Purpose: The association between GRIA1 rs548294 G>A and rs2195450 C>T polymorphisms and migraine risk has been reported in several case–control studies. However, the results of studies are inconsistent. Thus, we conducted a meta-analysis to more precisely estimate the association of the two polymorphisms with migraine risk.

Methods: Eligible studies were retrieved and screened from the online databases (EMBASE, PubMed, Web of Science, Wanfang, and Chinese National Knowledge Infrastructure). The pooled odds ratio (OR) with corresponding 95.0% confidence intervals (CIs) was assessed using random- or fixed-effects model.

Results: A total of 1233 cases and 1374 controls from four eligible studies were included. The pooled analysis showed that GRIA1 rs548294 G>A polymorphism was not significantly associated with migraine risk. GRIA1 rs2195450 C>T polymorphism was significantly associated with migraine risk under heterozygous model (CT vs. CC, OR = 1.23, 95%CI = 1.02–1.48, PZ = 0.03). Further subgroup analysis based on ethnicity showed a significant association of GRIA1 rs2195450 C>T polymorphism with migraine risk in Asian population, but not in Caucasian population.

Conclusions: Our results indicates that GRIA1 rs2195450 C>T polymorphism is significantly associated with migraine risk. However, the number of studies included in the meta-analysis was small. Thus, more high quality case–control studies with a large sample size are still required to confirm these findings.

Migraine is a common neurovascular disorder affecting over 17% of women and 5–8% of men around the world [1]. This disorder is characterized by recurrent moderate-to-severe headaches, and often accompanied by nausea, vomiting, photo- and phonophobia [2]. Clinically, migraine is divided into two main subtypes: migraine with aura (MA) and without aura (MO). MA is an idiopathic, recurrent disorder related to neurologic symptoms that are localized to the cerebral cortex or brain stem. MO is an idiopathic, recurring headache disorder typified by painful attacks lasting 4–72 h [3]. Although the pathogenesis of migraine is still unclear, studies suggests that inheritance factor plays an important role in the occurrence of this disease [4–6].

GRIA1 gene is located on chromosome 5q33.2 and belongs to a family of α-amino-3-hydroxy-5-methyl-4-isoxazole propionate (AMPA) receptors. This gene encodes a subunit of ionotropic glutamate receptor regulating neuronal excitability. Synaptic hyperexcitability has been recognized as one major factor causing migraine [7]. Glutamate is an excitatory neurotransmitter in the mammalian central nervous system and can activate the trigeminovascular system and increase the risk of cortical spreading depression, which plays an important role in the pathophysiology of migraine [8–10]. Therefore, GRIA1 gene was inferred as a migraine-related gene. Single-nucleotide polymorphism (SNP) is one of the most common forms of human DNA variations. The potentially functional SNPs in migraine-related genes have been reported to be associated with migraine risk [4–6]. Thus GRIA1 polymorphisms may also be involved in migraine risk. In order to confirm this hypothesis, some case–control studies were conducted to investigate the association of GRIA1 polymorphisms (rs548294 G>A, rs2195450 C>T) with migraine risk. However, the results were inconsistent [7,11–13]. To get a more accurate result, we collected all eligible studies and conducted the present meta-analysis.

Search strategy and selection criteria

The online databases including EMBASE, PubMed, Web of Science, Wanfang, as well as Chinese National Knowledge Infrastructure (CNKI) were searched in order to obtain all the relevant articles up to August 3, 2018. Keywords for search were: ‘glutamate ionotropic receptor AMPA type subunit 1 or GRIA1’, ‘polymorphism or SNP or variation’, and ‘migraine or headache’. In addition, reference lists in retrieved articles were also screened. All included studies were conducted on human subjects, and published in either English or Chinese.

All studies selected for further meta-analysis should meet the following criteria: (i) case–control studies investigating the association between GRIA1 polymorphisms (rs548294 G>A, rs2195450 C>T) and migraine risk; (ii) available data of genotype and allele distributions in cases and controls.

