Introduction

Bladder cancer (BC) is one of the most common urological malignancies in the worldwide, with an increasing incidence and death rate nowadays [1, 2]. An estimated 386,300 new cases of BC occurred worldwide in 2008, with 150,200 deaths annually [1]. The highest BC incidence rates are found in Western Europe and North America. In European countries, there were an estimated 0.14 million new cases of BC and 0.05 million deaths from these health care problems in 2008 [3]. Therefore, early identification of persons at risk and early detection of BC are the most appropriate means of prevention, and hence contribute to the improvement of the BC patient’s diagnosis and treatment.

In recent years, a large number of epidemiology studies have suggested that many genetic polymorphisms can affect the BC susceptibility [4–10]. And it is now commonly accepted that the cause of BC is a multi-factorial interaction of environmental triggers (e.g., exposure to certain chemicals, smoking, chronic urinary tract infections) and genetic susceptibility.

Glutathione S-transferases (GSTs) are members of a multi-gene family of isoenzymes expressed in almost all living organisms [11]. As the most important phase II metabolizing enzymes, GSTs catalyse the conjugation of potentially damaging chemical mutagens and protect against the products of oxidative stress [12, 13]. They are involved in the metabolism of many xenobiotics in mammals, including an array of environmental carcinogens and endogenously derived reactive oxygen species [12, 14]. Based on sequence similarities, human cytosolic GSTs superfamily have been grouped into at least 8 distinct classes, called GST α, μ, κ, π, σ, ω, θ, and ζ [15]. Functional polymorphism has been identified in the Glutathione S-transferase P1 (GSTP1) gene coding for GST-π. The GSTP1 A313G polymorphism may result in an amino acid variation of isoleucine/valine at codon 105 in the protein (Ile105Val). GSTP1 allelic variants may lead to increased organism highly susceptible to oxidative DNA damage and to the accumulation of DNA base adducts, which can allow tumor cells to acquire various other oncogenic genetic alterations in urinary bladder carcinogenesis.

Over the past few decades, a great number of studies were performed to clarify the true association between GSTP1 A313G polymorphism and BC risk, especially among Caucasians. However, previous case–control studies investigating the association have reported conflicting results. In order to investigate the real effect of GSTP1 polymorphism on the risk of developing BC, we conducted an updating meta-analysis from the available studies to better compare results between epidemiological studies.

Materials and methods

Literature search strategy

We did a systematic search in the following electronic databases: PubMed (1950 to August 2012), EMBASE (1950 to August 2012), and China National Knowledge Infrastructure (CNKI) (1979 to August 2012). The following key words were used: (“glutathione S-transferase” OR “GST” OR “GSTP1” OR “rs1695” OR “Ile105Val” OR “A313G”) AND (“bladder” OR “urinary” OR “urocyst” OR “urotheli*”) AND (“adenocarcinoma*” OR “carcinoma*” OR “cancer*” OR “tumour*” OR “tumor*” OR “neoplasm*”). No language restriction was used. The reference lists of the selected papers were screened by hand for potentially relevant new articles.

Furthermore, if more than one paper was published with identical author using the same case series, we selected the research with more sample size [14].

Inclusion and exclusion criteria

The criteria employed to select studies for this systematic review were as follows: (i) independent epidemiological studies (for humans only); (ii) a clear description of GSTP1 polymorphism in BC cases and controls. The exclusion criteria were: (i) not an original paper (e.g., review or letter etc.); (ii) duplicate publications; (iii) no control.

Data extraction

Two investigators (Ke Wu and Xianding Wang) independently extracted all the data from each study. Differences were resolved by a third investigator (Yiping Lu). The following data were extracted (please see Table 1): First author’s name, Publication year, Country, Design of study (hospital or population based case–control study), Majority race of study population, Number of cases and controls with different GSTP1 genotypes.

Table 1 Characteristics of primary studies in the meta-analysis
Full size table

Statistical analysis

Statistical analyses were conducted by use of STATA 11.0 (Stata-Corp LP, College Station, TX, USA) and Review Manager 5.1.6 (Cochrane Collaboration, Oxford, UK). A fixed- or random- effects model was used to calculate pooled effect estimates depending on statistical heterogeneity. The crude odds ratios (ORs) were pooled using the random-effects model (DerSimonian Laird method) when statistical heterogeneity was found(P < 0.05).

Also, subgroup analyses were performed on the basis of race, design of study and smoking behavior, and so on. Publication bias was assessed by visual inspection of funnel plots, the Begg’s rank correlation method and the Egger’s weighted regression method [16, 17]. In this study, P < 0.05 was considered statistically significant, and all statistical tests were two sided.

