Effects of varenicline and cognitive bias modification on neural response to smoking-related cues: a randomised control trial

Drug-related cognitive biases have been positively associated with drug-craving and increased likelihood of relapse. Cognitive bias modification paradigms have been developed to attenuate cognitive biases but there have been few studies that examined neural responses to these paradigms. This study compared neural responses following CBM and explored whether CBM effects were potentiated by varenicline administration. This was a double-blind placebo-controlled study with two between subject factors of drug (varenicline, placebo) and CBM (attend towards smoking cues, train away from smoking cues, control training) that recruited daily (≥ 10 cigarettes per day) non-treatment seeking smokers. Participants (n = 67, 53% female) were randomised to one-week of drug administration (varenicline or placebo) before attending a study session at which they were randomised to CBM condition, and underwent an fMRI scan were they were presented with smoking and neutral cues. Neural response to smoking (vs. neutral) cues, cognitive bias, craving and mood were assessed. There was no evidence of CBM effects on any outcomes. There was evidence of effects of varenicline on craving, with greater reductions in craving in the week preceding the study session in the varencline group (p = 0.04, ηp2 = .06). There was also evidence of a drug by CBM interaction for neural responses (z = 3.78, p <0.001). Compared to placebo, varenicline was associated with greater activation in the right temporal middle gyrus in the CBM control condition, compared to an opposite effect in the CBM “attend towards” condition. These data suggest that CBM does not modify cognitive bias, subjective craving and mood, or neural response to smoking cues. There was also no evidence that CBM effects were potentiated by varenicline.


86
Consequently, reduction in cognitive bias is a potential target for therapeutic 87 intervention.

88
There is evidence that it is possible to reduce cognitive biases using computer-89 based cognitive bias modification (CBM) paradigms that "train" individuals to allocate 90 attention away from disorder-relevant cues. CBM has been shown to reduce cognitive 91 and has also been associated with reduction in other symptoms such as low mood 92 (Baert, De Raedt, Schacht, & Koster, 2010). Attwood and colleagues (Attwood,

93
O'Sullivan, Leonards, Mackintosh, & Munafo, 2008) reported decreased cognitive bias in a group of smokers following one session of stimulus-avoidance CBM using a 95 modified dot probe task. Compared to a group who had been trained to attend to 96 smoking cues, there was evidence that the avoid group also showed attenuated 97 craving in response to in vivo smoking cues in a subsequent cue exposure test (male 98 participants only). A subsequent study in tobacco smokers found similar decreases in 99 cognitive bias following CBM, but did not observe generalisation of these effects of 100 other relevant behaviours (e.g., cigarette craving) or novel (untrained) stimuli (Field,

108
There has been growing interest in the development of combination drug-109 behavioural therapies, in which a drug is used to augment the outcomes of a 110 behavioural intervention (Swerdlow, 2012). This may offer a solution to the low efficacy 111 and reliability of CET effects, if a suitable pharmacological agent can be identified. The 112 smoking cessation pharmacotherapy varenicline acts as a partial agonist of the α4β2 113 nicotinic acetylcholine receptor and aids cessation by reducing cigarette craving and 114 withdrawal symptoms. However, it has also been associated with a reduction in cue-

120
The current study replicated earlier work by examining the effects of CBM on 121 behavioural measures of cognitive bias (visual dot probe and modified Stroop), and extended the work in two important ways. First, we examined whether 7-day pre-123 treatment of varenicline enhanced the effects of CET on smoking cue reactivity and 124 attentional bias. Second, using fMRI, we examined the neural responses to smoking 125 cues following treatment. Neuroimaging studies suggest that drug-related cognitive 126 biases are the result of a failure of cognitive regulatory systems to increase control in 127 the presence of salient cues that increase processing in the reward and emotional 128 centres of the brain (e.g., striatum, amygdala) (Hester & Luijten, 2013              presented. Before and after each block, a crosshair was presented for 5 s.

251
Participants were then asked to rate cigarette craving on an 8-point scale ("none at 252 all" to "extreme"). The scale was presented for 10 s followed by a crosshair for 253 another 10 s. Thus, the total interblock-interval was 25 s. The sequence of events 254 was controlled using EPrime version 2 software (Psychology Software Tools Inc.,

255
Pittsburgh PA), and total task time was approximately 10 min.

389
For error data, three participants were identified as outliers in the pre-CBM 390 condition and one participant was identified as an outlier in the post-CBM condition.

391
These data were removed from main analysis. After data removal error data were not 392 normally distributed and a square root transformation was applied to these data.  craving from session one (pre-drug) to session two (post-drug) in both drug groups, but 413 this effect was larger in the varenicline group (see Table 2). For cigarette craving VAS

514
Taken together these findings support a benefit of varenicline on tonic craving 515 and neural response to smoking cues (which may be driven by the craving effects).

516
While the effects of varenicline may be small, they are meaningful given the fact that 517 the dosing regime delivered in the study is substantially lower than the clinically 518 prescribed dose (i.e., 1 week compared to a standard 12-week course

534
There are some limitations of this study that should be considered when 535 interpreting these findings. First, our sample size was small for the analysis of 536 interactions. Our planned recruitment of 72 participants was achieved but not all 537 participants were tested to completion, and our final sample was lower (n = 67 for subjective and cognitive data; n = 64 for fMRI data). We also have a computer 539 malfunction for one of the conditions that was not identified until data were extracted.

540
We had to replace a number of participants in one CBM condition (avoid) and therefore 541 these individuals were tested outside of the randomisation sequence. We do not 542 however expect that this had a substantial effect on outcomes as these individuals 543 were testing in close time proximity to the rest of the sample. Furthermore, the 544 researchers collecting data were not aware of the reason for additional recruitment, 545 and therefore double-blinding was maintained. Third, our study recruited non-treatment 546 seeking smokers, and it is plausible that effects of CBM may be stronger in individuals 547 seeking treatment.

548
This study investigated neural responses to smoking cues following varenicline 549 and CBM treatment. There was little evidence of neural effects of either drug or CBM.

550
However, there was evidence of reductions in craving among smokers who completed 551 one-week of varenicline treatment. Drug by CBM interactions were exploratory due to 552 small sample sizes, but we observed an interaction on right temporal gyrus activity.

553
Specifically, varenicline appeared to attenuate cue-related activity in the right temporal