- Top of page
- Supporting Information
Nicotine and tonic dopamine (DA) levels [as inferred by catechol-O-methyl tranferase (COMT) Val158Met genotype] interact to affect prefrontal processing. Prefrontal cortical areas are involved in response to performance feedback, which is impaired in smokers. We investigated whether there is a nicotine × COMT genotype interaction in brain circuitry during performance feedback of a reward task. We scanned 23 healthy smokers (10 Val/Val homozygotes, 13 Met allele carriers) during two fMRI sessions while subjects were wearing a nicotine or placebo patch. A significant nicotine × COMT genotype interaction for BOLD signal during performance feedback in cortico-striatal areas was seen. Activation in these areas during the nicotine patch condition was greater in Val/Val homozygotes and reduced in Met allele carriers. During negative performance feedback, the change in activation in error detection areas such as anterior cingulate cortex (ACC)/superior frontal gyrus on nicotine compared to placebo was greater in Val/Val homozygotes compared to Met allele carriers. With transdermal nicotine administration, Val/Val homozygotes showed greater activation with performance feedback in the dorsal striatum, area associated with habitual responding. In response to negative feedback, Val/Val homozygotes had greater activation in error detection areas, including the ACC, suggesting increased sensitivity to loss with nicotine exposure. Although these results are preliminary due to small sample size, they suggest a possible neurobiological mechanism underlying the clinical observation that Val/Val homozygotes, presumably with elevated COMT activity compared to Met allele carriers and therefore reduced prefrontal DA levels, have poorer outcomes with nicotine replacement therapy.
Despite the recent development of several behavioral and pharmacological treatments for nicotine dependence (Cahill et al. 2011; Hughes et al. 2003; Silagy et al. 2004), long term abstinence rates remain low and most smokers relapse within the first year or sooner of treatment (Burke et al. 2008). However, like all addictions, smoking impacts domains that are important in maintaining abstinence differently across individuals (Gehricke et al. 2007). The use of imaging genetics to probe the relationship between the neurocircuitry of addiction-related intermediate phenotypes such as reward and error processing and the functional genetic polymorphisms affecting these neurocircuits has the potential to improve outcomes by matching individual patient characteristics with available smoking cessation treatments (Sweitzer et al. 2012). Response to reward (Martin-Soelch et al. 2003; Powell et al. 2002) as well as errors (Franken et al. 2010; Luijten et al. 2011) has been shown to be altered in smokers compared to controls. Alterations in prefrontal dopaminergic tone as inferred from the Val158Met polymorphism of the gene encoding the catechol-O-methyl tranferase (COMT) enzyme (Kaenmaki et al. 2010; Tunbridge et al. 2006) has been shown to impact cognitive (Bruder et al. 2005; Egan et al. 2001) and reward processing (Camara et al. 2010; Dreher et al. 2009; Marco-Pallares et al. 2009; Yacubian et al. 2007). As such, it is possible that this genetic variation also influences the neurocircuitry of reward processing that is altered as a consequence of chronic drug exposure (Robinson & Berridge 1993).
The COMT Val allele is associated with elevated COMT enzyme activity, resulting in more efficient catabolism of dopamine (DA) and a putative decrease, therefore, in extracellular DA, compared to the Met allele (Chen et al. 2004). The effect of this COMT polymorphism on DA levels is especially relevant in the prefrontal cortex (PFC), where COMT is therefore the primary mechanism of DA removal after it is released into the synaptic space (Garris & Wightman 1994).
With respect to reward processing, Val/Val homozygotes have been shown to exhibit more impulsive choices during a delayed discounting task (Boettiger et al. 2007) and less risk aversion (Farrell et al. 2012). This functional variant polymorphism has previously been shown to be associated with smoking behaviors, with the Val allele associated with increased susceptibility to nicotine dependence and greater risk for smoking relapse (Colilla et al. 2005; David et al. 2011; Enoch et al. 2006b; Johnstone et al. 2007; Munafo et al. 2008).
