Negative Mood Induction Increases Choice of Heroin Versus Food Pictures in Opiate-Dependent Individuals: Correlation With Self-Medication Coping Motives and Subjective Reactivity

Acute growth in negative affect is thought to play a major role in triggering relapse in opiate-dependent individuals. Consistent with this view, three lab studies have demonstrated that negative mood induction increases opiate craving in opiate-dependent individuals. The current study sought to confirm these effects with a behavioral measure of heroin seeking, and test whether the effect is associated with self-reported opiate use to cope with negative affect and subjective reactivity to mood induction. Participants were heroin-dependent individuals engaged with treatment services (n = 47) and control participants (n = 25). Heroin users completed a questionnaire assessing reasons for using heroin: negative affect, social pressure, and cued craving. Baseline heroin choice was measured by preference to enlarge heroin versus food thumbnail pictures in two-alternative forced-choice trials. Negative mood was then induced by depressive statements and music before heroin choice was tested again. Subjective reactivity was indexed by negative and positive mood reported at the pre-induction to post-test timepoints. Heroin users chose heroin images more frequently than controls overall ( p = .001) and showed a negative mood-induced increase in heroin choice compared to control participants (interaction p < .05). Mood-induced heroin choice was associated with self-reported heroin use to cope with negative affect ( p < .05), but not social pressure ( p = .39) or cued craving ( p = .52), and with subjective mood reactivity ( p = .007). These data suggest that acute negative mood is a trigger for heroin seeking in heroin-dependent individuals, and this effect is pronounced in those who report using heroin to cope with negative affect, and those who show greater subjective reactivity to negative triggers. Interventions should seek to target negative coping motives to build resilience to affective triggers for relapse.

Three studies have tested whether negative mood induction motivates opiate craving in opiate-dependent individuals. The study by Childress and colleagues (21) recruited 10 opiatedependent clients who had been abstinent for 30 days, exposed them to guided self-hypnosis of a depressive scene, and found that subjective opiate craving increased from pre-to post-induction. The second study by Hyman and colleagues (22) recruited 14 opiate-dependent individuals who had been detoxified and were undergoing naltrexone treatment. Exposure to guided imagery of a personalized stress situation increased subjective opiate craving from baseline, while exposure to neutral imagery had no effect. Positive correlations were found between stress effects on craving and subjective reactivity (anxiety, fear, and sadness). Finally, the third study by Stathopoulou and colleagues (23) recruited 76 opiate-dependent individuals who had been on methadone maintenance for 4 months and exposed them to short video clips to induce sadness. After excluding 10 participants who showed no increase in subjective negative mood, it was found that the increase in craving from pre to post was related to subjective negative mood, and was moderated by anxiety such that this relationship was only significant in those with high anxiety sensitivity. There was no relationship between mood-induced craving and selfreported opiate use to cope with negative affect. Overall, this work provides preliminary support for the notion that acute negative mood is an important trigger for opiate seeking.
One limitation of the existing literature is that there is no demonstration of negative mood induction increasing a behavioral measure of heroin-seeking behavior. The three prior studies all measured opiate craving which has an unknown relationship to behavior (24). To address this limitation, we employed a pictorial choice procedure in which opiate-dependent individuals had the choice to enlarge heroin versus food thumbnail pictures in a series of two-alternative forced choice trials. Prior studies have validated the pictorial drug choice task by demonstrating that percent drug choice was increased in drug users versus non-users, or as a function of dependence level in the drug user group, in cocaine (25,26), alcohol (27,28), and tobacco users (28,29), and was sensitive to the motivating effects of negative mood induction (10,27,30). In the current study, opiate users and control participants completed a concurrent pictorial choice task for heroin versus food pictures before and after mood induction. The first prediction was that heroin users would choose heroin images more frequently than control participants, validating the pictorial choice measure as an index of heroin value. The second prediction was that heroin users would show a mood-induced increase in heroin choice whereas control participants would not, suggesting that acute negative affect is an important trigger for heroin-seeking behavior.
The second limitation of the existing literature is that individual sensitivity to mood-induced opiate craving remains obscure. The two studies by Hyman et al. and Stathopoulou et al. (22,23) found that mood-induced opiate craving was associated with subjective mood reactivity, consistent with a range of other induction studies [e.g., Refs. (27,(31)(32)(33)(34), but see Refs. (35,36)]. Consequently, the third prediction of the study was that moodinduced heroin choice would be associated with subjective mood reactivity. More interestingly, however, Stathopoulou and colleagues (23) found that mood-induced opiate craving was not associated with self-reported opiate use to cope with negative affect. This finding is at odds with multiple studies that show that coping motives are associated with greater sensitivity to mood-induced drug-motivated behavior [Refs. (5,7,15,16,27,(37)(38)(39)(40)(41)(42)(43); but see Refs. (30,40,44)]. The fourth prediction of the current study, therefore, was that mood-induced heroin choice behavior would be greater in opiate users who reported using to cope with negative affect. Sensitivity to negative affect-triggered heroin seeking could be an important mechanism driving relapse (45,46).

