Cancer Causes and Control **: **** ****, ****. ***7

**** ****** ******** **********. ******* in the Netherlands.

Polymorphic variation of CYP1A1 is associated with the risk of gastric cardia

cancer: a prospective case cohort study of cytochrome P-450 1A1 and GST enzymes

Mark J. Roth1,*, Christian C. Abnet1, Laura Lee Johnson1, Steven D. Mark2, Zhi Wei Dong3, Philip R. Taylor1,

Sanford M. Dawsey1, You-Lin Qiao3,**

1

NIH, NCI, CCR, Cancer Prevention Studies Branch, Bethesda, MD 20892; 2NIH, NCI, DCEG, Biostatistics Branch,

Bethesda, MD 20892; 3Cancer Institute, Chinese Academy of Medical Science Beijing, P.R. China 100021

Received 18 May 2004: accepted in revised form 18 August 2004

Key words:cytochrome P-450, esophagus, gastrointestinal cancer, glutathione-S-transferase, Phase-I enzyme, Phase-

II enzyme, polycyclic aromatic hydrocarbons, stomach, xenobiotic.

Abstract

Objective: To determine if genetic polymorphisms of CYP1A1, GSTM1, GSTP1, or GSTT1 are associated with an

increased risk of developing esophageal squamous cell carcinoma (ESCC), gastric cardia cancer (GCC), or either in

a high-risk Asian population.

Methods: We conducted a case-cohort analysis with 5 years of prospective follow-up. The analytical cohort

contained 642 individuals who participated in either the Dysplasia Trial (DT) or the General Population (GPT) of

the Nutrition Intervention Trials conducted in Linxian, China, and included 131 cases of ESCC and 90 cases of

GCC. Genotyping analysis was performed on DNA extracted from red blood cells using a PureGene kit (Gentra

Systems, Inc., Minneapolis, MN) and real-time PCR analysis ampli cation (Taq-Man). Relative risks and 95%

con dence intervals were estimated using the case cohort estimator for the Cox proportional hazards models.

p-values from nested models with genotyping variables came from score tests.

Results: The relative risks for developing ESCC, GCC, or either cancer were calculated in the entire analytic cohort

for GSTM1, P1*B (A313G), and T1 and CYP1A1*2A (T3801C) and *2C (A2455G) genotypes, and no signi cant

associations were identi ed. However, because of the di erence in cancer risks between the DT (9.3 cases per 1000

person years) and the GPT (5.3 cases), the analytical cohort was strati ed by trial; the DT participants who were

heterozygous or homozygous for the variant-allele at CYP1A1*2A had a reduced risk for developing GCC

(adjusted RR (95%CI) 0.47 (0.23 1.00) p 0.037).

Conclusions: This study found an association for the CYP1A1*2A variant allele and a reduced risk of GCC in

people at high risk for development of this disease. This nding is consistent with previous studies suggesting that

substrates for the cytochrome P-450 1A1 metabolic pathway, such as polycyclic aromatic hydrocarbons, may be

etiologically signi cant in this high-risk region.

of inherited genetic factors and environmental in uences

Introduction

such as carcinogen exposure [1 6]. This hypothesis is

Regional differences in esophageal squamous cell supported by studies of the association between cancer

carcinoma rates may, in part, result from a combination risk and polymorphisms in genes that metabolize known

and potentially carcinogenic environmental exposures

*

Address correspondence to: Mark J. Roth NIH, NCI, CCR, CPSB, [7 12]. Polymorphisms in these genes affect enzymatic

6116 Executive Boulevard Suite 705, Bethesda, MD 20892-8314, USA.

activities and may alter an individual s ability to

Ph.: 301-***-****; Fax: 301-***-****; E-mail: abpqbb@r.postjobfree.com

metabolize pro-carcinogenic and related compounds,

You-Lin Qiao, Cancer Institute Chinese Academy of Medical

which may change the biologic effect of a given exposure

Sciences, Department of Epidemiology, P.O. Box 2252, Beijing, P.R.

[13].

