Association of genetic polymorphism of glutathione S-transferase(GSTM1, GSTT1, GSTP1) with bladder cancer susceptibility
Mohammad Reza Safarinejad, M.D.*, Saba Safarinejad, S.B., Nayyer Shafiei, B.Ch.E.,Shiva Safarinejad, M.S.
Private Clinical Center for Urological Disease Diagnosis and Private Clinic Specialized in Urological and Andrological Genetics, Teheran, Iran
Received 13 September 2011; received in revised form 23 October 2011; accepted 11 November 2011
Abstract
The glutathione-S-transferases (GSTs) comprise a class of enzymes that detoxify carcinogenic compounds by conjugating glutathione tofacilitate their removal. Polymorphisms in GSTM1, GSTT1, and GSTP1 genes have been related to risk for bladder cancer. Studies focusingon GSTs gene variants relationship with the risk of bladder cancer have produced conflicting and inconsistent results. We examine theassociation between genetic polymorphism of glutathione S-transferase P1, GSTM1, GSTT1 genes and development of bladder transitionalcell carcinoma (TCC). The study population consisted of 166 histologically confirmed male bladder TCC cases and 332 healthy malecontrols. Genotyping was done using the polymerase chain reaction restriction fragment length polymorphism (PCR-RFLP) method and alsoinvestigated combined gene interactions. The GSTP1 Val/Val genotype was significantly associated with bladder cancer (OR � 4.32, 95%
I: 2.64–6.34), whereas the association observed for GSTM1 null (OR � 1.32, 95% CI: 0.82–2.62; P � 0.67) and GSTT1 null genotypeOR � 1.18, 95% CI: 0.79–1.67; P � 0.74) did not reach statistical significance. There was a significant multiple interaction betweenSTM1, GSTT1, and GSTP1 genotypes in risk of bladder cancer (P for interaction � 0.02). The risk associated with the concurrentresence of GSTM1 positive and GSTP1 Ile/Val or Val/Val (OR � 3.71, 95% CI: 2.34–5.54) and GSTT1 positive and GSTP1 Ile/Val or
Urinary bladder cancer is the fourth most common typeof cancer in men and the seventh in women, with an annualincidence of 32 of 100,000 in men and 9 of 100,000 inwomen [1]. Bladder cancer is usually considered to becaused by environmental carcinogens. Tobacco smokingand occupational exposure to aromatic amines and polycy-clic hydrocarbons have been associated with risk of bladdercancer [2]. There is increasing evidence suggesting thatgenetic polymorphisms of xenobiotic-metabolizing en-
zymes could increase individual susceptibility to variousenvironmental and clinical conditions. The balance betweenactivation and detoxification of carcinogens affects theamount of deoxyribonucleic acid (DNA) damage that oc-curs in cells [3]. Glutathione-S-transferases (GSTs) consti-tute a superfamily of phase II enzymes that are involved inthe detoxification of exogenous substrates such as xenobi-otics, environmental substances, and carcinogenic com-pounds [4,5]. To date, human cytosolic GST superfamilycontains at least 16 genes subdivided into 8 distinct classesdesignated as: � (alpha), � (mu), � (kappa), � (omega), �(pi), � (sigma), � (theta), and � (zeta), which are encoded byhe GSTA, GSTM, GSTK, GSTO, GSTP, GSTS, GSTT,nd GSTZ genes, respectively [6,7]. Functional polymor-
hisms have been identified in the GSTM1, GSTT1, and
1194 M.R. Safarinejad et al. / Urologic Oncology: Seminars and Original Investigations 31 (2013) 1193–1203
GSTP1 genes coding for GSTs enzymes in the �, �, and �classes, respectively.GSTM1 gene is polymorphic, and atleast 4 different alleles exist [8]. The deleted genotypes,which result in the inactive form of the enzyme, have beennamed GSTM1 null [9]. The homozygous deletion ofGSTM1 has been shown to be associated with an increasedrisk of various types of cancer [10–13]. Studies investigat-ng the association between glutathione GSTs polymor-hism and bladder cancer risk report very conflicting andnconsistent results. Three studies reported that GSTM1 nullenotype is a risk factor for bladder cancer [14–16]. How-ver, a recent study found that the GSTM1 polymorphism isot associated with risk of bladder cancer [17]. The poly-orphism in the GSTT1 gene loci is also caused by a gene
eletion [18], and leads to virtual absence of enzyme activ-ty in individuals with the null genotype. A few studiesemonstrated nonsignificant diminished risk of bladder can-er with GSTT1 null genotype [15,19,20] nonetheless; somether studies demonstrated increased risk of bladder cancerith GSTT1 null genotype [21,22]. GSTMI and GSTTI nullenotypes have decreased capacity to detoxify certain car-inogens and have been linked with increased risks foreveloping bladder cancer [23,24].
The GSTP1313 A/G polymorphism at the nucleotideevel results in an amino acid variation of isoleucine/valinet codon 105 in the protein. Valine amino acid leads toiminished enzymes activity [25]. Törüner et al. reportedhat GSTP1Ile/Val or Val/Val genotype increases 1.75 foldf bladder cancer susceptibility [26]. In another study, Sriv-stava et al. [27] reported that the Val/Val genotype of theSTP1 gene polymorphism increases 7.1-fold of bladder
ancer susceptibility.Since Kempkes et al. [20] first presented the potential re-
ationship between GSTT1 and bladder cancer, an increasingumber of studies have investigated the association betweenSTs gene variation and bladder cancer risk in different ethnicopulations. However, the results from epidemiologic studiesre still inconsistent and conflicting. The purpose of this studyas to determine the frequencies of polymorphisms of GSTT1,STM, and GSTP1 genes and their association with bladderCC among Iranian population.
. Materials and methods
.1. Subjects
The study subjects consisted of 166 men with histopatho-ogically confirmed bladder TCC (mean age � SD, 63.8 �
9.2 years) and 332 cancer-free control subjects (mean age �SD, 63.4 � 9.7 years), recruited from our urology clinicbetween October 2006 and October 2009. No patient re-ceived chemotherapy or radiotherapy before recruitment.These patients were carefully paired with controls selectedfrom the blood donors with respect to environmental expo-
history, general health conditions, and previous diseases.All cancer patients were newly diagnosed with transitionalcell carcinoma (TCC) of the bladder. The control groupconsisted of non-related healthy men without history ofmalignant disease who were matched to those in the casegroup for geographic origin, smoking status, and age range(�2 years). For each case, 2 controls were randomly se-lected. All study subjects provided informed consent priorto participating in the study, which was conducted accord-ing to Helsinki declaration (2004).
