Skip to main content
NCBI home page
NIHPA Author Manuscripts logoLink to NIHPA Author Manuscripts
. Author manuscript; available in PMC: 2019 Aug 1.
Published in final edited form as: Addict Behav. 2018 Jan 10;83:72–78. doi: 10.1016/j.addbeh.2018.01.009

Real-Time Assessment of Alcohol Craving and Naltrexone Treatment Responsiveness in a Randomized Clinical Trial

Robert Miranda Jr 1, Hayley Treloar Padovano 1, Joshua C Gray 1, Stephanie E Wemm 1, Alexander Blanchard 1
PMCID: PMC5964002  NIHMSID: NIHMS937731  PMID: 29395188

Abstract

Introduction

This secondary data analysis examined whether and how the dopamine receptor D4 gene (DRD4) influenced naltrexone treatment responsiveness in a randomized clinical trial. We leveraged intensive experience sampling methods to test the hypothesis that craving recorded at drinking and non-drinking moments would mediate naltrexone effects on the likelihood of heavy drinking, but only among carriers of the DRD4 long (DRD4-L) allele.

Methods

Participants (Mage=29.8 years, SD=12.1) were non-treatment seeking heavy drinkers (n=104, 54.8% female, 61.5% alcohol dependent) randomized to 3 weeks of daily naltrexone (50mg) or placebo. During these 3 weeks, participants used handheld electronic devices to complete real-time reports of alcohol use and craving multiple times per day in their natural environments. This approach afforded intensive repeated assessment of focal variables and provided in-the-moment data to test whether craving when not drinking or early in drinking episodes explained naltrexone effects on drinking.

Results

Moderated-mediation multilevel structural equation models showed that craving during non-drinking moments mediated the treatment effect of naltrexone on heavy drinking but only among carriers of the DRD4-L allele. The same pattern of associations was not shown when evaluating craving while participants were consuming alcoholic beverages.

Conclusions

Findings provide the first in vivo evidence that, among carriers of the DRD4-L allele, naltrexone blunts craving in real-world settings, and this effect in turn reduces the likelihood of heavy drinking. This work highlights the utility of EMA methods for elucidating how treatments work and further demonstrates the importance of genetic factors for understanding individual differences in pharmacotherapy responsiveness.

Keywords: Alcohol, Naltrexone, Ecological Momentary Assessment, Craving, Genetic

1. INTRODUCTION

An estimated 29.1% of adults in the United States meet criteria for alcohol use disorder (AUD) during their lifetime (Grant et al., 2015), and excessive alcohol use has an adverse economic impact of $249 billion each year (Sacks, Gonzales, Bouchery, Tomedi, & Brewer, 2015). Yet, even the best available AUD interventions produce only modest benefit, and many individuals show little or no improvement. Thus, there is an urgent need to better understand how and for whom treatments work (Chung et al., 2016).

Naltrexone, an opiate receptor antagonist, is a frontline medication for treating AUD. Meta-analyses show that naltrexone reduces the quantity and frequency of alcohol use as well as relapse rates in heavy drinking adults (Jonas, Amick, Feltner, Wines, et al., 2014; Maisel, Blodgett, Wilbourne, Humphreys, & Finney, 2013); but it is not effective for all patients. Approximately 1 of every 12 people treated with naltrexone shows reductions in heavy drinking (Jonas, Amick, Feltner, Bobashev, et al., 2014). Given this considerable person-to-person variability, a growing focus of clinical research is to identify patient characteristics that predict naltrexone treatment responsiveness. The predictive value of previously studied moderators (e.g., sex, pretreatment drinking), however, is low (Garbutt et al., 2014).

Genetic variants associated with dopamine function may inform the types of individuals likely to benefit from certain medications, including naltrexone (Kranzler & Edenberg, 2010; Patriquin, Bauer, Soares, Graham, & Nielsen, 2015). The dopaminergic system plays a central role in reward processes and the pathogenesis of addiction (Di Chiara & Imperato, 1988; Nutt, Lingford-Hughes, Erritzoe, & Stokes, 2015; Wise, 1998). Alcohol increases the release and synaptic availability of mesolimbic dopamine, which promotes craving and further drinking (Di Chiara & Imperato, 1988). The dopamine receptor D4 gene (DRD4) contains a 48-base-pair (48bp) variable number tandem repeat (VNTR) sequence in exon III with three common length variants, including two, four, and seven or more repeats (Van Tol et al., 1992). The ≥7-repeat long allele (DRD4-L) is associated with reduced intracellular responsiveness to dopamine (Asghari et al., 1995). Compared to homozygotes for the <7-repeat short allele (DRD4-S), carriers of the DRD4-L allele show greater alcohol craving and consumption in laboratory (Hutchison, McGeary, Smolen, Bryan, & Swift, 2002; Mackillop, Menges, McGeary, & Lisman, 2007; McGeary et al., 2006) and real-world settings (Ray et al., 2010). DRD4-L carriers also show greater neural responses to alcohol cues (Filbey et al., 2008).

Naltrexone is thought to blunt dopaminergic transmission in mesolimbic pathways (Benjamin, Grant, & Pohorecky, 1993), thereby attenuating craving and the reinforcing effects of alcohol (Gonzales & Weiss, 1998; Valenta et al., 2013). In clinical trials, naltrexone reduced the frequency and intensity of weekly craving ratings (Maisel et al., 2013). A meta-analysis of laboratory studies found that naltrexone blunts craving caused by exposure to alcohol cues (Hendershot, Wardell, Samokhvalov, & Rehm, 2016). The magnitude of this effect was modest, however, and there is considerable variability in craving response that warrants further attention in subgroup analyses.

