Polymorphic deletions of the GSTT1 and GSTM1 genes and susceptibility to bladder cancer

INTRODUCTION

Urinary bladder tumours are the second most common cancer of the urogenital tract. The incidence varies worldwide but is higher in industrialized countries. In Spain, the raw incidence rate was in the range 30.0–55.1 per 100 000 inhabitants in 2002, according to geographical area [1].

The interaction between genetic susceptibility and environmental factors plays an important role in the development of bladder cancer. These factors classically include exposure to various chemicals, smoking, long-term use of analgesics, treatment with genotoxic chemotherapy agents, and the presence of irritating factors (e.g. chronic urinary infection and bladder stones). Bladder cancer has also been related to sweetener usage and coffee intake, although less consistently [2]. In short, bladder cancer is among those most closely-related to chemical exposure, which could explain 10–30% of cases [3,4].

Many toxic substances related to occupational exposure and smoking have been identified as urothelial carcinogens and most are aromatic amines: aniline dyes, 2-naphthylamine, 4-aminobiphenyl, 4-nitrobiphenyl, benzidine, 2-amine-1-naphthol, combustion gases and carbon soot, chlorinated aliphatic hydrocarbons, and certain aldehydes such as acrolein used in chemical dyes and in the rubber and textile industries [5]. Other compounds, such as polycyclic aromatic hydrocarbons from tobacco smoke, can also cause this type of tumour [6,7] and, therefore, smoking is a important risk factor in bladder cancer. The risk for smokers is fourfold compared to non-smokers, although the risk also correlates with the number of cigarettes, the duration of smoking habit and the degree of smoke inhalation [8,9].

The glutathione S-transferases (GST) represent a large family of enzymes that regulate the conversion of toxic compounds to hydrophilic metabolites excretable by glutathione conjugation. GST are expressed in many tissues and are a defence against cytotoxic and potentially carcinogenic chemicals [10]. Their level of activity depends on genetic differences and may influence individual susceptibility to cancer, particularly in cases of higher exposure to toxic and carcinogenic agents [11,12].

GST are part of a complex network of detoxifying enzymes that include cytochrome P450, which catalyses the oxidation of these compounds [13], and GST, which convert them into excretable water- soluble compounds [14]. The intermediate metabolites produced by cytochrome P450 are possibly genuine carcinogens able to react with DNA [15]. GST would act as protectors, eliminating these metabolites.

GST include four classes of enzymes: α (GSTA), µ (GSTM), θ (GSTT) and π (GSTP), each containing one or more homodimer or heterodimer forms. The main GST genes described as polymorphic in humans are GSTT1 and GSTM1[16,17]. Deletions that cause null alleles have been identified in both cases. In white individuals, the prevalence of null alleles is ≈50% in the GSTMI gene and 20% in GSTT1[10,18]. Homozygous carriers of null alleles of one or both genes are unable to produce the detoxifying protein and metabolize certain toxic compounds adequately; therefore, these individuals are at an increased risk of developing cancer.

The discrepancy between results regarding the association of common genetic polymorphisms and complex diseases such as cancer has raised skepticism in this area of research. Although the evidence generally supports the implication of GSTM1 and GSTT1 polymorphisms in bladder cancer, there is still some debate, with some studies in favour [19,20] and others against [21,22]. To enhance our understanding, the present study aimed to estimate the prevalence of null genotypes in the most polymorphic GST genes in the Spanish population and determine the relationship with bladder cancer. In addition, whether the relationship is influenced by the presence of known carcinogenic factors (e.g. smoking and exposure to occupational factors) was investigated. Lastly, the usefulness of these polymorphisms as a prognostic factor was determined.

PATIENTS AND METHODS

A cross-sectional, analytical study was conducted in two patient groups: a control group of patients who underwent surgery for benign urological (77.40%), general surgical (10.57%), traumatic (6.25%) and ophthalmological (5.76%) conditions, with a confirmed absence of neoplastic diseases or family history of cancer, and a study group comprised of patients diagnosed with urothelial bladder carcinoma. In this paired case–control study, the sample size was calculated based on a power of 80%, a significance level of 5% and a minimum detectable prevalence odds ratio (OR) of 1.6, which resulted in a requirement of 208 patients in each group. As subjects were recruited, the controls were selected and paired according to gender, age, smoking and occupational risk exposure. Subjects were recruited consecutively over an 18-month period (2007–2008).

The study was approved by the ethics committee for the Albacete health district, and all subjects gave written informed consent to participate.

