c

Rnmdmi

Cai

ARCHIVAL REPORT

Predictors of Marijuana Relapse in the HumanLaboratory: Robust Impact of Tobacco CigaretteSmoking StatusMargaret Haney, Gillinder Bedi, Ziva D. Cooper, Andrew Glass, Suzanne K. Vosburg, Sandra D. Comer,and Richard W. Foltin

Background: Few marijuana smokers in treatment achieve sustained abstinence, yet factors contributing to high relapse rates are unknown.

Methods: Study 1: data from five inpatient laboratory studies assessing marijuana intoxication, withdrawal, and relapse were combined toassess factors predicting the likelihood and severity of relapse. Daily, nontreatment-seeking marijuana smokers (n � 51; 10 � 5 marijuanacigarettes/day) were enrolled. Study 2: to isolate the effects of cigarette smoking, marijuana intoxication, withdrawal, and relapse wereassessed in daily marijuana and cigarette smokers (n � 15) under two within-subject, counter-balanced conditions: while smoking tobacco

igarettes as usual (SAU), and after at least 5 days without cigarettes (Quit).

esults: Study 1: 49% of participants relapsed the first day active marijuana became available. Tobacco cigarette smokers (75%), who wereot abstaining from cigarettes, were far more likely to relapse than non-cigarette smokers (odds ratio: 19, p � .01). Individuals experiencingore positive subjective effects (i.e., feeling “high”) after marijuana administration and those with more negative affect and sleep disruption

uring marijuana withdrawal were more likely to have severe relapse episodes (p � .05). Study 2: most participants (�87%) relapsed toarijuana whether in the SAU or Quit phase. Tobacco cigarette smoking did not significantly influence relapse, nor did it affect marijuana

ntoxication or most symptoms of withdrawal relative to tobacco cessation.

onclusions: Daily marijuana smokers who also smoke cigarettes have high rates of marijuana relapse, and cigarette smoking versus recentbstinence does not directly influence this association. These data indicate that current cigarette smoking is a clinically important marker for

ncreased risk of marijuana relapse.

ptimdpttl(d

ou(ilotjwSpep

M

S

md

Key Words: Cannabinoids, cannabis, self-administration, tobacco,treatment, withdrawal

M arijuana-use disorders are ubiquitous worldwide (1), andan increasing number of marijuana smokers are seekingtreatment (2–5). In the US, 6.9 million individuals smoke

near-daily (�20 days/month) (6), and approximately 16% of pa-tients entering treatment have a diagnosis of primary marijuanaabuse (7). Treatment outcome among these patients is poor. Clini-cal trials using psychological (8 –10), behavioral (11,12) and phar-macological (13) interventions report that only 15%–37% of pa-tients achieve continued abstinence. Given these poor outcomes, aclear understanding of the factors associated with high relapserates is crucial for improving marijuana treatment outcomes.

One factor presumed to contribute to marijuana relapse rates isa withdrawal syndrome. Controlled human laboratory studies haveshown that replacing active marijuana with placebo marijuana forseveral days produces a time-dependent increase in irritability, anx-iety, sleep disruption and decreased appetite (14 –19). Replace-ment with oral �9-tetrahydrocannabinol (THC; dronabinol) underdouble-blind conditions selectively reduces withdrawal symptomsin laboratory (16,20) and clinical (13) settings, demonstrating thepharmacological specificity of marijuana withdrawal.

From the Division on Substance Abuse, New York State Psychiatric Instituteand Department of Psychiatry, College of Physicians and Surgeons ofColumbia University, New York, New York.

Address correspondence to Margaret Haney, Ph.D., Professor of ClinicalNeurobiology, Division on Substance Abuse, New York State PsychiatricInstitute, College of Physicians and Surgeons of Columbia University,1051 Riverside Drive, Unit 120, New York, NY 10032; E-mail:mh235@columbia.edu.

aReceived May 9, 2012; revised Jul 13, 2012; accepted Jul 27, 2012.

0006-3223/$36.00http://dx.doi.org/10.1016/j.biopsych.2012.07.028

In an effort to improve treatment options, we have conductedlacebo-controlled studies testing the effects of a range of medica-

ions on marijuana intoxication, withdrawal, and relapse. Because mar-juana is administered, only individuals explicitly not seeking treat-

ent for their marijuana use are enrolled. Thus, relapse is operationallyefined as marijuana self-administration, at a financial cost, after aeriod of abstinence (17,21–23): participants purchase self-adminis-

ered marijuana using their study earnings. Although the motivationo relapse or abstain might differ from that of a patient in treatment,aboratory measures of self-administration predict clinical outcome13,21,24,25) and thus contribute controlled and clinically relevantata on factors affecting marijuana use.

After data were combined from the placebo medication phasesf five studies, a pattern emerged: despite closely similar marijuana-se and demographic profiles, only half of the participants relapsed

i.e., chose to pay the high financial cost to resume marijuana smok-ng after several days of abstinence). Given the detailed data col-ected on intoxication, withdrawal, and demographic variables, thebjective of Study 1 was to evaluate what predicted: 1) the odds

hat an abstinent marijuana smoker would relapse to active mari-uana use, and 2) the severity of relapse (i.e., how much marijuana

as self-administered among those who relapsed). The results fromtudy 1 showed that tobacco cigarette-smoking status robustlyredicted marijuana relapse. Thus, Study 2 evaluated the directffects of tobacco cigarette smoking on marijuana relapse com-ared with tobacco cessation.

ethods and Materials

tudy 1Data from five studies (n � 51 participants) using a similar model of

arijuana relapse were analyzed. Four medication studies (lofexidine,ronabinol, quetiapine, mirtazapine, nabilone) (17,22,23; M. Haney et

l., unpublished data, 2012) enrolled participants in up to four inpa-

BIOL PSYCHIATRY 2013;73:242–248© 2013 Society of Biological Psychiatry

gbc1

plepc

uF(r

tastbua((cdh

rp

oj(bw9mnostmmihlmfisPc

TLP

D

A

M

M. Haney et al. BIOL PSYCHIATRY 2013;73:242–248 243

tient phases, each testing a different medication dose in counter-bal-anced order and allowing time between phases for medication clear-ance. Only the placebo condition was analyzed. One non-medicationstudy (n�9) tested the effects of marijuana-paired cues and marijuanaexposure on relapse; only the condition in which neither cues normarijuana preceded the decision to relapse was analyzed herein.

