Association of GSTT1 and GSTM1 polymorphisms with early pregnancy loss in an Indian population and a meta-analysis

Transcription

1 Reproductive BioMedicine Online (2013) 26, REVIEW Association of GSTT1 and GSTM1 polymorphisms with early pregnancy loss in an Indian population and a meta-analysis Rohini R Nair a, Anuradha Khanna b, Kiran Singh a, * a Department of Molecular and Human Genetics, Banaras Hindu University, Varanasi , India; b Department of Obstetrics and Gynaecology, Institute of Medical Sciences, Banaras Hindu University, Varanasi , India * Corresponding author. addresses: singhk4@rediffmail.com, skiran@bhu.ac.in (K Singh). Kiran Singh, PhD is an Assistant Professor at Banaras Hindu University, Varanasi, India in the department of molecular and human genetics. She completed her PhD in 2005 in the cytogenetic laboratory, Banaras Hindu University. She specializes in reproductive genetics, including male and female infertility, recurrent pregnancy loss, and gynaecological malignancies. She has published book chapters and research articles in both national and international journals of repute. She is also a member of the editorial board of the Asian Pacific Journal of Reproduction. Abstract Glutathione S-transferase theta 1 (GSTT1) and glutathione S-transferase Mu 1 (GSTM1) enzymes of the glutathione detoxification pathway protect the embryo from oxidative stress. This study investigated GSTT1 and GSTM1 in relation to their role in conferring genetic susceptibility to pregnancy loss. In a case control study, 174 early pregnancy loss (EPL) patients, of which 130 were recurrent pregnancy loss (RPL) patients, and 180 healthy controls were investigated. Null genotypes of GSTT1 and GSTM1 were identified in duplex PCR reaction systems. Age-adjusted odds ratios (aor) were calculated by logistic regression analysis. A meta-analysis was also conducted. The GSTT1 null genotype was significantly associated with EPL (aor 4.47, P = 0.004) and RPL (aor 4.39, P = 0.006). No significant association of the GSTM1 null genotype was found with RPL. In a meta-analysis study, the presence of the GSTM1 null genotype was shown to be a risk for RPL. The GSTT1 null genotype was not found to be a risk factor for pregnancy loss in the pooled population but its association with RPL was found in the Indian population. This study suggests that women carriers of GSTT1 and GSTM1 null genotypes are more often at genetic risk of pregnancy loss. RBMOnline ª 2013, Reproductive Healthcare Ltd. Published by Elsevier Ltd. All rights reserved. KEYWORDS: detoxification pathway, early pregnancy loss, glutathione S-transferase, recurrent pregnancy loss Introduction Recurrent pregnancy loss (RPL) and early pregnancy loss (EPL) are multifactorial disorders with possible epidemiological risk factors, including gene mutations, structural abnormalities of the uterus, diabetes, maternal age, smoking, genital infections, caffeine or alcohol use and chemical exposures (Cramer and Wise, 2000; Fenster et al., 1991; Parazzini et al., 1991). The risk of pregnancy loss is increased by environmental as well as lifestyle factors such as stress, smoking, coffee and alcohol consumption (Cnattingius et al., 1985; Sokol et al., 1980). These toxins /$ - see front matter ª 2013, Reproductive Healthcare Ltd. Published by Elsevier Ltd. All rights reserved.

