Lower Ventral Striatal Activation During Reward Anticipation in Adolescent Smokers

IMAGEN investigates reinforcement-related behavior and its neural and genetic basis in a large group (approximately 2,000) of 14-year-old adolescents. Adolescents were recruited through local public schools at eight sites across Europe (Dresden, Berlin, Mannheim, and Hamburg, Germany; London and Nottingham, U.K.; Dublin, Ireland; and Paris, France). The subjects completed extensive neuropsychological and psychiatric assessments as well as structural and functional neuroimaging at these sites. Additional personality questionnaires were administered for home assessment. Details of these procedures are provided elsewhere (12).

The adolescent smokers were matched on age, gender, and IQ with comparison subjects from the IMAGEN study group (12). The smokers had smoked at least one cigarette during the last 30 days, whereas the comparison subjects had never smoked a cigarette in their lives. Furthermore, given that the subjects completed extensive behavioral testing and structural neuroimaging under continuous supervision prior to completion of the monetary incentive delay task (as described in the following), the minimum length of abstinence from smoking was around 2 hours in all smokers. No volunteer was taking psychoactive medication.

The subjects were screened for psychiatric disorders with the Development and Well-Being Assessment questionnaire (14). They also completed a range of questionnaires, reinforcement-related tasks, and structural and functional neuroimaging. The details of these procedures are summarized elsewhere (12). For the purpose of the present study, we focused on six measures. The Temperament and Character Inventory (15) and the Substance Use Risk Profile Scale (16) were used to assess novelty seeking and impulsivity. Alcohol use was assessed by using the Alcohol Use Disorders Identification Test (17), and nicotine and cannabis use were assessed with the European School Survey Project on Alcohol and Drugs questionnaires (18). For nicotine use, we also applied the Fagerström Test for Nicotine Dependence (19). An estimate of IQ was obtained for each participant by averaging the raw scores of four subtests (similarities, vocabulary, block design, and matrix reasoning) of the Wechsler Intelligence Scale for Children, 4th edition (WISC-IV) (20).

Delay discounting is the subjective (discounted) valuation of a future reward based on both its magnitude and its delay; the value typically declines in a hyperbolic manner over time. Delay discounting was assessed by using a previously published questionnaire (21) and consisted of 27 binary choices between a smaller, sooner and a larger, later hypothetical reward. We analyzed these data by applying the softmax choice rule (22),

to estimate the probability of choosing the selected option (o_i) in a trial, given the values of the available options (sv_o1 and sv_o2, corresponding to the smaller/sooner reward and the larger/later reward). The β parameter in this model describes the stochasticity in the subjects' choice behavior, corresponding to the steepness of the sigmoid function. The subjective value (SV) of each option was modeled by using the hyperbolic model of delay discounting (23):

In this model, A is the amount of the delayed reward, D is the delay in days, and k is a subject-specific discount rate. The best-fitting model parameters (k and β) were derived by maximum likelihood estimation (22) implemented by using MATLAB (MathWorks, Natick, Mass.) optimization functions. Discount rates were log transformed before the statistical analyses to account for their highly skewed distribution. High discount rates reflect steep discounting and impatient or impulsive choice behavior, whereas lower discount rates correspond to patient, more future-oriented preferences.

The participants performed a modified version of a monetary incentive delay task (24). In short, each trial involved an anticipation phase, a response phase, a feedback phase, and a fixation period. During the anticipation phase, cues signaling the amount of reward that could be won in a given trial (large, small, or none) were shown for 4 seconds. The subject could win large or small numbers of points (10 or 2) by responding as quickly as possible to a response cue. The points were converted to food snacks (small chocolate candies) following testing (5 points per candy). The subject completed 22 trials per condition, yielding 66 trials in total. One cue (circle with two lines) signaled that a large reward could be won, another (circle with one line) signaled that a small reward could be won, and a third cue (triangle) signaled that no reward could be won in the respective trial. Following a random time interval, a response cue was displayed, and the subject was instructed to respond as quickly as possible to this cue by means of a button press. The time window in which responses were counted as “wins” was adjusted dynamically during the course of the experiment according to the subject's performance, such that on average the subject won in 66% of all trials. The response and feedback phases had a total duration of 2 seconds. Four seconds of intertrial fixation separated the trials.

All group comparisons were conducted by using two-sample t tests with a two-tailed significance level of p<0.05. In the case of multiple subscales from the same test, correction for multiple comparisons was performed by using Bonferroni correction. Behavioral data from the monetary incentive delay task were analyzed by calculating the percentage of successful trials for each condition (large reward, small reward, and no reward). Reaction times were also analyzed for each condition separately. Because of technical problems, reaction times were unavailable for three comparison subjects and seven smokers.

For the correlation of reaction time with ventral striatal activation, we adopted a two-tailed significance level. The correlations of ventral striatal activation with smoking frequency and alcohol use were conducted by using a one-tailed significance level, because we tested the a priori hypothesis of a negative relationship between drug use and striatal activation, based on previous findings linking striatal hyporesponsivity to the severity of addiction (8).

