THE AMERICAN JOURNAL OF PSYCHIATRY

The Interface Between Borderline

Personality Disorder and Affective Disorder

John G. Gunderson, M.D., and Glen R. Elliott, Ph.D., M.D.

traits. Thus, the impulsive drug use and sexual promis-

The authors review the available literature on the cuity frequently found in borderline patients are seen
interface between borderline personality disorder and as efforts to alleviate chronic depression, and suicide
affective disorder. Three competing hypotheses have attempts are viewed as indications of despair. Like-
been offered to explain the substantial overlap wise, unstable relationships might be secondary to the
between these diagnostic categories; they postulate poor self-esteem and morbid preoccupations arising
that borderline disorder arises from affective from the primary disorder. Even the poor reality
disorder, that affective disorder arises from testing and thought disorder present in some border-
borderline disorder, or that the two are independent line patients can be attributed to comparable distur-
and overlap coincidentally. None of these hypotheses bances that occur in patients who have affective disor-
satisfactorily explains the existing data. The authors ders.
propose a fourth hypothesis focusing on the multiple Hypothesis II posits that borderline personality dis-
etiologies of the signs and symptoms used to order itself can produce diagnosable affective disorders
diagnose both affective and borderline disorders and in some individuals (4-8). In this view, the primary
suggesting that some patients in the resulting defect manifests itself as impulsivity, characteristic
heterogeneous population have symptom clusters attempts to gain control of and support from impor-
that fit both syndromes. tant relationships, hypersensitivity to separation and
(Am J Psychiatry 142:277-288, 1985) loss, and a tendency when stressed to exhibit impaired
reality testing. Given these difficulties, such individuals

T he upsurge
disorder
in studies
and primary
of borderline
affective disorder
personality
has led to
might
op other
hypothesis
secondarily
signs
is stated,
be chronically
and symptoms
such signs and symptoms
dysphoric
of depression.
and devel-
As the
might be
competing hypotheses about possible interrelation- identical to or differ substantially from those in other
ships between these diagnostic groups. In this review types of depression.
we summarize three existing hypotheses about this Hypothesis III postulates that affective and border-
apparent overlap, examine the available literature test- line personality disorders are unrelated, with each
ing some of their predictive implications, and suggest a having a relatively high incidence in the population (9,
fourth hypothesis that appears to be in better accord 10). From this perspective, what appears to be a
with existing data. meaningful correlation is more like one that might be
Hypothesis I proposes that affective disorder, proba- found in a search for tall people with blue eyes: such
bly depression, is the primary problem in patients with people exist, but not because being tall is related to
both affective syndrome and borderline character pa- having blue eyes. This hypothesis is consistent with
thology (1-3). From this viewpoint, a defect in mood DSM-III, which classifies the affective disorders as axis
regulation leads to certain pathological character I and borderline personality disorder as axis II, with no
suggestion of a particular relationship between them.
However, when the two disorders occur in the same
Received Nov. 8, 1983; revised April 20, 1984; accepted May 21, person, it would not be unreasonable to anticipate that
1984. From McLean Hospital and Harvard Medical School. Ad-
they can interact.
dress reprint requests to Dr. Gunderson, McLean Hospital, 1 iS Mill
St., Belmont, MA 02178. To facilitate comparisons among studies, we will try
Copyright © 1985 American Psychiatric Association. throughout this review to identify whether a study

