ARCHIVAL REPORT

Behavioral Responses to Catecholamine Depletion inUnmedicated, Remitted Subjects with BulimiaNervosa and Healthy Subjects

Simona Grob, Jair Stern, Lara Gamper, Hanspeter Moergeli, Gabriella Milos, Ulrich Schnyder,and Gregor Hasler

Background: Bulimia nervosa (BN) has been associated with dysregulation of the central catecholaminergic system. An instructive wayto investigate the relationship between catecholaminergic function and psychiatric disorder has involved behavioral responses toexperimental catecholamine depletion (CD). The purpose of this study was to examine a possible catecholaminergic dysfunction in thepathogenesis of bulimia nervosa.

Methods: CD was achieved by oral administration of alpha-methyl-para-tyrosine (AMPT) in 18 remitted female subjects with BN (rBN)and 31 healthy female control subjects. The study design consisted of a randomized, double blind, placebo-controlled crossover, single-site experimental trial. The main outcome measures were bulimic symptoms assessed by the Eating Disorder Examination—Questionnaire. Measures were assessed before and 26, 30, 54, 78, 102 hours after the first AMPT or placebo administration.

Results: In the experimental environment (controlled environment with a low level of food cues) rBN subjects had a greater increase ineating disorder symptoms during CD compared with healthy control subjects (condition � diagnosis interaction, p � .05). In theexperimental environment, rBN subjects experienced fewer bulimic symptoms than in the natural environment (uncontrolledenvironment concerning food cues) 36 hours after the first AMPT intake (environment � diagnosis interaction, p � .05). Serumprolactin levels increased significantly, and to a comparable degree across groups, after AMPT administration.

Conclusions: This study suggests that rBN is associated with vulnerability for developing eating disorder symptoms in response toreduced catecholamine neurotransmission after CD. The findings support the notion of catecholaminergic dysfunction as a possible traitabnormality in BN.

Key Words: Behavioral effects, bulimia nervosa, catecholaminedepletion, dopamine, norepinephrine, pathophysiology

Bulimia nervosa (BN) is a psychiatric disorder characterizedby recurrent episodes of binge eating and inappropriatecompensatory behavior to prevent weight gain. The path-

ophysiology of BN is poorly understood; however, there isgrowing evidence that neurobiological vulnerabilities contrib-ute to the pathogenesis of BN. Bulimia nervosa has beenassociated with dysregulation of central catecholaminergicsystem especially with decreased norepinephrine neurotrans-mission (1,2). Dopamine (DA) has been implicated in thevaluation of the rewarding properties of food (3) and inaddiction (4), which are likely related to the pathogenesisof BN.

One instructive technique for assessing the relationshipbetween catecholaminergic function and psychiatric disordershas involved the behavioral responses to catecholamine deple-tion (CD) achieved by oral administration of alpha-methyl-paratyrosine (AMPT) (5–7). AMPT is a competitive inhibitor ofthe rate-limiting enzyme in catecholamine synthesis, tyrosine

From the Department of Psychiatry and Psychotherapy (SG, LG, HM, GM,US), University Hospital; Collegium Helveticum (JS), Swiss FederalInstitute of Technology Zurich and University of Zurich, Zurich; andPsychiatric University Hospital (GH), University of Bern, Bern,Switzerland.

Address correspondence to Simona Grob, M.Sc., Department of Psychiatryand Psychotherapy, University Hospital, Culmannstrasse 8, 8091 Zurich,Switzerland; E-mail: simona.grob@bluewin.ch.

Received Apr 19, 2011; revised Sep 5, 2013; accepted Sep 6, 2013.

0006-3223/$36.00http://dx.doi.org/10.1016/j.biopsych.2013.09.013

hydroxylase (8), and temporarily decreases catecholamine trans-mission by depleting central dopamine and norepinephrinestores, evidenced by reduced concentrations of catecholaminesand their metabolites in plasma, urine, and cerebrospinal fluid(9,10) and decreased occupancy of striatal DA receptors byDA (11).

Most studies using CD have been conducted in affectivedisorders (7,12–14). In BN, several studies using tryptophandepletion (TD) have demonstrated a relationship betweendiminished serotonin activity and lowered mood, irritability,body image concerns, and loss of control of eating (15–18).Thus far, no study has used CD to evaluate the roles playedby norepinephrine and dopamine in the pathophysiologyof BN.

Monoamine depletion may not induce psychiatric symptomsin untreated acutely ill patients (15,19), possibly because of aceiling effect. The marked depressive responses following CD insubjects in the remitted phase of major depressive disorder whoeither were medicated with norepinephrine reuptake inhibitingantidepressant drugs (12–14) or were drug free (5,7) raised thepossibility that manifesting specific symptoms following catechol-amine depletion may constitute a neurobiological trait marker fordepression (6).

The purpose of this study was to identify a potential traitlikehypersensitivity to catecholamine depletion in BN by measuringthe CD-induced behavioral responses in remitted subjects withBN (rBN). We hypothesized that CD would induce more eatingdisorder symptoms in rBN subjects than in healthy controlsubjects. Given that the risk of BN is associated with the risk ofmood and anxiety disorders (20), we also predicted that CDwould induce mood and anxiety symptoms in remitted subjectswith BN.

