Tobacco smoking, GSTP1 polymorphism, and bladder carcinoma




Although cigarette smoking is considered a major risk factor for bladder carcinoma, little is known about the interaction between metabolic genes such as glutathione-S-transferase P1 and tobacco smoking in this process. GSTP1 may play a role in detoxification of tobacco-related carcinogens.


In this case–control study of 145 cases with bladder carcinoma (male:female = 7.5:1) and 170 noncancer controls (male:female = 3.7:1), the relation between genetic polymorphisms of GSTP1 and susceptibility to bladder carcinoma was investigated and the gene–environment interaction between tobacco smoking and GSTP1 polymorphism was evaluated. Epidemiological data were collected for all cases and controls by a standard questionnaire. Polymorphisms of GSTP1 were measured by polymerase chain reaction–restriction fragment length polymorphism. The logistic regression model in SAS was used to estimate odds ratios (ORs) and 95% confidence intervals (95% CIs).


Cigarette smoking was confirmed as a risk factor of bladder carcinoma with an OR of 3.1 (95% CI: 1.7–5.9) after controlling for potential confounding factors. The OR for pack-years of smoking as a continuous variable was 2.4 (95% CI: 2.0–2.8). The ORs were 7.6 (95% CI: 1.18–49.51) for isoleucine/valine (Ile/Val) and 6.5 (95% CI: 1.01–41.56) for Ile/Ile when the homozygous Val/Val was considered as comparison group after adjusting for age, gender, race, and education. The adjusted OR for interaction between smoking and the GSTP1 (any Ile genotype) was 11.42 (95% CI: 0.53–248.15).


The results indicate that the Ile 105 allele is associated with an increased risk of bladder carcinoma and suggest that individuals who smoke and possess the Ile allele might be at increased risk for bladder carcinoma. Cancer 2005. © 2005 American Cancer Society.

Epidemiological studies suggest that the risk of bladder carcinoma is strongly associated with tobacco smoking.1, 2 In the U.S. and many other Western countries, tobacco smoking is thought to account for almost one-half of bladder carcinoma cases in men. Among heavy smokers, the risk of bladder carcinoma in women is approximately twice that in men.3, 4 There are approximately 3800 chemicals in tobacco; the most important carcinogens include benzo[a]pyrene, polycyclic aromatic hydrocarbons (PAHs), aromatic amines (2-naphthylamine[2-NA] and 4-ABP), and N-nitroso compounds.5 However, the mechanism by which smoking increases bladder carcinoma is unclear. Metabolic enzymes are considered to be involved in either the activation or detoxification of chemical carcinogens in tobacco smoke, which have been suggested as possible genetic susceptibility factors for smoking-related carcinomas.6–9

The glutathione S-transferase supergene family consists of four gene subfamilies (GSTA, GSTM, GSTT, GSTP) that play a central role in the inactivation of toxic and carcinogenic electrophiles.10, 11 Their roles usually are considered to be the detoxification and protection of cellular macromolecules from damage induced by carcinogenic agents, which include benzo[a]pyrene (a major cigarette carcinogen) and PAHs.5, 12, 13 The GSTP1 family consists of a single gene (GSTP1), which has been shown to be overexpressed in different human epithelial tissue and tumors tissues, including in bladder carcinoma.14–17 In a healthy population, the proportions of GSTP1 gene polymorphism have been reported as, respectively, 51%, 43%, and 6% for homozygous Ile/Ile, heterozygous isoleucine/valine (Ile/Val), and homozygosity for the variant Val/Val.8

Harries et al.18 reported that there are two alleles, GSTP1a and GSTP1b, at the GSTP1 locus. GSTP1a was initially thought to be the more active enzyme. Recently, Sundberg et al.19 found GSTP1a to possess up to five-fold greater enzymatic activity to some PAHs in GSTP1 Ile/Val or Ile/Ile. Several studies have been performed to explore the associations between the various GSTP1 polymorphisms with bladder carcinoma; however, the results have been inconsistent.18, 20, 21 In some studies, the GSTP1 Ile/Ile genotype was found to be associated with increased risk of bladder and testicular carcinomas. Some studies have suggested that there is a higher susceptibility for carcinoma in individuals carrying the Val/Val genotype.22, 23 The previous studies have not shown a clear pattern of GSTP1 polymorphism and the risk of bladder carcinoma. Little is known about the interaction between metabolic genes such as GSTP1 and tobacco smoking on bladder carcinoma risk. In this study, we investigated the GSTP1 polymorphism as both a potential risk factor for bladder carcinoma and as a potential modifier of the association between tobacco smoking and bladder carcinoma in a case–control study.


