Alcohol and Alcoholism, 2020, 1–11 doi: 10.1093/alcalc/agaa089 Article

Article

A Combined Alcohol and Smoking Downloaded from https://academic.oup.com/alcalc/advance-article/doi/10.1093/alcalc/agaa089/5911709 by guest on 29 September 2020 Cue-Reactivity Paradigm in People Who Drink Heavily and Smoke Cigarettes: Preliminary Findings Carolina L. Haass-Koffler 1,2,*, Rachel D. Souza2, James P.Wilmott 3, Elizabeth R. Aston2 and Joo-Hyun Song3

1Department of Psychiatry and Human Behavior, Center for Alcohol and Addiction Studies, Warren Alpert Medical School, Brown University, 121 South Main Street, Providence, RI 02912, USA , 2Department of Behavioral and Social Sciences, Center for Alcohol and Addiction Studies, School of Public Health, Brown University, 121 South Main Street, Providence, RI 02912, USA , and 3Department of Cognitive, Linguistic and Psychological Sciences; Brown University, 121 South Main Street, Providence, RI 02912, USA

*Corresponding author: Department of Psychiatry and Human Behavior, Center for Alcohol and Addiction Studies, Brown Uni- versity, 121 South Main Street, Providence, RI 02912, USA. Tel.: +1 401-863-6630; E-mail: carolina_haass-koffl[email protected]

Received 12 May 2020; Revised 20 August 2020; Accepted 20 August 2020

Abstract Aims: Previous studies have shown that there may be an underlying mechanism that is common for co-use of alcohol and tobacco and it has been shown that treatment for alcohol use disorder can increase rates of smoking cessation. The primary aim of this study was to assess a novel methodological approach to test a simultaneous behavioral alcohol-smoking cue reactivity (CR) paradigm in people who drink alcohol and smoke cigarettes. Methods: This was a human laboratory study that utilized a novel laboratory procedure with individuals who drink heavily (≥15 drinks/week for men; ≥8 drinks/week for women) and smoke (>5 cigarettes/day). Participants completed a CR in a bar laboratory and an eye-tracking (ET) session using their preferred alcohol beverage, cigarettes brand and water. Results: In both the CR and ET session, there was a difference in time spent interacting with alcohol and cigarettes as compared to water (P’s < 0.001), but no difference in time spent interacting between alcohol and cigarettes (P > 0.05). In the CR sessions, craving for cigarettes was significantly greater than craving for alcohol (P < 0.001), however, only time spent with alcohol, but not with cigarettes, was correlated with craving for both alcohol and cigarettes (P < 0.05). Conclusion: This study showed that it is feasible to use simultaneous cues during a CR procedure in a bar laboratory paradigm. The attention bias measured in the integrated alcohol-cigarettes ET procedure predicted participants’ decision making in the CR. This novel methodological approach revealed that in people who drink heavily and smoke, alcohol cues may affect craving for both alcohol and cigarettes.

INTRODUCTION develop AUD (Grant et al., 2004). The co-use of both alco- Individuals with alcohol use disorder (AUD) are three times more hol and tobacco has a multiplicative effect on adverse health likely than the general public to smoke tobacco products and outcomes due to the increased risks associated with using both those with tobacco use disorder are four times more likely to substances (Berg et al., 2016).

© The Author(s) 2020. Medical Council on Alcohol and Oxford University Press. All rights reserved. 1 2 Alcohol and Alcoholism, 2020

A study of 259 people who smoke cigarettes that reported to assess and measure attentional bias can be executed. Rapid eye frequently drinking alcohol examined the subjective consequences movements, or saccades, are used to align the high resolution fovea of naturally occurring simultaneous use of alcohol and tobacco. area of the retina with objects of interest and predict attentional shifts The study utilized ecological momentary assessment techniques (Zhao et al., 2014). Additionally, volitional control of gaze position (Moskowitz and Young, 2006). This work showed that co-use of via saccades has been shown to predict choice behavior in decision alcohol and tobacco products elicits enhanced craving for both making tasks. When observers are presented with multiple items and substances (Piasecki et al., 2011). Similarly, the release felt from tasked to choose an item, individuals choose items that are viewed for smoking may make an individual more likely to crave an alcoholic longer overall and immediately before their response (Shimojo et al.,