BIOL PSYCHIATRY 2013;]:]]]–]]]& 2013 Society of Biological Psychiatry

2 BIOL PSYCHIATRY 2013;]:]]]–]]] S. Grob et al.

Methods and Materials

Female subjects aged 19 to 39 years who had previously metDSM-IV criteria for BN and had been in remission from BN for atleast 6 months (index subjects; n ¼ 18; length of illness, mean ¼53.7 months; time in remission, mean ¼ 29.2 months) or had nohistory of any psychiatric disorder and no major psychiatriccondition in first-degree relatives (control subjects; n ¼ 31) tookpart in this study. The screening visit included a diagnosticinterview with a psychiatrist, the Structured Clinical Interviewfor DSM-IV (21), and a physical examination. Both study groupswere recruited by advertisements in local newspapers andannouncements at the University of Zurich and the Swiss FederalInstitute of Technology Zurich. Exclusion criteria for participationwere current Axis I psychiatric disorders, a lifetime diagnosis ofpsychosis, major medical or neurological illness, psychoactivemedication exposure within the previous 6 months, lifetimehistory of substance dependence, pregnancy, suicidal ideation,and a history of suicide attempts. Remitted subjects with a historyof BN (rBN) had been in remission for at least 6 months; moreprecisely they had no recurrent episodes of binge eating and norecurrent inappropriate compensatory behavior to preventweight gain during the last 6 months (mean time in remissionfrom BN ¼ 29.2 months [SD ¼ 23.6], range: 6–84 months) at thetime of study participation.

All subjects gave written, informed consent before participa-tion. The study protocol was approved by the ethics committee ofthe Canton Zurich. The sample of this study overlaps with thesample of previous published data (22,23). We used a random-ized, double-blind, placebo-controlled, crossover design in whicheach subject underwent two identical sessions separated by atleast 7 days in which they received either AMPT or placebo. Eachsession included 2 days at the Department of Psychiatry andPsychotherapy of the University Hospital of Zurich. On a segre-gated floor, a one-bed room with separate lavatory was availablefor all participants. Thus, participants had no contact with otherhospitalized subjects. None of the rBN subjects had beenhospitalized at the Department of Psychiatry and Psychotherapybefore. During hospitalization, participants received regularnonvegetarian meals with standardized amounts of calories(Day 1 at 7 PM 650 kcal; day 2 at 7:30 AM 650 kcal, and at 12 PM

700 kcal).For the subsequent 3 days after each trial, subjects were

contacted daily by telephone for follow-up interviews. To avoidany risk of adverse reaction, a body weight–adjusted AMPT doseof 40 mg/kg body weight orally, to a maximum of 4 g, over 22hours (on Day 1 at 9 AM, 12 PM, and 7 PM; on Day 2 at 7 AM) wasadministered. During sham depletion, subjects received inactiveplacebo on Day 1 at 9 AM and 12 PM and 25 mg diphenhydramineorally on Day 1 at 7 PM and on Day 2 at 7 AM because AMPTfrequently induces mild sedation. To prevent the formation ofcrystalluria during AMPT administration, subjects were instructedto drink at least 2 L of water daily. Possible adverse reactions wereassessed regularly (26, 30, 54, 78, 102 hours after the first AMPT orplacebo administration) during hospitalization by a medicalexamination including blood pressure measurement and forsubsequent 3 days after each trial within the daily telephonefollow-up interview.

In each session, blood samples were drawn 26 hours after thefirst AMPT dose to measure serum prolactin levels.

Behavioral ratings were conducted immediately before thefirst AMPT or placebo intake (prechallenge) and 26, 30, 54, 78, 102hours after the first AMPT or placebo administration.

www.sobp.org/journal

Bulimic symptoms were assessed using the German Version ofthe Eating Disorder Examination—Questionnaire (EDE-Q) (24).EDE-Q is a 28-items self-report scale originated from the EDEinterview (25,26) designed to measure behavioral and cognitivefeatures of eating disorders. Respondents indicate the value ofparticular feelings and attitudes toward eating behavior and bodyconcerns over a definite time frame. Six of the 28 items assess thefrequencies of eating disorder-related behavior in terms ofnumber of binge-eating episodes and compensatory behavior.These items do not contribute to scale score. An EDE-Q globalscore and four separate EDE-Q subscales scores for the subscales:1, “control of eating/drive for thinness”; 2, “urge to eat/fear ofbinge eating”; 3, “weight concerns”; 4, “feeling fat/body dissat-isfaction” can be derived from the instrument. For the purpose ofthis study, the EDE-Q was adapted to a shorter time frame frompast 28 days used in the original version (24) to past 12 hours tomeasure bulimic symptoms in response to AMPT or placebo. Wetested this adapted scale in a pilot study. On the basis of ourclinical assessments, we came to the conclusion that it providesvalid data. Additional behavioral ratings included theMontgomery-Åsberg Depression Rating Scale (MADRS), YoungMania Rating Scale (YMRS), Beck Anxiety Inventory (BAI), Snaith-Hamilton Pleasure Scale (SHAPS), and Stanford SleepinessScale (SSS).

