Training alcohol-related attentional biases in alcohol-dependence. What are the neurocognitive effects?

Recent estimates within the EU, place the yearly economic burden from alcohol
abuse and dependence at 155 billion euros. Despite these enormous costs, less
than 10% of individuals diagnosed with alcohol dependence receive formal
treatment, and of those, only 28% achieve abstinence, suggesting that further
research in treatment-development is warranted (Rehm et al., 2012). Ideally,
treatment evaluation should be multifactorial, and include indices able to
assess the neurocognitive effects of an intervention (e.g. neuroimaging during
performance of cognitive tasks), in addition to typical measures of
treatment-outcome (i.e. degree of alcohol craving, and relapse propensity).
Such designs would allow the assessment of whether changes in real-life
drinking behaviours are mediated by treatment-induced ameliorations of neural
and cognitive processes known to be impaired in alcohol use disorders (Marhe et
al., 2014).
Alcohol use disorders are characterised by augmented attributions of
motivational salience to alcohol itself and to stimuli that predict alcohol
availability, and by deficits in frontal brain regions implicated in *top-down*
cognitive control processes (e.g. Goldstein & Volkow, 2011). Augmented
attributions of salience are reflected in stimulus-driven, or *bottom-up*,
neural, behavioural, and cognitive responses to alcohol-predicting cues, while
deficits in cognitive control processes are reflected in, for example, the
inability to regulate these stimulus-driven responses. *Biased* responses to
alcohol-associated cues develop over the course of alcohol abuse. For example,
both non-dependent heavy alcohol-drinkers and alcohol-dependent patients show
*attentional biases* to environmental stimuli associated with alcohol (i.e.
they preferentially allocate their attention to, and are distracted by
alcohol-associated cues; Cox et al., 2000). Biases are thought to contribute to
the development of binge-drinking behaviours, and the maintenance of alcohol
addiction, and have been related to alcohol craving, and the propensity to
relapse following periods of abstinence (Wiers et al., 2013).
Consequently, recent research has focused on developing interventions aimed at
moderating drinking behaviour, and facilitating abstinence, by reducing these
alcohol-associated biases. *Attentional bias modification* (ABM), is an
attentional re-training technique developed, and originally applied clinically,
for the amelioration of symptoms of anxiety (e.g. Amir et al., 2009). Applied
to heavy-drinkers, a single session of ABM training reduced alcohol-associated
attentional biases (Schoenmakers et al., 2007; Field et al., 2005), but the
effect did not generalize to untrained stimuli. Multiple sessions of ABM in
problem drinkers (Fadardi & Cox, 2009) reduced heavy drinking, and decreased
attentional biases to untrained stimuli, but there was no control group. Our
lab conducted the first small Randomized Controlled Trial of ABM in detoxified
treatment-seeking alcohol-dependent in-patients. The training was successful in
(a) reducing attentional biases to alcohol-related stimuli; (b) increasing the
time-to-relapse; and (c) reducing the time before in-patients were released
from treatment centres (Schoenmakers et al., 2010). Thus, initial data suggest
that reducing attentional biases using ABM, results in clinically-relevant
changes in alcohol-associated behaviours.
However, the data have not always been consistent, and findings suggest that
behavioural change through ABM is possible only in the cases where attentional
biases are successfully reduced (Clarke et al., 2014). ABM in alcohol research
has typically been implemented using adapted versions of the visual-probe
paradigm. Participants see pairs of alcohol-related and neutral images (usually
pictures of soft-drinks), and are asked to respond via key-press to the
direction of a subsequently presented probe (e.g., an arrow pointing up or
down). Attentional allocation is trained away from the alcohol-associated
images, because on each trial of training the probe replaces the neutral image.
Recently, researchers in the area of anxiety developed a novel ABM method based
on the visual search paradigm. Participants are trained to search for an image
depicting positive emotion within arrays of images depicting negative emotions.
Preliminary data from our lab suggests that visual search ABM training,
relative to control training, is effective in reducing both attentional biases
to anxiety-provoking images, and self-reported feelings of social phobia in
anxious adolescents, with stronger effects than the more standard dot-probe ABM
(De Voogd et al., 2014).
We recently adapted De Voogd and colleagues* (De Voogd et al., 2014) visual
search ABM, and ran an online pilot study on its effects on heavy-drinking
university students* alcohol-associated attentional biases and weekly alcohol
use. The *alcohol-visual-search-training* (AVST) requires searching for the
image of a non-alcohol-containing drink in arrays of images depicting
alcohol-containing drinks. The control training requires searching for a
5-petal flower in arrays of 7-petal flowers (as in De Voogd et al. 2014). The
group that received the AVST training, relative to the group that received the
control training, showed a significant reduction in the amount of time it took
to select the image of the non-alcohol-containing drink from arrays with
distracting alcohol-associated images, suggesting a significant improvement in
the ability to ignore alcohol-associated distracters in this group. In
addition, the same group showed significant reductions in attentional biases as
measured using the dot probe task, suggesting that AVST training generalized to
influence performance on other tasks of attentional biases. With respect to the
amount of units participants consumed in the week before and the week after the
training, we found a trend for reduced drinking in the AVST-training group that
approached significance (0.075). In addition, when we examined drinking during
training, the AVST group showed a significant reduction in drinking units
relative to the control group.
An attractive aspect of the AVST is that it may produce its effects by
strengthening *top-down* cognitive control processes in the presence of
alcohol-associated stimuli. Neurocognitive research suggests that stimuli which
are not task/goal-relevant, but possess strong motivational salience, can
interfere with selective attention to less salient yet task-relevant stimuli
(Krebs et al., 2010). A possible mechanism underlying this effect is that the
*top-down* cognitive control processes, necessary for resolving distracter
interference by enhancing the attentional processing of the task-relevant
information (Botvinick et al., 2001), are weakened, when the
task/goal-irrelevant distracters represent motivationally-salient stimuli.
Indeed, a recent functional imaging study (fMRI) demonstrated that the
interfering effect of task-unrelated alcohol-associated stimuli in heavy
drinkers, was coupled with reduced neural responses in areas of the prefrontal
cortex that were also activated by the need for cognitive control in the
absence of alcohol-associated cues (Nikolaou et al., 2013). AVST training
consistently renders alcohol-associated stimuli as task/goal-irrelevant
distracters, and non-alcohol-associated stimuli as task-related targets.
Additionally, by contrast to most dot-probe training paradigms visual search
training does not only involve learning to allocate attention to the
non-drug-associated cue, but it also requires learning to resolve the
attentional interference that is produced by the alcohol-associated
distracters. Thus, AVST should train the recruitment of *top-down* cognitive
control in the presence of alcohol-associated stimuli. At the same time
however, the control training trains cognitive control more generally. Thus,
given that cognitive control processes are thought to be deficient in alcohol
dependent patients, both the control and the real training conditions should
ameliorate deficits associated with cognitive control, but the real training
condition does this in the presence of alcohol-associated stimuli when the
deficient control resources are weakened by the presence of these stimuli. This
reasoning is corroborated by our findings in students, whereby both training
conditions resulted in reductions in drinking post-, relative to pre-training,
however the biggest effect was seen in the real training group.