A questionnaire was used to compile the following variables: (i) sociodemographic data: age, gender and place of residence; (ii) smoking habit: subjects were classified as non-smokers (individuals who had never smoked or who reported smoking fewer than 100 cigarettes over their lifetime) or smokers (ex-smokers as well as occasional or regular smokers). Among smokers, the number of cigarettes per day and the length of smoking habit were assessed; (iii) risk occupation: occupations considered to present the greatest potential exposure to carcinogenic substances related to bladder cancer were recorded using the International Standard Classification of All Industrial Activities [23].

The medical histories were reviewed to collect data on: tumour pathology, histological type, tumour stage (TNM), nuclear grade according to the WHO classification, number and size of tumours, primary or recurrent tumour, number of recurrences, disease-free period and synchronous or metachronous association with other urothelial tumours.

To analyse GSTT1 and GSTM1 polymorphisms, blood samples were drawn, then frozen and stored at −80 °C until processing. After DNA extraction, multiplex PCR amplification was performed. The primers used are available upon request.

The GSTT1 gene was analysed using 35 amplification cycles with a final reaction volume of 25 µL containing 100 ng of DNA, 200 mm of each desoxyribonucleotide (dNTP), 200 nm of each primer and 0.4 U of Taq polymerase. In the case of the GSTM1 gene, 30 amplification cycles were performed with a final volume of 25 µL containing the same quantities of DNA, dNTP and Taq polymerase described earlier, along with 400 nm of each primer.

In both cases, the amplification products were analysed by agarose gel electrophoresis. GSTT1 amplification yielded a 1460-bp product from a deleted allele and a 466-bp product from a normal allele. GSTM1 amplification produced a 237-bp fragment in the case of a normal allele but could not be amplified when a homozygous deletion was present.

A descriptive analysis of all the study variables was carried out. Bivariate analysis was performed with the Pearson chi-squared test or Student’s t-test, depending on the characteristics of the variables studied. Non-parametric tests were used when the conditions for these tests were not met. Attributable fractions in exposed individuals were also calculated. It was likewise determined whether allele frequencies were found in Hardy–Weinberg equilibrium using SNP & Variations Suite 7 software (Golden Helix®; Bozeman, MT, USA).

To detect any confounding variables or effect modifiers of the GST null genotype on bladder cancer, a stratified analysis was carried out with internal standardization for smoking and occupational exposure. A multivariate logistic regression model was constructed using the presence of bladder tumour as the dependent variable and stepwise inclusion of variables. Because the study was cross-sectional, the OR comprise prevalence ORs.

RESULTS

The sociodemographic characteristics and exposure factors of the study subjects are listed in Table 1. On average, control subjects were younger (61.7 years) than patients with bladder tumour (73.4 years). No differences in habitat or risk occupation were observed between the two groups. Smoking was more widespread among patients with bladder tumour (71.6%) than controls (47.2%). The percentage of smokers of more than 20 cigarettes a day was also higher in the patient group, as was the mean number of years they had been smoking (Table 1).

The most common histological type was transitional cell carcinoma (n= 198; 95.2%). The infiltration stage was usually superficial (pTa and pT1) (n= 142; 68.7%), and G1 or G2 tumours were more common (n= 122; 58.6%). All other features are listed in Table 2.

GSTM1 genotypes were determined in 201 patients and 193 controls and GSTT1 genotypes in 190 patients and 163 controls. The entire estimated sample was not completed as a result of technical problems concerning DNA extraction. The GSTM1 and GSTT1 genotypic distributions and the respective ORs are shown in Table 3. Significant differences were only found in the distribution of the GSTM1 null genotype, which was higher in patients (54.2%) compared to controls (40.4%) (P= 0.008). This gives a crude OR of 1.7, yielding an OR of 3.5 (P= 0.002) when both GST genes were considered together.

The attributable fraction in exposed individuals for the presence of GSTM1 null genotypes showed that 23.7% (95% CI, 7.3–37.2) of bladder cancer cases in our patients with null genotype in GSTM1 were precisely a result of the genotype. Regarding the genotypes of both GST genes, the attributable fraction in exposed individuals showed that 35.1% (95% CI, 20.7–46.2) of bladder cancers in our patients with both null genotypes are precisely a result of null genotypes. For the GSTT1 gene analysed, all confidence intervals were non-significant.

