Lack of Any Association of GST Genetic Polymorphisms with Susceptibility to Ovarian Cancer-a Meta-analysis

OBJECTIVE
Epidemiology studies have reported conflicting results between glutathione S-transferase Mu-1 (GSTM1), glutathione S-transferase theta-1 (GSTT1) and glutathione S-transferase pi-1 (GSTP1) and ovarian cancer (OC) susceptibility. In this study, an updated meta-analysis was applied to determine whether the deletion of GSTM1, GSTT1 and GSTP1 has an influence on OC susceptibility.


METHODS
A published literature search was performed through PubMed, Embase, Cochrane Library, and Science Citation Index Expanded database for articles published in English. Pooled odds ratios (ORs) and 95% confidence intervals (95%CIs) were calculated using random or fixed effects models. Heterogeneity between studies was assessed using the Cochrane Q test and I2 statistics. Sub-group analysis was conducted to explore the sources of heterogeneity. Sensitivity analysis was employed to evaluate the respective influence of each study on the overall estimate.


RESULTS
In total, 10 published studies were included in the final analysis. The combined analysis revealed that there was no significant association between GSTM1 null genotype and OC risk (OR=1.01, 95%CI: 0.91-1.12). Additionally, there was no significant association between GSTT1 genetic polymorphisms and OC risk (OR=0.98, 95% CI: 0.85-1.13). Similalry, no significant associations were found concerning the GSTP1 rs1695 locus and OC risk. Meanwhile, subgroup analysis did not show a significant increase in eligible studies with low heterogeneity. However, sensitivity analysis, publication bias and cumulative analysis demonstrated the reliability and stability of the current meta-analysis.


CONCLUSIONS
These findings suggest that GSTs genetic polymorphisms may not contribute to OC susceptibility. Large epidemiological studies with the combination of GSTM1 null, GSTT1 null and GSTP1 Ile105Val polymorphisms and more specific histological subtypes of OC are needed to prove our findings.


Introduction
Ovarian cancer (OC), a malignant cancer faced by females both in the developed and developing countries (Sagae et al., 2002), is a lethal gynecologic malignancy due to cancerous growth arising from the epithelial cells, stromal cells and germ cells in ovary tissue.There is no

Lack of Any Association of GST Genetic Polymorphisms with Susceptibility to Ovarian Cancer -a Meta-analysis
Li-Yuan Han 1& , Kui Liu 2& , Xia-Lu Lin 1& , Bao-Bo Zou 1 , Jin-Shun Zhao* effect of environmental factors and genetic factors plays a critical role in the development of carcinoma and complex diseases (Lichtenstein et al., 2000;Clayton et al., 2001).There are evidences suggesting that family history, infertility and age are risk factors for OC, while increased parity, oral contraceptive use, hysterectomy and tubal ligation are protective factors (Booth et al., 1989;Whittmore et al., 1992;Risch et al., 1994;Rosenberg et al., 1994;Purdie et al., 1995;Riman et al., 1998;Riman et al., 2002;Lukanova et al., 2005).However, the majority of risk factors of OC are not clear, and the existing screening methods for OC such as serum CA 125, pelvic examination and transvaginal sonography (TVS) are not reliable in any risk group (Van Nagell et al., 1995;Bell et al., 1998;Jayde et al., 2012).
The glutathione S-transferases (GSTs) enzymes are an important phase II isoenzyme group which can catalyze the conjugation of glutathione with a variety of electrophilic compounds (Whalen et al., 1998).Enzymes of the GST family in eukaryotic species are composed of multiple cytosolic and membrane-bound isoenzymes, among them GSTM1, GSTT1 and GSTP1 belong to human GSTs and have been proven to play an important role in human carcinogenesis.It has been assumed that GST functional variants that are related to a less effective detoxification of potential carcinogens may contribute to increased cancer susceptibility (Strange et al., 1998).Salinas-Sanchez et al. (2011) showed that GSTM1 null genotype is a risk factor for bladder cancer and dual GSTM1-GSTT1 null genotype increases the bladder cancer risk.Additionally, Ye et al found that GSTM1 deletion and GSTT1 deficiency were important risk factors for head and neck cancer from population-based and hospital-based studies (Ye et al., 2004).Chen et al revealed a relationship between GSTM1 genetic polymorphism and increasing susceptibility to gastric cancer (Chen et al., 2010).Liu et al also demonstrated that null genotype of GSTM1 and GSTT1 were linked to increased risk in developing hepatocellular carcinoma (Liu et al., 2013).While by the means of meta-analysis, Jie Peng et al indicated that GSTM1 null genotype had a significant effect on the susceptibility of oral cancer in the Indian population (Peng et al., 2014).Archana Krishna Murthy et al discovered a significant relationship for the GSTM1 and GSTT1 null genotypes and nasopharyngeal cancer risk (Murthy et al., 2013).
Various epidemiological research including casecontrol and cohort studies have been carried out on the association between GSTs genetic variants and OC risks in the general population and specific occupational groups, but the results are inconsistent.In addition, most of these studies had either small sample sizes or design limitations.Given that meta-analysis is an efficient and powerful statistical method to combine different studies, we carry out an updated meta-analysis to determine whether the deletion of GSTM1, GSTT1 and/or GSTP1 has an influence on OC susceptibility.

