Neurotransmission-Related Genetic Polymorphisms, Negative Affectivity Traits, and Gender Predict Tobacco Abstinence Symptoms across 44 Days with and without Nicotine Patch

Abstract

Genetic and personality trait moderators of tobacco abstinence-symptom trajectories were assessed in a highly controlled study. Based on evidence suggesting their importance in stress reactivity and smoking, moderators studied were serotonin transporter gene (5-HTTLPR) and dopamine D2 receptor gene (DRD2) polymorphisms, and negative affect-related personality traits. Smokers were randomly assigned to quit smoking with nicotine or placebo patches. Financial incentives resulted in 80% verified abstinence across the 44-day study. Individuals with one or two short alleles of 5-HTTLPR (S carriers) experienced larger increases in negative affect (NA) symptoms than those without a short allele. Nicotine replacement therapy (NRT) alleviated anxiety only in S carriers. NRT reduced NA to a greater extent in DRD2 A1 carriers than in A2A2 individuals during the first two weeks of treatment (when on the 21 mg patch); however, A1 carriers experienced a renewal of NA symptoms when switched to the 7mg patch and when off patch, while A2A2 individuals continued to benefit from NRT. The results suggest that the effects of genotype and treatment may vary across different durations of abstinence, treatment doses, and genotypes.

Differences in genetic makeup are a potentially important, yet little studied factor that may help explain individual differences in negative affect (NA)-related tobacco abstinence symptoms (TAS) (reviewed by Pergadia, Heath, Martin, & Madden, 2006). The large individual differences in the severity and time course of NA-related TAS (Piasecki, Jorenby, Smith, Fiore, & Baker, 2003a, 2003b, 2003c) appear to be important because relapse to smoking occurs more often in those experiencing high levels of NA (Shiffman & Waters, 2004). However, the factors that predict severity and nature of NA-related TAS have only been partially characterized. While NA-related personality traits (NATs) and history of major NA-related psychiatric disorder have been found to predict both severity of NA during abstinence and relapse (Covey, Glassman, & Stetner, 1990; Gilbert, McClernon, et al., 1999; Gilbert et al., 2002; Gilbert, Crauthers, Mooney, McClernon, & Jensen, 1999; Zvolensky, Lejuez, Kahler, & Brown, 2004), these factors and degree of nicotine dependence (Gilbert et al., 2002; Piasecki et al., 2003c) account for only a modest amount of the large individual differences in TAS. Identification of genetic moderators and predictors of NA-related TAS might provide a better understanding of individual differences in TAS and also lead to better treatments for nicotine dependence.

Twin studies indicate that genetic factors and NATs are associated with the likelihood of smoking and of severity of TAS (reviewed by Munafò, Clark, Johnstone, Murphy, & Walton, 2004; and Pergadia et al., 2006). Genetic factors account for approximately 50 percent of the variance in smoking initiation and persistence of smoking in the United States and other English-speaking countries (reviewed by Heath, Madden, Slutske, & Martin, 1995); and the effects of genes on smoking may in part be mediated by NATs (collectively referred to as “neuroticism”) (Kendler, Neale, MacLean, Heath, Eaves, & Kessler, 1993; reviewed by Gilbert & Gilbert, 1995). Given the initial evidence linking broad genetic influences and NATs to NA-related TAS, it is natural to assess how specific genetic polymorphisms and specific NATs moderate TAS severity and the effects of nicotine replacement therapy (NRT) on TAS. The study of NATs and NA is especially important given the recent convincing argument by Baker, Fiore, McCarthy, and Majeskie (2004) that NA is the “motivational core” of the nicotine withdrawal syndrome and nicotine dependence.

Genetic polymorphisms of neurotransmitter receptors and transporters likely account for much of the genetic moderation of nicotine abstinence symptoms (Munafò et al., 2004). Given that nicotine can modulate the release of both serotonin (Ribeiro, Bettiker, Bogdanov, & Wurtman, 1993) and dopamine (Benwell & Balfour, 1992; Corrigall, Cohen, & Adamson, 1994), it may be able to optimize serotonergic and dopaminergic functioning in individuals who find nicotine to be especially rewarding (dependent smokers). The serotonin transporter (5-HTT)-linked polymorphic region (5-HTTLPR) is a functional polymorphism in the promoter for the serotonin transporter gene that consists of two allelic forms: a short (S) variant with 14 copies of a 20–23 base pair repeat unit, and a long (L) variant consisting of 16 copies. At least two studies have found smoking-related behavior to be associated with interactions of 5-HTTLPR genotype and neuroticism (Hu et al., 2000; Lerman et al., 2000), though others have failed to find such relationships (e.g., Munafò, Roberts, Johnstone, Walton, & Yudkin, 2005).

Interest in 5-HTTLPR genotype is enhanced by the finding that carriers of the S variant are vulnerable to heightened NA in response to a variety of different stressors. In their review of the literature on genetic polymorphisms of serotonin transporter, Hariri and Holmes (2006) concluded that the S allele increases emotional reactivity to stressors. A review by Schmitz, Hennig, Kuepper, and Reuter (2007) also related the S allele to neuroticism (an overall measure of NATs). The S allele has been found to interact with stressful life events in predicting subsequent symptoms of clinical depression (Caspi et al., 2003). Given that most smokers experience quitting smoking as a stressor, one might expect the S allele to be associated with heightened NA-related TAS and for nicotine replacement therapy (NRT) to benefit S allele carriers more than LL individuals. In one of the few studies assessing this association of the S allele to smoking, Lerman et al. (2000) found that neuroticism was positively associated with smoking motivation to reduce NA only in individuals with an S allele. However, we are not aware of any study that has assessed potential moderating effects of the S allele on NA-related smoking abstinence symptoms or the effects of NRT on such symptoms in a prospective design.

