Tobacco Documents Online

Caner Caus~ and Contro/, 1997, 8, pp. 554-559 Polymorphisms in the glutathione S-transferase class rau and theta genes interact and increase susceptibility to lung cancer in minority populations (Texas, United States) Karl T. Kelsey, Margaret tL Spitz, Zheng-Fa Zuo, and John tL Wiencke (Received I0 September 1996; accepted in revised form 21 January 1997) The genes coding for separate isoforms of both the human glutathione S-trausferase class rau and class tbeta enzymes (GSTM1 and GSTT1) are polymorphic with a variable ethnic disu-ibution. These enzymes demxify reactive epoxides, including carcinogens produced by tobacco smoke. Because of this, the null polymorphism in the GSTM1 gene (coding for the glutathione S-~ransferase class mu enzyme) has been studied widely as a possible source of inherited susceptibility ~o smoking-related lung cancer. The more recently described null polymorphism in the GSTT1 gene also could contribute to an increased risk of smoking-related lung cancer. As the incidence of lung cancer is known to differ by ethnicity, we have conducted a case-control study in the United States of 108 African-Americans (Blacl~) and 60 Mexican-Americans (Hispanics) with lung cancer and 132 African-American (Black) and 146 Mexican- American (Hispanic) controls to investigate the association of the GSTT1 and GSTM1 polymorphisms with lung cancer in minority populations. In the unadjusted dam, there was a borderline significant association of the GSTM1 null polymorphism with lung cancer in Mexican-Americans (odds ratio [OR] -- 1.8, 95 percent confidence interval [CI] = 1.0-3-~ ) that was not observed in African-Americans. The GSTT1 null polymorphism also had a higher prevalence in cases than controls in both raclal/ethnic groups, but ~ increase was not statistically significant. When the data were analyzed u~ing logistic regression controlling for age, gender, race, and smoking, no significant association of either ~rait with lung cancer was observed, with Ot~ for both ~rair~ of approximately 1~. However, when the prevalence of individuals who were null for both polymorphisms was compared by case status, a significant interaction was observed. Logistic regression models showed the OR for the association of lung cancer and the presence of both null polymorph]sms compared with one (either GSTT1 or GSTM1) or no null genocype to be 2.9 (P < 0.04). These results suggest that there may be carcinogenic intermediates in cigarette smoke that are subs~rates for both the GSTF1 and GSTM1 enzymes, and thac lung cancer risk is increased more than additively for individuals who have both GSTTI and GSTM1 null polymorph/sms. Cancer Causes and Control 1997, 8, 554-559 Key words: African-American, Black, Hispanic, genetic susceptibility, GSTM1, GSTT1, glutathione S-transferase, lung cancer, Mexican-American, United States. Dr Kelsey is with the Departmen~ of Cancer Biology, and Drs Kelsey and Zuo are ~vitb the Department of Environmental Health, Harvard School of Public Health, Bouon, MA, USA. Dr Spitz is zuish the Departrnen~ of Epideraiology, The University of Texas, M.D. Anderson Cancer Cent~ Houston, TX, UStL Dr Wiencke is ¢nith the Laboratory for Molecular Epidemiology, Depar~raent of Epideralology and Biostatistics, School of Medicine, University of California at San Francisco, San Francisco, CA, USA. Address correspondence to Dr Kelsey, Department of Cancer Biology, Harvard School of Public Health, 665 Huntington Ave., Boston, MA 02115, USA. Thi~ study ¢vas supported by US National Institute of Health grants ES-O0002, ES-06717, ES-04705, CA-55769 and CA-51210. 554 Cancer Catne~ and Control. Vol 8. 1997 ~ 1997 Rapid Science Publishers

