67.9)	327 (32.1)
miR-499 (rs3746444)				AA	AG	GG	A	G	AA	AG	GG	A	G
Jeon et al16	2013	0.740	678/553	460	195	23	1,115 (82.2)	241 (17.8)	365	170	18	900 (81.4)	206 (18.6)
Liu et al18	2014	0.170	296/391	181	96	19	458 (77.4)	134 (22.6)	278	99	14	655 (83.8)	127 (16.2)
Huang et al20	2015	0.002	531/531	398	133	0	929 (87.5)	133 (12.5)	403	128	0	934 (87.9)	128 (12.1)
Lyu et al23	2016	0.621	378/378	257	110	11	624 (82.5)	132 (17.5)	250	113	15	613 (81.1)	143 (18.9)
Zhu24	2016	0.910	396/378	255	123	18	633 (79.9)	159 (20.1)	249	116	13	614 (81.2)	142 (18.8)
Luo et al25	2017	0.131	298/303	215	78	5	508 (85.2)	88 (14.8)	244	53	6	541 (89.3)	65 (10.7)
Zhu et al26	2017	0.380	505/510	349	124	32	840 (80.3)	206 (19.7)	328	158	24	814 (79.8)	206 (20.2)

Abbreviation: HWE, Hardy–Weinberg equilibrium.

Quantitative data synthesis

IS risk and miR-146a (rs2910164) polymorphism

The overall results for miR-146a (rs2910164) are summarized in Table 4 and Figure 2. On the basis of 5,726 cases and 7,175 controls from 13 studies,13,14,16–26 the overall results indicated that the GG genotype of miR-146a (rs2910164) may be associated with increased IS risk according to the recessive model (OR=1.20, 95% CI=1.02–1.42, P=0.03; Figure 2B).

Table 4

Overall meta-analysis of the association between ischemic stroke and polymorphisms in miR-146a (rs2910164), miR-196a2 (rs11614913), miR-149 (rs2292832), and miR-499 (rs3746444)

Genetic model	OR [95% CI]	Z (P-value)	Heterogeneity of study design			Analysis model
			χ2	df (P-value)	I2 (%)
miR-146a (rs2910164) from 13 case–control studies (5,726 cases and 7,175 controls)
Allelic model (G-allele vs C-allele)	1.10 [0.99–1.22]	1.74 (0.08)	47.91	12 (<0.001)	75	Random
Recessive model (GG vs GC+CC)	1.20 [1.02–1.42]	2.16 (0.03)	31.55	12 (0.002)	62	Random
Dominant model (CC vs GC+GG)	0.91 [0.80–1.04]	1.41 (0.16)	34.76	12 (<0.001)	65	Random
Homozygous model (GG vs CC)	1.24 [1.00–1.53]	1.95 (0.05)	43.43	12 (<0.001)	72	Random
Heterozygous model (GC vs CC)	1.06 [0.95–1.17]	1.00 (0.32)	20.79	12 (0.05)	42	Random
miR-196a2 (rs11614913) from 7 case–control studies (3,090 cases and 3,047 controls)
Allelic model (C-allele vs T-allele)	1.04 [0.97–1.12]	1.10 (0.27)	3.20	6 (0.78)	0	Fixed
Recessive model (CC vs TC+TT)	1.04 [0.93–1.17]	0.73 (0.46)	4.60	6 (0.60)	0	Fixed
Dominant model (TT vs TC+CC)	0.95 [0.85–1.08]	0.77 (0.44)	2.86	6 (0.83)	0	Fixed
Homozygous model (CC vs TT)	1.07 [0.92–1.24]	0.91 (0.36)	2.85	6 (0.83)	0	Fixed
Heterozygous model (TC vs TT)	1.07 [0.93–1.23]	0.90 (0.37)	2.72	5 (0.74)	0	Fixed
miR-149 (rs2292832) from 6 case–control studies (2,448 cases and 2,322 controls)
Allelic model (C-allele vs T-allele)	1.09 [1.00–1.18]	1.91 (0.06)	4.84	5 (0.44)	0	Fixed
Recessive model (CC vs TC+TT)	1.28 [1.08–1.52]	2.80 (0.005)	6.14	5 (0.29)	19	Fixed
Dominant model (TT vs TC+CC)	0.89 [0.79–1.00]	1.99 (0.05)	6.31	5 (0.28)	21	Fixed
Homozygous model (CC vs TT)	1,31 [1.09–1.58]	2.92 (0.004)	8.27	5 (0.14)	40	Fixed
Heterozygous model (TC vs TT)	1.07 [0.95–1.21]	1.12 (0.26)	4.22	5 (0.52)	0	Fixed
miR-499 (rs3746444) from 7 case–control studies (3,082 cases and 3,044 controls)
Allelic model (G-allele vs A-allele)	1.09 [0.95–1.25]	1.28 (0.20)	12.36	6 (0.05)	51	Random
Recessive model (GG vs AG+AA)	1.21 [0.91–1.61]	1.31 (0.19)	3.81	5 (0.58)	0	Fixed
Dominant model (AA vs AG+GG)	0.93 [0.78–1.12]	0.77 (0.44)	16.43	6 (0.01)	63	Random
Homozygous model (GG vs AA)	1.20 [0.90–1.60]	1.25 (0.21)	4.47	5 (0.48)	0	Fixed
Heterozygous model (AG vs AA)	1.06 [0.87–1.28]	0.56 (0.57)	17.10	6 (0.009)	65	Random