Data extraction and quality assessment

According to the selection criteria, two authors (Gao, X.R. and Wang, J.G.) carefully reviewed all included articles and independently extracted the relevant information, including the first author’s name, year of publication, country of origin, genotyping method, the total number of case and control, and the genotype and allele frequency distribution in case and control, and P value of HWE in the controls. In addition, quality assessment of each included study was conducted using the Newcastle–Ottawa Scale (NOS), which included three assessment categories (selection, comparability, and exposure) [14]. The categories for selection, comparability and exposure contained four, one and three items, respectively. In the selection and exposure categories, each eligible item was endowed with a star. In the comparability category, eligible item was endowed with at most two stars. Finally, any disagreements between two authors were solved by discussion.

Statistical analysis

For each study, the odds ratio (OR) and corresponding 95% confidence intervals (CIs) were calculated to assess the migraine risk associated with GRIA1 rs548294 G>A and rs2195450 C>T polymorphisms. The pooled ORs were calculated under the following genetic models: heterozygous model, recessive model, homozygous model, dominant model, and allelic contrast model. The significance of the pooled OR was examined by the Z-test, and PZ <  0.05 was considered statistically significant. Heterogeneity between studies was estimated by chi-square-based Q-test, and PH < 0.1 was considered statistically significant. The random-effects model was adopted to analyze the pooled ORs if PH value was less than 0.1. Otherwise, a fixed-effects model was exerted. Sensitivity analysis was used to assess the consistence and stability of the present meta-analysis. Additionally, funnel plots were used to estimate publication bias. All the statistical analyses were implemented with the Review Manager software (Version 5.2; Cochrane Collaboration, London, U.K.).

Characteristics of included studies

After an initial search, 12 potential articles were identified for further analysis. Eight articles unrelated to the present study were excluded after reviewing titles and abstracts. Finally, four eligible articles were included in the meta-analysis (Figure 1). Tables 1 and 2 presented the characteristics and quality of the included studies, respectively. Of the 4 studies with 1233 cases and 1374 controls, 1 was from Asian population, 3 were from Caucasian population.

Flow chart of literature search

Figure 1
Flow chart of literature search
Figure 1
Flow chart of literature search
Table 1
The main characteristics of the included studies in the meta-analysis
First authorYearCountryEthnicityGenotyping methodCasesControls
Jie Fang 2016 China Asian Multiplex SNaPshot 331 330 
Sarah Cargnin 2014 Italy Caucasian TaqMan 186 312 
Bridget H. Maher 2013 Australia Caucasian PCR-RFLP 472 472 
Daniela Formicola 2010 Italy Caucasian PCR-RFLP 244 260 
First authorYearCountryEthnicityGenotyping methodCasesControls
Jie Fang 2016 China Asian Multiplex SNaPshot 331 330 
Sarah Cargnin 2014 Italy Caucasian TaqMan 186 312 
Bridget H. Maher 2013 Australia Caucasian PCR-RFLP 472 472 
Daniela Formicola 2010 Italy Caucasian PCR-RFLP 244 260 

Abbreviation: PCR-RFLP, polymerase chain reaction-restriction fragment length polymorphism.

Table 2
The quality assessment of all included studies based on the Newcastle–Ottawa Scale
CategoriesItemsFang’s studyCargnin’s studyMaher’s studyFormicola’s study
Selection Adequacy of case definition 
 Representativeness of the cases 
 Selection of controls 
 Definition of controls 
Comparability Comparability of cases/controls 
Exposure Ascertainment of exposure 
 Same method of ascertainment for cases and controls 
 Non-response rate 
CategoriesItemsFang’s studyCargnin’s studyMaher’s studyFormicola’s study
Selection Adequacy of case definition 
 Representativeness of the cases 
 Selection of controls 
 Definition of controls 
Comparability Comparability of cases/controls 
Exposure Ascertainment of exposure 
 Same method of ascertainment for cases and controls 
 Non-response rate 

Meta-analysis results

As presented in Figure 2 and Table 3, there was no significant association between GRIA1 rs548294 G>A polymorphism and migraine risk under all genetic models. Further stratified analysis based on ethnicity also showed no statistical association between GRIA1 rs548294 G>A polymorphism and migraine risk under all genetic models.