Although laboratory studies are the standard for understanding medication mechanisms, they lack real-world, ecological validity. Researchers have turned to daily assessment methods to examine the effects of naltrexone on craving and drinking in the natural environment. This approach allows for a more complete understanding of treatment effects on the dynamics of craving and other potential mechanisms of behavior change while identifying important between-person characteristics that may moderate these associations. Helstrom and colleagues used a daily diary approach to test naltrexone’s effects on craving during a 28-day inpatient program (Helstrom et al., 2016). Patients rated their craving once per day and results showed that individuals who elected to take naltrexone had sharper decreases in craving than individuals who chose not to take the medication. Although such daily studies capture medication effects on putative mechanisms of behavior change, only the first link of the proposed causal chain from medication to putative mechanism was tested, and the second link from the mechanism to the outcome (naturalistic drinking) was assumed. Kranzler and colleagues took this work a step further by comparing the effects of daily versus targeted naltrexone on problem drinkers’ alcohol use and ‘desire to drink’ and examined whether associations between ‘desire to drink’ and alcohol use were moderated by a mu-opioid receptor gene and naltrexone treatment (Kranzler, Armeli, Covault, & Tennen, 2013). Participants completed telephone surveys once daily between 5:00 and 9:00 PM. Results showed that the OPRM1 gene moderated the association between ‘desire to drink’ and subsequent daily drinking. But this effect was only found for individuals treated with placebo; the interaction between OPRM1 and ‘desire to drink’ was not supported for individuals randomized to naltrexone treatment. These findings provide prospective evidence that craving may, in part, explain how naltrexone reduces drinking, at least for some patients.

One limitation of prior work, however, was that data were collected at single time points and participants retrospectively rated their level of craving over the course of the day. This approach cannot capture moment-by-moment variation in phenomena that are subject to rapid change, such as craving, and tests of whether medication effects on mechanisms depend on certain contextual influences, such as drinking, cannot be done. Repeated real-time measurement of focal constructs as they occur in the natural environment avoids problems with retrospective reports and allows for fine-grained analyses of individual variability by parsing within- and between-subjects variance (Shiffman, 2014). This approach is inherently idiographic and uniquely suited for examining individual variability in target constructs, such as craving, and testing the strength their associations with clinical outcomes (e.g., daily substance use).

This study aimed to further elucidate craving as a potential mechanism of naltrexone effects on heavy drinking by leveraging intensive experience sampling methods within the context of a randomized clinical trial. By combining clinical trial and ecological momentary assessment (EMA) methods, we sought to conduct a sophisticated test of how and for whom naltrexone reduces heavy drinking. Specifically, we examined the sequenced relationship between craving and heavy drinking. This approach is well suited to test intervention effects on substance use and, perhaps more uniquely, to evaluate which treatment- induced changes (i.e., mechanisms) account for intervention effects and under what conditions (Miranda & Treloar, 2016).

We conducted secondary data analyses of a 3-week clinical trial that found naltrexone, as compared to placebo, reduced the percentage of drinking days among community-recruited heavy drinkers but reduced heavy drinking only among carriers of the DRD4-L allele; there was no effect of naltrexone on heavy drinking among DRD4-S carriers (Tidey et al., 2008). Fine-grained tests evaluated: (1) naltrexone effects on the likelihood of heavy drinking each day—our initial report tested naltrexone’s effects on aggregated drinking data across each week (Tidey et al., 2008); (2) the indirect (mediational) effects of naltrexone treatment on the log odds of heavy drinking through craving captured in real-time when not drinking and early in drinking episodes, and (3) whether genotype moderates the mediation effects. We hypothesized that craving, recorded at non-drinking moments (and on drinking days, only prior to alcohol use) and immediately after 1–2 drinks, would mediate naltrexone effects on the likelihood of heavy drinking, but only among carriers of the DRD4-L allele. We also investigated the influence of sex. We disentangled craving in non-drinking moments from craving elicited by a priming dose of alcohol because we found pharmacotherapies for AUD can differentially affect craving under these distinct conditions (Miranda et al., 2016).

2. METHODS

2.1 Study Design

Following a 1-week placebo lead-in to establish EMA and medication compliance, participants were randomized to naltrexone (50 mg/day) or placebo under double-blind conditions for three weeks. Present analyses focused on the three weeks of active treatment versus placebo. Participants used a handheld electronic device to complete real-time reports of alcohol use and craving multiple times per day in their natural environments. The Brown University Institutional Review Board approved all procedures. Details for the parent trial are published elsewhere (Tidey et al., 2008).

2.2 Participants and Procedures

Participants for this secondary analysis were 104 non-treatment seeking heavy drinkers recruited from the community for a study of naltrexone effects (Tidey et al., 2008). Eligible participants were ≥21 years old and consumed alcohol ≥4 days/week in the past month. Men consumed >6 standard drinks and women >4 on at least two days/week in the past month. Individuals were excluded for abuse or dependence of substances other than nicotine or alcohol, past alcohol treatment, a positive toxicology screen for opiates, or medications or medical conditions that were contraindicated. Women were excluded if they were pregnant, nursing, or unwilling to use birth control.

Participants provided written informed consent, completed baseline measures, and received training in the EMA protocol, which was designed for this study and implemented on investigator-provided handheld electronic devices (Palm, Inc., Sunnyvale, CA). EMA reports were completed: (1) upon awakening; (2) in response to audible prompts presented at random times approximately five times per day (random prompts); and (3) immediately after each of the first two drinks of a drinking episode. EMA software permitted a delay (≤ 20min) or suspension (≤ 2hrs per day) of random prompts, and data were downloaded and reviewed for compliance at weekly appointments. Compensation was based on medication and EMA compliance.

2.3 Measures

2.3.1 Individual difference characteristics

Baseline information included a demographics questionnaire and the 90-day Timeline Follow-back interview (TLFB) to assess drinking patterns (Sobell & Sobell, 1992). Alcohol diagnoses were derived based the Structured Clinical Interview for DSM-IV Axis I Disorders–Patient Version (First, Spitzer, Gibbon, & Williams, 2002).

