Glutathione pathway genes and lung cancer risk in young and old populations
P.Yang1,5, W.R.Bamlet2, J.O.Ebbert3, W.R.Taylor4and
1Division of Epidemiology,2Division of Biostatistics,3Nicotine Research
Center and4Department of Laboratory Medicine and Cancer Genotyping
Facility, Mayo Clinic College of Medicine, 200 First Street SW, Rochester,
MN 55905, USA
5To whom correspondence should be addressed at: Department of Health
Sciences Research and Cancer Center, Mayo Clinic, 200 First Street SW,
Rochester, MN 55905, USA. Tel: þ1 507 266 5369; Fax: þ1 507 266 2478;
Multiple enzymes with overlapping functions and shared
substrates in the glutathione (GSH) metabolic pathway
have been associated with host susceptibility to tobacco
smoke carcinogens and in lung cancer etiology. However,
few studies have investigated the differing and interacting
roles of GSH pathway enzymes with tobacco smoke expo-
sure on lung cancer risk in young (550 years of age) and
old (480 years of age) populations. Between 1997 and 2001,
237 primary lung cancer patients (170 young, 67 old) and
234 controls (165 young, 69 old) were enrolled at the Mayo
Clinic. Using PCR amplification of genomic DNA, poly-
morphic markers for gGCS, GPX1, GSTP1 (I105V and
A114V), GSTM1 and GSTT1 were genotyped. Recursive
partitioning and logistic regression models were used to
build binary classification trees and to estimate odds ratios
(OR) and 95% confidence intervals for each splitting fac-
tor. For the young age group, cigarette smoking had the
greatest association with lung cancer (OR ¼ 3.3). For never
smokers, the dividing factors of recursive partitioning were
GSTT1 (OR ¼ 1.7), GPX1 (OR ¼ 0.6) and GSTM1 (OR ¼
4.3). For the old age group, smoking had the greatest asso-
ciation with lung cancer (OR ¼ 3.6). For smokers, the
dividing factors were GPX1 (OR ¼ 3.3) and GSTP1
(I105V) (OR ¼ 4.1). Results from logistic regression ana-
lyses supported the results from RPART models. GSH
pathway genes are associated with lung cancer develop-
ment in young and old populations through differing inter-
actions with cigarette smoking and family history.
Carefully evaluating multiple levels of gene--environment
and gene--gene interactions is critical in assessing lung
Around the world, the majority of lung cancer occurs in
tobacco smokers and is diagnosed on average in the sixth and
seventhdecades of life (1). Individuals younger than 50 (young
age group) or older than 80 years of age (old age group)
develop lung cancer infrequently (2). Factors resulting in
lung cancer development at these extreme ages cannot be
explained by smoking alone. Assessing individual susceptibil-
ity or resistance to exogenous carcinogens through the study of
lung cancer in young and old populations may be important in
understanding the etiology of lung cancer. Multiple enzymes
with overlapping functions and shared substrates in the glu-
tathione (GSH) metabolic pathway have been associated with
host susceptibility to carcinogens and toxic agents (3). This
system is dependent upon the concentration of GSH and the
level of activities of enzymes that catalyze the conjugation of
substrates to GSH (4). Therefore, the absence or low levels
of GSH-related enzyme activities could result in retention of
active carcinogens or toxic DNA-damaging compounds and
eventually lead to the development of cancer. Because of
the strong correlation between phenotype (enzyme activity
levels) and genotype assays for GSH pathway enzymes in
general (4,5), the use of polymorphic genotypes to identify
individuals who lack or have low activities of these enzymes is
The critical enzymes in this pathway are glutathione
S-transferase mu (GSTm), glutathione S-transferase theta
(GSTu), glutathione S-transferase pi (GSTp), glutathione
peroxidase (GPX) and gamma glutamylcysteine synthetase
(gGCS) (3,4,6). GSTm detoxifies tobacco-related and other
carcinogens. The GSTM1 gene, which encodes GSTm, has
three identified alleles (a, b and null). Based upon reviews of
meta-analyses (7--9), the GSTM1 null genotype confers an
increased risk for lung cancer [odds ratio (OR) ¼ 1.4--2.1].
GSTu detoxifies low molecular weight toxins as well as
tobacco-related carcinogens (10,11). The GSTT1 gene, encod-
ing GSTu, has the same allele system as GSTM1 and the
GSTT1 null genotype increases lung cancer risk (OR 1.2--1.6)
(12--17). Individuals null for both GSTM1 and GSTT1 appear
to be at substantially elevated risk (17), especially for squa-
mous cell lung carcinoma among people with low levels of
cigarette exposure (12). GSTP1, encoding GSTp, has two
polymorphic sites: exon 5 A1578G encodes I105V and exon
6 C2508T encodes A114V (13,18). The I105V variant associa-
tion with lung cancer risk has been inconsistent in the literature
(8,9,19), but the A114V variant has recently been reported to
increase lung cancer risk, particularly among smokers (20).
GPX1 encodes cellular GPX, a selenium-dependent detoxify-
ing enzyme that is important in the cellular defense against
cytotoxic lipid peroxidation products and in CD95-triggered
apoptosis to eliminate neoplastic cells (21,22). A GPX1 variant
(P198L) may contribute a 2-fold increase in lung cancer risk
(23). gGCS is the rate-limiting enzyme in GSH biosynthesis
(24) and a trinucleotide (GAG) marker has been identified
(25). Genotypes associated with lung cancer prognosis (6)
have been reported, but not with lung cancer risk. A few
studies have reported an association between GSH pathway
Abbreviations: CI, 95% confidence intervals; ETS, environmental tobacco
smoke; gGCS, gamma glutamylcysteine synthetase; GPX, glutathione
glutathione S-transferase; GSTP1, gene encoding GSTp; GSTM1, gene
encoding GSTm; GSTT1, gene encoding GSTu; OR, odds ratios; PCR,
polymerase chain reaction; RPART, recursive partitioning; ROS, reactive
Carcinogenesis vol.25 no.10#Oxford University Press 2004; all rights reserved.
Carcinogenesis vol.25 no.10 pp.1935--1944, 2004
by guest on June 8, 2013
genes and lung cancer risk in never smokers, but the results are
To date, only one study of GSTM1 and GSTT1 polymor-
phisms and lung cancer risk has focused on young populations
(32). The purpose of our study is to investigate the role of
multiple GSH pathway genes in lung cancer among young and
old populations and interactions with cigarette smoke exposure
and family history of lung cancer.
