Introduction

Consumption of tobacco is one of the world’s single largest avoidable causes of mortality, underlying 30% of all cancer deaths in high income and 10% in low income countries (WHO 1999). In the United States, a considerable reduction in adult use of tobacco has been achieved in recent years (Skaar et al 1997). In Japan, the smoking rate among males (47%) is still higher, as it is in China and Russia, than other developed countries, and the rate among females (10%), especially young women (21%), is on the increase (Japanese Health and Welfare Statistics Association 2002). Smoking behavior is clearly linked to an addiction to nicotine, which is the major psychoactive compound in cigarette smoke, and people smoke habitually to maintain nicotine levels in their body. Aside from this, smoking behavior may be affected by airway responsiveness to cigarette smoke. People who feel unpleasant when exposed to cigarette smoke, either from their own smoke or the smoke of others, are less likely to become persistent users. Unpleasant feelings may be related to bronchial inflammatory reactions to cigarette smoke.

Exposure to cigarette smoke modifies the balance of inflammatory cytokines released in the airways. IL-8 is known to be proinflammatory and one of the strongest chemotactic factors and activators of neutrophils, and its production has been shown to be stimulated by smoking (Kuschner et al 1996; Mio et al 1997).

IL8, located on human chromosome 4, exhibits functional polymorphisms. Among the 15 polymorphisms so far characterized (Hull et al 2004), the presence of IL8 −251T/A in the transcription start site exerts one of the greatest influences on IL-8 production. Stimulation assays with whole blood using lipopolysaccharides have shown that IL8 −251A tends to result in a greater expression of IL-8 in vitro, the highest values being observed for the AA genotype, the lowest with the TT genotype, and intermediate values are seen with AT (Hull et al 2000). It should be noted, however, that there is evidence to suggest that the IL8 −251T/A polymorphism may not be functional itself but rather linked with a functional polymorphism elsewhere in IL8. Recently, Hacking et al evaluated the allele-specific functions of six polymorphisms on IL8 (−251T/A, 396G/T, 781C/T, 1238del/insA, 1633T/C and 2767T/A) and identified T781C but not −251T/A as having an actual impact on protein–DNA binding (Hacking et al 2004). At the same time, they examined haplotype-specific functions and found haplotype 1 (−251T/396G/781T/1238delA/1633T/2767T, frequency 0.52) to be significantly increased in transcript level relative to the mirror haplotype (−251A/396T/781C/1238insA/1633C/2767A, frequency 0.41) in human respiratory epithelial cells (Hacking et al 2004). Frequencies of these two mirroring haplotypes suggest a strong linkage between the two loci. Supporting evidence for a linkage between these two polymorphisms is available for a European population: 99% of −251T were found to carry 781C and 89% of −251A were found to carry 781T (Hull et al 2001).

To test the hypothesis that the functional polymorphism, IL8 −251T/A, relating to inflammatory cytokine production, may influence disposition to cigarette smoking, the present study was conducted on a population of visitors to Aichi Cancer Center Hospital.

Subjects and methods

Study participants

The subjects were first-visit outpatients of Aichi Cancer Center Hospital (ACCH), aged 20–79 years, who were consecutively invited to fill in a lifestyle questionnaire and provide a peripheral blood sample within the framework of the Hospital-based Epidemiologic Research Program at Aichi Cancer Center II (HERPACC-II) (Hamajima et al 2001b). Participants in HERPACC-II between April and June 2001 were enrolled. Among 1,022 first-visit outpatients, 590 provided 7 ml of peripheral blood. They visited the ACCH on their own initiative (30.8%), at the family’s suggestion (20.4%), because of referral from other clinics (32.0%) or for secondary screening after primary screening (14.3%). 137 participants who had a present or past history of cancer, respiratory or cardiovascular disease at first visit (84 cancer cases, 7 cancer and respiratory disease, 34 respiratory disease and 12 cardiovascular disease) were excluded, because having these diseases is a strong motivation to quit smoking in general (Pinto et al 1999).

Data on smoking, including basic history (age at smoking initiation, number of cigarettes consumed, duration of smoking) and results of the Fargaström test for nicotine dependence (FTND) (Heatherton et al 1991) were available for all of the participants. Participants who currently smoked, information on attempts to quit, number of times that they did quit, and maximal duration of quitting were collected at the baseline. Smoking status at interview was classified into three categories; current smokers (individuals who were currently smoking), never smokers (those who had smoked less than 100 cigarettes in their lifetime) and former smokers (those who had quit smoking at the interview). Ever smokers (current and former smokers combined) were categorized for further analysis. The Institutional Review Board of Aichi Cancer Center approved the study before it was commenced (Approval Number 41-2).