2 314 RR Nair et al. increase oxidative stress through their activation by phase I enzymes, and their elimination involves members of the phase II glutathione detoxification system. Abnormal placentation leads to the generation of reactive oxygen species, resulting in effects damaging to the invading embryo. Oxidative stress has been proposed to be involved in the aetiopathogenesis of abortion (Hempstock et al., 2003; Jauniaux et al., 2000). Physiological redox concentrations may be important for embryogenesis. Culture of early embryos at low O 2 concentration can influence both cellular mechanisms and gene expression, which improves embryonic development (Harlow and Quinn, 1979; Quinn and Harlow, 1978; Rinaudo et al., 2006; Umaoka et al., 1992). An excessive increase in reactive oxygen species production is detrimental to the embryo. The superoxide anion, hydrogen peroxide and hydroxyl radicals can have detrimental effects on the fetus (Guerin et al., 2001; Harvey et al., 2002). Excessive oxidative stress can result in DNA fragmentation, changes in gene expression, and organelle and membrane disturbances. Thus, oxidative stress-mediated damage of macromolecules plays an important role in fetal loss (Burton et al., 2003). Allelic variants of the detoxification genes that have impaired biotransformation functions may increase susceptibility to reproductive toxicity and lead RPL. Glutathione S-transferases (GST) are a family of enzymes, comprising eight distinct classes: alpha, kappa, mu, omega, pi, sigma, theta and zeta. GSTT1 belongs to the theta GST family and GSTM1 belongs to the mu family (Strange et al., 2001). GST catalyse the conjugation of a wide variety of toxic or carcinogenic electrophiles to the tripeptide glutathione. The majority of the GST substrates are xenobiotics or products of oxidative stress. GST is believed to exert a critical role in cellular protection against toxic foreign chemicals and oxidative stress (Hayes and Strange, 2000). Phenotypic absence of enzyme activity is due to inheritance of a homozygous deletion of the whole gene (i.e. the null genotype). This makes GST genes important candidates for exploration in the disorders linked with detoxification pathway. Genetic variability in GST isoenzymes among women with prenatal exposure to tobacco smoke or other chemicals may disrupt their redox status and may therefore be responsible for pregnancy loss or EPL risk. The role of GST in complex disorders is not limited to cancers (Anantharaman et al., 2007; Demir et al., 2005; Sabitha et al., 2008) or type 2 diabetes (Nowier et al., 2009) but also to other diseases such as cardiovascular or neurodegenerative disorders (Piacentini et al., 2012; Polimanti et al., 2011). To date there have been only few studies related to the association of GSTT1 and GSTM1 polymorphisms with unexplained RPL. Sata et al. (2003) reported in a study of 115 Japanese patients that the polymorphic variant of GSTM1 was associated with increased risk of RPL, whereas no association of the disease was found with the null genotype of GSTT1. Zusterzeel et al. (2000) reported no association of both GSTM1 and GSTT1 with RPL. Among Indian subjects, there are only two previous studies that have analysed GSTM1 and GSTT1 polymorphism with RPL: Suryanarayana et al. (2004) reported no association of GSTT1 and GSTM1 with RPL; on the other hand, Parveen et al. (2010) showed that the risk of RPL was significantly associated with GSTT1, whereas no association of the disease was found with the null genotype of GSTM1. Looking at the importance of GSTT1 and GSTM1 in pregnancy, this study postulated that polymorphism in this gene may affect the outcome in pregnant women, particularly the risk of miscarriage. Therefore, patients with EPL were recruited to find if the null genotypes of GSTT1 and GSTM1 gene affects the risk of pregnancy loss. This study also conducted a meta-analysis of GSTT1 and GSTM1 polymorphism to support the findings that the presence of null genotype increases the risk of pregnancy loss. Materials and methods Subjects Approval of the University s Ethical Committee for research on human material was obtained (Reference No. Dean/ /556, granted on 8 July 2009). EPL was defined as miscarriage before 12 weeks of gestation and RPL was defined as three or more consecutive EPL after conceiving from the same partner. In a hospital-based case control study, 191 EPL women who had miscarriages before 12 weeks of gestation were recruited as cases and 180 healthy fertile women with successful pregnancy outcome and no history of pregnancy-related complications as controls. The patients were enrolled between July 2009 and March Patients and controls were recruited from the outpatient department of the University Hospital, Department of Obstetrics and Gynaecology, Institute of Medical Sciences, Banaras Hindu University (BHU), Varanasi, India. A questionnaire for each patient was completed to record details of their lifestyle, habits and family history. Informed consent was obtained from all participants before their enrolment in the study. Women having prior miscarriage were investigated by a routine work up, including testing for chromosomal aberrations, uterine abnormalities, hormonal status and positive lupus anticoagulant or anticardiolipin or TORCH antibodies. Patients who experienced miscarriage for first time were analysed only for chromosomal abnormality. Seventeen patients were excluded after standardized clinical and laboratory evaluation due to anatomic, hormonal, chromosomal, infectious, autoimmune or thrombotic causes: four presented polycystic ovaries, three were anticardiolipin antibodies positive, one was both lupus anticoagulant and anticardiolipin antibodies positive, four were TORCH positive, one with arcuate uterus, one with septate uterus and one had hypothyroidism, and karyotype analysis revealed two patients with abnormal numerical chromosome complement (mosaic 46,XX and 45,XO). After exclusion of these 17 patients, genotype analysis was conducted on remaining 174 patients and all 180 controls. Among the 174 cases thus selected, 130 had at least three prior miscarriages (RPL group), 33 had two prior miscarriages and 11 had experienced miscarriage for the first time. All the subjects included in this study belong to the eastern Uttar Pradesh province of North India and fall within the same ethnic group. The characteristics of patients and controls are presented in Table 1.