Participants and Procedures
Participation was open to males and females aged 18-65 being treated for current heroin addiction by opioid medication at the Royal Prince Alfred (RPA) Hospital Drug Health Clinic in Sydney, Australia. Data were collected from 47 opiatedependent outpatients (male = 32, female = 15) after they received opiate medication. In total, 2 participants (4.3%) were aged 19-24, 14 (29.8%) were 25-39, 16 (34.0%) were 40-49, and 15 participants (31.9%) were 50+ years of age. Thirty-five participants were receiving methadone (mean dose = 79 mg), 2 participants received buprenorphine (mean dose = 6 mg), and 10 participants were receiving suboxone (mean dose = 21 mg). The majority of these participants were currently unemployed, educated to high school level, and single. Eligibility criteria included: 1) current attendance in treatment for heroin addiction, 2) over 18 years of age, 3) English speaking, and 4) receiving opiate medication for the last 30 days. Healthy controls that did not have a history of opiate addiction were recruited via word of mouth from the community. Exclusion criteria included history of substance dependence or any other DSM-IV axis I disorders. Participants were matched for gender (opiate users = 33% female; controls = 48% female, Fishers exact p = .21), and age, t(34.26) = 1.66, p = .11. A chi-square comparing three categories of educational attainment (below high school, high school, greater than high school achievement) was non-significant, χ 2 (2, 71) = 4.48, p = .11, suggesting the two groups were matched for educational attainment. One opiate-using participant was excluded for showing an extremely outlying reduction in heroin choice from pre-to post-induction (>3 times the inter quartile range), leaving 46 opiate users and 25 control participants in the analyzed data set. The study was approved by the Western Sydney University Human Research Ethics Committee, and participants provided informed written consent.

Questionnaires
Participants reported gender and age. Heroin users completed the Reasons for Drinking Questionnaire (RFDQ), adapted for heroin use (47). Instructions stated "The following 16 questions are a list of reasons why people take illicit opiates. Please rate each of these reasons on how important each is for you. " Within the questionnaire, the word "alcohol" from the original was replaced with the word "heroin. " Responses were scored on a 1-10 scale ranging from "not at all important" to "very important. " The RFDQ has three subscales reflecting heroin use to cope with negative affect, social pressure, and cued craving, obtained by averaging relevant items, giving a subscale score range of 1-10. We adapted the RFDQ because the drinking to cope subscale in the original version has been shown to be associated with greater sensitivity to negative mood-induced alcohol choice in two of our prior studies with student drinkers in a task similar to the present (27,37).