However, the relationship between PFC DA levels and behavioral performance is not linear. Previous studies have showed an inverted-U-shaped relationship in which mid-range DA levels are associated with optimal PFC function, whereas high and low DA level extremes impair PFC function (Goldman-Rakic et al. 2000). Such DA-related inverted-U curves have been extended to the COMT Val158Met polymorphism, with the Val and Met alleles differentially responding to pharmacologically induced increases in extracellular DA. Specifically, in the Val allele (with low to suboptimal baseline DA), artificially enhancing DA activity increases PFC function to more ‘optimal’ levels, while in the Met allele carriers (with high to optimal baseline DA), boosting synaptic DA does not change and in some cases, impairs PFC function (Tunbridge et al. 2006). This pattern is consistent across various PFC-dependent tasks in the presence of agents that increase DA levels, including amphetamine (Mattay et al. 2003) and the central COMT inhibitor, tolcapone (Giakoumaki et al. 2008). In a recent study, Farrell et al. (2012) show that administering tolcapone not only increases working memory (WM) performance but also decreases risk taking behavior on a gambling task in Val/Val homozygotes with the opposite effect noted in Met/Met homozygotes.
Given that nicotine and tonic DA levels (as inferred by COMT genotype) (Bilder et al. 2004) interact to affect PFC processing (Loughead et al. 2009) and that nicotine addiction also moderates performance feedback which is known to be related to DAergic signaling (Holroyd & Coles 2002), one might expect similar nicotine × COMT genotype interactions on performance feedback in smokers.
We therefore used the Monetary Incentive Delay (MID) task (Knutson et al. 2000), to probe brain circuitry involved in trial by trial performance feedback in a cohort of current smokers distinguished by their COMT Val158Met genotype. We hypothesized that acute nicotine administration in smokers would differentially impact cortico-striatal circuitry and behavioral responses as a function of COMT genotype, such that Val/Val homozygotes will exhibit enhanced neurobehavioral responses, while Met allele carriers will exhibit no change or even decreased responding during performance feedback.
- Top of page
- Supporting Information
A total of 23 right-handed, daily smokers (≥10 cigarettes per day for at least one year; range 10–20 cigarettes) gave written informed consent to a National Institute on Drug Abuse Intramural Research Program Institutional Review Board approved study. Participants were recruited from the Baltimore area using print advertisements and referrals. Subjects had no Axis I diagnoses (other than nicotine dependence) as measured by the computerized Structured Clinical Interview for DSM-IV with follow-up clinical interview. Additional exclusionary criteria, as assessed by clinical history and exam, included pregnancy, claustrophobia, significant medical disorders such as cardiovascular disease or neurological diagnoses such as a history of head trauma, seizure disorder or migraine, together with current dependence on any drug other than nicotine. Participants were instructed to smoke as usual before their study session and not consume any alcohol for 24 h and no more than half a cup of caffeinated beverage 12 h before each study visit. Participants were excluded if they tested positive for current drug use (cocaine, amphetamine, THC, methadone, morphine, oxycodone, PCP, benzodiazepines, buprenorphine and MDMA) based on urine screening or alcohol use, based on breathalyzer testing.
In this within-subjects design, subjects underwent two scanning sessions – one with a placebo and one with a 21-mg nicotine patch. Subjects smoked as usual up until their experimental session. Patches were applied 30 min after the participant's last cigarette and approximately 2 h prior to the MRI sessions in a single-blind fashion. The single-blind design was to monitor subjects for adverse events during scanning. The session order was counterbalanced with 10 subjects receiving nicotine patches in sessions 1 and 13 receiving nicotine patches in session 2 (Table 1).