3 Association of GSTT1 and GSTM1 polymorphisms with EPL 315 Table 1 controls. Parameter Specimens Characteristics of north Indian RPL patients and Blood samples of all the cases and controls were drawn from the antecubital vein and stored at 4 C until genomic DNA isolation was performed. Genotyping for GSTT1 and GSTM1 null polymorphisms Genotype analysis of GSTM1 and GSTT1 were performed by previously used forward and reverse primers in duplex PCR reaction systems (Suryanarayana et al., 2004). In each reaction, 50 ng genomic DNA was amplified in 1 PCR buffer containing 200 lmol/l dntp and their respective forward and reverse primers with 0.5 U Taq DNA polymerase. All reactions were conducted in an oil-free thermal cycler (Thermal Cyclers, BioRad). The amplified products were electrophoresed on 2% agarose gels containing ethidium bromide and the product bands were visualized under ultraviolet light. The presence of functional GSTT1 and GSTM1 was determined by amplification of the band of the expected size (Figure 1). Individuals were determined null for GSTT1 or GSTM1 when a band of the expected size was absent in the presence of the positive internal control band (an exon fragment of methionine synthase). Statistical analysis Patients Controls Age (years) ± ± 4.56 Previous 4 (3 7) 0 miscarriages Live births 0 2 Smokers 0 0 Alcohol consumers 0 0 Values are mean ± SD, mean (range) or n. Statistical significance of the differences in the frequency of genotypes using the chi-squared test, and crude odds ratios (OR) and 95% confidence intervals (95% CI) were calculated to assess the relative risk conferred by a null genotype. In addition, unconditional logistic regression was performed to calculate age-adjusted odds ratios (aor). Logistic regression analysis and calculation of confidence intervals was performed using a logistic regression online statistical calculator ( Power of study was calculated using G* power (Faul et al., 2009). All statistical tests were two-sided. P < 0.05 was considered statistically significant. Meta-analysis Literature search Several databases (Pub-Med, Google Scholar and Ovid) were searched using the keywords GST, detoxification pathway, oxidative stress and recurrent pregnancy loss in different combinations. Broad search terms were used to facilitate identification of all relevant articles with the last search performed on 16 August The search results were screened further to identify studies analysing GST polymorphism in RPL. The inclusion criteria were as follows: case control studies that investigated the association between GSTT1 and GSTM1 polymorphism and risk of RPL; studies presented original data and the number of null genotypes of GSTT1 and GSTM1 in cases and controls; and selection of the patients done according to the standard and reliable diagnosis parameters. Two independent researchers (RRN and KS) extracted raw data according to the inclusion and exclusion criteria. The following information was collected from each study using a data extraction form: the surname of the first author, date of publication, country of origin, ethnicity, characteristics of cases and controls, number of cases and controls and genotype distribution of cases and controls. Quantitative data synthesis The association of GSTT1 and GSTM1 polymorphism with RPL was estimated by calculating pooled OR and 95% CI. The significance of the pooled OR was estimated by Z test (P < 0.05 was considered statistically significant). The inconsistency value (I 2 ) given by Higgins et al. (2003) and the chi-squared-based Q statistic test by Cochran (1950) were performed to evaluate variations due to heterogeneity rather than chance. A random-effects or fixed-effects methods model was used to calculate pooled-effect estimates in the presence or absence of heterogeneity. In the presence of heterogeneity the random-effects model is more appropriate, while in case of homogeneity both models will show similar outcome pattern. Therefore, an analysis model for inference was chosen on the basis of the level of heterogeneity. High-resolution plots were generated corresponding to OR and 95% CI for both models. Sensitivity analysis was not conducted due to the low number of studies and lack of Figure 1 Genotype patterns of the four polymorphisms displayed by agarose gel electrophoresis. (A) Genotype patterns of the GSTT1 polymorphisms; lanes 2, 3, 6 and 7 show samples displaying the functional genotype (+); lanes 4 and 5 show samples displaying null genotypes ( ); M = puc/hinf1 marker. (B) Genotype patterns of the GSTM1 polymorphisms; lanes 3, 5, 8 and 9 show samples displaying the functional genotype (+); lanes 2, 4, 6 and 7 show samples displaying null genotypes ( ); M = 100 bp DNA ladder.