beverage. People who both drink and smoke cigarettes are also 2003; Krajbich et al., 2010). This is consistent with the notion that Downloaded from https://academic.oup.com/alcalc/advance-article/doi/10.1093/alcalc/agaa089/5911709 by guest on 29 September 2020 at a heightened risk not only due to the multiplicative morbidity attention is biased toward valuable items during visually mediated of their substance use habits, but also due to the resistance to decision making. Laboratory studies using ET procedures have been interventions aimed at traditional substance using populations, utilized to evaluate the effect of alcohol and cigarette cues on selective because the experienced cravings in these individuals are often attention and motivation. For example, ET was used to validate the stronger (Berg et al., 2016). reduction of attention to a neutral image during the presentation of an Experimental presentation of simultaneous cues has been utilized alcohol image (Laude and Fillmore, 2015). Also, it was shown that in animal studies (Larson and Sieprawska, 2002). Conditioned place the eye gaze dwell time was longer in response to cigarette-related preference (CPP) is a form of Pavlovian conditioning that is used cues compared to neutral cues, confirming that there is attentional to measure subject motivation due to objects or experiences (Luck- bias towards smoking-related cues (Kang et al., 2012). While selective e-Wold, 2011). By measuring the amount of time a subject spends attention can be routinely measured in the ET laboratory procedure, with a stimulus, we can infer the subject’s liking for that specific there are however no ET studies that have evaluated the simultaneous object. Therefore, the CPP represents a subject’s preference due to presentation of alcohol and cigarette-related cues. continuous association with that stimulus (Bardo and Bevins, 2000) The goal of this study was to test a novel methodological and can be translated into human laboratory studies. Furthermore, approach for a simultaneous behavioral alcohol-smoking CR bar laboratories provide a naturalistic environment to assess alcohol- paradigm in people who drink and smoke cigarettes. We paired an related behavior within a controlled setting (Fox et al., 2012; Thomas alcohol-cigarette CR (real cues) with an ET (virtual cues) procedure et al., 2012; Haass-Koffler et al., 2015, 2017; Kenna et al., 2016), to evaluate mechanisms by which substance-associated cues influence where a CPP approach can be applied. alcohol and cigarette related responses (individuals’ substance Although the CR paradigm has been used in many different interaction and selective attention). domains, a single CR session combining preferred alcohol and smok- We hypothesized that people who drink and smoke cigarettes in ing cues in a bar laboratory has its challenges when assessing simulta- the CR will spend more time interacting with, and in the ET will pay neous cravings. Human laboratory studies that evaluated the effects more attention to, alcohol and/or cigarettes rather than water.Finally, of alcohol cues on cigarettes craving and cigarettes-related cues we also explored if this novel integrated protocol will provide a more on alcohol craving have cast mixed results (for extensive review, comprehensive human laboratory model with improved quantitative see Verplaetse and McKee, 2017). For example, alcohol CR proce- and predictive value that may help to determine which substance dures to assess alcohol and cigarettes craving (Gulliver et al., 1995; induces cravings for alcohol and/or cigarettes in the bar laboratory. Rohsenow et al., 1997) have demonstrated that exposure to alcohol cues resulted in significantly greater craving for alcohol and cigarettes in individuals with AUD who also smoke. The effect of alcohol cues MATERIALS AND METHODS also resulted in increased craving for cigarettes, but did not increase alcohol craving (Colby et al., 2004) in individuals experiencing Study setting tobacco deprivation who are moderate to heavy users of cigarettes This was a human laboratory study that combined alcohol and smok- and alcohol. Virtual smoking-related cues (pictures) increase alcohol ing CR with an ET procedure in non-treatment seeking individuals craving in individuals with AUD who also smoke (Drobes, 2002). who drink and smoke (N = 32). The CR procedure was conducted at Fixed-dose alcohol administration in individuals who smoke has been the Center for Alcohol and Addiction Studies (CAAS) and the ET pro- utilized to assess addictive-related behaviors such as a decreased cedure was conducted at the Department of Cognitive, Linguistic and latency to start smoking (Kahler et al., 2012) Psychological Sciences (CLPS) at Brown University, Providence, RI, However, these studies presented only one cue (alcohol or USA. The study was approved by the Brown University Institutional cigarettes-related product) and did not provide simultaneous Review Board. All participants signed an informed consent document exposure to real cues within the same laboratory procedures. A prior to enrollment in the study. cross CR that presented alcohol and cigarettes pictures (Oliver and Drobes, 2015) has also shown that exposure to alcohol cues resulted in significantly greater craving for alcohol and cigarette Inclusion and exclusion criteria in individuals with AUD who smoke. Clearly, literature has shown Participants were recruited from the community using flyers, iden- that alcohol and cigarette-related products influence alcohol and tified after a telephone pre-screening assessment and enrolled after tobacco cravings, however, no human laboratory studies have an in-person screening in the laboratory. Males and females, age administered simultaneous real alcohol- and cigarette-related cues 18–65 years old, were enrolled according to the Center for Disease in a bar laboratory to examine the subjective substance-related Control and Prevention (CDC) guidelines for heavy drinking (≥15 behavioral response. drinks/week for men; ≥8 drinks/week for women) (DeSalvo, 2016). Eye tracking (ET) has been validated as an indicator of attention, Participants were eligible to participate in this study if they also cognitive processing and memory and is widely used in tobacco con- smoked at least 5 cigarettes/day (<10 light; 10–19 moderate; ≤ 20 trol regulation (Meernik et al., 2016). With ET,an objective approach heavy smokers) (Okuyemi et al., 2001). Participants had to have a Alcohol and Alcoholism, 2020 3 Downloaded from https://academic.oup.com/alcalc/advance-article/doi/10.1093/alcalc/agaa089/5911709 by guest on 29 September 2020