China 100021. Ph.: 86-10-677*-****; Fax: 86-10-677*-****; E-mail:

abpqbb@r.postjobfree.com

1078 M.J. Roth et al.

Linxian, China, currently known as Linzhou, has May 1991, at the conclusion of the interventions, whole

some of the highest rates of esophageal squamous cell blood was collected from approximately 6000 NIT

participants. We successfully extracted 1.5 lg DNA

carcinoma (ESCC) and gastric cardia adenocarcinoma

(GCC) with a combined incidence of approximately 100 from RBCs of 4005 subjects and these were considered

per 105 person years [1, 2]. Recent studies have eligible for participation in the current study. A

suggested that environmental exposure to carcinogenic strati ed case-cohort design [26 28] was used to select

polycyclic aromatic hydrocarbons (PAHs) may be individuals for inclusion from the cohort of all eligible

etiologically related to these high rates. Supporting this participants. All eligible incident cases of ESCC

(n 131) and GCC (n 90) that occurred between

possibility are the nding of histologic changes in

esophageal cancer resections consistent with high ex- May 1991 and April 1996 were included in the study.

posure to PAH associated air pollution [14]; high During the study period, mortality from all causes and

concentrations of benzo(a)pyrene (B(a)P), a carcino- incidence of all cancers were ascertained by monthly

genic PAH, in local staple foods [15]; high concentra- contact with village health workers, with little or no loss

tions of 1-hydroxypyrene glucuronide, a PAH to follow up. Diagnostic materials for cancers, including

metabolite in the urine of local residents [16]; and the X-ray lms and cytological, pathological and surgical

extensive use of domestic coal-burning stoves from specimens, were collected from local hospitals and from

which a variety of PAHs in soot extracts were recovered a study medical team that provided clinical and

[17]. diagnostic services, and cancers were con rmed by a

The cytochrome P-450 and glutathione-S-transferase committee of international experts [21]. An age- and sex-

enzymes metabolize PAHs formed during the combus- strati ed random sample of all eligible trial participants

tion of organic material and, thus, are among the irrespective of case status was selected to serve as the

reference control group termed the sub-cohort (n 454).

pathways with the greatest potential to modify the

effect of PAH exposure [18]. Cytochrome P-450 1A1 is The six strata included both sexes and three age

the Phase-I enzyme that is most important in metabo- categories (de ned as age at the start of the trial

lically activating B(a)P and other PAHs, and intervention), 60. The within-strata

glutathione-S-transferase M1 (GSTM1) and glu- ratios of control subjects to case subjects for the incident

tathione-S-transferase P1 (GSTP1), are the most site-speci c cancers ranged from 2.3 to 5.8 for ESCC,

important enzymes involved in detoxifying reactive 3.7 7.0 for GCC, and 1.6 2.5 for the combined endpoint

PAH metabolites [13, 19, 20]. (ESCC or GCC).

The current study was conducted to determine if Genotyping analysis was performed at a commercial

genetic polymorphisms in CYP1A1, GSTM1, GSTP1, laboratory (BioServe Biotechnologies Ltd., Laurel,

or GSTT1 alone, or in combination, are associated with MD), using a PureGene kit (Gentra Systems Inc.,

an increased risk of developing ESCC, GCC, or either in Minneapolis, MN) and real-time PCR analysis ampli -

the high-risk population of Linxian, China. The study s cation (Taq-Man; PE Biosystems, Foster City, CA) on

analytical cohort represents a sub-cohort of the Linxian DNA extracted from frozen blood and the CYP1A1,

Nutrition Intervention Trials (NIT), including both the GSTM1, GSTT1, and GSTP1 genotyping was success-

Dysplasia Trial (DT) and the General Population Trial ful in at least 90% of the cases (Table 1). PCR primers

(GPT) [21]. Both trials were randomized, double-blind, and dual-labeled allele discrimination probes were

placebo-controlled trials which tested the effects of designed using Primer Express software (version 1.5;

vitamin/mineral supplements on the rates of ESCC and Perkin-Elmer). All laboratory personnel were blinded to

GCC in Linxian. DT participants (n 3318) had a case control status. A blinded repeat genotyping of

pretrial balloon cytology diagnosis of dysplasia, a 10% of the DNA samples yielded 100% concordance

neoplastic precursor lesion for the development of for all four genes. Both GSTM1 and GSTT1 were coded

ESCC [22, 23], whereas GPT participants (n 29,584) as having the gene either absent or present, while

had no cytologic evidence of dysplasia (10%) or were GSTP1 was coded as homozygous for the common

not tested [21]. We analyzed the most common allele, heterozygous, or homozygous variant. Because of

polymorphic variants of CYP1A1, GSTM1, GSTP1 the reported altered functionality for the B variant allele

and GSTT1. [29, 30] and the small number of individuals possessing

the variant allele, the GSTP1 analysis compared

Material and methods homozygous for the common allele versus the combina-

tion of heterozygotes and homozygous variants. The

A detailed description of the methods and results of the relatively rare GSTP1*C and *D alleles were not

DT and GPT Trials has been reported [21, 24, 25]. In evaluated.

