Unique Brain Areas Associated with Abstinence Control Are Damaged in Multiply Detoxified Alcoholics

Background The ability to abstain from drinking, despite incentives to imbibe, is essential to recovery from alcoholism. Methods We used an incentive conflict task to investigate ability to abstain from responding during presentations of incentive cues. Both alcoholic (n = 23) and healthy subjects (n = 22) were required to withhold responding during the simultaneous presentation of two visual stimuli in which the individual presentation allowed responding for monetary reward. Brain structures activated during performance of the task were studied using functional magnetic resonance imaging in healthy volunteers (n = 8), and changes in gray matter volume were studied in a separate group of patients (n = 29) compared with control subjects (n = 31) in regions of interest identified on functional magnetic resonance imaging. Results Abstinent alcoholic patients were severely impaired on the incentive conflict task. The impairment was greater in patients with experience of several versus a single detoxification. Healthy volunteers, during the same incentive conflict task, showed distinct patterns of brain activation (including gyrus rectus, ventromedial prefrontal cortex, and superior frontal gyrus). Reduction of gray matter volume in ventromedial prefrontal cortex and superior frontal gyrus of patients was more extensive in those with multiple detoxifications. Conclusions Performance deficits in alcoholics are associated with withdrawal-induced impairments in prefrontal subfields, which are exacerbated following repeated episodes of detoxification. Detoxification thus compromises functional and structural integrity of prefrontal cortex and may thus impair the ability to control future drinking. Performance in the incentive conflict task is a sensitive biomarker for such deficits.

. Brain activation to reward and reversal. Activity enhancement within medial orbitofrontal cortex (A), and insula (B) associated with reward [A+ or B+ versus C-or D-] and within lateral orbitofrontal gyrus/frontal pole (C) associated with reversal [A-versus A+]. Scale represents T statistic. Contrasts were used as control conditions for masking brain activations during incentive conflict performance to reveal areas activated which are not associated to learning about reward outcomes (reward versus non reward) or to reversal. Data are presented in mean ± SEM.  Table S1. Test procedure in the scanner: Stimulus presentation, number of trials per stimulus and total trials at different phases of performance. Each part was completed before proceeding to the next part. Trials within each part were randomly allocated to each of the stimuli, except that, in Part 2, during the last 24 trials, only the stimuli A+, B+, C-,and D-were presented to allow evaluation of the reversal for stimulus A (A A-) during Part 3.   the incentive conflict task and also underwent structural MRI.

Incentive conflict task
The incentive conflict task in patients comprised initial training followed by a test phase ( Figure   S1). Participants were required to press a computer space bar to obtain a small monetary reward (10 pence) following the presentation of two single element visual stimuli, A+ and B+ (see example Figure S1), each presented 24 times on a computer screen in random sequence.
Following each stimulus presentation, subjects were asked "How likely are you to gain 10 pence?" and responded on the keyboard using a 1 -9 scale anchored with 1 = unlikely, 5 = don't know, 9 = likely. The four final presentations of A and B were used to determine 'awareness' of the cue-reward relationship. Participants were labelled as 'aware' if the mean of their expectancy ratings for both A+ and B+ was greater than 5. There were no differences between patients and control subjects in expectancy ratings, or in probability of response during training. All subsequent analyses were performed on data from aware participants only (i.e., all those who had successfully learned the first stage of the task). In the next phase, a compound stimulus (AB-) was introduced, intermixed with presentations of the rewarded single element stimuli (A+ and B+). Pressing the space bar following AB-resulted in loss of 10 pence ( Figure 1).

Participants
Imaging data from eight healthy participants were included in the imaging analysis. Population characteristics are given in Table S3.

Incentive conflict task
In this version of the task, during training, in addition to the rewarded A+ and B+ stimuli, we included two additional stimuli, C-and D-, that resulted in loss of money, thus enabling us to include a compound stimulus, CD-(to be used as control), in the testing stage, that continued to inform of monetary loss (i.e., no change of valence from C-or D-alone, and the same outcome as compound AB-). Since the stimuli used for A, B, C and D were counterbalanced across subjects, by comparing the CD-response with the AB-response we were able to isolate those brain responses specific to the change in valence from reward predictors A+ or B+ to punished AB-. From the fourteen healthy participants trained in the incentive conflict task, ten were selected on the basis of successful acquisition of the task to be included in the fMRI; scanning data from two participants were not included in the analysis due to technical reasons.
Following a training session outside the scanner to establish awareness of stimuli-reward contingencies, a single element presentation phase in the scanner (Part 1; see Table S1) was For the fMRI analysis three statistical models were computed, one for each part of the task (Table S1). Functional data were acquired in one continuous session (315 volumes per subject, discarding the initial 4 volumes to ensure steady state B0 magnetization).
Anatomical images of each subject's brain were collected using a T1-weighted magnetizationprepared rapid acquisition gradient echo (MP-RAGE) sequence (56 X 256 matrix, 0.9 mm isotropic voxels).
The fMRI data were preprocessed and statistically analyzed using SPM5 (Wellcome Trust Centre for Neuroimaging, University College London, UK) and MATLAB 7 (The MathWorks, Inc., Natick, MA). Functional images were realigned and motion corrected, spatially normalized to standard MNI (Montreal Neurological Institute) space (7); re-sampled to 2 mm isotropic voxels and smoothed (8 mm full-width half-maximum Gaussian kernel).

Participants
Sixty participants were included in the analysis of whom 29 were alcohol dependent (see Table   S2). Social drinkers were recruited from the local community in the central London area. All evaluation procedures and inclusion criteria were the same as with participants (alcoholic patients and social drinkers) who were examined in the incentive conflict task. Five alcohol dependent patients and 1 control participant completed both the incentive conflict task and also underwent structural MRI. The study was approved by the Kings College Hospital NHS Research Ethics Committee.

Structural MRI methods
High resolution T1-weighted structural images from each participant were segmented into gray matter, white matter and cerebrospinal fluid (CSF) volumes using unified segmentation in SPM5.
Images were normalized into standard stereotactic space and modulated by the deformation parameters such that each voxel retained local volume information of the original scan. All images were checked manually for gross structural abnormalities before analysis, and flipped into the axial plane. Analysis was performed using voxel-based morphometry (VBM) with unified segmentation in SPM5 (www.fil.ion.ucl.ac.uk/spm/software/spm5) (8). Unified segmentation performs image registration, magnetic resonance imaging bias field correction and tissue segmentation in one generative model. Information regarding regional volume was entered into the segmented data using modulation by the deformation parameters required to normalize the images.