Authors: Dylan E. Kirsch (1Department of Psychology, University of California, Los Angeles, Los Angeles, CA 90095, USA), Erica N. Grodin (1Department of Psychology, University of California, Los Angeles, Los Angeles, CA 90095, USA; 2Department of Psychiatry and Biobehavioral Sciences, University of California Los Angeles, Los Angeles, CA 90095, USA), Lara A. Ray (1Department of Psychology, University of California, Los Angeles, Los Angeles, CA 90095, USA; 2Department of Psychiatry and Biobehavioral Sciences, University of California Los Angeles, Los Angeles, CA 90095, USA)
Categories: Article, Alcohol use disorder, Cue reactivity, Craving, Pharmacotherapy, Alcohol drinking
Source: Addictive behaviors
Authors: Dylan E. Kirsch, Erica N. Grodin, Lara A. Ray
Exposure to alcohol-related cues is thought to elicit a conditional response characterized by increased craving in individuals with alcohol use disorder (AUD). In the context of AUD research, it is important to consider that not all individuals with an AUD are cue reactive. This study systematically examined subjective alcohol cue reactivity and its clinical and drinking correlates in individuals with an AUD enrolled in a human laboratory pharmacotherapy trial.
Individuals with current moderate-to-severe AUD (N=52) completed a standard alcohol cue exposure paradigm and individual difference assessments as part of a human laboratory pharmacotherapy trial (NCT04249882). We classified participants as cue reactive (CR+) and cue non-reactive (CR−), as indicated by self-reported, subjective alcohol urge, and examined group differences in baseline clinical characteristics and drinking outcomes over the course of the trial.
Twenty participants (38%) were identified as CR+, while 32 participants (62%) were identified as CR−. The CR+ and CR− groups did not differ in baseline drinking and AUD clinical characteristics, but the groups differed in race composition (p=0.02) and smoking prevalence (p=0.04) such that the CR+ group had lower prevalence of smokers. The CR+, versus the CR−, group drank more during the titration period (p=0.03). Both groups reduced drinking across the trial (p’s<0.001), but the CR+ group exhibited a smaller reduction in drinking, compared with the CR− group (time × group, p=0.029; CR−, p<0.0001; CR+: p=0.01).
Results indicate that cue reactivity is a heterogenous construct. Recognizing this heterogeneity, and the clinical factors associated with it, is critical to advancing this paradigm as an early efficacy marker in AUD research.
Alcohol use disorder (AUD) is a chronic relapsing condition characterized by an impaired ability to stop or control use despite clinically significant impairment, distress, and/or adverse consequences [1, 2]. In individuals with an AUD, exposure to alcohol-related cues, such as the sight, smell, or taste of alcohol, is thought to elicit a conditional response characterized by heightened subjective cravings/urges and psychophysiological response (i.e., salivation, heart rate, blood pressure) [3, 4]. This phenomenon is known as alcohol cue-induced craving, or cue reactivity, and has been associated with AUD severity and drinking outcomes [5–9]. Given the ubiquity of alcohol cues in the real world, understanding and managing reactivity to alcohol-related cues is critical for the treatment of AUD. To that end, alcohol cue reactivity has been a focal point of AUD research for decades [4, 10–12].
Alcohol cue exposure paradigms are widely used in experimental laboratory studies to examine alcohol cue reactivity and to model high-risk situations that may lead to relapse [4, 13, 14]. These paradigms involve systematically exposing individuals to alcohol-related and control (i.e., water, juice) cues and assessing their “reactivity,” or their subjective urge to drink alcohol, and in some studies, psychophysiological response [3, 4, 15]. These paradigms have also been tailored to explore factors influencing cue reactivity, such as negative mood and stress [9, 16, 17]. Alcohol cue reactivity is commonly used as an endpoint in human laboratory medication trials to establish initial efficacy for promising pharmacotherapies for AUD [18–20]. Medication-related reductions in alcohol cue reactivity are thought to signal early efficacy and inform whether a medication should advance to the next stage of clinical testing [18–21]. Despite the widespread utilization of cue exposure paradigms to screen pharmacotherapies for AUD [14], the degree to which it is actually predictive of clinical efficacy remains unclear [18, 22, 23]. As recently described in a systematic review and quantitative synthesis of the literature [12], there are a number of factors in cue reactivity testing that introduce heterogeneity, which may ultimately influence the translational and predictive utility of this paradigm.