Table 1. Genotype demographic data
| ||Val/Val homozygotes||Met allele carriers||P-Value|
|Age (mean ± SD)||32.2 ± 10.8||31.6 ± 8.60||0.89|
|BAI (mean rank)||10.3||13.4||0.24|
|FTND (mean ± SD)||5.10 ± 1.66||5.08 ± 1.55||0.97|
|Order of sessionsa||4/6||6/7||0.77|
|Self-reported race||4 AA, 1AS, 5CC||3AA, 0AS, 10CC||0.22c|
|European ethnicityb||0.41 ± 0.45||0.74 ± 0.39||0.18d|
|African ethnicityb||0.31 ± 0.40||0.18 ± 0.34||0.15d|
|East Asian ethnicityb||0.10 ± 0.30||0||0.61d|
|WASI IQ (mean ± SD)||102 ± 6.90||112 ± 11.8||0.02e|
Participants completed the Beck Anxiety Inventory (BAI) (Beck et al. 1988; Knutson et al. 2000), the Fagerström Test for Nicotine Dependence (FTND) (Heatherton et al. 1991), a nicotine craving assessment modified from the short form of the Tobacco Craving Questionnaire (TCQ) (Heishman et al. 2003,2008) and the Parrott Mood Assessment (PARROTT et al. 1996). The TCQ and Parrott Mood Assessment were administered both before and after each of the two scan sessions to investigate possible changes in affect and craving as a function of nicotine or placebo patch application. Note that the nicotine craving assessment we used differs from the TCQ in that we used a continuous VAS (0–800) while the TCQ used a Likert type scale, ranging from 1 (strongly disagree) to 7 (strongly agree). Thus, the maximum value for any nicotine craving factor is 2400. Each measure in the Parrott was also rated on a VAS ranging from 0 to 800. VAS values were not visible to participants during ratings.
The MID task
We used a modified version of the MID task (Knutson et al. 2000; Nielsen et al. 2008) to examine the neural circuitry involved in performance feedback (Fig. 1a). The goal of this task is to respond to a target within a dynamically adjusted time window in order to maximize monetary rewards. In each trial of our version of the MID task (Fig. 1a), participants first viewed a cue indicating the trial type (gain, loss or neutral), and then, after an interstimulus interval (ISI) of 500–3350 milliseconds, responded to a target with a button press. After a second ISI, the outcome of their response during target presentation was displayed. On gain trials, participants could win $10 if they responded within the acceptable time window or $1 if they responded outside this window; on loss trials, they could lose $1 if they responded within the acceptable time window or lose $10 if they responded outside this window; on neutral trials, they neither lost nor gained money regardless of whether they responded in the allotted time, although they were given feedback regarding whether they responded to the target in time.
Figure 1. MID task schematic and hypothesized DRUG × GENOTYPE effects. (a) Participants were presented with a cue indicating the subsequent trial type (e.g. the ‘win’ cue circled in red). After an ISI of 350–3350 milliseconds, participants then saw a target stimulus (white star on a black background), in which they had to make a button press within a specific time window. The time window ranged from 200 to 650 milliseconds was modified as a function of previous response history to ensure that 67% of responses were in-time. After another ISI of 350–3350 milliseconds, participants were presented with the outcome of the target press (i.e. whether they successfully responded within the allotted window). For gain trials, participants win $10 if they responded within the time window (i.e. ‘hit’) and win $1 if they did not (i.e. ‘miss’). For loss trials (leftmost cue), participants lose $1 if they hit and lose $10 if they missed. For neutral trials (middle cue), participants would win $0 regardless of the outcome, but were still presented with hit or miss feedback. Participants earned 10% of their MID winnings for each session (maximum = $53). (b) In a subset of trials, participants completed valence and arousal ratings. Sometimes these were presented immediately after the cue presentation (i.e. ‘Anticipation ratings’) and, at other times they were presented after outcome presentation (i.e. ‘Outcome ratings’). Other trials proceeded without any subjective ratings. (c) Hypotheses of DRUG × GENOTYPE interaction on performance feedback. Val/Val homozygotes (red) will show increased activation to performance feedback with nicotine, while Met allele carriers (blue) will show no change or decreased activation with nicotine.
