Copyright © 1988 PMA Publishing Corp.The Eating DisordersEdited by BJ Blinder, BF Chaitin, & R Goldstein

Chapter 31

The Relationship of the Eating and Affective DisordersBaru F. Chaitin

INTRODUCTION

t the turn of the century, Freud [1] speculatedabout the relationship between anorexia ner-vosa and melancholia. Abraham, in writing of a

patient suffering from excessive hunger and binge eat-ing, was impressed with the addictive intoxicating aspectof the behavior, which he interpreted as substitute libid-inal gratification [2]. Over the ensuing years, much ofthe psychoanalytic focus in anorexia nervosa in adoles-cent females revolved around fear of oral impregnation,denial of genitality, and the concept of organ neurosis inwhich the ego is fixated at the oral phase and producessecondary hormonal changes. Fenichel, in his psychoan-alytic compendium, points out that anorexia nervosacan be the result of a variety of dynamic processes andmay be an affective equivalent of a depression that ante-dates other signs of depression [3]. He also mentions cy-clic food addiction in woman and its relationship tomanic depressive illness as well as the relationship ofbulimia to mourning. Working from an existentialframework, Binswanger detailed the case history of hisanorexic patient, Ellen West, and described her over-whelming sense of despair and alienation [4]. Bruch,also in moving beyond the standard psychoanalytic con-flict formulations, emphasized the self-despair and per-vasive sense of personal ineffectiveness in patients withanorexia nervosa [5]. From a phenomenologic orienta-tion, these depictions may equate with a depressive ill-ness.

In recent years, growing theoretical and clinical inter-est in the eating disorders has led to increased precisionin diagnosis. As the eating disorders have been morecarefully studied, more distinct subtypes have been de-fined [6-9]. Now it is perhaps more correct to talk of pri-mary restrictive anorexia nervosa, bulimia nervosa, andnormal-weight bulimia. It is also important to recognizethat these syndromes are not completely static, and theclinical course for any particular patient may cross diag-nostic boundaries. However, with this fractionation ofthe eating disorders and the emergence of bulimia as adiagnostic entity, the relationship to the affective dis-orders has become more visible.

At the same time, the boundaries of the affective dis-orders have come under new scrutiny and revision.What was previously passed off as personality andcharacter, and ergo not susceptible to biological inter-vention, is being reexamined—albeit with a high-resolu-tion clinical microscope. Akiskal, in his study of subaf-fective disorders, has uncovered many treatable condi-tions that formerly were ascribed to personality [10].Keller and Shapiro's work on "double depression" hasalso been enlightening by recognizing the coexistence ofa depressive personality core with a superimposed affec-tive episode [11]. The borderline quality of manypatients with eating disorders has been frequently ob-served and may rivet the clinician's attention. Carroll etal caution that the presence of dramatic psycho-pathology may mask the more traditional signs andsymptoms of major depressive disorders [12].

345

346 The Eating Disorders

As in all diagnostic schema, how hard one looks andwhere one draws the line determines the yield. It is ex-tremely important to take a careful and detailed historyfrom the patient and, when possible, obtain corrobora-tive history from family members, as the compelling na-ture of the eating disorder symptoms may indeed ob-scure other pathology. If one insists on full DSM-III cri-teria for a major affective disorder in eating disorderpatients before considering a codiagnosis and pharma-cologic intervention, a number of treatable patients maybe missed.

In this chapter, I will briefly review the current linesof inquiry and data concerning an affective disorder-eat-ing disorder linkage without attempting a thorough cri-tique of research strategies.

STUDIES OF PSYCHOPATHOLOGY

In an often-cited study, Cantwell followed 26 patientswho had been hospitalized as adolescents for anorexianervosa [13]. The distinction was not made betweenpure restricters and bulimics. Almost 50% of thepatients, by parent and self-report, were diagnosed ashaving an affective disorder at follow-up. In a retrospec-tive evaluation of premorbid psychiatric symptoma-tology, 61% of the patients reported dysphoric mood,and 71% reported being fearful. These findings agreewith three earlier studies by Dally [14], Theander [15],and Morgan and Russell [16], who reported significantpsychiatric morbidity at initial presentation and follow-up.

Stonehill and Crisp found that anorexia nervosapatients scored significantly higher than healthy femalestudents on the anxiety, obsessional somatic, and de-pression scales of a standardized instrument [17]. Whenpatients who vomited were separated out, the depres-sion scores were even higher. However, the anorexicgroup scored significantly lower on the anxiety, somatic,phobic, and depression scales than a matched group ofdepressed outpatients. Stonehill and Crisp felt that thislent support to the view that anorexia nervosa is not asimple variant of depression. They ascribed the highlevel of scores on the depression scale to the overrepre-sentation of early morning awakening, which is quiteprominent in anorexia nervosa supposedly apart fromany affective disorder.

In discussing what he termed "bulimia nervosa" asan ominous variant of anorexia nervosa, Russell re-ported on 30 patients (28 females and 2 males), 24 ofwhom suffered at least moderate depression [8]. Rus-sell doubted that there were primary depressives, be-cause the eating disorder did not appear to improve withthe relief of depressive symptoms, and the depressive

symptoms did not seem to be of an endogenous variety.Ben-Tovin et al used the Present State Examination

to evaluate the mental status of 21 anorexics and foundthat they did not closely resemble any other publishedgroup [18]. The predominant affect was depressive,which differentiated them from patients with anxietystates. There was some similarity to an inpatient sampleof depressives in terms of the mixture of symptoms, butthe anorexic group could be separated by the absenceof depressive delusions, ideas of reference, and retarda-tion. Out of this anorexic group, which included 12 ab-stainers and nine vomiters, a case for a clinical depres-sion using operational criteria could be made for certainin only three patients, with seven patients seen as prob-able cases of depression. In a depressive sample of 23patients, 16 definites and three probable cases wereidentified. The spread and diagnostic frequency be-tween these two groups was significant and lead theauthors to conclude that the findings they uncoveredwere possibly related to starvation states and their sec-ondary effects, the ongoing existential struggles of theanorexic, a reflection of underlying personality, or acoincidental finding. Despite the differences in depres-sive symptoms reported by anorexics and depressives,Ben-Tovin et at were impressed by the qualitative de-pressive feelings identified by their patients and sug-gested that the measures of depression used were notable to tap the existential despair and guilt of the ano-rexic with its focus on the loathing of the body and fearof losing control over eating and resultant weight gain.

