Chester Clinic

 

Childhood abuse and platelet tritiated-paroxetine binding in bulimia nervosa: implications of borderline personality disorder

Abuse and Paroxetine Binding in Bulimia Childhood Abuse and
Platelet Tritiated-Paroxetine Binding in Bulimia Nervosa:
Implications of Borderline Personality Disorder
Howard Steiger, Ph.D.; Stéphanie Léonard, M.Sc.; N. M. K. Ng Ying Kin, Ph.D.;
Copyright 2000 Physicians Postgraduate Press, Inc.
Cécile Ladouceur, B.A.; Dhunraj Ramdoyal, M.Sc.; and Simon N. Young, Ph.D.
Received Feb. 24, 1999; accepted Nov. 8, 1999. From the Eating Disorders Program, Douglas Hospital, Verdun (Dr. Steiger and Mss.
Léonard and Ladouceur); the Psychiatry Department, McGill University,Montreal (Drs. Steiger, Ng, and Young and Ms. Léonard); and the Research Background: Co-occurrence of bulimia ner-
Centre, Douglas Hospital, Verdun (Drs. Steiger and Ng and Mr.
Ramdoyal), Quebec, Canada. vosa and borderline personality disorder has been Some of the data reported here were collected as part of a Masters in attributed to shared factors, including childhood Psychiatry thesis prepared by Stéphanie Léonard at McGill University, abuse and disturbances in central serotonin with additional work conducted on a grant from the Fonds de la Recherche (5-hydroxytryptamine; 5-HT) mechanisms. To en Santé du Québec and the Conseil Québécois de la Recherche Sociale explore this notion, we conducted a controlled (no. RS3019). One personal copy may be printed assessment of childhood abuse and 5-HT function Preliminary findings from this study were presented at meetings of the Canadian Council of Neuropsychopharmacology, Montreal, Quebec, in bulimics with and without borderline personal- Canada, on June 8, 1998, and of the Eating Disorders Research Society, ity disorder.
Boston, Mass., on November 7, 1998. Method: Forty patients with bulimia nervosa,
Reprint requests to: Howard Steiger, Ph.D., Eating Disorders Unit, confirmed with the Eating Disorders Examination Douglas Hospital, 6875 LaSalle Blvd., Verdun (Quebec) Canada, H4H interview (14 with borderline personality disorder 1R3 (e-mail: [email protected]). and 26 without), and 25 normal-eater controlswere assessed for clinical symptoms (eating dis-turbances, mood lability, impulsivity, and disso- ulimia nervosa is defined by dietary dyscontrol and ciation) and childhood sexual and physical abuse.
Bbodily concerns, but is generally a polysympto-
We also conducted tests of platelet tritiated-paroxetine binding in blood samples from 27 of matic syndrome with a strongly characterological flavor.
the bulimics (11 with borderline personality dis- From 20% to 30% of persons with bulimia nervosa are, order and 16 without) and 16 of the controls.
for example, reported to have borderline personality dis- Results: Relative to normal eaters, bulimics
order,1,2 for which dysregulation of affects, impulsivity, showed greater affective instability, overall im- recurrent self-harm, and transient dissociative states are pulsivity, and a history of physical abuse. How-ever, borderline bulimics alone showed elevated pathognomonic.3 Co-aggregation of bulimia nervosa and motor impulsivity, dissociation, and rates of "borderline-spectrum" pathology has been attributed to sexual abuse. Paroxetine-binding tests indicated shared factors—thought to explain concurrent dysregu- no differences attributable to comorbid borderline lation of impulse controls and mood and eating behav- personality disorder, instead linking bulimia iors1,4,5—and recent attention has focused on (1) child- nervosa with or without borderline personalitydisorder to substantially reduced 5-HT transporter hood sexual and physical abuse6 and (2) disturbances in central serotonin (5-hydroxytryptamine; 5-HT) mecha- Conclusion: Results suggest relatively au-
nisms.4,5 The present study examined the specificity of as- tonomous pathologic entities: one, relevant to sociation, for bulimic and borderline syndromes, of child- bulimia nervosa, being associated with abnormal hood abuse and 5-HT disturbances.
5-HT transporter function and affective instabil-ity, but relatively independent of childhood sexual Developmental abuse and bulimia nervosa. Studies
abuse; another, relevant to borderline personality indicate 30% to 45% of persons with bulimia nervosa disorder, onto which sexual abuse, dissociative report childhood sexual abuse, and more still, physical symptoms, and behavioral impulsivity converge.
abuse.6 Such associations need not, however, imply a We propose that abnormal 5-HT function may, bulimia-specific link, given studies (1) reporting height- however, constitute one basis for the frequentco-occurrence of bulimic and borderline ened prevalences of childhood abuse in bulimic individu- als showing comorbid personality pathology, and espe- (J Clin Psychiatry 2000;61:428–435) cially borderline personality disorder,7,8 and (2) showinghalf or more of patients with borderline personality disor-der to have a positive history of childhood sexual abuse.9In light of such findings, the question arises: Is childhood J Clin Psychiatry 61:6, June 2000 Steiger et al.
abuse associated with bulimia nervosa, or with borderline interview, and concurrent psychiatric symptoms (affective personality disorder found in only some bulimic patients? instability, impulsivity, and dissociation) by question- The present study addressed this question.