Study Population

A hospital-based case–control study was conducted at Memorial Sloan-Kettering Cancer Center (MSKCC). Data included questionnaire data and blood and tissue samples for assaying molecular markers from April 1, 1994, to June 30, 1997. The study was approved by the Institutional Research Board on Human Subjects of MSKCC and study participants were asked to sign an informed consent form.


Eligible cases were all patients diagnosed at MSKCC from April 1, 1994, to June 30, 1997, with pathologically confirmed bladder carcinoma. Cases were either newly diagnosed or undergoing surgical cystectomy. We approached 233 consecutive patients with bladder carcinoma. Among these, 145 patients had bladder carcinoma specimens for isolation of high-quality DNA for the proposed assays. Patients consented in writing to participate in the study. Using a standard questionnaire, a research nurse interviewed patients when admitted in the medical/surgical ward, which took approximately 1 hour. Pathology reports were obtained after surgery or upon discharge. All cases with pathologically confirmed diagnoses of bladder and available tissue samples were included in the study.


Eligible controls were healthy and cancer-free males and females recruited from the blood bank at MSKCC. They were approached and interviewed in the same manner as cases and met the same criteria, except for absence of a cancer history or diagnosis. The nurse interviewer approached blood donors, explained the study, and asked them to read the study description and sign the informed consent form if they agreed to participate. Controls also donated a 20 mL blood sample. A total of 204 participants were approached, and consented to be interviewed for the study. The DNA samples were obtained from 170 controls.

Epidemiologic Data Collection

A nurse interviewer approached and interviewed participants using a standardized questionnaire and asked participants to donate blood and tissue samples for biomarker measurements. By using a detailed questionnaire, we collected the following: 1) demographic information: name, gender, race, birth date and birthplace, marital status, education, address, telephone number, etc.; 2) occupational titles held and occupational and environmental exposures; 3) personal habits summary: cigarette smoking, passive smoking, alcohol consumption. The time period for alcohol use was the usual alcohol intake 1 year before the cancer diagnosis for cases and a year before the interview for controls. An epidemiological database was established and maintained to store patient data. Pathologic and laboratory data were collected, recorded, and linked to the epidemiological database. Medical charts and pathology reports were examined to ensure that the controls did not have a prior history of carcinoma.

Blood and Tissue Sample Collection and Storage

Twenty milliliters of blood sample per person were collected from cases and controls who were willing to donate their blood samples for laboratory assays. Bladder carcinoma tissues from cystectomy were collected and stored in a −70 °C freezer. Blood and tissue samples were packed in Styrofoam containers with a large amount of dry ice and shipped from MSKCC to UCLA in 1997. All blood and tissue specimens were stored in a −70 °C freezer at the Molecular Epidemiology Laboratory of the UCLA Jonsson Comprehensive Cancer Center.

DNA Extraction

DNA was isolated from tissue and peripheral blood samples (1–2 mL) using the phenol-chloroform method, which was modified according to Cao et al.24 Briefly, to digest the sample the pellet was resuspended in 200 μL of digestion buffer (100 mM Tris.Cl [pH 8.0], 5 mM EDTA [pH 8.0], 1% SDS) containing freshly thawed proteinase K (500 μg/mL). The sample was then incubated at 55 °C in a water bath overnight. After the incubation, DNA was extracted with an equal volume of Tris-saturated phenol-chloroform-isoamylalcohol solution (25:24:1) and precipitated with two volumes of ice-cold absolute ethanol. The sample was placed in the freezer (−20 °C) for at least 1 hour and then centrifuged at 14,000 rpm for 20 minutes. The supernatant was carefully removed without touching the pellet or the area where the pellet was expected to be. The pellet was washed by adding 1 mL of 70% ethanol and then centrifuged for 15 minutes at 14,000 rpm. The supernatant was carefully removed, then washed again with absolute ethanol. The sample was centrifuged again for 15 minutes at 14,000 rpm. Finally, the DNA was resuspended in 100 μL of distilled water and stored at 4 °C or −20 °C.