Figure 2

Forest plot describing the association between the miR-146a (rs2910164) polymorphism and ischemic stroke risk according to different genetic models: (A) allelic (G-allele vs C-allele), (B) recessive (GG vs GC+CC), (C) dominant (CC vs GC+GG), (D) homozygous (GG vs CC), and (E) heterozygous (GC vs CC).

IS risk and miR-196a2 (rs11614913) polymorphism

The overall results are summarized in Table 4 and Figure 3. On the basis of 3,090 cases and 3,047 controls from seven studies,16,18–20,24–26 miR-196a2 (rs11614913) polymorphism did not show significant association with IS risk in any of the following five genetic models: allelic model, OR=1.04, 95% CI=0.97–1.12, P=0.27 (Figure 3A); recessive model, OR=1.04, 95% CI=0.93–1.17, P=0.46 (Figure 3B); dominant model, OR=0.95, 95% CI=0.85–1.08, P=0.44 (Figure 3C); homozygous model, OR=0.95, 95% CI=0.85–1.08, P=0.44 (Figure 3D); and heterozygous model, OR=1.07, 95% CI=0.93–1.23, P=0.37 (Figure 3E).

Figure 3

Forest plot describing the association between the miR-196a2 (rs11614913) polymorphism and ischemic stroke risk according to different genetic models: (A) allelic, (B) recessive, (C) dominant, (D) homozygous, and (E) heterozygous.

IS risk and miR-149 (rs2292832) polymorphism

The overall results for miR-149 (rs2292832) are summarized in Table 4 and Figure 4. On the basis of 2,448 cases and 2,322 controls from six studies,16,18,20,23–26 the overall results indicated that the CC genotype of miR-149 (rs2292832) may be associated with increased IS risk according to the recessive model (OR=1.28, 95% CI=1.08–1.52, P=0.005; Figure 4B) and homozygous model (OR=1.31, 95% CI=1.09–1.58, P=0.004; Figure 4D).

Figure 4

Forest plot describing the association between the miR-149 (rs2292832) polymorphism and ischemic stroke risk according to different genetic models: (A) allelic, (B) recessive, (C) dominant, (D) homozygous, and (E) heterozygous.

IS risk and miR-499 (rs3746444) polymorphism

The overall results are summarized in Table 4 and Figure 5. On the basis of 3,082 cases and 3,044 controls from seven studies,16,18,20,23–26 miR-499 (rs3746444) polymorphism did not show significant association with IS risk in any of the following five genetic models: allelic model, OR=1.09, 95% CI=0.95–1.25, P=0.20 (Figure 5A); recessive model, OR=1.21, 95% CI=0.91–1.61, P=0.19 (Figure 5B); dominant model, OR=0.93, 95% CI=0.78–1.12, P=0.44 (Figure 5C); homozygous model, OR=1.20, 95% CI=0.90–1.60, P=0.21 (Figure 5D); or heterozygous model, OR=1.06, 95% CI=0.87–1.28, P=0.57 (Figure 5E).

Figure 5

Forest plot describing the association between the miR-149 (rs2292832) polymorphism and ischemic stroke risk according to different genetic models: (A) allelic, (B) recessive, (C) dominant, (D) homozygous, and (E) heterozygous.

Sensitivity analysis

Sensitivity analysis was conducted for miR-146a (rs2910164) by excluding the studies by Li et al14 and Qu et al;22 the P-value for HWE was less than 0.05 for these two studies. The recessive model gave different results (OR=1.19, 95% CI=0.98–1.45, P=0.07) than those obtained when all studies were meta-analyzed. Sensitivity analysis was conducted for miR-146a (rs2910164) by excluding one study by Jeon et al.16 Again, the recessive model gave different results (OR=1.18, 95% CI=0.99–1.41, P=0.07) than when all studies were included. Therefore, the results for miR-146a (rs2910164) should be interpreted with caution.

Sensitivity analysis was conducted for miR-196a2 (rs11614913) by excluding the study by Jeon et al.16 The results were similar to those obtained with all studies, regardless of the genetic model. This implies that our meta-analysis results for miR-196a2 (rs11614913) are robust. Similar robustness was observed when we performed sensitivity analysis for miR-149 (rs2292832) and for miR-499 (rs3746444) by excluding the study by Jeon et al.16

Sensitivity analysis was conducted for miR-499 (rs3746444) by excluding a study by Huang et al,20 in which the P-value of HWE was less than 0.05. The results were not altered in any of the five genetic models.

Publication bias

Begg’s funnel plot and Egger’s test were performed to detect potential publication bias in this meta-analysis. No obvious asymmetry was observed in Begg’s funnel plots in the recessive model, and Egger’s tests (Figure 6) indicated no publication bias.

