the glutathione s-transferase p polymorphism as a marker ... · gst-u activity toward tso is...

[CANCER RESEARCH 53. 2313-2318. May 15. 1993]

The Glutathione S-Transferase p Polymorphism as a Marker for Susceptibility toLung Carcinoma1

Valle Nazar-Stewart,2 Arno G. Motulsky, David L. Eaton, Emily White, Sigrid K. KÃ¶rnung, Zhong-Tai Leng,

Pat Stapleton, and Noel S. Weiss

Departments of Epidemiology Â¡V.N. S.. E. W., Z. T. L. N. S. W.Â¡,Medicine and Genetics Â¡A.G. M.. S. K. H.Â¡.and Environmental Health Â¡D.L E.. P. S. I. University ofWashington. Seattle 98195, and the Fred Hutchinson Cancer Research Center. Seattle 9SI94 Â¡V.N. S.. E. W.. N. S. W.Â¡,Washington

ABSTRACT

Glutathione S-transferase (GST) enzymes detoxify carcinogens in to

bacco smoke. Interindividual variation in GST function may be related todifferences in risk for smoking-related cancer. Leukocytes from 50% ofCaucasians lack GST activity toward frans-stilbene oxide (TSO), due to adeletion of the gene for the GST-M enzyme. Presence of GST-TSO activity

in leukocytes has been associated with low risk for lung cancer amongcigarette smokers. We sought to determine whether GST activity in lungtissue is determined by the same gene polymorphism and whether it isassociated with risk for lung cancer.

Subjects were cigarette smokers, identified at the time of lung resectionor autopsy in Seattle hospitals. Uninvolved lung tissue was obtained from35 patients with lung carcinoma and 43 control patients and assayed forGST-M activity with TSO, for the presence of the GST-M gene product withan inununological assay, and for the GST-u gene with Southern blotting.Mailed questionnaires were used to collect information on subjects' smok

ing histories and exposures which might alter enzyme activity.Interindividual results from the three assays correlated well. Smokers

with high GST-TSO enzyme activity present in their lung tissue had a

lower risk for lung carcinoma than did smokers with no or low activity(relative risk = 0.30; 95% confidence interval, 0.11-0.79), as did smokerswith GST-M antigen identified in lung tissue versus those with no antigen(relative risk = OJO; 95% confidence interval, 0.11-0.79). Smokers withboth maternal and paternal copies of GST-u DNA (n = 7) had a lowercancer risk than smokers lacking GST-M DNA (n = 30; relative risk =0.35; 95% confidence interval, 0.06-2.10). High GST-M activity appeared

to be associated with a greater decrease in lung cancer risk among 38heavy cigarette smokers (relative risk = 0.15; 95% confidence interval,0.03-0.64) than among 38 light smokers (relative risk = 0.61; 95% confidence interval, 0.14-2.60).

Presence or absence and number of copies of the GST-M 8ene appear todetermine activity of the GST-M enzyme in lung. Smokers with the GST-Menzyme have approximately one-third of the risk for lung carcinoma of

smokers without the enzyme.

INTRODUCTION

PAHs3 are believed to be an important class of oncogenic com

pounds in tobacco smoke and to contribute to lung carcinoma incigarette smokers (1). PAHs occur as inactive procarcinogens in tobacco smoke. Once inhaled, certain PAHs undergo enzyme-mediated

oxidation to form active electrophilic epoxide metabolites, which canbind with DNA (1). This oxidation is catalyzed by the cytochromeP450 enzymes. Formation of PAH-DNA adducts modifies normalDNA structure and function (2) and has been associated with carci-

Received 11/17/92; accepted 3/11/93.The costs of publication of this article were defrayed in part by the payment of page

charges. This article must therefore be hereby marked advertisement in accordance with18 U.S.C. Section 1734 solely to indicate this fact.

1Supported by Grants R35 CA39779 and T32 CA09168 from the National CancerInstitute. NIH Grant ES-05780, and a Dana Foundation Grant in Ecogenetics and Envi

ronmental Health.2 To whom requests for reprints should be addressed, at Department of Epidemiology

SC-36, School of Public Health and Community Medicine, University of Washington,

Seattle, WA 98195.3 The abbreviations used are: PAHs, polycyclic aromatic hydrocarbons; GST, glu-

tathione S-transferase; GST1, GST-n gene; TSO, frani-stilbene oxide; RR, relative risk;95% CI, 95% confidence interval.

nogenic potency in several animal and Â¡nvitro systems (3-7). How

ever, active electrophiles, including PAH metabolites, may undergo analternative reaction, in which they are enzymatically conjugated withintracellular glutathione to form inactive, water-soluble metabolites.

This detoxification reaction is catalyzed by the GSTs. It has beenhypothesized that interindividual variation in the activity of enzymeswhich catalyze either oxidation of PAHs (aryl hydrocarbon hydroxy-lase, Refs. 8-13; debrisoquine hydroxylase, Refs. 14-16) or conjugation of mutagenic PAH metabolites with glutathione via GSTs (17-19)

may account for differences among cigarette smokers in risk forsmoking-induced lung cancer.

A genetic polymorphism for the glutathione S-transferase u enzymehas been described in humans (20-21). GST-u activity in leukocytes,

measured with TSO, varies several thousandfold across individuals inthe Caucasian populations tested (22). Family studies have shown thatGST-u activity toward TSO is determined by a single gene (GST1)and that individuals lacking detectable GST-u activity in leukocytes

are homozygous for the null alÃeleat the GST1 locus (22). Because thegene frequency for the deletion is so high (q2 = 0.5), most individuals

with GST-u activity are hÃ©tÃ©rozygotes(2pq = 0.42) and a few individuals with high activity are homozygous for the active alÃele(p1 =

0.09) (23). Recent evidence has raised the possibility that three alÃelesmay occur at the GST1 locus (including a null alÃele),resulting in sixgenotypes and four or more phenotypes (24-28).