4 316 RR Nair et al. any potential outlier study. Begg s funnel plot, a scatter plot of effect against a measure of study size, was generated as a visual aid to detecting bias or systematic heterogeneity; an asymmetric funnel plot indicated a relationship between effect and study size, which suggested the possibility of either publication bias or a systematic difference between smaller and larger studies (small study effects). Furthermore, publication bias was assessed quantitatively using various statistical tools such as Egger s regression intercept, Duvall and Tweedie trim and fill procedure and Classic fail safe N. Classic Fail safe N suggests the number of null studies which have to be included to subsidize the effect to non-significant. Comprehensive Meta-Analysis version 2 (Biostat, Engelwood, New Jersey, USA) was used for conducting the quantitative data synthesis. Results Association of GSTT1 and GSTM1 null genotypes with pregnancy loss The frequency distributions of the GSTT1 and GSTM1 genotypes among the pregnancy loss and control groups are presented in Table 2. As the study groups differed in age, aor Figure 2 Flow diagram of the literature search. was calculated. The GSTT1 null genotype was found to be significantly higher in the EPL and RPL groups (10.92%, P = 0.004; 10.77%, P = 0.006, respectively) as compared with the control group (2.78%). The frequencies of the GSTM1 null alleles were Table 2 Genotype and allele frequencies of GSTT1 and GSTM1 polymorphisms among cases and controls and combined association of GSTT1 and GSTM1 polymorphisms with early and recurrent pregnancy loss. Polymorphism Cases Controls Crude OR P-value Age-adjusted OR P-value Early pregnancy loss GSTT / / /174 5/ ( ) ( ) GSTM / / /174 52/ ( ) NS 1.63 ( ) 0.03 GSTM1/GSTT1 +/+ 100/ / / 55/174 50/ ( ) NS a 1.41 ( ) NS a /+ 9/174 3/ ( ) NS a 3.87 ( ) NS a / 10/174 2/ ( ) a 7.06 ( ) 0.02 a Recurrent pregnancy loss GSTT / / /130 5/ ( ) ( ) GSTM1 + 83/ / /130 52/ ( ) NS 1.39 ( ) NS GSTM1/GSTT +/+ 77/ / / 39/130 50/ ( ) NS a 1.17 ( ) NS a /+ 6/130 3/ ( ) NS a 3.11 ( ) NS a / 8/130 2/ ( ) a 7.13 ( a Differences in frequencies between the case and control groups were analysed for statistical significance at the 95% confidence level using the chi-squared test. Odds ratios are reported with 95% confidence limits. Age-adjusted odds ratios were calculated using logistic regression analysis. NS = not statistically significant (P > 0.05). a Bonferroni correction was used for multiple comparison.

5 Association of GSTT1 and GSTM1 polymorphisms with EPL % and 36.15% in EPL and RPL patients, respectively, and 28.89% in the control group. However, no significant association of the GSTM1 null genotype was found with either EPL or RPL. The combined analysis for GSTT1 and GSTM1 showed that the GSTT1 null genotypes in combination with the GSTM1 null genotypes conferred a higher risk both for EPL (approximately 4.74-fold) and RPL (approximately 5.76-fold; Table 2). No statistically significant interaction of genotype with the confounding factor age was observed (data not shown). Meta-analysis Following the above-mentioned search criteria; only eight relevant studies on RPL were retrieved (Figure 2). Of the 29 publications initially identified, 20 articles were irrelevant and one was found to be a review article. All eight studies investigated the association between GSTM1 polymorphism whereas six studies investigated the association between GSTT1 polymorphism and risk of RPL (Table 3). The meta-analysis included two studies from Japan and Table 3 Study Description of the studies included in the meta-analysis. Country (region) Ethnicity Polymorphism evaluated Sample size (cases/ controls) Inclusion/exclusion criteria Hirvonen et al. (1996) Mendola et al. (1998) Sata et al. (2003) Zusterzeel et al. (2000) Suryanarayana et al. (2004) Parveen et al. (2010) Nonaka et al. (2011) Polimanti et al. (2012) North Carolina Black and White GSTM1 29/29 Inclusion criteria: two or more spontaneous consecutive miscarriages Exclusion criteria: NA New York Caucasian GSTM1 26/150 Inclusion criteria: two or more spontaneous consecutive miscarriages regardless of other pregnancy outcome Exclusion criteria: NA Japan Japanese GSTT1, GSTM1 115/160 Inclusion criteria: two or more spontaneous consecutive miscarriages and still births Exclusion criteria: anatomical abnormalities of genital tract, chromosomal abnormality, uterine conformational abnormality such as septate uterus Netherlands Caucasian GSTT1, GSTM1 187/109 Inclusion criteria: two spontaneous consecutive miscarriages before 17 weeks of gestation Exclusion criteria: chromosomal translocation, uterine abnormalities or positive lupus anticoagulant or cardiolipin antibodies South India North India Indo- Dravidian Indo- Aryan GSTT1, GSTM1 160/63 Inclusion criteria: three or more miscarriages with no live birth Exclusion criteria: chromosomal abnormality, positive TORCH test, positive antiphospholipid antibodies, uterine abnormality GSTT1, GSTM1 200/300 Inclusion criteria: three or more miscarriages with no live birth Exclusion