The current study used an fMRI approach to examine the effects of abstinence and cue exposure on tobacco dependent individuals, with mixed findings. As anticipated, abstinence resulted in greater craving and withdrawal, however, contrary to our hypothesis, greater VS/NAc activation was associated with smoking-related (vs. neutral) cues when smokers were in the non-abstinent state compared to when they were abstinent. This effect was remarkably consistent in this sample of eight female smokers, and was evident in both whole brain and ROI analyses. We had predicted greater activation in these brain regions during nicotine abstinence, reflecting increased nicotine craving. The rationale underlying this hypothesis was that the craving following deprivation would be a more salient catalyst for VS/NAc brain activation than the actual reinforcing properties of nicotine, particularly among chronic smokers. This was not the case, as smokers exhibited greater relative activation associated with smoking cues when they were satiated with nicotine. That VS/NAc activation was greater in the non-abstinent condition than the abstinent condition may seem surprising given a previously published report of abstinent smokers who demonstrated differentially greater VS/NAc activation (David et al. 2005
). However, the two studies are not comparable because of substantial differences in methodology (i.e., use of enforced abstinence, within-subjects vs. between-subjects designs). In addition, additional sample characteristics, such as the different sex composition, age, and years of smoking between smokers make the two studies even more difficult to compare.
Despite being unanticipated, a potential explanation for these findings can be drawn from the previous literature on the neurobiological effects of nicotine abstinence on smokers. Given that substance cue reactivity appears to be mediated by dynamic changes in mesolimbic dopamine (Katner et al. 1996
; Katner and Weiss 1999
; Melendez et al. 2002
; Weiss et al. 1993
) and given extensive evidence that nicotine abstinence results in deficits in accumbal dopamine release [for a review, see Kenny and Markou (2001)
], it is possible that acute nicotine abstinence reduced the capacity of the mesolimbic dopamine system to produce acute phasic responses to cigarette stimuli, resulting in greater reactivity to smoking cues in the non-abstinent condition.
At first glance, this may seem paradoxical, suggesting that deprived individuals would be less reactive to environmental stimuli, and therefore less motivated to smoke. However, this is not our contention. Rather, these data suggest that abstinent smokers may be less reactive to cues relative to when they are satiated, but smoking behavior is not motivated by cue reactivity alone and self-reported craving does not exclusively reflect reactions environment cues, so much as the individual’s overall motivational state. Smoking is driven by both positive and negative reinforcement (Tiffany and Drobes 1991
), and, during abstinence, negative reinforcement plays a particularly prominent role (Willner et al. 1995
). Thus, a potential interpretation of current data is that during abstinence, smokers may experience less VS/NAc cue reactivity resulting from relatively reduced burst firing of dopamine in the nucleus accumbens, but may nonetheless be highly motivated to smoke to alleviate aversive withdrawal symptoms (i.e., negative reinforcement). In contrast, under normal smoking conditions, substance-related cues may play a more prominent role because of minimal need for negative reinforcement due to ongoing smoking.
Such an interpretation is consistent with studies by Teneggi et al. who demonstrated that tobacco craving and withdrawal are modulated via distinct neural circuits (Teneggi et al. 2002
). In addition, it is consistent with human laboratory studies that have not demonstrated potentiation of cue-elicited craving by nicotine enforced abstinence (e.g., Palfai et al. 2000
; Tidey et al. 2005
). Finally, it is consistent with a recent study by (Waters et al. 2004
) who found that when nicotine replacement alleviated withdrawal symptoms, cue-elicited craving appeared to play a more prominent role in relapse.
