bromocriptine reduces cigarette smoking

Addiction (2000) 95(8), 1173–1183

RESEARCH REPORT

Bromocriptine reduces cigarette smoking

MURRAY E. JARVIK1,2, NICHOLAS H. CASKEY1,2,WILLIAM C. WIRSHING1,2, DAMIAN C. MADSEN1,2,PAULA N. IWAMOTO-SCHAAP1, JULIE L. ELINS1,NAOMI I. EISENBERGER1 & RICHARD E. OLMSTEAD1,2

1University of California Los Angeles School of Medicine & 2Veterans Affairs, Greater LosAngeles Healthcare System, Los Angeles, CA, USA

AbstractAims. Animal studies have shown that nicotine releases dopamine, a neurotransmitter implicated in drugreinforcement. We hypothesized that bromocriptine would decrease smoking behavior in humans. Design.The study was conducted double blind and subjects’ order of dose exposure was randomized. Participants.The smoking behavior of 20 heavy smokers was recorded for 5 hours after ingesting placebo or one of two dosesof bromocriptine (2.50 mg, 3.75 mg) over three sessions (one dose per session). Findings. There was asigni� cant negative linear trend by dosage indicating shorter total puf� ng time with increasing bromocriptinedosages (p , 0.02). Other signi� cant negative linear trends by increasing dosage include fewer number ofpuffs, fewer number of cigarettes smoked and mean latency to smoke after 3 hours (expected CMAX on the drug(all ps , 0.05). There was a negative signi� cant linear trend showing decreased plasma nicotine (p ,0.02) and cotinine (p , 0.005) with increasing dosages of bromocriptine. Shiffman/Jarvik WithdrawalScale (SJWS) cigarette craving subscale scores decreased signi� cantly across increasing dosages (linear trendp , 0.02). There was a signi� cant negative linear trend (p , 0.05) on the Pro� le of Mood States(POMS) Vigor and Depression subscales, with subjects reporting decreased vigor and depression withincreasing bromocriptine doses. No other mood effects were observed. Conclusion. These results support thehypothesis that dopaminergic mechanisms mediate cigarette smoking reinforcement.

IntroductionAnimal research (Clarke, 1990, 1992; Corrigall,1991; Corrigall & Coen, 1991; Pontieri et al.,1996; Rose & Corrigall, 1997) has pointed to thepossibility that nicotine reinforcement may be, atleast in part, controlled by dopaminergic mecha-nisms. Dopamine has been previously hypothe-sized to play a central role in mediating thereinforcing effects of other stimulant drugs, e.g.

cocaine and amphetamines (Koob & Bloom,1988; Wise & Rompre, 1989). In a repeatedmeasures study, McEvoy et al. (1995) reportedincreased smoking in schizophrenics (evidencedby higher expired carbon monoxide and plasmanicotine levels) when they were given haloperi-dol, a dopamine antagonist, in comparison withtheir baseline levels when they were taking noantipsychotic medications. Recent studies exam-

Correspondence to: Murray E. Jarvik MD, Phd, VAMC-WLA T-350 (691/B151D), University of California LosAngeles School of Medicine, Los Angeles, CA 90073, USA. Fax: 310 268 4498; e-mail: [email protected]

Submitted 9th September 1998; initial review completed 9th November 1998; � nal version accepted 17th March2000.

ISSN 0965–2140 print/ISSN 1360-0443 online/00/081173–11 Ó Society for the Study of Addiction to Alcohol and Other Drugs

Carfax Publishing, Taylor & Francis Ltd

1174 Murray E. Jarvik et al.

ining the effect of haloperidol on smoking innon-psychiatric subjects have shown that it in-creases a number of indices of smoking behavior.Dawe et al. (1995) reported signi� cantly in-creased nicotine intake in postprandial smokingwhen subjects were pretreated with 5 mg halope-ridol vs. placebo. Caskey et al. (1999) have re-ported increased rates of smoking innon-psychiatric subjects with acute administra-tion of haloperidol vs. placebo (1.0 mg and 2.0mg vs. placebo) and with higher vs. lower dosesof haloperidol (1.0 and 0.5 mg).