Am] Psychiatry 142:3, March 1985 277

BORDERLINE PERSONALITY DISORDER AND AFFECTIVE DISORDER

used strict or broad criteria for diagnosing affective however, these results probably are biased by their
disorder and borderline personality disorder. In line admission selection process. In summary, estimates of
with most recent research, we use the term “strictly overlap range from 8% to 61%, with a suggestion that
defined affective disorder” for that diagnosed accord- patients who have nonmelancholic unipolar depres-
ing to DSM-III criteria
or Research Diagnostic Criteria sion may have an especially high prevalence of the
(RDC) (11). These two diagnostic systems are quite borderline syndrome. To our knowledge, no study so
similar, but many of the patients classified as having far has assessed the full spectrum of axis II disorders.
schizoaffective illness by RDC are diagnosed as either Also, Hirschfeld et al. (17) indicated that diagnoses of
manic or depressed with psychotic features in DSM- personality disorder are artificially elevated if made
III. Other definitions of affective disorders are noted in when patients are clinically depressed. Efforts by Char-
the text. “Strictly defined borderline personality disor- ney et at. (13) and Kroll et al. (15) to avoid this effect
der” refers to criteria in DSM-III or in the Diagnostic may partially account for their markedly low preva-
Interview for Borderlines (DIB) (12). Other definitions lence figures.
of borderline personality disorder typically include a Table 2 summarizes studies of the prevalence of
wider range of personality disorders and are referred affective disorder in patients with borderline personal-
to as “broadly defined.” ity disorder. Stone et al. (18) reported a very high
Each of the following sections addresses a different prevalence of broadly defined affective disorder in a
approach to the interface between major affective broadly defined borderline population. Kroll et al. (15)
disorders and borderline personality disorder. In each, found a much lower prevalence in strictly defined,
available evidence is described and analyzed with borderline patients. However, also using strict diag-
respect to hypotheses I-Ill. nostic criteria, Carroll et al. (19), Akiskal (3), and Pope
et al. (10) all found a high prevalence of affective
disorder in their borderline patients. The proportion of
PREVALENCE affective disorder in the borderline populations of the
above reports ranges from 14% to 83%. Pope et al.
Many clinicians have had patients with concurrent (10) found that major depression accounted for most
diagnoses of major depression and borderline person- of the affective disorders in their subjects, as did
ality disorder. The availability of reliable diagnostic Carroll et al. (19). The latter study provided data on
criteria has enabled investigators to ask whether there subgroups: endogenous and primary major depres-
are more of these patients than would be expected. sions were common but not predominant; for patients
They have taken two approaches-looking for person- with secondary depressions, alcoholism and drug
ality disorders in patients with major affective disease abuse were common primary diagnoses. In contrast to
and looking for major affective disease in borderline these two studies, Akiskal (3) suggested that atypical
patients. bipolar and dysthymic disorder were especially preva-
Table 1 summarizes four studies of the prevalence of lent.
personality disorders in patients with a major affective The available evidence suggests a 40%_60% over-
disorder. Using chart reviews, Charney et al. (13) lap between major affective disorder and borderline
found many more patients with diagnosable personal- personality disorder among hospitalized and clinic
ity disorders among patients with unipolar nonmelan- patients. The prevalence of major affective disorders in
cholic depression than among those with either unipo- the general population is estimated to be S%-1O%
tar melancholic or bipolar affective depression. For (20). Baron has estimated the prevalence of strictly
unipolar nomelancholic patients with a personality defined borderline personality disorder to be 1.6%-
disorder, the most common personality traits were 4% (unpublished 1983 paper); this estimate was mdi-
borderline and histrionic; in contrast, subjects with a rectly supported by Loranger et al. (21). According to
personality disorder in the other two groups were Medical World News (22), Kernberg, using a much
more likely to have obsessional traits. Unfortunately, broader concept, has suggested that the prevalence
the investigators were able to assess reliably only may be as high as 15%. Thus, compared with the
borderline, compulsive, histrionic, and hostile traits, general population, inpatients and clinic patients are at
making it impossible to give specific axis II diagnoses. much greater risk than expected of having both diag-
Using both chart review and patient interviews, Ga- noses. However, patients are not reflective of the
viria et al. (14) found a similarly low prevalence of general population. For example, Kroll et al. (15)
borderline personality disorder in outpatients with estimated that the prevalence of borderline personality
bipolar affective disorder. Kroll et al. (15), studying disorder for inpatients is 8.5% (DSM-III criteria) to
consecutive admissions to a general psychiatric unit, 18% (DIB criteria). According to DSM-III criteria, the
found a low prevalence of borderline personality disor- prevalence of affective disorder in psychiatric popula-
der in patients with affective disorder (subtypes un- tions has been estimated to be 22% (23) to 40% (24).
specified). In contrast, Friedman et al. (16), assessing Important questions remain about the comparability
consecutive admissions to an adolescent depression of the patient populations of these various studies.
unit, reported that most met DSM-III criteria for Still, existing data indicate an unexpectedly high con-
major depression and borderline personality disorder; currence of affective disorder with borderline personal-

278 Am] Psychiatry 142:3, March 1985

 JOHN G. GUNDERSON AND GLEN R. ELLIO1T

TABLE 1. Studies of Borderline Personality Disorder in Patients With Affective Disorder

Percentage of Patients Wit h Personality Disorder

Study N Type of Affective Disorder Borderline Other

Charney et al. (13)a 30 Bipolar 0-6 17-23

66 Melancholic unipolar 0-3 11-14
64 Nonmelancholic unipolar 25 36
Gaviria et al. (14)b 100 Bipolar 13 12
Kroll et al (1S)c 39 Unspecified 8 -
Friedman et al. (16)a 42 Major depressive 67 24
7 Atypical depressive and dysthymic 43 0
3 Bipolard 100 0

‘DSM-III affective
criceria disorder
for and for personality “traits,” including borderline and histrionic, compulsive, and hostile (“other”).
bRDC for affective disorder; broad criteria for personality disorders.
CDSMJIJ criteria for affective disorder, with no breakdown within that classification, and for nonborderline personality disorder; Diagnostic Interview for
Borderlines (DIB) for borderline disorder.
dTwo atypical and one mixed.