Genotyping procedure

DNA was extracted from 200 μl of buffy coat preserved at −80 °C using a QIAamp Blood Mini Kit (Qiagen, Valencia, CA, USA). PCR-CTPP (PCR with confronting two-pair primers), used for the genotyping (Hamajima et al 2000), was performed according to established methods for IL8 −251T/A (NCBI SNP number; rs4073) (Hamajima et al 2003). Extracted DNA was amplified with four primers and TaKaRa TaqTM (Takara Bio Inc., Shiga, Japan): F1, 5′-CAT GAT AGC ATC TGT AAT TAA CTG-3′; R1, 5′-CAC AAT TTG GTG AAT TAT CAA A-3′; F2, 5′-GTT ATC TAG AAA TAA AAA AGC ATA CAA-3′; and R2, 5′-CTC ATC TTT TCA TTA TGT CAG AG-3′. PCR conditions were 3 min denaturation at 94 °C followed by 35 cycles of 94 °C for 1 min, 61 °C for 1 min, and 72 °C for 1 min, with 5 min final extension at 72 °C. Primer pairs F1 and R1 for the T allele (−251T) and F2 and R2 for the A allele (−251A) produced allele-specific bands of 169- and 248-bp, respectively, as well as a 349-bp common band.

Statistical analysis

All statistical analyses in the present study were performed using STATA version 8 (STATA Corporation Inc., College Station, TX, USA) statistical software. Accordance with the Hardy–Weinberg equilibrium, indicating an absence of discrepancy between genotype and allele frequencies, was checked for the total participants using the chi-square test, which was also employed to examine the association between individual polymorphisms and smoking status. We defined three sets of cases and controls as follows: (a) ever and never smokers; (b) current and never smokers; and (c) current and former smokers. Odds ratios (ORs) and 95% confidence intervals (CIs) were estimated using an unconditional logistic model and adjusted for age and sex (aORs). Age adjustment was conducted with age as a continuous variable. Age at smoking initiation, cigarettes consumed, duration of smoking, pack-years of smoking and FTND scores were compared between individuals with the TA/AA and TT genotypes of IL8 using the two-sample Wilcoxon rank-sum test. The frequencies of having tried to quit smoking were compared between the two groups (TA/AA and TT) using the chi-square test.

Results

A total of 453 Japanese outpatients who visited to Aichi Cancer Center Hospital were the subjects of the present research. Table 1 shows their characteristics according to age and sex. The age ranged from 20 to 79 years. Current smokers accounted for 37.7% of the males and 17.2% of the females. Three of the subjects (0.7%) could not be genotyped for IL8 −251T/A. The allele frequencies of IL8 −251T/A were 0.686 and 0.314 and the genotype distribution was in Hardy–Weinberg equilibrium (P=0.155).

Table 1 Characteristics of the subjects according to sex
Full size table

Table 2 shows the genotype distribution according to smoking status. Individuals with the IL8 −251TT genotype were more frequent among current smokers (52.5%), former smokers (51.0%) and ever smokers (51.9%) than among never smokers (45.3%), although this was not significant (P=0.408 and 0.165, respectively). For males, the frequencies of IL8 −251TT among current, former and ever smokers were not significantly different from that among never smokers. On the other hand, for females, the frequency among ever smokers (63.2%) was significantly higher than among never smokers (47.9%, P=0.030) but not for current smokers (60.0%, P=0.145). Among former smokers, the frequency of the IL8 −251TT genotype (69.6%) was higher than that among never smokers, but this was not statistically significant, possibly due to the small sample size (P=0.05). Table 3 lists the age-, sex- and drinking habit-adjusted ORs for the IL8 −251T/A polymorphism with reference to smoking behavior. The aORs for being an ever smoker for IL-8 TA and IL-8 AA were 0.50 (95% CI 0.30–0.82, P=0.006) and 0.56 (0.26–1.23, P=0.164) in total, respectively. Similarly, those for being a current smoker were 0.54 (95% CI 0.31–0.95, P=0.033) and 0.40 (0.15–1.10, P=0.076) in total, respectively. The participants with more A allele tended to be ever smokers (P for trend=0.016) and current smokers (P for trend=0.016). These tendencies were more evident in females than in males. When the IL8 −251TT genotype was defined as the reference, the aOR for being an ever smoker for IL8 −251TA/AA combined was 0.51 (0.32–0.81, P=0.005). A stronger association with ever smoking was noted for females than for males; the aORs for being an ever smoker for the IL8 −251TA/AA genotypes combined were 0.48 (0.26–0.89, P=0.019) for females and 0.56 (0.26–1.18, P=0.125) for males. A significant association with current smoking was observed in total (aOR 0.52, 95% CI 0.30–0.89, P=0.032) but the sex-stratified ORs were not significant. No significant difference was found between current and former smokers.