Candidate genotyping was conducted using previously published methods (McGeary et al., 2006). Genomic DNA was isolated from saliva (Freeman et al., 1997; Lench, Stanier, & Williamson, 1988). The DRD4 gene was assayed using modifications of previously reported methods (Sander et al., 1997). Conventional methods for grouping DRD4 status were used (Hutchison et al., 2002; Ray et al., 2010); DRD4-L had at least one copy of the ≥7 repeat allele and the DRD4-S had neither copy >6 repeats. Only a subset of the larger sample was genotyped (Tidey et al., 2008); DNA collection began 12 months after study initiation. The observed frequency was DRD4-L n=39 and DRD4-S n=65.

2.3.2 Ambulatory measures

Each morning, the EMA device prompted participants to record the number and type of standard drinks consumed on the previous day. For analyses, responses were coded to identify heavy drinking days (non-heavy drinking day=0; heavy drinking day=1), defined as ≥5 drinks for men and ≥4 for women. We focused on heavy drinking because it is a clinically meaningful endpoint and because our initial report showed that DRD4 moderated naltrexone’s effects on this outcome (Tidey et al., 2008). The well-established single- item craving assessment was completed via EMA. Specifically, participants rated their urge to drink on an 11-point Likert-type scale rating from 0 (No urge) to 10 (Strongest ever). Separate models examined craving reported at (1) non-drinking random prompts, and (2) during the first two standard drinks of a given episode. Random prompts collected after participants began drinking (n = 1,224; 12.8%) were excluded to avoid confounding craving at non-drinking moments with craving assessed after drinking. These assessments were not pooled with drinking-moment data due to variable delays between drink reports and random prompts delivered post-drinking.

2.4 Data Analytic Plan

Moderated-mediation multilevel structural equation models (MSEM) were implemented with Mplus software, Version 7.2 (Muthen & Muthen, 1998–2014). MSEM is preferable to multilevel models for testing mediation because putative mediating associations are allowed to vary across levels of analysis, i.e., temporal (daily) relations between variables are disaggregated from static (person-level) differences. A conceptual moderated-mediation MSEM (Figure 1) illustrates hypothesized associations at each level of analysis. This work combines the logic of statistical papers on multilevel SEM (Preacher, Zhang, & Zyphur, 2011; Preacher, Zyphur, & Zhang, 2010) and moderated mediation (Preacher, Rucker, & Hayes, 2007).

Figure 1.

Figure 1

Open in a new tab

Conceptual illustration of moderated-mediation multilevel structural equation model showing disaggregation of effects at the between- and within-person levels. The subscript j indicates variables assessed only once at the baseline laboratory visit, and the subscript ij indicates variables assessed repeatedly in the natural environment. Craving and heavy drinking days are latent variables defined by observed variables. As a conceptual illustration, however, this figure does not perfectly correspond to the model that is mathematically estimated.

Genotype and medication condition only had one value per person, and thus existed only at the highest level of analysis (level 2) and had only between-person variability. Heavy drinking day (HDD) was the primary outcome and had variability at the daily level (level 1) and between persons (level 2).1 The focal mediator, craving, was assessed multiple times per day but ultimately aggregated as a latent variable at the daily level, thus having variability similar to HDD, i.e., at the daily level (level 1) and between persons (level 2). We might refer to this as a 2(2)-1-1 data structure, with medication condition and genotype at level 2, craving at level 1, and HDD at level 1. As advised by Preacher and colleagues (Lachowicz, Sterba, & Preacher, 2015; Preacher et al., 2011; Preacher et al., 2010) mediation in our 2(2)-1-1 model is considered a between-person mediation effect, because there is not variability in medication condition or genotype within persons. Thus, moderation by genotype of the relation of medication condition with the mediator, craving, was also hypothesized at the between level.

Sequenced data is a necessary condition for testing mediation. Our approach allowed for sequencing of the data structure by including only those craving assessments that occurred when not drinking (and prior to drinking on drinking days) or immediately after each of two first standard drinks of a drinking day. Thus, we tested whether craving captured at nondrinking moments or early in drinking episodes predicted the likelihood of that day later becoming a HDD. Still, it is important to bear in mind that the indirect effects are ultimately calculated from between-person effects.

Bayesian estimation with diffuse (non-informative) priors and 95% highest posterior density (HPD) credibility intervals accounted for the asymmetric sampling distribution of the proposed indirect effect (Muthen & Asparouhov, 2012). Moderated mediation using Bayesian methods with diffuse priors is shown to yield unbiased estimates with improved power over maximum- likelihood (ML) with delta method standard-error estimates, ML with bootstrapped percentile confidence intervals, and similar power to bootstrapped bias-corrected confidence intervals (Wang & Preacher, 2017). Convergence was evaluated through inspecting the Proportional Scale Reduction (PSR) factor and trace plots. The FBITERATIONS option (20,000 iterations) confirmed that parameter values remained stable and PSR values remained close to 1. Technical aspects of the Bayesian MSEM analysis in Mplus are available online (Muthén, 2010).

The HDD outcome was dichotomous (0=non-HDD; 1=HDD). Model constraints calculated moderated indirect effects. Since the value of the moderator (genotype) was dichotomous, with DRD4-S=0 and DRD4-L=1, the indirect effects at each value of the moderator were reflected by a1×bbetween and (a1+a3)×bbetween, respectively.2 Sex was included as a covariate to account for our observed sex differences in DRD4 variants and potential non-treatment-related common causes of craving and heavy drinking (0=men; 1=women). Additional models included a random slope term for the “b” path to test whether allowing the effect of craving on HDD to vary within persons altered our pattern of moderated mediation effects. Including a random slope term for the link of craving and HDD also allowed us to test whether genotype moderated the influence of medication on the within-person, craving-to-HDD association. The pattern of coefficients was examined to determine whether mediation was supported, and formal tests of mediation were considered to reach a common threshold of statistical “significance” when the Bayesian 95% credibility interval did not include zero.