Bromocriptine is an ergot alkaloid currentlyused for the treatment of amenorrhea, galactor-rhea, prolactin secreting adenomas and Parkin-son’s disease. A single 2.5 mg dose ofbromocriptine has been shown to suppress circu-lating prolaction levels by 47% to 96% (Thorneret al., 1980); when given chronically, dosagesrange from 5.0 mg/day to over 100 mg/day de-pending on the indication (Vance, Evans &Thorner 1984). In tests with native D1 (calfbrain striatum/calf parathyroid) and D2 receptors(pig anterior pituitary/canine brain striatum) andcloned D1-like (D1, D5) and D2-like (D2, D3, D4)receptors, it appears that bromocriptine is equiv-alent to dopamine in af� nity in binding to D2

and D3 receptors (Seeman, 1994). It therebyfunctions in much the same way that endoge-nous dopamine does, except that bromocriptineis more speci� c to certain types of receptors. Itputatively exerts its therapeutic action throughits potent agonism of post-synaptic D2 receptors(Vance et al., 1984). Stimulation of the postsy-naptic D2 receptor with bromocriptine causes adecrease in � ring of the postsynaptic neuron.

Pilot data (Caskey et al., 1999) indicated thatacute administration of bromocriptine, a do-pamine agonist, to a small sample of non-psychi-atric subjects (n 5 5) resulted in a signi� cantlyslower rate of smoking (2.5 mg vs. placebo). Thecurrent study was designed to replicate and ex-tend the results of the bromocriptine pilot exper-iment (Caskey et al., 1999) by increasing boththe sample size and the dose range. It was hy-pothesized that there would be dose-dependentdecrease in smoking behavior and nicotine intakewith increasing doses of bromocriptine.

MethodsSubjectsSubjects were initially screened during a brief

telephone interview. Subjects who reported anyhistory of cardiac or respiratory/pulmonary ill-ness or disease, endocrine or metabolic disor-ders, seizure disorders, treatment with anyantipsychotic, anti-depressant or other psy-chotropic medication and/or taking any medi-cation which could interact with bromocriptinewere excluded. To be included in the study,subjects had to report that they had been smok-ing a minimum of 15 cigarettes per day for atleast the past 2 years. Twenty heavy smokers (13males, seven females) were recruited from theWest Los Angeles Veterans Affairs Medical Cen-ter (n 5 11) and general community (n 5 9).Their mean age was 34.6 years (SD 5 11.6;range 5 18–58; one missing value). The meanlevel of education in years was 13.3 (SD 5 1.8;range 5 10–17). Subjects smoked a mean of 21.3cigarettes per day (SD 5 6.3; range 5 12.5–35)and had smoked for a mean of 19.8 years(SD 5 11.4; range 5 4–42). Two subjects wereincluded who smoked less than the requiredminimum of 15 cigarettes per day (12.5 and 14cigarettes per day). These two subjects had re-ported smoking at least 15 cigarettes per dayduring the telephone interview, but subsequentlyreported smoking fewer cigarettes when theycompleted written questionnaires at the labora-tory. Both subjects had CO levels greater than25 p.p.m. when tested at 5:30 p.m. during theinitial baseline visit.

Design and procedureThis study utilized a randomized, double-blind,repeated-measures design in which subjectsserved as their own controls. Subjects partici-pated individually in one screening baseline visitand three experimental sessions spaced one weekapart. Sessions were conducted at the West LosAngeles Veterans Administration Medical Cen-ter. Subjects were paid a total of $180 for partic-ipating in the baseline visit and threeexperimental sessions.

Baseline visitA baseline visit was conducted with all subjectsto provide an opportunity for more in-depthhealth screening and to familiarize potential sub-jects with experimental procedures. At the base-line visit subjects gave written informed consent,had a physical examination (including an ECG)

Bromocriptine reduces cigarette smoking 1175

and completed background questionnaires.Health screening included a medical history andphysical examination (including blood pressure,heart rate and weight). Smoking status wasveri� ed by carbon monoxide (CO $ 20 p.p.m.).Subjects smoked a cigarette through an exper-imental smoking apparatus to acquaint themwith the experimental procedure. The apparatus(described below) was designed to measuresmoking topography (e.g. puff duration). Sub-jects also completed both the Pro� le of MoodStates (POMS) and the Modi� ed Shiffman–

Jarvik Withdrawal Scale (SJWS) to familiarizethem with these questionnaires which were to beused repeatedly throughout experimental ses-sions.