The aim of the proposed study is to examine the effectiveness of this new form
of attentional bias training as an add-on to treatment-as-usual in facilitating
abstinence from alcohol use in patients diagnosed with alcohol dependence.

Additionally, the study aims to explore the neurocognitive mechanisms by which
this new form of ABM training has its effects, by collecting fMRI data both
before the start of the training and after the training has been completed.

Design
Each patient will be randomly assigned to one of two ABM training groups: a
control group and an experimental group (see "intervention" section for
details). Allocation will be random, double blind and stratified by gender and
will be conducted by a computer.

Testing
Each participant will take part in 2 testing sessions, and a second session in
the week after the participant has completed his/her allocated training
regimen. Both, will take place at the Spinoza Centre in Amsterdam. In between
these two sessions, they will complete 6 x 20 minute training sessions online.
A month after the end of the scanning session, participants will complete a
final follow-up assessment.

Baseline Session
During the baseline session, informed consent will be taken and participants
will have the opportunity to experience the scanning environment by undergoing
a dummy scan. Once participants have declared that they would feel comfortable
in the scanner, and that they are happy to continue, we will administer the
following measures:
(1) Medical history questions
(2) The Beck Depression Inventory (BDI; Beck, 1961)
(3) The Barratt Impulsiveness Scale
(4) Zelf-beoordelings Vragenlijst (Van der Ploeg et al., 1980)
(5) Drug-use history using CORE full
(6) Drug-use previous month using CORE short
(7) Time-Line Follow back (full)
(8) The Fagerstrom Test for Nicotine Dependence (FTND; Fagerstrom, 1978)
(9) The Alcohol Use Disorders Identification Test (AUDIT; Saunders et al., 1993)
(10) Drinking Motives Questionnaire-Revised (DMQ-R; Cooper, 1994)
(11) Obsessive-Compulsive Drinking Scale (OCDS; Anton et al., 1995).
(12) Penn Alcohol Craving Scale (PACS, Flannery et al., 1999).
(13) Alcohol-use history (Age of onset of regular alcohol consumption; the age
of onset of problem drinking; the number of previous attempts at abstinence
with details of whether they were assisted attempts and conducted in inpatient
or outpatient settings; and the number of days/weeks since the last use).
(14) The Alcohol Dot Probe task (e.g. Schoenmakers et al., 2008)
(15) The Alcohol Visual search assessment task (e.g. Kruijt et al., 2013)
These will be followed by the completion of a backward digit span pen-and-paper
task, and a practice run of the Concurrent Flanker Alcohol Attentional bias
task (CFAAT; Nikolaou et al., 2013a; 2013b).
Then, patients will be placed in the MRI scanner and undergo the scanning
procedure (see below), which will include the following tasks:

Baseline assessment inside the scanner
(1) Baseline cue reactivity
(2) Baseline run of the Concurrent Flanker Alcohol Attentional bias task
Upon completion of scanning, participants will rate their craving to the images
presented in the cue reactivity task. At the end of the entire baseline
session, participants will be given an information booklet providing
information regarding the online training procedure. This booklet will inform
them which site they should visit, how often they should visit it, and what
they should do when they are on the site. Before they leave, patients will be
reimbursed for their travel. The baseline session should not last longer than
135 minutes.