Low or no GST-u activity in mononuclear leukocytes, as measured

with TSO, has been reported to be associated with approximately threetimes higher risk for lung cancer among cigarette smokers comparedwith smokers who have some GST-TSO activity (17-18). This association was stronger for heavy smokers with over 30 pack-years ofexposure (RR = 3.2) than for light smokers with 10 to 30 pack-years(RR = 1.9) and for adenocarcinoma (RR = 3.2) than for otherhistolÃ³gica! types (RR = 2.5) (18). In contrast, several of us havereported no difference in GST-TSO activity in leukocytes from 66lung cancer patients and 120 randomly selected population-based

controls (19). However, when that comparison was limited to heavysmokers (at least 20 pack-years), presence of GST-TSO activity wasassociated with reduced risk for lung carcinoma (RR = 0.60), al

though this result could have been due to chance (95% confidenceinterval, 0.3-1.1) (19). Others (29) have reported the absence of anassociation between GST-u genotype, measured in leukocytes, and

risk for lung cancer of all histolÃ³gica! types. However, in that studythere was an association between genotype and risk when cases werelimited to patients with adenocarcinoma. Most recently another studyreported failure to find a "trend of overrepresentation of GST classu-deficient individuals" (measuring genotype and phenotype) among

49 lung cancer patients when compared with 96 hospital controlpatients (30), but this statement was not supported by any presentationof data. Information regarding the selection or smoking history ofcontrol subjects was not presented for either of these two studies.

Interpretation of the results from these studies of GST activity maybe limited because GST activity was measured in leukocytes and notin lung tissue. GST enzymes are differentially expressed across humantissues (31, 32), and measurements of indicators for tobacco smoke

2313

on June 20, 2020. © 1993 American Association for Cancer Research. cancerres.aacrjournals.org Downloaded from

http://cancerres.aacrjournals.org/

GST-M POLYMORPHISM AND RISK FOR LUNG CANCER

metabolism in blood cells may not reflect metabolic activity in a tissuethat is a target for smoking-related tumors (26, 33, 34). One report haseven suggested that the GST-ja enzyme in human lung may be differ

ent from that in human leukocytes, based on mobility in sodiumdodecyl sulfate-polyacrylamide gels (35).

Our study was designed to determine whether GST-p activity in

lung tissue is determined by the same gene polymorphism that isexpressed in leukocytes and whether it is associated with risk for lungcancer. We measured GST in lung tissue from subjects with andsubjects without lung carcinoma, by performing simultaneous measurements of GST-Menzyme activity with rra/is-stilbene oxide, GST-ugene product expression with an immunological assay, and the GST-u

gene with Southern blotting. We sought to determine whether anymeasured differences in risk for lung carcinoma were due to interindividual differences in the presence of the gene, in expression of theenzyme in lung tissue, or in activity of the GST-u enzyme.

SUBJECTS AND METHODS

Subjects. Between January 1988 and January 1990 we selected subjectsfrom among persons who underwent major lung surgery and deceased individuals who were autopsied at one of several Seattle hospitals or at theSeattle-King County Medical Examiners' Office, as well as individuals whose

organs were donated to the Northwest Tissue Center. All eligible subjects wererequired to have a history of cigarette use (at least 100 cigarettes over theirlifetime), to be 30 years old or older, have a known address or phone numberavailable for either the patient or the next of kin, and have available at least 0.5g of fresh uninvolved lung tissue. The lung specimen must have been obtainedwithin 48 h of death for deceased subjects. Informed consent was obtainedfrom all subjects or their next of kin.

Our cases had histologically confirmed primary carcinoma of the lung orbronchus (International Classification of Disease-Oncology, site codes 162.2-

162.9) as their underlying cause of death or cause for surgery, with the following histologie diagnoses: squamous cell carcinoma (n = 13); adenocarci-noma (n = II); non-small cell carcinoma (n = 8); adenosquamous carcinoma(n = 1); small cell carcinoma (n = 1); and large cell carcinoma (n = 1). All

35 eligible cases or their next of kin agreed to participate in the study.Our controls were required to have a cause of death or cause for surgery

which is epidemiologically unrelated to tobacco smoking and be within the agerange of the cases (30 to 80 years of age). More than one-half of the 115

potential controls identified at surgery or autopsy were found to be ineligibleas a result of never having been a cigarette smoker, having a final diagnosis ofa smoking-related cause of death, having no next of kin, or being outside the

age range of the cases. Causes of death among excluded potential controls thatare or may be smoking-related included cardiac arrhythmia, congestive heart

failure, aortic aneurysm rupture, myocardial infarction or cardiac ischemia,atherosclerosis, arteriosclerosis, cerebrovascular accident, sarcoid-type granu-lomatous infiltration, squamous cell carcinoma of the skin, and lung carcino-

sarcoma. Three of the remaining 47 potentially eligible controls did not respond to the questionnaire (thus their smoking history and their eligibility wereunknown), and one specimen was lost, resulting in a total of 43 controls.Causes of death and causes for surgery among the eligible controls includedneoplasms (n = 16, including two tumors with unknown primary sites that

were metastatic to the lung); diseases of the circulatory system includingmyocarditis, cardiorespiratory failure, cardiac transplant rejection, cardiome-galy. rheumatic heart disease, and coagulopathy (n = 6); infection (n = 4,including one viral pneumonia); organ failure (n = 6); aspiration or relatedcomplications (n = 4); conditions of the central nervous system (n = 2); and

other conditions including angioimmunoblastic lymphadenopathy, dehydration, hernia repair surgery, colon perforation, and lung hamartoma (n = 5).