1079

CYP P-450 1A1 and GST polymorphisms in upper GI cancer

Table 1. Genotype polymorphisms with variant alleles Table 2. Linzhou NIT case cohort subject characteristics

Allele Nucleotide changes E ect Subcohort Esophageal Gastric cardia

cancer cancer

CYP1A1*1A None No change

CYP1A1*2A 3801T>C (Msp1) No change Number 454 131 90

CYP1A1*2C 2455A>G I462V Age (year (SD)) 58.8 (7.7) 57.4 (6.9) 60.4 (6.8)

CYP1A1*3 3205T>C No change Male 55.5 50.4 41.1

GSTP1*A None No change Smoking 39.3 36.6 47.8

GSTP1*B 313A>G I105 V Drinking 26.4 22.1 23.3

GP trial 63.7 41.2 46.7

Portions adapted from the human cytochrome P450 (CYP) allele Follow-up 4.6 (1.0) 3.2 (1.6) 3.0 (1.7)

nomenclature committee [http://www.imm.ki.se/CYPalleles/]. time (year (SD))

Covariate de nitions

proportional hazards assumption for each main e ect

Smoking was de ned as a dichotomous variable, never (genotype) with each of the three case de nitions

versus ever smoking for 6 months. Drinking was (esophagus, gastric cardia, or combined) using a time-

de ned as a dichotomous variable, none versus any dependent covariate (genotype*follow-up time). This test

drinking in the previous 12 months. Follow-up time was was non-signi cant (p > 0.05) in all cases.

de ned as days from May 1, 1991 to incident cancer or The statistical analyses include the following: (1)

April 30, 1996. tabulation of subject personal characteristics (Table 2),

(2) tabulation of genotype frequencies by case status

Statistical analysis (Table 3), (3) correlation of genotypes with other

covariates, (4) calculation of crude and adjusted

Pearson correlation coef cients, Tajima s D-prime, relative risks and 95% con dence intervals for esopha-

t-test and and chi-square tests for association were geal, gastric cardia, and combined cancer (Table 4).

calculated and Hardy-Weinberg equilibrium was tested For some genotype-cancer site comparisons we tested

with the chi-square in the subcohort. All p-values were whether the cancer RRs varied by the covariates age,

2-sided tests. sex, tobacco use, alcohol use (i.e., interactions). We did

When analyzing cancers at a speci c site we treated this by comparing a model with the main e ect of a

persons with cancers at other sites as censored at the time covariate (e.g., sex) and a single risk parameter for the

of cancer occurrence. We estimated relative risks (RR) analyte to a model with the main e ect term for the

and 95% con dence intervals (CI) using the case-cohort covariate and separate risk parameters for each sub-

estimator for the Cox proportional hazards models [26 group (e.g., females and males) de ned by the covariate.

28, Preston Epicure 1998 Computer Program]. All Since all models were strati ed on sex and age, there was

estimates came from models strati ed on the six sex-age no main e ect term when testing and estimating a sex

sampling strata. Additional stratum-speci c age terms interaction. Cases and the entire subcohort were

for continuous age were used to adjust for variation included in analyses strati ed by trial.

within age strata. p-values from nested models with A model was also t with variables for all three GST

genotyping variables came from score tests. We tested the genotypes simultaneously, with variables for the two

Table 3. Percent of each genotype by case status

Subcohort Esophageal cancer Gastric cardia cancer

Present Absent Present Absent Present Absent

GST M1 68.0 32.0 68.3 31.2 73.4 26.1

GST T1 46.4 53.6 41.5 58.5 52.3 47.7

Wtb Htb Hvb Wt Ht Hv Wt Ht Hv

GST P 62.4 31.2 6.3 65.4 29.2 5.4 62.2 30.0 7.8

Cyp 1A1*2A 34.3 48.5 17.2 38.8 42.2 19.0 41.8 39.2 19.0

Cyp 1A1*2C 59.3 33.0 7.7 61.5 32.0 6.6 59.3 34.6 6.2

a

The Hardy-Weinberg equilibrium assumption was tested for each genotype using a chi-square test with the known stratum-speci c sampling

weights. p-values were as follows GST p: 0.062, Cyp 1A1*2A = 0.269, Cyp 1A1*2C = 0.040.

b

Wt: Homozygous common allele; Ht: Heterozygous; Hv: Homozygous variant.