Full factorial linear mixed models with restricted maximumlikelihood estimation were applied to determine the effects ofcondition (cond), diagnosis (dx), and time on each behavioralmeasure. SPSS subcommand for fixed effects: /FIXED ¼ cond dxtime cond*dx cond*time dx*time cond*dx *time | SSTYPE (3).Mixed models effects were computed based on change scores foreach behavioral measure. For each condition (AMPT or Placebo),and each participant change scores were calculated by subtract-ing the baseline (time point 0 hours) from the score of each timepoint (26, 30, 54, 78, and 102 hours). For all models, a randomeffect for the subjects was included.

Because of the crossover study design, period was handled as2*2 (and 5, respectively) repeated measures. SPSS subcommand:/REPEATED ¼ sequence*time♠SUBJECT(id) COVTYPE(covst), wheresequence captures the real sequence of AMPT or Placebo admin-istration (randomly assigned to each patient). For each behavioralmeasure an appropriate covariance structure (covst) for theresiduals was chosen considering the lowest Akaike’s InformationCriterion. A first-order ante-dependence covariance structure (AD1)was best for the EDE-Q, a first-order factor analytic structure (FA1)for MADRS, a heterogenous first-order autoregressive structure(ARH1) for YMRS and BAI, a first-order autoregressive structure(AR1) for SHAPS, and an AD1 for SSS.

Estimated marginal means regarding the interaction betweendiagnosis and condition allowed analysis of the diagnostic groupsseparately.

The analyses of the behavioral ratings EDE-Q, MADRS, BAI,SHAPS, and SSS refer to the time in the controlled environment(time points 26 and 30 hours) since the main hypothesis appliedto this condition and time frame. An exception represented theYMRS for which a different time frame was considered (timepoints 26, 30, 54, 78, and 102 hours), given that hypomanicsymptoms were not observed before 48 hours after the first AMPTadministration in our previous study (27).

To determine whether the environment had an effect onbulimic symptoms, the experiment time frame of 102 hours wasdivided into two parts: controlled environment versus uncon-trolled environment. The controlled environment refers to thetime of the stay at the Department of Psychiatry and

S. Grob et al. BIOL PSYCHIATRY 2013;]:]]]–]]] 3

Psychotherapy (time points 26 and 30 hours). By contrast, theuncontrolled environment refers to the time after the stay at theDepartment of Psychiatry and Psychotherapy (time points 54, 78,and 102 hours).

To assess details regarding the type of BN symptoms, we ranexploratory analyses on the EDE-Q subscales.

To verify whether past episodes of depression (n ¼ 4), historyof anorexia nervosa (n ¼ 6), and the intake of psychotropic drugs(n ¼ 6) had an impact on the obtained results within the rBNgroup, the effect of these variables was also analyzed using linearmixed models.

To evaluate the relationship between CD-induced bulimicsymptoms (EDE-Q) and other clinical scales (MADRS, YMRS, BAI,SHAPS, and SSS) in the controlled environment, mixed modelsincluding clinical scales as covariates were applied.

The inclusion of a period effect and its interactions did notlead to important changes in the results; as result, we did notinclude a period effect in the analyses reported in this article. Theresults of analyses on raw scores instead of change scores alsodid not change the main findings of this study. As a result, wereport only the analyses on change scores, assuming they aremore conclusive than the raw score analyses.

Analyses were performed using SPSS 18.0 statistical software(SPSS, Chicago, Illinois). The statistical significance level was set atalpha ¼ .05.

Results

Demographic and clinical characteristics of the subject sampleare summarized in Table 1. Figure 1 shows the behavioralresponses in rBN and control subjects after AMPT/placeboadministration. In the controlled environment (time points 26and 30 hours), there was a significant condition � diagnosisinteraction for bulimic symptoms assessed by the EDE-Q (EDE-Qglobal score, F1,41.9 ¼ 4.19, p ¼ .047). rBN subjects reported morebulimic symptoms in the conditions in which they received AMPTcompared with the placebo condition (F1,41.4 ¼ 4.78, p ¼ .03),whereas AMPT did not induce bulimic symptoms in controls

Table 1. Demographic and Clinical Characteristics of Unmedicated Sub-jects with rBN and Healthy Controls*

rBN Subjects ControlsCharacteristic (n ¼ 18) (n ¼ 31)

Sex F FAge, Mean (SD), y 25.6 (4.7) 25.8 (3.8)Age at Onset, Mean (SD), y 14.4 (4.2) NABody Mass Index, Mean (SD), kg/m² 21.2 (1.7) 22.4 (2.2)Range, kg/m² 18.3–24.7 18.6–26.6

Time in Remission from BN, Mean (SD), mo 29.2 (23.6) NARange, mo 6–84 NA

Major Depression Preceding or During BN, n 4 NAPrevious AN, n 6 NATime in Remission from AN, Mean (SD), mo 98 NARange, mo 36–132 NA

Previous Psychoactive Medication,y n 6 0Time Medication Free, Mean (SD), mo 10 (3.5) NARange, mo 6–12 NA

AN, anorexia nervosa; BN, bulimia nervosa; NA, not applicable; rBN,bulimia nervosa in remission.

nSubjects with rBN and control subjects did not differ significantly inage and body mass index.

ySelective serotonin reuptake inhibitor or tricyclic antidepressant.