GSTP1 Restriction Fragment Length Polymorphism Analysis

Genetic polymorphism at the GSTP1 locus was detected by Alw261 digestion of an amplified 176 basepair (bp) region of genomic DNA flanking the exon 5 fragment.18 Briefly, 100 ng of the DNA sample was amplified using 0.2 μM primers 5′-ACC CCA GGG CTC TAT GGG AA-3′ and 5′-TGA GGG CAC AAG AAG CCC CT-3′, 20 μM dNTPs in a total volume of 20 μl, 1 U Taq DNA polymerase (Promega, Madison, WI), 1.5 mM MgCl2. Thermal cycling was carried out under the following conditions: initial denaturation at 95 °C for 5 minutes; subsequently, 35 cycles of 95 °C for 30 seconds, 55 °C for 30 seconds, and 72 °C for 30 seconds. A final polymerization step of 72 °C for 5 minutes was carried out to complete the elongation processes. The polymerase chain reaction (PCR) product of 17.5 μL was then digested with 5 U Alw261 (Promega) in a total volume of 20 μL and the products separated on a 3.5% agarose gel.

Statistical Analysis

All analysis was performed using SAS 8.0 software (SAS Institute, Cary, NC). Age and education were analyzed as continuous variables. A history of never smoking (‘never smoked’) was defined as fewer than 100 cigarettes in a lifetime, and ‘ever smoked’ was defined as ≥100 cigarettes in a lifetime. A pack was considered to be 20 cigarettes and a pack-year is the equivalent of smoking 20 cigarettes a day for a year. A history of alcohol drinking (‘ever drink’) was defined as ≥20 drinks in a lifetime, and ‘never drink’ was fewer than 20 drinks in a lifetime. Years of alcohol consumption were categorized as follows: no, 1–20, 20–40, ≥40 years.

The relations between bladder carcinoma and putative risk factors were estimated with odds ratios (ORs) and their 95% confidence interval (95% CI), derived from logistic regression analysis. On the basis of the distributions of variables and prior knowledge of the risk factors for bladder carcinoma, age, gender, race, and education were adjusted for in the logistic regression model. We used the quartile distribution for four categories and median for binary variables as cutoff points. Crude OR was calculated for all of the independent variables. Dummy variables were used to estimate the OR for each category of exposure. A more than additive and multiplicative interaction were indicated, respectively, when: additive interaction: OR11>OR10+OR01-1; multiplicative interaction: OR11>OR10*OR01. OR11 represents the OR for combined effect when both factors were present, OR10 represents the OR when only factor 1 was present, OR01 represents the OR when only factor 2 was present. The departures from multiplicative effects were assessed by including main effect variables and their product term in the logistic regression model, when adjusting for potential confounding factors.

Logistic regression was used to estimate ORs and their 95% CIs to investigate the relation between genetic polymorphisms of GSTP1 and susceptibility to bladder carcinoma, and when evaluating gene–environment interaction between tobacco smoking and GSTP1 polymorphism. The GSTP1 genotyping categorized subjects into three distinct groups: Ile/Ile, Ile/Val, and Val/Val. The homozygous Val/Val was designated as the reference category. Two types of analysis were performed: first, by comparing genotypes with at least one Ile allele (Ile/Val or Ile/Ile) to the referent (Val/Val), and second by comparing Ile/Ile or Ile/Val to the referent. In addition to the analysis of interaction between GSTP1 and smoking, a stratified analysis by smoking status was performed.