Fig. 1. Diagram of the simultaneous (neutral, alcohol and cigarette cues) CR procedure and the ET procedure and flow of participants. After pre-telephone screening and in-person screening, participants were assigned to either the CR or ET session. Both sessions were completed on the same day and were counterbalanced. (A) The CR session was comprised by four trials: two fixed-order interaction trials (T1 and T2) and two free-choice interaction trials (T3 and T4). After cue exposure, participants completed the ACQ and TCQ in a counterbalanced manner and were then directed to the ET laboratory. (B) The ET session was comprised by 160 trials. On each trial of the ET procedure, pictures of water, alcohol and cigarette cues were presented at locations above, below left and below right of the screen center. For 80 trials, participants were presented with the same water, alcohol and cigarette cues they interacted with during the CR task. The colored trace plots the participant’s eye position throughout an example trial. After completing the ET procedure, the participants were directed to the CR bar laboratory. breath alcohol concentration of 0.00 mg/L at each session. Individu- Questionnaire (TCQ-SF) (Heishman et al., 2003) were utilized to als with a self-reported diagnosis of a severe Substance Use Disorder assess craving in the bar laboratory. In order to ensure that the (SUD) for a substance other than alcohol, caffeine, marijuana or participants understood the task and test for cognitive and neurolog- nicotine were excluded from the study and if they had a history of ical impairments that could influence the response to the procedure, seizures in adulthood. For the ET session, individuals had to have we used a three-part explanation protocol that was standardized 20/20 normal vision or corrected to normal vision. across all participants. First, we verbally explained the task and re- iterated the instructions until we received verbal confirmation from the participant that they understood the task. Second, we showed Study procedures them how to perform the task by demonstrating the stimulus setup A diagram of the simultaneous (neutral, alcohol and cigarettes cues) (e.g. how to use the buttons to make their selection). Third, we CR and ET procedures are depicted in Fig. 1. The study consisted had participants perform a few example trials where they used the of the following phases: (a) telephone pre-screening, (b) signing of experimental setup themselves (e.g. for the ET procedure they put the consent document and in-person screening, (c) CR session in the their head in the chin rest and practiced the task) to ensure they felt bar laboratory at CAAS with real cues and (d) an ET session in comfortable using the setup. the virtual laboratory at CLPS with virtual cues. Both sessions were completed in the same day and were counterbalanced. Participants were asked if they prefer alcohol or cigarettes and their preferred The CR in the bar laboratory alcoholic beverage and cigarettes brand to prepare the sessions in the The CR was comprised of four trials: two fixed-order interaction laboratories. trials (T1–T2) and two free-choice interaction trials (T3–T4). The Assessments included: a demographic questionnaire and a time entire procedure took about 1 hour. Participants were presented life follow-back for alcohol consumption and cigarettes use during with three simultaneous real cues: neutral (water), favorite brand the 90 days prior to their first session (Pedersen et al., 2012); the of alcoholic drink and favorite cigarette brand for 2 minutes. They Structured Clinical Interview for Diagnostic and Statistical Manual were asked to interact with (smelling and touching) each substance of Mental Disorders, fifth Edition (DSM-5) (SCID-5) short form by following the direction of an audio recording. After exposure to questionnaire, Alcohol Dependence Scale (ADS) and Fagerstrom Test T1, the cues were removed and participants completed the ACQ for Nicotine Dependence (FTND) (Etter et al., 1999; Doyle and (Tiffany and Carter, 1998) and TCQ-SF (Heishman et al., 2003). Donovan, 2009; Shankman et al., 2018). Alcohol Craving Ques- After a 2-minute relaxation period to avoid carryover effect, a second tionnaire (ACQ) (Tiffany and Carter, 1998) and Tobacco Craving cue fixed-interaction trial was repeated for another 2-minute session 4 Alcohol and Alcoholism, 2020 to expose participants to each substance and administer the craving while in the ET procedure, participants performed a three-alternative questionnaires in a counterbalanced manner (Fig. 1A). forced-choice decision. The three cues were then reintroduced and participants proceeded Craving was assessed by repeated measures across the entire to the 2-minute free-interaction trial (T3–T4). They were asked to CR session (T1–T4). Since all the cues were administered simul- touch and smell any of the substances that they preferred and then taneously during each trial, two fix-order trials (T1–T2) and two they completed the ACQ and TCQ. This free-interaction trial was free-interaction trials (T3 and T4), it was not possible to capture also repeated twice, with a relaxation period in-between T3 and a difference in craving due to a neutral trial (water) compared to T4 (Monti et al., 1999). Dependent variables measured in the CR substance (alcohol and cigarettes) trials. Alcohol craving (ACQ) was