2.2. Inclusion/exclusion criteria
Patients with previous cancer, with cancer metastasizedto bladder from another origin, and those with previouschemotherapy or radiotherapy were excluded. Selection cri-teria for controls were no evidence of any personal or familyhistory of cancer or other serious illness. Control subjectswho had a previous diagnosis of any type of cancer wereexcluded.
2.3. Evaluations
All individuals underwent a complete physical examina-tion and answered a structured questionnaire that includeddemographic and ethnic background, height, weight, life-time occupational history, smoking habits, use of drugs andmedicines, general health conditions, and personal medicalhistory. Smoking status was defined as: (a) an ‘eversmoker’, who had smoked more than100 cigarettes in hislifetime; and (b) a former smoker, who had stopped smokingmore than 1 year before diagnosis for the patients and morethan 1 year before entry to the study for the control subjects.For each individual, the number of pack-years (PY) was cal-culated to demonstrate the cumulative smoking dose.
PY � (number of cigarettes smoked per day/20 ciga-rettes) � number of years smoked.
The clinical information about tumor size, number, stage,and grade, radiotherapy, and chemotherapy were obtainedfrom medical records. In all of the bladder cancer patients,the tumors were diagnosed histologically as transitional cellcarcinomas (TCC). The tumors were classified as superficial(pTa–pT1) or invasive (pT2–pT4) according to the 1997TNM staging system of the American Joint Committee onCancer. The histologic grades were subdivided into lowgrade and high grade using the grading systems of theWorld health Organization (WHO) and the InternationalSociety of Urological Pathology (ISUP).
2.4. Genotyping
Genotyping was done exactly as the methods we used inour previous study [28].
Genomic DNA was extracted from the peripheral bloodof subjects using the QIAmp blood kit (Qiagen, Chatsworth,
CA) and according to the method describe by Liu et al. [29].
A
c
Gp6
L
1195M.R. Safarinejad et al. / Urologic Oncology: Seminars and Original Investigations 31 (2013) 1193–1203
For the GSTM1 gene, DNA samples were amplified withthe primers: 5=-CTG CCC TAC TTG ATT GAT GGG-3=and 5=-CTG GAT TGT AGC AGA TCA TGC-3=, usingpolymerase chain reaction (PCR). The paired primers forGSTT1 were: 5=-TTC CTT ACT GGT CCT CAC ATCTC-3= and 5=-TCA CCG GAT CAT GGC CAG CA-3=.
mplification of human -globin (110 bp) with the primers5=-ACA CAA CTG TGT TCA CTA GC-3= and 5=-CAACTT CAT CCA CGT TCA CC-3= was used as an internalcontrol. PCR was performed in a final volume of 50 �l,onsisting of DNA (0.5 �l), and a PCR buffer containing
200 ng of the following reagents: dNTP (0.2 mM each),MgCl2 (1.5 mM), KCL (50 mM), Tris–HCl (10 mM, pH8.3), and 0.1% of bovine serum albumin. Reaction mixtureswere heated at 94°C for 5 minutes followed by 35 cycles ofamplification as follows: a denaturing step at 94°C for 90seconds, an annealing step at 59°C for 1 minute, and finalextension step at 72°C for 1 minute. Reaction products wereanalyzed by 2% agarose gel. The fragment lengths of thePCR products were 273-bp for individuals with one or moreGSTM1 alleles, and 480-bp for individuals with 1 or moreGSTT1 alleles.
Fig. 1. Representative screening for the GST genotypes. Patterns for eacSTM1 (273-bp fragment) genes, detected by PCR-RLFP analysis are shoroducts samples: 100 bp ladder (lane L); absence of null genotypes (lane, 7, 8, 10, 11).
Fig. 2. Cleavage of PCR products of GSTP1 gene by the Alw26I restriction
), Ile/Ile allele (lanes 2, 3, 5, 6); Ile/Val allele (lanes 1, 7, 8, 9), and Val/Val al
The GSTP1 genotype was also determined by themethod of Liu et al. [29] using a PCR-restriction fragmentlength polymorphism (RFLP) technique. An 177 bp frag-ment of the GSTP1 gene containing Ile to Val substitutionin exon 5 (codon 105) was amplified using the primer pair5=-ACC CCA GGG CTC TAT GGG AA-3= and 5=-TGAGGG CAC AAG AAG CCC CT-3=. The amplification cy-cles included an initial denaturing at 94°C for 30 seconds,followed by an annealing step at 61°C for 30 seconds, anda final extension step at 72°C for 30 seconds. The PCRproducts were then digested with Alw26I. HomozygousIle/Ile individuals had a single fragment of 177-bp, andhomozygous Val/Val individuals had both 92- and 85-bpfragments (Figs. 1, and 2). The presence of all 3 fragmentscorresponded to heterozygous Ile/Val individuals.
2.5. Quality control
Quality control was maintained by genotyping 10% du-plicates for cases and controls with 100% concordance tothe genotype by PCR-RFLP. Laboratory personnel were
genotypes, GSTT1 (480 bp fragment), -globin (268-bp fragment) andsence of the PCR product indicates the null genotype. Representative PCR9); GSTT1-null allele (lanes 2, 5), and GSTM1-null allele (lanes 1, 2, 5,
uclease. Representative PCR-RFLP products samples: 100 bp ladder (lane
h of thewn. Abs 3, 4,
endon
lele (lane 4).
oaTtfpb
gmnaaaV
cteidPI
OO
E
M
M
S
TT
1196 M.R. Safarinejad et al. / Urologic Oncology: Seminars and Original Investigations 31 (2013) 1193–1203
unable to distinguish among case, control, and quality con-trol samples.
2.6. Statistical analysis
The power of the study was determined using Quantosoftware, ver. 1.0 (available from: http://hydra.usc.edu/gxe)with consideration of following variables: case-controlstudy design, significance level (�) � 0.05 (2-sided), modelf inheritance was log additive, allele frequency was 0.18,nd the genetic effect for odds ratio (OR) was 1.8 or greater.he present study achieved 80% of the statistical power for
he Val allele of GSTP1, which showed the lowest allelerequency among the 3 polymorphisms. Results are ex-ressed as mean � standard deviation (SD). Hardy-Wein-erg equilibrium was examined using the goodness-of-fit 2
test to compare the observed allele frequencies with theexpected frequencies determined from control subjects. The2 or Fisher’s (F) exact test was used to assess the homo-eneity between cases and controls regarding ethnicity,edication use, cigarette smoking, disease extent, and ge-
otypes. Associations of polymorphisms with susceptibilitynd clinicopathologic characteristics of bladder cancer werenalyzed by unconditional logistic regression to estimatedjusted odds ratios and 95% confidence intervals (CI).