Table 1 shows the distribution of age, gender, race, education, drinking, type of drinking, smoking, pack years, and year of smoking cessation in subjects with and without tissue or blood samples. The presentation of these distributions was to evaluate possible confounding factors and to justify the procedure for controlling these variables in the multivariate analysis. Tissue or blood samples were available for 145 of 233 (62.23%) cases and for 170 of 204 (83.33%) controls. We also presented the selected variable for those who had tissue or blood samples and those who did not to assess whether the individuals with tissue or blood samples were representative of the original study population. This attempt was to show whether there was possible selection bias due to missing samples that may threaten the validity of the study. For cases with bladder and for controls, except gender, no obvious differences were observed between those with and without tissue or blood samples for variables listed in Table 1. For cases with bladder carcinoma, patients with tissue were more often male than those without tissue samples.

Table 1. Distribution of Selected Variables in Bladder Carcinoma Cases and Controls by Tissue or Blood Sample Availability
 Cases, n (%)Controls, n (%)
Age groups (yrs)    
≤5013 (14.8)16 (11.1)25 (73.5)122 (71.8)
51-6012 (13.6)26 (17.9)4 (11.8)30 (17.7)
6063 (71.6)103 (71.0)5 (14.7)18 (10.6)
χ2, P1.2366, 0.5389 1.0266, 0.5985 
Male64 (72.7)128 (88.3)21 (61.8)134 (78.8)
Female24 (27.3)17 (11.7)13 (38.2)36 (21.2)
χ2, P9.1307, 0.0025 4.5178, 0.0335 
Caucasian79 (92.9)133 (92.4)31 (96.9)164 (97.0)
Other6 (7.1)11 (7.6)1 (3.1)5 (3.0)
χ2, P0.0262, 0.8715 NA, 0.9595 
Education (Yrs)    
≤1238 (43.2)54 (37.2)4 (11.8)30 (17.7)
13-1634 (38.6)68 (46.9)14 (41.2)82 (48.2)
>1616 (18.2)23 (15.9)16 (47.0)58 (34.1)
χ2, P1.5191, 0.4679 2.1963, 0.3335 
Alcohol Drinking    
Yes68 (82.9)124 (87.9)30 (90.9)135 (80.4)
No14 (17.1)17 (12.1)3 (9.1)33 (19.6)
χ2, P1.0901, 0.2964 2.0888, 0.1484 
Type of drinking    
non-drinking14 (18.4)17 (13.0)3 (9.1)33 (19.8)
Beer5 (6.6)11 (8.4)2 (6.1)21 (12.6)
Wine8 (10.5)10 (7.6)5 (15.1)21 (12.6)
Liquor6 (7.9)11 (8.4)2 (6.1)7 (4.2)
Mix43 (56.6)82 (62.6)21 (63.6)85 (50.9)
χ2, P1.9234, 0.7498 3.9577, 0.4118 
Yes72 (83.7)119 (82.6)18 (54.6)74 (44.3)
No14 (16.3)25 (17.4)15 (45.4)93 (55.7)
χ2, P0.0448, 0.8324 1.1618, 0.2811 
Pack yrs    
014 (18.4)25 (18.3)15 (50.0)93 (58.5)
0 to <2012 (15.8)18 (13.1)9 (30.0)38 (23.9)
20 to <406 (7.9)26 (18.9)4 (13.3)15 (9.4)
40 to <6014 (18.4)20 (14.6)2 (6.9)7 (4.4)
≥6030 (39.5)48 (35.0)0 (0.0)6 (3.8)
χ2, P4.9519, 0.2923 2.4816, 0.6479 
Years of quit smoking    
014 (21.9)25 (21.9)15 (57.7)93 (65.9)
1 to <1014 (21.9)33 (29.0)3 (11.5)12 (8.5)
10 to <2014 (21.9)18 (15.8)6 (23.1)17 (12.1)
≥2022 (34.4)38 (33.3)2 (7.7)19 (13.5)
χ2, P1.6340, 0.6517 2.9755, 0.3954 

Table 2 shows the general characteristics and the association of possible risk factors for bladder carcinoma according to case and control status. The bladder carcinomas were more concentrated among the older age groups, compared with the controls. The >60-year age group was the largest group among the total cases while the ≤50-year age group was the largest group for the controls. With regard to gender and race, there were more men and Caucasians than women and others (white Hispanic 8, black non-Hispanic 8, black Hispanic 1, Asian/Pacific Islander 2, Native American 2, other 2). For education, a higher percentage of cases were in the lower level of education group in comparison to the controls. Low education level was strongly associated with an increased risk of bladder carcinoma.