Download figure to PowerPoint
Three runs of the MID task, each consisted of 54 trials, were completed in each fMRI session. In a subset of trials in each run, subjective ratings of emotional state were included immediately after cue presentation (8 ‘Anticipation ratings’) or immediately after the outcome presentation (16 ‘Outcome ratings’) (Fig. 1b). Emotional state ratings were completed on a visual analogue scale (VAS) for arousal ratings (anchors of ‘excited’ or ‘distressed’) and for valence ratings (anchors of ‘down’ or ‘great’). The remaining 30 trials had no ratings. Ratings of valence and arousal in gain and loss trial types were normalized against ratings for neutral trials by subtracting mean neutral ratings from mean gain and loss ratings. Similar normalizations were also performed for hit and miss outcome types. Total time to complete these three runs was 33 min and 48 seconds.
Whole brain EPI data consisting of 39, 4 mm oblique axial slices were acquired 30° to coronal on a 3 T Siemens Allegra scanner with the following parameters: repetition time = 2 seconds, echo time = 27 milliseconds; flip angle = 80°, field of view = 220 × 220 mm at 64 × 64 matrix. A T1-weighted MPRAGE image was also collected at each scan session with a voxel size of 1 × 1 × 1 mm; repetition time = 2.5 seconds, echo time = 4.38 milliseconds, flip angle = 8°.
Genomic DNA was isolated from blood using standard protocols. The COMT Val158Met (rs4680) functional polymorphism was genotyped on a custom-designed ‘addictions array’ using the Illumina GoldenGate platform (Hodgkinson et al. 2008). The sample of smokers included 10 Val/Val homozygotes, 10 Val/Met heterozygotes and 3 Met/Met homozygotes; the latter two were grouped to form a Met allele carrier group. Although multiple single nucleotide polymorphisms were determined with this ‘addictions array’, the only variant analyzed in relation to behavioral and imaging outcomes on the MID task was the Val158Met polymorphism.
Assessment of population stratification using ancestry informative markers
This is an admixed sample; moreover, the frequency of the Val158Met polymorphism varies considerably by ethnicity. Therefore we avoided the possibility of confounding our results by population stratification by genotyping ancestry informative markers (AIMS) (Enoch et al. 2006a) and deriving the proportions of ethnicity (ethnic factor score) for each individual. A total of 186 AIMs (Hodgkinson et al. 2008) were genotyped in the study sample and in the HGDP-CEPH Human Genome Diversity Cell Line Panel (1051 individuals from 51 worldwide populations) (http://www.cephb.fr/HGDP-CEPH-Panel). PHASE Structure 2.2 (http://pritch.bsd.uchicago.edu/software.html) was run simultaneously using the AIMS data from our sample and the 51 CEPH populations to identify population substructure and to compute individual ethnic factor scores (0–1). For example, an individual might have 0.70 European ancestry (European factor score = 0.70) and 0.30 African ancestry (African factor score = 0.30). The ethnic factor scores that had a mean value ≥ 0.1 in either group are provided in Table 1.
Gene group differences in average age, BAI, FTND and Wechsler Abbreviated Scale for Intelligence (WASI) were determined using independent-sample, two-sided t-tests. Gene group gender differences were determined using chi-square analysis. Genotype group self-reported race differences were determined using Fishers Exact Test and AIMs score differences were determined using a Mann–Whitney test due to non-normal distribution of scores in the gene groups. Significance was set at P ≤ 0.05. See Table 1.
Nicotine craving and Parrott scores were analyzed in a linear mixed effects (LMEs) model. The nicotine craving assessment generates four factors of smoking behavior: emotionality, expectancy, compulsivity and purposefulness. The Parrott includes nine questions on mood related to alertness, contentedness, energy, focus, happiness, hunger, nervousness, satisfaction and tension. Each of these variables from the two assessments was analyzed in a 2 (DRUG) × 2 (GENOTYPE group) × 2 (TIME: pre- or postscan) design.
Average reaction time (RT) to the target in the MID task and total amount of money were analyzed in a 2 (DRUG) × 2 (GENOTYPE group) LME model. Monetary Incentive Delay task ratings of emotional state (arousal and valence) were each analyzed separately for the anticipation and outcome phases of the task (Fig. 1b). Anticipation ratings were analyzed in a 2 (DRUG) × 2 (GENOTYPE group) × 2 (TRIAL type: gain or loss) design. Outcome ratings were analyzed in a 2 (DRUG) × 2 (GENOTYPE group) × 2 (TRIAL type) × 2 (OUTCOME type: hit or miss) design.