1080 M.J. Roth et al.

Table 4. Association between genotype and esophageal squamous cell carcinoma (ESCC) and gastric cardia adenocarcinoma (GCC) for the

entire NIT sub-cohort, General Population Trial (GPT) and the Dysplasia Trial (DT) in Linzhou, China

All casesc pb GPTc pb DTc pb

RRa

GST M1

ESCC 1.15 (0.73 1.81) 0.48 0.79 (0.41 1.53) 0.39 1.39 (0.72 2.69) 0.29

GCC 0.83 (0.49 1.40) 0.45 0.72 (0.34 1.51) 0.36 0.86 (0.39 1.88) 0.63

RRa

GST P

ESCC 0.93 (0.60 1.42) 0.61 1.28 (0.70 2.32) 0.40 0.68 (0.36 1.29) 0.23

GCC 0.98 (0.61 1.58) 0.80 1.04 (0.53 2.07) 0.81 0.94 (0.47 1.89) 0.82

RRa

GST T1

ESCC 1.43 (0.93 2.19) 0.094 1.17 (0.63 2.19) 0.47 1.55 (0.86 2.82) 0.13

GCC 0.80 (0.50 1.28) 0.33 0.93 (0.47 1.84) 0.71 0.72 (0.37 1.41) 0.31

RRa

CYP1A1*2A

ESCC 0.92 (0.59 1.43) 0.38 0.64 (0.35 1.19) 0.13 1.23 (0.66 2.30) 0.47

GCC 0.69 (0.41 1.15) 0.13 1.05 (0.48 2.26) 0.47 0.48 (0.23 1.00) 0.037

RRa

CYP1A1*2C

ESCC 0.98 (0.64 1.51) 0.45 0.71 (0.37 1.36) 0.22 1.34 (0.73 2.47) 0.25

GCC 1.02 (0.62 1.67) 0.73 1.70 (0.85 3.41) 0.11 0.56 (0.26 1.19) 0.12

a

Relative risks and 95%CI calculated using a Cox model strati ed on sex and adjusted for continuous age, smoking, drinking, and trial.

b

p-values are from score tests for the addition of the main e ect term to the base model.

c

Range of cases available for analysis after exclusion for missing PCR or covariate data, All ESCC cases: n = 122 131, GPT n = 49 54, DT

n = 69 76; All GCC cases: n = 81 90, GPT n = 37 42, DT n = 42 48.

CYP1A1 genotypes, or with all ve variables simulta- consequently, further analysis of this genotype was

neously. not conducted.

Association between genotypes and with covariates of

age, sex, smoking, drinking, and trial were examined

and an association between the CYP1A1*2A and *2C

Results

genotypes was identi ed (D 0.64). No other strong

The analytical cohort consisted of 642 individuals, associations were identi ed.

including 347 (54%) males and 295 (46%) females. 275 The crude and adjusted relative risks for developing

(43%) participated in the DT and 367 (57%) partici- ESCC, GCC, or either cancer were calculated in the

pated in the GPT. entire analytic cohort for each of the GSTM1, P1, T1

The 131 cases of ESCC were similar in age, gender, and CYP1A1*2A and *2C genotypes and no signi -

smoking and drinking exposure to the sub-cohort cant associations were identi ed. Because of the

(Table 2). The 90 cases of GCC included more females different risks for cancer in the two trials (Cancers

and more smokers than the other groups. Both the per 1000 person-years, DT 9.3 versus GPT 5.3; (24,25))

ESCC and GCC cases were more likely than the sub- we strati ed our analytic cohort by trial. In the lower

cohort members to be in the DT, a re ection of the risk GPT participants, we found no signi cant associa-

higher rate of cancer in the DT participants (ESCC DT tions between cancer risk and genotype. In the higher

47% (77/165) versus GPT 19% (54/289)); GCC 29% risk DT, however, the participants who were hetero-

(48/165) versus 14% (42/289)). The subcohort was zygous or homozygous for the variant-type allele at

comprised of 64% GPT participants and 36% DT CYP1A1*2A appeared to have a reduced risk of

participants. developing GCC (adjusted RR (95%CI) 0.47 (0.23-

1.00) p 0.037, Table 4). No other signi cant genotype

Genotype analysis showed that the most common

genotypes in the sub-cohort were GSTM1*1 (68%), associations between ESCC risk or risk of the

GSTT1*0 (54%), GSTP1 homozygous for the common combined cancer variable were found in subjects from

allele (62%), CYP1A1*2A heterozygotes (48%) and either trial.

