Background

Nicotine is the primary addictive component of tobacco, and its addictive effects operate through dopamine neurotransmission in the brain's mesolimbic "reward" pathway. Genes that code for constituents of the dopaminergic system are considered candidates for involvement in addictive behaviors, including smoking. Within a cohort of women who participated in a randomized smoking cessation trial, we examined the relation of polymorphisms in genes related to dopamine synthesis (tyrosine hydroxylase [TH]), receptor activation (DRD2, DRD3, DRD4), transport and synaptic reuptake (the dopamine transporter [SLC6A3]) and metabolism (catechol-O-methyltransferase [COMT]) with the ability to quit smoking.

The genes and polymorphisms were selected based on prior reports of an association with smoking behavior as well as reports of associations with other addictive behaviors such as alcoholism. We hypothesized that alleles associated with reduced dopaminergic neurotransmission (whether due to decreased dopamine production, increased metabolism, increased synaptic reuptake, or reduced receptor activation) would be less common among women who quit smoking than among continuing smokers. The DRD2 receptor gene, for example, is polymorphic in the promoter region where the gene can exist as an insertion or deletion variant at position -141 (rs1799732). Because the deletion variant is associated with lower promoter activity [1], we expected that the polymorphic form would presumably decrease dopamine neurotransmission and would be associated with a decreased ability to quit smoking. Similarly, downstream on the same DRD2 receptor gene exists a polymorphism identified by the Taq1A restriction enzyme (A1 allele) (rs1800497), which is associated with reduced receptor binding [2, 3]. Because of this reduced receptor binding ability, we hypothesized that the A1 allele would also be associated with a decreased ability to quit smoking. In contrast, the single nucleotide substitution of G by A in codon 158 of the COMT gene resulting in an amino acid change from valine to methonine (rs4680) is associated with low enzyme activity and thermolability [4]. We hypothesized that a reduced ability of this COMT polymorphic enzyme to metabolize dopamine would presumably increase dopamine neurotransmission and be associated with a higher likelihood of quitting smoking.

We also selected other polymorphisms that have been associated with smoking or addictive behaviors in prior reports, and based our hypothesis of their affect on smoking cessation from existing literature. These include DRD3 Ser9Gly (rs6280) DRD4 variable number tandem repeat (VNTR) in third exon, SLC6A3 VNTR in 3' untranslated region, and TH VNTR in first intron. Table 1 summarizes the genes and polymorphisms we assessed, the possible effect of the variant allele on dopaminergic neurotransmission, and the hypothesized direction of effect on smoking cessation. Because results of prior studies of the relation between these selected polymorphisms and smoking behavior have been inconsistent, our assessment of these polymorphisms in a population of female smokers who were highly motivated to quit could contribute to a greater understanding of genetic influences on smoking cessation.

Table 1 Summary of candidate genes, polymorphisms, and their presumed effect on smoking cessation under the hypothesis that reduced dopaminergic neurotransmission is associated with lower likelihood to quit smoking.
Full size table

Methods

Study population

Eligible individuals were female residents of Washington State who had participated in a double-blinded randomized controlled trial conducted at the Fred Hutchinson Cancer Research Center (FHCRC) in Seattle, Washington in 1993 and 1994, investigating a pharmacologic agent, d,l-fenfluramine, as a component of smoking cessation treatment [5]. The 723 participants were between 26 and 65 years of age, weighed within 85–150 percent of ideal weight for height based on the Metropolitan Life Tables, and smoked more than 10 cigarettes per day at randomization. At entry into the trial, each woman completed questionnaires regarding: smoking behavior; other behaviors, including patterns of food and alcohol consumption; smoking-related beliefs; confidence in quitting; and a variety of psychological measures. All women received a self-help smoking cessation program from the American Lung Association. Women were randomized to receive either d,1-fenfluramine or placebo control in a double-blinded fashion. Quit outcome was assessed at 12-month follow-up.

For the current genetic study, we contacted clinical trial participants to request a biological specimen (blood or buccal cells) and to re-assess smoking behavior several years after completion of the trial (median, 3.3 years). Data and specimens were obtained from 593 women, 517 of whom provided blood and 76, buccal cells. DNA extraction and genotyping were conducted at the Functional Genomics Laboratory of the Center for Ecogenetics and Environmental Health at the University of Washington. The following polymorphisms were assessed: COMT Val158Met (rs4680), DRD2 -141C Ins/Del (rs1799732), DRD2 Taq1A (rs1800497), DRD3 Ser9Gly (rs6280), and VNTRs in the DRD4, SLC6A3, and TH genes. Laboratory personnel were blinded to smoking status and quality control samples were used to ensure accurate genotyping techniques. Primers, probes and conditions for genotyping assays are available upon request. All study activities received human subjects approval by the Fred Hutchinson Cancer Research Center Institutional Review Board.