Table 1Distribution of selected demographic variables and risk factors among the
ariables that were significantly associated with bladder
ancer incidence or outcome by univariate analysis wereested by multiple logistic regression model in order toxamine the effect of all genotypes and clinical risk factors,ncluding, age, and cigarette smoking as independent pre-ictors. Statistical analysis was performed by Statisticalackage for the Social Sciences (SPSS) ver. 17.0 (SPSSnc., Chicago, IL).
1197M.R. Safarinejad et al. / Urologic Oncology: Seminars and Original Investigations 31 (2013) 1193–1203
3. Results
3.1. Characteristics of the study population
The frequency distributions of demographic and clin-ical characteristics of the cases and controls are presentedin Tables 1 and 2. The cases and controls appeared to bewell matched on age: mean age was 63.8 � 9.2 years forcases and 63.4 � 9.7 years for controls (P � 0.71).Throughout our analyses, we consistently found no im-portant confounding by age, BMI, smoking status, oreducation. In the following paragraphs all of the reportedORs are adjusted for confounding factors in multivariateanalysis.
3.2. Genotyping
3.2.1. Genotype distributions of the GSTM1polymorphisms between the cases and controls
The genotype frequencies of GSTM1 polymorphismsamong the controls were in agreement with the Hardy-Weinberg equilibrium (2 test: P � 0.842). Genotype fre-quencies of the GSTM1 polymorphisms among the casepatients and control subjects and their associations with riskof bladder cancer are shown in Table 3. There was no mainffect on risk of bladder cancer.
.2.2. Genotype distributions of the GSTT1 polymorphismsetween the cases and controls
Distribution of GSTT1 genotype among our controls metardy-Weinberg equilibrium (2 test: P � 0.232). Theverall prevalence of the GSTT1 null genotype was similarmong cases and controls (21.1% vs. 20.8%, respectively).
tendency of increased risk was also observed betweenSTT1 null genotype and bladder cancer, but this associa-
ion failed to reach statistical significance (OR � 1.18, 95%
Table 3The distribution, as n (%), of GSTM1, GSTT1, and GSTP1 genotypes in
Genotype and allele frequency Control (n � 332) (%)
a Adjusted OR: adjusted in multivariate logistic regression models inclgenotypes.
I: 0.79–1.76; P � 0.74) (Table 3).
.2.3. Genotype distributions of the GSTP1olymorphisms between the cases and controls
The GSTP1 genotypes were also in Hardy-Weinbergquilibrium for the controls (2 test: P � 0.228).
The Ile/Val genotype was more frequent in cancer pa-tients with a significant trend for an OR of 2.32 (95% CI:1.24–3.65; P � 0.007). The proportion of the individualswith GSTP1 Val/Val genotype was higher in cases (14.5%vs. 2.4%), the ORs being 4.32 (95% CI: 2.64–6.34; P �0.001).
3.3. Distribution of double GST genotypes among cancerpatients and controls
Significant linkage disequilibrium was found among allpairs of GSTs gene. A total of 11 double combinations wereobserved in the participants. Combinations of these poly-morphisms were also evaluated (Table 4). There was astatistically significant multiple interaction betweenGSTM1, GSTT1, and GSTP1 genotypes in risk of bladdercancer (P for interaction � 0.02). The frequency of theGSTM1 positive � GSTP1 Ile/Ile or Val/Val combination,was higher in patients (45.2%) compared with controls(33.7%; OR � 3.71, 95% CI: 2.34–5.54; P � 0.004; Table 4).A comparison between individuals with GSTT1 positive �GSTP1 Ile/Ile or Val/Val combination and those withGSTT1 positive � GSTP1 Ile/Ile combination indicated anOR of 2.66 (95% CI: 1.45–4.72; P � 0.007). The combi-nation of GSTM1 null genotype with GSTP1 Ile/Val orVal/Val genotype, also resulted in 4.76-fold increased riskfor bladder TCC (95% CI: 2.68–8.72; P � 0.002). Theopposite was the case for GSTP1 Ile/Ile genotype. Thecombination with GSTP1 Ile/Ile genotype was associatedwith a 66% lower decreased risk of bladder TCC forGSTM1 null (OR � 0.44; 95% CI: 0.23–0.74; P � 0.001),and a 35% lower decreased risk of bladder TCC for GSTT1null (OR � 0.65; 95% CI: 0.42–0.84; P � 0.007) (for
ge, BMI, occupational status, educational level, smoking status and GST
the pat
Cas
11650
13135
548824
112
uding a
details see Table 4).
1198 M.R. Safarinejad et al. / Urologic Oncology: Seminars and Original Investigations 31 (2013) 1193–1203
3.4. Distribution of triple GST. genotypes among cancerpatients and controls
Since none of the null genotypes alone seem to representa significant risk factor, associations between the mutantgenotypes were also evaluated. Individuals with a combinedgenotype GSTM1 positive/GSTT1 positive/GSTP1 Ile/Valor Ile/Val exhibited an elevated risk for bladder cancer withan OR of 2.2 (95% CI 1.10 � 4.57 and 1.12 � 4.20,respectively). The combination with highest risk wasGSTM1 null/GSTT1 null/GSTP1 Ile/Val or Val/Val (OR �6.42, 95% CI: 4.76–14.72; P � 0.0001). Any combinationwith GSTP1 Ile/Ile, irrespective of GSTM1 and GSTT1genotype, resulted in a decreased risk for bladder TCC(Table 5). Lowest risk was among subjects with GSTM1present/GSTT1 null/GSTP1Ile/Ile (OR � 0.66, 95% CI:0.47–0.79; P � 0.02).
3.5. Association of GSTs genotypes with tumorstage/grade
After stratification for grade and stage, neither was theGSTM1 null genotype frequency statistically different be-
Table 4Distribution of double GST genotypes among cancer patients and control
GSTM1 and GSTP1M1(�/�) and P1(Ile/Ile) 41 (24.7)M1(�/�) and P1(Ile/val or Val/Val) 75 (45.2)M1(–/–) and P1(Ile/Ile) 13 (7.8)M1(–/–) and P1(Ile/Val or Val/Val) 37 (22.3)
GSTT1 and GSTP1T1(�/�) and P1(Ile/Ile) 43 (25.9)T1(�/�) and P1(Ile/Val or Val/Val) 88 (53.0)T1(–/–) and P1(Ile/Ile) 11 (6.6)T1(–/–) and P1(Ile/Val or Val/Val) 24 (14.5)
a Adjusted OR: adjusted in multivariate logistic regression models inclgenotypes.