Experimental sessionsExperimental sessions began at 8:30 a.m. (At thebaseline session, subjects were instructed not toeat anything or drink any caffeinated beveragesprior to coming to the experimental sessions inorder to facilitate drug absorption. Subjects whoreported having had food or caffeinated bever-ages after midnight were rescheduled.) Uponarrival each subject reported the time of com-pletion of his or her most recent cigarette. Bloodpressure was then measured. Subjects then com-pleted a baseline questionnaire battery (includ-ing the Pro� le of Mood States [POMS] andShiffman–Jarvik Withdrawal Scale [SJWS] (seebelow)). Next, subjects provided a baseline ex-pired-air carbon monoxide sample. At this timesubjects smoked a “loading cigarette” to start theexperimental session. They were instructed tosmoke as much or as little of one of their owncigarettes as they pleased. Blood was drawn ex-actly 2 minutes after completion of the loadingcigarette. The drug (placebo, 2.5 mg or 3.75 mg)was administered after the blood draw. AnotherCO measurement, taken 20 minutes after theloading cigarette, was used as the baselinemeasurement for subsequent experimentalanalyses on CO changes. Breakfast was served45 minutes after drug ingestion. Breakfast itemsincluded a 12-ounce bowl of corn� akes, 16ounces of low-fat milk, 16 ounces of orangejuice, 6 ounces of low-fat mixed fruit yogurt andthree small cinnamon rolls.

Experimental sessions lasted a minimum of 5to a maximum of 5.5 hours (see below). Subjectswere instructed to smoke freely (ad libitum) using

the smoking topography apparatus during theexperimental sessions. Subjects smoked theirown brand of cigarettes. Subjects watchedvideotaped movies for the remainder of the ses-sion by themselves. The movies were light-hearted comedies, which did not feature anyscenes of actors smoking. The questionnaire bat-tery (POMS, SJWS, additional measures of de-sire to smoke and two nausea measures) wasgiven 30 minutes after completion of the loadingcigarette. Before each successive cigarette, sub-jects were required to notify the experimenter viawireless intercom when they wanted to smokeanother cigarette. Carbon monoxide and re-peated questionnaire measures were taken beforeevery cigarette. Time at which cigarette was lit,number of puffs, duration per puff, and timecigarette was extinguished were recorded. Imme-diately after each cigarette the subjects com-pleted the two nausea scales. Carbon monoxidelevels were measured again 20 minutes after eachcigarette. Subjects completed the repeated ques-tionnaire measures battery 30 minutes after eachcigarette. In cases where subjects desired tosmoke again prior to the repeated measures bat-tery (i.e. less than 30 minutes) they � rst com-pleted the battery, then were allowed to smoke.This procedure was repeated for each cigarette.

Blood was again drawn precisely 2 minutesafter the � rst cigarette smoked after 3 hourspost-drug ingestion (time of expected Cmax) or 5hours after drug ingestion if no cigarettes weresmoked after the 3-hour mark. The sessionended 5 hours after the drug was administeredexcept to complete the CO and questionnaireassessments that followed cigarettes smoked dur-ing the half-hour immediately before the 5-hourmark. For example, if the subject requested tosmoke at the 5-hour mark, they remained in thelaboratory for another 30 minutes for CO testingat 5 hours 20 min and questionnaires at 5 hours30 min. Subjects’ blood pressure and heart ratewere measured to ensure that they were withinnormal limits before subjects were released.

ApparatiA thermistor puff-detecting device was used tomeasure cigarette smoking topography (puff dur-ation). This device consists of a thermistor (Vic-tor Engineering) embedded in a commerciallyavailable cigarette holder (Aqua Filter). Thethermistor is heated to 200°C by electrical cur-


rent. Cigarette smoke passing over the thermis-tor causes a drop in the thermistor’s temperaturecausing a change in the thermistor’s electricalresistance. Changes in the thermistor’s resistanceare converted into a voltage signal. Combiningthe thermistor apparatus with an electronic tim-ing device enabled us to measure the time ofonset and completion for each cigarette, puffduration, number of puffs per cigarette, etc. ABedfont II Microsmokelyzer was used to mea-sure expired air carbon monoxide. Blood press-ure was measured manually using asphygmomanometer.

MeasuresBackground measures. One-time-only back-

ground questionnaires included a measure ofsmoking history and demographic information(Smoker’s Pro� le), the Fagerstrom Test for Nic-otine Dependence (FTND; Heatherton et al.,1991), the Smoking Motivation Questionnaire(SMQ; Russell et al., 1974) and the Smoker’sBeliefs Questionnaire (SBQ).

Smoking topography. Total number ofcigarettes smoked and total number of puffstaken during the session were noted. Puff timeswere recorded by the puff detector, obtainingtotal puf� ng time and mean length of puffs overthe session. The time in minutes to the � rstcigarette smoked after 3 hours on the drug wasderived from the record of cigarette and drugonset times. If no cigarettes were smoked afterthe 3-hour mark, the time until the end of thesession was substituted for the censored datapoint (e.g. 120 minutes).