Participants. Healthy marijuana smokers solicited through ad-vertisements in New York were enrolled from August 2006 to Sep-tember 2011. Inclusion criteria included: 21–50 years of age andcurrent marijuana use (minimum two marijuana cigarettes/day, 5days/week). No participant could: 1) regularly use any other illicitdrug or be dependent on alcohol; 2) meet criteria for a current AxisI disorder requiring medical intervention; 3) be taking medication;or 4) be seeking marijuana treatment. Assessments of health in-cluded physical examination, psychiatric evaluation, electrocardio-gram, urinalysis, and blood chemistry. Participants signed a consentform approved by The New York State Psychiatric Institute Institu-tional Review Board, which described the study, outlined possiblerisks, and indicated that two different-strength marijuana ciga-rettes would be tested. Volunteers were compensated for partici-pation.

Procedures. Participants, in groups of three or four, lived in aresidential laboratory for 9 –11 days, with four private rooms, twobathrooms, and a recreational area; video- and audio-monitoringsystems allowed for continuous observation of participants (14).Before the inpatient stay, participants sampled an active marijuanacigarette (labeled “Dose A”) and a placebo marijuana cigarette (la-beled “Dose B”) using procedures described in the following sec-tion. They were told that the strength of Dose A and Dose B wouldnot change and that they should pay attention to how each dosemade them feel, because they would later make decisions withregard to self-administration.

On each inpatient day, participants completed one sleep scale,eight mood scales (17), and six 30-min psychomotor/cognitive taskbatteries (26) between 8:15 AM and 11:30 PM. A variety of food to beconsumed ad libitum was available. The recreation area was availableat lunchtime and from 5:00 to 10:00 PM. At 11:30 PM, participants were

iven faux money representing a portion of their earnings that coulde used to purchase marijuana on self-administration days or ex-hanged for cash upon study completion. Lights were turned off by2:00 midnight.

Marijuana. Marijuana was administered with a cued-smokingrocedure, where duration of inhalation, holding smoke in the

ungs, and inter-puff interval was controlled (27). Marijuana wasither experimenter-administered at no cost or was available tourchase for self-administration; participants were informed of theondition for that day at 9:50 AM each morning.

During experimenter-administered days (first day of each inpa-tient phase), participants smoked three puffs of active marijuana(3.3%– 6.2% THC [depending on study]; Dose A) six times through-out the day. The purpose of this day was to standardize marijuanaexposure before abstinence. On the subsequent 2–3 inpatient days,Dose B (placebo marijuana) was available for self-administration(withdrawal), followed by 1– 4 days when Dose A was available forself-administration (relapse). During self-administration days, par-ticipants had six opportunities to purchase up to three puffs of theavailable dose. The cost was $9 –10 for the first puff of the day and$2–3 for all subsequent puffs, depending on the study; prices werederived from pilot data showing that marijuana self-administrationafter abstinence varies as a function of cost (22). Participants paidfor marijuana puffs with money representing their earnings. Theysmoked self-administered marijuana in private to keep other par-

ticipants blind to their choice.

Sleep. Subjective ratings of sleep were obtained each morningsing a seven-item visual analog scales (VAS) sleep questionnaire.or objective sleep measurement, participants wore the Nightcap

n � 8) or the Actiwatch Activity Monitoring System (n � 43) (Respi-onics, Bend, Oregon).

Subjective-Effects Battery. A 44-item subjective-effects ques-ionnaire— comprising a series of 100-mm VAS anchored with “Nott all” (0 mm) and “Extremely” (100 mm)—included mood, physicalymptom, and drug effect descriptors; participants rated the extento which each descriptor applied to them at that moment. On theasis of cluster analyses, we employed arithmetic means of individ-al item scores to produce seven subscales: miserable (e.g., “miser-ble,” “irritable”); anxious (e.g., “anxious,” “on edge”); bad effecte.g.,“muscle pain,” “chills”); sedated (e.g., “sleepy,” “tired”); sociale.g., “friendly,” “talkative”); high (e.g., “high,” “good effect”); andonfused (e.g., “forgetful,” “confused”). We also analyzed ratings ofrug craving with VAS ratings of “I Want Marijuana,” “I Want Alco-ol,” and “I Want Cigarettes.”

Tobacco Cigarette Smoking. The number of tobacco ciga-ettes smoked was recorded by counting cigarette butts in eacharticipant’s ashtray each evening.