naire. Serotonin function was assessed by measuring bind- Serotonin dysfunction and bulimia nervosa. Evi-
ing, in blood platelets, of the selective 5-HT reuptake in- dence shows 5-HT to moderate mood, impulsive behavior, and satiety,4,5 and this creates a rationale for the hypothesis There are various reasons for the assumption that plate- that central 5-HT mechanisms act in the predisposition to let paroxetine binding models central 5-HT transporter (or perpetuation of) bulimia nervosa. Empirical support (or reuptake) mechanisms16,17: (1) Platelets possess high af- for this notion has been impressive. Jimerson et al.10 found finity-uptake sites for 5-HT, which seem morphologically Copyright 2000 Physicians Postgraduate Press, Inc.
high-frequency binge eaters (in a normal-weight bulimic and kinetically comparable with 5-HT reuptake sites in sample) to have significantly lower levels of 5-HT me- brain.16,17 (2) Platelet binding is selectively associated with tabolites in cerebrospinal fluid than did low-frequency binding in brain tissue.16 (3) Antidepressant response in binge eaters or controls. Goldbloom et al.11 reported 22 ac- depressed outpatients coincides with normalization of 5-HT tive bulimics to have higher platelet 5-HT uptake rates reuptake inhibitor binding in the periphery.18 (4) Platelet than did 20 age-matched controls, and interpreted this to paroxetine binding has been applied as a model of 5-HT imply an adaptation to reduced 5-HT. Similarly, several function in various clinical syndromes.19,20 While ours is (to studies in bulimia nervosa have documented blunted pro- our knowledge) the first application of paroxetine binding lactin responses to 5-HT agonists or partial agonists4 (im- in bulimia nervosa, Marazziti and colleagues21 have used plying down-regulation at postsynaptic 5-HT sites). Fi- platelet imipramine binding as a model of 5-HT function nally, the selective serotonin reuptake inhibitor (SSRI) and found transporter density (B ), but not affinity (K ), One personal copy may be printed fluoxetine is found to yield clinically significant reduc- to be reduced in bulimic versus nonbulimic women.
tions in binge-eating episodes.12 While such findings indi-cate association between bulimia nervosa and 5-HT anomalies, they need not imply bulimia-specific effects.
Compared with healthy controls, patients with borderline personality disorder also show signs of decreased 5-HT Bulimic group. Forty women with bulimia nervosa
tone, or anomalous hormonal responses to 5-HT agonists,9 were recruited through a specialized outpatient service.
and clinical trials show fluoxetine to be effective in treat- Eating-disorder status was confirmed at the start of the ment of dysphoria, impulsivity, and self-mutilation in study using the Eating Disorders Examination (EDE) in- some patients who have this disorder.13 The possibility ex- terview.22 On the basis of the EDE, 33 (82.5%) women ists, therefore, that "borderline" phenomena may account met Diagnostic and Statistical Manual for Mental Disor- for some aspects of the 5-HT anomalies observed in bu- ders, Fourth Edition (DSM-IV)3 criteria for bulimia ner- limia nervosa. Our study also addressed this second issue.
vosa, purging subtype; 1 (2.5%) for bulimia nervosa, Limited data are available that bear upon the implica- nonpurging subtype; and 4 (10.0%) for a subclinical tions of borderline features for 5-HT function in bulimia bulimia nervosa purge type (bingeing once versus the nervosa: Verkes and colleagues14 found bulimics with bor- requisite twice weekly). According to interviews, our bu- derline personality disorder (N = 5) to show elevated limic participants binged on a mean ± SD of 16.96 ± 7.03 platelet 5-HT content relative to bulimics without border- days monthly at a frequency of 24.55 ± 14.35 episodes line personality disorder (N = 10) and argued that this monthly. Those who vomited did so on a mean of might reflect increased uptake associated with reduced 16.74 ± 9.95 days monthly, at a mean frequency of circulating 5-HT. Likewise, Waller and colleagues15 noted 47.17 ± 50.80 times monthly. Mean ± SD age and body self-reportedly impulsive bulimics in a small (N = 6) mass index (BMI) in this sample were 26.30 ± 6.19 years sample to show greater blunting of prolactin responses and 22.01 ± 3.48 kg/m2, respectively.
following buspirone treatment (which they presumed to Normal-eater control. Members of the normal-eater
be largely a 5-HT agonist).
control group (N = 25) were recruited through advertise- The present study. A first goal in this study was to de-
ments or university classes and were admitted to the study termine whether bulimics with and without borderline per- if they had no past or present eating disorder upon inter- sonality disorder spanned a continuum of disturbances view and no overt psychiatric history upon inquiry. All de- (with respect to psychiatric symptomatology, childhood nied bingeing, purging, or use of psychoactive medica- abuse, and 5-HT function) or showed distinct areas of dis- tions. Mean ± SD age and BMI in this group were turbance (as might suggest distinct psychopathologic 20.80 ± 3.69 years and 20.72 ± 1.85 kg/m2, respectively.