Forest plots of the association of GRIA1 rs548294 G>A polymorphism and migraine risk (A: Homozygous model; B: Dominant model; C: Recessive model; D: Heterozygous model; E: Allelic contrast model)

Figure 2
Forest plots of the association of GRIA1 rs548294 G>A polymorphism and migraine risk (A: Homozygous model; B: Dominant model; C: Recessive model; D: Heterozygous model; E: Allelic contrast model)
Figure 2
Forest plots of the association of GRIA1 rs548294 G>A polymorphism and migraine risk (A: Homozygous model; B: Dominant model; C: Recessive model; D: Heterozygous model; E: Allelic contrast model)
Table 3
Meta-analysis of the association between GRIA1 rs548294 G>A polymorphism and migraine risk
ComparisonSubgroupPHModelPZOR (95%CI)
Homozygous model (AA vs. GG) Overall 0.01 0.49 1.19 (0.72–1.97) 
 Asian – – 0.51 1.18 (0.72–1.95) 
 Caucasian 0.004 0.63 1.19 (0.58–2.46) 
Dominant model [(AA+AG) vs. GG] Overall 0.009 0.52 1.11 (0.81–1.54) 
 Asian – – 0.91 0.98 (0.72–1.33) 
 Caucasian 0.004 0.51 1.17 (0.74–1.84) 
Recessive model [AA vs. (AG+GG)] Overall 0.08 0.46 1.15 (0.79–1.66) 
 Asian – – 0.40 1.22 (0.76–1.97) 
 Caucasian 0.03 0.69 1.12 (0.66–1.89) 
Heterozygous model (AG vs. GG) Overall 0.06 0.58 1.08 (0.82–1.42) 
 Asian – – 0.66 0.93 (0.67–1.29) 
 Caucasian 0.04 0.48 1.14 (0.79–1.67) 
Allelic contrast model (A vs. G) Overall 0.003 0.49 1.10 (0.84–1.43) 
 Asian – – 0.75 1.04 (0.82–1.31) 
 Caucasian 0.0009 0.55 1.12 (0.77–1.63) 
ComparisonSubgroupPHModelPZOR (95%CI)
Homozygous model (AA vs. GG) Overall 0.01 0.49 1.19 (0.72–1.97) 
 Asian – – 0.51 1.18 (0.72–1.95) 
 Caucasian 0.004 0.63 1.19 (0.58–2.46) 
Dominant model [(AA+AG) vs. GG] Overall 0.009 0.52 1.11 (0.81–1.54) 
 Asian – – 0.91 0.98 (0.72–1.33) 
 Caucasian 0.004 0.51 1.17 (0.74–1.84) 
Recessive model [AA vs. (AG+GG)] Overall 0.08 0.46 1.15 (0.79–1.66) 
 Asian – – 0.40 1.22 (0.76–1.97) 
 Caucasian 0.03 0.69 1.12 (0.66–1.89) 
Heterozygous model (AG vs. GG) Overall 0.06 0.58 1.08 (0.82–1.42) 
 Asian – – 0.66 0.93 (0.67–1.29) 
 Caucasian 0.04 0.48 1.14 (0.79–1.67) 
Allelic contrast model (A vs. G) Overall 0.003 0.49 1.10 (0.84–1.43) 
 Asian – – 0.75 1.04 (0.82–1.31) 
 Caucasian 0.0009 0.55 1.12 (0.77–1.63) 

Abbreviations: F, fixed-effects model; R, random-effects model.

As shown in Figure 3 and Table 4, there was a significant association between GRIA1 rs2195450 C>T polymorphism and migraine risk under heterozygous model (CT vs. CC, OR = 1.23, 95%CI = 1.02–1.48, PZ = 0.03). In addition, subgroup analysis based on ethnicity showed that GRIA1 rs2195450 C>T polymorphism was significantly associated with migraine risk in Asian population ((TT+CT) vs. CC, OR = 1.83, 95%CI = 1.26–2.66, PZ = 0.001; CT vs. CC, OR = 1.75, 95%CI = 1.19–2.57, PZ = 0.005; T vs. C, OR = 1.79, 95%CI = 1.28–2.49, PZ = 0.0006), but not in Caucasian population.