One central consideration is that not all individuals with an AUD are cue reactive [7, 16, 17]. A handful of studies revealed that ~22–65% of individuals with an AUD are cue reactive, as indicated by subjective cue reactivity [11, 15, 16, 19, 24]. Despite these findings, the application of alcohol cue exposure paradigms often assumes uniform cue reactivity among individuals with an AUD (i.e., all individuals with AUD uniformly exhibit greater reactivity following exposure to alcohol, compared with control, cues) [5]. This assumption may be problematic, particularly in the context of treatment studies that assess changes in cue reactivity before and after treatment. If individuals are not cue reactive to begin with, the study will lack power to determine whether a medication can reduce cue reactivity. Although not widely implemented, some studies have screened for cue reactivity prior to enrollment in order to minimize heterogeneity and enhance signal detection [16, 25]. These screening procedures typically involve an abbreviated cue reactivity paradigm and consider individuals cue reactive if they exhibit a ≥1 standard deviation increase in subjective alcohol craving when exposed to alcohol, compared with control, cues [11, 22, 25]. This approach may produce a stronger signal for investigating treatment effects, but limits generalizability of findings to individuals with AUD who are cue reactive. Therefore, it is critical to understand differences in the clinical presentation of individuals with AUD based on their alcohol cue reactivity.
The present study characterized alcohol cue reactive (CR+) and cue non-reactive (CR−) individuals with current moderate-to-severe AUD by comparing them on clinical variables and drinking outcomes during a quit attempt. This study leveraged data from a recent human laboratory pharmacotherapy trial (NCT04249882) involving two weeks of treatment with varenicline, naltrexone, or placebo [26]. Randomized participants completed a standard alcohol cue exposure paradigm, followed by a 1-week medication titration period and 1-week “practice quit” attempt. The objectives of the present study were 1) identify CR+ and CR− groups, as determined via a self-reported, subjective assessment of alcohol urge, during an alcohol cue-reactivity paradigm, and their clinical correlates (i.e., alcohol use/problems and demographic, smoking, and mood characteristics); and 2) determine how CR+ and CR− individuals differ in drinking over the course of the study. Based on research suggesting alcohol cue-induced craving is an important mechanism underlying alcohol use disorders (including alcohol use and relapse) [27], we hypothesized that CR+ individuals would 1) report heavier drinking and higher levels of AUD severity compared with CR− individuals and 2) report more drinks per drinking day and fewer percent days abstinent during the trial compared with CR− individuals.
The current study is a secondary analysis of a randomized, double-blind, placebo-controlled pharmacotherapy trial of 53 individuals with current moderate or severe AUD randomized to receive varenicline (VAR; 2 mg; n=19), oral naltrexone (NTX; 50 mg, n=15), or matched placebo (n=19) for two weeks (NCT04249882). All participants self-reported motivation to reduce or quit drinking. Participants were recruited from the community between February 2020 and January 2022 through online and newspaper advertisements, campaigns in mass transit in the greater Los Angeles metropolitan area, and targeted recruitment through a UCLA Addiction’s Laboratory database of previous study participants who agreed to be contacted for future studies. Study procedures were approved by the Institutional Review Board at UCLA. All participants provided written informed consent after receiving a full explanation of study procedures and discussed study medications with a physician.