Data Analysis. The primary outcome measure was the numberf marijuana puffs (0 –18) purchased on the first day active mari-

uana was available after abstinence. Of the 51 participants, 2651%) bought zero puffs of marijuana and 25 participants (49%)ought at least 1 marijuana puff. Puffs of marijuana were analyzedith a Poisson hurdle regression model with PROC NLMIXED in SAS

.2 (SAS, Cary, North Carolina). The Poisson hurdle, or two-partodel, models whether the number puffs of marijuana is zero or

on-zero with logistic regression and models the number of puffsf marijuana purchased with a truncated-at-zero Poisson regres-ion, conditional on non-zero outcomes. This model reflects thewo-stage decision-making process present: deciding to self-ad-

inister the first puff of marijuana, and then deciding how manyarijuana puffs to self-administer. Table 1 lists the covariates exam-

ned in both parts of the model; those included in the final modelad a p value � .05. The logistic regression component used a logit

ink function and predicted the odds of abstinence (i.e., buying zeroarijuana puffs). The factors included in the logistic portion of the

nal model were: 1) whether participants were tobacco cigarettemokers, and 2) age of onset of daily marijuana use. The truncatedoisson regression part used a log link function and predicted theonditional incidence rates of marijuana puffs purchased. The fac-

able 1. Factors Included in Logistic Regression Equation Predictingikelihood of Relapse and Truncated Poisson Regression Equationredicting Severity of Relapse

emographic VariablesAgeAge of daily marijuana useYears smoking marijuana regularlyNumber of marijuana cigarettes smoked/dayCigarette smoker (yes/no)Cigarettes/day (#)Alcohol drinks/week

cute Marijuana EffectsPeak ratings of “High”arijuana Withdrawal SymptomsPeak ratings of marijuana cravingPeak cluster ratings of “Miserable”Peak cluster ratings of “Anxious”Peak ratings of fell asleep early, sleep satisfaction, woke Early

Sleep latency, % time asleep

www.sobp.org/journal

pcwss

cersdwtedpRH

R

S

d

p1w1lT1

mucc9Ttts5u

lmot

T

NREAMMACCAAE

244 BIOL PSYCHIATRY 2013;73:242–248 M. Haney et al.

tors included in the Poisson portion of the model were: 1) peakratings of “high” after experimenter-administered marijuana; 2)peak cluster ratings of miserable during marijuana withdrawal; and3) sleep latency during marijuana withdrawal. Akaike informationcriterion was used to assess goodness of fit.

Study 2Study 1 demonstrated that the factor having the largest impact

on marijuana relapse was tobacco cigarette-smoking status. Partic-ipants were free to smoke tobacco cigarettes, so marijuana relapsedid not reflect an effort to attenuate nicotine withdrawal (28). Onepossibility is that nicotine directly enhances the effects of marijuanaand that repeatedly pairing tobacco cigarettes and marijuana re-sults in one drug cueing use of the other (29,30). There is closeoverlap in the distribution of nicotinic and cannabinoid receptors(31), and pretreatment with a nicotine patch (21 mg) has beenshown to increase marijuana “high” (32). Furthermore, some indi-viduals report smoking tobacco cigarettes immediately after mari-juana, purportedly to produce greater intoxication (33).

Thus, the purpose of Study 2 was to test the direct effects oftobacco cigarette smoking on marijuana relapse. We used the samemodel of marijuana intoxication, withdrawal, and relapse describedin Study 1 to compare daily marijuana and cigarette smokers acrosstwo inpatient phases: when they were smoking cigarettes as usual(SAU), and after they had undergone a period without smokingcigarettes (Quit). The order of Quit and SAU phases was counter-balanced.

Participants. Screening procedures and inclusion criteria wereidentical to Study 1 except that eligible volunteers smoked �5cigarettes/day and could not be seeking tobacco cessation treat-ment.

Procedures. Each 8-day inpatient SAU or Quit phase was pre-ceded by at least 7 outpatient days to allow time to either quit orresume smoking tobacco cigarettes. For the Quit phase, our objec-tive was to have participants quit smoking tobacco cigarettes atleast 1 week before study onset to reduce the impact of peaknicotine withdrawal on measures of marijuana effects, althoughsome symptoms would likely persist (34). Participants underwentmodified contingency management procedures to foster tobaccocessation. They were given a date to quit tobacco cigarettes andthen visited the lab every 2–3 days to contribute a urinary samplefor cotinine measurement (NicAlert; Global Business Support Sys-tems, San Diego California) and a breath sample for carbon monox-ide measurement and to complete ratings of nicotine withdrawal.Participants were given $50 for each outpatient day in which theyachieved low cotinine levels (�100 ng/mL) (35). They were told toavoid nicotine replacement products and smoking marijuana inblunts (cigar tobacco leaves), which can increase urinary cotininelevels (36). For ethical reasons, participants who had the Quit phasefirst were not instructed to resume SAU; they were told that theywould be re-enrolled, regardless of tobacco smoking status. How-ever, all resumed SAU.

As in Study 1, participants smoked experimenter-administeredactive marijuana (5.5% THC) on the first inpatient day. For the next 3days, inactive marijuana (.0% THC) was available for self-administra-tion (withdrawal), followed by 4 days when 5.5% THC was availablefor self-administration (relapse).

Data Analysis. Repeated measures analyses of variance withlanned comparisons were used to determine the effect of tobaccoigarette smoking on the direct effects of marijuana, marijuanaithdrawal, and relapse. An advantage of using repeated-mea-

ures, within-subjects designs is that they are well-powered, due to

ubstantial correlations between levels. Behavioral outcomes in- e

www.sobp.org/journal

luded: the number of marijuana puffs purchased, peak subjectiveffects, drug craving, task performance, number of tobacco ciga-ettes smoked per day, and objective and subjective sleep mea-ures. There were two within-group factors: tobacco smoking con-ition (SAU, Quit), and marijuana condition. Planned comparisonsere conducted to determine whether the Quit versus SAU condi-

ion affected marijuana: 1) intoxication when active marijuana wasxperimenter-administered; 2) withdrawal (mean peak values onays 2 and 3 of abstinence); and 3) relapse (number of marijuanauffs purchased on the first day of active marijuana availability).esults were considered statistically significant at p values � .05.uynh-Feldt corrections were used, when appropriate.

esults

tudy 1Participant Characteristics. Table 2 portrays demographic

ata on the participants (n � 51) included in the analysis.Odds of Marijuana Relapse. Non-cigarette smokers had ap-

roximately 19� the odds of not relapsing to marijuana (odds ratio:9.02; 95% confidence interval [CI]: 2.18 –165.95, t � 2.73, p � .009),ith age at daily use held constant. This effect is portrayed in Figure

, which shows the number of daily marijuana smokers who re-apsed as a function of their tobacco cigarette use. As shown inable 2, 75% of participants smoked tobacco cigarettes (range:–18 cigarettes/day) and continued to do so while inpatient.