From each subject we collected at least 0.5 g of fresh lung tissue, whichappeared to be uninvolved upon gross examination. The specimen was obtained as often as possible from the right upper lobe (60%), since this is theorigin of the slight majority of lung carcinomas, and the remaining specimenswere taken from various other lobes. Specimens were snap-frozen in liquidnitrogen and stored at -70Â°C until assayed.

We mailed questionnaires to our subjects or their next of kin to collectdetailed data on each subject's history of tobacco use (cigarettes, pipes, cigars).

demographic characteristics, and exposures in the 3 weeks prior to specimencollection which might have induced enzyme production and affected enzymemeasurements (tobacco and alcohol use; use of sedatives, barbiturates, anti-

convulsants, and illicit drugs; chemotherapy or radiation therapy; weight loss;and time elapsed since diagnosis of the cause of death or surgery as a measureof cachexia). Records of the pathology services and hospital charts were usedto collect data on time of death, the underlying reason for surgery or cause ofdeath, any comorbid conditions, and whether any abnormal lung conditionswere noted.

Enzyme Measurements. Lung tissues were homogenized and cytosolicfractions were collected by centrifugation. Blood tissue specimens were notobtained. All assays were performed with no knowledge regarding the subject's

disease status. We measured GST activity with TSO using methods reported bySeidegard and DePierre (36). Samples were diluted to a protein concentrationof 1-2 ug/ml with 175 mM sodium phosphate buffer (pH 7.2). The reaction

mixture contained 90 jil lung sample, 5 ul reduced glutathione (final concentration, 5 ITIM),and 5 |il tritiated /rans-stilbene oxide (final concentration, 50UM).Two duplicate samples were incubated at 37Â°Cfor 5, 15, and 30 min. The

linear increase in TSO-glutathione conjugate over 30 min was averaged for twoduplicate samples and expressed as pmol TSO-glutathione conjugate/mgprotein/min. ['Hjrrani-stilbene oxide was generously provided by Dr. Bruce

Hammock (University of California. Davis, CA). The mean amount of timeover which specimens were frozen at -70Â°C before assay with TSO was 25

weeks (Â±17 SD).We measured the occurrence of GST-(j antigen in lung homogenates with a

solid-phase sandwich enzyme-linked immunosorbent assay, originally de

signed for use with whole blood and sold as a kit by Medlabs (n Kit; Unit 1C.Stillorgan Industrial Park, Stillorgan, Co. Dublin, Ireland). Homogenized lungspecimens were incubated in wells coated with standard GST-|a antibody, rabbitanti-human antibody specific for the bound antigen was added, and peroxidase-conjugated antibody (goat anti-rabbit) specific for the primary (rabbit) anti

body was added. Hydrogen peroxide and chromagen solutions were added.Specimens were characterized as positive for GST-p antigen if a blue perox-

idase reaction product was seen and negative if no blue color was seen. Foreach specimen assayed, a positive control consisted of lyophilized, lysedhuman blood positive with respect to GST-(j antigen, and a negative controlconsisted of similarly treated human blood with no detectable GST-p antigen.

Southern blot hybridization was performed on DNA from homogenizedlung specimens using the restriction enzyme Sst. Specimens were placed in 10ml Poncz buffer and incubated for 30 to 60 min at 37Â°C.Two hundred ul

proteinase K (10 mg/ml) were added, and specimens were incubated overnightat 37Â°Cand then centrifuged at 2800 rpm for 20 min. DNA extraction and

radioautography were carried out with standard methods (37). The complementary DNA GST-M pMP98 (p GST-18) (29) probe was generously provided

by Dr. Julie Moss in the laboratory of Dr. C. R. Wolf (Laboratory of MolecularPharmacology and Drug Metabolism, Edinburgh, Scotland). Four bands couldbe seen on the Southern blots. Bands I and 2 did not differ in intensity. Band3 was completely absent in homozygotes for the GST-|j deletion and was

diminished in intensity in hÃ©tÃ©rozygotes.The ratio of density between bands 3and 4 served to distinguish hÃ©tÃ©rozygotesfor the GST-(a deletion (single gene

copy) from homozygotes without the deletion (two gene copies). HÃ©tÃ©rozygoteand homozygote (both null and double gene) human controls were used oneach blot. The deletion polymorphism also was apparent using the followingrestriction enzymes: Hinc, Xba, Pst, and double digest with BamH-EcoR.

Densitometry was carried out with a Hoefer densitometer and analyzed with aMacintosh computer. Because we had limited amounts of tissue from some ofour subjects, we could perform Southern blot assays only for a subset of oursubjects (n = 54). Each of the three assays was performed on all specimens by

a different person.Statistical Analysis. To determine whether we had obtained enzyme mea

surements from our autopsied subjects that were valid estimates of their an-

temortem levels, we examined relationships between several indicators ofproteolysis (occurring either between death and specimen collection or between collection and assay when the specimens were frozen) and GST measurements. We also evaluated intrasubject agreement between results from

each of the assays.Risks for lung carcinoma associated with high enzyme levels relative to low

enzyme levels were estimated with odds ratios and 95% confidence intervals,using unconditional logistic regression to control for the possible confounding

2314



GST-n POLYMORPHISM AND RISK FOR LUNG CANCER

influence of other risk factors for lung cancer. Estimates were derived withmaximum likelihood estimation (38). Estimates for trends in risk were used todescribe the average change in relative risk which occurs in moving from oneGST category to the next, and the statistical significance of linear trends in riskwas calculated using an extended x2 procedure. Relative risks were estimated

for several subgroups of subjects by gender, ethnic background (white andnonwhite), age (<64 or >64 years), smoking history (<54 or >54 pack-years),

and histolÃ³gica! type of lung carcinoma .