criteria: chromosomal abnormality, positive TORCH test, positive antiphospholipid antibodies, uterine abnormality Japan Japanese GSTT1, GSTM1 103/101 Inclusion criteria: three or more miscarriages after conceiving with same partner Exclusion criteria: chromosomal abnormality, positive antiphospholipid antibodies, uterine abnormality Italy Italian GSTT1, GSTM1 121/113 Inclusion criteria: two or more miscarriages after conceiving with same partner Exclusion criteria: chromosomal abnormalities, thrombophilia, metabolic disorders, anatomical anomaly and immune factors Current study North India Indo- Aryan GSTT1, GSTM1 130/180 Inclusion criteria: three or more miscarriages with no live birth Exclusion criteria: chromosomal abnormality, positive TORCH test, positive antiphospholipid, lupus anticoagulant antibodies, uterine abnormality

6 318 RR Nair et al. Figure 3 Forest plots for the overall association between GSTM1 (A) and GSTT1 (B) polymorphisms and RPL risk. one each from New York, the Netherlands, North Carolina, north India, south India and Italy. The genotypes were classified into wild type and null for statistical calculations. GSTM1 polymorphism and RPL In total, eight studies met the inclusion criteria and were selected for the meta-analysis (Hirvonen et al., 1996; Mendola et al., 1998; Nonaka et al., 2011; Parveen et al., 2010; Polimanti et al., 2011; Sata et al., 2003; Suryanarayana et al., 2004; Zusterzeel et al., 2000), which in combination with data from the current study included 1071 cases and 1205 controls. The forest plot of the meta-analysis of GSTM1 is shown in Figure 3A. The I 2 value of 19.87% for the between-studies comparison showed a low level of true heterogeneity. The genotype data for the studies included were less heterogeneous (Q 9.98, not statistically significant, I %, tau-squared 0.01), favouring the use of the fixed-effects model for overall inference. The effects sizes for different studies were distributed around the centre, further favouring the use of the fixed-effects model. However, all this could be due to the small number of studies available for meta-analysis. Therefore, the results are presented for both models and there was a significant difference for both models, showing a significantly increased RPL risk in the presence of null genotypes (Figure 3A): fixed-effects OR 1.24, 95% CI , P = 0.016; random-effects OR 1.24, 95% CI , P = Egger s test for intercept showed no evidence of publication bias (intercept 0.05, SE 1.57, 95% CI 3.77 to 3.66, P two-tailed not statistically significant). Classic Fail safe N was 4 studies missing from the analysis for every published study included, which further supported absence of publication bias. A funnel plot of standard error by log odds ratio was generated; the almost-symmetrical distribution of studies showed absence of publication bias (Figure 4A). GSTT1 polymorphism and RPL This study retrieved six relevant studies on RPL and, including the data from the present study, there were 1016 cases and 1026 controls in this meta-analysis. True heterogeneity existed between studies for the GSTT1 null genotype (Q 17.86, P = 0.007, I %, tau-squared 0.16) comparison. The I 2 value of more than 50% for the between-studies comparison in both allele and genotype analysis showed a high level of true heterogeneity. The genotype data for the studies included were heterogeneous (I %), favouring the use of random-effects model for overall inference (Figure 3B). The overall result showed that no association between the GSTT1 null genotype and risk for RPL (OR 1.13, 95% CI ). Egger s test for intercept showed no evidence of publication bias (intercept 1.06, SE 3.07, 95% CI 6.84 to 8.98, P two-tailed not statistically significant). A funnel plot of standard error by log odds ratio was generated: the almost-symmetrical distribution of studies showed absence of publication bias (Figure 4B).

7 Association of GSTT1 and GSTM1 polymorphisms with EPL 319 Figure 4 Funnel plots for GSTM1 (A) and GSTT1 (B) polymorphisms and RPL risk. Figure 5 Relationship between GSTT1 polymorphisms and RPL risk by ethnicity in the Indian population. A subgroup analysis was performed on studies on the Indian population (Parveen et al., 2010; Suryanarayana et al., 2004 and the current study; Figure 5). After stratification, the heterogeneity was not statistically significant (Q 2.21, I , tau-squared 0.01, SE 0.14) and a significantly increased risk of the GSTT1 null genotype was found both in the fixed-effects (OR 2.08, 95% CI ) and random-effects models (OR 2.09, 95% CI ). The other studies could not be stratified due to the small sample sizes. Egger s test for intercept showed no evidence of publication bias (intercept 1.52, SE 2.07, 95% CI to 27.89, P two-tailed not statistically significant). Classic Fail safe N was 10 studies missing from the analysis for every published study included, which further supported absence of publication bias. GSTM1 GSTT1 interaction with RPL The association between GSTM1 GSTT1 interaction and RPL risk was evaluated in Sata et al. (2003), Parveen et al.

