This is the first study to use pharmacokinetically modeled intravenous alcohol infusion to study how the brain’s response to conditioned alcohol cues is modified during priming exposure to alcohol. The most robust result to emerge in this sample of hazardous drinkers was the potentiation of the response to alcoholic drink odors in the posterior cingulate/retrosplenial area from alcohol priming.
However, it is increasingly clear that the posterior cingulate area is part of the brain’s reward circuit, and critically involved in subjective reward value (Kable and Glimcher, 2007
), reward receipt (Taylor et al., 2006
), and reward choice (McClure et al., 2007
; McCoy and Platt, 2005
). Consistent with these findings, this area has been previously identified as responding to photographs of alcoholic beverages in adolescents with alcohol use disorders, and is an area in which activity from alcohol-related stimuli correlated with the number of drinks consumed per month (Tapert et al., 2003
). Hermann et al. (2006)
similarly found that activity in posterior cingulate induced by the visual images of alcoholic beverages discriminated between alcoholic patients and controls.
Given Tapert’s particular finding, we examined our own data post hoc
for correlations between BOLD response and drinking patterns. While we found no correlation between the number of drinks consumed per month and BOLD signal activation in the posterior cingulate, we did observe a significant positive correlation between the number of heavy drinking days (for men, greater than 4 drinks; for women, greater than 3 drinks) and BOLD signal activation in the retrosplenial-posterior cingulate cortex (peak maximum p
= 0.002 at [−8, −34, 28]; see ), which is within 10 mm of the effect reported by Tapert et al. (2003)
. This positive correlation between heavy drinking days and BOLD response in the posterior cingulate reinforces the concept that this area is involved in subjective reward value
(Kable and Glimcher, 2007
). That is, individuals with a higher number of heavy drinking days presumably place a higher subjective value on alcohol and find it more rewarding. The correlation was, however, present only with the [AO > ApCO]alcohol
contrast, and not with the [AO > NApO]alcohol
contrast (or in any of the comparisons under placebo), which is similar to Tapert’s results where the neutral condition consisted of appetitive stimuli (non-alcoholic drinks).
Fig. 7 Plot of the [AO > ApCO] effect under alcohol as a function of number of heavy drinking days within a 90-day period (mean-centered across the sample) in the retrosplenial cortex peak maximum [−8, −34, 28]. Blue arrow indicates the (more ...)
There was highly suggestive evidence that priming exposure to alcohol may also enhance the contrast between responses to alcohol-related odors and those of non-appetitive control odors in other areas known to be important in appetitive drive. In the largest available sample for both alcohol and placebo infusion conditions, the difference between AO and NApO was largest under alcohol in the NAc, and in medial frontal and orbitofrontal cortices, both of which project to the NAc/ventral striatum (Haber et al., 2006
) and to the posterior cingulate/retrosplenial area (Kobayashi and Amaral, 2003
). Activity in medial prefrontal cortex, in particular, correlates positively with subjective reward value (Kable and Glimcher, 2007
), and reward receipt (Taylor et al., 2006
). Similar effects were also present in comparing AO to ApCO, although in this case there were no differences in the NAc/olfactory tubercle area. Moreover, the effects were largely specific to AO, without significant differences between the two classes of control odors under either alcohol or placebo.
Orbitofrontal cortex, in particular, plays a central role in the coding of reward (e.g., Breiter et al., 1997
; Kringelbach et al., 2003
; O’Doherty et al., 2001
), with its responses varying as a function of satiety. In primates, orbitofrontal neurons that respond to food odors decrease their responses to the odor of a food eaten to satiety (Critchley and Rolls, 1996
). In humans, O’Doherty et al. (2000)
and Kringelbach et al. (2003)
showed that when a particular food is eaten to satiety, the perceived reward value of its odor decreases, as does the activation induced by the odor in orbitofrontal cortex; perceived reward/pleasantness and activation nevertheless remain unchanged to the odor of unconsumed food. Likewise, Gottfried et al. (2003)
showed that orbital activation to a conditioned visual stimulus (an abstract design previously paired with a food odor) also varied as a function of satiety in a similar manner, as did activation in the ventral striatum, cingulate cortex, amygdala, and insula.