Table 5Distribution of triple GST genotypes among cancer patients and controls
Triple GST genotypes Cases (n � 166)
M1 (�/�) and T1(�/�) and P1(Ile/Ile) 22 (13.3)M1 (�/�) and T1(�/�) and P1(Ile/Val or Val/Val) 74 (44.6)M1(–/–), T1(�/�), and P1(Ile/Ile) 12 (7.2)M1(–/–), T1(�/�), and P1(lIe/Val or Val/Val) 23 (13.9)M1(�/�), T1(–/–), and P1(Ile/Ile) 18 (10.8)M1(�/�), T1(–/–), and P1(Ile/Val or Val/Val) 2 (1.2)M1(–/–), T1(–/–), and P1(Ile/Ile) 2 (1.2)M1(–/–), T1(–/–), and P1(Ile/Val or Val/Val) 13 (7.8)
a Adjusted OR: adjusted in multivariate logistic regression models incl
genotypes.
tween lower and higher grade and stage of diseases (Fig. 3).Similarly to GSTM1 null genotype, the frequency of theGSTT1 null genotype did not significantly differ betweenvarious subgroups of cases (Fig. 3). The GSTP1 Val/Valgenotype was observed to be associated significantly withan increased risk of having high-grade (OR � 7.68, 95% CI:4.73–19.25; P � 0.0001) and invasive (OR � 10.67, 95%CI: 6.34–21.75; P � 0.0001) tumors of bladder cancer(Table 6). Individuals with both null genotypes showed asignificantly increased risk of developing high grade (OR �7.48, 95% CI: 4.58–14.62; P � 0.0001), and muscle inva-sive (OR � 6.84, 95% CI: 4.45–12.18; P � 0.0001) tumorsof the bladder (Table 7).
GSTM1 present/GSTP1 Ile/Val or Val/Val, and GSTT1present/GSTP1 Ile/Val or Val/Val patients had significantlylower risk for high grade (OR � 0.21, 95% CI: 0.14–0.37;P � 0.0001, and OR � 0.51, 95% CI: 0.32–0.77; P �0.0006, respectively), and invasive (OR � 0.27, 95% CI:0.18–0.47; P � 0.0001, and OR � 0.47, 95% CI: 0.36–0.72; P � 0.001, respectively) bladder TCC; a similarassociation was also found for GSTM1 null/GSTP1 Ile/Ile, and GSTT1 null/GSTP1 Ile/Ile patients (for detailssee Table 7).
ls (n � 332) (%) Odds ratioa (OR) 95% CI P value
.7) 1.0 (referent)
.3) 1.49 0.81–2.34 0.66
.0) 1.16 0.84–1.76 0.79
.3) 1.0 (referent)
.7) 3.71 2.34–5.54 0.004
.5) 0.44 0.23–0.74 0.001
.5) 4.76 2.68–8.72 0.002
.6) 1.0 (referent)
.7) 2.66 1.45–4.72 0.007
.2) 0.65 0.42–0.84 0.007
.5) 3.88 2.86–5.82 0.001
ge, BMI, occupational status, educational level, smoking status and GST
Controls (n � 332) (%) Odds ratioa (OR) 95% CI P value
ge, BMI, occupational status, educational level, smoking status and GST
s
Contro
165 (4994 (2873 (22
127 (38112 (33
45 (1348 (14
138 (41125 (37
34 (1035 (10
uding a
(%)
uding a
wGe
GGaa
3p
cops
le6
just
edod
dsra
tios
for
rela
tion
betw
een
GST
geno
type
san
ddi
ffer
ent
grad
esan
dcl
inic
alst
ages
atdi
agno
sis
ofbl
adde
rca
ncer
ymor
phis
mG
rade
no.
(%)
OR
a(9
5%C
I)(n
�59
)P
valu
eC
linic
alst
ages
no.(
%)
OR
a(9
5%C
I)P
valu
e
Low
grad
eH
igh
grad
e(n
�10
7)Su
perfi
cial
(n�
119)
Inva
sive
(n�
47)
TM
1re
sent
95(8
1.9)
21(1
8.1)
1.0
(Ref
eren
t)83
(71.
6)33
(28.
4)1.
0(R
efer
ent)
ull
24(4
8.0)
26(5
2.0)
1.83
(0.6
5–2.
47)
0.64
24(4
8.0)
26(5
2.0)
1.82
(0.6
7–2.
57)
0.66
TT
1re
sent
103
(78.
6)28
(21.
4)1.
0(R
efer
ent)
90(6
8.7)
41(3
1.3)
1.0
(Ref
eren
t)ul
l16
(45.
7)19
(54.
3)1.
77(0
.62–
2.72
)0.
0817
(48.
6)18
(51.
4)1.
81(0
.71–
2.42
)0.
56T
P1 le/I
le49
(90.
7)5
(9.3
)1.
0(r
efer
ent)
40(7
4.1)
14(2
5.9)
1.0
(ref
eren
t)le
/Val
63(7
1.6)
25(2
8.4)
0.42
(0.2
7–0.
61)
0.00
0165
(73.
9)23
(26.
1)0.
40(0
.18–
0.62
)0.
0001
al/V
al7
(29.
2)17
(70.
8)7.
68(4
.73–
9.25
)0.
0001
2(8
.3)
22(9
1.7)
10.6
7(6
.34–
21.7
5)0.
0001
le/V
alor
Val
/Val
70(6
2.5)
42(3
7.5)
0.62
(0.4
8–0.
79)
0.00
267
(59.
8)45
(40.
2)0.
72(0
.51–
0.87
)0.
002
Adj
uste
dO
R:
adju
sted
inm
ultiv
aria
telo
gist
icre
gres
sion
mod
els
incl
udin
gag
e,B
MI,
occu
patio
nal
stat
us,
educ
atio
nal
leve
l,sm
okin
gst
atus
and
GST
geno
type
s.
1199M.R. Safarinejad et al. / Urologic Oncology: Seminars and Original Investigations 31 (2013) 1193–1203
Combinations of GSTP1, GSTM1, and GSTT1 showed asignificant linear gene-dosage relationship with grade (P fortrend � 0.001) and stage (P for trend � 0.001), withincrease in number of putative risk genotypes (GSTM1null, GSTT1 null, GSTP1 Ile/Val or Val/Val) (Tables 7,and 8).