RESULTS

Subjects. The majority of cases (n = 24, 68.6%) were identified atsurgery, and the majority of controls (n = 38, 88.4%) were identified

at autopsy. The mean age (and SD) in each of the two groups wassimilar: 65 (Â±8) years among cases and 63 (Â±9) among controls. Agreater proportion of the cases than controls were male (83% versus74%). Five subjects were nonwhite: two cases (one black and oneAsian) and three controls (two blacks and one Native American). Wehad detailed information on smoking history from all but two (97.4%)of our subjects. Cases had smoked cigarettes more intensely (meanand SD of 80 Â±43 pack-years) than had controls (52 Â±52 pack-

years). One case and 12 (28%) controls had smoked less than 20pack-years. However, none of our estimates of relative risk waschanged by adjustment for pack-years, gender, race (examined by

exclusion and inclusion of nonwhite subjects), or the length of timewhich elapsed between the last cigarette and collection of the lungspecimen.

Validity of Laboratory Measurements. To determine whetherour enzyme measurements were lowered by proteolysis in specimensobtained postmortem, we compared the distributions of GST-u activity

and antigen measurements among deceased controls with those amonglive controls (and made the same comparison among cases). Deceasedand living subjects had broadly similar distributions, although thesmall number of subjects (particularly living controls) limits anyinferences (Table 1). To a small extent, adjustment for vital statusmade the observed associations stronger. Additional analyses indicated that enzyme measurements did not decrease with increasinginterval between death and collection of the specimen at autopsy.There was also no association between the length of time our specimens were frozen and subsequent enzyme measurements.

Intrasubject results from the activity, protein, and DNA assays werein close agreement (Table 2). Absence of enzyme activity correlatedwith the absence of GST-u genes. For a few subjects, there was someinconsistency in the assignment of either heterozygous or homozy-

gous GST status by the activity and Southern blot assays. This smallamount of misclassification was expected because definite antimodesthat distinguish one phenotype from another are not well established

Table 1 Distribution of GST markers according to vital status andcase-control status

Table 2 Agreement berufen laboratory measurements (number of subjects)

GST-Mu antigen No. of GST 1 genes

Cases Controls

Autopsy(n= 11)Surgery (n = 24)Autopsy (n = 38)Surgery (n = 5)

GST-TSO activity (pmol/mg/min)"

<270 8(72.7%) 18(75.0%) 18(47.4%) 2(40.0)>270 3(27.3%) 6(25.0%) 20(52.6%) 3(60.0)

GST-u antigen

Absent 7(63.6%) 19(79.2%) 20(52.6%) 0Present 4(36.4%) 5(20.9%) 18(47.4%) 5(100.0%)

GSTl-u gene

No copies 6(66.7%) 10(62.5%)One copy 3 (33.3%) 4 (25.0%)Two copies 0 2(12.5%)

14(53.8%) 08 (30.8%) 2 (66.7%)4(15.4%) 1(33.3%)

Absent Present 0

GST-TSO activity(pmol/mg/min)"0-270271-10001001-2740GST-u

antigenAbsentPresent387181772910282313151203416

*Categories determined by visual inspection of distribution of activity among controls.

a Categories determined by visual inspection of distribution of activity among controls.

(and may vary across tissues). Agreement between the commercialGST-u immunoassay and the other assays was less satisfactory, pos

sibly because the immunoassay was designed for use with blood cellsor because some of the other GST enzymes cross-reacted with theGST-ja enzyme-linked immunosorbent assay.

The generally good agreement between the genotype and phenotype(activity) assigned to each subject (Table 2) also suggests that ourenzyme activity measurements were representative of true baselineactivities in our subjects, uninfluenced by exogenous exposures, presence of a lung neoplasm or other disease, or smoking habits. Furthermore, adjustment for recent exogenous exposures did not affect any ofour estimates of risk.

Association between GST-u Markers and Risk for Lung Carcinoma. GST-TSO activity across controls (Fig. 1) ranged from 0 to

2740 pmol/mg/min, similar to the range of activity reported for leukocytes (22, 19). Subjects were categorized into two groups (<270 or>270 pmol/mg/min) and three groups (<270, 271-1000, >1000

pmol/mg/min) according to visual inspection of the distribution ofactivity and selection of antimodes among the controls. Subjects werealso assigned to activity categories defined by the proportion expectedin each category, based on the proportion of our controls determinedby Southern Blot to have no (48%), one (34%), or two GST1 genes(17%): <277 pmol/mg/min (48%); 278-694 pmol/mg/min (34%);

>694 pmol/mg/min (17%).Presence of GST-u activity was associated with a 70% reduction in

risk for lung carcinoma (Table 3), and increasing GST-u activity wasassociated with decreasing risk in a dose-response pattern (Table 4).

This finding was not dependent on the scheme used to categorizesubjects as having low, medium, or high GST activity (Table 4). Thedistribution of GST activity among control subjects, using either categorization, was similar to that which would be expected under Har-dy-Weinberg equilibrium. Estimates for risk associated with GST-

TSO activity were unchanged by adjustment for any of the potentialconfounders we had identified, including vital status; age; ethnicbackground; hospital; which lung lobe was sampled; whether anysedatives, barbiturates, or other drugs were taken in the 2 weeksbefore specimen collection; or whether any lung condition was notedon the pathology report. Our risk estimates did not change when welimited the analysis to Caucasians. High GST-TSO activity was asso

ciated with a lower risk for lung carcinoma among the 38 subjects whohad a cigarette smoking history of more than 54 pack-years (unadjusted RR = 0.15; 95% confidence interval, 0.03-0.64) than among38 subjects with a history of 54 pack-years or less (unadjusted RR =0.61; 95% confidence interval, 0.14â€”2.60).Relative risks did not vary

substantially across subgroups defined by age, sex, or histolÃ³gica!type of tumor (13 squamous cell and 11 adenocarcinoma cases).