Introduction Tobacco smoke contains a large number of reactive com- pounds that are directly genotoxic as well as numerous combustion products that are metabolically activated by cytochrome P450 enzymes to generate reactive inter- mediates. These highly reactive products can be metabolized further to water soluble intermediates through conjugation with glutathione (GSH). Genetic variation in GSH-mediated detoxification has been studied widely, including now-well-described detetion polymorphisms in the glutathione S-transferase class mu and class theta. The glutathione S-transferases [EC 2.5.1.18] are a family of enzymes that detoxify reactive electrophiles, such as epoxides, that can act as mutagens,t There are mukiple classes of glutathione S-transferases (GSTs) found in humans, including the mu, pi, theta, and alpha. Phenotypic activity of both the glutathione S- transferase rau and theta enzymes have been found to be polymorphic in the p0pulation,z" The variation in conjugation of epoxide substrate intermediates has been observed to segregate with inherited loss of the GSTM1 gene as well as with the GSTT1 gene. That is, individuals who inherit the homozygous form of the null polymor- phisms in these genes are less capable of conjugating and detoxifying specific substrate intermediates.~'~ There has been considerable interest in the possible association of the null GST genotypes with susceptibility to environmental cancer. Since the known substrates for the GSTM1 protein products include PAHs, specifically including metabolically generated epoxide intermediates of benzo(c~)pyrene,~ several groups have investigated the association of the null genoLTpes and smoking-induced lung cancer? Similarly, substrates for the GSTT1 enzyme include alkyl haLides found in cigarette smoke.7. However, there has been only limited study of the association of these genes and lung cancer susceptibility in African- Americans~ (B!acks) and, to our knowledge, no investigation of lung cancer risk modification by these traits in Mexican-Americans (Hispanics). Therefore, we have genotyped 206 Mexican-Americans (146 controls and 60 lung cancer cases) and 240 African-Americans (132 controls and 108 lung cancer cases) for the GSTM1 and GSTT1 deletion polymorphisms to investigate their association with susceptibility to lung cancer. Materials and methods Study populations A detailed description of the study population has been presented elsewhere..° Cases were newly diagnosed, histologicaI]y confirmed, lung cancer patients who had not received prior treatment for their disease with radia- tion or chemotherapeutics. The patients were recruited Genetic susceptibility to lung cancer from the University of Texas M.D. Anderson Cancer Center, from county, community, and Veterans Admini- stration hospitals in the Houston and San Antonio metropolitan areas and from Galveston, Texas (United States). There were no age, histologic, or stage restrictions. Controls were identified from a convenience sample recruited from community centers, cancer screening pro- grams, churches, and employee groups. Only individuals without a history of cancer were eligible to participate as controls. The controls were frequency-matched to the cases by gender, ethnicity, and age (± 5 years). A structured interview of approximately 45 minutes was conducted by trained interviewer/phlebotomists. Bilingual interviewer/phlebotomists conducted the inter- views for the Mexican-American cases and controls using a questionnaire translated into Spanish and backtranslated into English by the Department of Mexican Studies at the University of Houston. Mexican-American study participants could choose either English or Spanish for the interview. Data were collected on sociodemographic charac- teristics, recent and prior tobacco use, other lifestyle habits, and family history of cancer. Blood was drawn into heparinized tubes for use in cytogenetic and molecu- lar analyses. Laboratory methods Genotyping for the GSTM1 deletion polymorphism and GSTT1 deletion polymorphism was completed using polymerase chain reaction (PCR)-based methods pre- viously published."''z Laboratory personnel