Outcome measurement

We assessed smoking status at two points in time. First, women were considered "short-term" quitters if they had not smoked for at least the seven day period before their 12-month follow-up in the original clinical trial. Second, we defined "long-term" quitters as those who reported smoking abstinence for at least six consecutive months immediately before their 3-year interview for the current study. We analyzed plasma samples from self-reported long-term quitters for cotinine to validate reported smoking status. Women initially categorized as long-term quitters who had a cotinine level of 14 nanograms per milliliter or greater [6] and who did not report recent use of nicotine gum and/or patch (n = 4) were re-categorized as continuing smokers.

Statistical analysis

For VNTR polymorphisms, we categorized alleles based on number of repeats according to groupings used in prior studies. For DRD4, the alleles containing 2–5 repeats were considered short and alleles containing 6–10 repeats were considered long. For SLC6A3, alleles were categorized based on the presence or absence of a 9-repeat. For TH, alleles were categorized based on the presence or absence of a 10-repeat. With the exception of the COMT Val158Met polymorphism, for which there were a sufficient number of individuals in the homozygous variant (Met/Met) category to consider separately, we grouped individuals who were heterozygous with those who were homozygous variants for analysis. We computed relative risks (RR) of short-term and long-term smoking cessation using the Mantel-Haenszel estimator to compute crude and adjusted RRs. Ninety-five percent confidence intervals (CI) were calculated for all RR estimates. Analyses were conducted for the entire study population and also after restricting to non-Hispanic Whites. We assessed potential confounding by age, race, and intervention arm of the randomized trial, and conducted analyses separately among younger and older women. Because prior studies reported an effect modification between SLC6A3 and DRD2 Taq1A polymorphisms on smoking cessation, we examined the influence of the combination of these two polymorphisms on smoking cessation in our population. We also explored whether any of the polymorphisms of interest modified the effect of treatment intervention on smoking cessation in this population. Data management and statistical analyses were conducted using STATA Version 8 (StataCorp, College Station, TX, USA) and SAS software Version 6.12 (SAS Institute Inc., Cary, NC, USA).

Results

Participants (n = 593) and non-participants (n = 130) did not differ on demographics, intervention arm, baseline smoking and drinking characteristics, or short-term quit status (21 percent and 19 percent, respectively, were considered quitters; p = 0.7). Long-term quit status was only assessed in those who participated in the long-term interview (median, 3.3 years after enrollment in clinical trial). The proportion of women randomized to d,l-fenfluramine was similar between those who did and did not quit smoking for both cessation outcomes. The majority of participants (93%) were non-Hispanic Caucasian. Compared to women who continued to smoke at either time point, those who were short- or long-term quitters had higher income and educational levels, consumed less alcohol, started smoking at a later age, smoked fewer numbers of cigarettes per day, and had smoked for a shorter period of time when assessed at enrollment in the trial (Table 2).

Table 2 Characteristics of 593 women by short-term and long-term smoking cessation status, Seattle, WA, 1993–1998.
Full size table

Information on short-term quit was available for 563 women, 116 (21%) of whom were categorized as nonsmokers and 447 (79%) as smokers at this time point. For long-term cessation outcome, 93 (16%) women had quit, and 500 (84%) continued to smoke. Fifty-nine of 116 women (51%) were non-smoking at both time points. Those who had quit for the short-term were 6.9 times (95% CI: 4.7, 10.0) more likely to be non-smoking at the later time point.

Distributions of genotypes for all the polymorphisms did not deviate to any appreciable extent from expectation predicted by Hardy-Weinberg equilibrium. We observed no association between any of the individual polymorphisms examined and smoking cessation either for the short- or long-term quit outcomes (Table 3). Although, in a few instances, RR estimates appeared to be suggestive of a weak association with smoking (albeit with wide confidence intervals), either the direction of effect was in the opposite of that hypothesized or results of the short-term and long-term cessation endpoints were inconsistent. For example, while we observed a decreased likelihood of quitting in the short-term (RR = 0.9; 95 percent CI: 0.6, 1.5) and long-term (RR = 0.7; 95% CI: 0.4, 1.3) among women with two copies of the COMT Met allele, we had hypothesized that the lower activity Met allele would occur more commonly among women who quit smoking. While a 30% increase in likelihood of quitting in the short-term was observed among women with one or more 9-repeat alleles of the SLC6A3 gene (RR = 1.3; 95% CI: 0.9, 1.8), no association with long-term quitting was observed (RR = 1.1; 95% CI: 0.7, 1.6).

