Glutamate ionotropic receptor NMDA type subunit 2A (GRIN2A) gene polymorphism (rs4998386) and Parkinson's disease susceptibility: A meta‐analysis

Abstract Objective Dopaminergic neuronal degeneration seen in Parkinson's disease (PD) might result from a single nucleotide polymorphism (SNP) in the glutamate ionotropic receptor NMDA type subunit 2A (GRIN2A) gene. We thus performed a meta‐analysis exploring the relationship between the rs4998386 SNP of the GRIN2A gene and PD susceptibility. Methods We searched PubMed, EMBASE, Web of Science, Google Scholar, and China National Knowledge Infrastructure for studies published between January 2005 and January 2019. The association between the rs4998386 polymorphism and PD susceptibility was evaluated by calculating the pooled odds ratios (ORs) and 95% confidence intervals (CIs). Results Meta‐analysis results did not show a significant association between the rs4998386 polymorphism of the GRIN2A gene and PD susceptibility when assuming an allelic model (OR, 0.90; 95% CI, 0.76‐1.07; P = .22; I 2 = 53%), a dominant model (OR, 0.96; 95% CI, 0.82‐1.12; P = .62; I 2 = 64%), or a recessive model (OR, 1.14; 95% CI, 0.93‐1.38; P = .22; I 2 = 0%). Conclusion Our meta‐analysis found that the rs4998386 polymorphism of the GRIN2A gene is not associated with risk of PD in either Europeans or white Americans. However, large sample studies with different ethnicities should be conducted to establish the role of the rs4998386 polymorphism in PD pathophysiology.

Current research hypothesizes PD as a compilation of complex interactions between genetic and environmental factors. PD was historically considered sporadic in nature, without genetic origin. In the past decade, however, genetic studies performed in various geographical regions worldwide have strengthened the hypothesis that PD has a substantial genetic component. Numerous genes have unequivocally been tied to familial PD, including SNCA, PRKN, PINK1, DJ1, LRRK2, ATP13A2, and PLA2G6. 4,5 This "familial form" of PD, however, accounts for only a small percentage of overall PD diagnoses.
The majority of PD cases are considered a "sporadic form" of disease, indicating that additional factors may also contribute to the risk of PD.
Current evidence suggests an association between glutamate-mediated excitatory neurotoxicity and neurodegenerative diseases. [6][7][8] The excitatory effects of glutamate are exerted via activation of ionotropic receptors, mainly at NMDA receptors (NMDARs). NMDAR channels are heteromers composed of the key receptor subunit, NMDAR1, with one or more of the four NMDAR2 subunits: NMDAR2A, NMDAR2B, NMDAR2C, and NMDAR2D. 9 Continuous activation of large numbers of NMDARs can lead to elevated intracellular calcium and catabolic enzyme activity, which triggers a cascade of events leading to cell death. These event cascades include mitochondrial membrane depolarization, caspases activation, cellular toxicity, and production of reactive oxygen species and nitrogen free radicals. 8 A growing body of evidence supports an essential role for oxidative stress in the initiation and progression of neurodegeneration in PD. [10][11][12] The substantia nigra pars compacta (SNPC), the location of the primary pathological lesion, is predisposed to oxidative stress as well as toxic and metabolic insults. 8 One hypothesis suggests that mitochondria genetic alterations lead to impaired mitochondrial function and, therefore, degeneration of SNPC. 11,12 Oxidative stress may also be mediated by NMDAR-induced production of reactive oxygen species, such as peroxide, hydroxyl radical, and superoxide. Animal studies have shown that both competitive and non-competitive antagonism at the NMDAR lead to an anti-parkinsonism effect. 13 Currently, only one anti-glutamatergic pharmacotherapy, amantadine, is used clinically to treat PD. Amantadine has been shown to noncompetitively block NMDARs at therapeutic concentrations in the central nervous system, supporting the association between NMDA and PD. 13 Unfortunately, the precise role of NMDAR-mediated excitotoxicity in PD progression has not been established and thus remains unclear.
With this supporting data, we hypothesized that dopaminergic neuronal degeneration seen in PD could result from a single nucleotide polymorphism (SNP) in the NMDAR. The glutamate ionotropic receptor NMDA type subunit 2A (GRIN2A) gene encodes for an epsilon subunit of NMDAR (ie, NMDAR2A). No studies have focused on an association between the SNP of the GRIN2A gene and PD susceptibility. However, four studies to date have established an interaction between rs4998386 polymorphisms of the GRIN2A gene and caffeine on PD risk. Though these studies were not performed to establish association between GRIN2A rs4998386 SNP and PD, we extracted relevant data from these studies to perform a meta-analysis exploring the relationship between the rs4998386 SNP of the GRIN2A gene and PD susceptibility.

| MATERIAL S AND ME THODS
This meta-analysis was conducted in accordance with the guidance of the Preferred Reporting Items for Systematic Reviews and Meta-Analyses (PRISMA) statement. 14 All collected data were extracted from published studies, without ethical issues.

| Literature search
We searched PubMed, EMBASE, Web of Science, Google Scholar, and China National Knowledge Infrastructure for studies published between January 2005 and January 2019. Searches were conducted using the following keywords and phrases: "GRIN2A gene polymorphism," "glutamate ionotropic receptor NMDA type subunit 2A gene polymorphism," "NMDAR2A gene polymorphism," or "rs4998386 polymorphism" in combination with "Parkinson's disease," "idiopathic disease," or "neurodegenerative disease." Titles and abstracts were screened and full text was acquired for all papers. Those that met inclusion criteria were manually reviewed for confirmation of their eligibility. Screening, report retrieval, and inclusion/exclusion of studies were conducted by two authors (G. N. and R. O.). Any uncertainty about study eligibility was resolved through discussion with the third author.