(F1,42.6 ¼ .27, p ¼ .61). Moreover, there was a significant condition �time interaction for bulimic symptoms assessed by the EDE-Q(EDE-Q global score, F1,81.7 ¼ 6.81, p ¼ .01).

The triple interaction of diagnosis � condition � time (EDE-Qglobal score, F1,81.7 ¼ 4.71, p ¼ .03) was significant as well. Theeffects of diagnosis (EDE-Q global score, F1,51.7 ¼ 2.52, p ¼ .12),condition (EDE-Q global score, F1,41.9 ¼ 2.01, p ¼ .16) and time(EDE-Q global score, F1,83.9 ¼ 1.44, p ¼ .23) did not reachsignificance. There was no significant diagnosis � time interaction(EDE-Q global score, F1,83.4 ¼ 2.77, p ¼ .10). The type ofenvironment (controlled vs. uncontrolled) had a significantinfluence on the experience of bulimic symptoms (environment �diagnosis interaction, F1,70.3 ¼ 4.77, p ¼ .03). In the controlledenvironment, rBN subjects experienced significantly fewer bulimicsymptoms than in the uncontrolled environment (F1,71.4 ¼ 17.84,p � .001), whereas this effect was much smaller in controlsubjects (F1,68.0 ¼ 4.24, p ¼ .04).

In the controlled environment, none of the subjects experi-enced any binge and/or purge episodes in response to AMPTversus placebo, whereas in the uncontrolled environment (54hours after the first AMPT intake), one rBN subject in the AMPTcondition showed a binge-eating episode without subsequentcompensatory behavior.

Exploratory analyses regarding EDE-Q subscales showed atriple interaction of diagnosis � condition � time for subscale 2,“urge to eat/fear of binge eating” (F1,91.3 ¼ 6.15, p ¼ .01), andsubscale 4, “feeling fat/body dissatisfaction” (F1,70.3 ¼ 6.69, p ¼ .01).Moreover, there was a significant condition � time interaction onsubscale 3, “weight concerns” (F1,77.5 ¼ 4.56, p ¼ .04), whereasthere were no significant interactions on EDE-Q subscale 1“control of eating/drive for thinness.” In none of the four EDE-Qsubscales was a significant condition � diagnosis interactionevident (subscale 1, “control of eating/drive for thinness,” F1,46.5 ¼2.35, p ¼ .13; subscale 2, “urge to eat/fear of binge eating,”F1,41.9 ¼ 1.76, p ¼ .19; subscale 3 “weight concerns,” F1,40.4 ¼ .72,p ¼ .40, and subscale 4, “feeling fat/body dissatisfaction,” F1,46.0 ¼1.98, p ¼ .16).

The type of environment (controlled vs. uncontrolled) had asignificant influence on the experience of bulimic symptoms alsoregarding EDE-Q subscale 1, “control of eating/drive for thinness”(environment � diagnosis interaction, F1,218.7 ¼ 4.02, p ¼ .04),and subscale 4, “feeling fat/body dissatisfaction” (F1,210.8 ¼ 4.19,p ¼ .04).

In subsequent analyses, the effects of past episodes ofdepression, anorexia nervosa, and intake of psychotropic drugson EDE-Q scores were analyzed within the rBN group. Nosignificant effects were evident.

Baseline mean (SD) EDE-Q global scores were for rBN subjects:drug ¼ 4.05 (5.66), placebo ¼ 3.62 (3.12), and for controls: drug ¼1.79 (2.14), placebo ¼ 1.99 (2.40). A significant effect of diagnosiswas evident in EDE-Q baseline scores (F1,47.0 ¼ 5.05, p ¼ .029),while there was no diagnosis � condition interaction (rBN, F1,47.0 ¼.495, p ¼ .485).

Regarding depressive symptoms assessed by the MADRS, inthe controlled environment, the effects of AMPT on MADRSscores did not differ between diagnostic groups (condition �diagnosis interaction F1,42.2 ¼ 1.02, p ¼ .32). A significant effect ofdiagnosis (F1,47.1 ¼ 6.63, p ¼ .01) and an effect of drug with ahigher mean value in the AMPT condition compared with theplacebo condition (F1,42.1 ¼ 4.92, p ¼ .03) were evident. Toinvestigate the depressive response to AMPT in rBN individualswith a history of major depressive disorder (MDD), we specificallycompared rBN-MDD subjects with rBN-only subjects. In the

www.sobp.org/journal

Figure 1. Behavioral responses to catecholamine depletion and placebo in unmedicated subjects with bulimia nervosa in remission (rBN group, n ¼ 18)and healthy control subjects (n ¼ 31). *Significant diagnosis effect (rBN vs. controls, p � .05). †Significant condition effect for rBN. ‡Significant conditioneffect for controls. Behavioral ratings are shown as changes scores that were calculated by subtracting baseline scores from raw scores of each time pointto increase visibility of depletion effects. AMPT, alpha-methyl-para-tyrosine; BAI, Beck Anxiety Inventory; MADRS, Montgomery-Åsberg Depression RatingScale; SHAPS, Snaith-Hamilton Pleasure Scale; SSS, Stanford Sleepiness Scale; YMRS, Young Mania Rating Scale.