Literature and search strategy
We searched PubMed, Cochrane Library, Embase, and Science Citation Index Expanded databases for case-control and cohort studies on GSTs polymorphisms and OC susceptibility in English.The following search terms were used: ("glutathione S-transferase" or "GST" or "GSTM1" or "GSTT1" or "GSTP1") and ("ovarian cancer" or "oophoroma" or "ovarian carcinoma" or "carcinoma of ovary").In addition, reference lists were also reviewed manually.The latest research was performed on February 28, 2014.

Inclusion and exclusion criteria
In order for articles to be included in our study, the following criteria must be met: 1) case-control or cohort studies; 2) evaluating the relationship between GSTM1, GSTT1 or GSTP1 and OC risk; 3) providing raw data, or relevant information which could be used to calculate an odds ratios (ORs) with 95% confidence intervals (CI); 4) genotype distribution in controls were in Hardy-Weinberg equilibrium.The exclusion criteria included: 1) repeated reports; 2) the genotype frequency was not reported; 3) case reports, editorials, review articles, conference papers and meta-analysis.

Data extraction and synthesis
All publications retrieved from the databases were examined by two independent reviewers (Liyuan Han and Kui Liu) and disagreements were solved by a third researcher (Jinshun Zhao).For each eligible study, the following characteristics were collected: first author, year of publication, the relevant polymorphisms, study design, sample size, ethnicity and geographical location of the populations, sources of control, characteristics of cases and controls.

Statistics and analysis
Meta-analysis using STATA software (version 11; Stata Corporation, College Station, Texas) was conducted.The Z test was used to calculate the P value of the overall effect for the meta-analysis.Heterogeneity assessment employed the Cochran's Q test and I 2 test.If the P value of the Q test was above 0.1 (P>0.1), the fixed-effect model was used to evaluate the pooled ORs and 95% CIs, otherwise, the random-effect model was used.Subgroup analysis (<200 vs.≥200, the source of controls: hospital-based vs population-based studies), sensitivity analysis and cumulative meta-analysis were performed to assess the heterogeneity and change trends in all inclusive articles.Publication bias was assessed by Begg's funnel plot and Egger's linear regression.The leave-one-out sensitivity was performed, in which the meta-analysis estimates were computed each time that a study was omitted.All P values are two-tailed with a significant level at 0.05.

The results of meta-analysis GSTM1 null genotype with OC risk
Using the fixed-effect model, metaanalysis showed no significant association between the GSTM1 null genotype and OC risk (OR=1.01,95% CI: 0.91 to 1.12, P =0.16) (Table 2), the forest plot was shown in Figure 2.

GSTT1 null genotype with OC risk
The meta-analysis showed no significance association between the GSTT1 and OC risk (OR=0.98,95% CI: 0.85 to 1.13, P=0.75) under the fixed-effect model (Table 3), the forest plot was shown in Figure 3.

Subgroup analysis and sensitivity analysis
Considering the small sample size of GSTP1 meta-analysis, we only performed subgroup analysis in GSTM1 and GSTT1 groups by the number of OC cases (<200 vs.≥200) and the source of controls (hospitalbased vs population-based studies).The subgroup analysis results showed no significant associations (Table 2 and Table  3).Meanwhile, we checked the pooled ORs in each study by excluding sequentially one article each time with sensitivity analysis, the pooled ORs did not change significantly.

The cumulative meta-analysis
Cumulative meta-analyses were also conducted via assortment of studies by     chronological order.The GSTM1 cumulative metaanalysis tended to gradually become more consistent after the 2008 studies were added.Similarly, the GSTT1 cumulative meta-analysis tended to be more consistent with the addition of studies added after 2006.The cumulative analysis showed that the results tended to remain stable over time.The results of cumulative meta-analysis for GSTM1 and GSTT1 were shown in (Supplementary Figure 4).

Potential publication bias
Publication bias was measured by Begg's and Egger's test.The outlines of funnel plots did not demonstrate any asymmetry in all comparison models by Begg's test (data not shown).Furthermore, we used Egger's test to supplement statistical evidence of funnel plot symmetry.The results were shown in Tables 2 and Table 3.