As genotype groups differed in WASI IQ scores (Apud et al. 2007; Giakoumaki et al. 2008; Wechsler 1999) (Table 1), the above behavioral analyses were carried out using de-meaned WASI score as a covariate. Behavioral analyses were performed with SPSS 15.0 (SPSS Inc., Chicago, IL, USA).
All preprocessing and first-level magnetic resonance imaging analyses were performed using the AFNI software package (Cox 1996). Preprocessing steps included volume registration for motion correction, slice-timing correction and temporal normalization. Whole brain data were spatially normalized and smoothed to a 8-mm full width at half maximum (FWHM) (Friedman et al. 2006). The six time-locked trial outcome regressors for hit or miss on each of the three trial types (gain, loss or neural): Gain Condition-Hit outcome, Loss Condition-Hit outcome, Neutral Condition-Hit outcome, Gain Condition-Miss outcome, Loss Condition-Miss outcome, Neutral Condition-Miss outcome were then convolved with a model of the hemodynamic response and its first temporal derivative. In addition, six head motion parameters as well as 3 cue (gain, loss and neutral) and 2 target (fail or succeed to press target in time window) presentations were included as regressors of no interest. Additional regressors of no interest were as follows: 2 for rating arousal and valence during anticipation; and 4 for ratings of arousal and valence for hits and misses at outcome.
As the focus of this study was on performance feedback, our group level imaging analysis was restricted to the outcome phase of the task using LME modeling (AFNI program 3dLME). The model was a 2 (DRUG) × 2 (GENOTYPE group) × 3 (TRIAL type) × 2 (OUTCOME type) design. Further, de-meaned WASI and average RT for ‘hit’ or ‘miss’ on each trial were used as covariates. An omnibus P ≤ 0.05 (defined as a minimum cluster size of 27 voxels (729 µl) at a voxel-wise threshold of P < 0.001) was considered significant throughout. Post-hoc analyses were conducted at the whole brain level (Pcorrected ≤ 0.05) and significant clusters were examined for spatial overlap with the original clusters showing significant interactions. Specifically, for significant DRUG × GENOTYPE group interactions, post-hoc comparisons of drug were conducted separately by genotype group. For significant DRUG × GENOTYPE group × OUTCOME type interactions, post-hoc comparisons of drug were conducted separately by genotype group and outcome type. To examine the patterns of significant effects and interactions (i.e. Figs. 2, 3), data were averaged in ROI's functionally defined by significant results from the whole brain analysis. Figure 1c graphically summarizes the hypothesized linkage between genotype, synaptic DA, behavioral performance and task activation.
Figure 2. DRUG × GENOTYPE interactions in three representative cortico-striatal areas. For all three regions, Val/Val homozygotes showed greater activation in the nicotine condition compared to placebo condition to outcome regardless of whether outcome was ‘hit’ or ‘miss’. Met allele carriers, however, showed a reversed effect of drug condition—that is, they showed greater activation in the placebo compared to nicotine condition. Data in bar graphs represent average beta weight across subjects. Error bars represent standard errors. FG, frontal gyrus; IFG, inferior frontal gyrus. *P < 0.05.
Download figure to PowerPoint
Figure 3. DRUG × GENOTYPE group × OUTCOME type interaction. For miss trials, Val/Val homozygotes showed greater activation in the left cingulate/superior frontal gyrus (SFG) in the nicotine vs. the placebo condition (a similar pattern to that reported in Fig. 2), compared with the change in activation in Met allele carriers across patch conditions. For both gene groups, during hit trials there was no differential activation between patch conditions. Peak activation is at (−9, 12, 42). Data in bar graphs represent average beta weight across subjects. Error bars represent standard errors. *P < 0.05.