We obtained similar results when the enzyme was assayed immuno-logically. Cases were substantially less likely to have GST-u antigen

identified in their lungs than were controls (Table 3). The size of theassociation between the presence of GST-u antigen and risk for lungcarcinoma differed slightly according to the subjects' long-term smok-

2315




14

12

10

8

6

4

2

0 .III Lui,â€¢¿�iiiO 350 700 1050 1400 1750 2100 2450 2800

GSTTSOActivity(pmol/mg/min)Fig. I. Distribution of glutathione S-transferase activity toward fraws-stiibene oxide in

lung tissue from 43 control subjects.

Table 3 Relative risk for lung carcinoma among smokers according to presence ofi activity, GST-^Â¡antigen, and GSTI gene in lung tissue

Casesn(%)GST-TSO

activity (pmol/mg/min)

Â£270 26 (74.3)>270 9(25.7)GST-u

antigen

AbsentPresentGST-u

gene

No copiesOne or two copies26

(74.3)9(25.7)16(64.0)

9 (36.0)Controls

n(%)20(46.5)

23(53.5)20(46.5)

23(53.5)14(48.3)

15(51.7)Relative

risk,unadjusted(95%CI)1.00

0.30(0.11,0.79)1.00

0.30(0.11,0.79)1.00

0.52(0.18.1.57)Relative

risk,adjusted forvital status(95%CI)1.00

0.29(0.09,0.%)1.00

0.23(0.06,0.80)1.00

0.31 (0.07,1.33)

ing history, with heavier smokers having a somewhat lower risk(unadjusted RR = 0.25; 95% confidence interval, 0.06-1.02) associated with the presence of GST-u than did lighter smokers (unadjustedRR = 0.35; 95% confidence interval, 0.07-1.60).

Presence of the gene for the GST-|j enzyme, as determined by

Southern Blot analysis of lung tissue DNA, was also associated withreduced risk for lung cancer. There was a suggestion in the data thatthe presence of two copies of the GST-u gene was associated with a

greater reduction in risk than was the presence of one copy (unadjusted RR = 0.35; 95% CI, 0.06-2.10 versus unadjusted RR = 0.61;95% CI, 0.18-2.04). The average change in risk per additional copy ofthe GSTI gene, unadjusted, was 0.60 (0.27-1.32) with P = 0.20 for

linear trend. Because we found agreement between intraindividualresults from each of the three assays, we believe that these confidenceintervals include one because genotype was assayed in a subset of oursubjects (69% of subjects for whom we have activity and immunoas-say data). The small number of patients assayed for the GST-u gene

limited our ability to evaluate differences in relative risk across subgroups. The relative risk for lung carcinoma associated with absenceof the GST-u gene was similar in magnitude to the risks associatedwith absence of the GST-u antigen and with lack of GST-TSO activity.

DISCUSSION

Subjects with GST-TSO activity, GST-u antigen, or GSTI DNA

were at lower risk for lung carcinoma than were subjects without thesemarkers for the GST-u enzyme. Because the basic defect in the GST-u

polymorphism appears to be a gene deletion, it is unsurprising that ourresults from the three GST-u assays were consistent. Regardless of

whether the gene, the gene product, or the enzyme activity was assayed, there was a consistent relationship with risk for lung carcinoma. Similarly, BrockmÃ¶ller et al. (30) recently reported that theabsence or presence of GST-TSO activity in lymphocytes corre

sponded with the absence or presence of the GST-u gene in all but 3

of 145 subjects (30). Thus, we and others have shown that futurestudies may rely on DNA techniques to detect the GST-u polymorphism. If GST-TSO activity in lung and GST genotype in DNA are

measurements of the same biological phenomenon, any nucleatedtissue will be suitable for molecular study of the role of GST-u in

determining susceptibility to lung cancer.The association between the presence of GST-TSO activity in lung

tissue and a relatively low risk for lung carcinoma reported here issimilar, in direction and magnitude, to that reported by Seidegard et al.using leukocytes (17, 18) and in contrast to the overall lack of association found by Heckbert et al. (19) and by Zhong et al. (29) withleukocytes. Our findings, and those of Seidegard et al., differ from theresults of Heckbert et al. inasmuch as different proportions of controlslacked GST activity. With each of our categorizations, the proportionof our controls with no or low GST-TSO activity was between 46%and 49%, which is consistent with 41% and 42% reported in Seide-gard's original (17) and follow-up studies (18) but lower than the 58%

found in controls in the study of Heckbert et al. (19). The controls inthe latter study were population based, whereas this study and Seide-gard's study used hospital-based controls (although with different

criteria for eligibility). While we cannot explain why these controlpopulations differed, we do not believe it is due to different racialcompositions or smoking histories of the study populations or todifferences in methods used to classify subjects as having low, intermediate, or high GST activity. Our finding of a stronger negativeassociation among heavy smokers is consistent with the results of bothSeidegard and Heckbert. The association may be more evident inheavy smokers because the importance of GST detoxification increases as the importance of tobacco in causing tumorigenesis increases.

We also measured GST-TSO activity in 9 potential control patients

who were subsequently found to be ineligible for the study becausethey had not smoked 100 or more cigarettes. Six of these subjects(66.7%) had the null phenotype, in contrast to 46.5% of eligiblecontrols with a history of smoking. Heckbert et a!, found a similardifference in the proportions with the null phenotype among 22 of 33controls who were nonsmokers (66.7%) and 34 of 66 smokers (51.5%). These results argue for careful attention to cigarette smokingstatus when selecting controls for studies of this type.