4 BIOL PSYCHIATRY 2013;]:]]]–]]] S. Grob et al.

controlled environment, there was a difference in the MADRSresponse to AMPT between these two groups (condition �

www.sobp.org/journal

diagnosis interaction, F1,17.1 ¼ 12.5, p ¼ .003). Compared withplacebo medication rBN-MDD subjects reported a significant

S. Grob et al. BIOL PSYCHIATRY 2013;]:]]]–]]] 5

increase in depressive symptoms under AMPT (F1,16.4 ¼ 12.1,p ¼ .003), whereas this was not the case for rBN-only subjects(F1,15.1 ¼ .85, p ¼ .37). Baseline mean (SD) MADRS scores were forrBN subjects: drug ¼ 2.83 (3.48), placebo ¼ 2.61 (2.77); and forcontrols: drug ¼ .52 (.99), placebo ¼ 1.16 (2.83). A significanteffect of diagnosis was evident in MADRS baseline scores (F1,47.0 ¼9.39, p ¼ .004), whereas there was no diagnosis � conditioninteraction (rBN, F1,47.0 ¼ .996, p ¼ .323).

Hypomanic symptoms as assessed by the YMRS (across alltime points), were significantly higher in rBN subjects relative tocontrol subjects (F1,61.0 ¼ 7.07, p ¼ .01). A significant effect ofdrug was evident across groups (F1,94.8 ¼ 8.82, p ¼ .004).However, there was no significant condition � diagnosis inter-action (F1, 94.8 ¼ 1.96, p ¼ .16).

Regarding anxiety symptoms assessed by the BAI, in thecontrolled environment no effect of diagnosis (F1,51.1 ¼ .17, p ¼.68), no effect of drug (F1,89.1 ¼ .15, p ¼ .70) and no significantcondition � diagnosis interaction (F1,89.1 ¼ .39, p ¼ .53) wasevident.

Concerning the ability to experience pleasure as assessed bythe SHAPS, in the controlled environment, there was no effect ofdiagnosis (F1,39.2 ¼ .23, p ¼ .64), no effect of drug (F1,111.2 ¼ .62,p ¼ .43) and no significant condition � diagnosis interaction(F1,111.2 ¼ 1.70, p ¼ .19).

In the controlled environment, rBN subjects had highersleepiness scores assessed by the SSS than control subjects(F1,46.7 ¼ 11.74, p ¼ .001). Across groups, AMPT induced moresleepiness than the active placebo with a total of 100 mgdiphenhydramine (F1,47.3 ¼ 9.49, p ¼ .003). However, no signifi-cant condition � diagnosis interaction was evident (F1,47.3 ¼ .48,p ¼ .49).

In the controlled environment, the covariates MADRS (F1,117.4 ¼.21, p ¼ .65), YMRS (across all time points; F1,106.5 ¼ 1.68, p ¼ .19),BAI (F1,116.2 ¼ .93, p ¼ .34), SHAPS (F1,77.2 ¼ .45, p ¼ .50), and SSS(F1,97.9 ¼ .79, p ¼ .37) had no significant effect on bulimicsymptoms as assessed by the EDE-Q global score.

Serum prolactin levels were significantly higher in the AMPTcondition versus the placebo condition (mean [SD], 42.0 [2.5] vs.29.5 [2.6] μg/L; F1,36.5 ¼ 20.93, p � .001). There was no diagnosiseffect (F1,39.3 ¼ .09, p ¼ .76) and no diagnosis � conditioninteraction (F1,36.5 ¼ .16, p ¼ .69) regarding serum prolactinconcentration.

Discussion

To our knowledge, this is the first study that examined thebehavioral effects of CD in BN. The results indicate that CDinduced a transient reappearance of mild eating disorder symp-toms in remitted subjects with a history of BN. Moreover, theinduction of mild BN symptoms through exposure to theuncontrolled environment is consistent with persistent vulner-ability to BN in fully remitted subjects (28). Contrary to ourhypothesis, we did not find any significant condition � diagnosisinteractions regarding depressive, manic, anxiety, anhedonia, andsleepiness symptoms in rBN. As in our previous CD study in fullyremitted subjects with MDD (7), rBN subjects with a MDD historyshowed a significant return of depressive symptoms after CD.These results suggest that depression comorbid with BN hasneurochemistry no different from major depression alone.