Table 2 Genotype distributions of IL8 −251T/A according to smoking status
Full size table
Table 3 Adjusted odds ratios (aORs) and 95% confidence intervals (95%CI) for IL8 −251T/A
Full size table

The characteristics of ever smokers are shown in Table 4. Males with IL8 −251TT started significantly earlier; median age at smoking initiation was 19.2 and 19.9 years, respectively, for TT and TA/AA (P=0.042). In addition, they had a tendency to consume more cigarettes; median cigarettes/day was 27.3 for TT and 22.9 for TA/AA, respectively (P=0.077). Likewise, the FTND score for those with IL8 −251TT was significantly higher than those with −251TA/AA (4.9 and 4.0, respectively, P=0.023). On the other hand, females showed no differences in the characteristics of ever smokers between IL8 −251TT and TA/AA. Furthermore, no significant association was observed after age and sex- or age-adjustment in the logistic model.

Table 4 Characteristics of ever smokers according to IL8 −251T/A genotype
Full size table

Discussion

The results obtained in the present study indicate significant associations between IL8 −251T/A polymorphism and smoking behavior among a Japanese sample population. First, we found significantly low aORs for ever or current smokers with the IL8 −251TA or AA genotypes, known to cause high production of IL8 compared to the TT genotype (Hull et al 2000), when never smokers were employed as controls. On the other hand, when former smokers were the controls, the aORs for current smokers were not found to be significant. These results suggested that the IL8 −251T/A genotype might be associated with the initiation, but not with the continuation of smoking, especially in females. Second, male smokers with the TT genotype of IL8 −251 smoked earlier and appeared to have higher nicotine dependence than those with the TA/AA genotypes, although these were not statistically significant. The results, however, suggested that IL8 −251T/A might influence the characteristics of habitual smokers. Our data support the hypothesis that a constitutional predisposition to inflammatory reactions to cigarette smoke influences smoking behavior, and this is in line with a previous report that a polymorphism in IL1B, another inflammatory cytokine, impacts on habitual smoking (Hamajima et al 2001a).

Exposure to cigarette smoke has been shown to induce IL8 release from bronchial epithelial cells and cause neutrophil accumulation in the airways (Chalmers et al 2001; Kuschner et al 1996; Mio et al 1997). This may contribute to airway inflammation in smokers. However, to the best of our knowledge, the association between smoking status and either IL-8 activity in the airway or degree of airway inflammation has been clearly elucidated. Besides, the influence of IL8 −251T/A polymorphism on the inflammation induced by tobacco smoking is still unclear.

IL8 −251T/A is found to influence IL-8 production, tending to result in the highest levels for the AA genotype and the lowest levels for the TT genotype, with AT producing an intermediate amount (Hull et al 2000). Fifteen polymorphisms have been identified on IL8 to date (Hull et al. 2004). T781C but not −251T/A was found to have an actual impact on protein-DNA binding (Hacking et al 2004) and −251T/A might be a surrogate marker for functional T781C. For this Japanese population, we examined A-251T and G396T polymorphisms (90.0% of −251TT carry 396TT, 90.5% of −251TA carry 396TG and −251AA carry 396GG, our unpublished data). This relation correlates with observations in a European population (99% of −251T carry 396G and 95% of −251A carry 396T (Hull et al 2001), which supports the applicability of −251T/A as a surrogate marker for other functional polymorphisms. Further examination of this question in future studies is warranted.

