In the previous sections we described an example of the application of the imaging genetics approach to understanding the neural basis of inter-individual variability in impulsivity; a behavioral phenotype closely related to addictive disorders. We demonstrated the development of a mechanistic framework in which variability in the phenotype of impulsivity could be traced through patterns of neural activity to genetically driven alterations in molecular signaling pathways. However, the above analyses were conducted in healthy, non-addicted adult populations, and thus likely speak only to the pathways associated with early processes (e.g., initiation of substance use). Although it is possible that the same mechanisms underlying impulsivity in healthy populations may also interact with drug exposure to confer risk for drug dependence, additional work is needed to dissect the processes acting in addicted populations. Such research would help to identify the pathways which may place some individuals at greater risk for neuroadaptations leading to dependence and relapse. In the next section, we turn to discussion of building upon this knowledge and applying this approach within substance dependent populations.
As described above, addiction is a complex, multidimensional disorder. As with other complex traits such as impulsivity, it is necessary to refine the behavioral phenotype to focus on specific dimensions or aspects of addictive disorders in order to begin to isolate the specific neural processing mediating inter-individual differences. Similarly, characterization of a neural phenotype which is associated with variability in behavior paves the way for further investigation of the genetic sources of variance and can be used to understand the pathways mediating ultimate clinical outcomes, such as treatment response. Given that research has begun to describe the long-term neuroadaptations resulting from chronic drug exposure, we can begin to examine how variability in these neuroadaptations may predict severity of addictive phenotypes, as well as identify the genetic sources of this variance. Here, we focus on variability in nicotine dependence and incentive processing of non-drug rewards.
As described above, neuroadaptations resulting from chronic exposure to drugs of abuse, including nicotine, are known to involve midbrain dopaminergic function, leading to sensitization of the dopamine response to drugs of abuse and heightened sensitivity to drug related cues (Robinson and Berridge, 1993
). By contrast, behavioral evidence suggests that withdrawal from drugs of abuse leads to deficits in reward functioning. For example, intracranial self-stimulation (ICSS) experiments demonstrate increased reward thresholds during withdrawal from multiple drugs of abuse including cocaine (Markou and Koob, 1991
), amphetamine (Cryan et al., 2003
), alcohol (Schulteis et al., 1995
), and nicotine (Epping-Jordan et al., 1998
). Indeed, opponent-process theory posits that chronic drug exposure results in a compensatory alteration in reward processing in an attempt to correct the imbalance that is produced by constant stimulation of the reward pathways (Koob and Le Moal, 1997
). Consistent with this framework, research with both animals and humans suggests that nicotine facilitates the reinforcing properties of other stimuli (Barr et al., 2008
; Chaudhri et al., 2006
; Donny et al., 2003
; Olausson et al., 2003
), while abstinence from nicotine appears to attenuate the value of other reinforcing stimuli (Weaver et al., 2012
). Indeed, abstinent smokers experience diminished capacity for reward relative to both satiated smokers and non-smokers including less enjoyment from ordinarily pleasurable events and reduced response to financial reward (Dawkins et al., 2006
; Powell et al., 2002
; Powell et al., 2004
). Compared with satiated smokers, abstinent smokers also demonstrate less interference from and report lower levels of happiness in response to positive or pleasure-related stimuli (Dawkins et al., 2006
While some symptoms described above, such as loss of hedonic experience of pleasure, may be mediated by a variety of neurotransmitter systems including endogenous opioids or serotonin, other symptoms related to deficits in motivated behavior during nicotine withdrawal may be related to attenuation of dopamine related activation in the VS. Indeed, Martin-Soelch and colleagues (2003)
reported that although there was a significant correlation between magnitude of monetary rewards and VS activity, this relation was not observed in smokers, suggesting they were much less responsive to even the largest rewards presented. Likewise, data from a PET study revealed a negative correlation between metabolic activity in the VS and abstinence-induced withdrawal, suggesting that a component of withdrawal may be a reduction in activation within this region (Rose et al., 2007
). Together, these data support theories of addiction which suggest that adaptations in dopamine reward pathways due to chronic stimulant exposure, including nicotine, mediates the development of an abstinence-induced withdrawal syndrome characterized by a decreased incentive motivation (Koob and Le Moal, 1997
; Volkow et al., 2004
). Genetically driven variation in dopamine signaling and associated reward circuitry, in interaction with the changes wrought by chronic nicotine exposure, could contribute to individual differences in the extent of reward deficits observed during abstinence, and in turn, the manifestation of nicotine dependence in chronic smokers.