Table 3 Relation of gene polymorphisms with short-term and long-term smoking cessation among 593 women, Seattle, WA, 1993–1998.
Full size table

Adjustment for race, age, or intervention arm of the randomized trial had no appreciable influence on the risk estimates. Analyses restricted to non-Hispanic Caucasian women yielded similar results, as did analyses conducted separately among younger (<50 years) and older (≤ 50 years) women. We also conducted analyses in which we compared women who were categorized as both short-term and long-term quitters to women who had not quit smoking at either of these two time points; again, we observed no relation of any of the polymorphisms assessed with the likelihood of quitting smoking. We further observed no association of any set of combinations of dopamine receptor polymorphisms with smoking cessation for either outcome (data not shown).

There was no relation between having a SLC6A3 9-repeat allele and short-term (RR: 0.9; 95% CI: 0.6, 1.4) or long-term (RR: 0.8; 95% CI: 0.5, 1.4) quitting among women with the more common A2/A2 genotype for DRD2 Taq1A. However, women with one or two copies of the SLC6A3 9-repeat allele who also had either the DRD2 Taq1A A1/A1 or A1/A2 genotypes were significantly more likely to quit in both the short- (RR = 2.5; 95% CI: 1.3, 4.7) and long-term (RR = 1.8; 95% CI: 0.9, 3.6). We also assessed the moderating effects of genotype on the pharmacologic agent, d,l-fenfluramine, offered in the clinical trial in which these women previously participated. Except for DRD3 Ser9Gly and DRD4 VNTR polymorphisms, d,1-fenfluramine did not appear to have an effect on smoking cessation in any subgroup for any of the polymorphisms we assessed (data not shown). The likelihood of quitting smoking associated with the use of d,1-fenfluramine was slightly increased for those who carried at least one Gly allele in the DRD3 gene (RR = 1.5; 95% CI: 0.9, 2.6), but not for those homozygous for the Ser allele (RR = 1.05; 95% CI:, 0.62, 1.78). Among those with two short-repeat alleles in the DRD4 VNTR, d,1-fenfluramine was significantly associated with a higher chance of quitting (RR = 1.8; 95% CI: 1.0, 3.9). Women with one or two long-repeat alleles, however, did not benefit from the intervention (RR = 0.8; 95% CI 0.4–1.4). Table 4 summarizes our findings regarding these modifying effects.

Table 4 Gene-gene and gene-treatment interaction for smoking cessation in 593 women, Seattle, WA, 1993–1998.
Full size table

Discussion

Our study has several strengths. Relative to cross-sectional studies that assess smoking status at a single time point, the cohort design of this study allowed us to focus on cessation independent of other aspects of smoking behavior (such as initiation) and to assess quitting behavior at two widely separated time points. Examining genetic influences within a group of moderate to heavy smokers who were sufficiently motivated to quit that they chose to participate in a randomized trial might be expected to enhance our ability to isolate and identify genetic effects on smoking cessation (independent of psychosocial factors related to motivation), relative to studies based on comparisons of smokers and nonsmokers. The conduct of this study within a female population, coupled with the lack of effect of the d,l-fenfluramine intervention on smoking cessation [7], minimizes the extent to which our results might be influenced by gender or treatment arm.

Nevertheless, the study is vulnerable to several potential sources of error and issues of generalizability. Some misclassification of smoking status might be expected, since determination of short-term quitting was based entirely on self-report. Long-term quitting was verified with plasma cotinine levels. However, because cotinine levels of a typical habitual smoker (200 ng/ml) [8] decline to nonsmoking values by the third day of abstinence (based on the average half-life of 16–19 hours) [9], individuals who had smoked at some point within the six month time period used to define this endpoint may still be misclassified. Furthermore, our results may not be generalizable to populations that were either excluded or not well-represented in our study such as men, women who are not within 85–150% of ideal body weight, and non-white racial or ethnic populations.