3. RESULTS

Participants (Mage=29.8 years, SD=12.1; 54.8% women) consumed, on average, 30.2 drinks/week at baseline (SD=14.2). The majority met criteria for alcohol abuse (18.3%) or dependence (61.5%), and most were White (95.2%) and non-Hispanic (99.0%). Men were more likely to carry DRD4-S (72.3%) than DRD4-L (27.7%); these variants were more equally distributed in women (DRD4-S: 54.5%; DRD4-L: 45.6%). Other demographic characteristics (i.e., race, ethnicity, age, baseline drinking patterns) did not differ by genotype. Genotype was also balanced within treatment group, χ2 (df=1) = .003, p = .960.

Participants completed 3.3 non-drinking random prompt reports per day, on average (SD=1.7), which produced a total of 8,361 random assessments during the 3-week period. On the whole, participants drank on 1,409 of 2,480 study days (56.8%), over half of which were HDDs (60.9%). The average number of standard drinks consumed per day was 6.2 (SD=4.2), with men consuming greater daily volumes than women, Mdifference=1.59, t(1319)=6.9, p<.001. On drinking days, participants completed 1,430 first-drink reports and 1,070 second-drink reports. Morning reports of total drinks were obtained for 1,321 drinking episodes (92.4%). Consistent with the results of the parent trial, naltrexone did not affect the likelihood of heavy drinking on a given day when sex was included in the model, 95%CI[−0.39,0.06], or removed, 95%CI[−0.36,0.09].

An initial model explored variability in craving at the within and between levels, without any predictors. The intraclass correlation coefficient was calculated as the ratio of between-level variance to total variance, i.e., 3.94/(3.94+6.28)=.386, suggesting that 38.6% of the variability in craving was due to between-level influences. Inclusion of Medication, Genotype, and their interactive effect altered the variance ratio, i.e., 3.61/(3.61+6.27)=.365, suggesting a reduction in between-level craving variability after accounting for these putative between-level influences. The difference in between residual variance from 3.94 to 3.61 reflects an 8.4% reduction in craving variability when Medication, Genotype, and their interactive effect are included in the model (i.e., 3.61 is 91.6% of 3.94).

Table 1 presents results from tests of moderated mediation via craving assessed at non-drinking times. The pattern of coefficients supported the hypothesized relations, and the 95% CI did not include zero.3 Greater craving at non-drinking times was a strong predictor of the log odds of heavy drinking (bwithin and bbetween paths, Table 1). As expected, naltrexone’s effects on craving were moderated by Genotype, in the negative direction (a3 path, Table 1), such that naltrexone produced the greatest reductions in craving among DRD4-L carriers, 95%CI[−3.07,0.38]. Indirect effects tests supported mediation in DRD4-L carriers, 95%CI[−0.30,−0.02], but not DRD4-S carriers, 95%CI[−0.03,0.14]. The same pattern of indirect-effect estimates was shown when sex was not included as a covariate (DRD4-S: 95%CI[−0.07,0.15]; DRD4-L: 95%CI[− 0.28,0.01]).

Table 1.

Between, Within, and Indirect Effects (with Bayesian Credibility Intervals) from a Moderated-Mediation Multilevel Structural Equation Model Predicting Heavy Drinking Days through Craving at Non-Drinking Times

95% C.I.b
Parameter Est. / OR Posterior
S.D.		 Lower
2.5%		 Upper
2.5%
Between
a1: Medication Conditionj → Cravingj 0.39 0.479 (− 0.45, 1.17)
a2: Genotypej → Cravingj (a2) 0.73 0.59 (− 0.21, 2.02)
a3: Medicationj × Genotypej → Cravingj − 1.48 0.84 (− 3.07, 0.38)
bbetween: Cravingj → Heavy Drink Dayj 0.10 / 1.10 0.04 (0.03, 0.16) *
c′1: Medication Conditionj → Heavy Drink Dayj − 0.02 / 0.98 0.14 (− 0.32, 0.23)
c′2: Genotypej → Heavy Drink Dayj 0.22 / 1.24 0.16 (− 0.14, 0.52)
c′3: Medicationj × Genotypej → Heavy Drink Dayj − 0.28 / 0.76 0.23 (− 0.80, 0.16)
Sexj → Cravingj − 0.13 0.38 (− 0.82, 0.61)
Sexj → Heavy Drink Dayj − 0.24 / 0.79 0.13 (− 0.47, 0.01)
Variance Medication Conditionj 0.26 0.04 (0.19, 0.34) *
Residual variance Cravingj 3.26 0.49 (2.41, 4.15) *
Residual variance Heavy Drink Dayj 0.33 0.06 (0.22, 0.44) *
Within
bwithin: Cravingij → Heavy Drink Dayij 0.08 / 1.08 0.01 (0.06, 0.09) *
Variance Cravingij 4.93 0.09 (4.74, 5.12) *
Indirect Effects
a1 × bbetween: Indirect Effect DRD4-S 0.04 0.05 (− 0.03, 0.14)
(a1 + a3) × bbetween: Indirect Effect DRD4-L − 0.10 0.08 (− 0.30, − 0.02) *
Open in a new tab

Note. Est. = estimate. OR = odds ratio. S.D. = standard deviation. C.I. = credibility interval. p-value = Bayesian one-tailed p-value. The heavy drinking day outcome is a binary variable (0 = non-drinking day or non-heavy drinking day; 1 = heavy drinking day); thus, treated as a categorical outcome in Mplus, with exponentiated estimates calculated and interpreted as odds ratios. Sex is coded as 1 for women and 0 for men.

a

For positive values, the Bayesian one-tailed p-value is the proportion of the posterior distribution below zero; for negative values, the Bayesian one-tailed p-value is the proportion of posterior distribution above zero.

b

The Bayesian credibility interval is based on the percentiles of the posterior and allows for a strongly skewed distribution.

The final columns show the 2.5 and 97.5 percentiles in the posterior distribution. Intervals that do not include zero are noted with an asterisk.