Materials and methods
Study participants and data collection
The research protocol and consent form were approved by the Mayo Clinic
Institutional Review Board. Details about the study design and methods have
been provided elsewhere (6,33--35). Cases were patients diagnosed as having
and/or being treated for pathologically confirmed primary lung cancer at Mayo
Clinic (Rochester, MN) and were enrolled between April 1997 and March
2001. Mayo Clinic is a tertiary medical center serving Olmsted County resi-
dents as a major primary care provider. Eligible patients were invited to
participate in an interview and provided a peripheral blood sample.
The controls were drawn from a pool of 2335 Olmsted County residents
enrolled as general controls in the Mayo Clinic Cancer Center’s Population
Science Program between 1997 and 2001 (33). This design was based
on findings from the Rochester Epidemiology Project (36) showing that in a
3 year period, 490% of Olmsted Countyresidents will visitthe Mayo Clinic at
least once for a blood test. Control subjects had no current or previously
diagnosed malignancy (except non-melanoma skin cancer) as of the date of
phlebotomy. Eligible controls received a self-administered questionnaire
(equivalent to the case interview) and a request for permission to use their
peripheral blood sample obtained during clinic visits. Of the respondents, 78%
gave permission to use their blood sample, of whom 84% completed the
questionnaire; 11% declined to participate and 1% were dead or could not be
located. Ethnic backgroundof the cases and controlswere very similar(93% of
the young cases, 92% young controls and 99% old cases and controls), mostly
Caucasians of non-Hispanic origin. Minority groups included African-Amer-
icans, Native Americans and Inuits, Asians and Pacific Islanders, Hispanic and
Data collected from case interviews or control questionnaires regarding
each first degree relative included vital status and health history of malignant
and non-malignant diseases with age at diagnosis and cause of death. Ancestral
background of each subject’s paternal and maternal grandparents was also
obtained(37). Cigarettesmokingdata collectedfrom ever smokers(formerand
current) included age of smoking initiation, years of smoking (duration),
cigarettes smoked per day (intensity) and date of smoking cessation if applic-
able. A detailed environmental tobacco smoke (ETS) history was obtained
(34). Never smokers were defined as those who smoked 5100 cigarettes
during their lifetime.
Each subject’s blood sample was assigned a blind identification number and
tested systematically at the Mayo Clinic Cancer Center’s Genotyping Labora-
tory. Specific quality control procedures included: isolated DNA was stored at
4?C until analysis was performed; positive and negative control samples were
used to confirm proper system operation and the absence of cross contamina-
tion; and amplified products were directly sequenced with the primers used for
the PCR to confirm reaction fidelity, as needed. The contrasting genotypes for
the six polymorphic markers used in our analysis were gGCS (repeat 77 versus
GSTP1-I105V(II versus other),
GSTP1A114V (CC versus other), GSTM1 (null versus present) and GSTT1
(null versus present).
Determination of GSTM1, GSTT1 and gGCS genetic polymorphisms
Patient genomic DNA, extracted from a peripheral blood sample, was ampli-
fied with the primer sets listed in Table I. Samples were amplified in 25 ml
reactions containing 50 ng genomic DNA, 0.5 mM forward and reverse pri-
mers, 200 mM deoxynucleotide triphosphates (Applied Biosystems, Foster
City, CA), 1.5 mM MgCl2, 50 mM KCl, 10 mM Tris--HCl (pH 8.3) and
0.625 U Taq polymerase (AmpliTaq Gold; Applied Biosystems). GSTM1
and GSTT1 were amplified in a multiplex format along with CYP1A1 as a
positive control (38). Reaction conditions were 35 cycles of 95?C for 30 s,
59?C for 30 s, 72?C for 30 s following a10 min 95?C ‘hot start’ activation and
followed by a 7 min 72?C extension. The reaction conditions for gGCS were
similar with a 65?C annealing temperature. Upon completion, amplicons from
both PCRs were mixed and 1 ml aliquots loaded onto DNA500 microcapillary
chips and run on a model 2100 bioanalyzer (Agilent, Palo Alto, CA). The
GSTT1 and GSTM1 polymorphisms are characterized by a homozygous dele-
tion of the respective gene, easily distinguishable from the bioanalyzer elec-
tropherograms (6). The gGCS polymorphism involves a GAG trinucleotide
repeat in clusters of either 7, 8 or 9 subunits, each of which is able to be
discriminated from the ordering of potential 132, 135 and/or 138 bp electro-
pherogram peaks (6).
Determination of GSTP1, CYP1A1 and GPX1 genetic polymorphisms
Patient genomic DNA was amplified under standard PCR conditions, as above,
with the exception of using 50to 30biotinylated primers. Primer sequences and
specific annealing temperatures are also shown in Table I. The 72?C extension
step was increased from 30 s to 2 min for the longer GSTP1 amplicon contain-
ing both the I105V and A114V polymorphisms. Samples were then purified
using nucleic acid filtration plates (MultiScreen-PCR; Millipore, Bedford,
MA) and resuspended in ddH2O. Histidine (Sigma, St Louis, MO) was added
to each of the samples to a final concentration of 50 mM and a final volume
of 80 ml.
The NanoChip Molecular Biology Workstation (Nanogen, San Diego, CA)
was used to determine patient sample genotypes in a two-step procedure. First,
purified controls and patient samples (5--40 nM) were electronically addressed
to discreet microarray test sites on NanoChip cartridges. Second, 50-Cy5 and
50-Cy3 labeled probes with sequences specific to the individual polymorph-
isms were hybridized to the arrays. Adjacent unlabeled probes were added to
stabilize shorter labeled probes. The concentration of the probes varied from
250 to 500 nM in 50 mM NaPO4(pH 7.0) and 500 mM NaCl. The cartridges
were washed with 50 mM NaPO4 (pH 7.0) at discriminating temperatures
[30?C for CYP1A1 and GSTP1 (I105V); 42?C for GSTP1 (A114V) and GPX1],
scanned and analyzed. Heterozygote controls were used to effect proper
equilibration of fluorescent signals and determination was completed based
upon signal ratios.
Demographic information stratified by case/control status within each age
group was tabulated as a mean ? standard deviation for continuous variables
and a number (and percentage) for categorical variables. Pearson’s w2test was
used to assess group differences on categorical variables and a two-sample
t-test was used to assess group differences for continuous variables. In all cases
two-tailed P values ? 0.05 were considered statistically significant. We
employed recursive partitioning, a method of tree-based classification, to
investigate the following risk factors: tobacco smoke exposure (cigarette
smoking and/or ETS exposure), GSH-pathwaygenes, gene ? gene interactions
and cigarette smoking ? genes ? family history interactions in lung cancer
risk. We applied the RPART function written in the S-Plus software to
construct a classification tree accounting for potential confounders (39,40).
Table I. Primer sequences for polymorphic markers in a case--control study of 237 lung cancer patients and 234 community-based controls
Gene Forward primera
cCYP1A1 was used as a control marker.