The only other significant predictor was the age when dailyarijuana use was initiated, with those who started daily marijuana

se later in life being more likely to relapse. The odds of not pur-hasing any marijuana were estimated as 87.3% of what it would beompared with starting marijuana use 1 year older (odds ratio: .87;5% CI: .77–.99, t � �2.15, p � .036), cigarette smoking held fixed.able 2 shows that, on average, participants were 18 years old whenhey started daily marijuana use (range: 9 – 40 years). Everyone inhe non-relapsing group was under the age of 25 when they startedmoking marijuana every day, whereas the relapse group includedindividuals who were older when commencing daily marijuana

se (28 – 40 years).Severity of Marijuana Relapse. Among the group who re-

apsed, three factors predicted relapse severity (i.e., the number ofarijuana puffs self-administered after a period of abstinence). All

ther factors were held fixed when each of these factors was inves-igated. The first factor was peak cluster ratings of “High” after

able 2. Demographic Characteristics of Participants

Study 1 Study 2

umber of Participants 51 (46 M; 5 F) 15 (13 M; 2 F)ace (Black/White/Mixed/Pacific Islander) 41/7/2/1 9/2/4thnicity (Hispanic/Non-Hispanic) 11/40 10/5ge (Years) 28 � 7 30 � 6arijuana Use, n (Days/Week) 6.9 � .4 6.9 � .3arijuana Cigarettes/Day 9.9 � 5.6 8.9 � 5.7ge Beginning Daily Marijuana Use 17.9 � 5.8 16.1 � 4.5igarette Smokers, n 38 15igarettes/Day 9.0 � 4.8 10.3 � 3.8lcohol Drinkers, na 29 7lcohol (Drinks/Week) 4.4 � 3.5 12.3 � 8.0ducation (Years) 12.6 � 2.4 12.5 � 1.3

Data are presented as means (� SD) or as frequency.F, female; M, male.aOnly included those reporting at least 1 drink/week.

xperimenter-administered marijuana (range: 0 –100 mm). For ev-

Til8pdea2

S

dSromsc

tm6

sSdmwmwtpt

smocttDc(

meb

M. Haney et al. BIOL PSYCHIATRY 2013;73:242–248 245

ery 10-mm increase in peak ratings of “High,” the estimated numberof puffs purchased during the relapse increased by 12.4% (Inci-dence Rate Ratio [IRR] � 1.12; 95% CI: 1.05–1.20, t � 3.50, p � .001).

he second predictive factor was an objective sleep measure dur-ng marijuana withdrawal. For every 10-min increase in sleep onsetatency, the estimated number of puffs purchased increased by.1% (IRR � 1.08; 95% CI: 1.03–1.13, t � 3.31, p � .002). The thirdredictive factor was cluster ratings of “Miserable” during the with-rawal phase. For every 10-mm increase in “Miserable” ratings, thestimated number of puffs purchased after a period of marijuanabstinence increased by 4.3% (IRR � 1.04; 95% CI: 1.002–1.09, t �.13, p � .038).

Relapsers Non-relapsers

0

5

10

15

20

25

Num

ber

Smoker Non- Smoker Non- Smoker

Smoker

Figure 1. Total number of participants in Study 1 (n � 51) relapsing toarijuana (MJ) as a function of tobacco cigarette smoking status. “Relaps-

rs” (left panel) purchased at least one puff of active MJ on the first day itecame available. “Smokers” smoked at least one tobacco cigarette/day.

Slept Well

Rat

ings

(m

m)

Hou

rsR

atin

gs (

mm

)

Hours S

####

####

Woke Ofte

Intoxication Withdrawal

Fell Asleep Easily

Intoxication

Intoxication

SAU Quit

Rat

ings

(m

m)

25

50

75

100

**

0

25

50

75

100

SAU QuitSAU Quit

Intoxication Withdrawal

6.5

7.0

7.5

8.0

SAUSAU Quit0

SAU QuitSAU Quit6.0

tudy 2Participant Characteristics. Table 2 portrays demographic

ata on the 15 participants who completed the study. During theAU phase, participants averaged 10 –12 tobacco cigarettes/day,egardless of marijuana condition. At the onset of the Quit phase, 13f the 15 participants met criteria for tobacco cessation (carbononoxide: � 4 ppm and/or cotinine � 100 ng/mL); 2 participants

howed a reduction in cigarette use but did not achieve eitherriteria before move-in.

Marijuana Intoxication. Tobacco cigarette smoking condi-ion did not significantly affect peak ratings of “High” when active

arijuana was experimenter-administered (SAU: 67 � 7 mm; Quit:8 � 6 mm).

Marijuana Withdrawal. Craving for tobacco cigarettes wasignificantly higher during the Quit phase (55 � 6 mm) relative toAU [45 � 8 mm: F (1,98) � 5.53, p � .03] during marijuana with-rawal, but cigarette smoking condition did not significantly affectarijuana craving or the mood symptoms characterizing marijuanaithdrawal. Figure 2, which portrays sleep ratings as a function ofarijuana and tobacco cigarette condition, shows that marijuanaithdrawal significantly worsened sleep ratings relative to intoxica-

ion. Tobacco cigarette smoking had few effects on sleep, althougharticipants reported waking more often during the Quit phase

han SAU during marijuana withdrawal.Marijuana Relapse. Tobacco cigarette smoking status did not

ignificantly influence marijuana relapse. On the first day of activearijuana availability, 93% of participants purchased at least 1 puff

f marijuana in the SAU condition, compared with 87% in the Quitondition; the number of puffs purchased was identical (8 � 1) inhe two conditions. By contrast, placebo marijuana self-administra-ion was significantly influenced by tobacco cigarette condition.uring abstinence from active marijuana, participants in the Quit

ondition purchased significantly more placebo marijuana puffs3 � 1) compared with SAU [1 � 1; F (1,84) � 3.97, p � .05].