spectra). Another goal of this study was to allow an explo-ration into the association between abuse history and 5-HT function. We assessed borderline personality disor- Rating scales. Well-known interviews and question-
der, childhood abuse, and eating symptoms by structured naires were selected for demonstrated psychometric J Clin Psychiatry 61:6, June 2000 Abuse and Paroxetine Binding in Bulimia strengths and relevance to constructs of interest. We [3H]-paroxetine in the presence and absence of an excess used the EDE22 interview and the Eating Attitudes Test amount of citalopram (3 µM), was found to be between (EAT-26)23 to tap clinical eating-disorder symptoms and 70% to 90% of total binding. The apparent B BMI to reflect nutritional status. We also measured per- were obtained by Scatchard analysis of binding curves for sonality disorders using the Structured Clinical Interview the different concentrations of [3H]-paroxetine.
for DSM-IV Axis II (SCID-II),24 which we used to clas-sify all patients as either having or not having borderline personality disorder. The borderline personality disorder All participants provided written informed consent criterion referring to overeating was excluded. Interrater for research. Measures of psychopathology and child- Copyright 2000 Physicians Postgraduate Press, Inc.
reliability checks on a subsample of 17 interviews (se- hood abuse were obtained from all participants, and lected pseudorandomly to represent adequate numbers of blood samples from a subset of 27 bulimics (11 with bor- probable "borderline" and "nonborderline" diagnoses) derline personality disorder [BN/BPD] and 16 without yielded a kappa of 0.68 (representing 88.2% agreement) [BN/nonBPD]) and 16 normal-eater control (NC) partici- for a borderline/nonborderline distinction.
pants. The 5-HT indices thus represented diagnostic clas- Additional psychopathologic characteristics were evalu- sifications well. Potential sources of extraneous variation ated using the Dissociative Experiences Scale (DES)25; the on 5-HT measures necessitate controls or comment: Barrat Impulsivity Scale (BIS; version 10),26 producing (1) Contraceptive use: Given reports suggesting absence scores measuring cognitive, motor, and nonplanning im- of marked effects of oral contraceptives on blood 5-HT pulsivity; and the affective instability subscale from the Di- indices,30 we did not treat contraceptive use as an exclu- mensional Assessment for Personality Pathology-Basic sion criterion. We did, however, test for differences (on Questionnaire (DAPP-BQ).27 One personal copy may be printed Finally, to assess childhood paroxetine-binding indices) among individuals who were abuse, we used the Childhood Trauma Interview (CTI).28 or were not taking contraceptives and found no significant We used CTI severity and age indices to isolate experiences effects. (2) Seasonal effects: Seasonal variations have involving frankly inappropriate sexual or physical contacts been observed on various 5-HT indices, with studies in occurring prior to age 13 years and then up to age 18 years.
healthy volunteers reported to yield reduced paroxetine Given a bilingual population, we employed official, vali- binding in summer/fall.31 Our recruitment of participants dated French translations of the DES and EAT-26 and de- was skewed over time in such a way that any bias due to veloped French translations for other scales using careful seasonal variations should have run toward reduced bind- forward and back translation techniques. On global indi- ing in normal controls versus bulimics. Nevertheless, we ces, translations were psychometrically equivalent to cor- applied statistical controls for possible confounds due to responding English questionnaires.
seasonal effects, using previously published values31 for Paroxetine binding. Blood samples were always drawn
seasonal variations in platelet paroxetine binding (see Re- between 8:30 and 9:00 a.m., after an overnight fast. Par- sults). (3) Menses: To optimize sample size, we combined ticipants were asked to refrain from alcohol or nonpre- one group of participants tested during follicular phase scription drug use for 48 hours prior to testing and from only with another in whom testing took place on non- binge eating for 24 hours prior to testing. Whole blood was menstrual days. We tested for (and ruled out) potential collected in Vacutainer tubes containing the anticoagulant confounding effects of menstrual phase on paroxetine- EDTA and kept on ice (for no more than 30 minutes) until binding findings. (4) Medication: Six cases providing platelets were isolated by differential centrifugation. Plate- blood samples (5 BN/BPD and 1 BN/nonBPD) had let rich plasma was first isolated at 280g for 15 min at 4°C.
started medication (always an SSRI) at the time of recruit- Platelets were then isolated from the platelet rich plasma ment. To optimize sample sizes for data on childhood at 18,000g for 15 min. Next, the pellets were washed in abuse, we retained these participants and applied statisti- buffer containing EDTA/Tris/NaCl, pH 7.5, and homog- cal procedures (described below) to rule out confounds enized using a Polytron (Brinkman Instruments, Roxdale, attributable to medication effects. We note, also, that a re- Ontario, Canada). The lysed membranes were stored in a cent report indicates absence of acute effects of various small volume of buffer at –80°C until analyzed. Blood antidepressants (including paroxetine) upon platelet par- work was done under blind conditions. The binding ex- oxetine binding in healthy volunteers.32 periment was performed as described by Langer et al.29Lysed membranes (0.8 to 2.0 mg protein) were incubated in a Tris/EDTA/NaCl/KCl buffer containing 0.05 to 10 nMof [3H]-paroxetine (26.5 Ci/mmol [980.5 GBq], NEN [Life Science Products, Boston, Mass.]) for 90 min at 20°C. The According to SCID-II criteria, none of our NC partici- bound and free ligands were separated by filtration on pants had borderline personality disorder. A more sizable GF/B Whatman filters, washed 3 times with buffer, and number of our bulimic participants met borderline person- counted. Specific binding, determined by incubating ality disorder criteria (N = 14; 35.0%). When a borderline J Clin Psychiatry 61:6, June 2000 Steiger et al.