Forest plots of the association of GRIA1 rs2195450 C>T polymorphism and migraine risk (A: Homozygous model; B: Dominant model; C: Recessive model; D: Heterozygous model; E: Allelic contrast model)

Figure 3
Forest plots of the association of GRIA1 rs2195450 C>T polymorphism and migraine risk (A: Homozygous model; B: Dominant model; C: Recessive model; D: Heterozygous model; E: Allelic contrast model)
Figure 3
Forest plots of the association of GRIA1 rs2195450 C>T polymorphism and migraine risk (A: Homozygous model; B: Dominant model; C: Recessive model; D: Heterozygous model; E: Allelic contrast model)
Table 4
Meta-analysis of the association between GRIA1 rs2195450 C>T polymorphism and migraine risk
ComparisonSubgroupPHModelPZOR (95%CI)
Homozygous model (TT vs. CC) Overall 0.004 0.46 1.30 (0.64–2.62) 
 Asian – – 0.06 2.74 (0.95–7.90) 
 Caucasian 0.003 0.85 1.08 (0.48–2.46) 
Dominant model [(TT+CT) vs. CC] Overall 0.03 0.09 1.29 (0.96–1.74) 
 Asian – – 0.001 1.83 (1.26–2.66) 
 Caucasian 0.11 0.12 1.17 (0.961.41) 
Recessive model [TT vs. (CT+CC)] Overall 0.003 0.54 1.24 (0.622.46) 
 Asian – – 0.10 2.45 (0.857.02) 
 Caucasian 0.002 0.91 1.05 (0.472.37) 
Heterozygous model (CT vs. CC) Overall 0.14 0.03 1.23 (1.02–1.48) 
 Asian – – 0.005 1.75 (1.19–2.57) 
 Caucasian 0.49 0.34 1.11 (0.901.37) 
Allelic contrast model (T vs. C) Overall 0.0006 0.16 1.27 (0.911.77) 
 Asian – – 0.0006 1.79 (1.28–2.49) 
 Caucasian 0.002 0.48 1.14 (0.791.67) 
ComparisonSubgroupPHModelPZOR (95%CI)
Homozygous model (TT vs. CC) Overall 0.004 0.46 1.30 (0.64–2.62) 
 Asian – – 0.06 2.74 (0.95–7.90) 
 Caucasian 0.003 0.85 1.08 (0.48–2.46) 
Dominant model [(TT+CT) vs. CC] Overall 0.03 0.09 1.29 (0.96–1.74) 
 Asian – – 0.001 1.83 (1.26–2.66) 
 Caucasian 0.11 0.12 1.17 (0.961.41) 
Recessive model [TT vs. (CT+CC)] Overall 0.003 0.54 1.24 (0.622.46) 
 Asian – – 0.10 2.45 (0.857.02) 
 Caucasian 0.002 0.91 1.05 (0.472.37) 
Heterozygous model (CT vs. CC) Overall 0.14 0.03 1.23 (1.02–1.48) 
 Asian – – 0.005 1.75 (1.19–2.57) 
 Caucasian 0.49 0.34 1.11 (0.901.37) 
Allelic contrast model (T vs. C) Overall 0.0006 0.16 1.27 (0.911.77) 
 Asian – – 0.0006 1.79 (1.28–2.49) 
 Caucasian 0.002 0.48 1.14 (0.791.67) 

Abbreviations: F, fixed-effects model; R, random-effects model.

Bold terms indicate significant OR.

Sensitivity analysis and publication bias

Each study was sequentially removed to assess the impact of single study on the pooled ORs. The result showed that the omission of any one study did not significantly change the overall estimation result on the association between GRIA1 rs548294 G>A polymorphism and migraine risk. However, after removing Cargnin’s study, the significant association between GRIA1 rs2195450 C>T polymorphism and migraine risk was observed under homozygous model (TT vs. CC, OR = 1.73, 95%CI = 1.25–2.40, PZ = 0.009), heterozygous model (CT vs. CC, OR = 1.33, 95%CI = 1.08–1.65, PZ = 0.007), dominant model ((TT+CT) vs. CC, OR = 1.45, 95%CI = 1.19–1.76, PZ = 0.0002), and allelic contrast model (T vs. C, OR = 1.46, 95%CI = 1.12–1.91, PZ = 0.005). After removing Fang’s study or Formicola’s study, no significant association was observed between GRIA1 rs2195450 C>T polymorphism and migraine risk under heterozygous model.

Funnel plots were used to assess the publication bias. As shown in Supplementary Figure S1, the funnel plot seemed symmetric and did not show obvious publication bias.