Table 2. Distribution of Demographic Factors in Cases and Controls and Corresponding Odds Ratios and 95% Confidence Intervals
VariableCases n (%)Controls n (%)Crude OR (95% CI)Adjusted ORa (95% CI)
  • a

    Adjusted for age (continuous), gender (M/F), race, and education (continuous).

Age group    
≤5029 (12.5)147 (72.0)1.01.0
51-6038 (16.3)34 (16.7)5.7 (3.08-10.43)6.3 (3.26-12.00)
60166 (71.2)23 (11.3)36.6 (20.27-66.04)39.5 (20.98-74.47)
Trend  <0.0001<0.0001
Gender   1.0
Female41 (17.6)49 (24.0)1.0 
Male192 (82.4)155 (76.0)1.5 (0.93-2.36)1.7 (0.84-3.43)
Other17 (7.1)6 (3.0)1.01.0
Caucasian212 (92.9)195 (97.0)0.4 (0.15-0.99)0.2 (0.04-0.59)
Education (Yrs)    
≤1292 (39.5)34 (16.7)1.01.0
13-16102 (43.8)96 (47.0)0.4 (0.24-0.64)0.8 (0.44-1.65)
>1639 (16.7)74 (36.3)0.2 (0.11-0.34)0.4 (0.17-0.74)
Trend  <0.00010.0059

Both case and control cohorts contained more drinkers than nondrinkers. In type of drinking, more cases were in the drinking liquor compared with controls. The proportion of smoking was much higher among the cases than controls, especially among those smoking ≥60 pack-years: while 36.6% were in cases, only 3.1% were in controls.

Crude and adjusted ORs and 95% CIs of smoking and drinking exposure on the risk of bladder carcinoma were estimated and are presented in Table 3. No obvious association was observed between alcohol drinking and risk of bladder carcinoma after controlling for age, gender, race, and years of education. The frequencies of all kinds of drinking type were not associated with bladder carcinoma. Tobacco smoking was associated with a strongly increased risk with an adjusted OR of 3.1 (95% CI: 1.7–5.9). The OR for pack-years of smoking as a continuous variable was 2.4 (95% CI: 2.0–2.8). An obvious dose–response relation was observed between pack-years of smoking and risk of bladder carcinoma (P trend < 0.0001).

Table 3. Associations between Tobacco Smoking, Alcohol Drinking, and the Risk of Bladder Cancer
 Cases n (%)Controls n (%)Crude OR (95% CI)Adjusted ORa (95% CI)
  • a

    Adjusted for age (continuous), gender (M/F), race, and education (continuous).