8 320 RR Nair et al. Figure 6 Forest plot of GSTM1 GSTT1 interaction (null/null versus present/present). (2010) and the current study. The dual null genotype was associated with a significantly increased risk of RPL when compared with the dual present genotype using the fixed-effects model (OR 3.08, 95% CI ; Figure 6). No significantly increased risk was detected in any other comparison group. Discussion The oxidative status of the cell modulates angiogenesis, which is critical for endometrial differentiation and embryonic growth (Geva and Jaffe, 2000; Zbucka et al., 2004). Oxidative stress results in functional changes associated with idiopathic RPL (Hempstock et al., 2003; Jauniaux et al., 2000). It induces cellular damage, endothelial damage and impaired placental vascularization (Wang et al., 1991; Myatt and Cui, 2004). The glutathione S-transferases are one of the most important parts of phase II enzymes. In human, there are several GST classes, among them GSTM1 and GSTT1 have been pointed out because a polymorphic deletion in these genes influences GST activity. Based on these backgrounds, the present study investigated the association GSTM1 and GSTT1 polymorphisms and risk of pregnancy loss and found that the null genotype of GSTT1 increases the risk of pregnancy loss. The phase I and phase II variants included in this study have been extensively characterized previously in various populations and disease association studies (Anantharaman et al., 2007; Demir et al., 2005; Nowier et al., 2009; Piacentini et al., 2012; Polimanti et al., 2011; Sabitha et al., 2008). Prior to this, there were only a few studies which investigated GSTM1 and GSTT1 polymorphisms with RPL. These genetic polymorphisms have been shown to vary considerably among studies. GSTM1 null genotypes were found to be associated with RPL in Japanese and North Carolina populations (Hirvonen et al., 1996; Sata et al., 2003). Parveen et al. (2010) on north Indian subjects revealed an association between the GSTT1 null genotype and the risk of RPL. However, other studies have not found any significant association of GSTT1 and GSTM1 with RPL (Nonaka et al., 2011; Polimanti et al., 2012; Suryanarayana et al., 2004; Zusterzeel et al., 2000). This study conducted meta-analysis on GSTT1 and GSTM1 null genotypes incorporating data from all eight published studies and the present one. All the nine studies were found to have contrasting results. Ethnic differences have been reported to have an effect on genetic predisposition to human diseases. The studies done by Mendola et al. (1998) and Zusterzeel et al. (2000) on Caucasian populations found no association of GSTT1 and GSTM1 with RPL, but the present study found results similar to those of Parveen et al. (2010) on north Indian subjects. However, two studies in Japan reported discrepancy in their results; in Sata et al. (2003), the GSTM1 null genotype showed a significantly increased risk of pregnancy loss among women with three or more pregnancy losses but not among women with two or more pregnancy losses; on the other hand, Nonaka et al. (2011) found no association of this polymorphism among women with three or more pregnancy losses. The discrepancy could be due to the different ethnicities in the populations: Nonaka et al. studied a Japanese population and Sata et al. studied a Sapporo population which includes the ethnic groups the Ainu and the Yamato, which differ from the general Japanese population. Therefore, this study conducted a meta-analysis on the data from all eight published studies as well as the present case control study to investigate the role of GSTT1 and GSTM1 polymorphisms in RPL. The meta-analysis showed an association between GSTM1 and the risk of RPL. For the GSTT1 null genotype, the risk was found to be significant in the Indian population rather than the pooled population. The association between RPL and smoking or consumption of coffee or alcohol has been discussed since the late 1970s. Zusterzeel et al. (2000) studied the distribution of GSTP1 genotypes in RPL patients and controls in relation to smoking or consumption of alcohol. Nonaka et al. (2011) studied the difference in distribution of GSTM1 and GSTM1 genotypes in RPL patients and controls in relation