Initial screening procedures were conducted via telephone interview. Following telephone screening, eligible participants were invited for an in-person assessment. Eligible participants (1) were 21 – 65 years old; (2) had a current (past 12-months) DSM-5 moderate-to-severe AUD; (3) reported drinking >7 drinks/week for women and >14 drinks/week for men in the 28-days prior to screening visit; (4) had self-reported intrinsic motivation to reduce or quit drinking within the next 6-months; and (5) had reliable internet access to complete electronic daily diary assessments. Exclusion criteria (1) current (past 12-months) DSM-5 diagnosis of substance use disorder for any psychoactive substances other than alcohol or nicotine; (2) lifetime DSM-5 diagnosis of schizophrenia, bipolar disorder, or any psychotic disorder; (3) positive urine screen for drugs excluding cannabis; (4) clinically significant alcohol withdrawal symptoms as indicated by a score ≥ 10 on the Clinical Institute Withdrawal Assessment for Alcohol-Revised (CIWA-Ar); (5) intense fear of needles or any adverse reaction to needle puncture; (6) pregnancy, nursing, planning to become pregnant while taking part in the study, or refusal to use reliable method of birth control (if female); (7) medical condition thought to interfere with safe study participation; and (8) current use of psychotropic medications that would interfere with safe study participation. Additional details of study procedures are described elsewhere [26].
On Day 1 of the study, randomized participants completed a standard alcohol cue exposure paradigm and battery of clinician-administered and self-report assessments. All assessments were completed prior to ingesting the first dose of study medication. Participants then began a week-long medication titration (Days 1–7) and returned to the lab for an in-person visit on Day 8 before beginning a week-long quit attempt (Days 8 – 14). On the final day (Day 14), participants returned to the lab for their last in-person visit. Participants were required to have a breath alcohol concentration of 0.00 g/dl and test negative on a urine toxicology screen for all drugs (except cannabis) at each in-person visit.
Alcohol use/problems and individual differences were measured using a battery of clinician-administered and self-report assessments. All assessments were completed prior to ingesting the first dose of study medication. Clinician administered assessments included the (1) Structured Clinical Interview for DSM-5 (SCID-5) [28], (2) Clinical Institute Withdrawal Assessment for Alcohol Scale – Revised (CIWA-Ar) [29], and (3) 30-day Timeline Followback (TLFB) Interview [30] for alcohol use. The TLFB was also administered on study Day 1, Day 8, and Day 14 to obtain daily reports of alcohol use between screening and randomization and throughout the trial. Drinks per drinking day (DPDD) and percentage of days abstinent (PDA) during each period were calculated. Drinking variables were selected because they were the primary drinking outcomes in the parent study [26].
Self-report measures included the (1) Alcohol Dependence Scale (ADS) [31], which measures severity of alcohol use problems; (2) Penn Alcohol Craving Scale (PACS) [32], which measures alcohol craving during the past week; (3) Obsessive Compulsive Drinking Scale (OCDS) [33], which also measures alcohol craving during the past week; (4) Readiness to Change (RTC) Ladder [34], which measures motivation to reduce or cut back on drinking; (5) UCLA Reward Relief Habit Drinking Scale (RRHDS) [35], which measures reward, relief, and habit drinking sub-types; (6) Beck Depression Inventory (BDI-II) [36], which measures depressive symptomatology over the past two weeks; and (7) Beck Anxiety Inventory (BAI) [37], which measures anxiety symptomatology over the past two weeks. Participant demographic, alcohol use/problem, smoking, and mood characteristics are detailed in Table 1.
Participants completed a standard cue exposure paradigm on Day 1. Alcohol cue exposure followed a well-established experimental procedure [3, 38], which (1) 3-minute relaxation period (baseline); (2) 3-minute water exposure period, where participants held and smelled a glass of water to control for the effects of simple exposure to any potable liquid (water condition); and (3) 3-minute alcohol exposure period, where participants held (visual, tactile cues) and smelled (olfactory cue) a glass of their preferred alcohol beverage and were asked to recall sensory and psychological memories (guided imagery) associated with their alcohol use. Cue presentation order was not counterbalanced to avoid carryover effects [3]. Participants who identified as cigarette smokers were allowed to take a smoke break immediately prior to the cue reactivity assessment to avoid potential confounding effects of nicotine withdrawal. The baseline and water conditions were administered in a standard testing room, and the alcohol condition was administered in a simulated bar room.
