Interaction between cytochrome P450 1A2 genetic polymorphism and cigarette smoking on the risk of hepatocellular carcinoma in a Japanese population

Limited epidemiological evidence suggests that genetic polymor-phisms of drug-metabolizing enzymes such as cytochrome P450(CYP), glutathione S-transferase (GST) and N-acetyltransferase(NAT) may be involved in tobacco-related hepatocarcinogenesis.We conducted a case–control study, including 209 incident caseswith hepatocellular carcinoma (HCC) and two different controlgroups [275 hospital controls and 381 patients with chronic liverdisease (CLD) without HCC], to investigate whether CYP1A1,CYP1A2, CYP2A6, CYP2E1, GSTM1 and NAT2 polymorphismsare related to the risk of HCC with any interaction with cigarettesmoking. Overall, no significant associations with HCC were ob-served for any genotypes against either control group. However,we found a significant interaction (P 5 0.0045) between CYP1A2-3860G>A polymorphism and current smoking on HCC riskwhen we compared HCC cases with CLD patients; adjusted oddsratios [ORs; and 95% confidence intervals (CIs)] for G/A and A/Agenotypes relative to G/G genotype were 0.28 (0.12–0.66) and 0.18(0.04–0.94), respectively, amongcurrent smokers (Ptrend 50.002),as compared with 1.28 (0.80–2.06) and 0.76 (0.34–1.71), respec-tively, among never/former smokers (P trend 5 0.96). Similarly,in CYP1A2 G/G genotype, significant risk increase was observedfor current smoking (OR 5 4.08, 95% CI 5 2.02–8.25) ormore recent cigarette use (e.g. pack-years during last 5 years,P trend 5 0.0003) but not in G/A and A/A genotypes combined(OR for current smoking 5 1.39, 95% CI 5 0.63–3.03; P trendfor pack-years during last 5 years 5 0.40). These results suggestthat the CYP1A2 -3860G>A polymorphism modifies the smoking-related HCC risk among CLD patients.IntroductionCigarette smoking has lately been accepted as a risk factor for primaryliver cancer (1,2), consisting mostly of hepatocellular carcinoma(HCC) (3), although this linkage has not yet been widely recognizedby clinicians or the general public. The International Agency forResearch on Cancer added primary liver cancer to a list of tobacco-related malignancy in 2004 (1). According to a recent review ofepidemiological studies among Japanese populations, cigarette smok-ing probably increases the risk of primary liver cancer in this ethnicgroup as well (2). Similarly, our case–control study demonstrated thatcurrent smoking or more recent cigarette use (e.g. pack-years duringthe last 10 or 5 years) was associated with higher HCC risk amongpatients with chronic liver disease (CLD), and thus, such patients maybenefit from their earliest smoking cessation (4). However, the bi-ological mechanisms underlying smoking-related hepatocarcinogen-esis remain to be elucidated.Many constituents in tobacco smoke (e.g. polycyclic aromatichydrocarbons, nitrosamines, aromatic amines and heterocyclicamines) are known to be carcinogenic (5). Most such carcinogensrequire metabolic activation by phase I drug-metabolizing enzymes,such as cytochrome P450s (CYPs), before binding to DNA and othercellular macromolecules, and they are detoxified by phase II enzymessuch as glutathione S-transferase (GST) and N-acetyltransferase (NAT)(6). Over the past two decades, numerous studies have explored geneticpolymorphisms of these enzymes as potential candidates for individualsusceptibility to smoking-related cancers (6,7); once a certain geneticpolymorphism of some enzyme has been established as a risk factor oreffect modifier, some substrate of the enzyme contained in cigarettesmoke would be regarded as playing an etiologically important role.For HCC, however, data remain sparse on this issue (8).A limited number of case–control studies evaluated both smokinghabits and genetic polymorphisms of the above drug-metabolizingenzymes in the causation of HCC, as well as related gene–smokinginteractions (9–14). According toa seriesofcase–controlstudies nestedin a cohort of hepatitis B virus (HBV) carriers in Taiwan, the poly-morphisms of CYP1A1, CYP2E1 and NAT2, but not GSTM1 deletionpolymorphism, were significantly associated with HCC risk amongsmokers, whereas any polymorphisms posed no risk change amongnon-smokers (9–11). However, other studies did not confirm such find-ings for CYP2E1 or NAT2 polymorphisms (12,13). A recent case–control study in China reported that homozygous carriers of the majorhaplotype (-3860G/-3113G/5347C) of CYP1A2 showed around 3-foldincrease of HCC risk in heavy smokers (14); yet, no other data existon this association. Moreover, data on those positive gene–smoking interactions came from HBV-endemic areas, and it remainsessentially unknown whether such interactions hold true where hepati-tis C virus (HCV) infection represents a dominant risk factor of HCC,like in Japan (15).We conducted a case–control study of HCC in a Japanese popula-tion, in order to examine whether genetic polymorphisms of CYP1A1(MspI site in the 3#-flanking region) (16), CYP1A2 (DdeI site in the5#-flanking region) (17), CYP2A6(whole deletion type) (18), CYP2E1(RsaI site in the 5#-flanking region) (19), GSTM1 (homozygous de-letion) (20) and NAT2 (acetylation polymorphisms) (21) are related tothe risk of HCC with any interaction with cigarette smoking. All thesepolymorphisms are typical candidates for susceptibility to smoking-related cancers (6,7), including HCC as described above, with somedata on their positive associations with enzyme activity or inducibility(7,22,23), and the corresponding genes are known to be expressed inthe liver (24–27). Two different control groups (hospital controls andpatients with CLD without HCC) were employed in this study; theformer represents a conventional control group, and the latter waschosen because of the clinically established finding that the majorityof HCC develops from CLD in Japan (3).Materials and methods