In the present experiment, the data suggested the inverse. That is, the relative difference between responses to AO and responses to non-alcohol-related odors was larger
with exposure to alcohol, at least in the abusive drinkers studied here. In fact, the orbital area showing this effect (34, 34, −14) was highly similar to areas identified by others as showing effects related to satiety (Gottfried et al., 2003
; [24, 33, −12]; Kringelbach et al., 2003
; [−22, 34, −8], although somewhat more anterior to the satiety-specific regions reported by O’Doherty et al. (2000
; [18, 18, −17]). Similar to Gottfried et al. (2003)
, we also found that activation induced by the sensory properties of a reward was sensitive to reward satiety in the ventral striatum and cingulate cortex. We believe that, in these hazardous drinkers, the relatively greater response to the AO compared to control odors reflects the reinforcing effects of the priming alcohol, and increased desire to drink following that prime (De Wit, 1996
; De Wit and Chutuape, 1993
) particularly insofar as the targeted BrAC of 0.05 would be a relatively low concentration for subjects who habitually drink an average of 6 drinks per drinking day. While we did not directly examine desire to drink according to stimulus exposure or infusion type during scanning, it was the case that perceived high and intoxication were not significantly different. As these subjects habitually drank to intoxication, it seems unlikely that they would have reported feeling sated by the alcohol infusion—an assumption consistent with their reported subjective effects of alcohol during imaging1
Thus, low-concentration alcohol exposure appears to have enhanced the contrast between responses to alcohol’s conditioned cues and cues unassociated with alcohol in regions of the mesocorticolimbic pathway, at least in these hazardous drinkers. Where this effect was observed (as apparent from ), it was often a decrease in the NApO rather than an increase in AO (except in the precuneus and posterior cingulate/retrosplenial regions, where AO increased and NApO decreased). Such drug-related effects on the sensory properties of reward may be important to alcohol seeking. For example, Bäckström and Hyytiä (2006)
found that alcohol priming significantly enhanced cue-reinstated responding for alcohol, which mirrors human data showing that low-dose alcohol exposure increases desire to drink in both social drinkers and alcoholic subjects (De Wit, 1996
; De Wit and Chutuape, 1993
Frontal lobe dysfunction is thought to underlie the risk for alcoholism (Giancola and Tarter, 1999
; Justus et al., 2001
), with orbital dysfunction particularly suggested insofar as behavioral disinhibition and impulsiveness are risks for alcoholism (Finn et al., 1994
; Finn and Hall, 2004
). Dysfunction comprising the orbital system and its reciprocal targets involved in satiety and incentive salience might therefore be implicated, as well. Thus, while decreases in orbital activation as a function of food exposure may signal satiety and govern a balanced diet (Rolls, 1999
), dysfunction in this system could be one factor that leads to loss of control in drinking in the early phase of alcohol consumption. Study of social drinkers who do not drink abusively, as well as alcoholic subjects, are required to test this hypothesis.
Since Kringelbach et al. (2003)
and Small et al. (2001)
reported correlations between orbitofrontal activation and perceived pleasantness, the greater activation caused by AO compared to NApO could be related to differences in perceived pleasantness. Two lines of reasoning suggest otherwise. First, the differential activity between these stimulus types occurred primarily in the alcohol infusion condition, whereas pleasantness ratings (made after each infusion) did not differ between alcohol and placebo infusion days. Second, the stimuli perceived as most pleasant were the ApCO. However, the AO still caused greater orbitofrontal activation than ApCO. Secondary analyses, in fact, suggest that the odors perceived as least pleasant (NApO) induced somewhat more activation during alcohol infusion in nearby orbital areas than the odors perceived as most pleasant (NApO > ApCO; 14 voxels at p
< 0.01 at [20, 36, −20]).
There remain important limitations to consider in this study, particularly in making inferences about activity in the NAc and in medial/orbitofrontal areas. Specifically, not all subjects had available imaging data for both the alcohol and placebo sessions, with only 7 subjects having both alcohol and placebo session data. Analyses in this smaller data set also have more limited power to detect effects. While this direct paired analysis principally showed significant differences between alcohol and placebo sessions in the posterior cingulate/retrosplenial area, medial-frontal cortex continued to show significant trends suggesting that AO evoked more activity under alcohol than under placebo, mainly in contrast to ApCO (607 voxels at [−6, 34, 20] at p
< 0.05). Orbitofrontal cortex also showed a small cluster of 27 voxels ([24, 32,−22] at p
< 0.05) in which activity from the [AO > ApCO] comparison was greater under alcohol than under placebo. Of the 7 subjects in the smaller analyses, only 2 had a pronounced right lateral orbital [AO > ApCO] response under placebo, while four had a pronounced effect under alcohol (the remaining subject had a nonsignificantly greater response under alcohol). Finally, compared to our prior study (Kareken et al., 2004
), we did not find NAc activation without exposure to alcohol. The reasons for this remain unclear, although one important difference between the two studies is that the priorsample of hazardous drinkers was composed uniquely of individuals with a family history of alcoholism (e.g., see Murphy et al., 2002
We are continuing to study how olfactory cue-induced activity is modified by alcohol exposure at higher magnetic field strength, and in a larger sample of hazardous drinkers stratified by family history of alcoholism who are compared to socially drinking controls. As a whole, however, these initial results suggest that a low level of exposure to alcohol can sensitize a limbic brain network related to reward cue processing in subjects who routinely seek these rewards.