Figure 6

Begg’s funnel plot and Egger’s test to assess publication bias in the meta-analysis of potential associations between ischemic stroke risk and (A and B) miR-146a (rs2910164), (C and D) miR-196a2 (rs11614913), (E and F) miR-149 (rs2292832), and (G and H) miR-499 (rs3746444).

Note: All analyses were performed using a recessive genetic model.

Discussion

Previous studies have demonstrated that mutations in the pre-miRNA of miR-146a, miR-499, miR-149, and miR-196a2 decrease the levels of the corresponding mature miRNAs.20,29,30 These four miRNAs affect thrombosis or inflammation pathways in the circulatory system by regulating tumor necrosis factor-α (TNF-α),31 methylenetetrahydrofolate reductase,32 annexin A1,33 C-reactive protein,34 the NF-κB pathway, and the MAP kinase pathway.35 Many studies have been conducted to reveal the impact of SNPs on precursor and mature miRNAs and their associations with IS risk.13–26 In fact, several meta-analyses have been conducted to explore the association between miRNA polymorphisms and IS risk. The results have been inconsistent, largely due to limited sample size.36–39 Therefore, we conducted this meta-analysis on all eligible studies to provide a more precise estimate of the association of IS risk with miR-146a (rs2910164), miR-196a2 (rs11614913), miR-149 (rs2292832), and miR-499 (rs3746444). Interestingly, all the case–control studies in our meta-analysis analyzed Asian populations.

A previous meta-analysis by Zhu et al39 found the C allele of miR-146a (rs2910164) to be associated with lower IS risk, but this trend was observed only in Koreans according to the allelic model. Our meta-analysis, in contrast, suggests that this C allele is not significantly associated with IS risk; instead, we found the GG genotype of miR-146a (rs2910164) to be associated with increased risk. Our result may be more reliable than that of the previous meta-analysis by Zhu et al39 because our meta-analysis contained nine more case–control studies14,15,17,21–26 with larger samples. Our subgroup analysis suggesting a significant relationship between the C allele of miR-146a (rs2910164) and lower IS risk contained only one case–control study, which was by Jeon et al.16

While the meta-analysis by Zhu et al39 reported an association between the A allele of miR-499 (rs3746444) and decreased IS risk in Chinese, our meta-analysis did not detect this association, either across Asian populations or specifically in the Chinese population (data not shown). Our result may be more reliable because our meta-analysis included four more case–control studies23–26 than the one by Zhu et al.39 The results of our meta-analysis are consistent with those reported in the meta-analysis by Xiao et al.37

Our meta-analysis suggests a significant association between the CC genotype of miR-149 (rs2292832) and increased IS risk. In contrast, the meta-analysis of Xiao et al37 based on two case–control studies indicated that the TT genotype and T allele of miR-149 (rs2292832) are associated with significantly lower IS risk, whereas another meta-analysis36 based on three case–control studies found the CC genotype and C allele of miR-149 (rs2292832) to be significantly associated with IS risk. Our meta-analysis contained three more case–control studies24–26 than either of these other meta-analyses, which may make it more reliable.

Our meta-analysis did not find a significant association between miR-196a2 (rs11614913) polymorphism and IS risk. This result confirms other meta-analyses37–39 based on smaller samples.

To the best of our knowledge, the current meta-analysis involves the largest sample (6,083 cases and 7,248 controls) than previous studies36–39 investigating the possible association of IS risk with miR-146a (rs2910164), miR-196a2 (rs11614913), miR-149 (rs2292832), and miR-499 (rs3746444) in Asian populations. Nevertheless, the meta-analysis is limited by the designs of the included studies. First, the P-value for HWE was less than 0.05 in two studies14,22 on miR-146a (rs2910164) and one study26 on miR-499 (rs3746444). These results suggested that these study populations may not be representative of the broader target population. Second, the results may be affected by both genetic and environmental factors, but most studies did not report environmental exposure, making it impossible to include them in the meta-analysis. Third, our exclusion of unpublished data and of papers published in languages other than English and Chinese may have biased our results. Fourth, the studies may be subject to performance bias, attrition bias, and reporting bias, although Newcastle–Ottawa scores were at least 5 for all 14 studies, indicating high quality. Fifth, stroke is a heterogeneous disease and has different subtypes that may affect the results of genetic association studies, but most case–control studies in our meta-analysis appeared not to use a well-phenotyped population. This may make the results less accurate. Finally, all the patients in this meta-analysis were Asian and this may limit the relevance of the results to other populations. Thus, more large and well-designed studies are warranted in non-Asian populations.

Conclusion

This meta-analysis suggests that the GG genotype of miR-146a (rs2910164) and the CC genotype of miR-149 (rs2292832) may confer increased susceptibility to IS in Asian populations, whereas polymorphism in miR-196a2 (rs11614913) and miR-499 (rs3746444) may not be associated with IS risk. These conclusions should be verified in large and well-designed studies.