Clear and consistent findings have not emerged in research examining candidate gene polymorphisms in relation to smoking behavior. Results of studies of the relation between the COMT Val158Met polymorphism and smoking behavior have been inconsistent [10–12]. Several studies that have examined the relation of this polymorphism with smoking cessation or nicotine dependence among smokers participating in cessation trials reported positive findings. Collila and colleagues observed that women recruited from a smoking cessation trial who had the Met/Met genotype were more likely to have abstained from smoking for at least seven days (OR = 2.96; 95% CI: 1.07, 8.14) and for a prolonged period of time (OR = 3.23; 95% CI: 1.13, 9.20) [10]. In another report, investigators observed a positive relation between nicotine dependence and the presence of the high activity (Val) allele of COMT (p = 0.0072) in a population of male and female smokers enrolled in cessation trials; however, they were unable to replicate these findings in an independent group of trial participants [13].

Observational studies assessing the DRD2 Taq1A polymorphism and smoking characteristics have also had inconsistent findings. Whereas the results of some studies suggest an association between the A1 genotype and current or past smokers relative to nonsmokers, [14–17]. Studies of subjects recruited from clinical trials have also reported associations of smoking behavior with the DRD2 Taq1A polymorphism in non-Hispanic Caucasian populations. In a study of 134 male and female smokers from a trial of venlafaxine or placebo plus standard of care (e.g., counseling and nicotine patch), those with the A2/A2 genotype were more likely to quit smoking than smokers with at least one A1 allele [18]. In contrast, among enrollees of a clinical trial of buproprion plus counseling, women with at least one DRD2 Taq1 A1 allele were somewhat less likely to report smoking at 12-month follow-up compared to women who carried two A2 alleles (OR = 0.76; 95% CI: 0.56, 1.03); no such association was observed among men [19]. In a study of 600 male and female smokers enrolled in a clinical trial of befloxatone or placebo plus counseling, individuals with genotypes containing the A1 allele were less likely to sustain abstinence during the last four weeks of the treatment period (OR = 0.74; 95% CI: 0.46, 1.18) [20]. Overall, a meta-analysis of 12 studies of DRD2 Taq1A polymorphisms found no association between this polymorphism and smoking cessation (OR = 1.17; 95% CI: 0.89, 1.55) [21].

Opposite directions of associations with the DRD2 Taq1A polymorphism have been observed in some studies of ethnic populations. In a study of lung cancer, the A1/A1 genotype was associated with current smoking status among Mexican-American but not in African-American control subjects [22]. Among 332 Japanese individuals recruited from an outpatient clinic of Aichi Cancer Center, those with the A2/A2 genotype displayed a greater likelihood of having ever smoked (OR = 3.68; 95% CI: 1.50, 9.05) or were more likely to be current smokers (OR = 3.72; 95% CI: 1.23, 11.2) relative to those with the A1/A1 genotype after adjusting for sex and age [23]. Another study of Japanese individuals recruited from the same clinic partially replicated this finding. Compared to men with the A1/A1 genotype, men with the A2/A2 genotype were more likely to be current smokers (OR = 2.32; 95 percent CI: 1.02, 5.29). This association, however, was not apparent in women [24]. In a population of Chinese smokers, those with the A2/A2 genotype smoked a greater number of cigarettes per day than smokers with at least one A1 allele [25]. Among 187 healthy Koreans genotyped for the DRD2 Taq1A polymorphism, men carrying the A1 allele were more commonly smokers (p = 0.049) whereas women with the A1 allele were more likely to be nonsmokers (p = 0.018), with no association observed when men and women were pooled together [26].

Results of studies on the SLC6A3 VNTR, TH VNTR, DRD2 -141C Ins/Del and DRD3 Ser9Gly polymorphisms are either lacking or conflicting. SLC6A3 9-repeat genotypes were observed to correlate with non-smoking status, smoking cessation, later age of smoking initiation, and long periods of quitting in early studies [27, 28], but these associations were not observed in subsequent studies [29, 30]. In a meta-analysis of four studies, the SLC6A3 VNTR polymorphism was not associated with smoking cessation in a fixed-effects model (OR = 0.85; 95% CI: 0.68, 1.08) or random-effects model (OR = 0.89; 95% CI: 0.63, 1.28) [21]. An examination of the TH VNTR polymorphism found no association with the likelihood of smoking in either Caucasians or African-Americans. Among smokers, however, the TH1 allele (the longest polymorphic repeat, corresponding to the 10-repeat allele in our study) was associated with a higher smoking rate and the TH4 allele (corresponding to the 7-repeat allele) was associated with a lower smoking rate [31]. The TH4 allele was also observed to be inversely associated with nicotine dependence among adolescent smokers in two independent populations in Australia [32, 33]. The DRD2 -141C Ins/Del polymorphism was examined in one study among a Japanese population, but was not associated with smoking status [23]. DRD3 Ser9Gly has not been previously examined in relation to smoking behaviors, although it has been assessed, with inconsistent results, in studies of other addictive behaviors.