In their comprehensive study of 105 female inpatientanorexics, Casper et al gave further support to the no-tion that bulimics constitute a subgroup of anorexiahighly impacted by depression, obsessionalism, and so-matization [6]. They make the point that bulimia is notsimply a disorder of appetite but a highly complex symp-tom with depression, anxiety, and guilt interwovenwithin its matrix. To some degree, their conclusions areat variance with Russell [8] and Stonehill and Crisp [17],who tended to minimize the affective linkage of the ill-ness.

Using the same 105 patient sample, Eckert et at con-cluded that anorexia nervosa patients manifest a clini-cally significant level of depression as measured bysymptoms and mood [19]. The severity of the depres-sion correlated with other features of the illness felt tobe indicative of greater severity, such as more bulimicvomiting, greater disturbance in body image, greaterdenial, more impairment in relationship with theirfathers, and sexual disturbance. Depression was ob-served to be reduced but not eliminated over time, andpatients who gained more weight were observed to ex-perience a greater reduction in depression. While

The Eating & Affective Disorders 347

clearly recognizing that weight loss and depression existtogether, the authors did not feel that their data allowedthem to draw any etiological conclusions.

Using DSM-III criteria, Strober was able to assign adiagnosis of major depression to 41% of his bulimicpatients but to only 9% of the restricters [20]. Premor-bid personality characteristics differed between the twogroups, with the bulimic patients often described as un-happy, fearful, clinging, and quarrelsome during child-hood. Strober saw these qualities as implying affectiveinstability.

Pyle et al studied 34 bulimic patients of whom mostreported feelings of depression and problems with inter-personal relationships and self concept [21]. MeanMMPI scores on the D, Pd, Pt, and Sc scales were atleast two standard deviations above the norm. Therealso appeared to be a high rate of substance abuse inthese patients. Over 75% of them reported a decreasein sexual interest since the start of their illness, and 90%identified some traumatic event being associated withthe onset of the bulimia, with the most frequent eventbeing a loss or separation. This latter finding may be ret-rospective elaboration or could truly represent thebeginning of a depressive episode.

Hendren reviewed the case records of all patients di-agnosed as having anorexia nervosa at the Mayo Clinicover a seven-year period [22]. From the 230 cases withan anorexia nervosa diagnosis, 84 women were eventu-ally evaluated on the basis of research diagnostic criteria(RDC) for major depressive disorder and endogenousmajor depressive disorder. Fifty-six percent met theRDC criteria for a major depressive disorder and 35%met RDC criteria for an endogenous depression.MMPI data was also reviewed and agreed in large partwith the results reported above.

The trend of the above studies is in the direction of asignificant relationship between eating disorders—par-ticularly bulimia and affective illness. The lack of con-sistency in defining depression from study to studymakes the drawing of general conclusions difficult.

FAMILY STUDIES

The idea that affective disorders run in families hasreceived greater credence over the last 20 years. Re-cently, the Yale-NIMH collaborative family study of de-pression reported on a study of psychiatric disorders in2,003 first-degree relatives of 335 probands and foundconsiderable support for a familial relationship withdifferential rates and types of illness observed depend-ing on the diagnosis of the proband [23]. Of additionalpossible significance for our discussion was the findingof Leckman and his colleagues that appetite disturbanceand excessive guilt in depressed probands was as-

sociated with increased rates of major depressionamong family members with a 2.5 times greater risk forthe same subtype [24]!

While not directly addressing the relationship of eat-ing disorders to affective illness, Crisp et al studied therelationship of parental psychoneurotic characteristicsto the prognosis of patients with anorexia nervosa [25].They found that after the patient's weight restoration,maternal anxiety and paternal depression increased ifthe marital relationship was judged to be poor. Also, inpatients who demonstrated bulimic symptoms, thefathers were seen as having an undue preoccupationwith discipline and self-control and having a propensityfor developing affective illness. Increased levels ofanxiety and depression emerged with the daughter's re-covery. Poor outcome was associated with high levels ofparental psychiatric morbidity particularly in respect todepression. This study, while seeming to make a strongstatement about anorexia nervosa in adolescents and itsdynamic relationship to parental psychopathology,could also be interpreted as supporting the idea that af-fective loading and a particularly severe form of eatingdisorder may be related.

Cantwell et al reported a high degree of affective ill-ness in the families of the anorexic patients he studied[13]. Out of 26 patients studied, 2 fathers, 15 mothers,6 siblings, and 3 maternal grandparents were given a di-agnosis of affective disorder. Four of the mothers hadmade suicide attempts. After affective disorders, alco-hol abuse was diagnosed most frequently, appearing in16 relatives.

Winokur et al studied the presence of primary affec-tive disorders in the relatives of 25 patients with anorexianervosa and 25 normal controls according to RDC cri-teria [26]. Affective disorders were present in 22% ofthe relatives of the anorexic patients and in only 10% ofcontrols. Of the 43 relatives of anorexics who had an af-fective disorder, 34 were given a diagnosis of unipolardepression, and 9 were given a bipolar diagnosis. In bothgroups, more female than male relatives had a historyof affective disorder, but the rate was significantly higherin female relatives of anorexic patients. Seventy-six per-cent of the anorexic families had at least one relativewith affective disorder compared to 48% in the controlgroup.

In an attempt to further validate the bulimia-restric-tion distinction in anorexia nervosa, Strober et al ana-lyzed MMPI scores of parents of bulimics and restricters[27]. Their data suggested that the severity of bulimia inanorexia nervosa is associated with more pronouncedaffective disturbance in both parents along with greaterimpulsivity in fathers. The morbid risk of affective dis-order in first- and second-degree relatives of the com-bined group was computed to be 15% which is more

348 The Eating Disorders

than two times the average expected lifetime risk for af-fective disorder in the general population. Seventy-onepercent of the bulimic group versus 40% of the re-stricter group had positive family histories. Affective dis-order was almost four times greater in the mothers ofthe bulimic group and was more prevalent in the fathersof the bulimics, but not at a statistically significant level.Alcoholism was shown to be present in 83% of the fami-lies of bulimics and 49% of the restricters, which mayalso point to a linkage with affective disorders.

Hudson and his group evaluated ten normal-weightnonanorexic bulimic patients for familial affective dis-order [28]. Their results were similar to other investiga-tors and revealed a risk factor for major depression ofabout 22% in first-degree relatives of the bulimicpatients. In another study, Hudson et al evaluated 420first-degree relatives of 14 patients with anorexia ner-vosa, 55 patients with bulimia, and 20 patients with bothdisorders [29]. The found the prevalence of familial af-fective disorder to be significantly greater in patientswith anorexia nervosa and/or bulimia than in patientswith bipolar disorder. The morbid risk of affective dis-orders in relatives of the total eating disorder group was27%.