personality disorder diagnosis was present, Table 1. Mean ± SD for Borderline-Bulimic (BN/BPD),
we assigned the participant to the BN/BPD Nonborderline-Bulimic (BN/nonBPD), and Normal-Eater Control (NC)
group, and when not, to the BN/nonBPD Groups on Indices of Eating and Psychopathologic Symptoms†
group. Mean ± SD age (26.50 ± 6.25 and 26.20 ± 6.28 years, respectively) did not dif- ferentiate BN/BPD from BN/nonBPD groups. Bulimic participants were, however, slightly (and significantly) older than were Binge episodes/mo control participants (F = 7.96, df = 2,62; p < .01). Copyright 2000 Physicians Postgraduate Press, Inc.
Where the age variable was corre- lated with other indices (affective instability, BIS attention and nonplanning, and B findings were confirmed using analyses of covariance with age as a covariate. BMI yielded no group differences: mean ± SD val- ues across BN/BPD, BN/nonBPD, and NC groups were 21.75 ± 3.16, 22.15 ± 3.70, and Motor impulsivity 20.72 ± 1.85 kg/m2, respectively.
impulsivity (BIS) Table 1 shows mean ± One personal copy may be printed impulsivity (BIS) BN/BPD, BN/nonBPD, and NC groups on Total impulsivity EDE mean monthly binge and vomit indices Affective instability (the latter values computed for cases who were vomiters only) and the EAT-26. Results †Abbreviations: BIS = Barrat Impulsivity Scale, DAPP-BQ = Dimensional of t tests revealed significant borderline/ Assessment for Personality Pathology-Basic Questionnaire, DES = DissociativeExperiences Scale.
nonborderline differences on mean monthly a,b,cMeans with different letters in their superscripts differ at the .05 level or better.
binge episodes, borderline patients showing *p < .05.
**p < .001.
the higher frequencies. No corresponding dif-ferences were obtained on measures of meandays of bingeing or vomiting per month, or of mean vom- ing: on dissociation and motor impulsivity (arguably the iting episodes per month. On EAT-26 scores, analysis of most pathognomonic features of borderline personality variance (ANOVA) revealed a significant group effect (see disorder measured), pathologic elevations occurred in Table 1), Newman-Keuls tests indicating reliable bulimic BN/BPD cases, but not in BN/nonBPD cases.
versus nonbulimic differences but no borderline/non-borderline differences.
Table 2 shows numbers (and proportions) of partici- Psychiatric Symptoms pants in each group who reported sexual abuse, physical Table 1 also provides mean ± SD results for BN/BPD, abuse, or any abuse (i.e., either form of abuse), both be- BN/nonBPD, and NC groups on measures of dissociation fore age 13 years and up to age 18 years. Group effects (total score), impulsivity (motor, cognitive, nonplanning, (or trends) were obtained for data reflecting sexual abuse and total scores), and affective instability (total score).
prior to age 13 (χ2 = 5.22, df = 2, p < .08) and up to age One-way multivariate ANOVA on the total dissociation, 18 (χ2 = 10.22, df = 2, p < .01). Pairwise group compari- impulsivity, and affective instability scores yielded an sons for prevalences prior to age 13 were (given low fre- omnibus group effect (Wilks lambda = 11.07, df = 6,120; quencies in some cells) conducted using Fisher exact p < .001), and we therefore proceeded to univariate tests, and a significant difference was obtained between ANOVAs. Reliable univariate group effects were obtained BN/BPD and NC groups (p < .03) alone. Hence, elevated on all but the nonplanning impulsivity variable (see Table childhood sexual abuse seemed to be characteristic 1); those on affective instability (F = 20.80, df = 2,61; largely of BN/BPD cases and only nonsignificantly el- p < .001) and cognitive impulsivity (F = 24.18, df = 2,61; evated among BN/nonBPD bulimics.
p < .001) remained after age effects were removed To further explore an apparent association between through analyses of covariance (ANCOVAs). Nonsignifi- sexual abuse and borderline personality disorder, we con- cant results on nonplanning impulsivity were unchanged ducted an analysis to reflect associations between each when age effects were removed through ANCOVAs.
diagnostic classification and type of abuse, computing Group comparisons (Newman-Keuls) showed the follow- proportions of cases in each group who reported J Clin Psychiatry 61:6, June 2000 Abuse and Paroxetine Binding in Bulimia in both bulimic groups, although here, too, BN/BPD cases Table 2. Number and Percentage of Cases in BN/BPD,
BN/nonBPD, and NC Groups Reporting Sexual Abuse,
showed extreme rates (see Table 2). Finally, we compared Physical Abuse, or Either Form of Abuse Prior to
the groups for proportions of any abuse (combined sexual Age 13 Years and up to Age 18 Years†
and physical abuse), and found significant effects for abuse prior to age 13 (χ2 = 16.28, df = 2, p < .001) and up to age 18 (χ2 = 18.16, df = 2, p < .001). Here, exact tests (for abuse prior to age 13) differentiated BN/BPD Prior to age 13 y from control (p < .001) and BN/nonBPD from control BN/nonBPD (N = 26) (p < .02), and tended to differentiate BN/BPD from Copyright 2000 Physicians Postgraduate Press, Inc.