Non-Smoking39 (17.0)108 (54.0)1.01.0
Yes191 (83.0)92 (46.0)5.8 (3.7-9.0)3.1 (1.7-5.9)
Non-Smoking39 (18.3)108 (57.1)1.01.0
2030 (14.1)47 (24.9)1.3 (0.8-2.3)1.5 (0.7-3.1)
20 to <4032 (15.0)19 (10.1)3.5 (1.8-6.7)1.6 (0.7-3.7)
40 to <6034 (16.0)9 (4.8)7.9 (3.5-17.5)3.2 (1.2-8.3)
≥6078 (36.6)6 (3.1)27.1 (11.2-65.8)7.3 (2.7-19.9)
Trend  <0.0001<0.0001
Years of quit smoking    
Non-Smoking39 (21.9)108 (64.7)1.01.0
1 to <1047 (26.4)15 (9.0)4.8 (2.6-9.1)3.0 (1.3-7.1)
10 to <2032 (18.0)23 (13.8)2.1 (1.2-3.9)1.1 (0.5-2.4)
≥2060 (33.7)21 (12.6)4.4 (2.5-7.7)0.6 (0.3-1.4)
Trend P  <0.00010.8925
No31 (13.9)36 (17.9)1.01.0
Yes192 (86.1)165 (82.1)1.4 (0.8-2.3)1.1 (0.5-2.3)
Type of drinking    
Non-drinking31 (15.0)36 (18.0)1.01.0
Beer16 (7.7)23 (11.5)0.5 (0.2-1.0)1.0 (0.3-3.1)
Wine18 (8.7)26 (13.0)0.7 (0.5-1.0)0.7 (0.4-1.2)
Liquor17 (8.2)9 (4.5)1.1 (0.8-1.5)0.9 (0.6-1.3)
Mix125 (60.4)106 (53.0)1.0 (0.8-1.1)1.0 (0.8-1.2)
Trend P  0.07110.5688
Years of drinking beer    
Non-drinking31 (21.1)36 (26.3)1.01.0
1 to <2019 (12.9)42 (30.7)0.4 (0.2-0.7)2.3 (0.7-7.1)
20 to <4037 (25.2)49 (35.8)0.7 (0.4-1.1)1.3 (0.7-2.6)
≥4060 (40.8)10 (7.3)5.3 (2.6-10.8)1.6 (0.7-3.9)
Trend P <0.00010.7862 
Years of drinking wine    
Non-drinking31 (25.8)36 (26.5)1.01.0
1 to <2016 (13.3)42 (30.9)0.3 (0.1-0.5)0.8 (0.3-2.1)
20 to <4032 (26.7)50 (36.8)0.7 (0.5-0.9)0.9 (0.6-1.2)
≥4041 (34.2)8 (5.9)1.5 (1.2-2.0)1.0 (0.7-1.4)
Trend P  0.00060.8758
Years of drinking liquor    
Non-drinking31 (21.8)36 (32.7)1.01.0
1 to <2016 (11.330 (27.3)0.6 (0.3-1.3)1.7 (0.5-5.8)
20 to <4045 (31.7)31 (28.2)1.7 (0.9-3.3)1.4 (0.5-3.5)
≥ 4050 (35.2)13 (11.8)4.52 (2.1-9.7)0.7 (0.2-2.0)
Trend P  <0.00010.7014

The distributions of GSTP1 genotype polymorphism in bladder carcinoma and controls are shown in Table 4. The prevalence of GSTP1 homozygous Ile/Ile was 53.1% in bladder carcinoma and 54.7% in controls. The prevalence of Ile/Val was 45.5% in bladder carcinoma and 38.8% in controls. For the homozygous Val/Val, the prevalence was 1.4% in bladder carcinoma and 6.5% in controls. The relations between genetic susceptibility markers and risk of bladder carcinoma are also presented in Table 4. After adjusting for age, race, gender, and years of education, the Ile/Ile and any Ile genotype showed increased risk with an adjusted OR of 6.5 (1.00–41.56) and 7.0 (1.11–43.88), respectively.

Table 4. Odds Ratios (ORs) and 95% Confidence Intervals (95% CI) of GSTP1 and the Risk of Bladder Carcinoma
GenotypeCasesControlsCrude OR (95% CI)Adjusted ORa (95% CI)Further adjusted ORb (95% CI)
  • a

    Adjusted for age (continuous), gender (M/F), race, and education (continuous).

  • b

    Adjusted for age (continuous), gender (M/F), race, education (continuous), drinking (Yes/No), pack-years of smoking.

Val/Val2 (1.4)11 (6.5)
Ile/Val66 (45.5)66 (38.8)5.5 (1.17-25.77)7.6 (1.18-49.51)6.8 (0.74-63.30)
Ile/Ile77 (53.1)93 (54.7)4.6 (0.98-21.16)6.5 (1.01-41.56)8.2 (0.89-75.15)
Val/Val2 (1.4)11 (6.5)
Any Ile143 (98.6)159 (93.5)4.9 (1.08-22.69)7.0 (1.11-43.88)7.5 (0.84-67.21)

The possible interactions that were explored between GSTP1 genotypes and smoking are presented in Table 5. Cases and controls were divided into four groups: the referent group is the nonsmokers with GSTP1 Val/Val, the second group included individuals who were nonsmokers with at least one Ile genotype, the third group included people who were smokers with Val/Val genotype, and the fourth group included individuals who were smokers with any Ile genotype. The adjusted ORs were 1.95 for the smokers and with any Ile genotype. More than a multiplicative interaction was indicated between GSTP1 and smoking. The adjusted OR for interaction between GSTP1 any Ile/Ile genotype and tobacco smoking on the risk of bladder carcinoma was 11.42 (0.53–248.15) after controlling for age, gender, race, and years of education.