Repeated subjective measures. Repeated-measurequestionnaires included: (1) the Urge to Smokequestionnaire (UTS; Jarvik et al., 2000), (2)Schuh & Stitzer’s index, a four-item visualanalog scale (SSI; Schuh & Stitzer, 1995), (3) amodi� ed version of the Shiffman–Jarvik With-drawal scale which contained only the non-craving related withdrawal items (SJWS-NC;Shiffman & Jarvik, 1976), (4) Pro� le of MoodStates questionnaire (POMS), (5) a 100-mmVisual Analog nausea scale, (6) � ve-point Likertnausea scale, and (7) a single-item Strength ofthe Urge to Smoke (SUTS; Jarvik et al., 2000).

Bioassays. Blood samples (10 ml) were assayedfor nicotine and cotinine concentrations by gaschromatography. Breath samples were assessedfor carbon monoxide content. Blood and breathsamples were not taken from the � rst two partic-ipants entered into the study.

Data analysisRepeated measures analysis of variance (r-ANOVA) was used to compare values differentacross the three sessions. Within each analysis,polynomial contrasts (linear and quadratic) ex-amining the effect of dose in detail were tested(The dose–response was predicted to have alinear shape.) For the smoking topography indi-ces (total number of cigarettes, total puf� ng time,total number of puffs, mean length of puff andlatency to smoke after 3 hours on drug) whichproduced only a single measurement per session,a simple three-condition repeated measuresanalysis was conducted. For plasma nicotine andplasma cotinine, the values from the sample afterthe 3-hour mark were compared across sessions;the sample values from the pre-loading cigarettewere used as within-session covariates to controlfor a differences in pre-drug values. The COmeasures taken concurrently with the blood sam-ples were also analyzed in this fashion. In ad-dition, since other CO measures were also taken(contingent upon smoking), all available datawere used to calculate mean CO values duringthree periods: within 1 hour of drug administra-tion (including prior to drug administration),1–3 hours after drug administration and 3–6hours after drug administration (3 3 3 time 3condition r-ANOVA). For each self-report mea-sure (UTS, SUTS, SSI, SJWS subscales, POMSsubscales, nausea), the mean value post-drugover all available data was calculated for eachsession. The r-ANOVA proceeded as with thetopography measures.

The planned analyses speci� ed above weretested with a directional hypothesis of greatereffect with increasing dosages of bromocriptine(signi� cance set at a 5 0.05 one-tailed). In ad-dition to statistical signi� cance, effect size values(partial- g 2; Cohen, 1988) were calculated to as-sess relative impact of the dose factor on thedependent variables. All statistical analyses wereexecuted using SPSS for Windows.


Table 1. Means and effect sizes for smoking topography for placebo and two doses of bromocriptine (standard deviationsin parentheses)

Number of Mean for Mean for Mean for Effect size Linearsubjects placebo 2.5 mg 3.75 mg g 2 trend

Total no. of cigarettes 20 3.0 2.6 2.3 0.20 p 5 0.044(1.8) (1.5) (1.6)

Total puf� ng time 20 72.3 55.4 50.6 0.29 p 5 0.011(57.8) (38.8) (51.0)

Total number of puffs 20 32.9 26.8 23.8 0.23 p 5 0.028(22.3) (14.4) (17.0)

Mean length of puffs 20 2.2 2.1 2.0 0.17 p 5 0.060(1.1) (0.8) (1.0)

Minutes to smoke after 3 hourson drug 20 50.7 71.3 73.5 0.23 p 5 0.028

(41.4) (40.8) (38.8)

ResultsSome data were lost due to events that occurredduring the study. Blood was not drawn if theparticipant had recently vomited (four samples).Fourteen blood samples were lost when theywere accidentally destroyed during preparationfor shipping. Overall, seven of the lost sampleswere for the initial blood draw and 11 were forthe draw after the 3-hour point. With the excep-tion of the four samples not drawn for subjectsafety, the missing samples occurred basically atrandom. Some of the lost samples were withinthe same subject; the total number of subjectswith a complete set of blood samples was nine.

Smoking topographySeveral signi� cant negative linear trends by in-creasing bromocriptine dose (placebo, 2.5 mg,3.75 mg) were observed (see Table 1). Thenumber of cigarettes subjects smoked decreasedsigni� cantly with increasing levels of bro-mocriptine (F[1,19] 5 4.63; p , 0.05). Sub-jects’ total puf� ng time (summed over all puffs)decreased signi� cantly with increasing bro-mocriptine levels (F[1,19] 5 7.88; p , 0.02)and total number of puffs also decreasedsigni� cantly with increasing levels of bro-mocriptine (F[1,19] 5 5.69; p , 0.05). As totalpuf� ng time is partly a function of the totalnumber of puffs, mean length of puff (totalpuf� ng time/total number of puffs) was exam-ined. Mean puff duration also decreasedsigni� cantly with increasing levels of bro-

mocriptine (F[1,19] 5 4.01; p , 0.05). Themean latency to smoke after the time of expectedCmax for bromocriptine (3 hours) increasedsigni� cantly with increasing bromocriptine(F[1,19] 5 5.66; p , 0.05).