Several types of indirect evidence suggest that dysfunction ofthe central noradrenergic system contributes to the risk of BN. Astudy with BN subjects abstinent from bingeing and vomiting (29)

showed reduced basal plasma and cerebrospinal fluids, suggest-ing that noradrenergic disturbance might be a traitlike character-istic of BN. The effectiveness of noradrenergic antidepressants inBN (30) underlines the clinical relevance of these findings. AlteredDA activity in BN has been identified, although the evidence isless consistent. Low DA metabolite concentrations in cerebrospi-nal fluid have been reported in BN subjects with frequent bingeepisodes (31,32). In addition, highly palatable foods induce potentrelease of DA into the nucleus accumbens (33), similar to drugs ofabuse, and DA is a key neurotransmitter in novelty seeking andthe development of addiction (4), which suggests that abnor-mal DA activity may contribute to binge-eating episodes andother impulsive behaviours associated with BN. A study usingpositron emission tomography with [11C]raclopride detecteddecreased striatal DA neurotransmission in patients with BNrelative to controls, a similar pattern to that described inaddiction disorders (34). This finding may relate to reward-processing dysfunctions that have been found in remitted BN.Moreover, in rBN, compared with control subjects, a functionalmagnetic resonance imaging study using a glucose tasteparadigm found reduced anterior cingulate cortex activity, anarea that is involved in error monitoring but also in theanticipation of reward (35). These results suggest that DA-related reward-processing dysfunctions are traitlike character-istics associated with BN.

Several strengths of the current study deserve mention.Whereas in previous studies using AMPT doses greater than 4g, subjects experienced adverse reactions such as dystonicreactions (36), restlessness (37), crystals in urine, and decreasedblood pressure (38), none of our participants reported anysignificant adverse reactions, probably because of the use of alow, body-weight-adjusted AMPT dose. The sample size wasrelatively large for a complex pharmacologic challenge study.The use of an active placebo (diphenhydramine) to mimic theside effects of the experimental drug (AMPT) contributed to aneffective blinding of the study drug and reduced the potentiallyconfounding effect of sleepiness, one of the main adverse effectsof AMPT, on psychiatric ratings. As expected, CD-induced sleepi-ness did not correlate with CD-induced bulimic symptoms. Thefact that CD induced the same amount of prolactin in rBNsubjects and healthy controls suggests that there was no differ-ence of the CD effect on catecholamine synthesis betweengroups (39).

Some limitations of this study should be considered. Thepsychiatric symptoms induced by CD were minor, which calls theclinical relevance of the findings into question. However, usingthe same low, body-weight-adjusted AMPT dose, a previous studyin fully remitted subjects with MDD also induced only milddepressive symptoms, which helped to elucidate a neurocircuitrythat plausibly relates to the catecholamine-related pathogenesisof depression (7). A previous TD study also induced only minorbulimic symptoms in rBN (17), suggesting that this is a generallimitation of modern monoamine depletion studies using safeand ethically justifiable dosages. Our cross-sectional design couldnot establish whether the bulimic response to CD in rBN reflectedan endophenotypic vulnerability to eating disorders or a con-sequence of illness. The lack of more objective outcome measure-ments represents a shortcoming of this study. The inclusion of afood challenge would have provided more reliable data. Theinclusion of lab values such as potassium would have contributedto a more objective definition of recovery. We did not reliablyassess the phase of the menstrual cycle, and subjects were testedin both the follicular and luteal phases, which may represent a

www.sobp.org/journal

6 BIOL PSYCHIATRY 2013;]:]]]–]]] S. Grob et al.

potential confounding factor. However, a previous study did notreveal any effects of the menstrual cycle on CD-induced symp-toms (7). As a result of the inclusion criterion that rBN subjectshad to be in remission and off medication, a selection bias mayhave been introduced, resulting in a sample with relatively mildforms of BN. Although the results of subsequent analysesconsidering the impact of past episodes of depression, anorexianervosa, and intake of psychotropic drugs on EDE-Q scores didnot indicate a significant effect, the influence of these potentiallyconfounding variables cannot be ruled out because of theinsufficient power of the stratified analyses. In the controlledenvironment the condition � diagnosis interaction was absent.However, in the uncontrolled environment, catecholamines werenot expected to be further depleted. As a result, this study doesnot provide insights into the interaction of drug condition,diagnosis, and uncontrolled environment. Not correcting forexperiment-wise error represents another limitation of the study;as a result, the findings should be considered exploratory. Finally,the specificity of our results was limited by CD’s effect of reducingsynthesis of norepinephrine as well as DA.

The provocation of mild BN symptoms by CD suggests that BN isassociated with persistent vulnerability for developing eating disordersymptoms in response to reduced catecholamine neurotransmission.The absence of significant condition � diagnosis interactions regard-ing mood and anxiety symptoms suggests that disease-specific riskfactors rather than reduced catecholamine neurotransmission per sedetermine the type of symptoms induced by CD.

Given the extensive interconnection among the monoaminesystems, the specificity of the contribution of serotonin, norepi-nephrine, and dopamine dysfunctions to psychiatric conditionshas been a matter of debate. In fully remitted subjects with MDD,CD induced similar symptoms as TD (7,40), although somedifferences emerged: CD induced less depressed mood andsadness but more lassitude and anxiety than TD (Hasler et al.,unpublished data, 2008). In the previous TD study in rBN (17), TDinduced significantly more depressive symptoms as measured bythe Hamilton Depression Rating Scale in rBN than in controls.Even though caution is required in combining findings from TDstudies with results of this study, behavioral responses to CD andTD seem to be similar in rBN subjects and controls. Differencesbetween studies may suggest that catecholamine activity mightbe more specifically related to eating disorder psychopathologythan serotonin activity in rBN. Stratified analyses suggest thatdifferences between the two studies were not related to thehigher rate of rBN subjects with a history of MDD in the TD thanin this CD study.