The large literature showing dopamine to be an important component of the reinforcing effects of nicotine (Benwell & Balfour, 1992; Corrigall, Cohen, & Adamson, 1994) makes genes associated with dopamine metabolism, receptors, and transporters leading candidates involved in the genetics of smoking motivation and abstinence symptoms (Munafò et al., 2004). It should be noted that recently the most widely analyzed in DRD2-related genetic polymorphism, Taq1A, was recently shown to reside within the coding region of the ankyrin repeat and kinase domain containing 1 (ANKK1) gene and located near the 3′ end of the DRD2 gene (Neville, Johnstone, & Walton, 2007). This close proximity allows linkage of the Taq1A polymorphism to DRD2 expression. The dopamine D2 receptor (DRD2) Taq1A A1 variant has been associated with an increased brain reactivity to experimental stress during smoking abstinence (Gilbert et al., 2004), reduced sustained abstinence (Cinciripini et al., 2004), and greater efficacy of NRT relative to placebo (Johnstone et al, 2004). However, a number of studies have failed to confirm these associations (reviewed by Munafò, Clark, et al., 2004). Nonetheless, in their literature review, Berrettini and Lerman (2005) concluded that variants in DRD2 genes may predict therapeutic response to NRT and other pharmacotherapies for nicotine dependence. Consistent with this possibility, recent studies have found the Taq1A A2A2 genotype to be uniquely associated with effects of bupropion on cessation (David, Strong, et al., 2007). The A1 allele appears to be associated with a reduced number of DRD2 receptors (Thompson et al., 1997), NATs, and compulsive behavior (Blum et al., 2000). The reduced number of DRD2 receptors associated with the A1 allele may lead to dysphoria and NA due to a lessened response to emotionally positive stimuli that normally inhibit NA (Blum et al., 2000; Gilbert, 1995). Thus, there is reason to rigorously test the hypothesis that the Taq1A polymorphism is associated with differential benefits from NRT.

While the above-noted evidence suggests that 5-HTTLPR and DRD2-related polymorphisms may moderate the effects of NRT on NA-related TAS, no studies have been explicitly designed to assess this possibility. Carefully designed tests are needed to limit dropout and relapse in studies in this area because of biases associated with the high dropout and relapse rates among those with high NA states and traits (reviewed by Gilbert & McClernon, 2000; Gilbert et al., 2002; Piasecki & Baker, 2000; Shiffman, West, & Gilbert, 2004). Individuals who experience the most severe NA-related TAS are least likely to maintain abstinence and remain in the study, thus NA-related TAS severity is often underestimated. The loss of data from high-TAS individuals may lessen the estimated associative strengths of TAS with genotypes and NATs.

The primary goal of the present investigation was to better characterize NA-related genetic influences on TAS with and without NRT over a multi-week period by maximizing abstinence and minimizing study dropout. A second goal was to replicate and extend previous findings suggesting that NATs predict withdrawal symptom severity by controlling the effects of two NA-related genetic variants. A priori hypotheses were that: a) 5-HTTLPR S allele carriers and Taq1A A1 allele carriers would experience greater increases in NA during abstinence and would benefit more from NRT; and b) NATs would be positively associated with severity of NA symptoms during abstinence. Specifically, in line with previous findings (e.g., Covey et al., 1990; Gilbert et al., 1998, 2002) and the Situation by Trait Adaptive Response (STAR)-model hypothesis that NAT dispositions to specific negative affects predict greater increases in corresponding affective states during smoking abstinence (Gilbert, 1995), it was hypothesized that trait depression would predict greater increases in depressive affect, while trait anger would predict greater increases in anger-irritability, and anxious traits would predict greater increases in anxiety during abstinence. The possibility that 5-HTTLPR and DRD2 genotypes may moderate the association of these traits with corresponding TAS was also explored.

Method

Participants

Smokers wanting to quit were recruited by newspaper ads and postings in a Midwestern university community from 1998 through 2004. Exclusion criteria included smoking fewer than seven cigarettes per day for the past two years, habitual cigarette nicotine deliveries of less than 0.6 mg, use of psychoactive drugs or medications other than alcohol and caffeine, alcohol use in excess of 28 alcoholic drinks per week, age less than 18 or greater than 50 years, non-English speaking, atypical sleep cycles, pregnancy, and serious medical, and visual problems.

Of the 209 participants who completed the baseline phase (Fig. 1), 171 were randomly assigned (by an 80:20 odds ratio using an urn technique) to the quit group and 38 were assigned to the group that continued to smoke. Group assignment occurred after completion of the final pre-quit experimental session. Those in the quit group were randomly assigned (50:50 chance) to the nicotine (NP) or placebo (PP) patch group. While patch assignment was double blind, assessment of blindness after the final experimental session revealed that 77 percent of the individuals in the NP patch group correctly guessed that they had been on the NP patch; while only 45 percent of on the PP group believed that they had been on the NP (χ² = 15.68, p < .001).

Figure 1.

Study flowchart.

Of those assigned to the NP group, 73 (81%) completed the study and smoked four or fewer cigarettes total across the 45-day abstinence period. Quitters who were not fully abstinent (N = 12), but smoked four or fewer cigarettes total, smoked an average of 1.92 (SD = 1.65) cigarettes total (0.04/day). Of those in the PP group, 68 (84%) met these criteria. Of the 38 assigned to the smoke group, 33 (87%) fulfilled requirements, including continuing to smoke at their habitual rate. Sample size was based on power estimates to detect moderate-sized effects with 95 percent likelihood.