to smoking or consumption of coffee or alcohol. In comparison with controls, significant differences were found in the frequency of GSTM1 deletion in patients who drink coffee everyday, but not in patients who smoked cigarettes or consumed alcohol (Nonaka et al., 2011). More similar studies are needed to determine the effect of the interaction of GSTM1 and GSTT1 polymorphisms and smoking on the development of RPL. Understanding the interaction between environmental exposure and the GSTM1 null polymorphism is an important priority; the lack of available data precluded examination of gene environment interactions in this meta-analysis. One limitation of this meta-analysis is that it is relatively small and might not provide sufficient power to estimate the association between the null polymorphisms of GSTM1 and GSTT1 and RPL. Moreover, a great heterogeneity is found in the studies included. This also limits the

9 Association of GSTT1 and GSTM1 polymorphisms with EPL 321 applications of the knowledge generated. Therefore, these findings cannot be generalized across populations. Having mentioned the limitations, this analysis nevertheless encourages analysis of GSTT1 and GSTM1 polymorphisms in other populations, especially ethnically divergent populations, to find they are clinically important. In conclusion, the present study suggests a significant risk of RPL is associated with GSTT1 and GSTM1 polymorphisms such that the presence of null genotypes increases the risk of pregnancy loss. The evidence is not conclusive but is sufficient to suggest that there is a risk associated with a polymorphism. Obtaining more evidence with other studies would be worthwhile, particularly in making future meta-analysis more meaningful. Gene environment interactions play very important roles in the pathogenesis of other diseases and further studies investigating the effect of gene environment interactions on RPL risk are required. Acknowledgements The authors are thankful to all the patients and volunteers for providing blood samples. This work was financially supported by the Banaras Hindu University, Varanasi, India and Department of Biotechnology, New Delhi, India. RRN is receiving her Doctoral Fellowship from the Department of Biotechnology, Government of India, New Delhi, India. References Anantharaman, D., Chaubal, P.M., Kannan, S., Bhisey, R.A., Mahimkar, M.B., Susceptibility to oral cancer by genetic polymorphisms at CYP1A1, GSTM1 and GSTT1 loci among Indians: tobacco exposure as a risk modulator. Carcinogenesis 28, Burton, G.J., Hempstock, J., Jauniaux, E., Oxygen, early embryonic metabolism and free radical-mediated embryopathies. Reprod. Biomed. Online. 6, Cnattingius, S., Axelsson, O., Eklund, G., Lindmark, G., Smoking, maternal age and fetal growth. Obstet. Gynecol. 66, Cochran, W.G., The comparison of percentages in matched samples. Biometrika 37, Cramer, D.W., Wise, L.A., The epidemiology of recurrent pregnancy loss. Semin. Reprod. Med. 18, Demir, A., Altin, S., Demir, I., Koksal, V., Cetincelik, U., Dincer, S.I., The frequency of GSTT1 null genotype in Turkish population and lung cancer risk. Indian J. Hum. Genet. 11, Faul, F., Erdfelder, E., Buchner, A., Lang, A.G., Statistical power analyses using G*Power 3.1: tests for correlation and regression analyses. Behav. Res. Methods 41, Fenster, L., Eskenazi, B., Windham, G.C., Swan, S.H., Caffeine consumption during pregnancy and spontaneous abortion. Epidemiology 2, Geva, E., Jaffe, R.B., Role of vascular endothelial growth factor in ovarian physiology and pathology. Fertil. Steril. 74, Guerin, P., El Mouatassim, S., Menezo, Y., Oxidative stress and protection against reactive oxygen species in the pre-implantation embryo and its surroundings. Hum. Reprod. Update 7, Harlow, G.M., Quinn, P., Foetal and placental growth in the mouse after pre-implantation development in vitro under oxygen concentrations of 5 and 20%. Aust. J. Biol. Sci. 32, Harvey, A.J., Kind, K.L., Thompson, J.G., REDOX regulation of early embryo development. Reproduction 123, Hayes, J.D., Strange, R.C., Glutathione S-transferase polymorphisms and their biological consequences. Pharmacology 61, Hempstock, J., Jauniaux, E., Greenwold, N., Burton, G.J., The contribution of placental oxidative stress to early pregnancy failure. Hum. Pathol. 