Variations in the definition or timing of smoking assessment may explain, in part, some of the dissimilar results between studies. For example, Vandenberg and colleagues observed differences in the relation of smoking with the SLC6A3 9-repeat allele when smokers were compared to "never smokers" (i.e., individuals who had never smoked a cigarette) versus "nonsmokers" (i.e., those who had ever smoked less than 100 cigarettes) [30], suggesting that important information may be gained by distinguishing between those who have never smoked and those who have smoked less than 100 cigarettes. Furthermore, smoking cessation is a dynamic process characterized by high rates of relapse, possibly influencing the power to detect effects at different time periods. In several smoking cessation trials, the effect of time on weekly abstinence rates was appreciable: abstinence rates rose through the first several weeks and declined over time [18, 20]. Even among individuals who abstain from smoking for seven months, relapse rates between 7% and 35% have been reported [34].

Gender appears to be related to smoking cessation in complex ways. Men have been reported to have lower rates of relapse relative to women [35], and smoking cessation may be more difficult for women for reasons pertaining to demographic and smoking history [36], weight control, or social support [37]. As described above, several studies have reported gender differences in the relation of the DRD2 Taq1 polymorphism with smoking behavior.

Other studies have observed gene-gene interactions with smoking status or cessation. Lerman and colleagues reported an interaction of the DRD2 Taq1A and SLC6A3 VNTR polymorphisms in a population of 289 smokers and 233 controls in which non-Hispanic Caucasian participants with SLC6A3 9-repeat genotypes were more likely to be nonsmokers relative to those without a 9-repeat allele among those with the DRD2 Taq1 A2/A2 genotype (p = 0.008). Among those with the DRD2 Taq1 A1 genotypes, the SLC6A3 9-repeat was not associated with smoking status (p = 0.51) [28]. These investigators also observed a gene-gene interaction between the same two polymorphisms and smoking cessation in a population of 418 smokers of European Caucasian descent enrolled in a clinical trial of bupropion [38]. In contrast to these findings, we observed an increased likelihood of smoking cessation among a different group defined by DRD2 Taq1A and SLC6A3 9-repeat genotypes: women with the SLC6A3 9-repeat genotypes had an increased likelihood of quitting in either the short-term or long-term, relative to women without a SLC6A3 9-repeat allele, only if they also possessed the DRD2 Taq1 A1 genotypes. There was no apparent association between the SLC6A3 9-repeat genotypes and smoking cessation for either time point among women with the DRD2 Taq1 A2/A2 genotype. The discrepancies of these various studies are not readily explainable by differences in the distribution of racial, ethnic or gender characteristics of the study populations.

In our exploration of whether d,1-fenfluramine may be more effective in certain subgroups defined by genotypes, we observed an increased likelihood of quitting smoking associated with d,1-fenfluramine among those with one or two DRD3 glycine alleles in the long-term (RR = 1.5; 95% CI: 0.9, 2.6), but not among smokers without the glycine allele (RR = 1.1; 95% CI: 0.6 1.8). For women with two short alleles in the DRD4 VNTR polymorphism, d,1-fenfluramine appeared more effective in helping them quit smoking than for women with one or two long alleles. Studies examining the influence of genotypes on d,1-fenfluramine's effectiveness in smoking cessation are not available, although a recent study has identified a more favorable outcome for bupropion treatment among European-American smokers with the COMT haplotypes (rs737865 and rs165599) [39]. Similar studies may be important in identifying likely responders for common smoking cessation treatments.

Conclusion

Our results, together with the inconsistent findings of prior studies, provide little support for an independent relation of the candidate polymorphisms we assessed with smoking cessation. However, a substantial literature based on twin studies suggests that aspects of smoking behavior are highly heritable. Findings from our exploration of effect modification suggest possible important gene-gene and gene-treatment interactions for smoking cessation. Future studies examining genetic influences on smoking cessation may prove more informative if they are designed to completely characterize variation in the genes of interest, or to more effectively identify subgroups of smokers who are more genetically inclined to quit smoking or for whom certain pharmacologic interventions may be more efficacious.