Structural and functional MRI data were acquired by using 3-T MRI scanners from a range of manufacturers (Siemens, Philips, General Electric, Bruker), and the scanning variables were specifically chosen to be compatible with all scanners (12). For the present task, 300 volumes were acquired for each subject. Each volume consisted of 40 slices aligned to the line connecting the anterior-posterior commissure (2.4-mm thickness, 1-mm gap, TR=2.20 seconds, TE=30 msec), and the total scanning time for the task was 11 minutes.

All preprocessing and analysis procedures were conducted by using SPM8 (Wellcome Trust Centre for Neuroimaging, London). In the first step, single-subject echo-planar images were coregistered with the T₁ structural image. Functional images were realigned and resliced to the first volume. A first-level model was constructed on the unsmoothed single-subject data by using the following regressors: 1) anticipation of large reward, 2) anticipation of small reward, 3) anticipation of no reward, 4) feedback indicating large reward, 5) feedback indicating small reward, 6) feedback indicating no reward. Each regressor was included twice, once for successful trials, i.e., hits, and once for unsuccessful trials, i.e., misses. Thus, there were a total of 12 regressors. Trials in which subjects failed to respond were modeled as separate error trials. Movement parameters from the realignment procedure were included as covariates in the first-level model for each subject. Next, contrast images of the parameter estimates were created for each subject. In the present investigation, we focused our analysis on the reward anticipation phase, i.e., greater neural responses during anticipation of rewards (pooled over large and small amounts) than during anticipation of no reward.

We created a custom template from the T₁ images of 552 adolescents by using the DARTEL toolbox (25) as implemented in SPM8. Single-subject contrast images were normalized to Montreal Neurological Institute (MNI) space by means of this custom template on the basis of the individual subjects' DARTEL flow fields, and they were smoothed with a 6-mm Gaussian isotropic kernel. The normalized and smoothed single-subject contrast images were then taken to a second-level random effects analysis (two-sample t test of comparison subjects versus smokers). In this analysis, the IQ estimate, gender, total score on the Alcohol Use Disorders Identification Test, and scanning site were included as covariates of no interest. A separate covariate for each site was included, coded 1 for subjects scanned at that site and 0 otherwise. For the reported conjunction analysis, we used a conservative approach (26) that requires that a given voxel exceed the threshold in each of the examined contrasts separately.

We conducted two additional second-level models. First, gender effects and gender-by-group interactions were investigated by means of a full factorial model as implemented in SPM8. In keeping with the two-sample t test approach already described, site, alcohol use score, and IQ were also included as covariates of no interest in this model. Second, we examined whole-brain effects of smoking frequency in a multiple regression model with the covariates gender, IQ, alcohol use score, and smoking frequency.

For all analyses, the threshold was set to p<0.05, corrected for family-wise error, by using 12-mm spherical search volumes based on activation peaks reported in previous studies (i.e., independent data). Correction for the ventral striatum was based on a 12-mm sphere centered at x, y, z values of ±14, 8, –8 (27). Correction for the midbrain was based on a 12-mm sphere centered at coordinates ±9, –15, –15 (28). All images are displayed with a threshold of p<0.001, uncorrected, with at least 10 contiguous voxels for visualization purposes, and projected onto the mean structural scan of all 552 volunteers who were used for the construction of the DARTEL template.

Smoking and neuropsychological characteristics of the participants are listed in Table 1 and Table 2, respectively. Notably, the scores for novelty seeking were significantly higher for the smokers than for the comparison subjects (Table 2 and Figure 1, top). The smokers also preferred smaller-and-sooner rewards over larger-and-later rewards significantly more than the comparison subjects, as reflected in significantly higher delay discount rates (Table 2 and Figure 1, bottom). The groups did not differ on IQ (see Method section for description of this measure) or gender distribution. Cannabis use, on the other hand, was more common in the smokers than in the comparison subjects (Table 1). Also, the smokers scored significantly higher on the Alcohol Use Disorders Identification Test, and the clinical ratings from the Development and Well-Being Assessment revealed a higher estimated probability of psychiatric disorders in the smokers (Table 2).

Behavioral data from the monetary incentive delay task are provided in Table 3. No differences in accuracy were observed between the smokers and comparison subjects, but the smokers had slower reactions, in particular for hits.

No group differences were observed for the feedback period at p<0.001, uncorrected, and analysis of the fMRI data therefore focused on the reward anticipation phase. Specifically, we focused on contrasts showing greater responses to the anticipation of gains than to anticipation of no gain (see Method section). Note that we applied normalization to MNI space through a custom template created from a set of 552 T₁ images of 14-year-old adolescents by using DARTEL (25) to account for the fact that standard neuroanatomical templates may be inappropriate for adolescent neuroanatomy. We next analyzed group differences in a two-sample t test second-level model in SPM8; the model included gender, the IQ estimate, the total score on the Alcohol Use Disorders Identification Test, and the scanning site as covariates of no interest.