Download figure to PowerPoint
- Top of page
- Supporting Information
Our preliminary results suggest that both nicotine and the functional COMT Val158Met polymorphism regulate PFC and striatal activity during performance feedback such that Val/Val homozygotes show increased activation while Met allele carriers show decreased activation in the presence of acute nicotine in the MID task. These imaging results are supported by a similar DRUG × GENOTYPE group interaction in the subjective report of satisfaction and relaxation levels, in which Val/Val homozygotes report more positive mood than Met allele carriers under the nicotine patch condition. Although this self-reported interaction on mood levels did not survive false discovery rate (FDR) corrections for comparisons across all nine Parrott subscales, the behavioral interaction supports the robust DRUG × GENOTYPE interaction found in the imaging data.
This main finding extends previous work that showed an interaction between COMT GENOTYPE and nicotine on PFC function during a WM task (Loughead et al. 2009) in which nicotine administration increases performance and activation, but only in Val/Val homozygotes. In particular, in addition to effects in the PFC, where COMT is hypothesized to exert direct influence on DAergic tone (Bilder et al. 2004), the present study also shows the effect of this polymorphism on ‘downstream’ striatal areas. To our knowledge, this is the first evidence of COMT genotype group differences being modulated by acute nicotine challenge during performance feedback in a cortico-striatal circuit.
While the finding of a COMT genotype × DRUG interaction in striatal regions during performance feedback processing is novel, others have showed that the COMT polymorphism may also affect function of downstream striatal areas in other cognitive domains. For example, COMT genotype has been shown to affect activation of ventral striatal regions during gambling (Camara et al. 2010; Dreher et al. 2009; Marco-Pallares et al. 2009; Yacubian et al. 2007) and reversal learning tasks (Krugel et al. 2009). Indeed, other human studies, both postmortem and in vivo imaging, suggest that the COMT polymorphism affects striatal DAergic tone (Akil et al. 2003; Meyer-Lindenberg et al. 2005). Finally, nicotine × COMT genotype interactions have been observed for DA release in ventral caudate and nucleus accumbens, with Val/Val homozygotes demonstrating greater smoking-induced striatal DA release as assessed by PET imaging (Brody et al. 2006). In the results reported here, the absence of a DRUG × GENOTYPE interaction for RT indicates that these results are not explained purely by PFC-related cognitive or attentional phenomena, as has been previously shown (Loughead et al. 2009).
A mechanism by which this polymorphism influences cortical and subcortical tonic and phasic DA transmission was proposed by Bilder et al. (2004). In this model, low PFC DAergic tone in Val/Val homozygotes results in low striatal tonic DA via reduced glutamatergic signaling from the PFC to striatum. As a consequence, reduced DAergic feedback on autoreceptors located on DA neuron terminals in the nucleus accumbens as well as perhaps enhanced COMT degradation of extrasynaptic DA in the striatum, would then be expected to reduce overall striatal DA tone. This leads to disinhibition of DA striatal neuronal phasic firing in Val/Val homozygotes. Synaptic D2 receptor transmission, thought to support performance on tasks involving flexible processing (Mehta et al. 2004) may then be enhanced in this genotype group. In contrast, in Met allele carriers, high PFC DA tone facilitates cortical D1 transmission and suppresses striatal phasic DA firing, a scenario that favors maintaining stable mental representations or sustained attention (Dickinson & Elvevag 2009). Indeed, the genotype groups in our study illustrate this behavioral difference in that the Met allele carriers had, on average, significantly higher IQ, while the Val/Val homozygotes performed faster overall on the task. The genotype difference in RT has also been noted in tasks requiring frequent updating and cognitive flexibility (Colzato et al. 2010).