Table 5. Assessments of Interactions between GSTP1 and Smoking in Bladder Carcinoma
SmokingaGSTP1Cases n (%)Controls n (%)Crude OR (95% CI)Adjusted ORb (95% CI)
  • a

    Smoking: Ever smoking is defined as a total of 100 or more cigarettes smoked lifetime prior to 1 year ago.

  • b

    Adjusted for age (continuous), gender (M/F), race, and education (continuous).

NeverVal/Val1 (0.7)4 (2.4)1.01.0
NeverIle/Val &Ile/Ile24 (16.7)89 (53.3)1.08 (0.12-10.10)0.47 (0.03-7.57)
EverVal/Val1 (0.7)7 (4.2)0.57 (0.03-11.86)0.08 (0.002-2.87)
EverIle/Val & Ile/Ile118 (81.9)67 (40.1)7.05 (0.77-64.33)1.95 (0.13-29.84)
InteractionSmoking any Ile  11.42 (0.53-248.15)


Before interpretation of our results, potential limitations of the study should be addressed. Potential selection bias is the first consideration, because only 62.2% of cases and 83.3% of controls had DNA samples for GSTP1 assays. We compared those with and without DNA samples in terms of age, gender, race, and education; except gender, no obvious differences were observed. Furthermore, the selection of blood donors, who may have healthy lifestyles, as controls could lead to an underestimation of the association between smoking, alcohol drinking, and the risk of the disease.

Another limitation of this study was that the sample size was not large enough to analyze gender effect modification. In controls, the prevalence of GSTP1 in males was 51.49%, 41.79%, and 6.72% for Ile/Ile, Ile/Val, and Val/Val, respectively; the prevalence in females was 66.67%, 27.78%, and 5.56% for Ile/Ile, Ile/Val, and Val/Val, respectively. The distributions of polymorphisms were not different in gender among controls (χ2 = 2.6887, P = 0.2607) and cases (Fisher exact test P = 0.0661). In comparison with females with the Val/Val genotype, the ORs were 2.0 (0.09–46.82) for females with any Ile genotype, 0.89 (0.03–28.20) for males with Val/Val type, and 4.03 (0.18–90.29) for males with any Ile type. The OR for interaction between gender and GSTP1 on the risk of bladder carcinoma was 2.26 with CIs included null value.

This study shows that the risk factors for bladder carcinoma are older age, poorer education, and smoking. OR of bladder carcinoma increased with age, consistent with other studies.25, 26 Although drinking was not obviously associated with risk of bladder carcinoma, a dose–response relation was seen in all kinds of drinking. Gender was not found to be a risk factor, again in contrast to results from other studies,26–28 but men and non-Caucasians have a higher OR.29 Another risk factor for bladder carcinoma, education, was found to be a risk factor, with less education being a greater risk factor. We observed that both smoking and the duration of smoking increased the risk of bladder carcinoma, consistent with Castelao et al.3 and Vineis et al.29

Many epidemiological studies find that smokers have an overall risk level about 2–4 times that of lifelong nonsmokers. Risk of bladder carcinoma tends to increase with increasing number of cigarettes smoked and with duration of smoking in years, and risk levels are reduced in ex-smokers compared with current smokers. High risk of bladder carcinoma appears to be limited largely to cigarette smokers.3, 4, 29 The cases in this study also had a high frequency of smoking (83.04% of current smokers). It is suggested that abstinence from smoking to avoid exposure to chemical carcinogens in tobacco is very important for bladder carcinoma prevention and control.