Prior to age 13 y BN/nonBPD groups (p = .10).
Prior to age 13 y Paroxetine Binding Results reflecting receptor B and K are shown for †Abbreviations: BN/BPD = borderline-bulimic,BN/nonBPD = nonborderline-bulimic, NC = normal-eater control.
the 3 groups (11 BN/BPD, 16 BN/nonBPD, and 16 NCcases) in Table 3. One-way ANOVAs revealed a signifi-cant group effect on B , but not on K (see Table 3).
Table 3. Mean ± SD for BN/BPD, BN/nonBPD, and NC
Group contrasts indicated mean B for both bulimic Groups on B
and K Indices From Platelet
groups to be significantly lower than that for the NC group. BN/BPD versus BN/nonBPD differences were not, however, obtained. To ensure that the group effect ob- One personal copy may be printed was not a function of age (which was corre- ), affective problems (known to be associ- ated with 5-HT function), or seasonal variations in platelet 566.13b 357.89 1047.25a 467.95 paroxetine binding,31 we repeated the analysis using as covariates age, affective instability, and finally, season †Abbreviations: Bmax = transporter density, BN/BPD = borderline-bulimic, of testing. In line with reported findings,31 we coded sea- BN/nonBPD = nonborderline-bulimic, Kd = binding affinity constant, son as a dichotomous winter/spring (high-binding) versus NC = normal-eater control.
a,bMeans with different letters in their superscripts differ at the .05 summer/fall (low-binding) distinction. Although covar- level or better.
iates never yielded significant effects, group effects in *p < .001.
each case remained significant: covarying age: F = 6.24,df = 2,39; p < .004; covarying affective instability: intrafamilial abuse (involving a first-degree relative as F = 5.58, df = 2,39; p < .01; and covarying season: perpetrator) or extrafamilial abuse (involving another per- F = 5.12, df = 2,39; p < .02.
petrator), and dividing each type into less-severe forms Similarly, to verify the possible impact of medication (involving nongenital contacts) and more-severe forms values, we repeated the ANOVA on B (involving genital contact). In this analysis, cases report- data from unmedicated subjects only (6 BN/BPD, 15 ing more than one class of abuse were counted more than BN/nonBPD, and 16 NC). The group effect remained reli- once. Respective proportions of cases in BN/BPD, able (F = 6.55, df = 2,34; p < .005), with corresponding BN/nonBPD, and NC groups who reported each class of values for BN/BPD, BN/nonBPD, and NC groups being abuse prior to age 13 were as follows: less-severe 562.00 ± 146.39, 580.00 ± 365.97, and 1047.25 ± 467.95 intrafamilial abuse: 35.7%, 15.4%, and 12.0%; more- fmol/mg protein, respectively. Newman-Keuls compari- severe intrafamilial abuse: 21.4%, 0.0%, and 0.0%; less- sons again indicated reliable differences between bulimics severe extrafamilial abuse: 14.3%, 7.7%, and 4.0%; and normal eaters, but no borderline/nonborderline differ- more-severe extrafamilial abuse: 0.0%, 7.7%, and 0.0%.
ences. Hence, results were quite comparable with those The pattern of results links intrafamilial abuse, especially obtained in our full sample (see Table 3). As a final in more-severe forms, with the BN/BPD classification test, we computed mean B scores for medicated (N = 6) (and not with BN/nonBPD).
and unmedicated (N = 21) participants who had bulimia Chi-square analysis also showed significant group ef- nervosa. Resulting values (467.67 ± 195.82 and fects for physical abuse prior to age 13 (χ2 = 21.55, 574.86 ± 314.93 fmol/mg protein, respectively) did not df = 2, p < .001) and up to age 18 (χ2 = 25.61, df = 2, differ (t = –0.79, df = 25, NS).
p < .001). Here, exact tests (performed on values prior toage 13) differentiated BN/nonBPD cases and BN/BPD Association Between Transporter Density cases from NC cases (p < .01 and p < .001, respectively), and Other Indices and BN/nonBPD and BN/BPD groups from each other To explore possible links between altered transporter (p < .05). Results thus indicated elevated physical abuse density and nutritional factors, we computed correlations J Clin Psychiatry 61:6, June 2000 Steiger et al.