Mood-Induced Heroin Picture Choice Task
The trial structure and timings of the heroin picture choice task are shown in Figure 1. At baseline, participants freely chose to enlarge a heroin or food thumbnail picture with a left or right key press, over 32 trials. In each trial, a heroin and food thumbnail was presented randomly in the left or right position, sampled from a set of 28 of each image type (obtained online from noncopyrighted images). Following baseline choice, pre-induction subjective mood was measured by participants reporting the extent to which they currently felt five negative (jittery, upset, distressed, sad, irritable) and five positive emotions (enthusiastic, happy, excited, inspired, alert), randomly ordered, on a five-point scale ranging from "not at all" to "extremely. " Sad music was then played through headphones (Barber's Adagio for Strings), and participants were instructed to carefully consider 16 negative statements (e.g., "I don't think things are ever going to get better") randomly ordered [for full list, see Ref. (34)]. The heroin choice test comprised 32 trials identical to baseline, except that the sad music continued to play and a negative statement (randomly selected from the set of 16) was presented prior to each choice (the same picture set was used as at baseline). Post-induction subjective mood was then measured in the same way as before.  Figure 2B shows the mean (and SEM) subjective negative and positive mood reported pre-induction and post-test. ANOVA on these data with the variables group (heroin users, controls), mood state (negative, positive), and timepoint (pre, post) yielded a significant interaction between group, mood state, and timepoint, F(1,69) = 12.40, p = .001, η p 2 = .152. In heroin users, there was a significant main effect of mood state, F(1,45) = 6.06, p = .02, η p 2 = .119, but no effect of timepoint, F(1,45) = .12, p = .73, η p 2 = .003, or interaction between mood state and timepoint, F(1,45) = .91, p = .34, η p 2 = .020. By contrast, in controls, there was a significant main effect of mood state, F(1,24) = 39.22, p < .001, η p 2 = .620, and a significant interaction between mood state and timepoint, F(1,24) = 26.79, p < .001, η p 2 = .528, with negative mood increasing significantly from pre to post, F(1,24) = 17.23, p < .001, η p 2 = .419, and positive mood decreasing significantly from pre to post, F(1,24) = 16.06, p < .001, η p 2 = .401. Thus, heroin users showed no change in subjective mood following mood induction, whereas controls showed the appropriate change in mood states.

Correlations
As shown in Figure 2C, the change in heroin choice from the pre to post mood induction correlated positively with RFDQ heroin use to cope with negative affect, r = .29, p < .05, but not with RFDQ heroin use due to social pressure, r = −.13, p = .39, or cued craving, r = .10, p = .52. Furthermore, as shown in Figure 2D, heroin users' change in heroin choice also correlated with the change in negative mood from pre-to post-induction, r = .39, p = .007, but not with the change in positive mood r = −.19, p = .20. RFDQ heroin use to cope with negative affect did not correlate significantly with the change in negative mood, r = .28, p < .06. Thus, heroin users' change in heroin choice from baseline to test was amplified in those who reported heroin use to cope with negative affect, and those who reported the greatest increase in negative mood following induction.