The number of cigarettes smoked during the quit period was estimated by a combination of self-report, carbon monoxide (CO) concentration (8+ ppm), plasma or salivary cotinine (20+ ng/ml), and plasma nicotine (1.5+ ng/ml) and by knowledge of the half-lifes of these substances. If self-report and biochemical indicators differed, the indicator suggesting the greater number of cigarettes was taken as the number of cigarettes smoked. Comparisons of lapsers (four or fewer cigarettes) versus complete abstainers indicated that these groups did not differ in terms of pre-quit number of cigarettes smoked per day or Fagerström Test of Nicotine Dependence score (FTND; Heatherton, Kozlowski, Frecker, & Fagerström, 1991) scores (ps >.14). However, compared to the combined group of lapsers and abstainers, the combined group of drop-outs and relapsers (smokers of 5+ cigarettes during the abstinence period) did have higher FTND scores (5.6 vs. 4.7, p = .019) and baseline cotinine concentrations (268.5 vs. 208.7, p = .037).

Demographics, gender, and smoking characteristics of the nicotine, placebo, and smoke group completers did not differ significantly in terms of FTND dependence, NATs, age, gender, ethnicity, or genotypes (Table 1). Genotype variants were not significantly related to nicotine dependence, negative affectivity, or baseline mood status prior to smoking abstinence.

Table 1.

Descriptives of subjects in nicotine patch, placebo patch and smoke groups

	Nicotine Patch	Placebo Patch	Smoke
Assigned	90 (49 female)	81 (42 female)	38 (20 female)
Completed	73 (39 female)	68 (33 female)	33 (18 female)
Age (years)	27.0 (8.6)	25.4 (7.6)	28.1 (9.8)
Ethnicity	African American 2	African American 2	African American 0
	American Indian 1	American Indian 1	American Indian 1
	Asian American 3	Asian American 0	Asian American 0
	Caucasian 65	Caucasian 64	Caucasian 31
	Hispanic 2	Hispanic 1	Hispanic 1
FTND	4.9 (2.1)	4.4 (2.3)	5.0 (2.1)
Pack Years	11.1 (10.5)	9.4 (8.5)	11.3 (9.4)
Cigarettes/day	19.1 (6.8)	18.8 (5.4)	21.5 (6.3)
5-HTTLPR	S carrier 49 (67.1%)	S carrier 49 (72.1%)	S carrier 20 (60.6%)
DRD2	A1 carrier 43 (58.9%)	A1 carrier 43 (63.2%)	A1 carrier 19 (57.6%)
NEO Neuroticism	103.9 (2.3)	102.8 (2.2)	97.8 (3.3)
NEO Anger	16.1 (0.5)	16.3 (0.5)	16.0 (0.7)
NEO Anxiety	19.5 (0.6)	18.7 (0.5)	17.3 (0.9)
NEO Depression	12.7 (0.5)	12.0 (0.5)	11.7 (0.8)
Education (yrs)	14.9 (1.7)	14.5 (1.6)	14.5 (1.7)
Students/staff	52/2	44/2	20/1
Community	19	22	12

Note. In parentheses are SEMs for NEOPI trait measures and SDs for other continuous variables.

Biochemical Confirmation of Smoking Status

Smoking was monitored using plasma and salivary nicotine and cotinine concentrations and breath CO concentrations. CO was assessed during the five prequit monitoring sessions and every 72 hours during the post-quit period. CO was measured with a MiniCO™ meter (Catalyst Research Corporation, Owings Mills, MD). The nicotine and cotinine concentrations reported below were determined from blood samples collected during the pre-quit baseline experimental session and then at days 1, 3, 10, 17, and 45 post quit. Cotinine and nicotine concentrations were assayed using gas chromatography as described by Jacob, Wilson, and Benowitz (1981).

Genotyping

DNA was isolated from blood samples according to standard techniques. Each sample was genotyped twice, using blinded duplicate codes. In the very small number of cases where discrepancies were found, additional blinded assays were performed to assure the reliability of the genotyping. The Taq1A restriction-length polymorphism of DRD2 gene was assessed as described by Spitz et al. (1998) and Gilbert et al. (2004). Categorization of the DRD2 genotypes for statistical purposes was as either homozygous A2A2 alleles or A1 allele carriers (A1A2 and A1A1). The 5-HTTLPR polymorphism was assessed by PCR based on the conditions described by Yonan, Palmerb, & Gilliam (2006). Specific primers and reaction conditions are available upon request. For statistical purposes, 5-HTTLPR genotype is categorized as either S allele (one or two short [S] variants with 14 copies of a 20–23 base pair [bp] repeat unit, 430 bp) or homozygous LL type (two long variants consisting of 16 copies, 474 bp).

5-HTTLPR polymorphism frequencies were 56 [32.3%] LL, 86 [49.4%] SL, 32 [18.4%] SS, χ²(1, N = 174) = 0.010, p >.5. The allele frequencies were in almost perfect Hardy-Weinberg equilibrium for the HTTLPR gene, in that frequencies were exactly what were expected with random mating as assessed with no difference between expected and observed allele distributions. However, the allele frequencies (69 [39.7%] A2A2, 102 [58.6%] A1A2, 3 [1.7%] A1A1) for DRD2 were not in equilibrium, χ²(1, N = 174) = 23.75, p < .05.

Questionnaires

Fagerström Test of Nicotine Dependence (FTND; Heatherton et al., 1991)

The FTND assesses nicotine dependence and is predictive of withdrawal symptoms and relapse to smoking (Piasecki et al., 2000).