Subjects carrying the GSTM1 positive/GSTT1 positive/GSTP1 Ile/Val or Val/Val combined genotype were pro-tected against high grade (OR � 0.35, 95% CI � 0.22–0.64; P � 0.0001) and muscle invasive (OR � 0.44, 95%CI � 0.23–0.67; P � 0.0005) bladder cancer development(Table 8). After stratification for grade and stage, a statis-tically significant increased risk of invasive (OR � 8.62,95% CI � 6.41–17.39; P � 0.0001) and high-grade (OR �8.21, 95% CI � 6.37–17.34; P � 0.0001) bladder cancer
as found for carries of the GSTM1 null/GSTT1 null/STP1 Ile/Val or Val/Val genotype, compared with refer-
nt genotype (Table 8).Due to small numbers of subjects in GSTM1 present/
STT1 null/GSTP1 Ile/Val or Val/Val, and GSTM1 null/STT1 null/GSTP1 Ile/Ile, category, we were unable to
nalyze the association between these combined genotypesnd tumor behaviors.
.6. Association of GSTM1, GSTT1, and GSTP1olymorphisms with smoking status
The median number of smoking pack-years among theases and controls was 14.7 and 14.1, respectively. More-ver, as with age and BMI, none of the 3 GST polymor-hisms was correlated with the number of pack-yearsmoked (all 3 Kruskal-Wallis test P values � 0.34),
indicating that the differences in smoking history be-tween cancer cases and controls cannot confound ourgenotypic association results. This finding was furtherconfirmed by multivariable logistic regression analyses
Fig. 3. Distribution of GSTM1 null, GSTT1 null, GSTP1 Ile/Val, andGDTP1 Val/Val genotypes in different grade and stage of bladder tumors.
(data not shown). The GSTP1 Val/Val genotype showed
Tab
Ad
Pol
GS P N
GS P N
GS I I V I a
Table 7Adjusted odds ratios for relation between double GST genotypes and different grades and clinical stages at diagnosis of bladder cancer
Double GST genotype Grade no. (%) ORa (95% CI) P value Clinical stages no. (%) ORa (95% CI) P value
GSTM1 and GSTP1M1(�/�) and P1(Ile/Ile) 36 (78.8%) 5 (12.2%) 1.0 (referent) 26 (63.4%) 15 (36.6%) 1.0 (referent)M1(�/�) and P1(Ile/val or Val/Val) 65 (86.7%) 10 (13.3%) 0.21 (0.14–0.37) 0.0001 63 (84.0%) 12 (16.0%) 0.27 (0.18–0.47) 0.0001M1(–/–) and P1(Ile/Ile) 8 (61.5%) 5 (38.5%) 0.58 (0.34–0.79) 0.004 9 (69.2%) 4 (30.8%) 0.38 (0.18–0.72) 0.002M1(–/–) and P1(Ile/Val or Val/Val) 10 (27.0%) 27 (73.0%) 7.32 (4.47–14.22) 0.0001 9 (24.3%) 28 (75.7%) 7.72 (6.64–15.11) 0.0001
GSTT1 and GSTP1T1(�/�) and P1(Ile/Ile) 37 (86.0%) 6 (14.0%) 1.0 (referent) 27 (62.8%) 16 (37.2%) 1.0 (referent)T1(�/�) and P1(Ile/Val or Val/Val) 68 (77.3%) 20 (22.7%) 0.51 (0.36–0.72) 0.001 66 (75.0%) 22 (25.0%) 0.47 (0.32–0.77) 0.0006T1(–/–) and P1(Ile/Ile) 6 (54.5%) 5 (45.5%) 0.84 (0.65–2.84) 0.08 7 (63.6%) 4 (36.4%) 0.41 (0.37–0.78) 0.003T1(–/–) and P1(Ile/Val or Val/Val) 8 (33.3%) 16 (66.7%) 4.82 (2.56–9.76) 0.001 7 (29.2%) 17 (70.8%) 7.15 (5.87–14.82) 0.0002
a Adjusted OR: adjusted in multivariate logistic regression models including age, BMI, occupational status, educational level, smoking status and GST genotypes.
Table 8Adjusted odds ratios for relation between triple GST genotypes and different grades and clinical stages at diagnosis of bladder cancer
Triple GST genotype Grade no. (%) ORa (95% CI) P value Clinical stages no. (%) ORa (95% CI) P value
Low grade(n � 119)
High grade(n � 47)
Superficial(n � 107)
Invasive(n � 59)
M1 (�/�) and T1(�/�) and P1(Ile/Ile) 20 (90.9) 2 (9.1) 1.0 (referent) 15 (68.2) 7 (31.8) 1.0 (referent)M1 (�/�) and T1(�/�) and P1(Ile/Val or Val/Val) 60 (81.1) 14 (18.9) 0.35 (0.22–0.64) 0.0001 53 (71.6) 21 (28.4) 0.44 (0.23–0.67) 0.0005M1(–/–), T1(�/�), and P1(Ile/Ile) 8 (66.7) 4 (33.3) 0.54 (0.36–0.86) 0.008 9 (75.0) 3 (25.0) 0.37 (0.21–0.76) 0.002M1(–/–), T1(�/�), and P1(lIe/Val or Val/Val) 11 (47.8) 12 (52.2) 1.82 (0.84–2.47) 0.24 10 (43.5) 13 (56.5) 1.82 (0.87–2.61) 0.27M1(�/�), T1(–/–), and P1(Ile/Ile) 13 (72.2) 5 (27.8) 0.42 (0.37–0.79) 0.003 13 (72.2) 5 (27.8) 0.39 (0.26–0.77) 0.003M1(�/�), T1(–/–), and P1(Ile/Val or Val/Val) 2 (100.0) 0 (0.0) NA 2 (100.0) 0 (0.0) NAM1(–/–), T1(–/–), and P1(Ile/Ile) 2 (100.0) 0 (0.0) NA 2 (100.0) 0 (0.0) NAM1(–/–), T1(–/–), and P1(Ile/Val or Val/Val) 3 (23.1) 10 (76.9) 8.21 (6.37–17.34) 0.0001 3 (23.1) 10 (76.9) 8.62 (6.41–17.39) 0.0001
a Adjusted OR: adjusted in multivariate logistic regression models including age, BMI, occupational status, educational level, smoking status and GST genotypes.
1200M
.R.