included: time spent interacting with each cue in each free-interaction compared to cigarettes craving (TCQ) within the same trial. Two Downloaded from https://academic.oup.com/alcalc/advance-article/doi/10.1093/alcalc/agaa089/5911709 by guest on 29 September 2020 trial (T3 and T4), proportion of time spent interacting with real questions from the ACQ were excluded to account for two questions substances, and alcohol craving and tobacco craving. All sessions that were not included in the TCQ when the questionnaires were were video recorded to accurately capture and quantify the time spent administered to participants in order to effectively correspond to interacting with each substance. data collected between the two measures. Two-way ANOVA was used to evaluate the effect of substance (alcohol or cigarettes) on craving according to trial (T1, T2, T3 and T4) and substance by The ET procedure trial interaction. Post hoc analyses were utilized to further assess the Participants sat at a desk with a computer monitor located 57 cm direction of the effect. Then, linear regressions were conducted to away and a keyboard placed in front of them. Eye position was determine the relationship between craving of each substance and measured with an Eyelink 1000 (SR Research, Ottawa, Ontario, total time spent with a substance (behavioral subjective response) Canada) infrared eye tracker at 1000 Hz. To ensure that head in the CR session. Statistical Package for the Social Sciences (v.24) movements did not interfere with our measurement of eye position, (Armonk, NY, U.S.) was used to conduct the analysis. participants sat with their forehead and chin braced against a chin For the ET procedure, participants were instructed to view three rest attached to the desk. Each trial consisted of a fixation display presented pictures and select one of them as if they were choosing and a stimulus display. The fixation display consisted of a gray from the actual cues in the bar laboratory. The images of the cues background and a white fixation cross located at the center of the were obtained from the actual substances utilized in the CR. Eye screen. After a variable period of 250–500 ms, the stimulus display position data were analyzed offline using custom Python scripts. Sac- appeared and remained on the screen until an item was chosen. Each cades were detected by using a velocity thresholding procedure. Eye picture was located 6◦ above, lower left or lower right and equidistant position data were first smoothed by subjecting 2D velocity scalars from screen center and spanned a maximum of 6◦ in height and 6◦ to a second-order low-pass Butterworth filter with a cutoff frequency in width. The ordering of picture locations was randomized in each of 100 Hz. The resulting velocity profile was then compared against trial (Fig. 1B). a threshold value to determine when saccades occurred throughout The ET procedure took about 1 hour and consisted of four each trial. Fixations were defined as time points where velocity did blocks of 40 trials for a total of 160 trials. For 80 trials (50%), not exceed the threshold. Saccade onsets were defined as the first we included pictures of the cues the participant interacted with time point after a fixation at which velocity exceeded the threshold. during the CR procedure along with a picture of water. Prior to Saccade offsets were defined as the final time point that velocity each block, participants completed a six-point eye tracker calibration exceeded the threshold before a fixation. Thresholds were chosen on routine to ensure accurate measurement of eye position. On each trial, a trial by trial basis via visual inspection of the data to ensure accuracy participants performed a three-alternative forced-choice decision. of saccade detection. Participants were instructed to view the three pictures and select one To measure the time spent viewing each item, we offline defined a of them as if they were choosing from the actual items. Participants circular region of the screen 4◦ from the center of each picture loca- chose an item by pressing a button corresponding to its location on tion and calculated the total time in the trial the participant fixated the screen. Eye position was measured during the time the stimulus within each region. We then averaged the total time spent viewing display was presented. Dependent variables measured in the ET each picture category across all trials. We report time measures for ET session included: time spent viewing each item and the proportion analyses in milliseconds (ms) or proportion of time (0–1) as described of time spent fixating on the virtual substances. in the analysis for the CR data. We conducted repeated measures ANOVAs. Analysis for the ET procedure was conducted using the R statistical programming language (Venables and Smith, 2013) and Data analytic strategy the ggplot2 graphing package (Wickham, 2016). For the CR procedure, data were analyzed to assure normal dis- For both CR and ET analyses, Cohen’s d (d) was used to report tribution. Means were calculated and compared across the three effect size. Results were expressed as means (M) and standard error substances in the two different sessions. One-way analysis of variance of the means. All statistical tests were two-sided, and statistical (ANOVA) with repeated measures was used to evaluate the time spent significance was accepted if P < 0.05 was obtained. Follow-up interacting with each cue in each free-interaction trial (T3 and T4) in Bonferroni tests were used to correct for multiple comparisons and the CR session. Post hoc analyses were utilized to further assess the GraphPad Prism (v.7) (San Diego, CA) was used to generate the direction of the effect. figures. To integrate the time spent interacting with each substance in the CR procedure (real cues) with the time fixating on each substance RESULTS in the ET procedure (virtual cues), we used the proportion of time (ranging from 0 to 1) spent with each substance. This approach was Participant characteristics used because during the free-interaction CR procedure, participants Of 51 individuals pre-screened by phone, 32 signed the informed con- were not always engaged with the substances using their hands, sent, were screened in person and found to be eligible to participate. Alcohol and Alcoholism, 2020 5

Table 1. Demographic and baseline characteristics

Participant characteristics M ± SD or n (%)

Session: CAAS first 17 (53.1) Male 13 (40.6) Age 45.4 ± 11.8 Race: White; African American; Others 13 (40.6); 14 (43.8); 5 (15.6) Strong alcohol craving, last year (SCID-5) 19 (15.4)

Heavy drinkers 30 (93.8) Downloaded from https://academic.oup.com/alcalc/advance-article/doi/10.1093/alcalc/agaa089/5911709 by guest on 29 September 2020 Drinks per week (Total; males; females) 35.8 ± 18.2; 29.54 ± 16.2; 40.0 ± 18.7 Cigarettes per week (Total; males; females) 116.6 ± 86.3; 119 ± 65.9; 115.0 ± 99.6 Alcohol dependence scale 10.0 ± 7.0 Fagerstrom test for nicotine dependence 5.5 ± 1.9 Self-reported preferred substance as cigarette 25 (78.1)