were blind to case-control status and multiple repeat samples were included in the PCR analysis to monitor quality control. Statistical analysis We calculated odds ratios (OR) and 95 percent confidence intervals (CI) for the association of the GSTM1 and GSTT1 null genotype with lung cancer using logistic regression. We also used the likelihood ratio test and Wald chi-square analysis as a measure of gene-gene interaction. Results We studied 455 volunteer subjects, including 168 lung cancer cases and 278 controls. Demographic information and smoking histories, stratified by genotype, are shown in Table 1. Cases were more likely to be male, but no differences were noted for age, education, or socioeco- nomic variables (data not shown). Cases also were more likely to be African-American than Hispanic as well as more likely to be current smokers and to have smoked more than controls. There was no association of the GSTM1 polymorphism or the GSTT1 polymorphism with age. Cancer Causes a~ad Control Vol 8. 1997 ~55

K. Z Keh~y et al Table 1. Demographic characteristics of the Texas study population by case status and GSTT1 and GSTM1 genotype Variable No. Cases Controls GSTTf GSTM1 GSTT1 GSTT1 GSTMI and GSTMI GSTT1 and GSTM1 Pos.a Null Pos.= Null Null Pos.a Null Pos.a Null Null Gender Male 318 Female .128 Ethnicity Mexican-Americans 206 African-Americans 240 Age (yrs) 446 Cigarette smoking Current smoker 177 Ex-smoker 144 Never smoked 125 Mean pack-yrs sn:~okedb 321 Mean yrs quit° 143 102 24 80 46 10 164 28 127 65 8 29 13 31 11 4 68 18 62 24 5 50 10 27 33 5 129 17 87 59 5 81 27 84 24 9 103 29 102 30 8 62.5 60.8 61.2 63.8 61.1 62.3 60.6 61.8 62.7 65.7 75 19 69 25 4 66 17 55 28 3 50 16 39 27 8 68 10 52 26 4 6 2 3 5 2 98 19 82 35 6 51.7 50.3 49.3 55.8 55.1 21.4 18.5 22.1 18.7 8.4 7,8 5.8 8.3 5.9 4,0 17.3 18.8 17,7 17.0 14.3 Pos. = positive. Current and former smokers only Former smokers only. Table 2. GSTM1 genotype by lung cancer case status and ethnicity Overall group Null Present Mexican-Americans African-Americans Null Preeent Null Present Lung canoer cases (%) 57 (34) 111 (66) Controls (%) 89 (32) 189 (68) Odds ratio 1,1 95% confidence interval (0.7-1.6) 33 (55) 27 (45) 24 (22) 34 (78) 59 (40) 87 (60) 30 (23) 102 (77) 1.8 1.0 (1.0-3.3) (0.5-1.8) Approximately 23 percent of the A~rican-American controls and 40 percent of the Mexican-American controls had the GSTMI null genotype (Tabh 2). This difference in prevalence was statisticaLly sigrfificant (P < 0.01). When the prevalence of the GSTM1 polymorphism was examined in cases and controls overall, there was a slight excess of the null genotype in the cases, with a crude OK of 1.1 (CI = 0.7-1.6). Stratifying the data by ethnicity showed that there was a borderline significant association of the null genotype with case status in the Mexican- Americans (OK: 1.8, CI = 1.0-3.3). There was no similar suggestion of an association of the GSTM1 null genotype with lung cancer case status in African-Americans (OR. = t.0, CI = 0.5-1.8). Logistic regression analysis of the association of the GSTM1 null genotype also was con- ducted to control for known risk factors, including age, ethnicity, gender, and smoking. The adjusted analysis did not differ significantly from the crude analysis, with an 556 Cancer Causes and Contz'ol. Vo[ 8. 1997 nonsignificant OR of 1.4 for the association of the GSTM1 null genotype with lung cancer. Stratification by" light and heavy smoking did not alter the OR signLqcandy (light smoking was defined as a less than 15 pack-year history). Approximately 22 percent of the African-American controls and 12 percent of the Mexican-American con- trols were found to have the GSTT1 null genotype; this difference was statistically significant (P < 0.01). A similar analysis of the prevalence of the GSTT1 null g~notype by tung cancer case status also showed no significant association of the GSTT1 null genotype with lung cancer overall, or stratified by" ethnicity (Table 3). While these comparisons did not reach significance in any case, the magnitude of the OR. was similar for both the GSTT1 and GSTMI, ranging from 1.2-1.5. Logistic regression analysis of the GSTT1 data, again controlling for age, gender, ethnicity, and pack-years of smoking, differed little from the crude analysis, with a nonsignificant OK