Participants completed the Alcohol Urge Questionnaire (AUQ) [39] after the 3-minute relaxation period (baseline), after the presentation of the water cue, and after the presentation of the alcohol cue. The AUQ is an 8-item self-report questionnaire that measures participants urge to drink alcohol at the present moment. Participants reported their agreement or disagreement with 8 statements related to the subjective experience of alcohol craving on a 7-point Likert scale ranging from 0 (strongly disagree) to 6 (strongly agree). A total AUQ score was generated by summing responses to all eight questions. Two items (“I do not need to have a drink now” and “If I had a chance to have a drink, I don’t think I would drink it”) were reverse coded so that higher scores on all items would indicate greater alcohol urge. The AUQ has demonstrated high internal consistency in a host of human laboratory studies [40].
Analyses were performed with JMP Pro Statistical Software 16.0.0. First, participants were identified as cue reactive (CR+) and cue non-reactive (CR−). Participants were considered CR+ if their AUQ score following the alcohol condition was ≥50% greater than their AUQ score during the water condition. Participants who did not meet this were considered CR−. To validate that CR groups differed on the outcomes in which they were dichotomized, we performed dependent samples t-tests to examine within-group changes in AUQ scores (urge) during the alcohol cue exposure paradigm. We also performed two samples t-tests to examine between-group differences in AUQ scores during the cue exposure paradigm. One participant did not complete the cue exposure paradigm and was excluded from all analyses. This final sample included 52 participants.
Next, we examined between-group (CR+ versus CR−) differences in demographic (biological sex, age, and race/ethnicity), alcohol use/problems (DPDD and PAD, DSM-V AUD severity, ADS, PACS, OCDS, and RTC), smoking (smoker versus non-smoker), and mood (BDI and BAI) characteristics. Between-group differences in continuous variables were assessed using a t-test or Wilcoxon test, as appropriate, and between-group differences in categorical variables were assessed using a Chi-square or Fisher’s exact test, as appropriate. Significant differences (p≤0.05) are reported below. There was no correction for multiple comparisons due to the exploratory nature of this analysis.
Lastly, we used one-way ANCOVA to examine between-group differences in alcohol use (DPDD and PAD) during the (a) medication titration (Days 1–7) and (b) practice quit (Days 8–14) period. Alcohol use (DPDD and PDA, modeled separately) was the dependent variable, while group (CR+, CR−) was the independent variable. We explored within-subject changes in alcohol use using repeated measures ANCOVA. Alcohol use (DPDD or PDA as assessed baseline, Day 8, and Day 14) were the repeated measures dependent variable, with time, group, and their interaction as the independent variables. All models included the following biological sex, medication condition (varenicline, naltrexone, or placebo), age, and smoking status (smoker versus non-smoker). Three participants did not complete the TLFB on Days 8 and 14; therefore, the final sample for these analyses included 49 participants (N=30 CR−; N=19 CR+). Significant differences (p≤0.05) are reported below. As described above, there was no correction for multiple comparisons.
Twenty participants (38%) were identified as CR+, while 32 participants (62%) were identified as CR−. The CR+ group reported greater urge during the alcohol, versus the water, condition (t=−8,3, p<0.0001, 95% CI of difference=−15.2 – −9.0 d=−1.9), while the CR− group reported lower urge during the alcohol, versus, condition (t= 2.3, p=0.03 95% CI of difference=0.4 – 7.2; d=0.4), thereby supporting the CR groups differed on the outcomes in which they were dichotomized. Additionally, the CR+ group reported greater urge during the alcohol, versus the baseline, condition (t=−7.8, p<0.0001, 95% CI of difference=−14.2 – −8.2, d=−1.8), but did not differ in urge during the baseline and water conditions (t=0.9, p=0.4, 95% CI of difference=−1.2 – 3.0, d=0.2). The CR− group did not differ in urge during the alcohol and baseline conditions (t=−1.4, p=0.15, 95% CI of difference=−4.4 – 0.8, d=−0.25), but reported greater urge during the water, compared with baseline, condition (t=−3.9, p=0.005, 95% CI of difference=−8,6 – −2,7, d=−0.69). Urge was significantly lower in the CR+, compared with the CR−, group during the baseline and water conditions (baseline: t=2.1, p=0.04 95% CI of difference=0.23 – 13.8, d=0.54; t=4.7, p<0.0001 95% CI of difference=7.8 – 19.4, d=1.2), but urge did not differ between groups during the alcohol condition (t=−0.6, p=0.50, 95% CI of difference=−9.6 – 4.9, d=−0.17). Alcohol urge scores during the cue reactivity paradigm are presented in table 2 and illustrated in figure 1.