CYP1A1*2C homozygous for the common allele A model with variables for all three GST genotypes

(59%) (Table 3). In addition, a test for Hardy- simultaneously was t and produced estimates for each

Weinberg Equilibrium showed CYP1A1*2C to be variable that were similar to the estimates produced by

marginally not in allelic equilibrium (p 0.040) due the simpler models. The same result occurred when

to under-representation of the variant allele. This tting the two CYP1A1 variables or when tting all 5

population was not polymorphic for CYP1A1*3; genotype variables simultaneously (data not shown).

1081

CYP P-450 1A1 and GST polymorphisms in upper GI cancer

Discussion between GSTT1 and GSTP1 genotypic frequencies and

esophageal cancer. In addition, in a small cross-sectional

We performed a case-cohort study to determine whether study of asymptomatic Linxian inhabitants with a

genetic polymorphisms of CYP1A1, GSTM1, GSTP1, biopsy-proven diagnosis of mild or moderate squamous

or GSTT1 were associated with an increased risk of dysplasia, we found that individuals with a GSTM1*0

developing ESCC, GCC, or either in a high-risk Asian genotype had a tendency for an increased risk of

population during 5 years of prospective follow-up. This esophageal squamous dysplasia [12]. Potential explana-

is the rst prospective study of P-450 related genetic tions for the lack of an association between GSTM1 and

polymorphisms and cancer of the upper aerodigestive cancer in the current study include the fact that not all

tract in this high-risk region. These polymorphic dysplastic foci progress to cancer and the nite number

enzymes were selected based on evidence that environ- of cases identi ed during the follow-up period [41].

mental exposures such as PAHs that are metabolized by The strength of this study s ndings stem from its

the P-450 1A1 pathway and GST enzymes may incorporation of many of the prerequisites considered

contribute to carcinogenesis of upper gastrointestinal necessary for an informative analysis of genetic poly-

cancers in this high-risk region [14 17]. morphisms as set forth by Bartsch et al. [19]. These

Differences in enzymatic activity between various include a clear de nition of the representative study

genotypes are seen in biomarkers of exposure and may population and controls, avoidance of confounding

help explain why, for a given exposure, some individuals created by the use of study subjects of mixed ethnic

have an increased risk for developing cancers of the background, and the inclusion of gene polymorphisms

upper aerodigestive tract [19, 31, 32]. For instance, the shown to lead to altered phenotypic expressions. In

number of PAH-DNA adducts formed in the bronchial addition, the current study s potential for selection bias

tissues of smokers lacking GSTM1 (i.e., GSTM1*0) can is limited by its prospective design and subsequent use of

be up to 100-fold higher than smokers possessing incident rather than prevalent cases.

GSTM1 (GSTM1*1) [33]. In addition, studies of GSTP1 However, as with any study, these results must be

allelotypes nd isoenzymes with a valine in position 105 considered in the context of potential design related

(GSTP1*B) to more e ectively metabolize the diol weaknesses. In the current setting, these include the

epoxides of polycyclic aromatic hydrocarbons and those limited number of cases identi ed during the follow-up

with a A114V transition (GSTP1*C) in the presence period and the inability to assess cytochrome P-450 and

of V105 to more e ciently metabolize benzo(a)pyrene GST related factors, other than genotype, that may

[29, 30]. GSTT1 is also in the metabolic pathway modify an individual s risk of disease. For example,

for several PAHs [20]. This is illustrated by the possessing an inducible form of the CYP1A1 enzyme

association of GSTT1*1 with increased levels of can result in an increase in the biologically effective dose

excreted 1-hydroxyprene glucuronide [34] in addition for a given exposure and has been associated with an

to an inverse association between erythrocytic activity increased risk for bronchial, laryngeal, and oral cavity

and DNA adducts in mononuclear leukocytes [35]. tumors in smokers [19, 31]. Similarly, some exogenous

The current study results show an association between chemicals (e.g., NSAIDS, some pesticides) can induce

CYP1A1*2A genotype and risk of GCC among the and locally increase the expression of GSTs [20, 42, 43]

higher-risk participants of the Linxian Dysplasia Trial and this induction may also a ect an individual s risk