DISCUSSION
The study found that opiate-dependent individuals chose heroin over food images more frequently than control participants. This  Figure 1). Opiate-dependent participants showed a higher rate of heroin choice overall compared to control participants, and showed a mood induced increase in heroin choice at test, whereas controls did not. (B) Subjective negative and positive mood states reported at pre-induction and post-test timepoints (see Figure 1). Opiate-dependent participants showed no overall change in subjective mood states, whereas control participants showed an increase in negative mood and a decrease in positive mood following mood induction. (C) Scatterplot and regression slope relating the mood-induced change in percent heroin choice to self-reported opiate use to cope with negative affect in opiate-dependent participants. (D) Scatterplot and regression slope relating the mood-induced change in percent heroin choice to self-reported change in negative mood in opiate-dependent participants.
accords with results from two studies with cocaine-dependent individuals, who chose cocaine over pleasant pictures more frequently than control participants (25,26). Percent drug choice has also been shown to increase with dependence level within drug user groups (25)(26)(27)(28)(29). These findings suggest that the pictorial drug choice task is a valid assay of the relative value ascribed to the drug by drug users. The pictorial choice task may have the advantage over subjective craving as an outcome measure, in being more readily translatable to animal models that also use behavioral measures rather than subjective report (48,49). This procedure also has an advantage over human concurrent drug self-administration procedures (50)(51)(52) in not requiring actual consumption, and so is technically simpler and ethically acceptable for clients who are currently abstinent. Finally, the pictorial drug choice task is superficially similar to attentional bias tasks, but appears more reliable in detecting group differences, and correlations with dependence severity (53). The second finding was that negative mood induction increased heroin choice in opiate-dependent individuals but not control participants. This extends prior induction studies with opiate users (21)(22)(23), by including control participants to demonstrate the specificity of the mood induction effect. The finding also confirms that negative mood acts as a trigger for heroin-seeking behavior (and not just craving), as has been found with other drug classes including alcohol (4-7), tobacco (8)(9)(10), and cocaine (11)(12)(13)(14).
Sensitivity to mood-induced heroin choice was also found to correlate with subjective changes in negative mood, consistent with two prior opiate studies (22,23) and induction studies with other drug classes (27,(31)(32)(33)(34). These findings accord with the prediction of affective negative reinforcement theory (54) in suggesting that the affective change produced by the induction procedure was responsible for the change in heroin-seeking behavior.
Finally, sensitivity to mood-induced heroin choice was found to correlate with self-reported opiate use to cope with negative affect, but not other opiate use motives (social pressure and cued craving). This finding contradicts the study by Stathopoulou and colleagues (23) which found no association between moodinduced opiate craving and opiate use to cope with negative affect, but corroborates multiple induction studies with other drug classes that have found this same association (5,7,15,16,27,(37)(38)(39)(40)(41)(42)(43). We may therefore accept our association as a true positive. It is possible that coping motives increase the risk of dependence by conferring sensitivity to negative affective triggers for drug-seeking behavior (45,46).
We might further speculate that individual sensitivity to mood-induced heroin choice is a risk factor for relapse. The basis for this claim is that such sensitivity is associated with relapse risk in alcohol- (15)(16)(17)(18) and cocaine-dependent individuals (19,20). With respect to opiate users, poorer stress tolerance (55) and abnormal cortisol (56) predict poorer treatment engagement or earlier lapse, and preliminary evidence suggests that learning to cope with negative affect may promote abstinence (57). The implication is that sensitivity to negative mood-induced heroin-seeking is also a risk factor for relapse, and that treatments targeting this sensitivity may have efficacy for maintaining abstinence.
One limitation of the current study was that we did not observe an overall change in subjective mood following negative mood induction in the heroin user group, whereas controls did show changes to self-reported positive and negative mood. Despite this, the increase in heroin choice at test for the heroin user group, as well as the correlation between this effect and their change in negative mood, indicated that the mood induction procedure did impact the heroin user group. However, these effects were small and were perhaps reduced by the opiate replacement medications taken shortly before the experiment, similarly to acute alcohol, which has been shown to reduce mood induction effects (58). Future studies may employ a stronger mood induction procedure that produces a reliable change in subjective mood in heroin users, and a larger magnitude of effect on heroin choice behavior.
The second limitation was that we did not employ a control condition to determine whether the change in heroin picture choice was due to the mood induction or time. Previous studies have shown that drug choice remains stable over time then jumps following induction (30). Similarly, percent heroin choice in heroin users of the current study was stable across the two halves of the baseline phase (means = 39% and 39%, respectively), then jumped following induction and was stable across the two halves of the test phase (means = 48% and 46%, respectively). These data, plus the correlation between subjective mood and heroin choice, suggest that the increase in heroin choice was caused by negative mood induction and not by time.
The third limitation was that we could not obtain indices of psychiatric state in the two groups, because we had access to drug-using clients for an extremely short period during their hospital visit. As a consequence, we are unable to test whether the differential mood induction effect between the two groups was due to drug user status, or confounded psychiatric symptoms, such as anxiety or depression, which are known to be associated with greater sensitivity to mood induction effects on alcohol seeking (10,27).

ETHICS STATEMENT
This study was approved by the Western Sydney University Human Research Ethics Committee.

AUTHOR CONTRIBUTIONS
LHo designed the procedure and wrote the first draft of the manuscript. LHa and AB programmed the task and contributed to the analysis. JM and SC ran the participants and contributed to the analysis. GW and AM oversaw the running of the experiment. All authors contributed to the writing of the manuscript.