These 32 individuals were assigned in a counterbalanced manner This was true across subjective response as stratifying by preferred to the CR and the ET session (n = 17 to CR first, n = 15 to ET substance showed no significant difference (P > 0.05). first) and all participants completed both sessions on the same day. There were no significant differences in demographic information Integration of real cues and virtual cues: time in the cue reactivity and between the group that started with the CR session first and the eye tracking laboratory The proportion of time spent interacting with group that started with the ET session first (P > 0.05). Participants real substances in the bar laboratory and the proportion of time spent all drank alcohol heavily according to CDC criteria and most smoked fixating on the virtual substances in the ET laboratory is depicted in moderately (44%) to heavily (38%). The majority of participants Fig. 2C. disclosed cigarettes as their self-reported preferred substance (78%) On average at T3, participants interacted with alcohol (28% ± compared to alcohol. Baseline characteristics of eligible participants 5.0%) more than with cigarettes (25% ± 4.6%) or water who completed both sessions are described in Table 1 and a flow (6.7% ± 1.7%). When we analyzed the portion of time spent chart during the CR and ET sessions is depicted in Fig. 1. interacting with either alcohol or cigarettes, there was a significant ∗∗∗ < effect on substance (F93 = 8.172, P 0.001). Post-hoc analysis revealed that there was a significant difference between the time spent ∗∗∗ < Primary aim: integration of alcohol and cigarettes cues interacting with alcohol (t31 = 3.739; P 0.001) and cigarettes (t = 3.202; P < 0.01) when compared to the proportion of time Real cues: time interacting with alcohol and cigarettes in the 31 spent interacting with water. There was no significant difference bar laboratory Time spent interacting with each substance in the between the proportion of time spent with alcohol (P > 0.05) when free-interaction trials (T3–T4) in the CR session is described in compared to the proportion of time spent with cigarettes. Fig. 2A. There was a main effect of interacting with each substance ∗∗∗ < On average during T4, participants interacted with cigarettes (F93 = 14.878, P 0.001, d = 0.498), but not a main effect (29% ± 4.1%) more than with alcohol (22.8% ± 3.1%) or water of trial (T3 vs T4) and there was not a significant substance by (9.7% ± 2.6%). When we analyzed the portion of time spent inter- trial interaction (P’s > 0.5). Post hoc analyses revealed that at T3, acting with either alcohol or cigarettes, there was a significant effect there was a significant difference between the time spent interacting ∗∗∗ < ∗∗∗ < on substance (F93 = 8.737, P 0.001). Post-hoc analysis revealed with alcohol (t31 = 4.147; P 0.001; d = 0.733) and cigarettes ∗∗∗ < that there was a significant difference between the proportion of time (t31 = 4.002; P 0.001; d = 0.707) when compared to time spent interacting with alcohol (t = 2.786; ∗P < 0.05) and cigarettes spent interacting with water. At T4, post hoc analyses showed a 31 (t = 4.092; ∗∗∗P < 0.001) when compared to the proportion of significant difference between the time spent interacting with alcohol 31 ∗∗ < time spent interacting with water. There was no significant difference (t31 = 3.452; P 0.002; d = 0.610) and cigarettes (t31 = 3.844; between the proportion of time spent with alcohol (P > 0.05) when ∗∗∗P < 0.001; d = 0.680) compared to time spent interacting with compared to the proportion of time spent with cigarettes. water. For both T3 and T4, there was no significant difference In the ET procedure, participants on average chose the cigarette between alcohol and cigarettes (P > 0.5). This was also confirmed cue (38.9% ± 2.5%) more than the alcohol (37.2% ± 2.5%) or across subjective response as stratifying by preferred substance shows neutral cues (21.5% ± 2.4%). When we analyzed the proportion of no significant difference (P > 0.05). time spent fixating either alcohol or cigarettes, there was a significant ∗∗∗ < effect on substance (F93 = 15.14, P 0.001). Post-hoc analysis Virtual cues: selective attention to alcohol and cigarettes cues revealed that there was a significant difference between the time ∗ < in the eye tracking laboratory Time fixating on each substance spent interacting with alcohol (t31 = 4.48; P 0.05) and cigarettes ∗∗∗ < in the ET session is described in Fig. 2B. We excluded trials (t31 = 4.96; P 0.001) when compared to time spent interacting where the participant blinked or looked away from the computer with water. There was no significant difference between the time screen (12.2% ± 2.4%) or trials at which no item was viewed spent with alcohol when compared to cigarettes (P > 0.05). (6.2% ± 2.1%). There was a difference in average time spent viewing ∗∗∗ < each substance (F84 = 14.596, P 0.001, d = 0.258). A post- hoc analysis showed that time spent looking at alcohol (t28 = 4.36, Exploratory aims: craving outcomes ∗∗∗ < ∗∗∗ < P 0.001) and at cigarettes (t28 = 4.41, P 0.001) was Results of the craving outcomes are depicted in Fig. 3. A substance significantly greater than time spent looking at water. There was (alcohol and cigarettes) × trial (T1–T4) ANOVA showed that > ∗∗∗ < no significant difference between alcohol and cigarettes (P 0.05). there was a main effect of substance (F1 = 16.753, P 0.001, 6 Alcohol and Alcoholism, 2020 Downloaded from https://academic.oup.com/alcalc/advance-article/doi/10.1093/alcalc/agaa089/5911709 by guest on 29 September 2020

Fig. 2. Time spent interacting with each substance in the CR and ET . (A) In the CR session, there was a main effect of interacting with each substance (F 93 = 14.878, ∗∗∗p < 0.001, d = 0.498), but not a main effect of trial (T3 vs T4) and no substance by trial interaction (∗p’s > 0.5). Post hoc analyses revealed that at both at T3 and T4 there was a significant difference between the time spent interacting with alcohol and cigarettes (∗∗∗p’s < 0.001) when compared to time spent interacting with water. However, there was no significant difference between the time spent interacting with each addictive substance at T3 and T4 (∗p’s > 0.5). ∗∗∗ (B) In the ET session, there was a difference in time fixating each substance (F 84 = 14.596, p < 0.001, d = 0.258). Post hoc analysis showed that while there was a statistical difference between alcohol and cigarettes compared to water (∗∗∗p’s < 0.001), there was no significant difference between the time spent fixating ∗ ∗∗∗ each addictive substance ( p > 0.05). (C) In the CR procedure, there was a significant effect of substance both during T3 (F 93 = 8.172, p < 0.001) and T4 ∗∗∗ ∗∗∗ (F 93 = 8.737, p < 0.001). Similarly, in the ET procedure there was a significant effect of substance (F 93 = 15.14, p < 0.001). In both the CR and ET procedures, post hoc analysis showed that while there was a statistical difference between alcohol and cigarettes compared to water (∗p < 0.05-∗∗∗p < 0.001), there was no significant difference between the time spent interacting with each addictive substance (p > 0.05). Results are expressed as the M ± SEM of the proportion of time, (∗p < 0.05, ∗∗p < 0.01, ∗∗∗p < 0.001).

∗∗∗ < d = 0.213) and a main effect of trial (F3 = 6.676, P 0.001, and positively correlated with alcohol craving (T3: r29 = 0.367, ∗ ∗∗ d = 0.097). Substance by time interaction was not significant P < 0.05; T4: r29 = 0.468, P < 0.01; Fig. 3B–C), as well as with ∗ < (F3 = 0.885, P = 0.450, d = 0.258) (Fig. 3A). cigarettes craving (T3: r29 = 0.4189, P 0.05; T4: r29 = 0.459, The relationship between time spent interacting with alcohol and ∗P < 0.05, Fig. 3D–E). Time spent interacting with cigarettes during craving during the CR procedure is depicted in Fig. 3B–E. Time T3 and T4 was not significantly correlated with alcohol or cigarettes spent interacting with alcohol during T3 and T4 was significantly craving (P’s > 0.05). Alcohol and Alcoholism, 2020 7 Downloaded from https://academic.oup.com/alcalc/advance-article/doi/10.1093/alcalc/agaa089/5911709 by guest on 29 September 2020

