Before inclusion of studies, we briefly searched PubMed, Embase, Web of Science and China National Knowledge Infrastructure, and found that most of studies examined association of GSTM1 or GSTT1 polymorphisms with cataract risk while very limited studies were related to other GST polymorphisms, e.g., GSTM3, GSTO or GSTP polymorphisms. Thus, this meta-analysis only evaluated the effects of GSTM1 and GSTT1 ploymorphisms on cataract risk. Our data showed that GSTT1 but not GSTM1 null polymorphism was associated with cataract risk in Asians. Although different subtypes of cataract have their own pathogenesis and clinical characteristics, our meta-analysis data indicate that GSTT1 null polymorphism may contribute to increased risk of posterior subcapsular cataract.
In 1995, Sekine and colleagues for the first time reported possible correlation of GSTM1 null genotype frequency with cataract risk . However, the following studies showed inconsistent results [18, 21–25, 32–35]. By pooling these early studies, previous meta-analysis by Sun et al., did not find an association of GSTM1 null genotype with cataract risk . Even including three more studies, we did not find positive relationship between GSTM1 null genotype and cataract risk. To be noted, although previous meta-analysis indicated an association of GSTM1 null genotype and increased risk of cataract in Asians , our data did not confirm this association when including one more study on Asians.
For GSTT1 polymorphism, pooled four early studies on Caucasian showed no association [18, 22, 24, 25] while one study on Asians  showed positive association between GSTT1 null genotype and cataract risk; however, by pooling these five studies, no association was found . By including four recent studies, our meta-analysis showed positive association of GSTT1 polymorphism with increased risk of cataract in all populations, and this association remained in Asians when two studies were pooling [15, 21]. Previous studies reported gender-dependent effects of GSTT1 null polymorphism on cataract risk [18, 22, 24]; however, recent two studies showed negative results [15, 16]. We performed a subgroup analysis stratified by gender with all five studies, and results showed no significant association, which was consistent with previous meta-analysis data based on three studies . In addition, our data showed positive association of GSTT1 null polymorphism with increased risk of posterior subcapsular cataract although previous pooled study indicated that this association did not reach significant (OR, 1.21; 95 % CI, 0.96–1.53) . Since the studies included for subgroup analyses were still limited, future studies are required to validate the association between GSTT1 null polymorphism and cataract risk.
To the best of our knowledge, the association between combination of GST polymorphisms and susceptibility to cataract has been assessed for the first time by our meta-analysis. The study by Juronen et al., firstly reported that the GSTM1 positive phenotype frequency was significantly higher in the cataract group than in the controls, and the cataract risk associated with the GSTM1 positive phenotype was increased in carriers of the combined GSTM1 positive and GSTT1 positive phenotypes . However, a later study by Saadat et al., showed that individuals with the null genotypes for GSTM1 and GSTT1, or combination of GSTT1 positive and GSTM1 null genotypes were at a significantly higher risk for developing cataract than individuals with both the genes positive genotypes . The following studies consecutively presented inconsistent results [15, 16, 18, 22]. By pooling seven studies, our meta-analysis results did not show a significant association between each combination of GSTM1 and GSTT1 genotypes and cataract risk. Two pooled studies with population-based controls showed that combination of GSTM1 null and GSTT1 positive genotypes played a protective role in cataract risk [16, 25]; however, this positive association was not found in other stratified analyses. Thus, the result should be interpreted with caution.
When compared to individual studies, the meta-analysis has a vital advantages. However, some potential limitations in our study should be considered. First, the inclusion of studies might not be sufficient since we only included published papers with language in English, or Chinese. It is possible that some papers published in other languages may not indexed by the database (e.g. PubMed, Embase, Web of Science). Thus, the publication bias for GSTM1 polymorphism detected in our study might be due to insufficient inclusion of published studies. Second, this meta-analysis was limited by the small sample size, especially in subgroup analyses aforementioned (e.g., studies on GSTT1 polymorphism in Asians), and this need further investigation. Third, basic methodological differences among the studies, e.g., ELISA vs. PCR assay for genotyping, might have affected the results. Fourth, most of the studies included did not categorize the cataract patients as cortical, nuclear, posterior subcapsular and mixed cataract. Although we found positive association between GSTT1 null polymorphism and increased risk of posterior subcapsular cataract, however, only four studies with available data were pooled [16, 18, 21, 25], and thus this association awaits further confirmation. Fifth, the primary outcome measure was calculated based on individual unadjusted ORs, which might affect the evaluation precision of the study. The lack of detailed data in each study prevented multiple testing for combined effects of gene-environment factors on cataract risk, and thus future studies should address this point. Last, the Caucasian and Asian subjects from different countries might have been genetically heterogeneous, e.g., different lifestyle and environment (e.g., European vs. Arabian). These factors may explain the heterogeneity in this meta-analysis for Caucasian subjects.