| Eligibility criteria
Studies conducted on human subjects were included, without limitation of study language. In this meta-analysis, all included studies met the following criteria: (a) studies focused on the association between rs4998386 polymorphism and PD susceptibility, (b) patients with PD were diagnosed using standard diagnostic criteria by a neurologist, and (c) there were sufficient data of the genotypes in the case-control groups to evaluate the odds ratios (ORs) and 95% confidence intervals (CIs). The exclusion criteria were: (a) publications with overlapping cases and controls from a similar study, (b) no genotypic data available, and (c) publications with patients with other forms of parkinsonism.

| Data abstraction and assessment of methodological quality
Relevant data from each study, along with their supplementary files, were independently extracted by two reviewers (G. N. and R. O.) using a standardized, structured methodology that included the following: first author, type of design, site of study, patient ethnicity, year of publication, sample size of cases and controls, control design, Hardy-Weinberg equilibrium, source of DNA, genotyping method, and data of the genotypes in the case-control groups. The methodological quality of each study was assessed independently by two reviewers (G. N. and R. O.) using the parameters previously used by Kumar et al. 15 In this scale, the following sections were included for assessment of methodological quality: representativeness of cases, source of controls, matching of controls, ascertainment of cases, ascertainment of controls, blinding while genotyping, genotyping methods, Hardy-Weinberg equilibrium, and association assessment. A maximum score of 2 can be provided for each section listed, except source of controls (maximum 3), ascertainment of controls (maximum 1), blinding while genotyping (maximum 1), and association assessment (maximum 1). Methodological quality of scale score varies from 0 to 16, in which higher scores represent better quality and lower scores represent lower quality. Any discrepancies during data extraction and quality assessment were resolved by discussion with the third author.

| Statistical analysis
The association between the rs4998386 polymorphism and PD susceptibility was evaluated by calculating the pooled ORs and 95% CIs.

Heterogeneity between the included studies was determined with
Cochran's Q test and the I 2 test. The presence of P values above .1 or I 2 more than 50% was considered an indicator of significant heterogeneity. When significant heterogeneity was present, we selected the random-effects model to calculate the effects size. Three genetic models were examined: the allelic model (T allele vs C allele), the dominant model (TT + TC vs CC), and the recessive model (TT vs CC + TC). Subgroup analyses were conducted according to geographic location. Sensitivity analysis was performed to examine the stability of analysis. Egger's linear regression test and Begg's test were used to identify publication biases. Statistical analysis was performed using Comprehensive Meta-Analysis software (CMA 3.3, Biostat, 2014).

| Literature search
The results of the systematic literature search and selection are summarized in Figure 1. We identified 58 articles through various database searching. After excluding duplicates, 26 articles remained.
Screening titles and abstracts of these 26 articles resulted in seven relevant abstracts. After inclusion and exclusion criteria, four studies were chosen for final data analysis.

| Study and patient characteristics
Our research included four studies with data from 12 different cohorts. All studies used a case-control design. There were 5783 cases and 8788 controls in our study. Hamza IIN). 19 Allelic and genotypic data from each cohort are illustrated in Table 1.
Of the 12 different cohorts included in our study, nine were from the United States, with all included participants of Caucasian race.
Methods of PD case and control ascertainment are given in Table 2.
All included cohorts reached Hardy-Weinberg equilibrium in their control population. For genotyping, the TaqMan assay method was most commonly used among cohorts, though some applied the Illumina

| Meta-analysis
Due to significant heterogeneity within studies in the allelic and dominant models, we used a random-effects model to estimate pooled ORs and 95% CIs. The recessive model was without heterogeneity, so we used a fixed-effects model to estimate pooled ORs and 95% CIs.

| D ISCUSS I ON
While the NMDAR is involved in synaptic plasticity and learning, its excessive activation results in excitotoxic neuronal damage, which is hypothesized to be a mechanism underlying several neurologic and neurodegenerative disorders. Increased glutamatergic transmission through the NMDAR is believed to play a role in the degeneration of the nigrostriatal dopaminergic pathway seen in PD pathophysiology. This deterioration leads to significant changes in striatal circuit structure and function, including modifications of the corticostriatal glutamatergic synaptic architecture and, consequently, loss of striatal synaptic plasticity and dopaminergic neurons. 20 In addition, continuous mass activation of NMDARs increases intracellular calcium loads and catabolic enzyme activities, triggering a cascade that leads to cell death. 8 Various animal models of PD have shown that NMDAR antagonists alleviate symptoms of PD. 13 In animal models, NMDAR antagonists are effective antiparkinsonian agents and can reduce the complications of chronic dopaminergic therapy, such as "wearing off" and dyskinesias. 21 Amantadine, an NMDAR antagonist, has been used clinically for treatment of dyskinesias and motor fluctuations in PD. 13 The neuronal NMDAR signaling has been found to be an important pathway in the development and progression of schizophrenia, 22,23 epilepsy, 24,25 Huntington's disease, [26][27][28] depression, 29 Alzheimer's disease, 30  No study has primarily studied the association between the rs4998386 polymorphism of the GRIN2A gene of NMDAR and PD risk. Therefore, we conducted a meta-analysis to examine the association between the rs4998386 polymorphism of the GRIN2A gene and PD risk.
Our meta-analysis found that the rs4998386 polymorphism is not associated with PD in Europeans or white Americans in any of the three genetic models, though our results may have been influenced by several factors. Given assumptions that PD is multi-determined and that the GRIN2A polymorphism alone is uncommon and small in effect, our power to detect genetic effects may have been reduced. PD is a multigenic

| CON CLUS ION
Our meta-analysis found that the rs4998386 polymorphism of the

CO N FLI C T S O F I NTE R E S T
The authors declare no conflicts of interest.