Results

Study characteristics

The literature search was updated on August 1st, 2012. The search terms resulted in 742 articles. At last, 20 studies (19 in English and 1 in Chinese) with 4,428 BC cases and 5,457 controls were identified (please see Table 1 and Fig. 1) [18–37].

Fig. 1
figure 1

Studies identification, inclusion, and exclusion

Full size image

Overall analysis

The pooled results based on all studies showed a statistically significant link between GSTP1 A313G polymorphism and BC risk (GG vs. AA +AG: OR = 1.50, 95 % CI = 1.13–2.00; AA vs. AG+GG: OR = 0.82, 95 % CI = 0.70–0.95) (Table 2, Fig. 2). Because the test for heterogeneity between eligible studies was significant (P < 0.001, I 2 = 64.4 %), the random-effects model was performed for the data analysis.

Table 2 Meta-analyses of the association between GSTP1 A313G polymorphism and the risk of bladder cancer
Full size table
Fig. 2
figure 2

Overall meta-analysis for GSPT1 A313G polymorphism (AA vs. AG+GG) and bladder cancer risk

Full size image

Furthermore, Begg’s rank correlation method and Egger’s weighted regression method were used to assess publication bias. Finally, we found that there was no evidence of publication bias in GSTP1 A313G polymorphism studies (P Begg = 0.09, P Egger = 0.15) (Figs. 3, 4).

Fig. 3
figure 3

Begg’s funnel plot of GSPT1 A313G polymorphism and bladder cancer risk (AA vs. AG+GG, P = 0.09)

Full size image
Fig. 4
figure 4

Egger’s publication bias plot of GSPT1 A313G polymorphism and bladder cancer risk (AA vs. AG+GG, P = 0.15)

Full size image

Ethnic origin (Asians and Caucasians) and control sources (hospital-based and population-based)

When stratifying by race, the combined ORs for GSTP1 A313G polymorphism (GG vs. AA+AG) were 1.59 (95 % CI = 1.01–2.51, P = 0.04) in the analysis among Asians, and 1.51 (95 % CI = 1.11–2.06, P = 0.01) in the analysis among Caucasians. Stratifying this meta-analysis by control sources, we also found a significant difference between GSTP1 genotype and BC susceptibility in studies with population-based controls. In hospital-based studies, GSTP1 A313G variants (AG or GG) showed a marked increase in BC risk with an OR of 1.26 (95 % CI = 1.03–1.53), compared to individuals carrying AA genotype, used as reference category.

Smoking status (ever-smokers and non-smokers)

Considering that smoking is a risk factor for BC, and that GST genes are involved in the metabolism of various carcinogens present in smoke [7], further analyses according to smoking status of subjects were performed. Only five studies provided the raw data on the relationship between smoking and BC risk. We found that smoking did not modify the association between the GSTP1 polymorphism and BC risk (AA vs. GG: OR = 0.9, 95 % CI = 0.53–1.53, P = 0.69) in ever-smokers (Table 3).

Table 3 Summary OR and 95 %CI of GSTP1 A313G polymorphism and bladder cancer susceptibility
Full size table

Combination of genotypes

Combination of the two high-risk genotypes (G allele of GSTP1 genotype and GSTM1 null or GSTT1 null) revealed that the risk increased up to 2.64 times (95 % CI = 1.90–3.65; P < 0.0001) for GSTP1 and GSTM1 and 2.39 times (95 % CI = 1.54–3.70, P < 0.0001) for GSTP1 and GSTT1 genotype.

Four studies reported the combination genotypes of GSTP1, GSTM1 and GSTT1 in subjects. We found that individuals with risk genotypes (null genotypes of GSTM1 and GSTT1 and the 313 AG/GG of GSTP1) had considerably increased BC susceptibility (OR = 6.64, 95 % CI = 3.63–12.16, P < 0.00001) compared with those who had non-risk genotypes (positive genotypes of GSTM1 and GSTT1 and 313 A/A genotype of GSTP1). All the results are presented in Tables 4, 5.

Table 4 Combination of double GST genotypes among bladder cancer patients and controls
Full size table
Table 5 Combination of triple GST genotypes among bladder cancer patients and controls
Full size table

Discussion

Nowadays, the exact mechanisms of bladder tumorigenesis remain unknown. There is a growing realization that the development of BC is caused by a complex interaction of both genetic and environmental factors. Procarcinogens are mainly metabolized by various metabolizing enzymes in the human body. Interindividual variations in the genetic and cellular mechanisms of detoxification of carcinogenic chemicals, such as sequence variations in genes coding for the GSTs family, might potentially confer different degrees of risk to BC [5].