34, Higgins, J.P.T., Thompson, S.G., Deeks, J.J., Altman, D.G., Measuring inconsistency in meta-analyses. BMJ 327, Hirvonen, A., Taylor, J.A., Wilcox, A., Berkowitz, G., Schachter, B., Chaparro, C., Bell, D.A., Xenobiotic metabolism genes and the risk of recurrent spontaneous abortion. Epidemiology 7, Jauniaux, E., Watson, A.L., Hempstock, J., Bao, Y.P., Skepper, J.N., Burton, G.J., Onset of maternal arterial blood flow and placental oxidative stress. A possible factor in human early pregnancy failure. Am. J. Pathol. 157, Mendola, P., Moysich, K.B., Freudenheim, J.L., Shields, P.G., Schisterman, E.F., Graham, S., Vena, J.E., Marshall, J.R., Ambrosone, C.B., Risk of recurrent spontaneous abortion, cigarette smoking, and genetic polymorphisms in NAT2 and GSTM1. Epidemiology 9, Myatt, L., Cui, X., Oxidative stress in the placenta. Histochem. Cell Biol. 122, Nonaka, T., Takakuwa, K., Tanaka, K., Analysis of the polymorphisms of genes coding biotransformation enzymes in recurrent miscarriage in the Japanese population. J. Obstet. Gynaecol. Res. 37, Nowier, S.R., Kashmiry, N.K., Hoiyda, A., Rasool, A., Morad, H., Ismail, S., Association of Type 2 diabetes mellitus and glutathione S transferase (GSTM1 and GSTT1) genetic polymorphism. Res. J. Medicine. Med. Sci. 4, Parazzini, F., Bocciolone, L., Fedele, L., Negri, E., La Vecchia, C., Acaia, B., Risk factors for spontaneous abortion. Int. J. Epidemiol. 20, Parveen, F., Faridi, R.M., Das, V., Tripathi, G., Agrawal, S., Genetic association of phase I and phase II detoxification genes with recurrent miscarriages among north Indian women. Mol. Hum. Reprod. 16, Piacentini, S., Polimanti, R., Squitti, R., Ventriglia, M., Cassetta, E., Vernieri, F., Rossini, P.M., Manfellotto, D., Fuciarelli, M., GSTM1 null genotype as risk factor for late-onset Alzheimer s disease in Italian patients. J. Neurol. Sci. 317, Polimanti, R., Piacentini, S., Lazzarin, N., Re, M.A., Manfellotto, D., Fuciarelli, M., Glutathione S-transferase variants as risk factor for essential hypertension in Italian patients. Mol. Cell. Biochem. 357, Polimanti, R., Piacentini, S., Lazzarin, N., Vaquero, E., Re, M.A., Manfellotto, D., Fuciarelli, M., Glutathione S-transferase genes and the risk of recurrent miscarriage in Italian women. Fertil. Steril. 98, Quinn, P., Harlow, G.M., The effect of oxygen on the development of preimplantation mouse embryos in vitro. J. Exp. Zool. 206, Rinaudo, P.F., Giritharan, G., Talbi, S., Dobson, A.T., Schultz, R.M., Effects of oxygen tension on gene expression in preimplantation mouse embryos. Fertil. Steril. 86, Sabitha, K., Reddy, M.V.V., Jamil, K., GST genotypes in head and neck cancer patients and its clinical implications. Afr. J. Biotechnol. 7, Sata, F., Yamada, H., Kondo, T., Gong, Y., Tozaki, S., Kobashi, G., Kato, E.H., Fujimoto, S., Kishi, R., Glutathione S-transferase M1 and T1 polymorphisms and the risk of recurrent pregnancy loss. Mol. Hum. Reprod. 9,

10 322 RR Nair et al. Sokol, R.J., Miller, S.I., Reed, G., Alcohol abuse during pregnancy: an epidemiologic study. Alcohol Clin. Exp. Res. 4, Strange, R.C., Spiteri, M.A., Ramachandran, S., Fryer, A.A., Glutathione-S transferase family of enzymes. Mutat. Res. 482, Suryanarayana, V., Deenadayal, M., Singh, L., Association of CYP1A1 gene polymorphism with recurrent pregnancy loss in south Indian population. Hum. Reprod. 19, Umaoka, Y., Noda, Y., Narimoto, K., Mori, T., Effects of oxygen toxicity on early development of mouse embryos. Mol. Reprod. Dev. 31, Wang, Y.P., Walsh, S.W., Guo, J.D., Zhang, J.Y., Maternal levels of prostacyclin, thromboxane, vitamin E, and lipid peroxides throughout normal pregnancy. Am. J. Obstet. Gynecol. 165, Zbucka, M., Koda, M., Tomaszewski, J., Przystupa, W., Sulkowski, S., Wolczynski, S., Angiogenesis in the female reproductive processes. Ginekol. Pol. 75, Zusterzeel, P.L.M., Nelen, W.L.D.M., Roelofs, H.M.J., Peters, W.H.M., Blom, H.J., Steegers, E.A.P., Polymorphism in biotransformation enzymes and the risk for recurrent early pregnancy loss. Mol. Hum. Reprod. 6, Declaration: The authors report no financial or commercial conflicts of interest. Received 12 September 2012; refereed 8 December 2012; accepted 11 December 2012.