The groups differed in race composition characteristics (X^2^=12.5, p=0.02, 95% CI of difference=11.8 – 13.2, V=0.49) such that the CR+ group included a greater percentage of individuals who self-identified as White and the CR− group included a greater percentage of individuals who self-identified as Black or African American (p’s<0.05). The groups also differed in smoking prevalence such that the CR+, compared with the CR−, group had a fewer number of cigarette smokers (X^2^=4.7, p=0.04, 95% CI of difference=−0.7 – 9.1, ϕ=0.3). The CR+ and CR− groups did not differ in any other demographic, alcohol use/problem, or mood characteristics. See table 1.
The CR+, compared with the CR−, group reported greater DPDD during the medication titration period (F=4.9, p=0.03, 95% CI of difference=−3.3 – −1.8, partial eta squared=0.105,). Groups did not differ in PDA during the titration period, nor did they differ in their drinking during the practice quit attempt (p’s>0.18). See table 3. Both groups exhibited a reduction in DPDD across the study (main effect, F=19.1,0 p<0.001, partial eta squared=0.15), however, the CR− group exhibited a greater reduction, compared with the CR+ group (time by group interaction, F=3.90, p=0.029, partial eta squared=0.1; CR−, F=19.40, p<0.0001; CR+: F=6.90, p=0.01). Both groups exhibited an increase in PDA across the study (main effect, F=54.50, p<0.0001, partial eta squared=0.32), but there was not a significant time by group interaction (p=0.40). See figure 2.
This study characterized individuals with AUD as CR+ or CR−, as indicated by self-reported, subjective urge for alcohol during an alcohol-cue reactivity task, post-hoc in a human laboratory study of AUD medications and placebo. Thirty-eight percent of participants were CR+, further supporting the notion that not all individuals with an AUD are reactive to alcohol-related cues in the laboratory. Contrary to our prediction, the CR+ and CR− groups did not differ in drinking or AUD-related clinical characteristics at baseline, but did differ in race composition and smoking prevalence. Specifically, the CR+ group had a greater proportion of individuals who self-identified as White, while the CR− group was comprised of a greater proportion of individuals who self-identified as Black or African American. Additionally, the CR+ group had lower prevalence of smokers relative to the CR− group. In line with our drinking outcome prediction, the CR+ group reported greater DPDD throughout the human laboratory trial, compared with the CR− group. Overall, these results indicate that cue reactivity is a clinically relevant, but heterogenous, construct. This heterogeneity must be considered when designing, and interpreting results from, cue reactivity studies [12].