The Shiffman Withdrawal Questionnaire (SWQ; Shiffman, 1979)

The widely validated SWQ has a number of scales, including irritability. Reports on this and other NA state measures were based on participant ratings of their typical/average state across the preceding 72 hours.

The Profile of Mood States (POMS; McNair, Lorr, & Droppleman, 1971)

This well-validated questionnaire assesses six basic mood states.

Beck Depression Inventory-II (BDI-II; Beck, Steer, & Brown, 1996)

The BDI is an extensively validated instrument that assesses dimensions of clinical and subclinical depression.

Beck Anxiety Inventory (BAI; Beck, Epstein, Brown, & Steer, 1988)

The BAI is a validated instrument that assesses anxiety symptoms. Items include a preponderance of somatic symptoms.

State-Trait Personality Inventory (STPI, Spielberger, 1995)

This measure assesses four states and corresponding traits: Anxiety, Anger, Curiosity, and Depression.

NEO-Personality Inventory-Revised (NEO PI-R; Costa & McCrae, 1992)

Trait depression, anger, and anxiety were assessed using the NEO PI-R.

Procedure

The study flowchart and time line are presented in Figures 1 and 2. Phase 1 was a two-week baseline during which participants smoked at their usual rate and attended biweekly monitoring sessions. Phase 1 ended with random assignment to treatment groups. Phase 2 included late afternoon monitoring sessions on the first day of abstinence (or corresponding day in the smoke group), at day 3 of abstinence, and then every 72 hours.

Figure 2.

Time line. The present analyses exclude data from the final day of the abstinence period (Day 45) because this day was devoted to a lengthy experimental session.

An IRB-approved consent was signed by each participant. Completion of the study resulted in earning $500 minus any penalties for smoking. The first cigarette penalty was $10, the second cigarette penalty was $25, and the third was $50, for a maximum total of $85 for three or four cigarettes. Participants were excluded from the study without payment for smoking more than four cigarettes total over the abstinence period. Abstinence failure was defined as smoking a total of more than four cigarettes after quitting, using the estimation procedures noted above in the Participant section.

Participants received an abbreviated form of the American Lung Association smoking cessation program. Nicoderm® or placebo patches of corresponding size were 21 mg for the first 17 days of abstinence, 14 mg for days 18 to 26, and 7 mg for days 27 to 38. The durations on the patches were based on our desire to achieve an abstinence success rate in excess of 80 percent by limiting the duration of our study and because GlaxoSmithKline recommends being on their 21 mg patch for a longer period of time than the 14 and 7 mg patches. The final assessment of mood one week after going off the patch was chosen to characterize rebound effects. Individuals were instructed to replace patches immediately upon wakening each morning.

Data Analysis

To enhance the reliability, composite indices were based on the aggregate measures of each of the three NAs. The composite state anger-irritability index was formed by averaging STP anger, POMS anger, and Shiffman irritability Z scores. Depression was the mean of BDI depression, STP depression, and POMS depression Z scores, and anxiety was a composite of BAI, POMS tension-anxiety, and STP anxiety Z scores. Z scores were generated for each of the 15 post-quit monitoring sessions by subtracting each individual’s score on a given measure from the group mean of the third and fourth pre-quit baseline scores for that measure (e.g., POMS anger) divided by the mean SD for the measure (e.g., POMS anger) across all three groups.

To characterize how treatments and individual genetic and trait differences predict trajectories of affective abstinence symptoms across 44 days, hierarchical linear modeling (HLM; Raudenbush & Bryk, 2002) was used to perform growth curve analyses for each of the three mood change indices (Anger, Depression, and Anxiety). Specifically, a two-level model was fitted, where a level-1 submodel describes how each individual’s mood changed across 15 monitoring sessions over time, and a level-2 submodel relates interindividual differences in symptom trajectories to predictors including treatments and individual genotypes, traits, interactions of interest, and other control personal variables. Because preliminary analyses indicated that including a linear slope and quadratic trend term was appropriate, level-1 submodels included an intercept, a linear slope and a quadratic term of Time, and a random error. To avoid collinearity of the linear and quadratic terms, we recentered Time to the midpoint of the 44-day period. Consequently, the level-1 intercept represents individual’s mood status at the midpoint; i.e., day 22.5, rather than the initial status at day 1 of smoking cessation. All multilevel models were fit using full maximum likelihood method and parameters estimates based on robust standard errors. Initially the same set of predictors was included for each of the three level-2 models. To obtain a more parsimonious representation from each multilevel model, level-2 fixed effect parameters that were not significant according to the single parameter hypothesis test (z-statistic but labeled as t-ratio in HLM) or did not account for significant variance according to the model deviance statistic were removed from the initial model sequentially.

First, changes in NA symptoms of the two abstinence groups relative to the smoke group were assessed to test whether experimental manipulations led to significant differences in affective symptoms. Control variables included gender, age, FTND, and baseline mood scores. Next, the effects of 5-HTTLPR and DRD2 genotypes and treatment × genotype interactions were tested in separate HLM models for each category of affective symptoms, controlling for treatment main effect, gender, age, FTND, and corresponding baseline symptom scores. Because of statistical power limitations associated with the small sample size of the smoke group, analysis was limited at this and next step to data from the two abstinence groups. Finally, personality trait and genotype × trait interactions were entered into the reduced level-2 models for each affective index that was further reduced using the stepwise approach. Thus, the final models provide a comprehensive representation of the effects of one genotype controlling for the other and of the additional effects of traits and genotype × trait interactions controlling for genotype. Because analyses performed with the entire sample did not differ from the results from the European-only subsample, only the results based on the entire sample are presented.