P.Yang et al.
by guest on June 8, 2013
As an alternative to logistic models, tree-based models rectify classification
analysis, and they are useful for non-linear dependent and for visualizing
Recursive partitioning (RPART) in this study involved the following steps.
Each variable under consideration was examined and dividing rules were used
to examine all possible binary splits for the full group of subjects (‘root node’)
to select a dichotomization of the variable that maximally discriminated dis-
ease status. Each binary split yielded two subgroups (‘descendant nodes’), one
that contained a relatively high proportion of cases (splitting to the right) and
the other that contained a relatively high proportion of controls (splitting to the
left) (Figures 1 and 2). The variable producing the best split (greatest discrim-
ination) was then used to partition the root node into two descendant nodes.
The process was applied to each descendant node to produce further splits. The
combination of these binary splits provided a set of prediction rules that was
used to classify subjects according to the probability of being a case or a
control. At the end of the recursive partitioning process, the initial tree was
pruned using the technique of cross-validation (39,40). RPART can minimize
the impact of missing data by using all ofthe non-missing data available for the
variables at each split. Since recursive partitioning is exploratory and not
hypothesis testing, multiple comparison tests were not applied. The end results
of RPART were terminal nodes representing combinations of risk factors
associated with an increased likelihood of lung cancer.
In addition, we incorporated the results of the exploratory analyses using
RPART into a series of logistic regression models and applied the Akaike
information criterion (36) to identify the model that best fitted our data using
SAS software (version 8.12). The variables considered included age (at diag-
nosis for cases or blood sample collection for controls), gender, smoking
status, ETS, family history of lung cancer and genotypes of the six GSH-
Individuals with a missing value on any of the variables or factors included in
the model were omitted in the RPART models. On the other hand, the SAS
procedurelogistic regression includedthe factorsof interestfor each univariate
analysis and omitted any individual who was missing information for a uni-
variate variable.Therefore,the oddsratios (OR) with95% confidence intervals
(CI) calculated using SAS may theoretically differ slightly from those obtained
from the RPART analysis due to the impact of missing values in the different
modeling techniques. In our analysis, missing values were minimal for all
variables, and this difference was negligible.
Among the young population, 170 cases and 165 controls were
identified. Among the old population, 67 cases and 69 controls
were identified. When comparing cases and controls within an
age group, there was no difference in gender ratio, mean age or
ethnic background. For the young age group, reported cigarette
smoking history, ETS exposure and family history of lung
Smokers Never Smokers
No Family HX*
Fig. 1. RPART analysis: young age group. The classification tree illustrates complex interactions of risk factors in the populations. Each node shows the
number of controls (left) and cases (right). Branches to the right are more likely to contain cases, whereas branches to the left are more likely to contain controls.
Terminal nodes are indicated by square boxes. The values 0 or 1 at the top of each box indicate whether there are more cases (1) or more or the same number
of controls (0) in a particular branch. The four terminal nodes on the left correspond to never smokers, and the five terminal nodes on the right correspond to
ever smokers. The shaded area indicates branches dropped in the cross-validation pruning process of the RPART analysis. HX means history.
Glutathione pathway genes in young and old
by guest on June 8, 2013
cancer were significantly different between cases and controls.
For the old age group, only smoking and ETS exposure history
was significantly different. Between the young and old age
groups, as shown in Table II, the male:female ratio, region of
residence, smoking status at diagnosis, family history of lung
cancer, stage of NSCLC and histology were significantly
Allele distribution of the six polymorphic markers for cases
and controls are presented in Table III. The GSTT1 present
genotype was more prevalent, and the GPX1 TT genotype was
underrepresented in cases compared with controls for the
young age group. No significant difference was found for the
other four markers in either age group. Table IV provides
overall ORs and 95% CIs for all six genes from univariate
analysis. Genotype other is a combined group of homozygous
alternative allele and heterozygous genotypes that are not
informative (e.g. too rare) as separate groups.
RPART analysis was carried out separately for the young
and old groups (Figures 1 and 2). The order and the level of
each split represented the relative importance of each factor
and the interactions among factors. Logistic regression models
were applied to calculate ORs and 95% CIs for the splitting
For the young age group RPART models produced a
classification tree with eight splits and nine terminal nodes
(Figure 1 and upper panel of Table V). A positive cigarette
smoking history, the first split, had the most important effect
on lung cancer risk (OR ¼ 3.29). For young smokers, the first
dividing factor was family history of lung cancer (OR ¼ 3.06).
Young smokers with no such family history were further
divided on gGCS, GSTP1 (Il05V) and GSTP1 (A114V). If
gGCS is not homozygous 77, then being homozygous CC at
GSTP1 (A114V) increases the lung cancer risk by 44-fold.
However, if gGCS is homozygous 77, then possessing the
GSTP1 (I105V) II genotype decreases the risk of lung cancer
by 42-fold. For young never smokers, the first dividing factor
was GSTT1 (OR ¼ 1.65); the next dividing factor was GPX1,
and then GSTM1. For young never smokers who have GSTT1
present, possessing at least one T allele at GPX1 increases the
lung cancer risk by almost 2-fold; for those with GSTM1
present possessing at least one T allele on GPX1 could modify
the risk by 44-fold. For the old age group, RPART models
produced a classification tree with five splits and six terminal
nodes (Figure 2 and lower panel of Table V). We observed an
interaction between smoking and GPX1 genotype, i.e. GPX1
CC was associated with an increased lung cancer risk among
smokers while the same genotype was associated with a
reduced risk among never smokers. Furthermore, old smokers
with the GSTP1 II genotype were at an increased risk only
when they carried a T allele at the GPX1 locus.
To refine or prune the classification models, a cross-
validation procedure yielded final trees that were similar to
but had fewer terminal nodes than the original trees (unshaded
areas of Figures 1 and 2). Specifically, the cross-validated tree
for the young age group follows the tree discussed above for
never smokers but does not further divide the smokers.
Fig. 2. RPART analysis: old age group. The classification tree illustrates complex interactions of risk factors in the populations. Each node shows the number
of controls (left) and cases (right). Branches to the right are more likely to contain cases, whereas branches to the left are more likely to contain controls.
Terminal nodes are indicated by square boxes. The values 0 or 1 at the top of each box indicate whether there are more cases (1) or more or the same number of
controls (0) in a particular branch. There are three terminal nodes each for never and ever smokers. The shaded area indicates branches dropped in the
cross-validation pruning process of the RPART analysis.