Genetic susceptibility to lung cancer Table 3. GSTT1 genotype by lung cancer case status and ethnicity Overall group Mexican-Americans African-Americans Null Present Null Present Null Present Lung cancer cases (%) 37 (22) 131 (78) 10 (17) 50 (83) 27 (25) 81 (75) Controls (%) 46 (17) 232 (83) 17 (12) 129 (88) 29 (22) 103 (78) Odds ratio 1.4 1.5 1.2 95% confidence interval (0.9-2.3) (0.7-3.5) (0.7-2.2) Table 4. Combined GSTT1 and GSTM1 null genotypea by lung cancer case status and ethnicity Overall group GSTM1 and GSTT1 both null GSTM1 and GSTT1 both present Either gene present Lung cancer cases (%) 14 (8.3) 88 (52.4) 66 (39.3) Controls (%) 13 (4.7) 156 (56.1) 109 (39.2) Odds ratio (both null to both present) = 1.9; 95% confidence interval = 0.9-4.2. Odds ratio (both null to either present) = 1.8; 95% confidence interval = 0.8-4.0. Odds ratio (both present to either present) == 0.9; 95% confidence interval = 0.6-1.4. ./,able 5. Odds ratios (OR) for the association of GSTTI and GSTM1 genotypas with lung cancer GSTT1 positive GSTT1 null GS'I'M1 positive Cases 88 23 Controls 166 33 OR 1.0 1.2 CIb (ref.) (0.5-2.5) GS'/'M1 null Cases 44 14 Controls 81 14 OR 1.1 2.9 CIb (0.6-2.1) (1.1-7.7) a OR adjusted for smoking, age, gender, ethnicity. b CI = 95% confidence interval. Wald chi-square = 4.4, P== 0.04 of 1.5 for the estimate of the magnitude of the association of the GSTT1 null geno .vype with lung cancer. Because other data investigating the role of these polymorphisms in head and neck cancer (also a tobacco- related cancer) have suggested that the risk associated with the combination of being deleted for both GSTT1 and GSTM1 may be more than additive,t~ we compared the patients who were deleted for both GSTT1 and GSTMI in lung cancer cases and controls. There were no differences in the age, cigarette pack-year smoking history., years since smoked in ex-smokers, or history of asbestos exposure by the both GSTT1 and GSTM1 null genotT, pe (data not shown). Table 4 shows that the crude O R for individuals who were doubly deleted at these gene loci was 1.9 (CI : 0.9-4.0). When this was examined by ethnicity, restricting the analysis to individuals who were GSTT1 and GSTM1 deleted compared with those who had both alleles of both genes, an OR of 3.41 (CI = 1.0-12.9) was observed in Mexican-Americans and an OR of 1.4 (CI = 0.5-3.8) in African-Americans. When logistic analysis was performed- controlling for age, gender, ethnicity, and smoking - the OR for the double null geno- type compared with carriers of at least one allele of each gene was elevated significandy, (with the OR estimated at 2.9) with evidence of a significant interaction of the null genotypes (Table 5). To determine further if this enhanced risk was associated with dose, we also examined the distribution of the GSTT1 and GSTM1 deletion in Cancer Caus~ and Control. Vol 8. 1997 357

K.T. K'e/se;y et al those with a greater than 30 pack-year history of smoking. Using this as a point for stratification (greater than 30 pack-years), there were 10 cases that did hoe have either the GSTT1 or GSTM1 gene, whereas there were no con- trol subiects who similarly lacked both genes. The background 08. for lung cancer in smokers for greater than 30 pack-years, adiusted for age, gender, and ethnicity, was approximately 50. Discussion Previous estimates from relatively large studies of the prevalence of the GSTM1 deletion polymorphism in African-Americans included 27 percent (n = 251 from Los Angeles County),* 31 percent (n = 87 primarily from the UCLA tissue typing