No significant differences were observed between GSTM1 or GSTT1 genotype distribution according to age, gender, habitat or risk occupation in either the control subjects or the study group (data not shown). However, GSTT1 null genotype was more common in non-smokers than smokers (32.7% vs 18.1%, P= 0.027), and there were no differences in other smoking-related characteristics.

All variables related to development of bladder tumour and the respective first- order interactions were introduced in the multivariate logistic regression analysis. GSTM1 null genotype, smoking and age remained in the model, whereas GSTT1 null genotype did not. The sensitivity of the model was 75.1%, the specificity was 64.2% and the percentage of properly classified individuals was 69.8%. The variability of the model was 28.6% (chi-squared = 94.96; P < 0.0001) (Table 4).

In a second model, the presence of both GST null genotypes was introduced. The ORs for the variables were practically identical in this model, whereas the OR for the presence of both null genotypes was twice as high: 4.2 (95% CI, 1.7–9.9). The sensitivity and specificity of this model and the percentage of correctly classified individuals were higher (77.5%, 66.4% and 71.0%, respectively). Model variability also increased slightly: 29.5% (chi-squared = 86.73; P < 0.0001) (Table 4).

The mean ± SD age at bladder cancer diagnosis was younger in patients with GSTM1 null genotype compared to those with no allele deletion: 70.6 ± 11.13 years vs 74.2 ± 10.2 years (P= 0.049). When stratified by gender, women with GSTM1 null genotype presented a higher risk of 4.8 (95% CI, 1.2–19.3) (P < 0.0001) with respect to the crude OR (Table 5). No differences were observed in men or in the stratification according to habitat (data not shown).

In the stratification according to smoking habit, the OR of bladder tumour in smokers with both null genotypes was 4.7 (95% CI, 1.1–22.4); therefore, smoking can be considered to influence the effect of being a carrier of both null genotypes on the development of bladder tumours. Table 6 shows the entire stratified analysis.

There was a greater incidence of GSTT1 null genotypes in patients with G3 tumours: 31.2% vs 15.9% in G1–G2 (P= 0.013). When the genotypes of the two GST genes were considered together, a higher percentage of both null genotypes were found in patients with G3 tumours: 25.3% vs 8.9% in G1–G2 (P= 0.002). There was no relationship between nuclear grade and GSTM1 genotypes.

A separate analysis of the GSTM1 and GSTT1 genotypes found no significant differences related to pT stage. By contrast, when both genes were considered together, patients with infiltrating tumours were more likely to have both null genotypes: 23.7% vs 11.7% in superficial types (P= 0.035). There were no differences related to lymph node involvement or presence of metastasis.

DISCUSSION

Although other genes are related to the GST family, the GSTA and GSTP genes were not analysed in the present study because there is little evidence to suggest their involvement in bladder cancer [24].

The data obtained in the present study show a prevalence of null genotypes in 47.5% of our entire study population for the GSTM1 gene and 19.0% for the GSTT1 gene, which is consistent with the values for the white population [10], which is predominant in our area. The prevalence of null genotypes in bladder tumour patients was higher than that of the controls for both the GSTM1 and GSTT1 genes (OR, 1.7 for GSTM1 and non-significant OR for GSTT1). When both null genotypes are present, the risk of developing bladder carcinoma is twice as high than when only one is present. Studies to date show that 31% (95% CI, 20–46) of bladder tumours among the white population could be caused by polymorphisms in GSTM1 and the NAT2 gene related to slow N-acetylation activity [24,25]. In this regard, the present study estimated that 23.7% of bladder cancer cases could have occurred because GSTM1 null genotype was present and 35.1% because both null genotypes were present. On the basis of the above considerations, these figures should be taken with caution because very long-term longitudinal studies would be needed to confirm the data. However, similar to most neoplastic diseases, bladder cancer is a complex condition possibly involving multiple genes. Genes that may play a role in bladder cancer include those related to detoxifying enzyme activity, in particular NAT2[24,26–29], and those of the cytochrome P-450 family, in particular, CYP1B1[29], CYP2A6[26] and CYP1A1[13]; as well as, although not so clearly, genes such as MTHFR[30], NQO1[30] and SULT1A1[29].