P.Yang et al.
by guest on June 8, 2013
The cross-validated tree for the old age group follows the tree
discussed above for smokers but does not further divide the
In addition, logistic regression models were fitted to explore
several hypotheses, from univariate models to models adjust-
ing for smoking status and hierarchical models that incorpo-
rated second order interactions with smoking status. The
Akaike information criterion (41) was used to identify the
models that best fitted our data: model 1 included an intercept
and a risk factor; model 2 included an intercept, a risk factor
and smoking status; model 3 included an intercept, a risk
factor, smoking status and their interaction. We found that
for both age groups smoking status is the most important
main effect. For the old age group, the model including
GPX1, smoking status and their interaction provided a con-
siderably better fit than all the other models. For the young age
group, models incorporating smoking status, family history
and GSTT1 provided the best fit, although we did not find
one clearly superior model. Results from logistic regression
models supported the results from our RPART models.
In this analysis, ETS was not identified as an independent
nor significant factor in any of the models while cigarette
smoking had the most important effect on lung cancer risk.
We also repeated our analyses with a US White subset; the
results were almost identical to the total study population for
both young and old age groups.
As summarized in Table VI, we observed that significant risk
factors for developing lung cancer among young smokers were
family history of lung cancer and presence of the GSTP1
(A114V) CC genotype. Among young never smokers, the
GSTM1 present genotype was associated with lung cancer.
Among old smokers, the GPX1 CC and GSTP1 (I105V) II
genotypes were associated with lung cancer. Among old
never smokers, GSTP1 (I105V) II genotype was associated
with lung cancer. These findings not only suggest that GSH
pathway genes are important in lung cancer development
Table II. Selected characteristics of 237 primary lung cancer patients, Mayo Clinic, MN, 1997--2002
Young patients (550 years)
No. (%) (n ¼ 170)
42.9 ? 5.5
Old patients (480 years)
No. (%) (n ¼ 67)
83.2 ? 2.1
Age at diagnosis
Region of residence
Other Midwest Stateb
US White (non-Hispanic)
Smoking status at diagnosis
Never smokers with ETS
11 (92) 0.441
Family history of lung cancer
26 (15) 14 (21)0.030
aFisher’s exact test.
bIowa, Illinois, Indiana, Kansas, Michigan, Missouri, North Dakota, Nebraska, Ohio, South Dakota and Wisconsin.
cNon-small cell lung cancer.
dSmall cell lung cancer.
eBronchoalveolar carcinoma, a subtype of adenocarcinoma.
Glutathione pathway genes in young and old
by guest on June 8, 2013
among young and old populations but also elucidate the differ-
ing role of specific enzymes that interact with tobacco carcino-
gen exposure and family history of lung cancer in the two
contrasting age groups.
Our results on allele distributions of the polymorphic
markers are consistent with other studies of all age groups.
The frequency of homozygous deletion (null) genotypes in
both GSTM1 and GSTT1 vary substantially by ethnicity (11).
For GSTM1 the null genotype ranges from 36 to 67% in
Caucasians, 33 to 63% in East Asians and 22 to 35% in
Africans and African-Americans. Approximately 45% of US
Caucasians lack a functional allele, and the null genotype is a
risk factor for multiple cancers, including lung (11). For the
GSTT1null genotype, the highestfrequency isobservedinEast
Asians (38--58%), a lower frequency is observed in Africans
and African-Americans (24--38%) and the lowest frequency is
observed in Caucasians. Twenty percent of US Caucasians
lack a functional allele, and the null genotype has been asso-
ciated with a group of cancers (4,11). The reported frequency
for Ile/Val (IV) or Val/Val (VV) at codon 105 of GSTP1 is
18--57% (13,20) and for Ala/Val (AV) or Val/Val (VV) at
Table IV. Unadjusted odds ratios for six genes associated with lung cancer
Polymorphic lociUnadjusted odds ratio (95% CI)
Young age group Old age group
Null versus present
Null versus present
II versus other
CC versus other
77 versus other
CC versus other
0.9 (0.6, 1.5)
0.5 (0.3, 0.9)
0.8 (0.5, 1.2)
1.2 (0.6, 2.1)
0.8 (0.5, 1.2)
0.9 (0.6, 1.5)
0.9 (0.5, 1.9)
0.9 (0.4, 1.9)
1.4 (0.7, 2.8)
0.9 (0.4, 2.3)
1.4 (0.7, 2.9)
1.3 (0.6, 2.6)
Table III. Allele distribution of genetic markers in a case--control study of 237 lung cancer patients and 234 community-based controls
Marker Young age group (550 years) Old age group (480 years)
Control (n ¼ 165)Cases (n ¼ 170)
P valueControl (n ¼ 69) Cases (n ¼ 67)
Table V. Odds ratios estimation for risk factors identified from recursive
partitioning (RPART) models in a case--control study of 237 lung
cancer patients and 234 community-based controls
Variable identified from RPARTSAS estimates OR (95% CI)
Young age group (550 years)
Family history (first degree relatives)
GSTP1 (I105V) II
GSTP1 (A114V) CC
3.3 (2.1, 5.2)
3.1 (1.0, 9.4)
0.7 (0.4, 1.4)
0.4 (0.1, 1.1)
4.2 (1.3, 14.2)
1.7 (0.8, 3.4)
0.6 (0.3, 1.2)
4.3 (1.5, 13.0)
Old age group (480 years)
3.6 (1.7, 7.6)
3.3 (1.3, 8.4)
4.1 (1.1, 14.8)
0.12 (0.02, 0.7)b
4.3 (0.8, 25.0)
aVariables remaining in cross-validated models.
bThe second decimal place is kept to distinguish the lower CI from 0.
P.Yang et al.
by guest on June 8, 2013
codon 114 is 14--19% (20). The gGCS polymorphic repeat site
cohort (6). The prevalence of the GPX1 P98L variant allele was
58% (23). These data suggest that our cases and controls are
representative of the general population in terms of allele fre-
quencies of the six polymorphic markers in the study.
In the young age group, family history of lung cancer in first
degree relatives presents as a significant risk factor only
among smokers, supporting a previous hypothesis that an
inherited predisposition is important in early onset cancer
(42,43). Because smoking as a behavior aggregates in families,
which could lead to familial aggregation of smoking-related
cancer, particularly lung cancer, it is not unexpected that our
findings show smoking and lung cancer family history are very
important risk factors even for young cases (3). Among smo-
kers who have no family history and possess the gGCS 77
genotype, the GSTP1 (A114V) CC genotype is associated with
a 4-fold increased risk for lung cancer (OR ¼ 4.2, 95% CI
1.3--14.2). Recently, Wang et al. (20) reported that among
smokers who are 562 years of age at diagnosis the TT/CT
genotypes are associated with a 5-fold increased risk for lung
cancer (OR ¼ 5.1, 95% CI 2.5--10.2). The discrepant results
between studies may reflect the differences in subsets of the
population with regard to family history and the genotype
distribution of gGCS, which were not analyzed in the study
of Wang et al. (20).