##

##*

hdrawal

#

thdrawal

Quit

Figure 2. Sleep ratings in Study 2 as a function of marijuanacondition (Intoxication vs. Withdrawal) and tobacco ciga-rette smoking condition (smoking tobacco cigarettes asusual [SAU] vs. tobacco cessation [Quit]). Intoxication: meanpeak ratings after repeated marijuana administration (5.5%).Withdrawal: mean peak ratings on the second and third dayof marijuana abstinence. *Significant difference betweenSAU and Quit condition (*p � .05). Significant difference be-tween Intoxication and Withdrawal condition, #p � .05, ##p �.01. Each bar represents 15 participants.

lept

#

n

#

Wit

Wi

SAU

Quit

www.sobp.org/journal

lcr(

ctDtawndaitda

Atcopephccmetu

stttrojt

gimtmadwoyte

Frchop

246 BIOL PSYCHIATRY 2013;73:242–248 M. Haney et al.

Discussion

Study 1 evaluated the factors predicting whether daily mari-juana users who have gone through several days of marijuanaabstinence will relapse, defined in this laboratory model as paying afinancial cost to reinitiate active marijuana smoking on the first dayof its availability. The results show that the factor best predictingmarijuana relapse was tobacco cigarette-smoking status. With-drawal symptoms, including marijuana craving, did not predict thelikelihood of relapse. However, the intensity of mood and sleepdisruption during withdrawal predicted more severe relapse, mea-sured in the amount of marijuana self-administered after an absti-nence period. Another factor predicting the severity of relapse wasratings of “high” after active marijuana smoking: greater intoxica-tion predicted more marijuana self-administration after abstinence.These findings, which parallel tobacco cessation data (37), suggestthat both the positive and negative reinforcing effects of marijuanainfluence the severity of relapse.

To follow-up, Study 2 assessed the direct impact of tobaccocigarette smoking by comparing marijuana withdrawal and relapsein the same individuals when they smoked cigarettes ad libitumand after a period of tobacco cessation. Consistent with findingsfrom Study 1, the vast majority of participants in Study 2 (�87%), allcigarette smokers, relapsed to marijuana whether they were smok-ing tobacco cigarettes as usual or were recently abstinent fromcigarettes. These findings rule out acute nicotine exposure or thecueing effects of tobacco cigarettes as an explanation for the highmarijuana relapse rates observed among cigarette smokers. To-bacco cigarette smoking, which was consistent across marijuanaconditions, had little influence on the direct or indirect effects ofmarijuana. In addition to not affecting marijuana relapse rates, to-bacco cigarette smoking did not alter marijuana intoxication ormost symptoms of marijuana withdrawal relative to tobacco cessa-tion.

These laboratory findings parallel marijuana treatment data. In alarge sample of adolescents, 69% of marijuana smokers whosmoked cigarettes relapsed to marijuana 1 year later, comparedwith 54% of non-smokers (38). Among adults in marijuana treat-ment (n � 174), marijuana and cigarette smokers provided approx-imately half as many cannabis-negative urines and had fewerweeks of abstinence than former smokers, despite no differences intreatment retention or marijuana dependence severity (29). To-bacco cigarette smoking also predicts treatment outcome for otherdrugs (39 – 41). A meta-analysis of 19 randomized clinical trials eval-uating tobacco treatment interventions during drug treatmentfound that smoking cessation was associated with a 25% increasedlikelihood of long-term abstinence from alcohol and illicit drugs(42,43). Thus, there is a clustering of quitting behavior, where stop-ping use of one drug increases the likelihood that other drug usewill stop (44).

Although clinical data show that quitting cigarettes predictsbetter treatment outcome than continued smoking, short-term to-bacco cessation did not decrease marijuana relapse in Study 2. Yet,it is worth noting that tobacco cessation was brief, and a desire toquit cigarette smoking was an exclusion criterion. Participants (ex-cept for two) complied with the requirements to not smoke ciga-rettes before admission, but none maintained this abstinence whennot reinforced for doing so. Given that the absence of nicotine didnot reduce marijuana relapse, perhaps it is the demonstrated abilityto quit smoking cigarettes that best predicts the ability to remainabstinent from other drugs (29). Anti-smoking initiatives are in-creasing, particularly in New York City, where cigarette packs cost

over $12, leading to a 35% drop in adult cigarette smoking preva- n

www.sobp.org/journal

ence from 2002 to 2010 (45). Perhaps those who persist in smokingigarettes despite mounting societal prohibitions are particularlyecalcitrant smokers, unable to maintain abstinence from cigarettes46) or from drug use overall.

Impulsivity might contribute to this recalcitrance. Delayed dis-ounting, or the devaluing of a reward obtained in the future rela-ive to one that is immediate, is one measure of impulsivity (47).rug-dependent individuals tend to discount distant rewards more

han healthy volunteers, consistent with the choice for immediatelyvailable drug intoxication or withdrawal relief over deferred re-ards of higher long-term value (e.g., healthy lungs, job opportu-ities). Among cigarette smokers, those with higher rates of delayiscounting chose more cigarette puffs over money (48). Similarly,lthough all of the current participants were daily drug users (mar-

juana), those who also smoked tobacco cigarettes were more likelyo choose the immediate reward of marijuana smoking over a moreelayed reward (higher study earnings), regardless of tobacco cig-rette availability.