Discussion
In order to reflect new developments and produce more powerful estimation on the associations between the GSTs gene polymorphisms and OC susceptibility, an updated systematic meta-analysis with more available data was conducted in this study.Consistent with the conclusion in a previous published meta-analysis (its latest searching time was 2009) (Economopoulos et al., 2010), our results also show no significant associations between the GSTs gene polymorphisms and OC susceptibility.
Baxter et al reported an excess of GSTM1 null genotype in Caucasian OC patients when compared to controls in England (Whalen et al., 1998).Gates et al found that GSTT1 null genotype was associated with OC in a particular group of women using genital talc (Gates et al., 2008).Cristiane et al showed that the GSTM1 null, GSTT1 null and GSTP1 Ile/Ile genotypes, especially in combination, constitute significant inherited OC determinants in subjects from Southeastern Brazil, which indicated that GSTs may act synergistically in OC development (Lallas et al., 2000).Therefore, it is more likely that the combination of GST genes may increase the OC susceptibility.In addition, Spurdle et al reported that GSTM1 null genotype was associated with endometrioid/ DOI:http://dx.doi.org/10.7314/APJCP.2014.15.15.6131 GST Genetic Polymorphisms and Susceptibility to Ovarian clear cell invasive cancer risk, due to the fact that GSTT1 and GSTM1 were more common in both endometrioid and cell OC histological subtypes, which indicates that the GST null genotypes may specifically increase the risk of OC histological subtypes (Hengstler et al., 1998).
Associations of GSTM1, GSTT1, and GSTP1 genotype have not been observed in most studies.Considering the fact that the difference in sample sizes and control sources in certain studies could influence the stability of the conclusions, we performed subgroup analysis for GSTM1 and GSTT1 genetic polymorphism and OC risk, and the results still showed no significance differences.Then, we further performed sensitivity analysis and cumulative meta-analysis to observe the heterogeneity of these articles, the conclusion of non-significant association between GSTM1 genetic variants and OC risk didn't change.The cumulative analysis showed that the results tended to be stable over time.The funnel plots were all symmetric for GSTM1, GSTT1 and GSTP1, suggesting no publication bias (data not shown).Additionally, Egger's test didn't show significance of publication bias.
There are some limitations in this meta-analysis that are worth noting in explaining the results.1) Limited number and scale of studies may influence the stability of the conclusion and there were no data on the Asian and African population in the available studies, which may affect the conclusion; 2) The range of our literature searching only included English, which would result in the language selection bias; 3) The control subjects of three articles were from hospital-based population, which may differ from the population-based controls; 4) Available studies did not provide sufficient data about specific histological subtypes of OC, so we could not conduct a further analysis on the relationship between the specific OC histological subtypes and GSTs polymorphisms; This is significant, because Maliheh Arab et al found that histological subtypes affected OC patient's survival (Arab et al., 2009).5) Complex environmental factors also played critical roles in developing carcinoma, we could not evaluate the interactive between genes and environment in our meta-analysis.These may all be factors that could affect the results of this meta-analysis.
In summary, our meta-analysis revealed no association between the GSTs gene polymorphisms and OC susceptibility, but the roles of GSTM1 null, GSTT1 null and GSTP1 Ile105Val polymorphisms in combination for OC risk is still unclear.Perhaps the combination of GSTM1 null, GSTT1 null and GSTP1 Ile105Val polymorphisms may confer a greater risk to OC, future studies designed to investigate the combinations is a promising area.Besides, studies that examine associations between the specific histological types and OC susceptibility are also very important for future research (Sagae et al., 2002).In order to provide a more precise evaluation between the GSTs gene polymorphisms and OC susceptibility, well designed epidemiological and molecular biology studies should be performed in general populations especially in Asian and African.Marzieh Rohani-Rasaf et al demonstrated that OC had a positive correlation with socio-economic position in Iran (Rohani-Rasaf et al., 2013), therefore given the difference of inheritance and living habits in various ethnicities, future studies should also give balanced attention to different ethnicity groups (Hamajima et al., 2002).In addition, identification of microRNAs may be helpful for disclosing the molecular mechanisms of OC and exploring the new treatments of OC (Wan et al., 2012).

Figure 2 .
Figure 2. Association between GSTM1 Null Genotype and OC Risk Analyzed by Forest Plot

FIGURE 3 .Figure 3 .
FIGURE 3. Association between GSTT1 null genotype and OC risk analyzed by forest plotFigure 3. Association between GSTT1 Null Genotype and OC Risk Analyzed by Forest Plot

Figure 4 .
Figure 4.The Cumulative Meta-Analysis of GSTM1 and GSTT1.(A) the metatrend test of GSTM1 by public time, (B) the metatrend test of GSTT1 by public time

Flow Chart of Study Selection
, 10 articles documented GSTM1 (2578 cases and 3423 controls) (Sarhanis et al., FIGURE 1. Flow chart of study selection Figure 1.

Table 1 . Characteristics of the Studies Correlating with the Effects of GSTs Genetic Polymorphisms on OC Risks
FIGURE 2. Association between GSTM1 null genotype and OC risk analyzed by forest plot