In a prospective study of 40 anorexic patients,Rivinus et al [30] found significantly more depression inthe relatives of the patient group compared with a con-trol group. When substance abuse diagnoses wereadded to depression, the familial impact of impairmentwas even greater. An increase in multigenerational im-pairment and increased frequency of affected relativeswithin families was also observed in the patient group.

Gershon et al [31] attempted to ascertain the pre-sence of an identifiable subgroup of anorexic patientswith a higher risk of affective illness in their relatives asa way of further evaluating the data that suggest ano-rexia nervosa and major affective disorder may shareetiological factors. The presence of affective disorder,self-induced vomiting, or bulimia in the patient did notpredict affective illness in the relatives, which suggestedthe absence of clinically observable genetic heterogene-ity with anorexia nervosa defined by these features. Thedata seemed to support the idea of a shared geneticvulnerability between anorexia nervosa and affectivedisorders.

The idea that familially mediated variables playa cru-cial role in the pathogenesis of the eating disorders wasfurther advanced by a recent report by Strober et al [32].They found a five-fold increased risk of eating disordersin the female relatives of anorexics compared with rela-tives of controls. Subclinical anorexia nervosa was themost frequent disorder observed in the relatives of an-orexics, with a greater total transmission of eating dis-

orders occurring in the families of bulimic anorexics. In-terestingly, all diagnoses of severe restrictive anorexianervosa were made in relatives of restricter probands,whereas the preponderance of bulimic diagnoses weremade in the relatives of anorexic/bulimic probands-suggesting familial segregation of the disorders. Theauthors offered several hypotheses for the familialtransmission of which the cotransmission of an affectivedisorder or other psychopathology may be one.However, they caution that the data do not support anyconclusions.

A recent study by Stern et al questions the relation-ship of familial affective disorder to normal-weightbulimia [33]. Their findings, which used strict diagnos-tic criteria, direct family interview, and blind raters, wereat odds with the above. They only found a 9% preva-lence of affective disorder in the relatives of bulimic pro-bands, compared with 10% in the relatives of controls.However, a history of affective disorder was considera-bly more common among the bulimic patients, but it didnot reach statistical significance and their study waslimited to normal-weight bulimia.

Again, the differing methodologies may account forthe variance in results, ie, patient report, family ques-tionnaire, and direct interview. However, the trend ofmost of the data is consistently in the same direction ex-cept for the last study cited—namely, that there doesseem to be some connection between eating disordersand affective illness.

BIOLOGICAL STUDIES

For convenience, under this heading are subsumedthe neuroanatomical, neurophysiologic, neuroen-docrine, neurochemical, and psychopharmacologicdata that have been forthcoming in relating the eatingdisorders to the affective disorders.

That such relationships should exist biologicallyshould not be very surprising when one considers theoverlap of systems involved in the maintenance of eat-ing and mood. The hypothalamus has been implicatedas being an important structure in the regulation of bothappetite and mood. Opioid peptides, neuropeptides,and monoamines are increasingly being suggested asimportant neuroregulators of both systems [34,35].

Catecholamine metabolism has been thought to beintegral to the understanding of affective illness forsome time. Though the limitations of this hypothesishave become more apparent, the monoamines are stillthought to be important neuroregulators even if notsolely responsible for affective illness [36-38]. Halmi etal studied MHPG during treatment in a group of 25 an-orexics with secondary depressive symptoms [39].

The Eating & Affective Disorders 349

MHPG is a major metabolite of brain norephinphrineand may correlate with central noradrenergic activity.An increase in MHPG correlated with a decrease in de-pression on two of three measures of depression.Gerner and Gwirtsman found that 24-hour urinaryMHPG was low for all 11 female anorexic patientstested compared with control subjects [40]. This findingappeared to be independent of depression.

Abraham et al measured urinary MHPG in seven an-orexics during a bed rest and refeeding program andfound a positive relationship between MHPG and bodyweight, but no correlation between body weight and de-pression [41]. They interpreted their findings as reflect-ing a general metabolic effect. In an attempt to resolvesome of the confusing findings concerning the signifi-cance of low urinary MHPG levels in anorexia nervosa,Biederman et al studied a group of anorexia nervosapatients before and after five weeks of treatment and agroup of matched controls [42]. A subgroup of thepatients who met RDC criteria for a concomitant majordepressive disorder was found to have lower pre- andposttreatment mean urinary MHPG levels than bothnormal controls and nondepressed anorexia nervosapatients, both of whom had similar values. The medianvalue of all urinary MHPG samples was used as a cut-off point to assess the distribution of values. They re-ported that significantly more depressed patients ex-creted low MHPG compared with nondepressedpatients and normal controls.

Kaye et al studied CNS dopamine and serotonin me-tabolites and norepinephrine in underweight anorectics,recently weight-recovered anorectics, and long-termweight-recovered anorectics [43]. They found that thedopamine metabolite homovanillic acid (HVA) and theserotonin metabolite 5-hydroxyindolacetic acid (5-HIAA) were reduced in underweight anorectics but re-turned to normal shortly after weight was normalized.Norepinephrine levels, on the other hand, were similarto normals in underweight anorectics and after weightrestoration. However, long-term weight-recovered an-orectics had a 50% decrease in CSF norepinephrinelevels compared with controls. The investigators feltthat these changes in neurotransmitter metabolism arean integral part of the neurobiology of anorexia nervosaand may influence the changes in mood, behavior, andneuroendocrine function that have been observed. Alongstanding disturbance in CNS norepinephrine maycertainly point to a coexisting affective disturbance.

Monoamine oxidase (MAO) is an important enzymein the degradation of many biogenic amines and thushas been the subject of studies attempting to link bio-chemical disturbances with a variety of psychiatric ill-ness. The data generated from these studies appear in-conclusive. Biederman et al studied platelet MAO ac-

tivity prospectively in a group of young anorexia nervosapatients with and without an RDC major depressive dis-order along with a matched control group [44]. Theyfound that the depressed anorexics had lower MAO ac-tivity compared with nondepressed anorexics and con-trols. This finding led them to conclude that MAO ac-tivity may be useful in identifying a depressed subtypeof anorexia nervosa that may be etiologically related toother low-platelet MAO conditions such as bipolar af-fective disorder and alcoholism.