Among young never smokers who have the wild-type at
GSTT1 and carry at least one T allele at GPX1, the GSTM1
present genotype is associated with a 4-fold increased risk for
lung cancer (OR ¼ 4.3, 95% CI 1.5--13.0). In a review of a
meta-analysis (9), among 420 published studies, the GSTM1
null genotype has been inconsistently associated with lung
cancer risk, ranging from no association (three studies), a
weak association (OR ¼ 1.04--1.25), to a moderate association
(OR ¼ 1.4--2.1), an effect which is stronger in heavy smokers.
London et al. (44) found a non-statistically significant associa-
tion of the GSTM1 null genotype and lung cancer risk among
African-Americans (OR ¼ 1.20, 95% CI 0.72--2.00) and Cau-
casians (OR ¼ 1.37, 95% CI 0.91--2.06). The association was
evident among light but not heavy smokers. A large case--
control study conducted in Boston also generated negative
results (45), as did a case--control study in Texas that included
African-Americans and Mexican-Americans (17). A similar
lack of association was observed in a Portuguese population
(46). A small study conducted in Spain (15) found a greater
frequency of the null genotype among cases than population-
based controls (OR ¼ 1.57, 95% CI 0.99--2.51). The associa-
tion appeared to be most evident for small cell carcinoma and
adenocarcinoma for light rather than heavy smokers (50 pack-
years as the cut-off point) and patients diagnosed at older ages
(?60 years). A Swedish study found a slightly inverse associa-
tion of GSTM1 null genotype and lung cancer risk in a sample
enriched for women (74.6%) and never smokers (48.2%) (27).
Our finding of a seemingly opposite relation between GSTM1
and lung cancer risk could be explained by the potential role of
GSTm among never smokers being different from its role
among smokers (10,11), particularly in the presence of the
GSTT1 null and GPX1 TT/CT genotypes.
Our results are consistent with several other investigations
with regards to the association between lung cancer and GSH
pathway genetic polymorphisms in never smokers. In a case--
control study of 198 lung cancer cases and 332 controls no
difference was observed between GSTP1 (I105V) variant
alleles in non-smoking cases and controls (26). No statistically
significant relationship was observed among non-smokers
either with or without ETS exposure and the GSTP1 (I105V)
genotype in a study of 66 and 413 non-smoking cases and
controls, respectively (28). In a Swedish case--control study
focusing on never smokers, the GSTM1 present genotype con-
ferred a higher risk for lung cancer in the presence of a slow
acetylator genotype (OR ¼ 3.1, 95% CI 1.1--8.6) (27). Results
that were most similar to ours were reported in a clinical
case--control study in Manchester, UK, where GSTM1 present
was associated with a 2-fold increased risk of lung cancer (29).
In the same study GSTT1 and GSTP1 (I105V) were not found
to be significant genetic markers.
Our data also contrast with several other studies evaluating
GSH related genes and lung cancer among never smokers. In a
study of archival tumor tissues never smokers with ETS expo-
sure who developed lung cancer were more likely to possess
the GSTM1 null genotype compared with never smokers with
no ETS exposure (30). However, this study included only non-
smoking females with a larger percentage of adenocarcinoma
than in our study. In a Japanese case--control study of 198 cases
with adenocarcinoma and 152 controls the GSTM1 null geno-
type was associated with lung cancer (OR ¼ 3.32, 95% CI
1.41--7.84) in non-smokers compared with non-smokers with
GSTM1 present (31). However, subjects were Japanese,
included adenocarcinoma cases only, had mean ages of 63
and 65 years for the cases and controls, respectively, and
510% of the cases were 550 years. Further, by the author’s
own admission, the risk of the GSTM1 null genotype may have
been enhanced by a biased allele distribution. We could not
identify any studies of the risk of lung cancer with GPX1 or
gGCS polymorphisms in never smokers.
In the older age group the presence of wild-type GPX1
interacts with cigarette smoking history significantly in lung
cancer risk. Among smokers the GPX1 CC genotype is
associated with a 3-fold increased risk (OR ¼ 3.3, 95%
CI 1.3--8.4), while among never smokers the same genotype
is associated with an 8-fold decreased risk of lung cancer
(OR ¼ 0.12, 95% CI 0.02--0.7). Our results are partly consis-
tent with the findings of Ratnasinghe et al. (47) from 315
case--control pairs in which GPX1 CT versus GPX1 CC
Table VI. Summary of multiple risk models: significant risk factors, odds
ratios and 95% confidence intervals in a case--control study of 237 lung cancer
patients and 234 community-based controls
Young age groupOld age group
GSTP1 (A114V) CCNS NSNS
GSTM1 present 4.3
GSTP1 (I105V) IINS NS
aRelative importance of and interactions among risk factors are depicted
in the figures and explained in the text.
bLung cancer in first degree relatives.
cNot statistically significant.
Glutathione pathway genes in young and old
by guest on June 8, 2013
(OR ¼ 1.8, 95% CI 1.2--2.8) and GPX1 TT versus GPX1 CC
(OR ¼ 2.3, 95% CI 1.3--3.8). These findings are consistent
with the role that the GPX enzyme plays in lung tissue and the
impact that the altered enzyme function may have on oncogen-
esis. Reactive oxygen species (ROS) are important in the
initiation and promotion of cells to neoplastic growth, and
cigarette smoking generates a high level of ROS within the
human airways. However, lung cells are equipped with an
integrated antioxidant defense system, which includes the anti-
oxidants GSH and GPX1 (48), which also plays an important
role in apoptosis (21). In a study of a normal human lung and
six major types of human lung carcinomas immunostained for
antioxidant enzymes (manganese and copper, zinc superoxide
dismutases, catalase and GSH peroxidase), none of the
carcinomas studied had significant levels of catalase or GPX,
supporting GPX being important in antioxidant defense
(49,50). Polymorphic genes for GPX1 map to loci on chromo-
some 3p, which is subject to frequent loss of heterozygosity in
lung tumors (51), where reduced GPX1 enzyme activity may
affect the prognosis of lung cancer patients due to
compromised oxidative defense mechanisms.
We also observed that GSTP1 (I105V) interacts with GPX1
genotype. Among old smokers who have the GPX TT geno-
type, possessing GSTP1 II is associated with a 4-fold increased
risk of lung cancer (OR ¼ 4.1, 95% CI 1.1--14.8), whereas no
association was observed among those who have GPX1 CC or
GPX1 CT. Reviews by Kiyohara et al. (8,9) reported that five
of the six studies did not show any association between GSTP1
and lung cancer risk, whereas Miller et al. (19,52) found
GSTP1 II is associated with a significantly increased risk of
lung cancer. Our study has provided new evidence regarding
GSTp in lung cancer development in an elderly population.