Biochemical assessmentsLevels of plasma nicotine and cotinine werecompared at the second blood draw acrossplacebo and the two bromocriptine doses, co-varying on the levels of nicotine and continine atthe baseline blood draw at the beginning of theexperimental sessions. Figures 1 and 2 show asigni� cant negative linear trends for both nic-otine and cotinine (F[1,7] 5 8.25; p , 0.05 andF[1,7] 5 27.39; p , 0.001, respectively) indicat-ing that subjects took in less nicotine with in-creasing levels of bromocriptine (see Table 2).

Unlike the results for plasma nicotine andcotinine, the results using the CO measurestaken 20 minutes after the blood draws failedto show any effect by drug condition(F[1,19] 5 0.64; p . 0.20). On the other hand,the analysis of the within session trends didindicate an apparent effect of the drug. Thistrend over time in carbon monoxide values ispresented in Fig. 3. There was a signi� cant timeby dose effect (F[1,17] 5 12.65; p , 0.005). Itcan be seen in the � gure that CO increasedthrough the session for the placebo group butdecreased in the two active drug conditions.Although the baseline values for the three groupsappear to be different, this was not signi� cant (p. 0.20).

0Placebo

Nic

otin

e in

ng/

ml +

SE

5

2.5 mg 3.75 mg

Dose

10

15

20

25

30

35


Table 2. Means and effect sizes for biological assessments for placebo and two doses of bromocriptine (standard deviationsin parentheses)


Plasma nicotine levels 9 23.9 22.6 18.3 0.54 p 5 0.012(14.9) (17.2) (17.5)

Plasma cotinine levels 9 229 214 192 0.80 p 5 0.000(54.9) (83.0) (42.9)

Expired carbon monoxide levels 20 25.0 25.2 23.9 0.01 p 5 0.495(4.9) (4.7) (4.9)

Figure 1. Plasma nicotine levels by dose (adjusted means), n 5 9, p , 0.05.

Self-report measuresTable 3 presents the results for the self-reportmeasures. There was a signi� cant linear trend(F[1,19] 5 7.50; p , 0.02) for desire to smoketo decrease with increasing bromocriptine asmeasured by the craving subscale of the SJWS.Despite the high intercorrelation of all the differ-ent measures of craving used in this study, thiswas the only signi� cant effect observed. Onlyone of the subscales of the POMS showed astrong bromocriptine effect. There was asigni� cant (F[1,19] 5 8.11; p , 0.02) negativelinear trend for the Vigor subscale such thatsubjects reported reduced vigor with increasinglevels of bromocriptine. There was also an ap-parent effect of the drug on the Depression

subscale of the POMS (F[1,19] 5 4.43; p ,0.05) however the clinical signi� cance of thiseffect was rather small, accounting for a changeof less than 5-hundredths of a point on a four-point scale.

Nausea and vomitingThere was a near signi� cant positive linear trend(F[1, 9] 5 3.17; p , 0.06) for increasing nausea(measured with the Likert scale immediately af-ter completion of cigarettes) with increasing bro-mocriptine levels. Five subjects vomited duringthe course of the experiment. All � ve vomitedwith the 3.75 mg dose and one of those � ve alsovomited with the 2.5 mg dose. No subjects vom-

0Placebo

Cot

inin

e in

ng/

ml +

SE

50

2.5 mg 3.75 mg

Dose

100

150

200

250

300

1-3 hours20

Within 1 hour of drug

Car

bon

mon

oxid

e (P

PM

)

21

Time (hours)

22

23

24

25

26

3-6 hours


Figure 2. Plasma cotinine levels by dose (adjusted means), n 5 9, p , 0.001.

Figure 3. Mean expired air carbon monoxide comparisons ( r placebo, j low dose, s high dose).