Over the last 15 years, there has been great activity inthe exploration of the neuroendocrine basis of depres-sion [45,46]. The development of the dexamethasonesuppression test (DST) for depression was enthusiasti-cally greeted initially as a long-awaited biological markerof psychiatric illness [47,48]. This neuroendocrine lineof inquiry has been naturally extended to the eating dis-orders as a way of establishing a connection to the affec-tive disorders.

As part of the study cited above, Gerner and Gwirts-man studied 21 female primary anorexics and one maleanorexic [40]. The 21 women demonstrated a uniformand marked abnormal DST. Interestingly, the male an-orexic had a normal DST. No correlation was observedbetween depression and abnormal DST leading the in-vestigators to conclude that the abnormality was a pro-duct of primary anorexia nervosa but not solely ofweight loss, because the patients were not starving. Inaddition, three drug-free schizophrenic patients whohad starved themselves to 75% of their ideal weight hadnormal DST. The findings of low MHPG and abnormalDST led the authors to the speculation that low hy-pothalamic norepinephrine, via its tonic inhibition ofcorticotrophin-releasing factor, might be responsible forthe abnormal DST and point to an adrenergic vulnera-bility in anorexic patients, which might explain the strongfamilial association of affective illness and high degreeof risk for depression in recovered anorexics.

Beside nonsuppression of cortisol by dexametha-sone, there are other measures of adrenocortical activ-ity that are disturbed in anorexia nervosa. Walsh et alstudied a group of anorexic women and reported an in-creased absolute cortisol production rate adjusted forbody size and weight, increased plasma cortisol concen-tration, and increased urinary free cortisol [49]. All ofthese findings were statistically significant when com-pared with controls matched for age and sex. The abovefindings, while similar to those observed in malnutrition,are different on several measures of adrenocorticalfunction. Walsh speculated that the increased adrenalactivity in anorexia nervosa is out of proportion to themalnutrition and might reflect response to any numberof stresses—depressive illness being one.

Given the fact that the malnourished state in anorexia

350 The Eating Disorders

nervosa confounds much of the data generated, the nor-mal-weight bulimic population became a promisinggroup for study. Hudson et al reported abnormal DSTresults in five of nine normal-weight bulimics without ahistory of anorexia nervosa [28]. This percentage com-pared favorably with the results obtained by Carroll andcoworkers on patients with major depression [46]. In anexpanded study, Hudson and coworkers reported on aslightly more mixed group of normal weight bulimics,some of whom were on psychotropic medication, andfound 47% nonsuppression on the DST [50].

More evidence is provided by Gwirtsman et al, whoreported that 67% of 18 normal-weight bulimic patientshad abnormal DST [51]. These findings lead theauthors to adjust their previous conclusion that neu-roendocrine abnormalities previously found in anorex-ics were solely an artifact of low weight. They now sug-gested that weight may only be a partial determinant ofcortisol nonsuppression. Though the neuroendocrineabnormalities observed were not associated with diag-nosable major depressive disorder, a diagnosis of dys-thymic disorder could be made in 50% of their patients,with over half of the dysthymic patients also carrying aborderline personality diagnosis. Thus, the neuroen-docrine abnormalities observed could be seen as a re-sult of disorders in the affective and borderline spec-trum rather than purely as a result of the bulimia itself.

Gadpaille et al studied amenorrheic and menstruat-ing runners and found that the amenorrheic runnershad an overwhelming representation of affective dis-orders, eating disorders and affectively ill relatives whencompared to menstruating runners. Their findings raisethe possibility of a shared neuroendocrine vulnerabilityin the eating and affective disorders [82].

In a study of 29 normal-weight bulimic women,Blinder et al found that there was a significantly greaterincidence of prior history of anorexia nervosa and cur-rent clinical diagnosis of depression among the DSTpositive subgroup [52]. This data suggested to them thatappetite disturbance may be closely linked to affectivedisturbance in certain patient subgroups and that per-haps the ingestive dysfunction may be a phasic manife-station of an underlying affective disorder.

The response of thyroid stimulating hormone (TSH)to thyroid releasing hormone (TRH) is thought to beblunted in depression, with generally normal circulatingthyroxin (T4) and triiodothyronine (T3). In anorexianervosa, the TSH response to TRH is normal, althoughthe peak response is most often delayed [53]. Gwirts-man et al found eight of ten normal weight bulimicpatients to have blunted TRH tests [51]. Unlike the sit-uation in depression, T3 is significantly decreased in an-orexia nervosa, which is probably secondary to a

diminished rate of conversion of T4 to T3, which alsomay be partially responsible for the decreased break-down of cortisol. Also, anorexics have a lower mean T4than normals [54,55]. Whybrow and Prange have sug-gested that thyroid hormones may be important in reg-ulating noradrenergic receptors in the brain and as a re-sult may influence the development of and recoveryfrom affective illness [56].

Activation is a fundamental neurophysiologically de-termined state of living organisms. In the affective dis-orders this state can be both underactive and overactive.It has been known for some time that hyperactivity is aclinical feature of anorexia nervosa [57]. This observa-tion has been used to design an operant conditioningparadigm for the treatment of this condition [58].Winokur et al were impressed by the presence of racingthoughts in their patients and felt that this suggestedsimilarity to bipolar affective disorder, particularly whenconsidered along with a history of distinct episodes ofboth depression and hypomania [26]. Mills and Med-licott [59] in exploring the theoretical basis of naloxonetreatment in anorexia nervosa, make some extremelyinteresting points relating arousal, perfectionism, com-pulsivity, and anorexia. The mechanism they posit is thatanorexia enhances arousal, which serves perfectionismand compulsivity. In their view, the depression observedin anorexia is related to the exhaustion produced fromfighting compulsion.

Crisp and Stonehill pointed out in a rather compli-cated study that there is a relationship between weightloss, reduced sleep time, sleep discontinuity on one handand weight gain, longer sleep time, sleep continuity andlater waking on the other [60]. These changes appearedto transcend psychiatric diagnosis and to be primarily re-lated to nutritional factors. The findings are similar tothose uncovered in relationship to anorexia nervosa andsuggest a common hypothalamic disruption in anorexianervosa and other psychiatric illness. Though theauthors did not observe a relationship between earlymorning awakening and depression, this is still felt to bea cardinal sign of endogenous depression.