Individuals null for multiple enzymes appear to be at
elevated risk (17), especially for squamous cell carcinoma
among those with low levels of cigarette exposure (12). Kelsey
et al. (17) found that individuals who lacked functional genes
at both the GSTP1 and GSTM1 loci were at a 2.9-fold elevated
risk (P 5 0.05). This result is similar to that observed by
Saarikoski (12) in Finland, who observed a 2.3-fold excess
risk (P 5 0.05). Lung cancer patients in a Norwegian study
who had the null genotype at the GSTM1 locus and GSTP1 II
had higher DNA adduct levels than cases with other genotypes
(14), suggesting biological plausibility for the association.
However, the results are not entirely consistent with other
studies (15,16). Substrate specificity could influence a
person’s capacity to metabolize different environmental carci-
nogens, which isoftenacomplexmixtureofchemicalswhichare
possibly substrates for multiple metabolizing enzymes (12--17).
Therefore, additional studies with larger sample sizes and more
detailed exposure histories are required to clarify these inconsis-
Merging data from multiple studies across the world, one
recent study examined the lung cancer risk in populations
younger than 45 years of age and the GSTM1 null and
GSTT1 null genotypes (27). Only GSTM1 null was found to
be moderately associated with lung cancer risk among never-
smokers (52 cases). No studies have been published to date
reporting data on populations older than 80 years of age.
There are several limitations to this study. First, the use of
mostly referral-practice and self-selected patients may bias our
study results. However, this bias is considered minimal
because self-selection was not predetermined by genotype
and, the allele distributions of the six markers in our study
are in agreement with data reported in the literature.
The second limitation is that our control group was not ideal.
It is always difficult to find proper controls in a tertiary referral
clinic where a large number of new cases can be rapidly
enrolled. Since our lung cancer patients are mainly referrals,
ideal controls should be matched with cases by age, gender,
race, geographic referral area and duration of care at Mayo
Clinic. However, finding such ideal controls is not feasible,
mainly because ~50% of our cases are from outside the tri-
state area (Minnesota, Wisconsin and Iowa) where a limited
number of eligible controls can be found and enrolled. The
nature and seriousness of co-morbid conditions may differ
between regional and long-distance referral patients. Alterna-
tive control groups include population-based samples, siblings
or other relatives, neighbors, co-workers or spouses. We have
chosen to use population-based community residents as con-
trols based on Mantel and Haenszel’s principle of control
selection, i.e. using a group representing a more general popu-
lation could be superior if the comparability of exposure is the
major concern (53). Because the exposures in our study are
two genetic traits, biases of preferential recall or exposure time
are no longer major concerns. Although geographic distribu-
tion by residence differs between cases and controls, race and
ethnic background is comparable (data not shown). Because
the main exposures in our study are genetic traits, preferential
recall or exposure time biases are no longer major concerns.
On the other hand, our population-based control group can
serve two purposes: being a reference in testing our hypotheses
and providing accurate estimates of the expected allele fre-
quencies of the candidate markers in the reference population.
The third limitation is our low statistical power to thor-
oughly assess the role of ETS and its interaction with GSH
pathway genes in lung cancer risk, which may have been
masked by the strong effect of smoking. Analyzing only
never smokers could be helpful to detect ETS effects but is
limited by sample size in our current study.
In this study we focused on extreme age populations to
examine the interactions among selected host and environmen-
tal risk factors. We have achieved sizeable samples for the
young group but a limited number of old subjects. Our results
demonstrate that GSH pathway genes interacting with family
history of lung cancer and cigarette smoke exposure may
influence lung cancer risk in young and old populations.
Although we do not know the mechanisms underlying the
significant gene--age--smoking interaction from our epide-
miologic observations, these results are biologically plausible.
Due to the overwhelmingly strong causal effect of tobacco
smoking on lung cancer risk, never smokers of any age are
expected to be at minimal risk of developing lung cancer. If a
never smoker does develop lung cancer, most likely the person
has had previous exposure to second-hand smoke and/or other
known lung carcinogens (34,54). Two obvious differences in
individual susceptibility to lung cancer between smokers and
never smokers are: (i) the lung carcinogenic process in never
smokers does not require active smoking but may be triggered
by other exogenous or endogenous carcinogens (55,56); (ii) as
a consequence, metabolic pathways in reacting to carcinogens
may be different depending on the specificity of physiologic
substrates and the effective enzymes (57). Carefully evaluating
multiple levels of gene--environment and gene--gene interac-
tions is critical in assessing an individual’s lung cancer risk.
P.Yang et al.
by guest on June 8, 2013
We would like to thank Ms Susan Ernst for her technical assistance with the
manuscript. We also acknowledge Drs M.S.Allen, C.Deschamps, M.C.Aubry,
R.Marks,S.OkunoandZ.Sunfortheirsupportand helpin variousstages ofthis
project. This work was supported by grants NIH CA-77118 (P.Y.), NIH CA-
80127 (P.Y.) and NIH CA92049 (J.O.E.).
1.Blot,W.J. and Fraumeni,J.F.,Jr (1996) Cancers of the lung and pleura. In
Schottenfeld,D. and Fraumeni,J.F.,Jr (eds) Cancer Epidemiology and
Prevention. Oxford University Press, New York, NY, pp. 637--665.
2.Weir,H.K., Thun,M.J., Hankey,B.F. et al. (2003) Annual report to the
nation on the status of cancer, 1975--2000, featuring the uses of
surveillance data for cancer prevention and control. J. Natl Cancer Inst.,
3.Sellers,T.A. and Yang,P. (2002) Familial and genetic influences on risk of
lung cancer. In King,R.A., Rotter,J.I. and Motulsky,A.G. (eds) The
Genetic Basis of Common Diseases. Oxford University Press, New York,
NY, pp. 700--712.
4.Zhong,S., Howie,A.F., Ketterer,B., Taylor,J., Hayes,J.D., Beckett,G.J.,
Wathen,C.G., Wolf,C.R. and Spurr,N.K. (1991) Glutathione S-transferase
m locus: use of genotyping and phenotyping assays to assess association
with lung cancer susceptibility. Carcinogenesis, 12, 1533--1537.
5.Kempkes,M., Wiebel,F.A., Golka,K., Heitmann,P. and Bolt,H.M. (1996)
Comparative genotyping and phenotyping of glutathione S-transferase
GSTT1. Arch. Toxicol., 70, 306--309.
6.Yang,P., Yokomizo,A., Tazelaar,H.D. et al. (2002) Genetic determinants
of lung cancer short-term survival: the role of glutathione-related genes.