Table 3. Means and effect sizes for self-report measures (craving, withdrawal, and mood) for placebo and two doses ofbromocriptine (standard deviations in parentheses)


UTS (1–7)a 18 3.91 3.61 3.71 0.03 p 5 0.462(1.26) (1.24) (1.30)

SUTS (1–7)a 18 3.41 2.72 3.03 0.05 p 5 0.354(1.28) (1.43) (1.74)

SSI (0–100) 20 39.9 35.9 37.3 0.01 p 5 0.639(18.6) (21.3) (22.9)

S-J craving (1–7) 20 4.46 3.63 3.72 0.28 p 5 0.013(0.99) (1.12) (1.25)

S-J psychological distress (1–7) 20 2.79 2.90 2.96 0.05 p 5 0.357(0.70) (0.74) (0.83)

S-J physical symptoms (1–7) 20 1.69 1.92 1.94 0.05 p 5 0.327(0.77) (0.92) (1.27)

S-J appetite (1–7) 20 3.56 3.56 3.29 0.05 p 5 0.356(0.77) (0.80) (1.08)

S-J stimulation/sedation (1–7) 20 2.55 3.40 3.12 0.15 p 5 0.085(1.19) (1.48) (1.23)

POMS vigor (1–5) 20 2.91 2.50 2.45 0.30 p 5 0.010(0.86) (0.93) (0.98)

POMS confusion (1–5) 20 1.59 1.62 1.62 0.00 p 5 0.791(0.46) (0.39) (0.55)

POMS fatigue (1–5) 20 1.40 1.52 1.51 0.06 p 5 0.286(0.50) (0.60) (0.47)

POMS anger/hostility (1–5) 20 1.12 1.07 1.10 0.05 p 5 0.320(0.20) (0.16) (0.18)

POMS depression (1–5) 20 1.16 1.13 1.14 0.19 p 5 0.049(0.25) (0.24) (0.25)

POMS tension (1–5) 20 1.44 1.56 1.60 0.07 p 5 0.241(0.40) (0.48) (0.72)

aUTS and SUTS were not given to � rst two participants in study.

ited with the placebo. No subjects had to bediscontinued from the study, and subjects whodid vomit typically reported that they felt betterafter vomiting. Besides the � ve subjects whoactually vomited, three additional subjects ex-perienced nausea. One subject experienced nau-sea with both the low and high dose, one subjecthad nausea only with the high dose and one hadnausea only with the low dose.

Because of the possible effects of nausea onurge to smoke and smoking behavior, the vari-ables which showed signi� cant � nding were re-analyzed with nausea ratings as a covariate.Unfortunately, nausea was only measured con-tingent on smoking so that ratings were notavailable if the subject stopped smoking duringthe session (possibly due to nausea). Hence themean value of nausea over the session likelyunderestimates the degree of nausea experiencedby the participants. A correction was imple-mented such that instead of mean nausea rat-ings, the slope of the available nausea ratings was

calculated and utilized as the covariate in themodi� ed analyses. Thus, if nausea were low butincreasing prior to the discontinuation of smok-ing, this would be captured in the slope measure.

Table 4 presents the changes in signi� canceand effect size with and without the nausea slopecovariate. Introducing nausea as an explanatoryvariable reduced the signi� cance levels of thelinear trends for most of the topography mea-sures and the effect sizes were reduced by ap-proximately 50%. Two effects, total puf� ng timeand latency to smoke, remained signi� cant. Thelinear trend for nicotine levels was no longersigni� cant after covarying for nausea, but thenegative linear trend for cotinine levels remainedhighly signi� cant, despite the covarying for nau-sea levels. The g 2 values associated with theself-report, i.e. non-behavioral, measures,showed reduced effects accounted for by thedrug much like the topography measures. De-spite the decrease in signi� cance values, the ef-fect size index, g 2 (Cohen, 1988), still indicates


Table 4. Effects of covarying nausea on positive � ndings

Before covarying After covarying

Effect size Linear Effect size Linearg 2 trend g 2 trend

Smoking topographyTotal no. of cigarettes 0.20 p 5 0.044 0.08 p 5 0.079Total puf� ng time 0.29 p 5 0.011 0.12 p 5 0.034Total number of puffs 0.23 p 5 0.028 0.10 p 5 0.051Mean length of puffs 0.17 p 5 0.060 0.08 p 5 0.080Minutes to smoke after 3 hours 0.23 p 5 0.028 0.12 p 5 0.031

Biological assessmentsPlasma nicotine 0.54 p 5 0.012 0.38 p 5 0.053Plasma cotinine 0.80 p 5 0.000 0.66 p 5 0.007

Mood measuresS-J craving 0.28 p 5 0.013 0.13 p 5 0.022S-J stimulation/sedation 0.15 p 5 0.085 0.06 p 5 0.124POMS vigor 0.30 p 5 0.010 0.15 p 5 0.017POMS depression 0.19 p 5 0.049 0.02 p 5 0.358

moderate to large effects for bromocriptine onthese variables overall (range 5 0.08 to 0.66).