Our genotype results differ from previously published Southern blotresults (29). In a large study (228 cases and 225 controls), Zhong et al.used a u-class GST gene probe and the EcoR and fiamHI restriction

enzymes and reported no association between the GSTI genotype andrisk for lung cancer. However, when his cases were limited to patients

Table 4 Relative risk for lung cancer in relation to level of glutathione S-transferaseactivity' toward trans-stilbene oxide

Cases ControlsActivity (pmol/mg/min)

Relative risk,unadjusted(95% CD

According to modes seenamong controls"

<270271-1000

>1000

According to % controls in eachGST-u gene category*Â»c

<277278-694>694

26(74.3) 20(46.5) 1.007(20.0) 17(39.5) 0.32(0.11,0.91)2(5.7) 6(14.0) 0.26(0.05,1.41)

26 (74.3)6(17.1)3 (8.6)

21 (48.8)14(32.6)8(18.6)

1.000.35(0.11.1.06)0.30(0.07.1.29)

" Average change in risk between categories is 0.42 (0.20, 0.89). P = 0.02 for linear

trend.* Average change in risk between categories is 0.48 (0.24. 0.%), P = 0.03 for linear

trend.' Categories defined by the proportion expected in each category, based on the pro

portion of controls determined by Southern blot to have no, one or two GST-u genes.

2316



OST-M POLYMORPHISM AND RISK FOR LUNG CANCER

with adenocarcinoma, there was an association between genotype andrisk. We found no modification of risk for lung cancer when welimited our cases to either adenocarcinoma (RR = 0.47; 95% CI,0.07-2.96) or squamous cell carcinoma (RR = 0.47; 95% CI, 0.10-

2.23). A greater proportion of our lung cancer cases (64%) thanZhong's cases (43%) were homozygous null for the GST1 gene. We

do not know the source of the subjects in the study by Zhong et al.,so we cannot suggest whether differences in study populations, perhaps in ethnic background, might account for the difference in thisproportion. BrockmÃ¶ller has stated that he failed to find a "trend ofoverrepresentation of GST-u deficient individuals" among 49 lung

cancer patients compared with 96 hospital-based control patients (all

of German origin) but did not present any data and did not describe thesmoking status of his subjects (30).

One limitation of our study was the extensive use of tissue fromdeceased subjects. We were able to obtain lung specimens from anumber of living patients with lung carcinoma who underwent majorlung resection. However, because few persons without lung carcinomaundergo this procedure, we had to identify another source for specimens from control subjects, and thus we included autopsied patients.For several reasons we are satisfied that our measurements of GST-uactivity in deceased patients were probably valid indicators for ante-mortem GST-u activity. First, there was no evidence of proteolytic

degradation of GST in tissues obtained postmortem. Second, adjustment in the analysis for vital status resulted in nearly unchangedrelative risks. Also, the risk estimates that were strengthened slightlywith adjustment for vital status were those associated with the presence of the GST-u antigen and the GST-u gene. Because we would

expect measurements of enzyme activity to be more sensitive toproteolysis than measurements of antigens or DNA, it is unlikely thatthis slight confounding was due to proteolysis.

Although our controls were identified through hospital pathologyservices, we sought to make them as similar as possible to the generalpopulation with respect to their GST-u status by excluding those

patients who were operated on or died from a condition which isepidemiologically related to smoking and, therefore, might be relatedto GST status (39). Whether or not we were successful in accomplishing our aim cannot be evaluated, since lung tissue specimens frommembers of the healthy general population cannot be obtained. However, we believe that these exclusions were appropriate because weassayed lung specimens from 21 ineligible patients with other diseasesthat likely were a consequence of their smoking (listed under "Subjects and Methods") and found that their GST values were interme

diate, between those of our cases and our eligible controls (52.6% hadthe null genotype), perhaps suggesting that GST-u status may beinvolved in determining risk for other smoking-related diseases. Our

study was limited by the small number of subjects. While we wereable to measure the main association of interest, our ability to examinepossible differences in the influence of GST-u status on lung cancer

risk in important subgroups (e.g., females versus males) was limited.The similarity in results from each of our three GST-u assays, as

well as the good intrasubject agreement between assay results, suggests that it is the presence or absence of the GST1 gene whichdetermines GST-u activity toward TSO in lung tissue and GST-relatedsusceptibility to smoking-induced lung carcinoma. Our findings correspond well with previous findings that lack of GST-TSO activity in

leukocytes is due to the absence of both maternal and paternal copiesof the GST 1 gene (23, 29, 30). Our data suggest that the ability of GSTto detoxify carcinogens in tobacco smoke is determined genetically,rather than by local metabolic needs of the lung. Given the complexmetabolism of the numerous carcinogens in tobacco smoke and thenumber of enzymes involved in PAH metabolism (1, 40-43), themagnitude of the association between GST-u activity and risk for lung

cancer that we detected is large. The strength of this association mayindicate that polycyclic aromatic hydrocarbons are important carcinogens in tobacco smoke or that GST-u detoxifies other tobacco car

cinogens in addition to the PAHs. There is some recent evidence thatGST enzymes may play a role in the metabolism of nitrosamines (44,45). Ultimately, measurements of GST, in addition to aryl hydrocarbon hydroxylase, debrisoquine hydroxylase, and other genetic polymorphisms that may be involved in carcinogen metabolism, may benecessary in order to accurately classify the effect of genetic factorson an individual smoker's risk for lung cancer.

ACKNOWLEDGMENTS

We are indebted to many of the pathologists in the Seattle area, and inparticular to David R. Thorning, Steven A. Bigler, Brian K. Stewart, Charles E.Murry, and Rodney A. Schmidt.

REFERENCES

U.S. Surgeon General. The Health Consequences of Smoking: Cancer. A Report of theSurgeon General. Rockville, MD: U.S. Department of Health and Human Services.1982.Pelkonen. O., and Vahakangas. K. Binding of polycyclic aromatic hydrocarbons toDNA: comparison of mutagenesis and tumorigenesis. J. Toxicol. Environ. Health, 6:1009-1020, 1980.Goshman, L. M.. and Heidelberger, C. Binding of tritium-labeled polycyclic hydrocarbons to DNA of mouse skin. Cancer Res., 27: 1678-1688, 1967.