(adjusted RR (95%CI) 0.47 (0.23 1.00)). This nding is [44, 45]. This inducibility underlies the importance of

consistent with in vitro studies that identify di erences quantifying the relevant exposures and is exempli ed by

among the catalytic activity of the CYP1A1 variants, Lan et al. [46] population based case control study of

with the highest activity identi ed for the wild-type lung cancer in a coal using, high-risk lung cancer region

enzyme, followed by CYP1A1*4 (T461N) (approx. in China in which a 2.4-fold increased risk was found

60%) and CYP1A1*2 (I462V) (approx. 40%) [36 39]. only among the heavy coal users (>130 tons) with the

However, our results show no e ect of the polymorph- GSTM1*0 genotype. The potential signi cance of such

isms measured here on ESCC risk or on GCC in the characterization in Linxian, given its extensive use of

total analytic group. coal for heating and cooking, is that identi cation of the

This prospective study s negative GSTM1 results heavy coal users in Linxian may be necessary to identify

contrast with those from a case-control study conducted a genotype associated with elevation in risk. Yet, the

in the same population by Tan et al. [40]. Tan et al. current study, and the original NIT upon which it is

found the GSTM1 non-null and +/0) genotype based, were not designed to exhaustively characterize

to be over-represented in cancer cases. However, similar amount (dose) and length (duration) of potentially

to our results, Tan did not identify an association signi cant exposures, and, therefore, is unable to

1082 M.J. Roth et al.

9. Gao C, Takezaki T, Wu J, et al. (2002) Interaction between

accurately evaluate these potential factors for hetero-

cytochrome P-450 2E1 polymorphisms and environmental factors

geneity in its assessment of risk. It is possible that people

with risk of esophageal and stomach cancers in Chinese. Cancer

at highest risk, as re ected in a cytology nding of Epidemiol Biomarkers Prev 11: 29 34.

dysplasia, may be the most exposed to PAHs, but this 10. Tan W, Chen GF, Xing DY, Song CY, Kadlubar FF, Lin DX

remains to be shown. (2001) Frequency of CYP2A6 gene deletion and its relation to risk

of lung and esophageal cancer in the Chinese population. Int J

In summary, there continues to be considerable

Cancer 95: 96 101.

interest in the interaction between polymorphisms of

11. Smith TJ, Liao A, Wang LD, et al. (1998) Characterization of

xenobiotic metabolizing enzymes and exposure to xenobiotic-metabolizing enzymes and nitrosamine metabolism in

environmental carcinogens and their effect on disease the human esophagus. Carcinogenesis 19: 667 672.

risk. For those hoping to lessen the morbidity and 12. Roth MJ, Dawsey SM, Wang G, et al. (2000) Association between

GSTM1*0 and squamous dysplasia of the esophagus in the high

mortality of high-risk populations this interest stems

risk region of Linxian, China. Cancer Lett 156: 73 81.

from the promise of successfully identifying etiologic

13. Nebert DW, Ingelman-Sundberg M, Daly AK (1999) Genetic

factors so that they can be avoided and identifying epidemiology of environmental toxicity and cancer susceptibility:

individuals at increased risk for disease. The current human allelic polymorphisms in drug-metabolizing enzyme genes,

study represents the rst prospective examination of the their functional importance, and nomenclature issues. Drug Metab

Rev 31: 467 487.

association between esophageal and gastric cardia

14. Roth MJ, Guo-Qing W, Lewin KJ, et al. (1998) Histopathologic

cancer and polymorphisms in CYP1A1 and GST

changes seen in esophagectomy specimens from the high-risk

enzymes in the high-risk population of Linxian. This region of Linxian, China: potential clues to an etiologic exposure?

study nds the CYP1A1*2A variant allele may reduce Hum Pathol 29: 1294 1298.

the risk of GCC in people at highest risk for developing 15. Roth MJ, Strickland KL, Wang GQ, Rothman N, Greenberg A,

Dawsey SM (1998) High levels of carcinogenic polycyclic aromatic

this disease. This is consistent with previous studies that

hydrocarbons present within food from Linxian, China may

suggest that substrates for the cytochrome P-450 1A1

contribute to that region s high incidence of oesophageal cancer.

metabolic pathway, such as PAHs, may be etiologically Eur J Cancer 34: 757 758.

signi cant in this high-risk region. Future larger studies 16. Roth MJ, Qiao Y-L, Rothman N, et al. (2001) High urine

including a comprehensive and quantitative environ- 1-hydroxypyrene glucuronide concentrations in Linxian, China,

an area of high risk for squamous oesophageal cancer. Biomarkers

mental analysis in conjunction with a genotypic evalua-

6: 381 386.

tion are still needed to identify potential variability in

17. Wornat MJ, Ledesma EB, Sandrowitz AK, et al. (2001) Polycyclic

the induction of these genes and an individual s level of aromatic hydrocarbons identi ed in soot extracts from domestic

relevant exposures. coal-burning stoves of Henan Province, China. Environ Sci

Technol 35: 1943 1952.