Instances of vomiting were also examined as abinary covariate in re-analyses of total puf� ngtime, plasma cotinine and POMS Vigor. In allcases, the linear contrast continued to be statisti-cally signi� cant when controlling for instances ofvomiting (F[1, 37] 5 4.92; p , 0.04 for totalpuf� ng time; F[1, 37] 5 44.94; p , 0.001 forplasma cotinine; F[1,37] 5 10.74; p , 0.005 forPOMS Vigor). Total puf� ng time and POMSVigor did not appear to be affected by vomiting(p , 0.20).

Caffeine dependenceSince the protocol required participants to beabstinent from caffeine, the effect of caffeinedependence, as measured by the self-reportednumber of caffeinated beverages consumeddaily, was used as a covariate in a re-analysis ofthe total puf� ng time measure to ascertain ifthere might be a moderating effect. While thosewho consumed more caffeinated beverages didtend to smoke more during the experimentalsessions (r 5 0.38, (p , 0.10), there was noapparent effect of moderation in that the effect ofbromocriptine remained unchanged (F[1,37] 5 8.10; p , 0.01; g 2 5 0.18) with caffeineintake as a covariate.

DiscussionThe results from this study are consistent withthe hypothesis that bromocriptine would de-crease smoking and nicotine intake. Given thedopaminergic speci� city of bromocriptine, thecurrent results also lend support to hypothesizeddopaminergic modulation of nicotine reinforce-ment. That is, the current data suggest thatsmoking decreases with increasing dopamine ag-onism.

These results do not appear to be attributableto withdrawal effects for several reasons. First,subjects were allowed to smoke ad libitum suchthat they were in control over their level ofnicotine deprivation and withdrawal. In general,subjects smoked more in the placebo conditionthan in the bromocriptine conditions (totalpuf� ng time, total number of puffs, mean puffduration, smoking rate), yet they simultaneouslywere reporting more urges to smoke in theplacebo condition.

Increased levels of nausea also appear to havebeen related to the decreased levels of smokingwith higher levels of bromocriptine. The increasein nausea and vomiting episodes were not unex-pected as nausea and vomiting are common sideeffects of bromocriptine. Vomiting is dopaminer-gically mediated. Thus cigarette smoking (andpresumably desire/urge to smoke) and vomitingare both in� uenced by dopaminergic activity.The attenuation of the � ndings by using nausea


(rate of change) as a covariate raises the questionas to whether the observed effects are the resultof dopamine agonism, subjective nausea or acombination of both. The fact that the g 2 valuesremained relatively large (Cohen, 1988) afterusing nausea as a covariate suggest that bro-mocriptine also has effects on smoking behaviorbeyond those of inducing nausea and that, witha larger sample size, the signi� cant effects onsmoking behavior that were observed before co-varying on nausea levels would still have beenobtained after covarying for nausea.

None the less, a simple effect of nausea cannotbe entirely ruled out due to limitations within thestudy. The two measures of nausea, the four-point scale and the visual analogue scale, whiledirectly worded (“nausea”) were formed ad hocand thus their reliability and validity are un-known. A thorough review of the literature re-vealed no validated self-report measure of nauseato exist. Vomiting provides a more direct behav-ioral measure of nausea but the overall incidenceof vomiting was relatively low so its use, per se, asa meaningful explanatory variable is limited (ithad no effect on selected key analyses). A greaterlimitation was the fact that nausea ratings (likemost of the self-report measures) were linked tocigarettes smoked; when participants stoppedsmoking, nausea scores were not collected. Theuse of estimated rate of change in nausea as acovariate was utilized as the best possible sol-ution to this problem, but it is certainly anabstraction of limited data. While it is clear thatbromocriptine affected smoking, the relativecontributions of dopaminergic activity that isnausea-related versus nausea-unrelated remainsto be properly tabulated. It can be said that therewas no clear evidence of supremacy of one modeof action over the other.

From a practical point of view, these resultssuggest that bromocriptine may have some valueas a smoking cessation treatment. The currentdata do indicate the possible problems with nau-sea induction. Bromocriptine prescriptions aretypically gradually titrated up to therapeutic doselevels such that tolerance may develop, therebyreducing likelihood of nausea and vomiting.However, it is also possible that tolerance to the“anti-smoking” effect observed here acutelywould also occur with long-term use. It is poss-ible that a dose lower than 2.5 mg might produceless nausea and still reduce smoking.