Shen. A. L., Fahl, W. E., and Jefcoate. C. R. Metabolism of benzo(a)pyrene byisolated hepatocytes and factors affecting covalent binding of benzol a Ipyrene metabolites to DNA in hepatocyte and microsomal systems. Arch. Biochem. Biophys.,204: 511-523, 1980.

Harris, C. C., Autrup, H., and Stoner, G. Metabolism of benzo(Ã¼)pyrene in culturedhuman tissues and cells. In: H. V. Gelboin (ed.), Polycyclic Hydrocarbons and Cancer.Vol. 2, pp. 331-342. San Diego: Academic Press, 1978.Nakayama, J., Yuspa, S. H.. and Poirier. M. C. Benzo(a)pyrene-DNA adduci formation and removal in mouse epidermis in vivo and in vitro: relationship of DNA bindingto initiation of skin carcinogenesis. Cancer Res., 44: 4087-4095, 1984.Brookes, P., and Lawley. P. D. Evidence for the binding of polynuclear aromatichydrocarbons to the nucleic acids of mouse skin: relation between carcinogenic powerof hydrocarbons and their binding to deoxyribonucleic acid. Nature (Lond.). 202:781-784, 1964.Kellerman, G.. Shaw, C. R., and Luyton-Kellerman, M. Aryl hydrocarbon hydroxylase inducibility and bronchogenic carcinoma. N. Engl. J. Med.. 289: 934-937, 1973.

Trell. L., Korsgaard. R., Janzon, L., and Trell. E. Distribution and reproducibility ofaryl hydrocarbon hydroxylase inducibility in a prospective population study of middle-aged male smokers and nonsmokers. Cancer (Phila.), 56: 1988-1994. 1985.

Gahmberg, C. G., Sekki, A., Kosunen, T. U.. Holsti, L. R.. and Makela. O. Inductionof aryl hydrocarbon hydroxylase activity and pulmonary carcinoma. Int. J. Cancer.23: 302-305, 1979.

Ward, E.. Paigen, B., Steenland, K., et al. Aryl hydrocarbon hydroxylase in personswith lung or laryngeal cancer. Int. J. Cancer, 22: 384-389. 1978.Paigen. B., Gurtoo, H. L., Minowanda, J.. et al. Questionable relation of aryl hydrocarbon hydroxylase to lung cancer risk. N. Engl. J. Med., 297: 346-350, 1977.McLemore. T. L., Martin. R. R.. Wray, N. P., Cantrell. E. T., Busbee. D. L. Reassessment of the relationship between aryl hydrocarbon hydroxylase and lung cancer.Cancer (Phila.), 48: 1438-1443, 1981.

Ayesh, R., Idle, J. R.. Ritchie, J. C., Crolhers. M. J.. and Hetzel, M. R. Metabolicoxidation phenotypes as markers for susceptibility to lung cancer. Nature iLond.l,312: 169-170. 1984.

Caparaso, N. E.. Tucker. M. A., Hoover. R. N.. et al. Lung cancer and the debrisoquinemetabolic phenotype. J. Nati. Cancer Inst.. 82: 1264-1271. 1990.

Spiers, C. J., Murray, S., Davies, D. S.. Biola Mabadeje. A. F.. Boobis. A. R.Debrisoquine oxidation phenotype and susceptibility to lung cancer. Br. J. Pharmacol..29: 101-109, 1990.Seidegard. J.. Pero, R. W.. Miller. D. G., and Beattie. E. J. A glutathione transferasein human leukocytes as a marker for the susceptibility to lung cancer. Carcinogenesis(Lond.), 7: 751-753, 1986.

Seidegard, J., Pero. R. W.. Markowitz. M. M.. Rousch, G., Miller. D. G., and Beattie.E. J. Isoenzyme(s) of glutathione transferase (class u) as a marker for the susceptibility to lung cancer: a follow-up study. Carcinogenesis (Lond.), //: 33-36, 1990.Heckbert, S. R., Weiss. N. S.. Hornung, S. K., Eaton. D. L., and Motulsky. A. G.Glutathione 5-transferase and epoxide hydrolase activity in human leukocytes inrelation to risk of lung cancer and other smoking-related cancers. J. Nati. Cancer Inst..84: 414-422. 1992.Board. P. G. Biochemical genetics of glutathione 5-transferase. Am. J. Hum. Genet..33: 36-43, 1981.Strange, R. C., Faulder, C. G., Davis, B. A., el al. The human gluthathione 5-trans-ferases: studies on the distribution and genetic variation of the GST1, GST2, andGST3 isozymes. Ann. Hum. Genet.. 48: 11-20, 1984.Seidegard, J., and Pero. R. W. The hereditary transmisiÃ³n of high glutathione transferase activity toward transtilbene oxide in human mononuclear leukocytes. Hum.

2317




Genet.. 69: 66-68, 1985.23. Seidegard. J.. Voracheck, W. R., Pero, R. W., and Pearson, W. R. Hereditary differ

ences in the expression of the human glutathione transferase active on frans-stilbeneoxide are due to a gene deletion. Proc. Nati. Acad. Sei., Ã„5.7293-7297. 1988.

24. Board, P. G. Gene deletion and partial deficiency of the glutathione 5-transferase<Ligandin) system in man. FEBS Lett., 135: 12-14, 1981.

25. Laisney, V., Van Cong, N., Gross, M. S., and Frezal, J. Human genes for glutalhioneS-transferases. Hum. Genet.. 68: 221-228. 1984.

26. Seidegard. J.. Gulhenberg. C., Pero. R. W., and Mannervik. B. The (ran.t-stilbeneoxide-active glutathione transferase in human mononuclear leukocytes is identicalwith the hepatic glutathione transferase u. Biochem. J.. 246: 783-785. 1987.