GSTP1 has been reported to possess two variant alleles. A single base substitution as position 313 of exon 5, guanine for adenine, results in the presence of valine (Val), where originally isoleucine (Ile) was present. Interestingly, the prevalence rates of these isoforms are entirely dependent on which ethnic group is being considered. In Caucasians, Ryberg et al.30 reported that the frequencies of variant genotype of GSTP1, Ile/Ile, Ile/Val, and Val/Val were 51.5%, 39.4%, and 9.1%, respectively. In contrast, the Japanese population had a slightly lower rate of the Val variant genotypes (69% of Ile/Ile, 29% of Ile/Val, 1.8% of Val/Val) than Caucasians.31 In this study, we found that among all control subjects the prevalence of Ile/Ile, Ile/Val, and Val/Val were 54.7%, 38.8%, and 6.5%, respectively, which is very similar to the reported prevalence in Caucasian people. These findings suggest that the distributional variability of GSTP1 is of considerable importance in determining an individual's susceptibility to the development of certain carcinomas.

A report suggests that GSTP1b (Val/Val) is far more efficacious in modifying benzopyrene diol epoxide (BPDE), the ultimate carcinogen of benzo[a]pyrene.32, 33 For example, the Vmax of glutathione conjugation of (+)-7β, 8α-dihydroxy-9α, 10α-Oxy-7,8,9, 10-tetrahydrobenzo[α]pyrene((+)-anti-BPDE), which among the four BPDE isomers is the most potent carcinogen, was 3.4-fold higher for the Val 105 wildtype of GSTP1 compared with the Ile 105 variant. Similarly, the Vmax of GSH conjugation of the carcinogenic diol epoxide of chrysene, (+)-anti-1,2-dihydroxy-3,4-Oxy-1,2,3,4-tetrahydrochrysene((+)-anti-CDE), was 5.3-fold higher for the Val 105 wildtype compared with the Ile 105 variant. These studies support our results that individuals with the GSTP1 Val 105 variant might be less susceptible to carcinogenic effects in tobacco.

Our results indicate that the Ile 105 allele was associated with an increased risk of bladder carcinoma. In previous articles, several types of carcinoma have been studied. Most notably stated by Harries et al.,18 there appeared to be an approximately threefold increase in risk between those with the GSTP1 (Val/Val) allele and those with GSTP1 (Ile/Ile) variant for bladder carcinoma. A similar increase in risk was demonstrated by Ryberg et al.30 while studying the association between the Pi class polymorphisms and squamous cell carcinoma of the lung. Studies of other types of carcinoma have shown the same results.18, 29, 34 Recently, Saarikoski et al.35 agreed with the suggestion that subjects homozygous for variant GSTP1 (Val/Val) alleles would actually be less susceptible to the carcinogenic effects of benzopyrene than heterozygotes or wildtype homozygotes. This finding supports our results. No relation, however, between the GSTP1 genotype and lung carcinoma risk has been observed.36 These different results may be explained by patients differing in exposure toward an unidentified carcinogen metabolized by GSTP1. Although these data have inconsistent associations with GSTP1 (Ile/Ile), they suggest an important role for the GSTP1 polymorphism in cancer susceptibility.

In our results, we found that the OR for any Ile genotype changed from a crude OR of 7.05 (95% CI: 0.72–64.33) to adjusted OR of 1.95 (95% CI: 0.13–29.84) after controlling for age, gender, race, and education (Table 5). To identify the major factor leading to this reduction of the adjusted OR, we adjusted potential confounding factors one at a time. The ORs were 7.50 (95% CI: 0.80–70.00) after adjusting for gender, 1.48 (95% CI: 0.11–20.61) after adjusting for age, 6.82 (95% CI: 0.75–62.3) after adjusting for race, and 7.70 (95% CI: 0.81–73.13) after adjusting for education. It seems that age is a responsible variable for the major reduction of the ORs.

For the first time a potential multiplicative interaction has been documented between the GSTP1 Ile allele and tobacco smoking on the risk of bladder carcinoma. Our result shows that the OR for multiplicative interaction was 11.42 (95%CI: 0.53–248.15), although the sample size is relatively small for interaction assessment. The data suggest that the Ile genotype of GSPT1 might modify the risk related to smoking.

In conclusion, genetic factors might be related to bladder carcinoma. Our data indicate the Ile/Ile homozygous 105 allele was associated with an increased risk of bladder carcinoma and suggest a potential gene–environment interaction between tobacco smoking and GSTP1 (any Ile) on the risk of bladder carcinoma. These results need to be interpreted with caution because of the relatively small sample size. Further large-scale studies are needed to confirm our findings.