18. Smith G, Stanley LA, Sim E, Strange RC, Wolf CR (1995)

References Metabolic polymorphisms and cancer susceptibility. Cancer Surv

25: 27 65.

19. Bartsch H, Nair U, Risch A, Rojas M, Wikman H, Alexandrov K

1. Munoz N, Buiatti E (1996) Chemoprevention of oesophageal

(2000) Genetic polymorphism of CYP genes, alone or in

cancer. IARC Sci Publ 27 33.

combination, as a risk modi er of tobacco-related cancers. Cancer

2. Ke L (2002) Mortality and incidence trends from esophagus cancer

Epidemiol Biomarkers Prev 9: 3 28.

in selected geographic areas of China circa 1970-90. Int J Cancer

20. Landi S (2000) Mammalian class theta GST and differential

102: 271 274.

susceptibility to carcinogens: a review. Mutat Res 463: 247 283.

3. Hu N, Dawsey SM, Wu M, et al. (1992) Familial aggregation of

21. Li B, Taylor PR, Li JY, et al. (1993) Linxian nutrition intervention

oesophageal cancer in Yangcheng County, Shanxi Province,

trials. Design, methods, participant characteristics, and compli-

China. Int J Epidemiol 21: 877 882.

ance. Ann Epidemiol 3: 577 585.

4. Hu N, Dawsey SM, Wu M, Taylor PR (1991) Family history of

22. Liu SF, Shen Q, Dawsey SM, et al. (1994) Esophageal balloon

oesophageal cancer in Shanxi Province, China. Eur J Cancer 27:

cytology and subsequent risk of esophageal and gastric-cardia

1336.

cancer in a high-risk Chinese population. Int J Cancer 57: 775 780.

5. Hu N, Roth MJ, Emmert-Buck MR, et al. (1999) Allelic loss in

23. Dawsey SM, Yu Y, Taylor PR, et al. (1994) Esophageal cytology

esophageal squamous cell carcinoma patients with and without

and subsequent risk of esophageal cancer: a prospective follow-up

family history of upper gastrointestinal tract cancer. Clin Cancer

study from Linxian, China. Acta cytol 38: 183 192.

Res 5: 3476 3482.

24. Li JY, Taylor PR, Li B, Dawsey S, et al. (1993) Nutrition

6. Yang CS (1980) Research on esophageal cancer in China: a review.

Intervention Trials in Linxian, China: Multiple vitamin/mineral

Cancer Res 40: 2633 2644.

supplementation, cancer incidence, and disease-speci c mortality

7. Nebert DW (2000) Drug-metabolizing enzymes, polymorphisms

among adults with esophageal dysplasia. J Natl Cancer Inst 85:

and interindividual response to environmental toxicants. Clin

1492 1498.

Chem Lab Med 38: 857 861.

25. Blot WJ, Li JY, Taylor PR, et al. (1993) Nutrition intervention

8. Wang AH, Sun CS, Li LS, Huang JY, Chen QS (2002) Relation-

trials in Linxian, China: supplementation with speci c vitamin/

ship of tobacco smoking CYP1A1 GSTM1 gene polymorphism

mineral combinations, cancer incidence, and disease-speci c

and esophageal cancer in Xi an. World J Gastroenterol 8: 49 53.

1083

CYP P-450 1A1 and GST polymorphisms in upper GI cancer

mortality in the general population. J Natl Cancer Inst 85: 37. Schwarz D, Kisselev P, Honeck H, Cascorbi I, Schunck WH,

1483 1492. Roots I (2001) Co-expression of human cytochrome P4501A1

26. Prentice RLA (1986) A case cohort design for epidemiologic (CYP1A1) variants and human NADPH-cytochrome P450

cohort studies and disease prevention trials. Biometrika 73: 1 11. reductase in the baculovirus/insect cell system. Xenobiotica 31:

27. Self SG, Prentice RLA (1988) Asymptotic distribution theory and 345 356.

ef ciency results for case-cohort studies. Annals of Statistics 16: 38. Schwarz D, Kisselev P, Cascorbi I, Schunck WH, Roots I (2001)

64 81. Differential metabolism of benzo[a]pyrene and benzo[a]pyrene-7,8-

28. Mark SD, Katki H (2001) In uence function based variance dihydrodiol by human CYP1A1 variants. Carcinogenesis 22:

estimation and missing data issues in case-cohort studies. Lifetime 453 459.