CD did not induce depressive and anxiety symptoms in rBNsubjects. This is in contrast to previous studies on the centralserotonergic dysfunction in BN, demonstrating that TD inducednonspecific mood symptoms in rBN (16,17). Together, these findingssuggest that the relationship between central catecholaminergicneurotransmission and bulimic symptoms is relatively specific in rBN.

In summary, our findings indicate that decreased catechola-minergic neurotransmission after CD may trigger the transientreappearance of eating disorder symptoms in individuals vulner-able to BN. This result supports the assumption of a catechola-minergic dysfunction in the pathophysiology of BN. Additionalinvestigations are encouraged to elucidate the neural and geneticunderpinnings of this potential traitlike dysfunction.

This research was supported by the Swiss National ScienceFoundation Nr. 32003B-117763.

www.sobp.org/journal

The authors report no biomedical financial interests or potentialconflicts of interest. Simona Grob and Gregor Hasler had full accessto all of the data in the study and take responsibility for the integrityof the data and the accuracy of the data analysis.

1. Buckholtz NS, George DT, Davies AO, Jimerson DC, Potter WZ (1988):Lymphocyte beta-adrenergic receptor modification in bulimia. ArchGen Psychiatry 45:479–482.

2. George DT, Kaye WH, Goldstein DS, Brewerton TD, Jimerson DC (1990):Altered norepinephrine regulation in bulimia: Effects of pharmaco-logical challenge with isoproterenol. Psychiatry Res 33:1–10.

3. Fulton S (2010): Appetite and reward. Front Neuroendocrinol 31:85–103.

4. Koob GF, Volkow ND (2010): Neurocircuitry of addiction. Neuro-psychopharmacology 35:217–238.

5. Berman RM, Narasimhan M, Miller HL, Anand A, Cappiello A, Oren DA,et al. (1999): Transient depressive relapse induced by catecholaminedepletion: potential phenotypic vulnerability marker? Arch Gen Psy-chiatry 56:395–403.

6. Hasler G, Drevets WC, Manji HK, Charney DS (2004): Discoveringendophenotypes for major depression. Neuropsychopharmacology 29:1765–1781.

7. Hasler G, Fromm S, Carlson PJ, Luckenbaugh DA, Waldeck T, Geraci M,et al. (2008): Neural response to catecholamine depletion in unmedi-cated subjects with major depressive disorder in remission andhealthy subjects. Arch Gen Psychiatry 65:521–531.

8. Nagatsu T, Levitt M, Udenfriend S (1964): Tyrosine hydroxylase. Theinitial step in norepinephrine biosynthesis. J Biol Chem 239:2910–2917.

9. Mignot E, Laude D (1985): Study of dopamine turnover by monitoringthe decline of dopamine metabolites in rat CSF after alpha-methyl-p-tyrosine. J Neurochem 45:1527–1533.

10. Stine SM, Krystal JH, Petrakis IL, Jatlow PI, Heninger GR, Kosten TR,et al. (1997): Effect of alpha-methyl-para-tyrosine on response tococaine challenge. Biol Psychiatry 42:181–190.

11. Verhoeff NP, Christensen BK, Hussey D, Lee M, Papatheodorou G,Kopala L, et al. (2003): Effects of catecholamine depletion on D2receptor binding, mood, and attentiveness in humans: A replicationstudy. Pharmacol Biochem Behav 74:425–432.

12. Bremner JD, Vythilingam M, Ng CK, Vermetten E, Nazeer A, Oren DA,et al. (2003): Regional brain metabolic correlates of alpha-methylparatyrosine-induced depressive symptoms: Implications forthe neural circuitry of depression. JAMA 289:3125–3134.

13. Delgado PL, Miller HL, Salomon RM, Licinio J, Heninger GR, GelenbergAJ, et al. (1993): Monoamines and the mechanism of antidepressantaction: Effects of catecholamine depletion on mood of patientstreated with antidepressants. Psychopharmacol Bull 29:389–396.

14. Miller HL, Delgado PL, Salomon RM, Berman R, Krystal JH, HeningerGR, et al. (1996): Clinical and biochemical effects of catecholaminedepletion on antidepressant-induced remission of depression. ArchGen Psychiatry 53:117–128.

15. Bruce KR, Steiger H, Young SN, Kin NM, Israel M, Levesque M (2009):Impact of acute tryptophan depletion on mood and eating-relatedurges in bulimic and nonbulimic women. J Psychiatry Neurosci 34:376–382.

16. Kaye WH, Gendall KA, Fernstrom MH, Fernstrom JD, McConaha CW,Weltzin TE (2000): Effects of acute tryptophan depletion on mood inbulimia nervosa. Biol Psychiatry 47:151–157.

17. Smith KA, Fairburn CG, Cowen PJ (1999): Symptomatic relapse inbulimia nervosa following acute tryptophan depletion. Arch GenPsychiatry 56:171–176.

18. Weltzin TE, Fernstrom MH, Fernstrom JD, Neuberger SK, Kaye WH(1995): Acute tryptophan depletion and increased food intake andirritability in bulimia nervosa. Am J Psychiatry 152:1668–1671.