Safarinejadet
al./
Urologic
Oncology:
Seminars
andO
riginalInvestigations
31(2013)
1193–1203
1201M.R. Safarinejad et al. / Urologic Oncology: Seminars and Original Investigations 31 (2013) 1193–1203
an increased risk among non-smokers (OR � 3.22, 95%CI � 1.76 –5.37; P � 0.016), smokers (OR � 5.72, 95%CI � 3.28 –7.36; P � 0.003), former smokers (OR �6.14, 95% CI � 4.26 – 8.57; P � 0.001) and currentsmokers (OR � 4.34, 95% CI � 2.41– 6.76; P � 0.002).No association of GSTT1 and GSTM1 polymorphismswas found with smoking. In particular, light smokers(�20 pack-years) had a 3.17-fold (95% CI � 1.82– 4.79)and heavy smokers (�20 pack-years) had a 4.76-fold(95% CI � 2.59 – 6.82) increased risk, compared withnonsmokers.
4. Discussion
We detected no association between GSTM1 and GSTT1and risk for bladder cancer; however, the GSTP1 Ile/Val orVal/Val genotypes were associated with a significantly in-creased risk. In our study population, the bladder TCC riskwas found to increase 4.3-fold in individuals with the ho-mozygous GSTP1 Val/Val variant genotype. Our finding isconsistent with studies of bladder cancer conducted in In-dian [27], Turkish [26], and British population [30]. Con-versely, Steinhoff et al. [31] in the German population andKatoh et al. [32] in the Japanese population found no asso-ciation for GSTP1 Ile/Val or Val/Val genotype with sus-ceptibility to bladder cancer. Also, significant associationwas found between the GSTP1 Val/Val genotype and thegrade and stage of cancer with an OR of 7.68 and 10.67,respectively. In the present study, we did not observe sig-nificant increased risk for bladder cancer with GSTT1 nullgenotypes compared with the controls. This finding agreeswith previous studies [27,31]; nonetheless, in studies per-formed in Slovakian and Egyptian [21,33] populations, sta-tistically significant risk of bladder cancer for GSTT1 nullgenotype was reported. Two meta-analyses demonstratedthat the GSTM1 null genotype associated with an overall40%–50% increased risk for developing bladder cancer risk[34,35]. Decreased risk for bladder cancer has also beenreported with GSTT1 null genotype [19]. Results pertainingto the GSTT1 polymorphism and the risk of bladder cancerare less consistent [33,36]. Salagovic et al. [21] and Lee etal. [22] found an increased risk of bladder cancer with theGSTT1 null genotype. Kempkes et al. [20] and Brockmolleret al. [19] reported a significantly increased risk of bladdercancer associated with GSTT1 null genotype in nonsmok-ers. In contrast, Sobti et al. demonstrated significant asso-ciation between GSTT1 null genotype as a risk to bladdercancer in the light smoker group with an OR of 2.19 [17]. Inthe most recent meta analysis, Zeng et al. suggest thatGSTT1 null status is associated with a modest increase inthe risk of bladder cancer [37].
We also found a nonsignificant risk for the GSTM1 nullgenotype. Our results concur with the observations made byKim et al. in the Korean population [38], and by Srivastava
et al. in the Indian population [27]. Conversely, several
other studies reported significant association of GSTM1 inbladder cancer [21,31].
Bladder TCC with different grades and stages may rep-resent entities with different etiology and prognosis, anddisparities in the tumor molecular genetics [39]. In thecurrent study, the results did not show a significant associ-ation between GSTT1 and GSTM1 polymorphisms andgrade and stage of the cancer. Our findings concur with theprevious studies that showed nonsignificant association ofGSTM1 genotypes with grade of the disease [27,40]. How-ever, when we analyzed the relation between double GSTgenotypes and different grade and clinical stages, we founda statistically significant association between both GSTM1/GSTT1 null genotype variant.
Effect modification by smoking status has been exam-ined in some studies, but with conflicting results [21,41].Our study revealed that the effects of GSTM1 and GSTT1polymorphisms in developing bladder cancer are indepen-dent of smoking status. This finding was in agreement with2 German studies [20,42], but disagrees with finding ofSobti et al. in the Indian population [17].
Data from our study do not support a substantial as-sociation between GSTM1 null and GSTT1 null geno-types and bladder cancer risk. One possible explanationis that the effect of the GSTM1 null and GSTT1 nullgenotypes may be detectable only in combination withother “at-risk” genotypes. Confirmation of this would,however, have required much larger study groups thanthe present ones.
When GSTM1 null and GSTP1 Ile/Val or Val/Val ge-notype were present together, risk increased by 4.76-fold incomparison to referent group. Cases with GSTP1 Ile/Val orVal/Val genotype along with GSTT1 null genotype were at3.88-times more risk of having TCC of bladder. This issuggestive of a synergistic function only between GSTM1null and GSTT1 null and GSTP1 Ile/Val genotype in theetiology of bladder cancer. Of importance, when we com-bined the 3-risk genotypes, risk increased more than 6-fold.
Some studies also evaluated the combination effects ofrisk genotypes (GSTM1 null, GSTT1 null, GSTP1 Ile/Valor Val/Val), and significant results could be found in mostof the studies [26,31,43]. The present study indicated esca-lation of risk with multiple risk alleles of GSTs. This findingsuggests that gene–gene interactions may be partially in-volved in genetic susceptibility for bladder cancer, whichcould be explained by various substrates used by differentGSTs inducing resulting in combined action.
We also attempted to examine the correlation betweenclinical stage and/or pathologic grade with GSTs genotypesfor the risk of bladder cancer, but no association was foundbetween GSTM1 null and GATT1 null genotypes (Figure).This finding agrees with previous study by Srivastava et al.[27]. In contrast, our findings showed a significant associ-ation of GSTM1/GSTT1 double null genotypes with tumorgrade and stage of bladder cancer. However, there is dis-
agreement in the published literature about this finding [27].
[
[
[
[
[
[
[
[
[
[
1202 M.R. Safarinejad et al. / Urologic Oncology: Seminars and Original Investigations 31 (2013) 1193–1203
The prevalence of GSTM1 and GSTT1 null genotypes varysignificantly among ethnic groups. The reported frequency ofthe GSTM1-null genotype is around 50% of the Caucasianpopulation in Europe [44,45]. In the present study, the fre-quency of the GSTM1-null genotype in the Iranian populationwas 28%. Therefore, the frequency of the homozygotic dele-tion of the gene (GSTM1 null) in the Iranian population ismuch lower than in all the European Caucasian population.The GSTT1 null allele is one of the most widely studied GSTpolymorphisms with a frequency of about 10%–20% in Cau-casians and 50%–60% in Asians [46]. The GSTP1 polymor-phism had strongest associations with bladder cancer. Thefrequency of GSTP1 genotypes in our controls is comparableto other larger studies with Caucasian populations [31,47],which demonstrate that potential selection bias may be relatedwith lifestyle factors, but not polymorphisms of GSTP1. Afteradjustment for disease stage or grade, GSTP1 Ile/Val genotypealone conferred protective effects against high-grade and high-stage tumor. The chance of having high-grade and muscleinvasive tumors decreases about 60%, in individuals harboringGSTP1 Ile/Val genotype.