This increased risk of neoplasm in subjects with altered GST enzyme function has been seen in other types of tumours at other sites and in other urinary tumours, particularly prostate adenocarcinoma and less often in renal cancer [10,12]. The present study found more evidence for involvement in bladder tumour development in the case of GSTM1 vs GSTT1 null genotype, as also supported in the literature [24]. These differences could result from the different substrates for the action of these enzymes [12]: GSTM1 mainly acts on substances derived from polycyclic aromatic hydrocarbons (e.g. benzopyran), whereas GSTT1 acts on small reactive hydrocarbons (e.g. ethylene oxide) [31]. Synergism of the detoxifying action of both genes appears evident, given that the risk of bladder cancer is twofold higher in the present study when both genotypes are combined. This synergism, also reported in the literature [26,32], is also related to the combined action with other detoxifying genes, mainly NAT2, and especially among smokers [29].

Smoking is a major risk factor for the development of bladder cancer. In the present study, the univariate analysis showed a higher risk in smokers that was even greater in subjects who were heavy smokers. After stratification by the number of cigarettes or the duration of smoking habit, the crude ORs of GSTM1 and GSTT1 did not change. Nevertheless, the OR was almost fivefold higher in smokers with both null genotypes, which clearly shows the repercussion of this association on bladder cancer. However, the multivariate analysis showed that both smoking and the presence of GSTM1 null genotype or both null genotypes were independent risk factors for bladder cancer and that, individually, smoking appears to carry the highest risk. The literature confirms this conclusion [26,27,30,33] and does not provide solid evidence to confirm a higher risk of bladder cancer in smokers with GSTM1 null genotype. It has been suggested that GSTM1 can also act by mechanisms other than detoxification of the polycyclic aromatic hydrocarbons from tobacco smoke (e.g. by protecting from oxidation damage caused by metabolism of reactive oxygen species) [24].

No clear evidence was found showing that occupations with carcinogen exposure influenced bladder cancer risk, either directly or in association with GST genotypes. The results of the present study suggest a higher risk of bladder cancer in exposed subjects with GSTT1 null genotypes, although the difference is not statistically significant. This may result from the approach used to classify the occupations; carcinogen contact may not always be the same in an occupation and can vary according to factors not considered in the present study (e.g. exposure time, severity or strictness shown with preventive measures). However, studies that analyse and quantify contact with carcinogenic substances do find a higher risk of bladder cancer in subjects with GSTM1 as well as GSTT1 and GSTP1 null genotypes, multiplying the risk when various null genotypes are combined [28,29].

Patients with bladder tumour and GSTM1 null genotype were younger than those without deletion of this gene. This suggests that subjects with a loss of function in this gene, who are more susceptible to non-metabolizing carcinogenic agents, will present bladder tumours at earlier ages. Similarly, stratification by gender showed that women had a higher risk in the case of GSTM1 null genotype, which was not observed in men. The data suggesting an effect modification of GSTM1 on the risk of bladder cancer according to gender have been attributed to genotype distribution differences between men and women in study and control populations [24,34]. In the present study, however, the GSTM1 and GSTT1 genotype distribution according to gender was similar in both groups.

The presence of null genotypes in both GST genes was related to a higher incidence of more highly undifferentiated and infiltrating tumours, which makes it a prognostic factor. Considering that both GSTM1 and GSTT1 are expressed in normal bladder mucosa [35] and that there is a tendency toward increased expression during tumour development [ 36] that has even been related to resistance to certain chemotherapeutic agents [37], immunohistochemical analysis of these proteins could have prognostic value in clinical practice. However, data obtained to date on the usefulness of this marker are contradictory [26,32–34,38,39].

The main limitation of the present study is common to many related studies published to date and often occurs in cross-sectional studies because it relates to a possible incidence–prevalence bias and temporal ambiguity. Standardization strategies carried out during data collection in both groups ensure that screening bias, a common situation in case–control studies, is minimized. This is also supported by the observation that the allele frequencies were in Hardy–Weinberg balance.

In conclusion, the present study shows a greater risk of bladder cancer in individuals with the GSTM1 null genotype, particularly women. This relationship was less evident with the GSTT1 null genotype. The presence of the null genotype in both genes appeared to be synergistic, particularly among smokers, and to increase the predisposition to more aggressive tumours. Nevertheless, despite these data and other reported evidence, the role of GSTM1 and GSTT1 polymorphisms in the predisposition to bladder cancer should be viewed with caution as a result of the multifactorial genetic origin of this condition and the need for long-term longitudinal studies to confirm these results.

ACKNOWLEDGEMENTS

This study was funded by a grant (♯06068-00) from the Ministry of Health, Junta de Comunidades de Castilla-La Mancha, Spain.

CONFLICT OF INTEREST

None declared.