Introduction

Alzheimer’s disease (AD) is becoming a heavy burden to the human society nowadays [1]. AD is a multifactorial disorder, involving genetic and environmental factors [1, 2]. One major contributor is the oxidative stress, which plays an important role in Alzheimer’s diseases (see reviews: [3, 4]). However, the antioxidants, e.g., glutathione (GSH), could be potentially therapeutic [5]. As a group of the key antioxidant enzymes, the Glutathione S-transferases (GSTs) regulate the maintenance of GSH and cellular detoxification, and are involved in the activation of signals in cell apoptosis [6]. GST contains several subtypes, i.e., GST alpha (A, α), mu (M, μ), Pi (P, π), omega (O, ω), theta (T, θ), etc. [6]. The levels of GST and enzymatic activity are reduced in brain and ventricular fluid in AD [7, 8]. Genetic variations in these enzymes could impact the risk of diseases [9].

There were several studies to reveal associations between polymorphisms of GSTs’ genes and AD. However, these findings were debated by other reports with inconsistent results. It was suggested that no pooled analysis concerning GSTs gene variants and risk of AD was documented. Therefore, we conducted the current meta-analysis to assess the effect of GSTs gene polymorphisms upon AD risk.

Materials and methods

Search strategy

To identify the relevant published articles based on the relationship of GSTs genetic polymorphisms and the risk of AD, we conducted a systematic literature search in PubMed, Embase, Cochrane library, and Alzgene databases up to March 2014. The following terms “genetic polymorphism or variants,” “Glutathione S-transferase or GST,” “Alzheimer’s disease or AD or Alzheimer” were used and language was not restricted. A manual search of references in the retrieved articles was also performed to find additional potential studies.

Inclusion criteria and exclusion criteria

Inclusion criteria should be met: (1) studies evaluated GSTs gene polymorphisms and AD; (2) case–control or cohort studies design; (3) genotype distributions could be obtained from articles, authors, or other sources (i.e., Alzgene database); (4) the largest sample size or the latest studies were selected in case of overlapped publications. Accordingly, exclusion would be made under any of the conditions: nonclinical studies; abstracts, reviews or opinions, conference reports, and case reports; articles with insufficient genotypic data that could not be calculated or obtained from other resource.

Data extraction

After reviewing the relevant articles thoroughly, the following data for each single nucleotide polymorphism (SNP) on GSTs were extracted from studies that met the inclusion criteria: first author, year of publication, country of participants, ethnicity of the population, and allele distributions in cases and controls.

Statistical analysis

We performed the combined analysis for the SNPs that had more than three studies involved. The association between GSTs genetic variants and risk of AD was measured by the odds ratio (OR) with 95 % confidence intervals (CI) according to our previous method [10]. The Cochran’s Q test and I 2 test were used to estimate heterogeneity across studies. If a significant heterogeneity (assigned as p value <0.1 and I 2 > 50 %) existed, the random effect Mantel–Haenszel model was chosen. To determine the statistical significance of the pooled OR, Z test was used with p < 0.05 as statistically significant. Pooled analysis was performed under different genetic models [10] (i.e., codominant model, dominant model, recessive model, additive model, and allele model) when necessary. Begg’s test and Egger’s test were applied to evaluate the evidence of publication bias; p value <0.05 in both tests was considered statistically significant. Further, subjects were stratified by ethnicity to perform subgroup analyses. And cumulative meta-analysis was conducted to evaluate the trend of pooled results from studies that subsequently accumulated until the recent year of publications [11]. Statistical analyses were performed using STATA software (SE 11.0 version, Stata Corporation, College Station, TX, USA).

Results

Screening and characteristics of related studies

The initial search yielded 281 potential papers (239 from PubMed, 10 from Embase, 0 from Cochrane library, and 32 from Alzgene database). There were 200 repeated literatures, 51 records of animal studies, and nonclinical research or not related with GSTs gene or Alzheimer’s disease among these articles. 13 documents were excluded because of insufficient genotypic data among cases and controls. According to the inclusion criteria, 17 articles [12–28], were finally selected for further analysis (see Fig. 1, literature selections in each stage). Among these articles, genetic polymorphisms of GSTM1, GSTM3, GSTT1, GSTP1, and GSTO1 were studied in more than three researches. Most of the related studies only provided the distributions of present (+/+ and +/−) and null (−/−) alleles in GSTM1 and GSTT1 genes. The basic characteristics of all the included studies are shown in Table 1.