Several genome-wide association studies have identified more than a dozen loci associated with migraine risk, which suggests novel genes and pathways involved in migraine pathophysiology that point toward neuronal and vascular mechanisms of disease [15–17]. In addition, a subsequent genome-wide meta-analysis also revealed five new susceptibility loci for migraine, indicating that considerable increases in the number of cases and controls could further increase the number of risk loci [18].

In 2010, Formicola et al. [7] first reported that GRIA1 polymorphisms (rs548294 G>A, rs2195450 C>T) were significantly associated with migraine risk in an Italian population. Subsequently, three independent case–control studies from different countries were conducted to confirm this findings [11–13]. The result of their studies argued that GRIA1 rs548294 G>A polymorphism was not significantly associated with migraine risk. For GRIA1 rs2195450 C>T polymorphism, the result of Cargnin’s study and Maher’s study was contrary to that of Formicola’s study [7,12,13]. However, the result of Fang’s study indicated that the GRIA1 rs2195450 C>T polymorphism was significantly associated with migraine risk [11]. These inconsistent results would not be helpful to our understanding of the association between the two polymorphisms and migraine risk. Therefore, an effective method should be adopted to reach a more accurate conclusion.

Meta-analysis is a very powerful tool for analyzing cumulative data of studies, in which the sample sizes are small and the statistical power is poor. In the present study, we used this method to systematically assess the association between GRIA1 rs548294 G>A and rs2195450 C>T polymorphisms and migraine risk, and found that GRIA1 rs548294 G>A polymorphism was not significantly associated with migraine risk, but GRIA1 rs2195450 C>T polymorphism was significantly associated with migraine risk. Due to the fact that rs2195450 polymorphism was located in GRIA1 intron (NM_000827.3) and had no impact on splicing by the prediction of Human Splicing Finder tool (http://www.umd.be/HSF3/), we inferred that rs2195450 polymorphism was probably not a functional variant. Further linkage analyses for rs2195450 would contribute to finding functional variant associated with migraine risk. Subgroup analysis based on ethnicity showed a significant association of GRIA1 rs2195450 C>T polymorphism with migraine risk in Asian population, but not in Caucasian population, suggesting that the effect of this polymorphism might depend on genetic background of different ethnicities.

As far as we know, this is the first meta-analysis reporting the association of GRIA1 rs548294 G>A and rs2195450 C>T polymorphisms and migraine risk. Although the present results could provide stronger evidence for the association, some limitations should not be ignored. First, the number of studies and samples included in the meta-analysis was small and the analysis was based on only four case–control studies, of which three included Caucasian population and one Asian population. The omission of one study, such as Cargnin’s study, Fang’s study or Formicola’s study, could change the overall estimation results of rs2195450 polymorphism under different genetic model, suggesting that the results were not robust and needed to be treated cautiously. In addition, our meta-analysis was based on unadjusted estimates, while a more precise analysis should be performed if individual data such as gender, age, lifestyle and environmental factors were available.

Our results indicates that GRIA1 rs2195450 C>T polymorphism is significantly associated with migraine risk. However, more high-quality case–control studies with large sample sizes are still required to confirm these findings.

Xueren Gao and Jianguo Wang did the literature search, screened the potentially eligible studies and evaluated the data from each included study. They critically revised this manuscript, including important intellectual contents and grammatical mistakes. All authors read and approved the final manuscript.

The authors declare that there are no sources of funding to be acknowledged.

The authors declare that there are no competing interests associated with the manuscript.