In the current study, we only recruited male patients. Aninteraction between gender, the GSTM1 variant, smoking, andlung cancer has been reported [48]. With respect to genderdifferences, a hospital-based study from Germany of 157 blad-der cancer cases and 223 controls reported that the GSTT1polymorphism was unrelated to bladder cancer risk in men andnonsignificantly inversely related in women. However, gender-specific effects were not observed in the meta- and pooled-analysis of GSTM1 genotype and bladder cancer risk [49].
Our study has some limitations. Of risk factors for bladderTCC, we only addressed the smoking history. Because smok-ing is a well known risk factor for bladder cancer, we madeevery effort to match both groups regarding smoking history.Control subjects did not undergo any investigation regardingpresence of bladder cancer. Therefore, some of them (espe-cially smoker men) might have carcinoma in situ or bladdercarcinoma when recruited into study, and perhaps some ofthem will develop bladder cancer in the future. Despite thefairly large number of subjects included in the study, thenumbers were still regrettably small in some of the subgroupanalyses. In addition, multiple comparisons may cause a ten-dency towards findings by chance alone. Finally, since thefindings from the present study were only from an Iranianpopulation, it is uncertain whether these results are relevant toother ethnic groups.
5. Conclusion
The results obtained in the present study show that individ-uals with a particular combination of genotypes (GSTM1 null,GSTT1 null, and GSTP1 Ile/Val or Val/Val) have a signifi-cantly increased risk for bladder TCC, particularly for high-grade and muscle-invasive tumors. The interesting findings of
this study needs confirmation from larger studies.
Acknowledgments
The authors are indebted to the participants.
References
[1] Ferlay J, Bray F, Pisani P, et al. GLOBOCAN 2000: Cancer Inci-dence, Mortality, and Prevalence Worldwide. Ver. 1.0. IARC CancerBase No. 5. Lyon (France): IARC Press, 2001.
[2] Cohen SM, Shirai T, Steineck G. Epidemiology and etiology ofpremalignant and malignant urothelial changes. Scand J Urol Nephrol2000;(Suppl):105–15.
[3] Stern MC, Johnson LR, Bell DA, et al. XPD Codon 751 polymor-phism, metabolism genes, smoking, and bladder cancer risk. CancerEpidemiol Biomarkers Prev 2002;11:1004–11.
[4] Sanyal S, Festa F, Sakano S, et al. Polymorphisms in DNA repair andmetabolic genes in bladder cancer. Carcinogenesis 2004;25:729–34.
[5] Schnakenberg E, Breuer R, Werdin R, et al. Susceptibility genes:GSTM1 and GSTM3 as genetic risk factors in bladder cancer. Cyto-genet Cell Genet 2000;91:234–8.
[6] Lo HW, Ali-Osman F. Genetic polymorphism and function of gluta-thione S-transferases in tumor drug resistance. Curr Opin Pharmacol2007;7:367–74.
[7] Strange RC, Spiteri MA, Ramachandran S, et al. Glutathione-S-transferase family of enzymes. Mutat Res 2001;482:21–6.
[8] Hoffmann D, Hecht SS. Advances in tobacco carcinogenesis, In:Cooper CS, Grover PL, editors. Handbook of Experimental Pharma-cology. Heidelberg: Springer-Verlag, 1990. p. 63–102.
[9] Landi S. Mammalian class � GST and differential susceptibility tocarcinogens: A review. Mutat Res 2000;463:247–83.
10] Benhamou S, Lee WJ, Alexandrie AK, et al. Meta- and pooledanalyses of the effects of glutathione S-transferase M1 polymor-phisms and smoking on lung cancer risk. Carcinogenesis 2002;23:1343–50.
11] Singh M, Shah PP, Singh AP, et al. Association of genetic polymor-phisms in glutathione S-transferases and susceptibility to head andneck cancer. Mutat Res 2008;638:184–94.
12] Smits KM, Gaspari L, Weijenberg MP, et al. Interaction betweensmoking, GSTM1 deletion, and colorectal cancer: Results from theGSEC study. Biomarkers 2003;8:299–310.
13] Ates NA, Tamer L, Ates C, et al. Glutathione S transferase M1, T1,P1 genotypes and risk for development of colorectal cancer. BiochemGenet 2005;43:149–63.
14] Srivastava DS, Kumar A, Mittal B, et al. Polymorphism of GSTM1and GSTT1 genes in bladder cancer: A study from North India. ArchToxicol 2004;78:430–4.
15] Katoh T, Inatomi H, Kim H, et al. Effects of glutathione S-transferase(GST) M1 and GSTT1 genotypes on urothelial cancer risk. CancerLett 1998;132:147–52.
16] Shao J, Gu M, Zhang Z, et al. Genetic variants of the cytochromeP450 and glutathione S-transferase associated with risk of bladdercancer in a south-eastern Chinese population. Int J Urol 2008;15:216–21.
17] Sobti RC, Al-Badran AI, Sharma S, et al. Genetic polymorphisms ofCYP2D6, GSTM1, and GSTT1 genes and bladder cancer risk inNorth India. Cancer Genet Cytogenet 2005;156:68–73.
18] Pemble S, Schroeder KR, Spencer SR, et al. Human glutathioneS-transferase � (GSTT1): cDNA cloning and the characterization of agenetic polymorphism. Biochem J 1994;300:271–6.
19] Brockmoller J, Cascorbi I, Kerb R, et al. Combined analysis ofinherited polymorphism in arylamine N-acetyltransferases, glutathi-one S-transferases M1 and T1, microsomal epoxide hydrolase, andCYP 450 enzyme as modulator of bladder cancer risk. Cancer Res
1996;56:3915–25.
[
[
[
[
[
[
[
[
[
[
[
[
[
[
[
[
[
[
[
[
[
[
[
1203M.R. Safarinejad et al. / Urologic Oncology: Seminars and Original Investigations 31 (2013) 1193–1203
[20] Kempkes M, Golka K, Reich S, et al. Glutathione S-transferaseGSTM1 and STT1 null genotypes as potential risk factors for urothe-lial cancer of the bladder. Arch Toxicol 1996;71:123–6.