Results

Verification of Abstinence and Nicotine Replacement

The pre-quit baseline indices of smoking (CO, plasma nicotine, and cotinine concentrations) were not different across treatment groups or gender (Fig 3). After quitting, the two patch groups had CO concentrations equivalent to nonsmokers (2 ppm). The 21 mg patch increased plasma nicotine and cotinine concentrations slightly above pre-quit levels in both men and women. Nicotine values in the placebo patch group approximated zero after quitting.

Figure 3.

Mean and (SEMs) of plasma nicotine and cotinine and expired carbon monoxide at the pre-quit baseline and at 3, 17, and 45 Days after quitting smoking (in the quit groups). Nic-nicotine patch, Pla-placebo patch.

Symptom Trajectories of Abstinence Groups in Comparison with Smoke Group

Abstinence and NRT resulted in the expected patterns of change in affective symptoms across time. Due to space limitations, detailed statistics and figures for results from these non-primary analyses involving comparison with the smoker group are not presented here, but are available from the authors. Briefly, there was a significant increase in midpoint affective symptoms in the PP group relative to the smoke group for each of the symptom categories (all ps < .05). Negative slopes of the affective symptoms in the PP group in contrast to the smoke group (all ps < .01) indicated that these symptoms declined over time. For anxiety only, there was a greater curvature coefficient in the PP relative to smoke group (p < .01), reflecting the fact that participants receiving placebo reported more anxiety shortly after cessation.

NP and smoke groups did not differ significantly in mid-point values. NP and smoke groups also did not differ in their symptom trajectory slopes for anger and depression categories, but there were a steeper decline and greater curvature in anxiety change over time for NP in contrast to the smoke group. There were also greater curvature coefficients in the NP group relative to smoke group for angry and depressive symptoms (both ps < .01) that reflected greater symptom levels at early cessation and during the final two weeks on the abstinence (when on the 7 mg patch and when off the patch). NP and PP groups differed significantly in midpoint levels and slopes of symptom trajectory, with the latter showing greater midpoint affective symptoms in all three categories (all ps < .05) and steeper decline trends from the elevated early angry and depressive symptoms (both ps < .01). Thus, smoking abstinence elevated NA symptoms, but the NP group reported significant fewer symptoms than the PP group, and NP and smoke groups differed mainly in the tendency for somewhat greater NA at early cessation and during the final two weeks of abstinence when on the 7 mg patch and when off the patch.

Effects of Genotypes and Personality Traits on Withdrawal Symptoms

Table 2 summarizes the primary findings of the study, the moderation of TAS trajectories of the three NAs (anger, depression, and anxiety) by genotypes and treatment × genotypes, controlling for main effects of treatment, gender, age, FTND, and prequit baseline symptom scores. NATs, and genotype × NAT were subsequently entered into the level-2 models. NAT moderators (anger trait, depression trait, and anxiety trait) were applied only to the corresponding affective states. That is, trait anger was used to predict anger state, trait depression predicted depressive state, and trait anxiety predicted anxiety state.

Table 2.

Final models with genotypes and NAT as predictors of NA

Variables/terms		Coefficients of changes in NA State
		Anger	Depression	Anxiety
Midpoint status	Treatment	0.3018** (0.1000)
	Gender	−0.2129* (0.0871)		−0.2337** (0.0707)
	5-HTTLPR	0.3601*** (0.0924)	0.1410* (0.0622)
	5-HTTLPR × Treatment			0.2582* (0.0994)
	NATa	0.3168*** (0.0648)
	DRD2			0.1290* (0.0644)
	DRD2×NAT		0.1895** (0.0606)
	Baseline	−0.5370*** (0.0739)	−0.2801** (0.0905)	−0.2136** (0.0593)
Linear slope	Intercept			−0.0112*** (0.0027)
	Age	0.0049* (0.0023)	0.0032* (0.0014)
	Gender	−0.0140** (0.0039)	−0.0081** (0.0030)
	FTND	−0.0061* (0.0027)
	DRD2	0.0099* (0.0049)	0.0071* (0.0030)	0.0095* (0.0039)
	DRD2 × Treatment	−0.0178** (0.0053)	−0.0086* (0.0034)	−0.0083* (0.0040)
	Baseline		−0.0051** (0.0016)	−0.0048* (0.0020)
Quadratic trend	Intercept		−0.0004* (0.0002)
	Gender		0.0005* (0.0002)
	5-HTTLPR	0.0007*** (0.0002)	0.0006* (0.0002)	0.0008*** (0.0002)

Note. In parentheses are SEMs.

^*p < .05;

^**p < .01;

^***p < .001.

NAT corresponds to the NA state listed in the column headings. For dichotomous predictors, negative coefficients correspond to smaller values in NA changes in placebo treatment, females, 5-HTTLPR homozygous S alleles, or DRD2 A1 carriers.

Anger

Anger symptoms in the NP group were significantly less than in the PP group, controlling for the genotypes and other personal variables (Fig 4A). 5-HTTLPR S carriers reported more midpoint anger symptoms than those with LL alleles (Fig 5A). As indicated by a greater quadratic trend, S carriers reported more severe anger symptoms immediately after quitting and their symptoms increased during the last two weeks of the study when on the 7 mg patch and when they were off the patch. In contrast, anger continued to decline over time in subjects with homozygous L alleles (Fig 5A). However, there were no significant interactions of 5-HTTLPR genotype with treatment for anger trajectories. In contrast, DRD2 genotype interacted with treatment in influencing the slope of the anger symptom changes (Table 2). As shown in Figure 5B, A2A2 individuals on NP exhibited a steady decline in anger such that anger elevation was almost completely resolved within Day 44 of quitting, while the other three groups remained substantially elevated at this time. A1 carriers in the same treatment condition (NP) experienced few anger symptoms while on the two higher patch doses, but exhibited a progressive increase in anger when placed on the 7 mg dose and during the final week, when off the patch. In contrast, A2A2 individuals in the PP condition experienced progressively decreased anger across time, but remained over .6 Z scores above baseline 44 days after quitting. The overall pattern was for A1 carriers to experience greater reductions in anger from NP, relative to PP, during early abstinence while on the 21 mg patch, but for A2A2 individuals to show greater benefit from NP during the final two weeks of abstinence when their anger approximated their pre-quit levels.