In this model, we first examined reward anticipation in the smokers and comparison subjects separately. This revealed significant activation in the ventral striatum in both groups. For the comparison subjects (Figure 2A) the peak x, y, and z coordinates (in millimeters) were –9, 5, 1 (t=10.64, df=41, p<0.001, corrected for whole-brain volume), and for the smokers (Figure 2B) the coordinates were –8, 6, 0 (t=4.02, df=41, p_SVC=0.04, corrected for small volume; see Method section); for complete fMRI results see Supplementary Tables 1 and 2 in the data supplement accompanying the online version of this article. Next we identified regions of the reward circuit (ventral striatum and midbrain) showing smaller responses to reward anticipation in the smokers than in the comparison subjects (Figure 2C). Smokers showed lower responses to reward anticipation in the left ventral striatum (–9, 3, 1 [t=4.73, df=74, p_SVC=0.002]) and a more lateral region of the left putamen (–21, 11, –3 [t=4.35, df=74, p_SVC=0.006]). Similar results were observed in two locations in the right ventral striatum (8, 17, –3 [t=3.91, df=74, p_SVC=0.03] and 8, 9, 0 [t=3.71, df=74, p_SVC=0.04]) and one location in the more lateral right putamen (18, 11, 3 [t=3.70, df=74, p_SVC=0.04]). A region in the right midbrain showed a similar pattern but did not survive small-volume correction (–10, –15, –15 [t=3.34, df=74, p_SVC=0.74]).

For completeness, we also performed an exploratory test for gender-by-group interactions, using an additional 2×2 full factorial second-level model, with gender (male or female) and group (smoker or comparison) (see Method section). Although the lower ventral striatal responses in the smokers appeared to be somewhat more pronounced in boys than girls, this interaction effect did not survive correction for multiple comparisons across the ventral striatum region of interest (p=0.08).

These findings raise the obvious question of whether lower striatal responses are the consequence or cause of smoking. We therefore next examined a subset of smokers with extremely mild smoking habits, i.e., fewer than 10 total lifetime occasions of smoking. Even in these very mild smokers, in whom an alteration of reward processing due to chronic nicotine use is extremely unlikely, the left striatal reward anticipation response was still significantly attenuated (two-sample t test at the peak voxel in the left ventral striatum: p=0.02, Figure 2D). The same was true for the non-cannabis-using subgroup of the smokers (p=0.003, Figure 2D). We also examined responses in this region in a subset of smokers matched to the comparison subjects on scores on the Alcohol Use Disorders Identification Test (mean total score=2.42, SD=1.46; paired t test of difference between groups: p>0.07). The smokers matched for alcohol use also showed significantly less ventral striatal activation than the comparison subjects (p<0.001, Figure 2D).

In the next step, we examined the relationship between smoking frequency and reward-anticipation responses in these regions in the smoker group. Smoking frequency scores were calculated by averaging the scores from the European School Survey Project on Alcohol and Drugs questionnaires (18) for recent smoking frequency (over the last 30 days) and lifetime smoking frequency, two measures that were highly correlated in our group of 43 smokers (r=0.55, p=0.0001). fMRI activity in all left striatal regions showed negative correlations with smoking frequency (left ventral striatum: r=–0.23, p=0.054; left putamen: r=–0.41, p=0.004; Figure 2C). These correlations were still significant when subjects with a history of cannabis use (N=14) were excluded from the smokers group (left ventral striatum: r=–0.32, p=0.05; left putamen: r=–0.43, p=0.01). Similar correlations with smoking frequency were observed in the right striatum (right ventral striatum, two clusters: r=–0.20, p=0.10; r=–0.25, p=0.06; right putamen: r=–0.23, p=0.07), linking hypoactivation of these regions to the frequency of nicotine use. For completeness, we report results from a whole-brain analysis (see Method section) of the correlation with smoking frequency in the smokers group in the online Supplementary Table 3.

Finally, we formally tested for a present but low reward anticipation signal in the smokers, performing a triple conjunction analysis testing for regions exhibiting main effects of reward anticipation in each group separately but also showing a significant attenuation in the smokers (p<0.001, uncorrected for each contrast). Only the left ventral striatum showed this pattern (coordinates –8, 5, 1; t=3.71, df=74, p_SCV=0.04; Figure 3A). Furthermore, this region showed the same pattern as the ventral striatal peak in the group comparison discussed earlier, i.e., lower responses in the smokers were also observed when cannabis users were excluded (p=0.04), when subjects were matched for alcohol use (p=0.003), and when only mild smokers were included (p=0.06).

In light of the observed differences in reaction times between groups, we correlated individual responses to reward anticipation in the ventral striatum (at the peak of the triple conjunction) with individual differences in reaction times. There was a negative correlation of ventral striatal activation with reaction time, such that faster-responding subjects showed greater ventral striatal activation. The correlations were similar in the two groups, and the relationships between activation and reaction times for hits are shown in Figure 3B; the correlation fell short of significance in the comparison subjects. Similar correlations were observed in analyses using mean reaction times for misses and for overall reaction times (data not shown).