The frequency of cue reactive individuals observed in this study (38%) is largely consistent with prior studies measuring subjective cue reactivity. Our laboratory recently found that ~42% of individuals with a moderate-to-severe current AUD are cue reactive [11]. Earlier studies reported a wide range prevalence of cue reactive individuals, as indicated by subjective response, with AUD (~22 – 65%) [15, 16, 19, 24]. Considering these studies were all comprised of relatively small samples, it is not surprising that there is variability in the observed rates between studies. Furthermore, a subset of studies included a negative mood induction component, which has been shown to enhance subjective alcohol cue reactivity [9, 16, 17]. Despite this variability, the data converge to suggest that a significant portion of individuals with an AUD do not have a subjective reaction to alcohol-related cues in a laboratory setting. Also consistent with our results, prior work has failed to identify differences in AUD-related clinical characteristics between individuals who are and are not cue reactive. In a review of several cue exposure studies, Rohsenow and colleagues (1992) found an inconsistent association between cue reactivity and alcohol dependence severity. While one study showed that cue reactive individuals, compared with non-reactive individuals, had more severe AUD (indexed by the ADS), this association did not replicate in a similar study with parallel procedures [24]. Our laboratory previously reported that treatment-seeking status did not impact alcohol cue reactivity (i.e., likelihood of being cue reactive versus non-reactive), despite recognized differences in AUD-clinical characteristics between treatment- and non-treatment-seekers [11, 41]. The current study did not detect group differences in AUD clinical characteristics but did observe differences in race composition and smoking status. These findings underscore the importance of considering demographic and smoking factors in the context of AUD research. Most cue reactivity research has not considered demographic and smoking factors; additional research in large, racially diverse samples is needed to replicate and extend these findings. It is possible that genetic factors contribute to the observed differences in cue reactivity. Studies have examined pharmacogenetic effects in the treatment of AUD [42]; as alcohol cue reactivity is commonly used as an endpoint in human laboratory medication trials, investigations examining the role of genetics on cue reactivity are warranted. Specifically, it is plausible that some of the genetic loading towards AUD operate by increasing one’s reactivity to alcohol cues. However, since genetic studies have moved to large samples sizes and genome wide association studies (GWAS), they often lack fine-grained phenotypic measures such as alcohol cue-reactivity.
The CR− group reported greater alcohol urge during the baseline and water conditions compared with the CR+ group. We speculate that that higher prevalence of cigarette smokers in the CR−, versus CR+, group (50% of CR− group versus 20% of CR+ group) is one factor contributing to these group differences. Participants who identified as cigarette smokers were allowed to take a smoke break immediately prior to the alcohol cue reactivity task. This was implemented in order to avoid potential confounding effects of nicotine craving; however, alcohol and tobacco have reciprocal effects on potentiating subjective craving [43]. It is plausible that cigarette smoking acutely increased alcohol craving, as reflected by higher alcohol urge scores during the baseline and water conditions in the CR− group. The CR+ and CR− groups did not differ in other self-reported measures of tonic craving (PACS and OCDS), further supporting that greater craving in the CR− group at baseline was an acute/phasic effect.
Additionally, the CR−, but not the CR+, group exhibited a significant decrease in alcohol urge from the water to alcohol condition. As the alcohol condition always followed the water condition (i.e., to avoid carryover effects), it is possible that, with time, the CR− group was able to regulate craving. A review of neuroimaging drug cue reactivity studies found that treatment seeking status and/or drug availability was a critical factor modulating neural alcohol cue reactivity, as this factor provides a situational context for drug cues [44]. Individuals not actively seeking treatment exhibited greater alcohol cue reactivity in critical frontal and ventral striatal reward regions, which the authors interpreted to reflect anticipation and preparation to engage in alcohol use. On the other hand, individuals actively trying to stop or reduce their drinking may have dampened neural alcohol cue reactivity because they are actively trying to suppress craving. While all participants in this study reported intrinsic motivation to reduce or quit drinking, the CR− group was more effective than the CR+ group in decreasing their alcohol use during the trial, as reflected by lower DPDD. Interpreted in the context of this neuroimaging review, our results suggest the CR− group may be able to more effectively regulate cue-induced craving by maintaining that alcohol is not available to them in the near future. However, another possibility is that individuals who are non-reactive to alcohol-related cues in the laboratory are also less reactive to such cues in their natural environment, ultimately facilitating decreased alcohol use during a quit attempt.