Fig. 3. Time spent interacting with each substance in the CR and ET . (A) There was a main effect of substance (F 1 = 16.753, p < 0.001, d = 0.213) and a main ∗∗∗ ∗ effect of trial (F 3 = 6.676, p < 0.001, d < 0.097). However, the substance by time interaction was not significant ( p < 0.05). Pairwise comparisons revealed that cravings for cigarettes were significantly greater than cravings for alcohol in all trials (∗∗∗p’s < 0.001). Time spent interacting with alcohol during (B) T3 ∗ (r29 = 0.367, p < 0.05) and (C) T4 (r29 = 0.468, p < 0.01) was significantly correlated with alcohol craving. Time spent interacting with alcohol during (D) T3 ∗ ∗ (r29 = 0.4189, p < 0.05) and (E) T4 (r29 = 0.459, p < 0.05) was significantly correlated with cigarette craving. Time spent interacting with cigarettes during T4 was not significantly correlated with alcohol or cigarette craving (p’s > 0.05). Results are expressed as the M ± SEM, (∗p < 0.05, ∗∗p < 0.01, ∗∗∗p < 0.001).

DISCUSSION a virtual environment to study the hierarchical value of multiple cues (alcohol and cigarettes) to predict the behavioral response, The findings of this study align with previous research confirming measured by time spent interacting with each substance. Consistent that in individuals with SUD, there is a preference for substance- with our hypothesis, the results obtained in the CR sessions were related cues over neutral cues (Carter and Tiffany, 1999). This was also validated in the ET sessions, as participants fixated on alcohol observed in the CR sessions of our study where more time was and cigarettes for a greater amount of time compared to their spent interacting with alcohol and cigarettes compared to time spent fixation on water. Data collected through the integrated CR and interacting with water during both T3 and T4. In this study, in ET paradigm showed that measures of behavioral interaction in an attempt to improve ecological validity, we have incorporated the CR session predicted the measure of selective attention in the 8 Alcohol and Alcoholism, 2020

ET session and vice versa, as the sessions were counterbalanced decision making in this population. For example, the prefrontal cor- for this study. The measure of attention towards alcohol and tex and the primary motor cortex have been linked to the attentional cigarettes compared to water in the ET session was consistent with bias related to smoking-related cues (Kang et al., 2012). Interestingly, the behavioral response across the two free-interaction trials in the orbitofrontal cortex, the insula and the temporal gyrus have been the CR session. linked to craving elicited by smoking-related cues (Kang et al., 2012). Linking a CR procedure with an ET procedure is helpful in This study should be evaluated based on its strengths and limi- combining the decision-making process (time interaction with sub- tations. The strengths include: all individuals drank alcohol heavily stance) to selective attention (bias to substance). The ET procedure and smoked cigarettes moderately to heavily, the CR was performed