Shared risk factors might also explain the current findings (49).dolescents who smoke tobacco cigarettes are over 9� more likely

o smoke marijuana than nonsmokers (50,51), and rates of tobaccoigarette use among marijuana smokers (47%–78%) (7,29,52) arever twice the national average (23%) (53). In terms of mechanism,reclinical data show that nicotine produces epigenetic effects thatnhance drug reward. In mice, repeated nicotine administrationroduced selective changes in striatal gene expression that en-anced the subsequent rewarding effects of cocaine (54). Preclini-al (55–57) and clinical ([58], but see [59,60]) studies show thatannabinoids also elevate striatal dopamine levels, albeit moreodestly than cocaine. Thus, it is intriguing to consider that one

xplanation for the current finding is that cigarette smoking beforehe onset of marijuana use resulted in more intractable marijuanase patterns.

Among those who relapsed to marijuana, two factors were as-ociated with the severity of relapse: marijuana withdrawal symp-oms and intoxication. Again, there are interesting parallels to theobacco literature. Among cigarette smokers, breath-holding dura-ion, which measures the ability to tolerate discomfort, predictedelapse to cigarettes during early abstinence, whereas the severityf nicotine withdrawal did not (61). Perhaps the intensity of mari-

uana withdrawal is less predictive of relapse than the ability toolerate the discomfort of withdrawal.

The present findings suggest several future studies to pursue:iven that clinical data support treatment for tobacco dependence

n substance-dependent participants (62), it is important to deter-ine whether patients seeking marijuana and tobacco cessation

reatment who successfully quit smoking cigarettes reduce theirarijuana use relative to patients who failed to quit cigarettes. In

ddition, participants in Study 1 began smoking marijuana everyay at approximately 18 years of age. Studies are needed to clarifyhy individuals who started daily marijuana use when they werelder were more likely to relapse than those who started at aounger age. This finding seems counter-intuitive, but perhapshose with a longer history of daily marijuana use have had morexperience attempting to moderate this behavior.

There are several issues to consider with the present design.irst, it might be that marijuana withdrawal severity did not predictelapse because the abstinence phase was too short (�4 days) toapture the full range of symptoms (23). Second, the sample wasomogenous, comprising primarily black men, potentially limitingur ability to generalize the findings to women or racially mixedopulations. Finally, our design was laboratory-based and used

on-treatment-seeking volunteers. However, although it might

1

1

1

1

1

2

2

2

2

2

2

2

2

2

2

3

3

3

3

3

3

3

3

3

3

M. Haney et al. BIOL PSYCHIATRY 2013;73:242–248 247

seem that such an approach is not relevant to patients, the validityof the model is supported by clinical evidence that cigarette smok-ing predicts marijuana relapse. Furthermore, medication effects ondrug self-administration in this model are consistent with clinicaloutcome for marijuana (13,16,21,22) as well as for treatment ofother drugs (24,25). Thus, human laboratory models of drug self-administration offer a controlled yet clinically relevant method tostudy factors underlying problematic drug use.

In sum, these data indicate that daily marijuana smokers haveconsiderable difficulty maintaining abstinence from marijuana, andthe factor best predicting this failure to maintain abstinence istobacco cigarette-smoking status. Whether individuals are cur-rently smoking tobacco cigarettes or have been recently abstinentdoes not alter these high relapse rates, suggesting that cigarettesmoking does not directly or indirectly influence marijuana relapse.Rather, cigarette smoking in concert with daily marijuana use mightreflect intrinsic factors, such as greater impulsivity or distress intol-erance, that render these individuals susceptible to relapse. Overall,the data suggest that current cigarette smoking is a clinical markerfor a greater risk of drug relapse (40), including marijuana relapse.

The US National Institute on Drug Abuse supported this research(DA19239, DA09236, DA031005) and supplied the marijuana ciga-rettes. The authors report no biomedical financial interests or potentialconflicts of interest.

1. United Nations Office on Drugs and Crime (2010): World Drug Report2010 (United Nations Publication, Sales No. E.10.XI.13). Vienna, Austria:United Nations Office on Drugs and Crime.

2. Data Spotlight: Nearly 40 Percent of Substance Abuse Treatment Admis-sions Report Alcohol-Drug Combinations. Available at: http://www.oas.samhsa.gov.

3. Hall W (2006): The mental health risks of adolescent cannabis use. PLoSMedicine 3:159 –162.

4. Bhana A, Parry CD, Myers B, Pluddemann A, Morojele NK, Flisher AJ(2002): The South African Community Epidemiology Network on DrugUse (SACENDU) project, phases 1-8-cannabis and Mandrax. S Afr Med J92:542–547.

5. EMCDDA (2009): Annual Report on the State of the Drugs Problem. Lisbon:European Monitoring Centre for Drugs and Drug Addiction.

6. SAMHSA (Substance Abuse and Mental Health Services Administration,Office of Applied Studies) (2011): Results from the 2010 National Surveyon Drug Use and Health: Summary of National Findings, NSDUH SeriesH-41, HHS Publication No. (SMA) 11-4658. Rockville, MD: SAMHSA.

7. Stinson FS, Ruan WJ, Pickering R, Grant BF (2006): Cannabis use disor-ders in the USA: Prevalence, correlates and co-morbidity. Psychol Med36:1447–1460.

8. Stephens RS, Roffman RA, Curtin L (2000): Extended versus brief treat-ment for marijuana use. J Consult Clin Psychol 68:898 –908.

9. Copeland J, Swift W, Roffman R, Stephens R (2001): A randomized con-trolled trial of brief cognitive-behavioral interventions for cannabis usedisorder. J Subst Abuse Treat 21:55– 64.