laboratory),~* and 35 percent (n = 280 from North Carolina)?s Thus, our finding of the prevalence of the null polymorphism of 23 percent is somewhat lower than other estimates. The prevalence of the GSTM1 null genotype in Hispanics was 40 percent, also slightly lower than the 49 to 53 percent reported by Linet aL ~ The reason for these differences is unclear;, it may be attributable to chance, or perhaps to regional variation in African-American and/or Mexican-American heritage. Because nested quality-control samples, which were incinded in the routine analysis, and all repeat samples were concordant, it is unlikely that any difference is attributable to laboratory error. The association of the GSTM1 gene-deletion polymor- phism and lung cancer has been studied by many investigators; this literature has been recently reviewed? Most studies have found a modest association (ORs of about 1.3 to 1.4) of the GSTM1 null polymorphism and lung cancer risk. London et al' also have studied the null polymorphism in the GSTM1 gene in African-Americans and reported an OR consistent with the present study. Thin, our finding of a small, nonsignificant association of the GSTM1 null genotype in both African-Americans and Mexican-Americans is consistent with published data. The association of the GSTT1 null genotype with lung cancer risk has been studied recently by Wise et att~ Sire/far to our result, in 63 cases and 55 controls, they found no association of the GSTT1 deletibn polymor- phism and lung cancer. FinallB Deak;n etal ~7 also studied the prevalence of the GSTT1 and GSTM1 deleted geno- type in 108 lung cancer patents and 129 padents with obstructive lung disease. They reported no association of tither genotype with the occurrence of lung cancer. Interestingly, when we examined the association of the absence of both the GSTMI and GSTT1 genes with lung cancer occurrence, we found the genes interacted and were associated with significant risk for lang cancer. The association of the absence of both genes with lung cancer may be stronger in the heavier smokers, but the small 558 Cancer Causes and Control. Vol 8. 1997 number of controls who were deleted for both genes and who had a heavy" smoking history limited this analysis. Our result is similar to that observed by Trizna el: al ~ in studying head and neck cancer. These authors found that individuals with both gene deletion polymorphisms were at the highest risk for this form of cancer. However, the results of the present study differ from those of Deakin el: al tr who did not report an interaction of the GSTT1 and GSTM1 genes, studying lung cancer. One reason for the different results that must be considered is chance; our study as well as that of Deakin el: al ,r includes rela- tively few people that have both null genotypes. At the same dine, another possible explanation for the different results of the studies may be confounding by cigarette smoking. While Deakin et al *~ did analyze their data for an interaction, it is unclear if they controlled for cigarette consumption. As can be seen in our data, only after con- trolling for smoking can the magnitude of the interaction of the genotypes be accurately assessed. Further, our data also suggest that increased cigarette consumption also increases the magnitude of the risk. Consequendy, careful evaluation of smoking is essential for assessing the asso- ciation of these genotypes with lung cancer risk. The mechanism responsible for this possible inter- action is unclear. Since substrate for both enzymes are known to be in tobacco smoke, it is possible that the level of genotoxic damage associated with the presence of one but not the other null polymorphism is sufficiendy small to have little impact upon cancer risk, but when both enzymes are absent, the DNA damage is enhanced significandy and then could result in important increases in the occurrence of mutations. Our study suggests that the deletion polymorphisms in the GSTTI gene and the GSTMI gene interact to produce a significant increase in tobacco-induced lung cancer risk. Additional, large studies are needed to confirm this observation and to extend it to other ethnic groups. Acknowledgement--The authors thank Lucille Pothier for programming assistance. 