Rapid eye movement (REM) latency has beenthought to be a psychobiologic marker for primary de-pression [61]. Katz et al studied a mixed group of eat-ing disorder patients with all-night sleep recordings andfound a significantly reduced REM latency in the eat-ing disorders group, which was approximately midwaybetween previously established values for normal con-trols and primary depressives [62]. When the eating dis-orders group was divided into short versus long REMlatency subgroups, Hamilton scores for depression weresignificantly higher in the former group; bulimia was apart of the clinical picture in the short latency group and

The Eating & Affective Disorders 351

DSM-III criteria for major depression were met by allthe short REM latency patients. Mean urinary cortisolwas higher in the short latency REM group, but becauseof a wide distribution of values, this finding did not reachstatistical significance.

The application of psychopharmacology to the eatingdisorders became a logical step as the affective com-ponents of the eating disorders began to be betterappreciated. As of this writing, almost every class of psy-chotropic agent has been used in the eating disorders.Many of the studies of psychotropic agents in the treat-ment of eating disorders are of an anecdotal or open ex-perimental structure, which mitigates the positive ef-fects reported. However, the experience of the investi-gators is important from a clinical empirical standpoint.

L-dopa was utilized in a group of six anorexics be-cause of what appeared to be clinical similarity topatients with Parkinson's Disease [63]. Four of thepatients were noted to be improved with this medica-tion. Amitriptyline was reported to promote weightgain, decrease weight focus and improve mood in sixyoung anorexics admitted to a pediatric ward [64]. Theeffects of the amitriptyline were noted between the sixthto twelfth day, and weight change was felt to bepreceeded by a brightening of mood and improved so-ciability. The authors questioned the representative na-ture of their sample, as it was a young sample with aneight-month average length of illness—all factors whichauger well for a good prognosis. The also questionedother treatment factors that might have been operative,because the response to the tricyclic medication wasquite rapid.

In one of the few double-blind studies, Pope et altreated 11 chronically bulimic women with imiprimineand assigned a similar group of 11 patients to placebo[65]. Eventually 19 patients completed the six-weekstudy with a 70% decrease in bingeing observed in theimipramine treated group and virtually no change in theplacebo group. After the six-week blind study periodwas completed, patients in the placebo group wereoffered antidepressant treatment and changes weremade in antidepressant medications in treated patientswho were not responding. Eventually 20 of 22 patientsreceived a complete course of therapy with at least oneantidepressant. On follow-up of one to eight months,90% of treated patients reported a moderate or markeddecrease in binge eating. Also, decrease in HamiltonDepression Scale scores and reduction in binge eatingwere significantly correlated, suggesting a connectionbetween antibulimic and antidepressant drug effects.Mitchell and Groat in a double-blind controlled studyusing amitriptyline observed similar results [66].

Brotman et al reported on an uncontrolled retrospec-tive study of 22 bulimics treated with at least one ther-apeutic trial of antidepressants [67]. Of 17 depressed

bulimics, 10 achieved a remission in depressive symp-toms. Only 4 of these 17 depressed bulimics maintainedboth a decrease in binge eating and remission of depres-sion at follow-up. The patients who decreased theirbingeing after the drug trial were less often depressedthan those patients whose bingeing did not decrease.This study suggests that antidepressant medication mayhave independent antibinge and antidepressant effects,which would imply a less direct linkage between bulimiaand affective illness. However, the fact that four of fivetrue responders had an underlying depression that wassuccessfully treated argues for a more intimate relation-ship.

In a recent controlled double-blind crossover study,Hughes et al found desipramine to be of value in thetreatment of nondepressed bulimics. Sixty-eight percentof their patients were eventually able to completely elim-inate their bulimic behaviors within the ten weeks oftreatment. Whether this effect could be sustained overa longer time frame is open to question. The desi-pramine blood levels did not seem to completely predictresponse but seemed to indicate a group of patients inwhich response could be enhanced by more aggressivetreatment and more rational management of side-ef-fects and compliance. A prior history of anorexia ner-vosa did not predict response to desipramine, which ledthe authors to regard the distinction between anorexianervosa and bulimia with less certainty. This study, whileseeming to make a strong statement concerning the in-dependence of antibulimic antidepressant drug effectsand perhaps the independence of bulimia and affectiveillness, is confounded by the fact that over 50% of thepatients had a history consistent with a major affectivedisorder. In addition, the Zung scores (though not in thedepressed range) improved significantly in the desi-pramine-treated group, pointing to a more general ef-fect [68].

The follow-up report of Pope et al based on thepatients cited above sheds some more light on this issue.They reported 95% partial improvement in bulimia and50% complete remission after two years. The improve-ment was noted not just on measures of bulimia but ina marked reduction in depressive symptoms as well.This suggested to them that antidepressants provide anoverall therapeutic effect for bulimmic patients [69].

A double blind study of pheytoin versus placebo inbulimia produced moderate to marked improvement inonly 42% of treated patients [70]. However, this was sig-nificant compared with placebo. On follow-up of fourphenytoin responders, two were observed to relapseafter two months while continuing the medication.

Of related interest is a recent case report of thesuccessful usage of another anticonvulsant, car-bamazepine, in bulimia [71]. This agent is finding in-

352 The Eating Disorders

creased applicability in the treatment of bipolar affec-tive disorder. The one bulimic patient who responded,out of the six treated, had a history of brief dramaticmood swings. The observation of mood instability in theeating disorder population has also led others to employlithium [72,73]. Hsu recently reported on 14 normalweight bulimics who were treated in an uncontrolledfashion with lithium and a cognitive behavioral psycho-therapeutic approach [74]. Twelve of these patients im-proved markedly or moderately, including four of sixpatients who relapsed or failed to improve with behaviortherapy alone. Hsu raises the possibility that lithiumachieved its effectiveness as an antidepressant acting ona variant of affective illness perhaps by dampening thedysphoria that precipitates a binge. Indeed, over 70%of his patients had elevated Beck inventories. Hsu wasclinically impressed by the mood swings and emotionalinstability of his patients and suggests that lithium mayhave provided a general calming effect in emotionallylabile patients.

Atypical depression is a classification that has beenoffered to describe a group of depressives who have areactive mood and reversal of some of the typical neu-rovegetative signs of depression. This group seems torespond more effectively to monoamine oxidase inhib-itors (MAOI) [75]. The overeating of bulimics and theirmood disturbance suggested to investigators that theymight indeed also respond to MAOIs. Walsh and co-workers studied six patients who met both criteria forbulimia and atypical depression [76]. All of the patientswere reported to respond dramatically to MAOIs. Theusefulness of MAOIs in bulimia was further reinforcedby Walsh et al in a double-blind controlled study ofphenelzine with significant positive effect reported [77].