Data Analysis 7: 331 334. 39. Pavanello S, Clonfero E (2000) Biological indicators of genotoxic

29. Hu X, Xia H, Srivastava SK, et al. (1997) Activity of four allelic risk and metabolic polymorphisms. Mutat Res 463: 285 308.

forms of glutathione S-transferase hGSTP1-1 for diol epoxides of 40. Tan W, Song N, Wang GQ, et al. (2000) Impact of genetic

polycyclic aromatic hydrocarbons. Biochem Biophys Res Commun polymorphisms in cytochrome P450 2E1 and glutathione

238: 397 402. S-transferases M1, T1, and P1 on susceptibility to esophageal

30. Eaton DL, Bammler TK (1999) Concise review of the glutathione cancer among high-risk individuals in China. Cancer Epidemiol

S-transferases and their signi cance to toxicology. Toxicol Sci 49: Biomarkers Prev 9: 551 556.

156 164. 41. Dawsey SM, Lewin KJ, Wang, GQ, et al. (1994) Squamous

31. Nebert DW, McKinnon RA, Puga A (1996) Human drug- esophageal histology and subsequent risk of squamous cell

metabolizing enzyme polymorphisms: effects on risk of toxicity carcinoma of the esophagus. A prospective follow-up study from

and cancer. DNA Cell Biol 15: 273 280. Linxian, China. Cancer 74: 1686 1692.

32. Nimura Y, Yokoyama S, Fujimori M, et al. (1997) Genotyping of 42. van Lieshout EM, Posner GH, Woodard BT, Peters WH (1998)

the CYP1A1 and GSTM1 genes in esophageal carcinoma patients Effects of the sulforaphane analog compound 30, indole-3-

with special reference to smoking. Cancer 80: 852 857. carbinol, D-limonene or relafen on glutathione S-transferases

33. Bartsch H (1996) DNA adducts in human carcinogenesis: and glutathione peroxidase of the rat digestive tract. Biochim

etiological relevance and structure-activity relationship. Mutat Biophys Acta 1379: 325 336.

Res 340: 67 79. 43. van Lieshout EM, Tiemessen DM, Peters WH, Jansen JB (1997)

34. Hong YC, Leem JH, Park HS, et al. (1999) Variations in urinary Effects of nonsteroidal anti-in ammatory drugs on glutathione

1-hydroxypyrene glucuronide in relation to smoking and the S-transferases of the rat digestive tract. Carcinogenesis 18: 485 490.

modi cation e ects of GSTM1 and GSTT1. Toxicol Lett 108: 44. Spitz MR, Duphorne CM, Detry MA, et al. (2000) Dietary intake

217 223. of isothiocyanates: evidence of a joint e ect with glutathione

35. Popp W, Vahrenholz C, Schell C, et al. (1997) DNA single strand S-transferase polymorphisms in lung cancer risk. Cancer Epidemiol

breakage, DNA adducts, and sister chromatid exchange in Biomarkers Prev 9: 1017 1020.

lymphocytes and phenanthrene and pyrene metabolites in urine 45. Zhang Y, Talalay P (1998) Mechanism of differential potencies of

of coke oven workers. Occup Environ Med 54: 176 183. isothiocyanates as inducers of anticarcinogenic Phase 2 enzymes.

36. Chernogolov A, Behlke J, Schunck WH, Roots I, Schwarz D Cancer Res 58: 4632 4639.

(2003) Human CYP1A1 allelic variants: baculovirus expression 46. Lan Q, He X, Costa DJ, et al. (2000) Indoor coal combustion

and puri cation, hydrodynamic, spectral, and catalytical proper- emissions, GSTM1 and GSTT1 genotypes, and lung cancer risk: a

ties and their potency in the formation of all-trans-retinoic acid. case-control study in Xuan Wei, China. Cancer Epidemiol

Protein Expr Purif 28: 259 269. Biomarkers Prev 9: 605 608.

Contact this candidate