Although we cannot say what neurobiological processes are responsible for the observed BOLD signal changes to explain the observed two-way COMT GENOTYPE × DRUG interaction and the three-way COMT genotype × DRUG × MISS trial interaction seen uniquely in ACC/SFG activation during miss trials, the Bilder et al. model (Bilder et al. 2004) would suggest that, in the face of low striatal DA tone, nicotine augments phasic striatal DA release in Val/Val homozygotes and enhances striatal D2 signaling. This effect would be expected to be absent in the presence of high DA tone/lower phasic DA firing in Met allele carriers. In other words, nicotine may act to optimize DA-related signaling in areas involved in habitual responding such as the dorsal striatum (Tricomi et al. 2009), which also has been shown to activate in response to both positive and negative valence (Delgado et al. 2003) as we show in the DRUG × GENOTYPE interaction on activation in caudate and putamen to outcome on the MID task. Val/Val homozygote smokers on placebo and Met allele carrier smokers on nicotine showed deactivation in this region. On nicotine, ‘normalizing’ feedback response in dorsal striatum may not be helpful to smokers trying to quit as their normal response to feedback, especially negative feedback, may involve smoking. Thus, activation of this area on nicotine patch may be relevant to the replicated clinical observation regarding poorer nicotine replacement therapy (NRT) treatment outcome in Val/Val homozygotes compared to Met allele carriers (David et al. 2011).
Further, the three-way interaction, seen in cingulate/SFG in which Val/Val homozygotes show increased activation to miss trials following nicotine administration, suggests enhanced sensitivity to negative feedback in this group as a function of presumed increases in DA. Negative performance feedback may act as an error signal that is encoded by phasic DA activity and, in turn, is enhanced by nicotine in Val/Val homozygotes, which may ultimately lead to increased activation of error detection areas like the ACC (Holroyd & Coles 2002). This is supported by results (Marco-Pallares et al. 2009) showing during losses (misses) on a reversal learning task, Val/Val homozygotes (compared to Met allele carriers) exhibit a larger medial frontal error negativity signal, thought to be related to ACC disinhibition. Increased activation of error detection areas may reflect an increased sensitivity to negative feedback, in Val/Val homozygotes similar to the pattern of medial frontal error negativity seen in depressed patients (Tucker et al. 2003). It is of course possible that this effect in smokers may be a consequence of drug-induced plasticity modifying the genetic effect. However, we found no evidence of a relationship between severity of addiction as measured by FTND and activation in the area showing a three-way interaction.
This study has several limitations. Our sample size is small, particularly for the analysis of behavioral data by genotype, and results should be interpreted with caution pending replication with larger samples. In particular, the small sample size limited our ability to analyze across all three gene groups to examine the nonlinear relationship between prefrontal dopamine (as inferred from genotype) and brain function (Goldman-Rakic et al. 2000). That said, our use of COMT Val158Met was intended as a surrogate for prefrontal DA rather than for the functional effects of the gene itself. Furthermore, in a small sample size the signal to noise ratio is lower perhaps increasing risk for false positive results. Although the duration of abstinence was short (under 3 h) in this sample of daily smokers, our results of increased craving in the placebo condition are supportive of using this brief abstinence period as a comparison for nicotine administration via patch, used as NRT. The lack of a main effect of drug condition on mood symptoms indicates that withdrawal symptoms did not confound behavioral responding or neural activation in the placebo patch condition. Theoretically, the single-blind design could have introduced error via unconscious biases on the part of the experimenters, however, the experimenters were not involved in the portion of the study sessions where the task performance and imaging data were collected and interaction with study staff was minimized at all other times. Lastly, inferences about these DRUG × GENOTYPE group interactions on subjective reporting should be limited to supporting the imaging data as previous COMT Val158Met gene group differences in smoking cessation outcome have not found that gene group differences in mood measures account for the smoking outcome differences (Munafo et al. 2008).
Taken together, our findings of COMT Val158Met modulation of nicotine-related changes in performance feedback highlight a possible mechanism underlying the observation that the Val allele is associated with increased susceptibility to nicotine dependence and greater risk for smoking relapse (Colilla et al. 2005; Enoch et al. 2006b; Johnstone et al. 2007; Munafo et al. 2008), particularly after NRT (David et al. 2011). As such, several implications for smoking cessation treatment may be informed by genotype group differences in brain activation in response to nicotine. As the COMT inhibitor tolcapone has been shown to improve cognitive function (Apud et al. 2007; Giakoumaki et al. 2008) and reduce risk taking behavior (Farrell et al. 2012) in Val/Val allele individuals, such a strategy in smokers may also reduce responses to negative feedback and decrease impulsive choice behavior, potentially improving smoking cessation outcomes when using NRT in this genotype group.