Attention and interoception are additional factors that may have influenced these alcohol cue reactivity findings. For example, it is possible the CR− group decreased task engagement by the time of the alcohol condition. Rohsenow and colleagues (1992) found that cue reactive, compared with non-reactive, individuals payed more attention to the sight and smell of alcohol during a cue exposure paradigm [24]. The current study did not collect a measure of attention; therefore, we cannot rule out that differences in levels of attention to, and engagement with, alcohol cues during the paradigm could have contributed to individual differences in cue reactivity. Interoception is another variable that may influence the ability to detect and report the subjective experience of craving [45, 46]. Individuals with AUD have difficulty identifying and describing their emotional experiences [47], which may impact AUD participants’ ability to detect and report changes in their subjective experience during cue exposure. Physiological indicators (i.e., changes in heart rate, skin conductance, blood pressure) of alcohol cue reactivity were not collected in this study, but represent a tantalizing approach to obtain objective measures of cue reactivity not impacted by interoception [48, 49]. Prior work has identified three typological subgroups of cue reactivity in AUD, including 1) subjective and physiological responders, 2) physiological-only responders, and 3) non-responders [15]. These data suggest the “physiological-only responders” represent a distinct group of individuals, not captured in the current study, who lack awareness of their subjective experience during cue exposure. However, a recent review found little support for the use of physiological outcomes [12]. A host of factors may complicate the interpretation of physiological response. Notably, physiological measure are highly sensitive to individual differences [50, 51] (i.e., AUD severity, withdrawal state, comorbid psychological and medical conditions) and medication-specific effects [52, 53], which may diminish their ability to detect meaningful signal [12]. Despite these limitations, physiological data, particularly if obtained in conjunction with subjective reports, could help to understand heterogeneity in cue reactivity. Assessments of attention and interoception (i.e., Toronto Alexythymia Scale [54]) could also help examine this potential confound.
It is also important to consider that the types of cues used may influence cue reactivity. This study used participant-specific (i.e., preferred alcohol beverage) and multisensory alcohol cues. While this approach provides strong ecological validity, the variability in drinking cues presented may add noise to the data [55]. The alcohol presentation in this study occurred in a simulated bar laboratory. It is possible that this setting influenced cue reactivity findings by serving as an additional salient alcohol-related cue for those who tend to drink in bar settings, but not for those who do not typically drink in bar settings. Additionally, we did not examine relevant mood-related factors. As discussed, negative mood induction has been shown to enhance cue reactivity (both subjective and physiological response) [16, 17]. The current study was not designed to assess the impact of negative mood. Negative mood plays a critical role in the clinical course of AUD [48], and future studies are needed to characterize cue reactivity in paradigms with a mood/stress induction component.
Results from this study should be interpreted in light of several other strengths and limitations, in addition to those discussed above. Study strengths include a within-subjects design, racially diverse sample of individuals with a moderate-to-severe AUD, a relatively equal number of males to females, and use of a well-validated paradigm, that participant-specific and multisensory alcohol cues, to measure subjective alcohol cue reactivity. This study, however, had a relatively small sample size, which could have limited our power to detect group differences in clinical characteristics. Further, considering the exploratory nature of this study, we did not control for multiple comparisons. Larger follow-up studies are needed to re-examine these questions while employing more robust statistical thresholding. Additionally, larger sample sizes would allow us to take data driven approaches to improve our classification of cue reactivity. Overall, this study has important implications for the design and interpretation of cue reactivity research. First, studies implementing this paradigm should not assume that all individuals are subjectively reactive to alcohol-related cues in the laboratory. This is particularly relevant for longitudinal pharmacotherapy trials seeking to evaluate a medication (or any treatment) effect on cue reactivity. If a substantial portion of participants are CR− to begin with, there could be a lack of signal (cue-induced subjective urge) for the medication to mitigate. Second, while screening for cue reactivity may enhance signal detection, it could also decrease generalizability of findings to the broader AUD population. While larger-powered studies are needed to fully flesh out the clinical characteristics associated with cue reactivity, results from this study suggest excluding cue non-reactive individuals could result in disproportionate exclusion of Black participants and/or individuals who smoke. The limited generalizability associated with focusing exclusively on CR+ individuals may compromise the predictive utility of this measure for clinical efficacy in the broader AUD population.
The current study sought to characterize alcohol CR+ and CR− individuals with a current moderate-to-severe AUD. Results further support that not all individuals with AUD are cue reactive, as indicated by self-report, subjective assessment, and suggest that CR+ and CR− individuals may differ in demographic (i.e., race) and smoking factors, as well as in drinking outcomes in the context of a human laboratory pharmacotherapy trial. Recognizing and understanding heterogeneity in cue reactivity, and the clinical factors associated with it, is critical to advancing this paradigm as an early efficacy marker in AUD research.