19. Miller HL, Delgado PL, Salomon RM, Heninger GR, Charney DS (1996):Effects of alpha-methyl-para-tyrosine (AMPT) in drug-free depressedpatients. Neuropsychopharmacology 14:151–157.

20. Kendler KS, Walters EE, Neale MC, Kessler RC, Heath AC, Eaves LJ(1995): The structure of the genetic and environmental risk factors forsix major psychiatric disorders in women. Phobia, generalized anxietydisorder, panic disorder, bulimia, major depression, and alcoholism.Arch Gen Psychiatry 52:374–383.

S. Grob et al. BIOL PSYCHIATRY 2013;]:]]]–]]] 7

21. First M, Spitzer RL, Williams JBW (2001): Structured Clinical Interview forDSM-IV-TR Axis I Disorders, Research Version, Patient Edition (SCID-I/P).New York: Biometrics Research, New York State Psychiatric Institute.

22. Grob S, Pizzagalli DA, Dutra SJ, Stern J, Morgeli H, Milos G, et al. (2012):Dopamine-related deficit in reward learning after catecholaminedepletion in unmedicated, remitted subjects with bulimia nervosa.Neuropsychopharmacology 37:1945–1952.

23. Homan P, Grob S, Milos G, Schnyder U, Hasler G (2013): Reduction intotal plasma ghrelin levels following catecholamine depletion: relationto bulimic and depressive symptoms. Psychoneuroendocrinology 38:1545–1552.

24. Hilbert A, Tuschen-Caffier B (2006): Eating Disorder Examination: Deutsch-sprachige Übersetzung. Münster, Germany: Verlag für Psychotherapie.

25. Cooper Z, Cooper PJ, Fairburn CG (1989): The validity of theeating disorder examination and its subscales. Br J Psychiatry 154:807–812.

26. Sysko R, Walsh BT, Fairburn CG (2005): Eating Disorder Examination—Questionnaire as a measure of change in patients with bulimianervosa. Int J Eat Disord 37:100–106.

27. Hasler G, Luckenbaugh DA, Snow J, Meyers N, Waldeck T, Geraci M,et al. (2009): Reward processing after catecholamine depletion inunmedicated, remitted subjects with major depressive disorder. BiolPsychiatry 66:201–205.

28. Uher R, Brammer MJ, Murphy T, Campbell IC, Ng VW, Williams SC, et al.(2003): Recovery and chronicity in anorexia nervosa: Brain activityassociated with differential outcomes. Biol Psychiatry 54:934–942.

29. Kaye WH, Gwirtsman HE, George DT, Jimerson DC, Ebert MH, Lake CR(1990): Disturbances of noradrenergic systems in normal weightbulimia: Relationship to diet and menses. Biol Psychiatry 27:4–21.

30. Walsh BT, Wilson GT, Loeb KL, Devlin MJ, Pike KM, Roose SP, et al.(1997): Medication and psychotherapy in the treatment of bulimianervosa. Am J Psychiatry 154:523–531.

31. Jimerson DC, Lesem MD, Kaye WH, Brewerton TD (1992): Lowserotonin and dopamine metabolite concentrations in cerebrospinalfluid from bulimic patients with frequent binge episodes. Arch GenPsychiatry 49:132–138.

32. Kaye WH, Ballenger JC, Lydiard RB, Stuart GW, Laraia MT, O’Neil P,et al. (1990): CSF monoamine levels in normal-weight bulimia:Evidence for abnormal noradrenergic activity. Am J Psychiatry 147:225–229.

33. Lutter M, Nestler EJ (2009): Homeostatic and hedonic signals interactin the regulation of food intake. J Nutr 139:629–632.

34. Broft A, Shingleton R, Kaufman J, Liu F, Kumar D, Slifstein M, et al.(2012): Striatal dopamine in bulimia nervosa: A PET imaging study. IntJ Eat Disord 45:648–656.

35. Frank GK, Oberndorfer TA, Simmons AN, Paulus MP, Fudge JL, YangTT, et al. (2008): Sucrose activates human taste pathways differentlyfrom artificial sweetener. Neuroimage 39:1559–1569.

36. McCann UD, Penetar DM, Belenky G (1990): Acute dystonic reaction innormal humans caused by catecholamine depletion. Clin Neuro-pharmacol 13:565–568.

37. Laruelle M, D’Souza CD, Baldwin RM, Abi-Dargham A, Kanes SJ,Fingado CL, et al. (1997): Imaging D2 receptor occupancy byendogenous dopamine in humans. Neuropsychopharmacology 17:162–174.

38. Brogden R, Heel RC, Speight TM, Avery GS (1981): Alpha-methyl-p-tyrosine: A review of its pharmacological and clinical use. Drugs 21:81–89.

39. Freeman ME, Kanyicska B, Lerant A, Nagy G (2000): Prolactin:Structure, function, and regulation of secretion. Physiol Rev 80:1523–1631.

40. Neumeister A, Nugent AC, Waldeck T, Geraci M, Schwarz M, Bonne O,et al. (2004): Neural and behavioral responses to tryptophan depletionin unmedicated patients with remitted major depressive disorder andcontrols. Arch Gen Psychiatry 61:765–773.

www.sobp.org/journal