Pope et al reported on their extensive experience intreating 65 consecutive bulimics with a variety of antide-pressants and found promising results with tricyclics,trazadone, and MAOIs, which produced by far the bestresponse [78]. The structure of this study in using tricy-clics as the initial drug probably led to the underestima-tion of the effectiveness of the nontricyclic antidepres-sants. The authors felt that they could safely concludethat antidepressant medications have a clear place in thetreatment of bulimia.

CONCLUSIONS

Certainly the preponderance of data would allow oneto conclude that the eating disorders carry with them, atthe minimum, an increased risk for developing an affec-tive disorder. The difference in risk between pure re-strictive anorexia and bulimia and the heavy loading foraffective illness in the families of bulimics point to the

potential importance of biologic endowment in this dis-order. Still, the nature of the relationship between theeating disorders and affective disorders remains to beclarified.

Hudson et al feel that their data suggest that bulimiamay indeed be a forme fruste of an affective disorder[28]. In more recent work, they studied a group of ac-tive and remitted bulimics and found high lifetime ratesof major affective disorders as well as anxiety and sub-stance abuse disorders. They interpreted their results asproviding further support for the existence of a phe-nomenologic relationship between bulimia and majoraffective disorder [83]. Strober, in reviewing his data,speculated that some adolescent cases of bulimia ner-vosa might represent a "phenotypically unique phase-bound manifestation of the affective disorder spectra"or that bulimia nervosa is the product of and final com-mon pathway for different but overlapping etiologic fac-tors of which affective biologic endowment is but one[20]. Gerner and Gwirtsman leaned toward an adrener-gic vulnerability model to explain the strong familial as-sociation of affective dysfunction in anorexic patients[40].

The successful treatment of some eating disorderswith antidepressant agents have led some to concludethat the eating disorders are manifestations of affectiveillness, but this view may be somewhat overzealous.Walsh et al observed the close variation of bulimicsymptoms and depression but backed away from con-cluding that they are one and the same [75]. Glassmanand Walsh caution against equating bulimia with an af-fective disorder on the basis of response to antidepres-sants and make the point that everything that improveswith an antidepressant is not necessarily depression[79]. They suggest that both conditions may arise froma unitary biological matrix with environmnental and in-trapsychic factors determining the final outcome.

In a recent commentary, Altshuler and Weinerstrongly question the relationship between anorexianervosa and depression and raise some troublesome is-sues concerning reliability of diagnosis, researchmethodology, and epidemiology [80]. However, they donot distinguish between restrictive anorexia nervosa,bulimia nervosa, and normal-weight bulimia. Also theirdiagnostic focus was on major depression rather than abroader notion of affective impairment. Both of thesefactors could certainly contribute to underestimatingthe potential for linkage between the eating disordersand affective disorders.

Swift et al have concluded that affective disordersand eating disorders are related but in an unclear way.They have proposed an interactive, multideterminedmodel for understanding this complex relationship [84].

The Eating & Affective Disorders 353

Hinz and Williamson argue that the data does not yetjustify the conclusion that bulimia is a variant of an af-fective disorder. They view bulimia as a chronic psychi-atric disorder which, like other chronic disorders, maybe accompanied by depression [85].

Even if eating disorders are not manifestations of af-fective disorders, the high incidence of affective impair-ment in eating disorder patients must be explained. Theissues are sufficiently complex that a multi-axial biopsy-chosocial approach is necessary for perspective. Garneret al have studied the question of the continuity of ano-rexia nervosa and have concluded that there are quali-tative differences between true anorexia nervosa andweight preoccupation in supposed normals [81].Despite a culturally reinforced "drive for thinness,"many young women flirt with anorexia nervosa andbulimia without succumbing. This leads one to specu-late about predisposing factors. Certainly episodic orchronic affective dysregulation could provide a fertileground for the growth of maladaptive culturally rein-forced coping strategies, as may other factors. The af-fective dysregulation serves to deprive the individual ofthe emotional stability necessary to surrender maladap-tive behaviors and establish less self-destructivedefenses.

That there should be greater representation of affec-tively disordered individuals among bulimics than re-stricting anorexics should not be all that surprising whenone considers the instability created by mood swings andpoor impulse control. The bulimia itself represents theshifting internal environment that may be attempting tocope with pubertal body changes, individuation, and au-tonomy. In addition, the bulimic behaviors are suffi-ciently complex that they can become self-perpetuatingand addicting, particularly when they are seen as provid-ing mood change. Restrictive anorexia, on the otherhand, would seem to require different constitutionalfactors, parenting experiences, and ego defenses.

The clinical encounter requires continuous reformu-lation as new information is uncovered and new impres-sions registered. In the treatment of patients with eatingdisorders, this process is magnified because of the manyproblems that present simultaneously. Addressing thebiological underpinnings is just one factor in what mustbe a comprehensive rehabilitative effort. However, it isimportant to appreciate how disorganizing an affectivedisorder can be-particularly in a young individual whois attempting to consolidate a sense of self. The success-ful treatment of the underlying affective disorder allowsnew learning to proceed.

In conclusion, it would seem reasonable for a prudentclinician to maintain a high index of suspicion for a con-current affective disorder when evaluating or treating apatient with one of the eating disorders, particularly in

the context of mood instability, substance abuse, or afamily history of affective dysfunction.