Multilevel models can be sensitive to the specification of random effects. As such, additional models included a random slope term for the “b” path to evaluate whether allowing the effect of craving on likelihood of heavy drinking to vary within persons altered our pattern of moderated-mediation effects. The pattern of indirect effects still supported mediation in DRD4-L carriers, 95%CI[−0.31,0.00], but not DRD4-S carriers, 95%CI[−0.04,0.16], and the pattern remained when sex was included as a covariate (DRD4-S: 95%CI[−0.05,0.17]; DRD4-L: 95%CI[− 0.31,0.01]). Additionally, the “b” path was not moderated by Medication, Genotype, or their interactive effect.

Greater craving after the first two standard drinks of an episode predicted increased likelihood of heavy drinking (bwithin: 95%CI[0.13,0.22]; bbetween: 95%CI[0.19,0.42]). But results did not support moderation for craving measured while drinking (a3: 95%CI[−1.73,0.52]), and indirect effects did not support moderated mediation (DRD4-L: 95%CI[−0.61,0.03]). In addition, including sex as a covariate or including a random slope for the association of craving and HDD did not alter our pattern of mediation effects. Further, the interactive effect of DRD4-L and naltrexone on the random slope term did not support moderation of the “b” path, 95%CI[−0.04,0.18].

4. DISCUSSION

Findings from moderated-mediation multilevel structural equation models showed that naltrexone reduced the likelihood of heavy drinking in the natural environment through alleviating craving at non-drinking times, but only for those participants with the DRD4-L genotype. The same pattern of associations was not observed, however, when evaluating craving while participants were consuming alcohol in their natural environments. This works highlights the utility of EMA for identifying treatment mechanisms and moves beyond prior work by simultaneously testing a full moderated-mediation model, rather than separate tests of component pathways. On the whole, these findings are consistent with studies that show naltrexone blunts craving (Helstrom et al., 2016; Kranzler et al., 2013; Miranda et al., 2014; Monti et al., 1999) and builds on this work by showing a temporally sequenced association between craving and drinking.

Craving is central to most contemporary models of addiction (Drummond, 2001), predicts drinking in the human laboratory (Leeman, Corbin, & Fromme, 2009; MacKillop & Lisman, 2005; O'Malley, Krishnan-Sarin, Farren, Sinha, & Kreek, 2002), and is associated with relapse in clinical trials that target alcohol and other substances (Abrams, Monti, Carey, Pinto, & Jacobus, 1988; Erblich & Bovbjerg, 2004; Rohsenow et al., 1994). Importantly, carriers of the DRD4-L allele exhibit higher levels of alcohol craving and drinking in the human laboratory in many (Hutchison et al., 2002; Hutchison et al., 2006; Hutchison et al., 2001; Mackillop et al., 2007) but not all studies (van den Wildenberg, Janssen, Hutchison, van Breukelen, & Wiers, 2007). This study adds to a growing body of literature that examines within-person variability of craving in real-world settings to disentangle the association between fluctuations in craving and drinking. Our findings suggest that the DRD4-L polymorphism is associated with treatment-induced reductions in craving during non-drinking times rather than the neuropharmacological effects of alcohol ingestion. This possibility coincides with prior EMA and neuroimaging research that showed DRD4-L exhibited greater subjective (i.e., craving) and neural reactivity prior to, but not after, consuming alcohol (Filbey et al., 2008; Ray et al., 2010).

Findings from this study should be considered within the context of its several limitations. Participants were non-treatment seeking and results may not generalize to individuals interested in reducing their alcohol use. Similarly, we excluded individuals with co-occurring substance use disorders other than nicotine dependence. Many individuals seeking treatment for AUD struggle with other substance misuse in addition to alcohol, which may further limit generalization of our findings. In addition, the present sample was almost entirely White and non-Hispanic. In terms of our assessments, we relied on self-reported alcohol use; obtaining biomarkers of recent alcohol use may increase confidence in these results. Finally, the lack of within-person variability in medication condition and genotype precluded testing mediation at the daily level. Rather, between and within effects are disaggregated here to obtain the best possible indirect effect estimates for between-person mediation.

Even with these limitations, the current study provides important ecological evidence to explain “how” and “for whom” naltrexone may operate to reduce drinking, i.e., through blunting craving in real-world settings, but only among carriers of the DRD4-L allele variant. This work highlights the utility of EMA methods for elucidating how treatments work and demonstrates the importance of pharmacogenetics for understanding treatment responsiveness. These findings add to a growing body of clinical research that supports tailoring treatments to individual characteristics to improve patient care. Future studies may build upon these findings through testing other genetic pathways or mechanisms.

Highlights.

  • Tests whether and how a dopamine receptor gene influenced naltrexone effects

  • Craving mediated naltrexone effects on heavy drinking but only among carriers of the DRD4-L allele

  • Highlights value of intensive experience sampling for revealing how treatments work

Acknowledgments

Role of Funding Source: NIAAA Grants R01-AA07850 (RMJ) and K23-AA024808 (HT) provided funding for this study. NIAAA had no role in the study design, analysis or interpretation of the data, writing the manuscript, or the decision to submit the paper for publication.

Footnotes

Publisher's Disclaimer: This is a PDF file of an unedited manuscript that has been accepted for publication. As a service to our customers we are providing this early version of the manuscript. The manuscript will undergo copyediting, typesetting, and review of the resulting proof before it is published in its final citable form. Please note that during the production process errors may be discovered which could affect the content, and all legal disclaimers that apply to the journal pertain.

1

The heavy drinking day outcome was the only variable with missing data. Only 6.3% of heavy drinking day outcomes were missing (i.e., data present for 93.7% of cases). Covariance coverage, therefore, was .937 for all covariances of covariates with heavy drinking days and 1.000 for all other combinations. This amount of missing data is deemed ignorable. If we exclude cases for which the heavy drinking outcome variable is missing, the same pattern of moderated mediation is supported (DRD4-S: 95%CI[−0.04,0.12], p=.250; DRD4-L: 95%CI[− 0.24,0.02], p=.040).