10. MTPRG (2004): Brief treatments for cannabis dependence: Findingsfrom a randomized multisite trial. J Consult Clin Psychol 72:455– 466.

11. Budney AJ, Moore BA, Rocha HL, Higgins ST (2006): Clinical trial ofabstinence-based vouchers and cognitive-behavioral therapy for can-nabis dependence. J Consult Clin Psychol 74:307–316.

12. Kadden RM, Litt MD, Kabela-Cormier E, Petry NM (2007): Abstinencerates following behavioral treatments for marijuana dependence. Ad-dict Behav 32:1220 –1236.

13. Levin FR, Mariani JJ, Brooks DJ, Pavlicova M, Cheng W, Nunes E (2011):Dronabinol for the treatment of cannabis dependence: A randomized,double-blind, placebo-controlled trial. Drug Alcohol Depend 116:142–150.

14. Haney M, Ward AS, Comer SD, Foltin RW, Fischman MW (1999): Absti-

nence symptoms following smoked marijuana in humans. Psychophar-macol 141:395– 404.

5. Haney M, Hart CL, Ward AS, Foltin RW (2003): Nefazodone decreasesanxiety during marijuana withdrawal in humans. Psychopharmacol 165:157–165.

6. Haney M, Hart CL, Vosburg SK, Nasser J, Bennett A, Zubaran C, Foltin RW(2004): Marijuana withdrawal in humans: Effects of oral THC or dival-proex. Neuropsychopharmacol 29:158 –170.

7. Haney M, Hart CL, Vosburg SK, Comer SD, Reed SC, Cooper ZD, et al.(2010): Effects of baclofen and mirtazapine on laboratory model ofmarijuana withdrawal and relapse. Psychopharmacology 211:233–244.

8. Hart C, Ward AS, Haney M, Comer SD, Foltin RW, Fischman MW (2002):Comparison of smoked marijuana and oral delta(9)-tetrahydrocannab-inol in humans. Psychopharmacol 164:407– 415.

9. Schierenbeck T, Riemann D, Berger M, Hornyak M (2008): Effect of illicitrecreational drugs upon sleep: Cocaine, ecstasy, and marijuana. SleepMed Rev 12:381–389.

0. Budney AJ, Vandrey RG, Hughes JR, Moore BA, Bahrenburg, B (2007):Oral delta-9 tetrahydrocannabinol suppresses cannabis withdrawalsymptoms. Drug Alcohol Depend 86:22–29.

1. Haney M (2009): Self-administration of cocaine, cannabis and heroin inthe human laboratory: Benefits and pitfalls. Addict Biol 14:9 –21.

2. Haney M, Hart CL, Vosburg SK, Comer SD, Reed SC, Foltin RW (2008):Effects of THC and lofexidine in a human laboratory model of marijuanawithdrawal and relapse. Psychopharmacol 197:157–168.

3. Cooper ZD, Foltin RW, Hart CL, Vosburg SK, Comer SD, Haney M (2012):A controlled human laboratory study investigating the effects of que-tiapine on marijuana withdrawal and relapse in daily marijuana smokers[published online ahead of print June 28]. Addiction Biology.

4. Haney M, Spealman R (2008): Controversies in translational research:Drug self-administration. Psychopharmacol 199:403– 419.

5. Comer SD, Ashworth JB, Foltin RW, Johanson CE, Zacny JP, Walsh SL(2008): The role of human drug self-administration procedures in thedevelopment of medications. Drug Alcohol Depend 96:1–15.

6. Foltin RW, Haney M, Comer SD, Fischman MW (1996): Effects of fenflu-ramine in food intake, mood, and performance of humans living in aresidential laboratory. Physiol Behav 59:295–305.

7. Foltin RW, Fischman MW, Pedroso JJ, Pearlson GD (1987): Marijuana andcocaine interactions in humans: Cardiovascular consequences. Pharma-col Biochem Behav 28:459 – 464.

8. Balerio GN, Aso E, Berrendero F, Murtra P, Maldonado R (2004): �9-tetrahydrocannabinol decreases somatic and motivational manifesta-tions of nicotine withdrawal in mice. Eur J Neurosci 20:2737–2748.

9. Moore BA, Budney AJ (2001): Tobacco smoking in marijuana-depen-dent outpatients. J Substance Abuse 13:583–596.

0. Vandrey RG, Budney AJ, Hughs JR, Liguori A (2008): A within-subjectcomparison of withdrawal symptoms during abstinence from cannabis,tobacco, and both substances. Drug Alcohol Depend 92:48 –54.

1. Viveros MP, Marco EM, File SE (2006): Nicotine and cannabinoids: Paral-lels, contrasts and interactions. Neurosci Biobehav Rev 30:1161–1181.

2. Penetar DM, Kouri EM, Gross MM, McCarthy EM, Rhee CK, Peters EN, et al.(2005): Transdermal nicotine alters some of marijuana’s effects in maleand female volunteers. Drug Alcohol Depend 79:211–223.

3. Ream GL, Benoit E, Johnson BD, Dunlap E (2008): Smoking tobaccoalong with marijuana increases symptoms of cannabis dependence.Drug Alcohol Depend 95:199 –208.

4. Hughes JR (2007): Effects of abstinence from tobacco: Valid symptomsand time course. Nicotine Tob Res 9:315–327.

5. SRNT Subcommittee on Biochemical Verification (2002): Biochemicalverification of tobacco use and cessation. Nicotine Tob Res 4:149 –159.

6. Murray A, Levin FL Bisaga A, Glass A, Brooks D. Higher cotinine levelsamong non-cigarette smoking blunt users versus joint users. In prepa-ration.

7. Perkins KA, Broge M, Gerlach D, Sanders M, Grobe JE, Cherry C (2002):Acute nicotine reinforcement, but not chronice tolerance, predictswithdrawal and relapse after quitting smoking. Health Psychology 21:332–339.