REFERENCES

Freud S. On the origins of psychoanalysis: Letters toWilhelm Fleiss. New York: Basic Books, 1954.Abraham K. The first pregenital stage of the libido(1916). Selected Papers of Karl Abraham. New York:Basic Books, 1954: 262-85.Fenichel 0. The psychoanalytic theory of neurosis.New York: WW Norton & Co, 1945.Binswanger L The case of Ellen West. In: May R,Angel E, Ellenberger H, eds. Existence. New York:Basic Books, 1958.Bruch H. Eating disorders: obesity, anorexia nervosaand the person within. New York: Basic Books, 1973.Casper RC, Eckert ED, Halmi KA, et al. Bulimia: itsincidence and clinical importance in patients with an-orexia nervosa. Arch Gen Psychiatry 1980; 37:1030-5.Garfinkel PA, Moldofsky H, Garner DM. The hetero-geneity of anorexia nervosa: bulimia as a distinct sub-group. Arch Gen Psychiatry 1980; 37:1037-40.Russell G. Bulimia nervosa: an ominous variant of an-orexia nervosa. Psychol Med 1979; 9:429-48.Vandereycken W, Pierloot R. The significance of sub-classification in anorexia nervosa: a comparative studyof crucial features in 141 patients. Psychol Med 1983;13:543-9.Akiskal HS. Dysthymic disorder: psychopathology ofproposed chronic depressive subtypes. Am J Psychi-atry 1983; 140:11-20.Keller MB, Shapiro RW. "Double Depression": super-imposition of acute depressive episodes on chronic de-pressive disorders. Am J Psychiatry 1982; 139:438-42.Carroll BJ, Greden JF, Feinberg MD, et al. Neuroen-docrine evaluation of depression in borderlinepatients. Psychiatr Clin North Am 1981; 4(1):88-99.Cantwell DP, Sturzenberger S, Burrought J, et al. An-orexia nervosa-an affective disorder? Arch Gen Psy-chiatry 1977; 34:1087-93.Dally P. Anorexia nervosa. London: William HeilemanMedical Books, 1969.Theander S. Anorexia nervosa. Acta Psychiatr Scand1970; Suppl 213.Morgan H, Russell G. Value of family background andclinical features as predictors of long-term outcome inanorexia nervosa: four year follow-up study of 41patients. Psychol Med 1975; 5:355-71.Stonehill E, Crisp AH. Psychoneurotic characteristicsof patients with anorexia nervosa before and aftertreatment and at follow-up 4-7 years later. J Psychoso-matic Res 1977; 21:127-93.Ben-Tovin DI, Maricov V, Crisp AH. Personality andmental state within anorexia nervosa. J Psychosom Res1979; 23:321-5.

19 Eckert ED, Goldberg SC, Halmi KA, et al. Depressionin anorexia nervosa. Psychol Med 1981; 12:115-22.

354

The Eating Disorders

Strober M. The significance of bulimia in juvenile an-orexia nervosa: an exploration of possible etiologic fac-tors. Int J Eating Disorders 1981; 1:28-43.Pyle RL, Mitchell JE, Eckert ED. Bulimia: a report of34 cases. J Clin Psychiatry 1981; 42:60-4.Hendren RL. Depression in anorexia nervosa. J AmerAcad Child Psychiatry 1983; 22:59-62.Weissman MM, Gershon ES, Kidd KK, et al. Psychi-atric disorders in the relatives of probands with affec-tive disorders. Arch Gen Psychiatry 1984; 41:13-21.

clunan JF, Caruso KA, Prosoff BA, et al. Appetitedisturbance and excessive guilt in major depression:the use of family study data to define depressive sub-types. Arch Gen Psychiatry 1984; 41:839-44.Crisp AM, Harding B, McGuinness B. Anorexia ner-vosa. Psychoneurotic characteristics of patients: rela-tionship to prognosis. J Psychosom Res 1974; 18:167-73.Winokur A, March V, Mendels J. Primary affectivedisorder in relatives of patients with anorexia nervosa.Am J Psychiatry 1980; 137:695-8.Strober M, Salkin B, Burroughts J, et al. Validity of thebulimia-restricter distinction in anorexia nervosa-parental personality characteristics and family psychi-atric morbidity. J New Ment Dis 1982; 170:345-51.Hudson JI, Laffer PS, Pope Jr HG. Bulimia related toaffective disorder by family history and response to thedexamethasone suppression test. Am J Psychiatry1982; 139:685-7.Hudson JI, Pope HG, Jonas JM, et al. Family historystudy of anorexia nervosa and bulimia. Brit J Psychiat1983; 142:133-8.Rivinus TM, Biederman J, Herzog DB, et al. Anorexianervosa and affective disorders: a controlled familyhistory study. Am J Psychiatry 1984; 141:1414-8.Gershon ES, Schreiber JL, Hamovit JR, et al. Clinicalfindings in patients with anorexia nervosa and affec-tive illness in their relatives. Am J Psychiatry 1984;141:1419-22.Strober M, Morrell W, Burroughts J, et al. A control-led family study of anorexia nervosa. J Psychiat Res1985; 19:239-46.Stern SL, Dixon KN, Nemzer E, et al. Affective dis-order in the families of women with normal weightbulimia. Am J Psychiatry 1984; 141:1224-7.MorleyJE, Levine AS. The central control of appetite.Lancet: Feb 19, 1983; 398-401.Kaye WH, Pickar D, Naber D, et al. Cerebrospinalfluid opioid activity in anorexia nervosa. Am J Psychi-atry 1982; 139:643-5.Bunney WE, Garland BL. A reevaluation of the cate-cholamine hypothesis in affective disorders. In: UsdinE, Carlson A, Dahlstrom A, Engel J, eds. Cate-cholamines: neuropharmacology and the central nerv-ous system-therapeutic aspects. New York: Alan RLiss, 1984.

37. Bunney WE, Garland BL. Selected aspects of amineand receptor hypothesis of affective illness. J Clin Psy-chopharmacol 1981; 1:3S-11S.

Siever IJ, Davis KL. Overview: toward a dysregulationhypothesis of depression. Am J Psychiatry 1985;142:1017-31.Halmi KA, Dekirmenjian H, Davis JM, et al. Cate-cholamine metabolism in anorexia nervosa. Arch GenPsychiatry 1978; 35:458-60.Gerner RH, Gwirtsman HE. Abnormalities of dex-amethasone suppression test and urinary MHPG inanorexia nervosa. Am J Psychiatry 1981; 138:650-3.Abraham SF, Beumone PJV, Cobbin DM. Cate-cholamine metabolism and body weight in anorexianervosa. Br J Psychiatry 1981; 138:244-7.Biederman J, Rivinus T, Herzog D. Urinary MHPG inanorexia nervosa patients. In: Abstracts of the 137thAnnual Meeting of the American Psychiatric Associa-tion. Los Angeles, 1984.Kaye WM, Ebert JH, Raleigh M, et al. Abnormalitiesin CNS monoamine metabolism in anorexia nervosa.Arch Gen Psychiatry 1984; 41:350-5.Biederman J, Rivinus TM, Herzog DB, et al. PlateletMAO activity in anorexia nervosa patients with andwithout a major depressive disorder. Am J Psychiatry1984; 141:1244-7.Sachar EJ, Hellman L, Fukushima D, et al. Cortisolproduction in depressive illness. Arch Gen Psychiatry1970; 23:289-98.Carroll BJ, Curtis GC, Mendels J. Neuroendocrineregulation in depression, II: discrimination ofdepressed from nondepressed patients. Arch Gen Psy-chiatry 1976; 33:1051-7.Asnis GM, Sachar El, Halbreich V, et al. Cortisolsecretion and dexamethasone response in depression.Am J Psychiatry 1981; 138:1218-21.Carroll BJ, Feinberg M, Greden JF. A specific labora-tory test for the diagnosis of melancholia: stand-ardization, validation and clinical utility. Arch GenPsychiatry 1981; 38:15-22.Walsh BT, Katz JL, Levin J, et al. Adrenal activity inanorexia nervosa. Psychosom Med 1978; 40:499-506.Hudson JI, Pope HG, Jonas JM, et al. Hypothalamic-pituitary-adrenal-axis hyperactivity in bulimia. Psychi-atry Research 1983; 8:111-7.Gwirtsman HE, Roy-Byrne P, Yager J, et al. Neuroen-docrine abnormalities in bulimia. Am J Psychiatry1983; 140:559-63.Blinder BJ, Chaitin BF, Hagman J. Two diagnosticcorrelates of dexamethasone non suppression in nor-mal weight bulimia. Hillside Journal of Clinical Psychi-atry, 1987; 9:211-216.Loosen PT, Prange AJ. Serum thyrotropin responseto thyrotropin releasing hormone in psychiatricpatients: a review. Am J Psychiatry 1982; 139:405-16.Walsh BT. Endocrine disturbances in anorexia ner-vosa and depression. Psychosom Med 1982; 44:85-01.Moshang T, Utiger RD. Low triiodothyronine euthy-roidison in anorexia nervosa. In: Vigersky RA, ed. An-orexia nervosa. New York: Raven Press, 1977.