Lung Cancer, 35, 221--229.
7.McWilliams,J.E., Sanderson,B.J.S., Harris,E.L., Richert-Boe,K.E. and
Henner,W.D. (1995) Glutathione S-transferase M1 (GSTM1) deficiency
and lung cancer risk. Cancer Epidemiol. Biomarkers Prev., 4, 589--594.
8.Kiyohara,C., Otsu,A., Shirakawa,T., Fukuda,S. and Hopkin,J.M. (2002)
Genetic polymorphisms and lung cancer susceptibility: a review. Lung
Cancer, 37, 241--256.
9.Kiyohara,C., Shirakawa,T. and Hopkin,J.M. (2002) Genetic polymorphism
of enzymes involved in xenobiotic metabolism and the risk of lung cancer.
Environ. Health Prev. Med., 7, 47--59.
10.Hayes,J.D. and Strange,R.C. (1995) Invited commentary: potential
contribution of the glutathione S-transferase supergene family to
resistance to oxidative stress. Free Radic. Res., 22, 193--207.
11.Rebbeck,T.R. (1997) Molecular epidemiology of the human glutathione S-
transferase genotypes GSTM1 and GSTT1 in cancer susceptibility. Cancer
Epidemiol. Biomarkers Prev., 6, 733--743.
12.Saarikoski,S.T., Voho,A., Reinikainen,M., Anttila,S., Karjalainen,A.,
(1998) Combined effect of polymorphic GST genes on individual
susceptibility to lung cancer. Int. J. Cancer, 77, 516--521.
13.Harris,M.J., Coggan,M., Langton,L., Wilson,S.R. and Board,P.G. (1998)
Polymorphism of the Pi class glutathione S-transferase in normal
populations and cancer patients. Pharmacogenetics, 8, 27--31.
14.Ryberg,D., Skaug,V., Hewer,A. et al. (1997) Genotypes of glutathione
transferase M1 and P1 and their significance for lung DNA adduct levels
and cancer risk. Carcinogenesis, 18, 1285--1289.
15.To-Figueras,J., Gene,M., Gomez-Catalan,J., Galan,M.C., Fuentes,M.,
Glutathione S-transferase M1 (GSTM1) and T1 (GSTT1) polymorphisms
andlung cancerrisk among
Carcinogenesis, 18, 1529--1533.
16.Deakin,M., Elder,J., Hendrickse,C. et al. (1996) Glutathione S-transferase
GSTT1 genotypes and susceptibility to cancer: studies of interactions with
GSTM1 in lung, oral, gastric and colorectal cancers. Carcinogenesis, 17,
Polymorphisms in the glutathione S-transferase class mu and theta genes
interact and increase susceptibility to lung cancer in minority populations
(Texas, United States). Cancer Causes Control, 8, 554--559.
18.Mannervik,B., Awasthi,Y.C., Board,P.G. et al. (1992) Nomenclature for
human glutathione transferases. Biochem. J., 282, 305--306.
19.Miller,D.P., Neuberg,D., De Vivo,I., Wain,J.C., Lynch,T.J., Su,L. and
Christiani,D.C. (2003) Smoking and the risk of lung cancer: susceptibility
with GSTP1 polymorphisms. Epidemiology, 14, 545--551.
20.Wang,Y., Spitz,M.R., Schabath,M.B., Ali-Osman,F., Mata,H. and Wu,X.
(2003) Association between gluthathione S-transferase p1 polymorphisms
and lung cancer risk in Caucasians: a case-control study. Lung Cancer, 40,
Frisach,M.F., Mirault,M.E. and Levade,T. (2002) Gluthathione perox-
idase-1 protects from CD95-induced apoptosis. J. Biol. Chem., 277,
22.Schomberg,S., Rudra,P.K., Noding,R., Skorpen,F., Bjerve,K.S. and
Krokan,H.E. (1997) Evidence that changes in Se-glutathione peroxidase
levels affect the sensitivity of human tumour cell lines to n-3 fatty acids.
Carcinogenesis, 18, 1897--1904.
23.Ratnasinghe,D., Tangrea,J.A., Forman,M.R. et al. (2000) Serum tocopher-
ols, selenium and lung cancer risk among tin miners in China. Cancer
Causes Control, 11, 129--135.
24.Walsh,A.C., Li,W., Rosen,D.R. and Lawrence,D.A. (1996) Genetic
mapping of GLCLC, the human gene encoding the catalytic subunit of
gamma-glutamyl-cysteine synthetase, to chromosome band 6p12 and
characterization of a polymorphic trinucleotide repeat within its 50
untranslated region. Cytogenet. Cell Genet., 75, 14--16.
25.Liu,W., Smith,D.I., Rechtzigel,K.J., Thibodeau,S.N. and James,C.D.
(1998) Denaturing high performance liquid chromatography (DHPLC)
used in the detection of germline and somatic mutations. Nucleic Acids
Res., 26, 1396--1400.
26.Lin,P., Hsueh,Y.M., Ko,J.L., Liang,Y.F., Tsai,K.J. and Chen,C.Y. (2003)
Analysis of NQ01, GSTP1 and MnSOD genetic polymorphisms on lung
cancer risk in Taiwan. Lung Cancer, 40, 123--129.
27.Nyberg,F., Hou,S.M., Hemminki,K., Lambert,B. and Pershagen,G. (1998)
Glutathione S-transferase mu1 and N-acetyltransferase 2 genetic poly-
morphisms and exposure to tobacco smoke in nonsmoking and smoking
lung cancer patients and population controls. Cancer Epidemiol.
Biomarkers Prev., 7, 875--883.
28.Miller,D.P., De Vivo,I., Neuberg,D., Wain,J.C., Lynch,T.J., Su,L. and
Christiani,D.C. (2003) Association between self-reported environmental
tobacco smoke exposure and lung cancer: modification by GSTP1
polymorphism. Int. J. Cancer, 104, 758--763.
29.Lewis,S.J., Cherry,N.M., Niven,R.M., Barber,P.V. and Povey,A.C. (2002)
GSTM1, GSTT1 and GSTP1 polymorphisms and lung cancer risk. Cancer
Lett., 180, 165--171.
30.Bennett,W.P., Alavanja,M.C., Blomeke,G. et al. (1999) Environmental
tobacco smoke, genetic susceptibility and risk of lung cancer in never-
smoking women. J. Natl Cancer Inst., 91, 2009--2014.
31.Sunaga,N., Kohno,T., Yanagitani,N. et al. (2002) Contribution of the
NQ01 and GSTT1 polymorphisms to lung adenocarcinoma susceptibility.
Cancer Epidemiol. Biomarkers Prev., 11, 730--738.