(in subjects with bulimia nervosa only) between B ality disorder, although existing independently of the bor- ues and indices of nutritional status (BMI) and severity of derline personality pathology per se, may be shaped or ex- eating-disorder symptoms (EAT-26 and mean monthly aggerated by certain borderline characteristics (like im- binge days, vomit days, binge episodes, and vomit epi- sodes). None of the resulting correlations (0.28, –0.02, In contrast to the preceding, measures of cognitive im- –0.11, –0.11, –0.20, –0.24, respectively) were significant, pulsivity and affective instability differentiated bulimic implying absence of direct connection between eating be- groups from normal controls, but yielded no marked bor- haviors and reduced transporter density. We also explored derline/nonborderline distinctions (see Table 1). One im- correlations (and partial correlations, after removing vari- plication here, we assume, may be that there exists a pro-pensity toward labile moods and unreflectiveness in even Copyright 2000 Physicians Postgraduate Press, Inc.
ance due to bulimic versus nonbulimic status) betweenB values and presence of childhood abuse (coded di- chotomously as present or absent). Resulting correlationsfor variables reflecting presence or absence of sexual abuse prior to age 13 (r = –0.18; partial r = –0.17) or up to Consistent with the proposal (raised above) that bu- age 18 (r = –0.21; partial r = –0.14) and for physical abuse limia nervosa and borderline personality disorder repre- prior to age 13 (r = –0.11; partial r = 0.15) or up to age 18 sent independent psychopathologic structures, we found (r = –0.15; partial r = 0.12) were nonsignificant. Hence, bulimia nervosa in the presence of borderline personality presence of childhood abuse did not seem to be strongly disorder to coincide with substantially greater risk of predictive of alterations in paroxetine binding.
childhood sexual abuse (especially in intrafamilial forms)than did bulimia nervosa without borderline personality DISCUSSIONOne personal copy may be printed disorder. Indeed, in the present findings, risk of childhoodsexual abuse was negligibly higher among bulimics who Clinical Symptoms were nonborderline (i.e., less characterologically dis- On certain psychopathologic indices applied in this turbed) than it was among our normal-eater control par- study, we found rather clear evidence of a phenomenologi- ticipants. Such findings replicate previous results that cal discontinuity between bulimic patients with and with- have shown childhood sexual abuse to be more typical of out borderline personality disorder. Relative to normal- persons who have bulimia with comorbid personality pa- eater control participants, bulimics with the disorder thology, and especially in those with borderline personal- displayed remarkable levels of motor impulsivity and dis- ity disorder.10,11 We infer from these that childhood sexual sociation; bulimics without it, on the other hand, showed abuse may have a more specific relevance to personality no striking (or statistically significant) elevations on these pathology (in particular, borderline personality disorder) characteristics (see Table 1). Hence, the bulimic patients than to bulimia nervosa. Results on indices of physical with borderline personality disorder seemed to show a rela- abuse differed somewhat in showing a progressive in- tively unique propensity toward psychopathology of a be- crease in prevalence of abuse across normal eaters, non- haviorally impulsive or dissociative type. With respect to borderline bulimics, and bulimics with borderline symp- impulsive/dissociative potentials, our findings therefore toms. While such findings highlight the pertinence of suggest that bulimia nervosa and borderline personality abuse experiences for bulimic syndromes (and probably disorder represent rather distinct psychopathologic spectra.
for many forms of maladjustment), even here, a particu- Corroborating the same theme, eating-symptom mea- larly strong association was indicated between borderline sures provided evidence of a similar separation between personality disorder and history of abuse. We add, as a bulimic and borderline components of disturbance. Al- note, that our data on childhood abuse do not support in- though bulimics with borderline personality disorder ferences about causality. In other words, it remains to as- tended to binge more repeatedly when they did binge, all certain whether findings imply causal effects of abuse for bulimics otherwise tended to display comparable propor- borderline personality disorder or isolate abuse as a tions of days per month on which they binged and vomited marker of processes associated with vulnerability to bor- and similar levels of attitudinal distortion pertaining to derline personality disorder.
eating (on the EAT-26). These trends again point to theconclusion that eating-specific and characterological Paroxetine Binding components of disturbance in bulimia nervosa are rela- Paroxetine-binding tests yielded a somewhat different tively independent and parallel several previous reports pattern of findings, showing B (i.e., platelet 5-HT reup- that have suggested absence of overall differences in bu- take) to be significantly (and substantially) lower in both limic symptoms attributable to Axis II comorbidity.1 Nev- of our bulimic groups than it was in normal-eater controls, ertheless, to explain trends toward more dyscontrolled without differing across borderline and nonborderline bu- bingeing observed in borderline patients, we propose that limic subsamples (see Table 3). The pattern of group dif- bulimic manifestations in patients with borderline person- ferences described seemed, furthermore, to exist indepen- J Clin Psychiatry 61:6, June 2000 Abuse and Paroxetine Binding in Bulimia dently of seasonal variations, medication effects, associa- Indeed, we speculate that the apparently reduced 5-HT tions with childhood abuse, or indices of nutritional status reuptake observed here could reflect a common end state (i.e., BMI, EAT-26 scores, binge/vomit frequencies). The associated with bulimic eating but resulting from different preceding invites the conjecture that we may be observing processes in different individuals. In some individuals, a serotonergic anomaly that is implicated relatively ubiq- vulnerability to binge eating may arise from reduced uitously in bulimia nervosa and upon which presence or 5-HT activity (and corresponding, adaptive reduction in absence of borderline personality disorder has little im- density of 5-HT transporter sites) resulting, in part, from pact. In interpreting these results, we remain aware that such factors as prolonged or excessive dieting. Available platelet measures need not reflect brain 5-HT functions evidence indicates that dieting can alter 5-HT function in Copyright 2000 Physicians Postgraduate Press, Inc.