Figure 4.

Effects of gender and patch type on symptom trajectories for anger (A), depressive affect (B), anxiety (C) across the 44 days of abstinence.

Figure 5.

Effects of 5-HTTLPR genotype (LL vs. S carrier) and patch type (A), effects of DRD2 genotype (A2A2 vs. A1) and patch type (B), and effects of trait NEO-PI trait anger and patch type on anger symptom trajectories (C) across 44 days of smoking abstinence. In C, subjects were classified into three groups of approximately each size according to the percentiles of their anger scores.

When trait anger and its interactions with 5HTTLP and DRD2 genotypes were subsequently added to the model, trait anger significantly contributed to the prediction of midpoint status of anger trajectories. For both NP and PP groups, higher trait anger scores were associated with more anger symptoms (Fig 5C). No other trait anger-related effects or interactions were found.

Depression

In the final growth model of depressive symptoms, 5-HTTLPR genotype had significant effects on both the midpoint status and the curvature of symptom change (Table 2; Fig 6A). Relative to LL individuals, S carriers experienced not only more midpoint depressive symptoms but also more early depressive symptoms and their symptoms elevated again when on low-dose nicotine patch and no patch. A significant interaction of DRD2 genotype with Treatment on slope (Fig 6B) reflected the fact that very few depressive symptoms were reported in A1 allele carriers in the NP condition during the first four weeks of treatment when on the two higher nicotine doses, and that depressive symptoms emerged in A1 carriers and exceeded those of the other three groups at the end of treatment when off the patch. In contrast, in A2A2 individuals NP and PP groups did not differ shortly after quitting, but beginning two weeks of abstinence the NP group members reported fewer depressive symptoms than the PP group; and this difference was maintained through the end of the study.

Figure 6.

Effects of 5-HTTLPR genotype (LL vs. S carrier) and patch type on depressive symptom trajectories (A), effects of effects of DRD2 genotype (A2A2 vs. A1) and patch type (B), and effects of trait NEO-PI trait depression, DRD2 genotype (A2A2 vs. A1 carrier) and patch type on depressive symptom trajectories (C) across 44 days of smoking abstinence. In C, subjects were classified into two groups according to the median-split of their trait depression scores.

Trait depression interacted with DRD2 genotype in predicting the midpoint status of the depressive symptom trajectories (Fig 6C). While higher trait depression was related to more symptoms at midpoint among A1 carriers, the opposite pattern was observed among those with A2A2 genotype. In combination with the effects of DRD2 genotype on the slope of depressive symptom changes, A2A2 carriers with higher depression trait scores reported fewer symptoms relative to others during the later treatment period.

Anxiety

5-HTTLPR genotype interacted with treatment to predict the midpoint status of anxiety symptom trajectories. As illustrated in Figure 7A, relative to PP, NP alleviated anxiety only in S carriers; in LL individuals anxiety symptom levels were very low in the PP condition and NP did not further reduce these symptoms. Additionally, in comparison with individuals with LL genotype, S carriers experienced substantially more anxiety symptoms during early and during late abstinence as indicated by the greater curvature associated with S genotype. Figure 7B shows the main effects of DRD2 genotype on the midpoint status and slope of the anxiety symptom changes, as well as an interaction between DRD2 genotype and treatment in predicting the slope. Regardless of patch type, A1 carriers reported higher levels of anxiety symptoms at midpoint and a lesser decline in their symptoms across time. The difference in slope of anxiety changes among DRD2 genotypes was particularly strong in the NP group, with A1 carriers having a weak or no overall decline trend in these symptoms relative to A2A2 individuals. As compared to A1 carriers, A2A2 individuals in the NP condition had similar anxiety symptoms levels during the first few weeks of abstinence, but fewer anxiety symptoms during the late phases with low-dose patches and without the patch. As was the case with anger and depressive symptoms, A1 carriers benefitted more from NRT than did A2A2 individuals during the first two weeks of abstinence when on the high-dose patch, but exhibited no benefit of NRT and a tendency for the renewal of symptoms during the final week of abstinence, when off the patch.

Figure 7.

Effects of 5-HTTLPR genotype (LL vs. S carrier) and patch type (A) and effects of DRD2 genotype (A2A2 vs. A1 carrier) and patch type (B) on anxiety symptoms across 44 days of smoking abstinence.

Gender

Among control variables, gender remarkably influenced NA symptom trajectories (Table 2). As illustrated in Figure 4A, men reported more severe midpoint anger symptoms than women in both NP and PP groups and their symptoms declined less than females. Women reported more acute depressive symptoms just after quitting than men (Fig 4B). Men reported greater increases in anxiety than women throughout the monitoring period after quitting (Fig 4C).

FTND

Of the three NA symptom indices, only anger was affected by nicotine dependence as measured by the FTND. Across patch types, individuals with higher FTND scores experienced more anger immediately after quitting followed by a steeper decline of anger over time.