allowed us to test and validate the effect of more than one cue by under a well-controlled paradigm in a bar laboratory to ensure a Downloaded from https://academic.oup.com/alcalc/advance-article/doi/10.1093/alcalc/agaa089/5911709 by guest on 29 September 2020 using an objective approach that measures attentional bias. The ET more naturalistic environment, and the two sessions (CR and ET) procedure has provided us with an objective indicator of attention were performed the same day. Limitations include the small sample and cognitive processing during choice (Shimojo et al., 2003; Kra- size, the different smoking groups (light, moderate and heavy) and jbich et al., 2010) and acted as support when deciphering preference the lack of administration of the craving questionnaire in the ET between substance-related and neutral stimuli in the CR session. The procedure. temporal pattern of gaze behavior in a virtual environment allowed At a behavioral pharmacology level, simultaneous CR paradigms us to determine attention directed to substances compared to neutral can elucidate mechanisms by which alcohol and cigarette potentiates stimuli. Furthermore, using a virtual environment (ET), incorporated addictive-related behaviors since it is well-known that there is a with a natural environment (CR in the bar laboratory), allowed us to shared reward pathway involving the nicotinic acetylcholine and the examine the relationship between measures of visual orienting and mesolimbic dopamine systems. A simultaneous CR procedure in a the motivational valence of each cue presented simultaneously. human laboratory also may inform clinical treatment outcomes in The ET dual cues paradigm supports the primary findings of the individuals who drink heavily and also smoke. In fact, treatment for CR study, which provides promising evidence suggesting that selec- smoking cessation can be less successful if individuals who drink tive attention toward alcohol and/or cigarettes leads to an increased and smoke are exposed to alcohol cues (Oliver and Drobes, 2015). interaction with these substances in the bar laboratory in individuals A larger study should be developed to gain further understanding who drink and smoke cigarettes. of drivers of craving in people who drink and smoke to help tailor The results of this study begin to show that time spent with more specified interventions since tendency to selectively attend to alcohol is a greater factor for craving of both alcohol and cigarettes substances can make recovery that much more difficult. than time spent with cigarettes. This is further compounded by the Finally, currently there are several tools available to measure fact that time spent with cigarettes did not correlate with craving for craving, but no single approach represents a ‘state-of-the-art’ instru- alcohol or cigarettes consistently in either of the two free-interaction ment to measure and capture craving from both the psycholog- trials. These data are in line with a human laboratory study, which ical and biological perspectives (Kavanagh et al., 2013). Craving showed that acute alcohol administration was able to limit the assessments include both subjective measures (questionnaires) and ability to resist smoking in individuals who drink heavily and smoke objective parameters (autonomic responses) with limited validity moderately to heavily (Kahler et al., 2014), the same population (Haass-Koffler et al., 2014). With this work, we could develop and analyzed here. test more objective tools that can assess attention to multiple cues in Future studies utilizing a larger sample are necessary to repli- order to develop and validate specific CR paradigms for comorbid cate these findings and create a better representation of individu- populations that may have stronger experimental and clinical value. als who use alcohol and cigarettes. More specifically, the sample should be refined to include more participants who report alcohol ACKNOWLEDGEMENTS as their preferred substance (as most participants listed cigarettes as their self-reported preferred substance). This could have been The authors thank Bianca Persoud for her support in the data analysis an important factor concerning how participants engaged with one and Zoe E. Brown for editing the manuscript. The images of alcohol substance as compared to the other during both the CR and ET and cigarettes for the manuscript were obtained from ClipArtMag. sessions. Individual differences on variables such as “preference” com (free clip art). may have skewed the results of a subjective measure associated with craving (Sayette et al., 2000). Previous studies have stated that alcohol cues during CR can FUNDING elicit craving for alcohol in individuals with AUD (Witteman et al., This study was supported by the Rohsenow Pilot Award (PI: Haass-Koffler). 2015). This is seen in our study as time spent with alcohol in both Dr. Haass-Koffler is supported by the National Institute on Alcohol Abuse free-interaction trials were significantly correlated with craving for and Alcoholism (K01 AA023867; R01 AA026589; R01 AA027760; R21 alcohol. Additionally, we found that there was a correlation between AA027614). Dr. Aston is supported by the National Institute on Drug Abuse time spent with alcohol and craving for cigarettes. This observation (K01 DA0239311). Drs. Haass-Koffler and Aston are also supported by the is consistent with previous work that showed that alcohol cues elicit National Institute of General Medical Sciences (NIGMS), Center of Biomedical smoking craving in individuals with AUD who also smoke (Gulliver Research Excellence (COBRE, P20 GM130414). Dr. Joo-Hyun Song and James Wilmott are partially supported by National Science Foundation (BCS et al., 1995; Rohsenow et al., 1997; McGrath et al., 2015). This 1555006). James Wilmott is also supported by the National Eye Institute (T32 suggests that exposure to alcohol may be a driving force for craving EY018080). of both alcohol and cigarettes. This study could be further improved by administering alcohol and smelling cigarettes smoke or lighting a cigarette during the CR sessions, as olfactory cues can elicit strong CONFLICT OF INTEREST craving responses. Also, an ET combined with functional magnetic The authors listed above have no conflict of interest related to this work. resonance imaging may highlight brain regions that are affected by There are no relationships between financial supporters or other organizations Alcohol and Alcoholism, 2020 9 and these authors that could have influenced the study and results. The data Kahler CW, Metrik J, Spillane NS et al. (2012) Sex differences in stimulus that support the findings of this study are available on request from the expectancy and pharmacologic effects of a moderate dose of alcohol on corresponding author (CLH-K). smoking lapse risk in a laboratory analogue study. Psychopharmacology (Berl) 222:71–80. Kang OS,Chang DS,Jahng GH et al. (2012) Individual differences in smoking- AUTHORS CONTRIBUTION related cue reactivity in smokers: an eye-tracking and fMRI study. Prog Neuropsychopharmacol Biol Psychiatry 38:285–93. C.L.H.-K. and R.D.S. designed the alcohol laboratory study; J.W. and J.