Alternatively, a second explanation for the unanticipated finding of greater VS/NAc signal in the non-abstinence condition is that these findings may be related to a ceiling effect in terms of relative activation. A high-magnitude effect of acute abstinence on craving was observed, resulting in craving levels at approximately the 25% and 70% of the scale maximum for the satiated and acutely abstinent conditions (see ), respectively. BOLD imaging is fundamentally a measure of relative activation during the abstinence condition, high levels of craving may have reflected high levels of absolute
VS/NAc activation, which itself may have prevented substantial increases in VS/NAc activation. Thus, with more possible variation, the participants during the non-abstinence scanning may have simply exhibited greater relative
VS/NAc change. This explanation is also consistent with the studies cited above reporting greater relative increases in craving in response to a smoking cue exposure under non-deprived conditions compared to after abstinence (e.g., Tidey et al. 2005
However, we acknowledge that this finding was in contrast to our original hypothesis and these explanations are a posteriori and speculative. As such, they should be considered provisional interpretations, and a basis for further neuroimaging examinations of cue reactivity and nicotine abstinence. Determining that greater VS/NAc cue reactivity is reliably evident during satiation in subsequent studies will be essential to bolster the validity of these findings. It is also important to acknowledge that these effects may be related to the methodology used. An argument could be made that our approach to ROI analysis of aggregating across functional voxels of normalized data on ROI masks that were generated from an average, normalized anatomical image may not be as accurate as using individually-tailored ROI masks. However, of equal concern to the investigators was the need to avoid bias in determining anatomical boundaries according to the statistical maps of each individual (Warfield et al. 2004
Of note, a number of additional findings also warrant discussion. Cue reactivity was associated with bilateral temporal-occipital activation. The most pronounced activation was in the posterior fusiform gyri and inferior temporal gyri. These findings are consistent with previous reports of craving-related posterior fusiform gyrus activation (Due et al. 2002
; McClernon et al. 2005
; Smolka et al. 2006
; Wilson et al. 2005
) and provides further confirmation of the notion that extrastriate visual pathways are integrated with the mesolimbic system to process information during motivational states (David et al. 2005
; Due et al. 2002
). In addition, a significant hemisphere by abstinence condition interaction was detected, however, given the sample size, it is premature to assert that there is clear lateralization of the effect of abstinence condition on VS/NAc activation. There were no significant differences in VS/NAc activation between hemispheres in either condition. Qualitatively, it appears that the mean activation is greater in the right hemisphere than the left hemisphere in the smoking condition and greater in the left hemisphere than the right hemisphere in the abstinent condition, which is consistent with animal studies (Bortolozzi et al. 2003
; Schneider et al. 1982
; Stein et al. 1998
The observation of statistical trends for cue type and cue type by condition interactions is reassuring, as post-hoc testing demonstrated that, as expected, reaction times to smoking-related cues were significantly greater than to neutral cues in the abstinent condition, which is consistent with results of other studies and suggestive of incentive sensitization to smoking-related stimuli (Havermans et al. 2003
; Trimmel and Wittberger 2004
). Furthermore, the observation of faster reaction times in the second session is expected, as subjects have become accustomed to the task. Although not statistically significant, the difference in reaction times in the non-abstinent condition was in the expected direction and does not obviate the possibility of a smaller effect of cue type not reflecting nicotine deprivation.
There are also a number of limitations of the current study that need to be considered. The sample used was small, and although multiple sessions were used to partially address this, the study had relatively low statistical power. In addition to the sample size, one of the weaknesses of the study was that, as a result of three male subjects not attending all scanning sessions, complete data was only available for females. Thus, the observations can only be generalized to females.
A recent systematic review of the literature by (Carpenter et al. (2006
) demonstrated a lack of consistency of findings with regard to sex differences in cue reactivity, but did find evidence of an effect of menstrual cycle phase on the severity of nicotine withdrawal. Specifically, in 13 studies, women in the luteal phase demonstrated heightened experiences of withdrawal or craving. The degree to which variation in menstrual cycle phase for the three of eight women in our sample who were pre-menopausal might have affected our results but this cannot be evaluated in the present study because precise dates regarding last menstrual periods or biomarkers were not available. We acknowledge, however, that menstrual cycle phase could present a potential confound.
Another limitation to the study is that without subtraction data using methods such as arterial spin labeling, we cannot ascertain the degree to which changes in regional cerebral blood flow resulting from nicotine vs. neural activation associated with smoking-related cues. However, we have no reason to expect that the BOLD contrast (signal during smoking vs. neutral images) would be differentially affected, and our data reflect relative activation rather than absolute activation. Theoretically, the known protective effects of endogenous estrogen on the cardiovascular system in women might indirectly affect fMRI BOLD signal comparisons of male and female subjects due to differences in regional cerebral blood flow. However, we were not able to compare our female smokers to male counterparts and thus cannot dray any conclusions regarding the influence of endogenous estrogen on our fMRI results—particularly because none of the participants reported oral contraceptive use nor use of hormone replacement therapy.
Yet another potential limitation is that the unbalanced nature of the design, with less frequent presentation of smoking-related pictures, might result in evoking a greater response to smoking pictures given the novelty of the target (smoking) pictures and that this could pose a confound to the contrast. However, we were interested in constructing a paradigm that might reflect the occasional smoking-related cues that occur on a daily basis, and thus be able to evaluate potentially less frequent but more salient cues. One of the reasons for including a reaction time task was to evaluate attentional bias toward smoking-related pictures and the effects of acute abstinence on reaction times to a simple gender discrimination task. Our reaction time data do suggest, with statistical trends, that attention to smoking-related cues was relatively impaired in the acute abstinence condition, and suggest greater attentional bias toward smoking-related cues given the longer reaction times associated with smoking vs. neutral cues. In summary, we believe that the unbalanced design presents advantages and disadvantages but acknowledge the distinct possibility that it could present a potential confound.
Despite these limitations, these data provide additional evidence implicating the VS/NAc as an important locus in the processing of salient drug-associated stimuli and revealed an unexpected attenuation of such reactivity following nicotine deprivation. As such, these results contribute to the sparse literature on the neurobiological effects of abstinence on motivation for tobacco in dependent smokers. If the current findings can be replicated, they may considerably clarify the differential processes that contribute to the maintenance of ongoing smoking and relapse following cessation.