As an initial step in investigating this line of inquiry, we tested whether variation in nicotine dependence, previously shown to be associated with delay discounting (Sweitzer et al., 2008
), was associated with magnitude of change in reward-related VS activation during abstinence compared with shortly after smoking. We utilized the same monetary reward task previously shown to elicit activation associated with impulsivity (Hariri et al., 2006
). It is important to note that these data are from a small sample and hence should be viewed and preliminary and hypothesis-generating. Because these data have not previously been reported, we provide a brief summary of the methods below1
3.1 Materials and methods
Complete data were available from 10 male subjects (mean age, 41.5 years +/− 6.6 SD). All subjects were non-treatment seekers who reported smoking between 10 and 30 cigarettes per day (mean CPD, 19.4 +/− 6.9 SD) for the past 10 to 30 years (mean years, 22.0 +/− 6.3), scored a minimum of 4 on the Fagerstrom Test for Nicotine Dependence (FTND), and had a baseline exhaled carbon monoxide (CO) level of 10 ppm or greater.
Subjects participated in an initial screening session plus two fMRI sessions separated by a minimum of five days. Prior to one fMRI session, subjects were instructed to smoke ad libitum up until the time of their appointment (non-abstinent condition). Prior to the other fMRI session, subjects were required to abstain from smoking for a minimum of 12 hours (abstinent condition). Compliance with instructions was verified using self-report and expired CO levels (abstinence verified as < 8 ppm or 50% of baseline). Order of sessions was randomly assigned and counterbalanced across subjects.
During the screening session, nicotine dependence was assessed using the Nicotine Dependence Syndrome Scale (NDSS; Shiffman et al., 2004
) and the Fagerstrom Test of Nicotine Dependence (FTND; Heatherton et al., 1991
). Procedures within each fMRI session were identical for both sessions. Subjects first completed a measure of nicotine withdrawal (Minnesota Nicotine Withdrawal Scale; MNWS) and were trained on the VS task to be completed in the scanner. After smoking a final cigarette (non-abstinent condition) or waiting an additional 10 minutes (abstinent condition), subjects were tested for expired CO level immediately prior to entering the scan; subjects also completed a 4-item craving questionnaire (Questionnaire of Smoking Urges; QSU-4) just before and after the scan, with the average taken to represent overall craving for each scanning session.
Details of our fMRI task are available through several earlier reports (Forbes et al., 2009
; Hariri et al., 2006
; Hariri et al., 2009
). Briefly, subjects were instructed that they would be guessing whether a hidden number was higher or lower than five, indicated by pressing their middle or index finger, respectively. For each trial, subjects were presented with a question mark indicating that they should make their guess. Following each trial, the actual number was shown, followed by a green up arrow indicating they won money if they got it right (positive feedback) or a red down arrow indicating they lost money if they got it wrong (negative feedback). The blocked design consisted of pseudorandom presentation of trials organized into blocks of mostly positive feedback (4 out of 5 trials) or mostly negative feedback (4 out of 5 trials), interleaved with control blocks in which subjects were presented with comparable visual stimuli and were required to press a button with either their middle or index finger to control for motor activity. Subjects were unaware of the fixed outcome probabilities associated with each block and were led to believe that their performance would determine their net monetary gain, although all subjects received $10 upon completion of the task.
Subjects were scanned using a Siemens 3T MAGNETOM scanner (Siemens AG, Medical Solutions, Erlangen, Germany). Whole-brain image analysis was completed using SPM5 (http://www.fil.ion.ucl.ac.uk/spm
). Following preprocessing, data sets were analyzed using second-level random effects models. For each subject and scan, predetermined condition effects at each voxel within a predefined VS region of interest were calculated using a t
-statistic, producing a statistical image for each contrast: (1) positive feedback > control (2) negative feedback > control and (3) positive feedback > negative feedback2
Analysis of behavioral measures revealed that abstinence was associated with significantly lower CO levels and greater craving relative to non-abstinence. Although the increase in withdrawal symptoms during abstinence was in the predicted direction, this difference did not reach significance (see ).
Mean scores (and standard deviations) on smoking measures for 10 smokers assessed during abstinence and non-abstinence, and t score for the test of significant difference between conditions.
Consistent with previous studies, we observed strong bilateral VS activity associated with both positive and negative feedback blocks, relative to control blocks, collapsed across both abstinent and non-abstinent conditions (). We also observed relatively greater right VS activation in response to positive compared with negative feedback blocks (). When analyzed separately by condition, bilateral VS activation was observed for both positive and negative feedback blocks relative to control blocks during both abstinence and non-abstinence. However, the differential effect of positive > negative feedback was significantly associated with VS activation only during abstinence, while no effect was observed during non-abstinence (). Despite these apparent differences between conditions when analyzed separately, direct comparisons of VS reward-related activation during abstinence compared with non-abstinence were not statistically significant.