under all circumstances. Nonetheless, if (for reasons re- the fashion described4,5 and might be responsible for viewed earlier16–21) we assume that platelet binding often 5-HT–mediated effects conducive to dietary dyscontrol.
provides an approximation to central mechanisms, then Conversely, individuals with borderline personality disor- our findings might be taken to imply generally reduced der might show a primary disturbance in 5-HT function density of the 5-HT transporter in bulimia nervosa—a that could underlie these patients' unique proclivities to- finding that would be compatible with various other ob- ward trait impulsivity, mood dysregulation, and related servations indicating reduced 5-HT tone in bulimia ner- symptomatology. If the hypothetical disturbance included vosa.4,5 We can envisage several accounts for the appar- alterations in 5-HT mechanisms regulating appetitive be- ently reduced reuptake of 5-HT observed. One could havior, it might account for concurrent susceptibility in argue that we are observing a compensatory reduction in patients with borderline personality disorder to problems 5-HT reuptake, associated with dietary factors that pro- with satiation (or binge eating). This proposal might ac- One personal copy may be printed duce periodic excesses in circulating 5-HT (e.g., repeated count for a relatively generalized involvement of 5-HT overloading with 5-HT precursors during binge episodes).
disturbances in bulimia nervosa, regardless of comor- While such explanations may be viable, our findings indi- bidity, and a special affinity between bulimia nervosa and cate relative independence of 5-HT findings from nutri- disorders like borderline personality disorder that are pre- tional indices and mitigate against any account couched sumed to be, in part, 5-HT mediated.
solely in terms of dietary sequelae. Alternatively, we We add a note concerning a limitation of the present might be observing an adaptive reduction in 5-HT reup- study. In interpreting findings showing borderline/ take, corresponding to a constitutional deficit in 5-HT nonborderline distinctions, it is necessary to consider the availability, or 5-HT reuptake levels that are themselves, possible impact of uncontrolled effects arising from co- for constitutional or other reasons, simply too low.
morbid Axis I disorders (e.g., major depression or post- Present findings provide no strong indications for prefer- traumatic stress disorder) that have not been accounted ence among these alternative explanations. Regardless, for here. While concern about such influences is indeed our findings link bulimia nervosa quite strongly to a re- legitimate, we believe that our findings, even if they re- duction in platelet paroxetine binding. If findings with flect such confounds, are likely still to be informative paroxetine binding correspond to underactivity at the cen- from a purely phenomenological standpoint about that tral presynaptic 5-HT terminal, our results might justify group of bulimic patients who meet formal borderline use of SSRIs in bulimia nervosa treatment, as such treat- personality disorder criteria, and about shared and unique ment would presumably boost 5-HT activity at an appro- factors (developmental and neurobiological) that may co- priate locus in the system.
incide with phenomenologies of a bulimic and borderline In the absence of a comparison group composed of pa- type. Nonetheless, future work in bulimia nervosa needs, tients with borderline personality disorder but without bu- wherever possible, to include fuller controls for various limia nervosa, we cannot ascertain whether our results in- forms of psychiatric comorbidity.
dicate a 5-HT anomaly that is associated specifically withbulimia nervosa (and hence found uniformly across our Drug names: buspirone (BuSpar), fluoxetine (Prozac), paroxetine(Paxil).
borderline and nonborderline bulimic groups) or isequipresent in borderline personality disorder and bulimia nervosa alike (i.e., present in both syndromes, without ad- 1. Steiger H, Séguin J. Eating disorders: anorexia nervosa and bulimia ditive effects when the 2 are present concurrently). How- nervosa. In: Millon T, Blaney PH, Davis RD, eds. Oxford Textbook ever, previous evidence of reduced tritiated-paroxetine of Psychopathology. New York, NY: Oxford University Press; 1999: binding in at least certain individuals with personality 2. Vitousek K, Manke F. Personality variables and disorders in anorexia ner- disorders33 encourages us to speculate that we may be vosa and bulimia nervosa. J Abnorm Psychol 1994;103:137–147 observing an anomaly that is common to both bulimia 3. American Psychiatric Association. Diagnostic and Statistical Manual of nervosa and borderline personality disorder, and further- Mental Disorders, Fourth Edition. Washington, DC: American PsychiatricAssociation; 1994 more, that this anomaly may account for frequent co- 4. Brewerton TD. Toward a unified theory of serotonin dysregulation in eat- aggregation between borderline and bulimic syndromes.
ing and related disorders. Psychoneuroendocrinology 1995;20:561–590 J Clin Psychiatry 61:6, June 2000 Steiger et al.