27. DeJong, J. L., Chang, C. M., Whang-Peng, J., Knutsen, T., and Tu, C. P. The humanliver glutathione 5-transferase gene superfamily: expression and chromosome mapping of an Hb subunit cDNA. Nucleic Acids Res., 16: 8541-8554, 1988.

28. Widersten. M., Pearson. W. R.. EngstrÃ¶m.A.. Mannervik. B. Heterologous expressionof the allelic variant u class glutathione transferases u and - Biochem. J., 276:519-524. 1991.

29. Zhong, S., Howie. A. H., Ketterer. B.. et al. Glutathione 5-transerase u locus: use of

genotyping and phenotyping assays to assess association with lung cancer susceptibility. Carcinogenesis (Lond.), 12: 1533-1537, 1991.

30. BrockmÃ¶ller. J.. Gross. D.. Kerb. R., Drakoulis, N., and Roots. I. Correlation between/ran.t-stilbene oxide-glutathione conjugation activity and the deletion mutation in theglutathione 5-transferase class u gene detected by polymerase chain reaction. Biochem. Pharmacol.. 43: 647-650. 1992.

31. Strange. R. C., Davis. B. A., Faulder, C. G., et al. The human glutathione S-tran-

ferases: developmental aspects of the GST1. GST2. and GST3 loci. Biochem. Genet.,23: 1011-1028. 1985.

32. Awasthi. Y. C., Singh, S. V., Ahmad, H., and Moller, P. C. Immunohistochemicalevidence for the expression of GST I, GST2, and GST3 gene loci for glutathioneS-transferase in human lung. Lung. 195: 323-332, 1987.

33. Cohen, G. M., and Moore, B. P. Metabolism of(3H) benzo(a)pyrene by differentportions of the respiratory tract. Biochem. Pharmacol., 25: 1623-1629, 1976.

34. Mannervik. B. The isoenzymes of glutathione transferase. Adv. Enzymol.. 57: 357-

417, 1985.

35. Carmichael, J., Forrester, L. M., Lewis. A. D., Hayes, J. D., Hayes. P. C., and Wolf.C. R. Glutathione S-transferase isoenzymes and glutathione peroxidase activity innormal and tumor samples from human lung. Carcinogenesis (Lond. 1.9: 1617-1621.

1988.36. Seidegard, J., and DePierre, J. W. Characterization of soluble glutathione transferase

activity in resting mononuclear leukocytes from human blood. Biochem. Pharmacol.,33: 3053-3058, 1984.

37. Maniatis. T., Fritsch, E. F., and Sambrook. J. Molecular Cloning: A LaboratoryManual. Cold Spring Harbor. NY: Cold Spring Harbor Laboratory. 1982.

38. Breslow, N. E., and Day, N. E. Statistical Methods in Cancer Research. Vol. 1. IARCScientific Publication no. 32, p. 207. Lyon: International Agency for Research onCancer, 1980.

39. Silverman, D. T. Hoover, R. N., and Swanson. G. M. Artificial sweeteners and lowerurinary tract cancer: hospital versus population controls. Am. J. Epidemici., 117:326-334, 1983.

40. Cooper, C. S., Grover, P. L., and Sims, P. The metabolism and activation of benzo-

(a)pyrene. In: Â¡.W. Bridges and L. F. Chasseaud (eds.). Progress in Drug Metabolism.Vol. 17. New York: John Wiley and Sons, Ltd.. 1983.

41. Pelkonen, O., and Neben, D. W. Metabolism of polycyclic aromatic hydrocarbons:etiologic role in Carcinogenesis. Pharmacol. Rev., 34: 189-222, 1982.

42. Conney, A. H. Induction of microsomal enzymes by foreign chemicals and Carcinogenesis by polycyclic aromatic hydrocarbons: G. H. A. Clowes Memorial Lecture.Cancer Res., 42: 4875^4917. 1982.

43. Gelboin, H. V. Benzo(a)pyrene metabolism, activation, and Carcinogenesis: role andregulation of mixed-function oxidases and related enzymes. Physiol. Rev.. 60: 1107-

1166. 1980.44. Jensen, D. E., and Stelman. G. J. Evidence for cytosolic GST-mediated denitrosation

of nitrosocimetidine and l-methyl-2-nitro-nitroso-guanidine. Carcinogenesis (Lond.),8: 1791-1800, 1987.

45. Smith, M. T, Evans, C. G.. Doane-Setzer, P., Castro, V. M., Tahir. M. K., andMannervik. B. Denitrosation of l,3-bis(2-chloroethyl)-l-nitrosourea by class mu glu

tathione transferases and its role in cellular resistance in rat brain tumor cells. CancerRes.. 49: 2621-2625, 1989.

2318



1993;53:2313-2318. Cancer Res Valle Nazar-Stewart, Arno G. Motulsky, David L. Eaton, et al. Susceptibility to Lung Carcinoma

-Transferase µ Polymorphism as a Marker forSThe Glutathione

Updated version

http://cancerres.aacrjournals.org/content/53/10/2313

Access the most recent version of this article at:

E-mail alerts related to this article or journal.Sign up to receive free email-alerts

Subscriptions

Reprints and

[email protected] at

To order reprints of this article or to subscribe to the journal, contact the AACR Publications

Permissions

Rightslink site. Click on "Request Permissions" which will take you to the Copyright Clearance Center's (CCC)

.http://cancerres.aacrjournals.org/content/53/10/2313To request permission to re-use all or part of this article, use this link



http://cancerres.aacrjournals.org/cgi/alerts

mailto:[email protected]



the glutathione s-transferase p polymorphism as a marker ... · gst-u activity toward tso is...

Documents