56. Whybrow PC, Prange AF. A hypothesis of thyroid-catecholamine-receptor interaction. Arch Gen Psychi-

The Eating & Affective Disorders 355

atry 1981; 38:106-13.Kron L, Katz JL, Gorzynski G, et al. Hyperactivity inanorexia nervosa: a fundamental clinical feature.Compr Psychiatry 1978; 19:433-40.Blinder EJ, Freeman DM, Stunkard Al Behaviortherapy of anorexia nervosa: effectiveness of activity asa reinforcer of weight gain. Am J Psychiatry 1970;126:1093-8.Mills IH, Medlicott L. The basis of naloxone treatmentin anorexia nervosa and the metabolic responses to it.In: Pirke KM, Ploog D, eds. The psycholbiology of an-orexia nervosa. Berlin: Springer-Verlag, 1984.Crisp All, Stonehill E. Aspects of the relationship be-tween sleep and nutrition: a study of 375 psychiatricoutpatients. Brit J Psychiat 1973; 122:379-94.Kupfer DJ. REM latency: A psychobiological markerfor primary depression. Biol Psychiatry 1976; 11:154-74.Katz JL. Kuperberg A, Pollack CP, et al. Is there a re-lationship between eating disorder and affective dis-order? New evidence from sleep recordings. Am J Psy-chiatry 1984; 141:7653-9.Johansen AJ, Knorr NJ. Treatment of anorexia ner-vosa by levodopa. Lancet 1974; 1:591.Needleman HL, Waber D. The use of amitriptyline inanorexia nervosa. In: Vigersky RA, ed. Anorexia ner-vosa. New York: Raven Press, 1977.Pope HG, Hudson JI, Jonas JM, et al. Bulimia treatedwith imipramine a placebo-controlled, double-blindstudy. Am J Psychiatry 1983; 140:554-8.Mitchell JE, Groat R. A placebo-controlled, double-blind trial of amitriptyline in bulimia. J Clin Psycho-pharmacol 1984; 4:186-93.Brotman AW, Herzog DV, Woods SW. Antidepres-sant treatment of bulimia: the relationship betweenbingeing and depressive symptomatology. J Clin Psy-chiatry 1984; 45:7-9.Hughes PL, Wells LA, Cunningham CJ, et al. Treatingof bulimia with desipramine. Arch Gen Psychiatry1986; 43:182-6.Pope HG, Hudson JI, Jonas JM, et al. Antidepressanttreatment of bulimia: a two-year follow-up study. J ClinPsychopharmacol 1985; 5:320-7.Wermuth BM, Davis KL, Hollister LE, et al. Pheny-toin treatment of the binge-eating syndrome. Am JPsychiatry 1977; 134:1249-53.Kaplan AS, Garfinkel PD, Darby PL, et al. Car-bamazepine in the treatment of bulimia. Am J Psychi-atry 1983; 140:1225-6.Barcai A. Lithium in anorexia nervosa: a prior reporton two patients. Acta Psychiatr Scand 1977; 55:97-101.Stein GS, Hartshorn S, Jones J, et al. Lithium in a caseof severe anorexia nervosa. Brit J Psychiatry 1982;140:526-8.Hsu LK. Treatment of bulimia with lithium. Am J Psy-chiatry 1984; 141:1260-2.

75. Liebowitz MR, Quitkin F, Stewart JW, et al. Phenel-zine and imipramine in atypical depression. Psycho-pharmacol Bull 1981; 17:159-61.

Walsh BT, Steward JW, Wright L, et al. Treatment ofbulimia with monoamine oxidase inhibitors. Am J Psy-chiatry 1982; 139:1629-30.Walsh BT, Stewart JW, Roose SP, et al. Treatment ofbulimia with phenelzine: a double-blind, placebo-con-trolled study. Arch Gen Psychiatry 1984; 41:1105-9.Pope HG, Hudson JI, Jonas JM. Antidepressant treat-ment of bulimia: preliminary experience and practicalrecommendations. J Clin Psychopharmacol 1983;3:274-81.Glassman A, Walsh BT. Link between bulimia and de-pression (letter). J Clin Psychopharm 1983; 3:203.Altschuler KZ, Weiner MF. Anorexia nervosa and de-pression: a dissenting view. Am J Psychiatry 1985;142:328-32.Gamer DM, Olmsted MP, Garfinkel PE. Does ano-rexia nervosa occur on a continuum? Int J Eating Dis-orders 1983; 2:11-20.Gadpaille WJ, Sanborn CF, Wagner Jr WW. Athleticamenorrhea, major affective disorders and eating dis-orders. Am J Psychiatry 1987; 144:939-942.Hudson JI, Pope Jr HG, Yugelun-Todd et al. A con-trolled study of lifetime prevalence of affective andother psychiatric disorders in bulimic outpatients. AmJ Psychiatry 1987; 144:1283-1287.Swift WJ, Andrews D, Barklage NF. The relationshipbetween affective disorder and eating disorders: a re-view of the literature. Am J Psychaitry 1986; 143:290-299.Hinz DL, Williamson DA. Bulimia and depression: areview of the affective variant hypothesis. Psychol Bul-letin 1987; 102:150-158.