Post-training scanning session (1 week after completion of training regimen)
The scanning session will begin with a practice run of the Concurrent Flanker
Alcohol Attentional bias task (CFAAT; Nikolaou et al., 2012; 2013), which will
be used in the scanner. This will be followed by the completion of: (1)
Time-Line Follow back (Sobell & Sobell, 1995), and (2) the Penn Alcohol Craving
Scale (PACS, Flannery et al., 1999) to assess drinking and craving
post-training. We will then assess alcohol-associated attentional biases for
the final time using the dot probe task and the visual search assessment task.
This will be followed by the scanning procedure, minus the structural scan (see
below).

Scanning procedure (both sessions: baseline and post-training session, though
no structural scan will be taken at the post-training session
The MRI protocol in the 3T Phillips MRI scanner will be as follows:
(a) A structural scan (10 minutes)
(b) EPI fMRI sequence during performance of the CFAAT (9 minutes)
(c) EPI fMRI sequence during an alcohol cue reactivity task (11minutes)

The fMRI sequences will be presented in the same order for each participant.

Upon completion of scanning, participants will rate their craving to the images
presented in the cue reactivity task. Thus the session will not take longer
than one hour and 60 minutes. At the end of the session, patients will be
reimbursed for their travel.

Final follow-up session
The follow-up session will take place one month after the Scanning session, and
will involve completing the Time-Line Follow-Back Questionnaire and the Penn
Alcohol Craving Scale online, to assess whether the intervention affected
real-life drinking behaviours and subjective reports of craving. Upon
completion of the final questionnaire assessment, participants will be sent a
Debriefing Brochure. At this time, they will also be sent a cadeaucheque worth
¤40, as reimbursement for their participation.

ABM Training
Each participant will be randomly assigned to one of two ABM training groups
based on the work by Kruijt et al (2013):
(1) Control-training group (Participants will be trained on an
alcohol-unrelated visual search task: Participants will be shown a 4x4 grid of
16 pictures, 15 of which will be images of 7-petal flowers and 1 will be an
image of a 5-petal flower. During the training, the participants will be asked
to search for the picture with the 5-petal flower and ignore the other
pictures. Thus visual selective attention will also be trained in this
condition)
(2) ABM-training group (Participants will be trained on an alcohol-related
visual search task: Participants will be shown a 4x4 grid of 16 pictures, 15 of
which will be images of drinks containing alcohol and 1 will be an image of a
drink that does not contain alcohol. Training will focus on asking patients to
locate the non-alcoholic drink. Thus visual selective attention will be trained
to specifically and consistently ignore alcohol-associated images)

There will be 6 training sessions in total, each lasting for 20-minutes. All
training sessions will be conducted online

All training sessions will be completed by each patient on their own at home.
Patients will be provided with information explaining which website they should
visit and how they should log on to the website, in the baseline session. They
will also be told that they will receive emails notifying them when they should
do a training session, and reminders emails if they have forgotten to complete
a sessio. Patients will also be aware that the main researchers will be able to
monitor that they have indeed completed each training session.

Neither the main experimenter nor the participants will be aware of the
assigned training, as the allocation will be conducted randomly by the computer
as soon as each patient logs on to the website. Patients will be aware that the
main experimenter is not aware of the type of ABM that they are given.

(a) Patients will be required to complete 6 x 20-minute online training
sessions. Due to the fact that the training sessions are conducted online,
patients are able to complete them at their own time and on their own computer.
In addition, the training will not impact in any way on the treatment that
patients will already be receiving as it will be unrelated. We therefore do not
foresee any risks associated with doing the training. On the contrary, based on
previous findings it is expected that this additional form of treatment will
have beneficial effects on real-life drinking behaviours for the patients as we
expect it to facilitate the maintenance of abstinence for a longer period of
time.

(b) We do not foresee any risks associated with completing the tasks or
undergoing scanning. The tasks are cognitive tasks that should not produce any
distress on the part of the patients. In addition, the patients will be
monitored closely and will be given the opportunity to stop the tasks and the
scanning if they so wish at any point in the experimental session. The scanning
session and the baseline session will however be time consuming as they will
require travelling to the appropriate centre , and spending at least an hour
and 30 minutes for each session to be completed. We will however include
several pauses throughout both sessions and participants will be reimbursed for
both the entire study and for their travel to and from the testing centres.