     
  • AMPA

    α-amino-3-hydroxy-5-methyl-4-isoxazole propionate

  •  
  • CI

    confidence interval

  •  
  • OR

    odds ratio

  •  
  • SNP

    single-nucleotide polymorphism

1
Cámara-Lemarroy
C.R.
,
Rodriguez-Gutierrez
R.
,
Monreal-Robles
R.
and
Marfil-Rivera
A.
(
2016
)
Gastrointestinal disorders associated with migraine: a comprehensive review
.
World J. Gastroenterol.
22
,
8149
8160
[PubMed]
2
Goadsby
P.J.
,
Lipton
R.B.
and
Ferrari
M.D.
(
2002
)
Migraine–current understanding and treatment
.
N. Engl. J. Med.
346
,
257
270
[PubMed]
3
Bernstein
J.A.
,
Fox
R.W.
,
Martin
V.T.
and
Lockey
R.F.
(
2013
)
Headache and facial pain: differential diagnosis and treatment
.
J. Allergy Clin. Immunol. Pract.
1
,
242
251
[PubMed]
4
Geyik
S.
,
Ergun
S.
,
Kuzudişli
S.
,
Şensoy
F.
,
Temiz
E.
,
Altunışık
E.
et al (
2016
)
Plasma urotensin-2 level and Thr21Met but not Ser89Asn polymorphisms of the urotensin-2 gene are associated with migraines
.
J. Headache Pain
17
,
36
[PubMed]
5
Sezer
S.
,
Kurt
S.
and
Ates
O.
(
2016
)
Analysis of dopamine beta hydroxylase gene polymorphisms in migraine
.
Clin. Neurol. Neurosurg.
145
,
96
100
[PubMed]
6
Rodriguez-Acevedo
A.J.
,
Ferreira
M.A.
,
Benton
M.C.
et al (
2015
)
Common polygenic variation contributes to risk of migraine in the Norfolk Island population
.
Hum. Genet.
134
,
1079
1087
[PubMed]
7
Formicola
D.
,
Aloia
A.
,
Sampaolo
S.
et al (
2010
)
Common variants in the regulative regions of GRIA1 and GRIA3 receptor genes are associated with migraine susceptibility
.
BMC Med. Genet.
11
,
103
[PubMed]
8
Kowalska
M.
,
Prendecki
M.
,
Kozubski
W.
,
Lianeri
M.
and
Dorszewska
J.
(
2016
)
Molecular factors in migraine
.
Oncotarget
7
,
50708
50718
[PubMed]
9
Vikelis
M.
and
Mitsikostas
D.D.
(
2007
)
The role of glutamate and its receptors in migraine
.
CNS Neurol. Disord. Drug Targets
6
,
251
257
[PubMed]
10
Ramadan
N.M.
(
2003
)
The link between glutamate and migraine
.
CNS Spectr.
8
,
446
449
[PubMed]
11
Fang
J.
,
An
X.
,
Chen
S.
,
Yu
Z.
,
Ma
Q.
and
Qu
H.
(
2015
)
Case-control study of GRIA1 and GRIA3 gene variants in migraine
.
J. Headache Pain
17
,
2
[PubMed]
12
Cargnin
S.
,
Viana
M.
,
Mittino
D.
et al (
2014
)
Lack of association between GRIA1 polymorphisms and haplotypes with migraine without aura or response to triptans
.
Neurol. Sci.
35
,
421
427
[PubMed]
13
Maher
B.H.
,
Lea
R.A.
,
Follett
J.
et al (
2013
)
Association of a GRIA3 gene polymorphism with migraine in an Australian case-control cohort
.
Headache
53
,
1245
1249
[PubMed]
14
Stang
A.
(
2010
)
Critical evaluation of the Newcastle-Ottawa scale for the assessment of the quality of nonrandomized studies in meta-analyses
.
Eur. J. Epidemiol.
25
,
603
605
[PubMed]
15
Anttila
V.
,
Stefansson
H.
,
Kallela
M.
et al (
2010
)
Genome-wide association study of migraine implicates a common susceptibility variant on 8q22.1
.
Nat. Genet.
42
,
869
873
16
Freilinger
T.
,
Anttila
V.
,
de Vries
B.
et al (
2012
)
Genome-wide association analysis identifies susceptibility loci for migraine without aura
.
Nat. Genet.
44
,
777
782
[PubMed]
17
Chasman
D.I.
,
Schürks
M.
,
Anttila
V.
et al (
2011
)
Genome-wide association study reveals three susceptibility loci for common migraine in the general population
.
Nat. Genet.
43
,
695
698
[PubMed]
18
Anttila
V.
,
Winsvold
B.S.
,
Gormley
P.
et al (
2013
)
Genome-wide meta-analysis identifies new susceptibility loci for migraine
.
Nat. Genet.
45
,
912
917
[PubMed]
This is an open access article published by Portland Press Limited on behalf of the Biochemical Society and distributed under the Creative Commons Attribution License 4.0 (CC BY).

Supplementary data