Discussion

The present findings are important because they are the first, in a study with minimal dropout, to demonstrate: 1) strong associations of 5-HTTLPR genotype with smoking abstinence-symptom trajectories; and 2) interactions of 5-HTTLPR and of DRD2 genotypes with NRT effects on TAS. In addition, results replicated previous findings showing that higher scores on trait anger predict greater abstinence-associated increases in anger-irritability, and further showed trait depression interacts with DRD2 genotype in predicting depressive affect-related symptoms. Overall, support was found for our a priori hypotheses. On the other hand, the hypothesis that 5-HTTLPR S allele carriers and DRD2 A1 carriers would exhibit greater benefits from nicotine patch (NP) than from placebo patch (PP) was only partially supported. While confidence in these findings is enhanced by the rigorous experimental design and minimal dropout, questions arise as to what mechanisms mediated the observed effects.

Genotype Predicts NA-Related TAS Trajectories

The present findings are consistent with previous data (reviewed in the introduction) suggesting a substantial genetic basis to tobacco abstinence symptoms (e.g. Pergadia et al., 2006). Because of this assumed causal complexity of abstinence response, the strength of the association of the 5-HTTLPR S allele with larger increases in abstinence-related NA is surprising and is in need of replication. However, this association was predicted based on the extensive literature concerning the S allele’s association with stress-related NA.

In the current sample, individuals with the S allele of the serotonergic transporter gene had larger increases in anger-irritability, depression, and anxiety symptoms (only among individuals receiving placebo for the latter) than those without this allele. This finding is in accordance with previous observations showing S carriers, who presumably have low expression of serotonergic transporters relative to LL individuals (Lesch, et al., 1996), to exhibit heightened stress reactivity and emotional dysregulation (Hariri & Holmes, 2006). Both negative affect and nicotine abstinence are associated with lowered brain serotonergic functioning (Harrison, Liem, & Markou, 2001; Lesch & Merschdorf, 2000). Furthermore, NRT attenuated anxiety only in S carriers; while in LL individuals anxiety symptom levels were very low in the placebo condition and nicotine did not further reduce these symptoms. Thus, S carriers may be more vulnerable to the effects of nicotine abstinence and may benefit more from additional individualized interventions, in contrast with LL individuals. Supporting this possibility, a cross-sectional study by Hu et al. (2000) showed high neuroticism S carriers to be characterized by low success in quitting smoking; and Lerman et al. (2000) found neuroticism to be positively associated with smoking motivation to reduce NA among smokers in 5-HTTLPR S carriers, but not among homozygous L smokers. The present results provide evidence for the unique benefits of NRT to attenuate anxiety symptoms among smokers who are S carriers.

However, if the larger increases in NA in S carriers were due simply to nicotine withdrawal leading to the unmasking of S-allele-related negative affectivity, one would expect treatment by genotype interactions for anger and depressive affect, as well as for anxiety. It is possible that the lack of nicotine treatment by genotype interactions for anger and depressive affect reflects greater sensitivity of S carriers to the loss of (withdrawal from) the monoamine oxidase inhibitory (MAOI) properties of tobacco smoke (Fowler et al., 1996) rather than an enhanced degree of nicotine withdrawal symptoms. That is, S carriers may experience greater benefits from smoking because of NA-reducing chemicals in smoke other than nicotine. It would be interesting to see if S carriers benefit more from MAO medications for smoking cessation than do others.

DRD2 Genotype ×Treatment Interactions

A1 carriers benefited more from NRT than did A2A2 individuals during the first two weeks of abstinence when on the high-dose patch, but exhibited weak or no benefit of NRT and a visually apparent tendency for the renewal of symptoms during the final week of abstinence, when off the patch. In contrast, A2A2 individuals benefited somewhat less from NRT early in treatment but these benefits were maintained and extended when they were given low-dose patches and when off the patch. The finding that A2A2 genotype was associated with better retained therapeutic benefits of NRT extends previous evidence for the unique effects of bupropion in attenuating NA-related TAS in A2A2 carriers (David et al. 2003; Cinciripini et al. 2004), suggesting that this genotype may be generally a predisposition for responsiveness to the NA-reduction effects of common treatments among smokers. The present results also seem to indicate that A1 carriers can benefit from the NA-reduction effects of high-dose NRT.

NATs Predict NA-Related TAS Trajectories

The observed ability of trait anger and depression to predict abstinence-induced increases in state anger-irritability and depression (respectively) above and beyond the assessed genotypes is consistent with the view that TAS trajectories are determined by multiple genetic, situational, and historical factors (Shiffman et al., 2004). The strength of the association between trait anger and state anger-irritability was greater than in previous studies (e.g., Gilbert et al., 1998), possibly due to enhanced reliability and sensitivity resulting from the multi-instrument assessment of state anger-irritability used in the present investigation. The association between trait depression and abstinence-related increases in depressive affect was more modest and in line with previous findings (Gilbert et al., 1998; Gilbert et al., 2002).

Support for the ability of NATs to predict TAS has grown in recent years (Covey et al., 1990; Gilbert et al., 1998, 2002; Jamner et al., 1999; McClernon et al., 2006; Zvolensky et al., 2004). The abilities of specific affective traits to predict corresponding state affects in response to the stress of smoking abstinence likely reflect in large part genetically based trait dispositions to specific NAs (Gilbert & Gilbert, 1995; Heath, Neale, Kessler, Eaves, & Kendler, 1992) that were only in part influenced by the genotypes assessed in the present study. Given that questionnaire-assessed NATs and their effects on TAS reflect a large range of different genes and gene by environment interactions (Glazier, Nadeau, & Aitman, 2002; Lerman et al., 2000; Pergadia et al., 2006), it will be important to continue assessing both genotypes and NATs in future studies.