Until recently, a number of studies on the polymorphisms of xenobiotic-metabolizing enzymes and BC risk have been reported, especially for GSTP1 [7, 8, 35, 38]. In 1997, Harries et al. [18] firstly reported the association between the GSTP1 A313G polymorphism and BC risk among individuals from the Edinburgh area. Following this first report, similar studies were conducted in different countries by other researchers. However, studies investigating the association have reported conflicting results. Moreover, most of these studies were based on relatively small sample sizes.

As a powerful statistical method, meta-analysis can help to summarize the effect size results from numerous independent epidemiology studies and to provide more reliable outcomes. In 2007, there has been only one meta-analysis that suggested that, as compared with GSTP1 Ile/Ile, the unadjusted summary OR for GSTP1 Ile/Val and Val/Val was 1.44 (95 % CI = 1.17–1.77) [39]. However, some limitations were found in the statistical data in this prior meta-analysis: (i) in Katoh et al. [40] study , the sample included 106 cases who were the patients with urothelial cancer (not just bladder cancer); (ii) more than one included study was performed by identical research team using the same case series [41–44]. The duplicated data of these studies should not be included in prior meta-analysis. On the other hand, another nine studies have investigated the association between GSTP1 polymorphism and BC susceptibility over the last nearly 6 years. As a result, an updated meta-analysis is needed.

We found that significant associations between GG genotype of GSTP1 and BC risk in all subjects (Asians and Caucasians), suggesting that carriers of homozygous variant in GSTP1 lack enzyme activity. However, there was no association between GSTP1 313 GG genotype and BC susceptibility in hospital-based case–control study (HCC). As for HCC, selection bias may not be avoidable, and the subjects may not be representative of the general population [35]. The data on hospital controls could provide relatively lower risk estimates if the diseases of the controls were associated with the gene variant being studied [9]. Therefore, further studies based on population design are necessary. In addition, differences between study designs were also reported in prior studies concerning GSTs genotypes at cancer risk [12].

When stratified according to gender, we found that a significant association between G allele of GSTP1 genotype (heterozygous or homozygous variant) and BC risk among male, but not female. The inconsistent findings may be due to the following two reasons: the genetic background of female is distinguished from that of male, and risk of the same kind of disease is obviously different [45]; there is only one research team that published the original data (90 cases and 77 controls) about GSTP1 AG+GG genotype and BC risk among female [43]. Because of the limited sample size and power, results from this stratified analysis should be considered with caution.

It has been known that smoking is one of the main independent risk factors for BC risk, accounting for half of the cases in male and nearly 35 % in female [46]. GSTs are involved in the metabolism of the multiple carcinogens contained in tobacco smoke, so subgroup analyses by smoking were performed. In our study, no statistically significant difference in genotype distribution of GSTP1 in ever-smokers was found. The possible reason we though is that the GSTs are a family of enzymes responsible for the detoxification of a wide range of chemical carcinogens, so even though only one gene of GST family linked a variant, the GSTs-activity is unlikely to have a significant down-effect on metabolic clearance. Moreover, other environmental factors such as diet, living habit and occupational exposure may affect this association.

Some studies also examined the combination effects of unfavorable GSTs. Our meta-analysis suggests that the GSTP1 polymorphism and its combination with GSTM1, and GSTT1 may be associated with BC risk. Therefore, gene–gene interactions might be primarily involved in the genetic susceptibility for BC, which could be explained by various substrates used by different GSTs inducing resulting in combined action [37, 47, 48]. We assume that the subjects possessing Val allele of GSTP1 and null allele of GSTM1 and GSTT1 have higher BC susceptibility mainly due to reduced detoxification of carcinogens [28].

There are some limitations in our study: First of all, relatively small sample size and significant heterogeneity were observed in some sub-analyses. Second, because of the lack of individual patient data, we could not perform an adjustment estimate. Third, because many environmental factors may affect the BC susceptibility, all our findings may be due to the context of the genetic background and interacting with multiple environmental factors. Finally, meta-analysis is just a statistical test that is subject to many methodological restrictions [8, 45, 49].

In conclusion, our study suggested that GSTP1polymorphism is associated with a high increase in the risk of BC. Also, the combination of three risk GSTs genotypes is strong predisposing risk factor for BC. No significant gene–smoking interaction association was found for the GSTP1 variant in the risk of BC in ever-smokers. Because heterogeneity among the included studies was extreme, the results of this meta-analysis may be confirmed by additional well-designed, high-quality case control studies with larger populations.