AcknowledgementsThis study was conducted while the � rst threethree authors were at UCLA and VA-GLAHS.Research was conducted at the VAMC-WLA.Supported by NIH grant no. RODA09570Aawarded to the � rst three authors. The authorswould like to thank Victory Engineering for sup-plying thermistors used for special equipmentdescribed within.

ReferencesCASKEY, N. H., JARVIK, M. E. & WIRSHING, W. C.

(1999) The effects of dopaminergic D2 stimulationand blockade on smoking behavior, Experimental andClinical Psychopharmacology, 7, 72–78.

CLARKE, P. B. S. (1990) Mesolimbic dopamine acti-vation: the key to nicotine reinforcement, Ciba Foun-dation Symposium, 152, 153–162.

CLARKE, P. B. S. (1992) Recent advances in under-standing the actions of nicotine in the central ner-vous system, Journal of the National Cancer InstituteMonograph, 229–238.

COHEN, J. (1988) Statistical Power Analysis for the Be-havioral Sciences (Hillsdale, NJ, Lawrence ErlbaumAssociates).

CORRIGALL, W. A. (1991) Understanding brain mecha-nisms in nicotine reinforcement, British Journal ofAddiction, 86, 507–510.

CORRIGALL, W. A. & COEN, K. M. (1991) Selec-tive dopamine antagonists reduce nicotine self-administration, Psychopharmacology, 104, 171–176.

DAWE, S., GERADA, C., RUSSELL, M. A. H. & GRAY, J.A. (1995) Nicotine intake in smokers following asingle dose of haloperidol, Psychopharmacology, 117,110–115.

HEATHERTON, T. F., KOZLOWSKI, L. T., FRECKER, R.C., FAGERSTROM, K. O. (1991) The FugerstromTest cfor Nicotine Dependence: a revision of theFagerstrom Tolerance Questionnaire, British Journalof Addiction, 86, 1119–1127.

JARVIK, M. E., MADSEN, D. C., OLMSTEAD, R. E.,IWAMOTO-SCHAAP, P. N., ELINS, J. L. & BENOWITZ,N. L. (2000) Nicotine blood levels and subjectivecraving for cigarettes, Physiology and Behavior, inpress.

KOOB, G. F. & BLOOM, F. E. (1989) Cellular andmolecular mechanisms of drug dependence, Science,242, 715–723.

MCEVOY, J. P., FREUDENREICH , O., LEVIN, E. D. &ROSE, J. E. (1995) Haloperidol increases smoking inpatients with schizophrenia, Psychopharmacology,119, 124–126.

MCNAIR, D. M., LORR, M. & DROPPELMAN, L. F.(1981) EITS Manual For the Pro� le of Mood States,(San Diego, CA, Educational and Industrial TestingService).

PONTIERI, F. E., TANDA G., ORZI, F. & DI CHIARA G.(1996) Effects of nicotine on the nucleus accumbensand similarity to those of addictive drugs, Nature,382, 255–257.


ROSE, J. E. & CORRIGALL, W. A. (1997) Nicotine self-administration in animals and humans: similaritiesand differences, Psychopharmacology, 130, 28–40.

RUSSELL, M. A. H., PETO, J. & PATEL, U. A. (1974)The general classi� cation of smoking by factorialstructure of motives, Journal of the Royal StatisticalSociety Series A: General, 137, 313–346.

SEEMAN, P. (1994) Dopamine receptors: clinical corre-lates, in: BLOOM, F. E. & KUPFER, D. J. (Eds) Psy-chopharmacology: the fourth generation of progress, pp.295–302 (New York, NY, Raven Press).

SCHUH, K. J. & STITZER, M. (1995) Desire to smokeduring spaced smoking intervals, Psychopharmacol-ogy, 120, 289–295.

SHIFFMAN, S. M. & JARVIK, M. E. (1976) Smoking

withdrawal symptoms in two weeks of abstinence,Psychopharmacology, 50, 35–39.

THORNER, M. O., SCHRAN, H. F., EVANS, W. S.,ROGOL, A. D., MORRIS, J. L. & MACLEOD, R. M.(1980) A broad spectrum of prolactin suppressionby bromocriptine in hyperprolactinemic women: astudy of serum prolactin and bromocriptine levelsafter acute administration of bromocriptine, Journalof Clinical Endocrinology and Metabolism, 50, 1026–

1033.VANCE, M. L., EVANS, W. S. & THORNER, M. O. (1984)

Drugs � ve years later: bromocriptine, Annals of Inter-nal Medicine, 100, 78–91.

WISE, R. A. & ROMPRE, P. P. (1989) Brain dopaminereward, Annual Review of Psychology, 40, 191–225.

bromocriptine reduces cigarette smoking

Documents