[21] Salagovic J, Kalina I, Stubna J, et al. Genetic polymorphism ofglutathione S-transferase M1 and T1 and risk factor in lung andbladder cancer. Neoplasma 1998;45:312–7.
[22] Lee SJ, Cho SH, Park SK, et al. Association of genetic polymorphismof glutathione S-transferase M1 and T1 and bladder cancer. J KoreanCancer Assoc 1999;31:548–55.
[23] Bell DA, Taylor JA, Paulson DF, et al. Genetic risk and carcinogenexposure: A common inherited defect of the carcinogen-metabolismgene glutathione S-transferase M1 (GSTM1) that increases susceptibilityto bladder cancer. J Natl Cancer Inst 1993;85:1159–64.
24] Lafuente A, Pujol F, Carretero P, et al. Human glutathione S-trans-ferase m (GST m) deficiency as a marker for the susceptibility tobladder and larynx cancer among smokers. Cancer Lett 1993;68:49 –54.
25] Ali-Osman F, Akande N, Mao J. Molecular cloning, characterization,and expression in Escherichia coli of full-length cDNA of threehuman glutathione S-transferase � gene variants. Evidence for dif-ferential catalytic activity of the encoded proteins. J Biol Chem1997;272:10004–12.
26] Törüner GA, Akyerli C, Uçar A, et al. Polymorphism of glutathioneS-transferase genes (GSTM1, GSTP1, and GSTT1) and bladder can-cer susceptibility in the Turkish population. Arch Toxicol 2001;75:459–64.
27] Srivastava DS, Mishra DK, Mandhani A, et al. Association of geneticpolymorphism of glutathione S-transferase M1, T1, P1, and suscep-tibility to bladder cancer. Eur Urol 2005;48:339–44.
28] Safarinejad MR, Shafiei N, Safarinejad S. The association of gluta-thione-S-transferase gene polymorphisms (GSTM1, GSTT1, GSTP1)with idiopathic male infertility. J Hum Genet 2010;55:565–70.
29] Liu YJ, Huang PL, Chang YF, et al. GSTP1 genetic polymorphism isassociated with a higher risk of DNA damage in pesticide exposedfruit growers. Cancer Epidemiol Biomarkers Prev 2006;15:659–66.
30] Harries LW, Stubbins MJ, Forman D, et al. Identification of geneticpolymorphisms at the glutathione S-transferase � locus and associa-tion with susceptibility to bladder, testicular, and prostate cancer.Carcinogenesis 1997;18:641–4.
31] Steinhoff C, Franke KH, Golka K, et al. Glutathione transferaseisozyme genotypes in patients with prostate and bladder carcinoma.Arch Toxicol 2000;74:521–6.
32] Katoh T, Kaneko S, Takasawa S, et al. Human glutathione S-trans-ferase P1 polymorphism and susceptibility to smoking related epi-thelial cancer, oral, lung, gastric, colorectal, and urothelial cancer.Pharmacogenetics 1999;9:165–9.
33] Abdel-Rahman SZ, Anwar WA, Abdel-Aal, et al.GSTM1 and GSTT1genes are potential risk modifiers for bladder cancer. Cancer DetectPrev 1998;22:129–38.
34] García-Closas M, Malats N, Silverman D, et al. NAT2 slow acety-
lation, GSTM1 null genotype, and risk of bladder cancer: Results
from the Spanish Bladder Cancer Study and meta-analyses. Lancet2005;366:649–59.
35] Johns LE, Houlston RS. Glutathione S-transferase �1 (GSTM1) sta-tus and bladder cancer risk: A meta-analysis. Mutagenesis 2000;15:399–404.
36] Guengerich FP, Their R, Persmark M, et al. Conjugation of carcin-ogen by � class glutathione S-transferase; mechanism and relevanceto variations in human risk. Pharmacogenetics 1995;5:103–7.
37] Zeng FF, Liu SY, Wei W, et al. Genetic polymorphisms of glutathi-one S-transferase T1 and bladder cancer risk: A meta-analysis. ClinExp Med 2010;10:59–68.
38] Kim WJ, Kim H, Kim CH, et al. GSTT1-null genotype is a protectivefactor against bladder cancer. Urology 2002;60:913–8.
39] Brockmoller J, Kerb R, Drakoulis N, et al. Glutathione S-transferaseM1 and its variants A and B as host factors of bladder cancersusceptibility: A case-control study. Cancer Res 1994;54:4103–11.
40] Georgiou I, Filiadis IF, Alamanos Y, et al. Glutathione S-transferasenull genotypes in transitional cell bladder cancer: A case-controlstudy. Eur Urol 2000;37:660–4.
41] Schnakenberg E, Lustig M, Breuer R, et al. Gender-specific effects ofNAT2 and GSTM1 in bladder cancer. Clin Genet 2000;57:270–7.
42] Anwar WA, Abdel-Rahman SZ, El-Zein RA, et al. Genetic polymor-phisms of GSTM1, CYP2E1, and CYP2D6 in Egyptian bladdercancer patients. Carcinogenesis 1996;17:1923–9.
43] Hung RJ, Boffetta P, Brennan P, et al. GST, NAT, SULT1A1,CYP1B1 genetic polymorphisms, interactions with environmentalexposures and bladder cancer risk in a high-risk population. Int JCancer 2004;110:598–604.
44] Lafunte A, Pujol F, Carretero P, et al. Human glutathione S-trans-ferase m (GST�) deficiency as a marker for the susceptibility tobladder and larynx cancer among smokers. Cancer Lett 1993;68:49 –54.
45] Gabbani G, Hou SM, Nardini B, et al. GSTM1 and NAT2 genotypesand urinary mutagens in coke oven workers. Carcinogenesis 1996;17:1677–81.
46] Nelson HH, Wiencke JK, Christiani DC, et al. Ethnic differences inthe prevalence of the homozygous deleted genotype of glutathioneS-transferase �. Carcinogenesis 1995;16:1243–5.
[47] Shepard TF, Platz EA, Kantoff PW, et al. No association between theI105V polymorphism of the glutathione S-transferase P1 gene(GSTP1) and prostate cancer risk: A prospective study. Cancer Epi-demiol Biomarkers Prev 2000;9:1267–318.
[48] Tang DL, Rundle A, Warburton D, et al. Associations between bothgenetic and environmental biomarkers and lung cancer: Evidence ofa greater risk of lung cancer in women smokers. Carcinogenesis1998;19:1949–53.
[49] Engel LS, Taioli E, Pfeiffer R, et al. Pooled analysis and meta-analysis of glutathione S-transferase M1 and bladder cancer: A HuGE