The fact that NA, especially anger-irritability in males high in trait anger, remained substantially elevated even after 44 days of smoking abstinence is consistent with self-medication models of smoking in this population (Carmody, 1989; Eysenck, 1980), as well as the idea that chronic nicotine exposure is associated with enduring cognitive deficits (Snyder, Davis, & Henningfield, 1989) and deficits in brain networks mediating cognitive functions (Gallinat et al., 2006; Neuhaus et al., 2006). Long-lasting cognitive impairments caused by chronic smoking and exacerbated by withdrawal could make work and daily life tasks more challenging and more likely anger-provoking, especially for anger-prone individuals.

In the placebo condition, anger-irritability exhibited larger Z-score increases than other symptoms and at 44 days of abstinence was still substantially elevated. This pattern of findings supports the view that either anger-irritability takes longer to resolve than 44 days and/or that nicotine or smoking has an inherent ability to reduce anger. Support for the latter interpretation comes from work by Jamner et al. (1999) who found evidence suggesting that nicotine patch can reduce anger in non-smokers, especially if they are high in trait anger.

Anxiety and depressive affect symptoms exhibited smaller elevations than did anger, but in S allele carriers they still failed to come close to baseline levels by Day 44. Earlier studies (Gilbert et al., 1998; 2002) using largely similar methods found that anxiety and depression were substantially above control group and baseline levels after 31 days of abstinence. The question of whether NRT has the ability to suppress NA independent of withdrawal symptom alleviation was addressed by Jamner et al. (1999), who found that NRT reduced anger in non-smokers, and by McClernon et al. (2006), who found that NRT reduced symptoms of depression in non-smokers. Thus, our findings showing a failure of NA to fully resolve are consistent with the view that nicotine has an inherent ability to reduce NA. However, the question remains as to why anger-irritability exhibited larger increases after quitting smoking than did anxiety and depression.

Renewal of Abstinence Symptoms when Off the Patch

The renewal of withdrawal symptoms (NA) after being reduced to 7 mg nicotine patch and subsequently during the no-patch week is interesting, as is the observation that even after going off the patch those previously on the nicotine patch reported fewer symptoms than those previously on placebo. The mechanisms underlying these post-patch benefits of NRT are not clear. They could be largely neurobiological or psychosocial. The renewal could reflect the unmasking of underlying temperament/psychopathology masked by higher doses of nicotine. The finding that renewal was maximal in S carriers supports the unmasking hypothesis. However, the interpretation of these curvilinear effects should be made with caution because our complex modeling procedure did not include follow-up comparison analyses to demonstrate statistically significant increases in TAS during weaning from nicotine. The analyses only demonstrate a significant quadratic curve that is consistent with this interpretation.

Effects of Gender

The finding that men exhibited larger increases in anger and anxiety than women is interesting given that a recent meta-analysis of long-term smoking abstinence (Perkins & Scott, 2008) concluded that men benefit more than women from NRT. However, the current study did not find significant gender differences in the effects of NRT on NA, though menstrual cycle effects were not assessed because of the limited sample size.

Clinical Implications

Each of the major findings of the present study could have important clinical implications. First, given the relationship of NA symptoms to relapse (Shiffman & Waters, 2004), the ability of stress-related genes and NATs to predict TAS trajectories suggests the utility of using knowledge of both genes and NATs when devising individualized interventions. For example, one should take into account the failure of symptoms to recover fully within 44 days, especially in 5-HTTLPR S carriers and those high in trait anger-irritability. Possible renewal of NA-related abstinence symptoms when on low-dose patches and after going off the patch, especially in DRD2 A1 carriers, is also something that needs much more attention by both clinicians and researchers. The findings suggesting that genetic factors may influence NRT weaning-related re-emergence of symptoms needs to be addressed in clinical trials and replication studies. If the curvilinear effects seen in this research are indeed due to patch weaning, then one might expect these genotypes to predict relapse at specific junctures in a medication regimen. That is, there might not be overall gene-outcomes associations at long-term follow-ups, but genetic effects might be evident when analyzed as a function of scheduled clinical events.

Study limitations and Future Directions

Participant selection factors limit the generalizability of study findings. The study excluded individuals with psychiatric disorders and psychoactive drug use. The time-demanding nature of the study also resulted in self-selection bias. The sample was more educated than smokers in the general population. While the present study sample size was large for controlled studies of abstinence symptoms, it was small and weakly powered for genetic studies. Also, while the DRD2 alleles were out of Hardy-Weinberg equilibrium, all genotyping was conducted in duplicate; thus genotyping errors appear unlikely. Finally, time was confounded with patch dose because the study used a nicotine dose step-down procedure. Despite these limitations, the rigorous study methods provide more confidence in the results than is possible for other studies of comparable length.

The present findings replicate and extend findings indicating that NA experienced during smoking abstinence symptoms are substantially heritable (Pergadia et al., 2006) and are also moderated by NATs. 5-HTTLPR S carriers experienced larger increases in NA symptoms than those without a short allele and DRD2 A2A2 genotype was linked to better maintenance of therapeutic effects of NRT on NA-related abstinence symptoms. Study findings concerning anger and depression support the view that NA abstinence symptoms are greater in those scoring high in NATs (Gilbert et al., 1998, 2002, Pergadia et al., 2006), and that specific NATs predict greater increases in the corresponding NA abstinence symptoms. Clearly, additional studies with larger samples are needed to replicate, clarify and better characterize the effects of genetic factors, NATs, and NRT on tobacco abstinence symptom trajectories.