8. de Dios MA, Vaughan EL, Stanton CA, Niaura R (2009): Adolescent to-bacco use and substance abuse treatment outcomes. J Subst AbuseTreat 37:17–24.

9. Frosch DL, Shoptaw S, Nahom D, Jarvik ME (2000): Associations between

tobacco smoking and illicit drug use among methadone-maintainedopiate-dependent individuals. Exp Clin Psychopharmacol 8:97–103.

www.sobp.org/journal

4

4

4

4

4

5

5

5

5

5

5

5

5

6

6

6

248 BIOL PSYCHIATRY 2013;73:242–248 M. Haney et al.

40. Kohn CS, Tsoh JY, Weisner CM (2003): Changes in smoking status amongsubstance abusers: Baseline characteristics and abstinence from alco-hol and drugs at 12-month follow-up. Drug Alcohol Depend 69:61–71.

1. Shoptaw S, Jarvik ME, Ling W, Rawson RA (1996): Contingency manage-ment for tobacco smoking in methadone-maintained opiate addicts.Addict Behav 21:409 – 412.

2. Prochaska JJ, Delucchi K, Hall SM (2004): A meta-analysis of smokingcessation interventions with individuals in substance abuse treatmentor recovery. J Consult Clin Psychol 72:1144 –1156.

3. Gulliver SB, Kamholz BW, Helstrom AW (2006): Smoking cessationand alcohol abstinence: What do the data tell us? Alcohol Res Health29:208 –212.

4. Metrik J, Spillane NS, Leventhal AM, Kahler CW (2011): Marijuana useand tobacco smoking cessation among heavy alcohol drinkers. DrugAlcohol Depend 119:194 –200.

5. New York City Department of Health and Mental Hygiene. Epi DataBriefs and Tables. Available at: http://www.nyc.gov/html/doh/html/episrv/epidata.shtml. Accessed March 21, 2012.

46. Hughes JR (2011): The hardening hypothesis: Is the ability to quit de-creasing due to increasing nicotine dependence? A review and com-mentary. Drug Alcohol Depend 117:111–117.

47. Bickel WK, Odum AL, Madden GJ (1999): Impulsivity and cigarette smok-ing: Delay discounting in current, never, and ex-smokers. Psychophar-macology 146:447– 454.

48. Dallery J, Raiff BR (2007): Delay discounting predicts cigarette smokingin a laboratory model of abstinence reinforcement. Psychopharmacol-ogy 190:485– 496.

49. Kendler KS, Myers J, Prescott CA (2007): Specificity of genetic and envi-ronmental risk factors for symptoms of cannabis, cocaine, alcohol, caf-feine, and nicotine dependence. Arch Gen Psychiatry 64:1313–1320.

50. Rigotti NA, Lee JE, Wechsler H (2000): US college students’ use of to-bacco products: Results of a national survey. JAMA 284:699 –705.

51. Mathers M, Toumbourou JW, Catalano RF, Williams J, Patton GC (2006):Consequences of youth tobacco use: A review of prospective behavioral

studies. Addiction 101:948 –958.

www.sobp.org/journal

2. Ford DE, Vu HT, Anthony JC (2002): Marijuana use and cessation oftobacco smoking in adults from a community sample. Drug AlcoholDepend 67:243–248.

3. SAMHSA (Substance Abuse and Mental Health Services Administration,Office of Applied Studies) (2009): Treatment Episode Data Set (TEDS)Highlights—2007 National Admissions to Substance Abuse Treatment Ser-vices. OAS Series #S-45, HHS Publication No. (SMA) 09-4360. Rockville, MD:SAMHSA.

4. Levine A, Huang Y, Drisaldi B, Griffin EA Jr, Pollak DD, Xu S, et al. (2011):Molecular mechanism for a gateway drug: Epigenetic changes initiatedby nicotine prime gene expression by cocaine. Sci Transl Med3:107ra109.

5. Cheer JF, Wassum KM, Heien ML, Phillips PE, Wightman RM (2004):Cannabinoids enhance subsecond dopamine release in the nucleusaccumbens of awake rats. J Neurosci 24:4393– 4400.

6. Fadda P, Scherma M, Spano MS, Salis P, Melis V, Fattore L, Fratta W(2006): Cannabinoid self-administration increases dopamine release inthe nucleus accumbens. Neuroreport 17:1629 –1632.

7. Tanda G, Goldberg SR (2003): Cannabinoids: Reward, dependence, andunderlying neurochemical mechanisms—a review of recent preclinicaldata. Psychopharmacology (Berl) 169:115–134.

8. Bossong MG, van Berckel BN, Boellaard R, Zuurman L, Schuit RC Wind-horst AD, et al. (2009): Delta 9-tetrahydrocannabinol induces dopaminerelease in the human striatum. Neuropsychopharmacology 34:759 –766.

9. Stokes PRA, Mehta MA, Curran HV, Breen G, Grasby PM (2009): Canrecreational doses of THC produce significant dopamine release in thehuman striatum? NeuroImage 48:186 –190.

0. Barkus E, Morrison PD, Vuletic D, Dickson JC, Ell PJ, Pilowsky LS, et al.(2011): Does intravenous �9-tetrahydrocannabinol increase dopaminerelease? A SPET study. J Psychopharmacol 25:1462–1468.

1. Brown RA, Lejuez CW, Strong DR, Kahler CW, Zvolensky MJ, CarpenterLL, et al. (2009): A prospective examination of distress tolerance andearly smoking lapse in adult self-quitters. Nicotine Tob Res 11:493–502.

2. Prochaska JJ (2010): Failure to treat tobacco use in mental health and

addiction treatment settings: A form of harm reduction? Drug AlcoholDepend 110:177–182.