Figure 2 Average ventral striatal (VS) activity associated with general feedback as well as the differential effect of reward (p < 0.05, 125 cluster extent threshold for all contrasts), collapsed across both conditions. All slices are presented at Y = (more ...)
Figure 3 Average ventral striatal (VS) activity associated with the contrast of positive feedback > negative feedback (p < 0.05, 125 cluster extent threshold for all contrasts) for each condition. Slices presented at Y=10. a. Differential effect (more ...)
Given the theoretical and behavioral evidence suggesting that abstinence from smoking may result in an attenuated response to reward, the lack of an overall effect of abstinence compared with non-abstinence was surprising. However, we hypothesized that substantial inter-individual variability in reward-related VS activation induced by abstinence could be masking any group-level differences. Consequently, we sought to examine whether variation could be observed across individuals, and whether this variability was related to severity of nicotine dependence. To do this, we extracted right VS activation values from the contrast of positive > negative feedback, including scans from both conditions (). Extracted values for both abstinent and non-abstinent scans for each subject are illustrated in , presented as a function of nicotine dependence as measured by the NDSS. As can be seen, individuals low in dependence appeared to show increases in VS reactivity during abstinence compared with non-abstinence, while individuals high in dependence showed the opposite pattern. This effect was further apparent when examining difference scores calculated by subtracting non-abstinent from abstinent scan VS activation values for each subject. Variability in difference scores was significantly related to NDSS scores (R2
=.002; )—an effect which remained significant when controlling for age, race, session order, and change in CO between abstinence and non-abstinence (Partial r
= −.900, p
= .015). To assess the extent to which these findings generalized to left VS activation, we also extracted activation values from a left VS cluster associated with differential response to reward identified using a more liberal threshold3
. Although relaxing the significance threshold means it is less clear that the extracted values truly reflect reward-related activation, analyses based on this cluster revealed a pattern identical to that observed with the right VS (Partial r
= −.907, p
= .013). Furthermore, the same pattern was observed with the FTND predicting right VS change in activation (Partial r
= −.828, p
= .042), although this did not reach significance in the left VS (Partial r
= −.735, p
Figure 4 Correlations between Nicotine Dependence Syndrome Scale (NDSS) scores and ventral striatal (VS) differential reward-related activity (right hemisphere cluster, maximal voxel coordinates: x = 8, y = 6, z = −6) from the contrast of positive > (more ...)
It should be noted that the high correlations observed within this small sample likely represent an overestimation of the true effect size for the association between nicotine dependence and withdrawal induced changes in VS activation (see Yarkoni et al., 2009
for a discussion of this issue), and replication in a larger sample is needed. However, these preliminary findings suggest that, although abstinence from smoking was not associated with a statistically significant generalized decrease in reward-related VS activation, substantial individual variability was observed in the degree of change in activation induced by abstinence from smoking compared with non-abstinence. Importantly, this variability was significantly related to nicotine dependence. Only the three most highly dependent individuals appeared to show decrements in VS response to reward as a function of abstinence, while those low in dependence appear to show the opposite pattern. While replication is clearly needed, these preliminary findings suggest that a relative decrease in neural sensitivity to reward during abstinence may be associated with high levels of dependence. These findings are consistent with a recent study demonstrating a negative association between reward-related BOLD activation among detoxified alcoholics and number of subsequent drinking days (Heinz et al., 2007
), suggesting a heightened vulnerability to relapse among those exhibiting the greatest reward decrements during abstinence. Furthermore, studies with heroin and cocaine dependent subjects have demonstrated similar blunting of BOLD activation to positive affective stimuli relative to healthy controls (Garavan et al., 2000
; Zijlstra et al., 2009
), suggesting a potential pathway of common liability to addiction.
As described above, multiple neural pathways and environmental factors are likely to contribute to complex behaviors such as difficulty quitting smoking. Although the findings described above require replication in a larger sample, they suggest one potential pathway which may be related to inter-individual variability in more distal processes. Characterization of this neural phenotype among chronic smokers paves the way for further analysis of the mechanisms underlying this variability. Given the influence of common genetic variants (e.g., DAT1, DRD2 −141 Ins/Del) on dopamine signaling associated with reward circuitry and impulsivity, it is possible that these or related genes may also be related to the severity of nicotine dependence by way of their impact on the susceptibility to neuroadaptations induced by chronic smoking. Extending research to behavioral and neural processes unique to addicted populations also opens the door for testing the impact of genetic variation in other pathways such as nicotinic cholinergic or opioid signaling with clear prior relevance to drug abuse and addiction.