32.Taioli,E., Gaspari,L., Benhamou,S. et al. (2003) Polymorphisms in
CYP1A1, GSTM1, GSTT1 and lung cancer below the age of 45 years.
Int. J. Epidemiol., 32, 60--63.
33.Taniguchi,K., Yang,P., Jett,J., Bass,E., Meyer,R., Wang,Y., Deschamps,C.
and Liu,W. (2002) Polymorphisms in the promoter region of the
neutrophil elastase gene are associated with lung cancer development.
Clin. Cancer Res., 8, 1115--1120.
34.de Andrade,M., Ebbert,J.O., Wampfler,J.A. et al. (2004) Environmental
tobacco smoke exposure in women with lung cancer. Lung Cancer, 43,
35.Visbal,A.L., Williams,B.A., Nichols,F.C. et al. (2004) Gender differences
in non-small cell lung cancer survival: an analysis of 4,618 patients
diagnosed between 1997--2002. Ann. Thoracic Surg., in press.
36.Melton,L.J.,III (1996) History of the Rochester Epidemiology Project.
Mayo Clin. Proc., 71, 266--274.
37.Yang,P., Wentzlaff,K.A., Katzmann,J.A. et al. (1999) Alpha1-antitrypsin
deficiency allele carriers in lung cancer patients. Cancer Epidemiol.
Biomarkers Prev., 8, 461--465.
38.Abdel-Rahman,S.Z., El-Zein,R.A., Anwar,W.A. and Au,W.W. (1996) A
multiplex PCR procedure for polymorphic analysis of GSTM1 and GSTT1
genes in population studies. Cancer Lett., 107, 229--233.
39.Therneau,T.M. and Atkinson,E.J. (1997) An Introduction to Recursive
Partitioning Using the RPART Routines, Technical Report no. 61. Mayo
Clinic, Rochester, MN.
40.Zhang,H. and Bonney,G. (2000) Use of classification trees for association
studies. Genet. Epidemiol., 19, 323--332.
41.Akaike,H. (1974) A new look at the statistical model identification. IEEE
Trans. Automatic Control, 19, 716--723.
42.Sellers,T.A., Potter,J.D. and Folsom,A.R. (1991) Association of incident
lung cancer with family history of female reproductive cancers: the Iowa
Women’s Health Study. Genet. Epidemiol., 8, 199--208.
Glutathione pathway genes in young and old
by guest on June 8, 2013
43.Yang,P., Schwartz,A.G., McAllister,A.E., Swanson,G.M. and Aston,C.E. Download full-text
(1999) Lung cancer risk in families of nonsmoking probands: hetero-
geneity by age at diagnosis. Genet. Epidemiol., 17, 253--273.
44.London,S.J., Daly,A.K., Cooper,J., Navidi,W.C., Carpenter,C.L. and
Idle,J.R. (1995) Polymorphism of glutathione S-transferase M1 and lung
cancer risk among African-Americans and Caucasians in Los Angeles
County, California. J. Natl Cancer Inst., 87, 1246--1253.
45.Garcia-Closas,M., Kelsey,K.T., Wiencke,J.K., Xu,X., Wain,J.C. and
Christiani,D.C. (1997) A case-control study of cytochrome P450 1A1,
glutathione S-transferase M1, cigarette smoking and lung cancer
susceptibility. Cancer Causes Control, 8, 544--553.
46.Moreira,A., Martins,G., Monteiro,M.J. et al. (1996) Glutathione S-
transferase mu polymorphism and susceptibility to lung cancer in the
Portuguese population. Teratog. Carcinog. Mutagen., 16, 269--274.
47.Ratnasinghe,D., Tangrea,J.A., Andersen,M.R., Barrett,M.J., Virtamo,J.,
Taylor,P.R. and Albanes,D. (2000) Glutathione peroxidase codon 198
polymorphism variant increases lung cancer risk. Cancer Res., 60,
Cowan,K.H. (1994) Loss of heterozygosity of the human cystosolic gluta-
thione peroxidase I gene in lung cancer. Carcinogenesis, 15, 2769--2773.
49.Coursin,D.B., Cihla,H.P., Sempf,J., Oberley,T.D. and Oberley,L.W.
(1996) An immunohistochemical analysis of antioxidant and glutathione
S-transferase enzyme levels in normal and neoplasti human lung. Histol.
Histopathol., 11, 851--860.
50.Jaruga,P., Zastawny,T.H., Skokowski,J., Dizdaroglu,M. and Olinski,R.
(1994) Oxidative DNA base damage and antioxidant enzyme activities in
human lung cancer. FEBS Lett., 341, 59--64.
Williams,G.I. and Wild,C.P. (2000) The effect of hOGG1 and
glutathione peroxidase I genotypes and 3p chromosmal loss on
Gnarra,J., Johnson,B. and
8-hydroxydeoxyguasonsine levels in lung cancer. Carcinogenesis, 21,
52.Miller,D.P., Liu,G., De Vivo,I., Lynch,T.J., Wain,J.C., Su,L. and
Christiani,D.C. (2002) Combinations of the variant genotypes GSTP1,
GSTM1 and P53 are associated with an increased lung cancer risk. Cancer
Res., 62, 2819--2823.
53.Mantel,N. and Haenszel,W. (1959) Statistical aspects of the analysis of
data from retrospective studies of disease. J. Natl Cancer Inst., 22,
54.National Cancer Institute (1999) Carcinogenic effects. In Health Effects of
Exposure to Environmental Tobacco Smoke. The Report of the California
Environmental Protection Agency. Smoking and Tobacco Control
Monograph no. 10. National Cancer Institute, Bethesda, MD, Ch. 7,
55.Lang,M. and Pelkonen,O. (1999) Metabolism of xenobiotics and chemical
Susceptibility to Cancer. IARC, Lyon, Ch. 3, pp. 13--22.
56.Pelkonen,O., Raunio,H., Rautio,A. and Lang,M. (1999) Xenobiotic-
metabolizing enzymes and cancer risk: correspondence between genotype
and phenotype. In Vineis,P.,
Caporaso,N.,E., Cuzick,J. and Boffetta,P. (eds) Metabolic Polymorphisms
and Susceptibility to Cancer. IARC, Lyon, Ch. 8, pp. 77--88.
57.Strange,R.C. and Fryer,A.A. (1999) The glutathione S-transferases:
influence of polymorphism on cancer susceptibility. In Vineis,P.,
Boffetta,P. (eds) Metabolic Polymorphisms and Susceptibility to Cancer.
IARC, Lyon, Ch. 19, pp. 231--249.
Received March 9, 2004; revised April 26, 2004; accepted May 27, 2004
P.Yang et al.
by guest on June 8, 2013