Fig. 1
figure 1

Flow diagram of article selection concerning GSTs polymorphisms in AD studies

Full size image
Table 1 Basic characteristics of the related studies in this meta-analysis
Full size table

Results of heterogeneity tests

Evidence of heterogeneity was observed among studies within the included GSTs SNPs(see Table 2, heterogeneity results of I 2 and p value). Therefore, random effects pooling model was used to calculate the combined ORs for all SNPs in AD.

Table 2 Pooled measures of the relevant GSTs polymorphisms in Alzheimer’s disease
Full size table

GSTM1 (present versus null) alleles

In summary, results of meta-analysis failed to show any influence of GSTM1 on the risk of AD among either overall population (OR = 0.988, 95 % CI 0.647–1.509; z = 0.06, p = 0.955), or Caucasians, Asians, or other populations (see Fig. 2; Table 2).

Fig. 2
figure 2

Forest plot about the effect of GSTM1, GSTT1 alleles upon Alzheimer’s disease risk

Full size image

GSTT1 (present versus null) alleles

A null association of GSTT1 in AD samples was found by the meta-analysis among overall population or Caucasians. A potential relationship was revealed in Asians for AD (OR = 0.405, 95 % CI 0.182–0.900; z = 2.22, p = 0.027); however, with only one study involved, this needs to be confirmed in the future (see Fig. 2; Table 2).

GSTM3 rs7483 (G/A)

Summary ORs and 95 % CIs were evaluated under different inherited models to estimate the correlation of GSTM3 rs7483 with AD risk. The results did not show any significant effect of GSTM3 rs7483 within each genetic model. Further subgroup analyses also found significant impact of GSTM3 rs7483 on AD risk in other population (one study was involved), but not in overall and Caucasian populations (see Table 2).

GSTP1 rs1695 (Ile105Val, A/G)

A meta-analysis was carried out for GSTP1 rs1695 in AD under different genetic models, and we did not find associations between GSTP1 rs1695 and AD risks among overall and Caucasian populations (see Table 2).

GSTO1 rs4925 (C/A), rs1804834 (A/G)

According to the pooling analysis and subgroup analysis, lack of significant effect of GSTO1 rs4925 on AD morbidity was concluded. The combined ORs and 95 % CI values under related inherited models are shown in Table 2.

Publication bias and cumulative analysis

Begg’s and Egger’s tests indicated no evidence of significant asymmetry for most of the related GSTs SNPs within Alzheimer’s disease (see Table 3).

Table 3 Results of Begg’s test and Egger’s test in the association between GSTs variants and AD risk
Full size table

In the cumulative meta-analyses, the evidences were observed to support the pooling analyses regarding relevant GSTs SNPs and AD (see Fig S1).

Discussion

Effects of GSTs genetic polymorphisms and ethnicity in neurodegenerative diseases [29] are of our interest. Combined analyses concerning GSTs SNPs in Alzheimer’s disease have not been reported yet. To the best of our knowledge, this is the first meta-analysis discussing GSTs polymorphisms among AD subjects. In this study, a total of 17 articles regarding SNPs of GSTM1, GSTT1, GSTM3, GSTP1, and GSTO1 genes were evaluated in AD. No significant effect of GSTM1 or GSTT1 present polymorphism was observed in Alzheimer’s disease in the pooled analyses. We used several genetic models to assess the roles of GSTM3, GSTP1, and GSTO1 genetic variants in AD; however, we found null associations between these GSTs SNPs and AD risk.

rs1332018 (C/A), rs1799735 (del/AGG) of GSTM3 [21]; GSTP1 Ala114Val (GSTP1*A/*B/*C/*D) [15, 24, 30]; GSTP1 C341T [24] and GSTO1 rs1804834 [27, 31] were also reported in AD. We did not perform any pooled analysis on these SNPs in AD because the relevant studies were limited.

Some potential limitations should be taken into consideration. First, the relative small global sample size and the number of cases in some ethnic subgroups were shortcomings, considering small or inadequate samples could lead to misleading results. Moreover, other GSTs genes, e.g., GSTA1, GSTO2, may also play a role in the development of Alzheimer’s disease. One research [18] suggested that GSTA1 rs3957356 was associated with the increased AD in Dominant model. Positive effect of rs156697 on GSTO2 was also found [32] among older (>80 years old) AD patients. Third, there was an evidence of heterogeneity in the current meta-analysis. One potential explanation to the heterogeneity is the influence of ethnic background or environment [2] in the correlation of gene variants with AD. Heterogeneity could also come from individual clinical study.

In conclusion, our meta-analysis suggested that SNPs of the relevant GSTs (GSTM1, GSTT1, GSTM3, GSTP1, and GSTO1) did not confer risk for AD. Further well-designed researches are needed to confirm these findings, especially in the subsets with limited studies involved in the current study.