1. K.etterer B, Christodoulides LG. Enzymology of cytosolic glutathione S-transferases. Adv Pharmaco11994; 27: 37-69. 2. Seidegard J, Vorachek WR, Pero RW, Pearson WR. He- reditary differences in the expression of the human glutath~one traasferase active on trans-stilbene oxide are due to a gene deletion. Proc Natl Acad Sci USA 1988; 85: 7293 -7. 3. Wiencke JK, Pemble S, Ketterer B, Ketsey" KT. Gene dele- tion of glutathione-S-transferase 0: correlation with induced genetic damage and potential role in endogenous rnutagene- sis. Cancer Epideraiol Biomark Prey i995; 4: 253-9. 4. Wiencke JK, Kelsey KT, Lamela R.A, Toscano WA. Human ¢o

Genetic susceptibility to lung cancer glutathione-S-transferase deficiency: a marker of suscepti- bility to expoxide-induced chromosome damage. Cancer Res 1990; 50: 1585-90. 5. Warholm M, Guthenberg C, Mannervik B, yon Bahr C. Purification of a new glutathione S-transferase (transferase ~a) from human liver having high activity with benzo(~)py- rene-4,5-oxide. Biochera Biophys Res Comraun 1981; 98: 512. 6. McWilliams JE, Sanderson BJS, Harris EL, Richert-Boe KE, Heuner W'D. Glutathione S-transferase M1 (GSTM1) deficiency and lung cancer risk. Cancer Epideraiol Biomark Pre-o 1995; 4: 58%94. 7. Pembh S, Schroeder KR, Spencer SR, et al. Human glu- tathione S-transferase theta (GSTT1): cDNA cloning and the characterization of a genetic polymorphism. Biochera J 1994; 300: 271-6. 8. Thier R, Taylor J-B, Pemble SE, et aL Expression of mam- malian glutathione S-transferase 5-5 in Salmonella ryphirau~ium TA1535 leads to base-pair mutations upon exposure to dihalomethanes. Proc Natl Acad Sci USA 1993; 90: 8576-80. 9. London SJ, Duly AK, Cooper J, Navidi WC, Carpenter CL, Idle JR. Polymorphism of glutathione S-transferase M1 and lung cancer risk among African-Americans and Caucasians in Los Angeles County, California.]NC11995; 87: 1246o53. - 10. Wu X, Deldos GL, Annegers JF, etal. A case-control study of wood dust exposure, mutagen sensitivity, and lung cancer risk. Cancer Epideraiol Biomark Prey 1995; 4: 583-8. 1 I. Cheng T~, Christiani DC, Xu X, Wain JC, Wiencke JK, Kelsey KT. GST-p genotype, diet, and smoking as deter- minants of sister chromatid exchange frequency in lymphocytes. Cancer Epideraiol Bioraark Pr~v 1995; 4: 535-42. 12. Nelson HH, Wiencke JK, Christiani DC, et al. Ethnic differences in the prevalence of the homozygous deletion genotype of glutathione-S-transferase theta. Carcinogenesis 1995; 16: 1243o5. 13. Trizna Z, Clayman GL, Spitz MR, Briggs KL, Goepfert H. Glutathione S-transferase genotypes as risk factors for head and neck cancer. AraJ Surg 1995; 170: 49%501. 14. Lin HJ, Hun C-Y, Bernstein DA, Hsiao W, Lin BK, Hardy S. Ethnic distribution of the glutathione transferase Mu 1-1 (GSTM1) null genotype in 1473 individuals and application to bladder cancer susceptibility. Carcinogenesis 1994; 15: 1077-81. 15. Bell DA, Taylor JA, Paulson DF, Robertson CN, Mohler JL, Luder GW. Genetic risk and carcinogen exposure: a common inherited defect of the carcinog~-metabolism gene glutathione S-transferase M1 (GSTM1) that increases susceptibility to bladder cancer. JNCI 1993; 85: 1159-64. 16. Wise Sr JP, Caporaso NE, Trivers GE, et M. The relationship of a glutathione-S-transferase theta (GSTT-I) polymor- phism to lung cancer risk. ProcAra Assoc Cancer Res 1996; 37:362. 17. Deakin M, Elder J, Hendrickse C, et al. Glutathione S- transferase GSTT1 genotypes and susceptibility to cancer. studies of interactions with GSTM1 in lung, oral, gastric and colorecr.al cancers. Carcinogenesis 1996; 17: 881-4. Cancer Causes and Control. Vol 8. 1997 559