2

Referring to Figure 1, the indirect effect a1×bbetween reflected the indirect effect of medication condition on heavy drinking through craving when the value of genotype was 0, i.e., the effect for DRD4-S. In turn, the indirect effect of medication condition on heavy drinking through craving when the value of genotype was 1, i.e., the effect of DRD4-L, was calculated as the sum of the between path from medication condition to craving with the between path from the interactive effect of medication condition and genotype, multiplied by the between path from craving to heavy drinking: (a1+a3)×bbetween.

3

Table 1 presents Bayesian credibility intervals based on 2.5 and 97.5 percentiles of the posterior and allows for a strongly skewed distribution. The 95% CI for the formal indirect effect did not include zero.

Contributions: RM and HT conceptualized the current study. HT and AB managed the data and HT conducted statistical analyses. All authors contributed to writing the manuscript and interpreting the findings; all authors approved the final manuscript.

Conflict of Interest: There are no conflicts of interest by any author

References

  1. Abrams DB, Monti PM, Carey KB, Pinto RP, Jacobus SI. Reactivity to smoking cues and relapse: two studies of discriminant validity. Behav Res Ther. 1988;26(3):225–233. doi: 10.1016/0005-7967(88)90003-4. [DOI] [PubMed] [Google Scholar]
  2. Asghari V, Sanyal S, Buchwaldt S, Paterson A, Jovanovic V, Van Tol HH. Modulation of intracellular cyclic AMP levels by different human dopamine D4 receptor variants. J Neurochem. 1995;65(3):1157–1165. doi: 10.1046/j.1471-4159.1995.65031157.x. [DOI] [PubMed] [Google Scholar]
  3. Benjamin D, Grant ER, Pohorecky LA. Naltrexone reverses ethanol-induced dopamine release in the nucleus accumbens in awake, freely moving rats. Brain Res. 1993;621(1):137–140. doi: 10.1016/0006-8993(93)90309-b. [DOI] [PubMed] [Google Scholar]
  4. Chung T, Noronha A, Carroll KM, Potenza MN, Hutchison K, Calhoun VD, Feldstein Ewing SW. Brain mechanisms of Change in Addictions Treatment: Models, Methods, and Emerging Findings. Curr Addict Rep. 2016;3(3):332–342. doi: 10.1007/s40429-016-0113-z. [DOI] [PMC free article] [PubMed] [Google Scholar]
  5. Di Chiara G, Imperato A. Drugs abused by humans preferentially increase synaptic dopamine concentrations in the mesolimbic system of freely moving rats. Proc Natl Acad Sci U S A. 1988;85(14):5274–5278. doi: 10.1073/pnas.85.14.5274. [DOI] [PMC free article] [PubMed] [Google Scholar]
  6. Erblich J, Bovbjerg DH. In vivo versus imaginal smoking cue exposures: is seeing believing? Exp Clin Psychopharmacol. 2004;12(3):208–215. doi: 10.1037/1064-1297.12.3.208. [DOI] [PubMed] [Google Scholar]
  7. Filbey FM, Ray L, Smolen A, Claus ED, Audette A, Hutchison KE. Differential neural response to alcohol priming and alcohol taste cues is associated with DRD4 VNTR and OPRM1 genotypes. Alcohol Clin Exp Res. 2008;32(7):1113–1123. doi: 10.1111/j.1530-0277.2008.00692.x. [DOI] [PMC free article] [PubMed] [Google Scholar]
  8. First MB, Spitzer RL, Gibbon M, Williams JBW. Structured Clinical Interview for DSM-IV-TR Axis I Disorders, Research Version, Patient Edition. (SCID-I/P) New York: Biometrics Research, New York State Psychiatric Institute; 2002. [Google Scholar]
  9. Freeman B, Powell J, Ball D, Hill L, Craig I, Plomin R. DNA by mail: an inexpensive and noninvasive method for collecting DNA samples from widely dispersed populations. Behav Genet. 1997;27(3):251–257. doi: 10.1023/a:1025614231190. [DOI] [PubMed] [Google Scholar]
  10. Garbutt JC, Greenblatt AM, West SL, Morgan LC, Kampov-Polevoy A, Jordan HS, Bobashev GV. Clinical and biological moderators of response to naltrexone in alcohol dependence: a systematic review of the evidence. Addiction. 2014;109(8):1274–1284. doi: 10.1111/add.12557. [DOI] [PubMed] [Google Scholar]
  11. Gonzales RA, Weiss F. Suppression of ethanol-reinforced behavior by naltrexone is associated with attenuation of the ethanol-induced increase in dialysate dopamine levels in the nucleus accumbens. J Neurosci. 1998;18(24):10663–10671. doi: 10.1523/JNEUROSCI.18-24-10663.1998. [DOI] [PMC free article] [PubMed] [Google Scholar]
  12. Grant BF, Goldstein RB, Saha TD, Chou SP, Jung J, Zhang H, Hasin DS. Epidemiology of DSM-5 Alcohol Use Disorder: Results From the National Epidemiologic Survey on Alcohol and Related Conditions III. JAMA Psychiatry. 2015;72(8):757–766. doi: 10.1001/jamapsychiatry.2015.0584. [DOI] [PMC free article] [PubMed] [Google Scholar]
  13. Helstrom AW, Blow FC, Slaymaker V, Kranzler HR, Leong S, Oslin D. Reductions in Alcohol Craving Following Naltrexone Treatment for Heavy Drinking. Alcohol Alcohol. 2016;51(5):562–566. doi: 10.1093/alcalc/agw038. [DOI] [PubMed] [Google Scholar]