The genotype distribution of IL8 −251T/A was here confirmed to be in accordance with the Hardy–Weinberg law of equilibrium, so that inappropriate subject selection and incorrect genotyping of the polymorphism were unlikely. The frequency of the A allele (0.31) in our study subjects was similar to that for other Japanese populations reported previously (0.28), and lower than those described for Caucasians (0.44–0.55) (Hacking et al 2004; Hull et al 2001, 2004; McCarron et al 2002) and Africans (0.89) (Hull et al 2001), but the observed associations were unlikely to be due to selection bias or information bias, for the following reasons. The cases (current or ever smokers) and controls (never and former smokers) were recruited within the same framework and the rate of current smokers was validated to be same as in the general population (Inoue et al 1997). Smoking habits aside, the non-cancer outpatients of the Aichi Cancer Center were representative of the general population in relation to general lifestyles (Inoue et al 1997). The medical conditions of all 590 participants did not statistically differ between ever and never smokers; 17.7% of never smokers and 13.2% of ever smokers had cancer, and 10.2% and 7.8% had lung and/or cardiovascular disease, respectively. In our previous study of females, more than 66% of the non-cancer outpatients at ACCH did not have any specific medical condition. The others did have specific diseases, but the most common were benign tumors and/or non-neoplastic polyps (Hamajima et al 1995). The situation is similar for males (own unpublished data), which differs considerably from the situation in other developed countries, where people visit local general clinics first, and are then referred to hospitals which function as secondary and/or specific facilities for further medical conditions. We investigators did not know the genotypes at the time of enrolment, and therefore could not be biased regarding the subject population. The procedure of collecting information would also not be expected to introduce bias, because the participants’ smoking status was masked when genotyping was conducted. Genotype itself, in general, does not appear to influence intention to join a study. The effect of population stratification appears very limited in Japan, in comparison with multiethnic countries such as the USA. Around 60% (590/1,022) of the first-visit outpatients provided DNA, and the distributions of smoking status did not differ significantly between the donors and non-donors (data not shown). Thus, participants’ smoking status was unlikely to influence the rate of enrolment and appropriate sampling was conducted in terms of smoking behavior in the present study. Cases of cancer, respiratory and cardiovascular disease may be strong motivatory factors for quitting smoking (Pinto et al 1999). Besides, we found that a history of cancer, respiratory or cardiovascular disease was a strong factor for quitting smoking in an intervention study for smoking cessation conducted at ACCH, while genetic polymorphisms were not (data not shown). In the present study, the distribution of smoking status (25.8% for current, 22.1% for former and 52.1% for never smokers) was completely different from that of participants who were excluded from the study (19.0% for current, 38.2% for former and 43.8% for never smokers, P=0.002). Exclusion of cases of cancer, respiratory and cardiovascular disease might contribute to eliminating selection bias.

Regarding the potential for genetic predisposition to habitual smoking, polymorphisms have hitherto been investigated with genes for neurotransmitters and factors in nicotine metabolism such as the dopamine receptors DRD2 (Hamajima et al 2002; Noble et al 1994; Spitz et al 1998; Yoshida et al 2001) and DRD4 (Hutchison et al 2002; Shields et al 1998), a dopamine transporter (SLC6A3) (Lerman et al 1999; Sabol et al 1999) and a serotonin transporter (5-HTT) (Ishikawa et al 1999), as well as dopamine metabolism-related genes (MAO) (Ito et al 2003; Johnstone et al 2002; McKinney et al 2000) and cytochrome p450 2A6 (CYP2A6) (Pianezza et al 1998). Taking the present results into account, one may hypothesize that smoking behavior is influenced not only by neuropsychological but also by inflammatory factors.

Although the findings for females must be regarded as preliminary because of the modest sample size, we performed an analysis of sex differences because this is reported to impact on tobacco sensitivity and tolerance (Eissenberg et al 1999; Gritz et al 1996). Indeed, the associations between smoking behavior and IL8 −251T/A among females were more notable than those among males, supporting the previous conclusion that bronchial hyper-responsiveness in women may have a greater role in susceptibility to initiation of smoking (Leynaert et al. 1997). Significant associations with IL8 −251T/A were observed when never smokers were defined as controls but not when comparing current smokers with former smokers, suggesting that the inflammatory responsiveness to cigarette smoke influences starting smoking more than quitting.

In conclusion, the present study shows that the IL8 −251T/A genetic polymorphism may influence smoking behavior. An understanding of the molecular basis of this may provide new insights into determinants of habit initiation and could lead to more effective strategies for preventing this and for helping smokers to quit.