5. Wolfe BE, Metzger ED, Jimerson DC. Research update on serotonin func- abuse and limitations. J Psychiatr Neurosci 1994;19:87–90 tion in bulimia nervosa and anorexia nervosa. Psychopharmacol Bull 20. Meltzer HY, Arora RC. Platelet serotonin studies in affective disorder: evi- dence for a serotonergic abnormality. In: Sandler M, Coppen A, Harnett S, 6. Wonderlich SA, Brewerton TD, Jocic Z, et al. Relationship of childhood eds. 5-Hydroxytryptamine in Psychiatry: A Spectrum of Ideas. New York, sexual abuse and eating disorders. J Am Acad Child Adolesc Psychiatry NY: Oxford University Press; 1991:50–89 21. Marazziti D, Macchi E, Rotondo A, et al. Involvement of the serotonin 7. Fullerton DT, Wonderlich SA, Gosnell BA. Clinical characteristics of eat- system in bulimia. Life Sci 1988;43:2123–2126 ing disorder patients who report sexual or physical abuse. Int J Eat Disord 22. Fairburn C, Cooper P. The Eating Disorders Examination. 12th ed. In: Fairburn C, Wilson G, eds. Binge Eating: Nature, Assessment and Treat- 8. Steiger H, Jabalpurwala S, Champagne J. Axis-II comorbidity and ment. New York, NY: Guilford Press; 1993:317–360 developmental adversity in bulimia nervosa. J Nerv Ment Dis 1996;184: 23. Garner DM, Olmsted M, Bohr Y, et al. The Eating Attitudes Test: psycho- metric features and clinical correlates. Psychol Med 1982;12:871–878 Copyright 2000 Physicians Postgraduate Press, Inc.
Paris J. Borderline Personality Disorder: A Multidimensional Approach.
24. First MB, Spitzer RL, Gibbon M, et al. Structured Clinical Interview for Washington, DC: American Psychiatric Press; 1994 DSM-IV Axis-II Personality Disorders (SCID-II, Version 2.0). New York, 10. Jimerson DC, Lesem MD, Kaye WH, et al. Low serotonin and dopamine NY: Biometric Research, New York State Psychiatric Institute; 1996 metabolite concentrations in cerebrospinal fluid from bulimic patients 25. Bernstein EM, Putnam FW. Development, reliability, and validity of a dis- with frequent binge episodes. Arch Gen Psychiatry 1992;49:132–138 sociation scale. J Nerv Ment Dis 1986;174:727–735 11. Goldbloom DS, Hicks LK, Garfinkel PE. Platelet serotonin uptake in bu- 26. Barrat E. Impulsive subtraits: arousal and information processing. In: limia nervosa. Biol Psychiatry 1990;28:644–647 Spence JT, Izard CE, eds. Motivation, Emotion and Personality. New 12. Fluoxetine Bulimia Nervosa Collaborative Study Group. Fluoxetine in the York, NY: Elsevier; 1985:137–146 treatment of bulimia nervosa. Arch Gen Psychiatry 1992;49:139–147 27. Livesley WJ, Jackson DN, Schroeder ML. Factorial structure of traits de- 13. Cornelius JR, Soloff PH, Perel JM, et al. Fluoxetine trial in borderline per- lineating personality disorders in clinical and general population samples.
sonality disorder. Psychopharmacol Bull 1990;26:149–152 J Abnorm Psychol 1992;101:432–440 14. Verkes RJ, Pijl H, Meinders AE, et al. Borderline personality, impulsive- 28. Bernstein DP, Fink L, Handelsman L, et al. Initial reliability and validity of ness, and platelet monoamine oxidase measures in bulimia nervosa and a new retrospective measure of child abuse and neglect. Am J Psychiatry recurrent suicidal behavior. Biol Psychiatry 1996;40:173–180 15. Waller DA, Sheinberger AL, Gullion C, et al. Impulsivity and neuroendo- 29. Langer SZ, Schoemaker H, Segonac A. (3H)-Paroxetine binding to One personal copy may be printed crine response to buspirone in bulimia nervosa. Biol Psychiatry 1996;39: human platelets: clues to selectivity paradox from temperature studies. Br J Psychiatry 1985;96:303–309 16. Stahl SM. Platelets as pharmacological models for the receptors and bio- 30. Ortiz J, Artigas F, Gelpi E. Serotonergic status in human blood. Life Sci chemistry of monoaminergic neurons. In: Longnecker GS, ed. Platelets: Physiology and Pharmacology. Orlando, Fla: Academic Press; 1985: 31. D'Hondt P, Maes M, Leysen JE, et al. Seasonal variation in platelet [3H]paroxetine binding in healthy volunteers: relationship to climatic 17. Lesch KP, Wolozin BL, Murphy DL, et al. Primary structure of the human platelet serotonin uptake site: identity with the brain serotonin transporter.
32. Kasahara T, Ishigooka J, Nagata E, et al. Long-lasting inhibition of 5-HT J Neurochem 1993;60:2319–2322 uptake of platelets in subjects treated with duloxetine, a potential anti- 18. Freeman AM, Stankovic MD, Bradley RJ, et al. Tritiated platelet imipra- depressant. Nihon Shinkei Seishin Yakurigaku Zasshi 1996;16:25–31 mine binding and treatment response in depressed outpatients. Depression 33. Coccaro EF, Kavoussi RJ, Sheline YI, et al. Impulsive aggression in per- sonality disorder correlates with tritiated paroxetine binding in the platelet.
19. Hrdina PD. Platelet serotonergic markers in psychiatric disorders: use, Arch Gen Psychiatry 1996;53:531–536 J Clin Psychiatry 61:6, June 2000

Table of Contents

Source: http://www.candlab.pitt.edu/wp-content/uploads/2014/11/Steigeretal2000_Bulimia.pdf