  14. Hendershot CS, Wardell JD, Samokhvalov AV, Rehm J. Effects of naltrexone on alcohol self-administration and craving: meta-analysis of human laboratory studies. Addict Biol. 2016 doi: 10.1111/adb.12425. [DOI] [PMC free article] [PubMed] [Google Scholar]
  15. Hutchison KE, McGeary J, Smolen A, Bryan A, Swift RM. The DRD4 VNTR polymorphism moderates craving after alcohol consumption. Health Psychol. 2002;21(2):139–146. [PubMed] [Google Scholar]
  16. Hutchison KE, Ray L, Sandman E, Rutter MC, Peters A, Davidson D, Swift R. The effect of olanzapine on craving and alcohol consumption. Neuropsychopharmacology. 2006;31(6):1310–1317. doi: 10.1038/sj.npp.1300917. [DOI] [PubMed] [Google Scholar]
  17. Hutchison KE, Swift R, Rohsenow DJ, Monti PM, Davidson D, Almeida A. Olanzapine reduces urge to drink after drinking cues and a priming dose of alcohol. Psychopharmacology (Berl) 2001;155(1):27–34. doi: 10.1007/s002130000629. [DOI] [PubMed] [Google Scholar]
  18. Jonas DE, Amick HR, Feltner C, Bobashev G, Thomas K, Wines R, Garbutt JC. Pharmacotherapy for adults with alcohol use disorders in outpatient settings: a systematic review and meta-analysis. JAMA. 2014;311(18):1889–1900. doi: 10.1001/jama.2014.3628. [DOI] [PubMed] [Google Scholar]
  19. Jonas DE, Amick HR, Feltner C, Wines R, Shanahan E, Rowe CJ, Garbutt JC. Genetic polymorphisms and response to medications for alcohol use disorders: a systematic review and meta-analysis. Pharmacogenomics. 2014;15(13):1687–1700. doi: 10.2217/pgs.14.121. [DOI] [PubMed] [Google Scholar]
  20. Kranzler HR, Armeli S, Covault J, Tennen H. Variation in OPRM1 moderates the effect of desire to drink on subsequent drinking and its attenuation by naltrexone treatment. Addict Biol. 2013;18(1):193–201. doi: 10.1111/j.1369-1600.2012.00471.x. [DOI] [PMC free article] [PubMed] [Google Scholar]
  21. Kranzler HR, Edenberg HJ. Pharmacogenetics of alcohol and alcohol dependence treatment. Curr Pharm Des. 2010;16(19):2141–2148. doi: 10.2174/138161210791516387. [DOI] [PMC free article] [PubMed] [Google Scholar]
  22. Lachowicz MJ, Sterba SK, Preacher KJ. Investigating multilevel mediation with fully or partially nested data. Group Processes and Intergroup Relations. 2015;18(3):274–289. [Google Scholar]
  23. Leeman RF, Corbin WR, Fromme K. Craving Predicts Within Session Drinking Behavior Following Placebo. Pers Individ Dif. 2009;46(7):693–698. doi: 10.1016/j.paid.2009.01.024. [DOI] [PMC free article] [PubMed] [Google Scholar]
  24. Lench N, Stanier P, Williamson R. Simple non-invasive method to obtain DNA for gene analysis. Lancet. 1988;1(8599):1356–1358. doi: 10.1016/s0140-6736(88)92178-2. [DOI] [PubMed] [Google Scholar]
  25. MacKillop J, Lisman SA. Reactivity to alcohol cues: isolating the role of perceived availability. Exp Clin Psychopharmacol. 2005;13(3):229–237. doi: 10.1037/1064-1297.13.3.229. [DOI] [PubMed] [Google Scholar]
  26. Mackillop J, Menges DP, McGeary JE, Lisman SA. Effects of craving and DRD4 VNTR genotype on the relative value of alcohol: an initial human laboratory study. Behav Brain Funct. 2007;3:11. doi: 10.1186/1744-9081-3-11. [DOI] [PMC free article] [PubMed] [Google Scholar]
  27. Maisel NC, Blodgett JC, Wilbourne PL, Humphreys K, Finney JW. Meta-analysis of naltrexone and acamprosate for treating alcohol use disorders: when are these medications most helpful? Addiction. 2013;108(2):275–293. doi: 10.1111/j.1360-0443.2012.04054.x. [DOI] [PMC free article] [PubMed] [Google Scholar]
  28. McGeary JE, Monti PM, Rohsenow DJ, Tidey J, Swift R, Miranda R., Jr Genetic moderators of naltrexone's effects on alcohol cue reactivity. Alcohol Clin Exp Res. 2006;30(8):1288–1296. doi: 10.1111/j.1530-0277.2006.00156.x. [DOI] [PubMed] [Google Scholar]
  29. Miranda R, Jr, MacKillop J, Treloar H, Blanchard A, Tidey JW, Swift RM, Monti PM. Biobehavioral mechanisms of topiramate's effects on alcohol use: an investigation pairing laboratory and ecological momentary assessments. Addict Biol. 2016;21(1):171–182. doi: 10.1111/adb.12192. [DOI] [PMC free article] [PubMed] [Google Scholar]
  30. Miranda R, Jr, Ray L, Blanchard A, Reynolds EK, Monti PM, Chun T, Ramirez J. Effects of naltrexone on adolescent alcohol cue reactivity and sensitivity: An initial randomized trial. Addict Biol. 2014;19(5):941–954. doi: 10.1111/adb.12050. [DOI] [PMC free article] [PubMed] [Google Scholar]
  31. Miranda R, Jr, Treloar H. Emerging Pharmacologic Treatments for Adolescent Substance Use: Challenges and New Directions. Curr Addict Rep. 2016;3(2):145–156. doi: 10.1007/s40429-016-0098-7. [DOI] [PMC free article] [PubMed] [Google Scholar]
  32. Monti PM, Rohsenow DJ, Hutchison KE, Swift RM, Mueller TI, Colby SM, Abrams DB. Naltrexone's effect on cue-elicited craving among alcoholics in treatment. Alcohol Clin Exp Res. 1999;23(8):1386–1394. [PubMed] [Google Scholar]