-H.S. Kavanagh DJ, Statham DJ, Feeney GF et al. (2013) Measurement of alcohol designed the ET study; C.L.H.-K. provided funding used to conduct both craving. Addict Behav 38:1572–84. studies. R.D.S. and J.W.conducted the study,including collection of the research Kenna GA, Haass-Koffler CL, Zywiak WH et al. (2016) Role of the alpha1 Downloaded from https://academic.oup.com/alcalc/advance-article/doi/10.1093/alcalc/agaa089/5911709 by guest on 29 September 2020 data. C.L.H.-K., R.D.S. and J.W. conducted all the analyses. C.L.H.-K., J.- blocker doxazosin in alcoholism: a proof-of-concept randomized con- H.S., J.P.W. and R.D.S. participated in the interpretation of data for important trolled trial. Addict Biol 21:904–14. intellectual contents. E.R.A. contributed through questionnaire design and data Krajbich I, Armel C, Rangel A (2010) Visual fixations and the compu- interpretation. R.D.S. and J.W. wrote the first draft of the manuscript. C.L.H.- tation and comparison of value in simple choice. Nat Neurosci 13: K., J.-H.S. and R.D.S. contributed to the writing of the manuscript. All authors 1292–8. approved the final version of the manuscript. Larson J, Sieprawska D (2002) Automated study of simultaneous-cue olfactory discrimination learning in adult mice. Behav Neurosci 116:588–99. REFERENCES Laude JR, Fillmore MT (2015) Alcohol cues impair learning inhibitory signals in beer drinkers. Alcohol Clin Exp Res 39:880–6. Bardo MT, Bevins RA (2000) Conditioned place preference: what does it add Lucke-Wold B (2011) The varied uses of conditioned place preference in behav- to our preclinical understanding of drug reward? Psychopharmacology ioral neuroscience research: an investigation of alcohol administration in (Berl) 153:31–43. model organisms. Impulse (Columbia) 2011. Berg CJ, Haardorfer R, Schauer G et al. (2016) Reasons for polytobacco use Mcgrath DS, Peloquin MP, Ferdinand JC et al. (2015) Acute effects of nicotine among young adults: scale development and validation. Tob Prev Cessat on alcohol cue-reactivity in nondependent and dependent smokers. Exp 2. Clin Psychopharmacol 23:29–36. Carter BL, Tiffany ST (1999) Meta-analysis of cue-reactivity in addiction Meernik C, Jarman K, Wright ST et al. (2016) Eye tracking outcomes in research. Addiction 94:327–40. tobacco control regulation and communication: a systematic review. Tob Colby SM, Rohsenow DJ, Monti PM et al. (2004) Effects of tobacco depriva- Regul Sci 2:377–403. tion on alcohol cue reactivity and drinking among young adults. Addict Monti PM, Rohsenow DJ, Hutchison KE et al. (1999) Naltrexone’s effect on Behav 29:879–92. cue-elicited craving among alcoholics in treatment. Alcohol Clin Exp Res Desalvo KB (2016) Public health 3.0: applying the 2015–2020 dietary guide- 23:1386–94. lines for Americans. Public Health Rep 131:518–21. Moskowitz DS, Young SN (2006) Ecological momentary assessment: what it Doyle SR, Donovan DM (2009) A validation study of the alcohol dependence is and why it is a method of the future in clinical psychopharmacology. J scale. J Stud Alcohol Drugs 70:689–99. Psychiatry Neurosci 31:13–20. Drobes DJ (2002) Cue reactivity in alcohol and tobacco dependence. Alcohol Okuyemi KS, Ahluwalia JS, Richter KP et al. (2001) Differences among African Clin Exp Res 26:1928–9. American light, moderate, and heavy smokers. Nicotine Tob Res 3:45–50. Etter JF, Duc TV, Perneger TV (1999) Validity of the Fagerstrom test for nico- Oliver JA, Drobes DJ (2015) Cognitive manifestations of drinking-smoking tine dependence and of the heaviness of smoking index among relatively associations: preliminary findings with a cross-primed Stroop task. Drug light smokers. Addiction 94:269–81. Alcohol Depend 147:81–8. Fox HC, Anderson GM, Tuit K et al. (2012) Prazosin effects on stress- and Pedersen ER, Grow J, Duncan S et al. (2012) Concurrent validity of an online cue-induced craving and stress response in alcohol-dependent individuals: version of the timeline followback assessment. Psychol Addict Behav preliminary findings. Alcohol Clin Exp Res 36:351–60. 26:672–7. Grant BF, Hasin DS, Chou SP et al. (2004) Nicotine dependence and psychiatric Piasecki TM, Jahng S, Wood PK et al. (2011) The subjective effects of alcohol- disorders in the United States: results from the national epidemiologic tobacco co-use: an ecological momentary assessment investigation. J survey on alcohol and related conditions. Arch Gen Psychiatry 61: Abnorm Psychol 120:557–71. 1107–15. Rohsenow DJ, Monti PM, Colby SM et al. (1997) Effects of alcohol cues on Gulliver SB, Rohsenow DJ, Colby SM et al. (1995) Interrelationship of smoking urges and topography among alcoholic men. Alcohol Clin Exp smoking and alcohol dependence, use and urges to use. J Stud Alcohol 56: Res 21:101–7. 202–6. Sayette MA, Shiffman S, Tiffany ST et al. (2000) The measurement of drug Haass-Koffler CL, Goodyear K, Zywiak WH et al. (2017) Higher pretreatment craving. Addiction 95:S189–210. blood pressure is associated with greater alcohol drinking reduction Shankman SA, Funkhouser CJ, Klein DN et al. (2018) Reliability and validity in alcohol-dependent individuals treated with doxazosin. Drug Alcohol of severity dimensions of psychopathology assessed using the structured Depend 177:23–8. clinical interview for DSM-5 (SCID). Int J Methods Psychiatr Res 27. Haass-Koffler CL, Leggio L, Davidson D et al. (2015) Effects of idazoxan Shimojo S, Simion C, Shimojo E et al. (2003) Gaze bias both reflects and on alcohol pharmacokinetics and intoxication: a preliminary human influences preference. Nat Neurosci 6:1317–22. laboratory study. Alcohol Clin Exp Res 39:594–602. Thomas S, Bacon AK, Sinha R et al. (2012) Clinical laboratory stressors used Haass-Koffler CL, Leggio L, Kenna GA (2014) Pharmacological approaches to study alcohol-stress relationships. Alcohol Res 34:459–67. to reducing craving in patients with alcohol use disorders. CNS Drugs Tiffany ST, Carter BL (1998) Is craving the source of compulsive drug use? J 28:343–60. Psychopharmacol 12:23–30. Heishman SJ, Singleton EG, Moolchan ET (2003) Tobacco craving question- Venables W, Smith D, the R Core Team (2013) An Introduction to R, Vol. naire: Reliability and validity of a new multifactorial instrument. Nicotine 34. Notes on R: A Programming Environment for Data Analysis and Tob Res 5:645–54. Graphics. Version 4.0.2 (2020–06–22). Kahler CW, Metrik J, Spillane NS et al. (2014) Acute effects of low and high Verplaetse TL, Mckee SA (2017) An overview of alcohol and tobacco/nico- dose alcohol on smoking lapse behavior in a laboratory analogue task. tine interactions in the human laboratory. Am J Drug Alcohol Abuse Psychopharmacology (Berl) 231:4649–57. 43:186–96. 10 Alcohol and Alcoholism, 2020

Wickham H (2016) Ggplot2: Elegant Graphics for Data Analysis. Switzerland: bined laboratory and field study. Psychopharmacology (Berl) 232: Springer. 3685–96. Witteman J, Post H, Tarvainen M et al. (2015) Cue reactivity and Zhao F, Schnotz W, Wagner I et al. (2014) Eye tracking indicators of reading its relation to craving and relapse in alcohol dependence: a com- approaches in text-picture comprehension. Frontline Learn Res 2:46–66. Downloaded from https://academic.oup.com/alcalc/advance-article/doi/10.1093/alcalc/agaa089/5911709 by guest on 29 September 2020