  33. Muthen B, Asparouhov T. Bayesian structural equation modeling: a more flexible representation of substantive theory. Psychol Methods. 2012;17(3):313–335. doi: 10.1037/a0026802. [DOI] [PubMed] [Google Scholar]
  34. Muthén BO. Bayesian analysis in Mplus: A brief introduction. Retrieved from. 2010 http://www.statmodel.com/download/IntroBayesVersion3.pdf.
  35. Nutt DJ, Lingford-Hughes A, Erritzoe D, Stokes PR. The dopamine theory of addiction: 40 years of highs and lows. Nat Rev Neurosci. 2015;16(5):305–312. doi: 10.1038/nrn3939. [DOI] [PubMed] [Google Scholar]
  36. O'Malley SS, Krishnan-Sarin S, Farren C, Sinha R, Kreek MJ. Naltrexone decreases craving and alcohol self-administration in alcohol-dependent subjects and activates the hypothalamo-pituitary-adrenocortical axis. Psychopharmacology (Berl) 2002;160(1):19–29. doi: 10.1007/s002130100919. [DOI] [PubMed] [Google Scholar]
  37. Patriquin MA, Bauer IE, Soares JC, Graham DP, Nielsen DA. Addiction pharmacogenetics: a systematic review of the genetic variation of the dopaminergic system. Psychiatr Genet. 2015;25(5):181–193. doi: 10.1097/YPG.0000000000000095. [DOI] [PMC free article] [PubMed] [Google Scholar]
  38. Preacher KJ, Rucker DD, Hayes AF. Addressing Moderated Mediation Hypotheses: Theory, Methods, and Prescriptions. Multivariate Behav Res. 2007;42(1):185–227. doi: 10.1080/00273170701341316. [DOI] [PubMed] [Google Scholar]
  39. Preacher KJ, Zhang Z, Zyphur MJ. Alternative methods for assessing mediation in multilevel data: The advantages of multilevel SEM. Structural Equation Modeling: A Multidisciplinary Journal. 2011;18(2):161–182. [Google Scholar]
  40. Preacher KJ, Zyphur MJ, Zhang Z. A general multilevel SEM framework for assessing multilevel mediation. Psychol Methods. 2010;15(3):209–233. doi: 10.1037/a0020141. [DOI] [PubMed] [Google Scholar]
  41. Ray LA, Miranda R, Jr, Tidey JW, McGeary JE, MacKillop J, Gwaltney CJ, Monti PM. Polymorphisms of the mu-opioid receptor and dopamine D4 receptor genes and subjective responses to alcohol in the natural environment. J Abnorm Psychol. 2010;119(1):115–125. doi: 10.1037/a0017550. [DOI] [PMC free article] [PubMed] [Google Scholar]
  42. Rohsenow DJ, Monti PM, Rubonis AV, Sirota AD, Niaura RS, Colby SM, Abrams DB. Cue reactivity as a predictor of drinking among male alcoholics. J Consult Clin Psychol. 1994;62(3):620–626. doi: 10.1037//0022-006x.62.3.620. [DOI] [PubMed] [Google Scholar]
  43. Sacks JJ, Gonzales KR, Bouchery EE, Tomedi LE, Brewer RD. 2010 National and State Costs of Excessive Alcohol Consumption. Am J Prev Med. 2015;49(5):e73–79. doi: 10.1016/j.amepre.2015.05.031. [DOI] [PubMed] [Google Scholar]
  44. Sander T, Harms H, Dufeu P, Kuhn S, Rommelspacher H, Schmidt LG. Dopamine D4 receptor exon III alleles and variation of novelty seeking in alcoholics. Am J Med Genet. 1997;74(5):483–487. doi: 10.1002/(sici)1096-8628(19970919)74:5<483::aid-ajmg5>3.0.co;2-p. [DOI] [PubMed] [Google Scholar]
  45. Shiffman S. Conceptualizing analyses of ecological momentary assessment data. Nicotine Tob Res. 2014;16(Suppl 2):S76–87. doi: 10.1093/ntr/ntt195. [DOI] [PMC free article] [PubMed] [Google Scholar]
  46. Sobell LD, Sobell MD. Timeline follow-back: A technique for assessing self-reported alcohol consumption. In: Allen RLJ, editor. Measuring Alcohol Consumption. The Humana Press Inc.; 1992. [Google Scholar]
  47. Tidey JW, Monti PM, Rohsenow DJ, Gwaltney CJ, Miranda R, Jr, McGeary JE, Paty JA. Moderators of naltrexone's effects on drinking, urge, and alcohol effects in non-treatment-seeking heavy drinkers in the natural environment. Alcohol Clin Exp Res. 2008;32(1):58–66. doi: 10.1111/j.1530-0277.2007.00545.x. [DOI] [PMC free article] [PubMed] [Google Scholar]
  48. Valenta JP, Job MO, Mangieri RA, Schier CJ, Howard EC, Gonzales RA. mu-opioid receptors in the stimulation of mesolimbic dopamine activity by ethanol and morphine in Long-Evans rats: a delayed effect of ethanol. Psychopharmacology (Berl) 2013;228(3):389–400. doi: 10.1007/s00213-013-3041-9. [DOI] [PMC free article] [PubMed] [Google Scholar]
  49. van den Wildenberg E, Janssen RG, Hutchison KE, van Breukelen GJ, Wiers RW. Polymorphisms of the dopamine D4 receptor gene (DRD4 VNTR) and cannabinoid CB1 receptor gene (CNR1) are not strongly related to cue-reactivity after alcohol exposure. Addict Biol. 2007;12(2):210–220. doi: 10.1111/j.1369-1600.2007.00064.x. [DOI] [PubMed] [Google Scholar]
  50. Van Tol HH, Wu CM, Guan HC, Ohara K, Bunzow JR, Civelli O, Jovanovic V. Multiple dopamine D4 receptor variants in the human population. Nature. 1992;358(6382):149–152. doi: 10.1038/358149a0. [DOI] [PubMed] [Google Scholar]
  51. Wise RA. Drug-activation of brain reward pathways. Drug Alcohol Depend. 1998;51(1–2):13–22. doi: 10.1016/s0376-8716(98)00063-5. [DOI] [PubMed] [Google Scholar]

RESOURCES