TABLE 2. Studies of Affective Disorder in Patients With Borderline PHENOMENOLOGY

Personality Disorder

Percentage of Grinker et al. (25) first noted the importance of

Type of Borderline depressive symptoms in the presentation of many
Study N Affective Disorder Patients borderline patients, a finding confirmed by Gunderson
Stone et al. (18)a 36 Unspecified 83 et al. (26). Typically, such patients have signs and
Kroll et al. (1S)b 21 Unspecified 14 symptoms of depression that closely resemble those of
Carroll et al. (19)C 21 Major depressive 62 unipolar nonmelancholic depression (3, 13, 27, 28).
Schizoaffective,
However, these and many other studies showed that
depressed 14
Minor depressive S other, nonaffective signs and symptoms-especially
Akiskal (3)d 100 Recurrent unipolar 6 impulsive and interpersonal ones-readily discrimi-
Atypical bipolar 17 nate unipolar depressed patients from borderline pa-
Dysthymic 14
tients (15, 29-32). Borderline patients have these latter
Cyclothymic 7
Pope et al. (10)d 33 Major depressive 39
characteristics whether or not they have prominent
Bipolar 9 affective symptoms, and unipolar depressed patients
Dysthymic 3 with no personality disorder do not, making them
aBroadly defined criteria for borderline disorder and for depression. unlikely to be secondary effects of depression.
bDiagnostic Interview for Borderlines (DIB) used for borderline disorder;
Hypothesis II suggests that affective symptoms in
DSM-IIt criteria for affective disorder, with no breakdown within that
classification.
borderline patients are a result of the character pathol-
cDIB used for borderline disorder; RDC for affective disorder; 48% of the ogy but does not specify whether these symptoms are
patients with major depressive disorder had endogenous disorder; 38% had identical to those of primary depression. In fact,
primary disorder.
dDIB and DSM-1!I criteria for borderline disorder; DSM-III criteria for several investigators have identified important qualita-
affective disorder. tive differences: the depressive experience of borderline
patients is marked by loneliness, emptiness, and bore-
it disorder, which argues against their being indepen- dom rather than by guilt, remorse, and acute failures in
dent, overlapping syndromes (hypothesis III). self-esteem (4, 25, 33, 34). Gunderson (29) found that
The existing evidence is inadequate to show whether depressed borderline subjects often complain of chron-
any specific type of affective disorder is uniquely ic dysphoria, boredom, and loneliness, but, in addi-
relevant to the borderline syndrome. To clarify this, tion, many report more typically depressive themes of
studies are needed that assess the prevalence of border- major loss and sustained feelings of worthlessness and
line and other axis II diagnoses among patients with hopelessness. Such a heterogeneous result might be
each of the axis I diagnoses, including particularly the expected, because the sample contained a mixture of
subtypes of the affective disorders. Studies are also patients with and without concurrent affective disor-
needed to assess the prevalence of affective disorders der (10). Still, such features as the inner sense of
among patients representing the full range of axis II badness, deprivation, and conscious rage from early
diagnoses. However, even these studies would not life seem to be distinctive components of depression in
control for the likely tendency of patients who fulfill borderline patients.
criteria for both an affective and a borderline disorder There have been a few comparisons of affective
to be especially prone to be in psychiatric treatment. symptoms in nondepressed borderline patients and
Measurement of the overlap of these disorders in the nonborderline depressed patients. Conte et at. (35)
general population is needed, combined with accurate administered a self-report questionnaire to inpatients
estimates of the prevalence of each alone, to resolve diagnosed according to DSM-III as having either
this issue conclusively. nonpsychotic major depression or borderline personal-

Am J Psychiatry 1 42:3, March 1985 279

BORDERLINE PERSONALITY DISORDER AND AFFECTIVE DISORDER

ity disorder. With respect to affectively relevant mate- 100 strictly defined borderline patients who were
rial, the two groups resembled each other in feeling followed for 6-36 months. In that time, 29 had an
hopeless, out of control, unhappy, lonesome, empty, episode of major depression; four, of mania; 1 1, of
overwhelmed, and indecisive. Borderline patients were brief hypomania; and eight, of mixed affective disor-
much more likely than depressed patients to feel der. Thus, within a 3-year period, 52% of this popula-
disappointed in and let down by others close to them tion reportedly developed some form of affective disor-
and to view themselves as failures. Of special interest der. As noted, many of these patients had a diagnosis
was the statement, “I want to hurt myself.” None of of concurrent affective and borderline disorders at the
the depressed patients but 40% of the borderline onset of the study; yet, even of the 55 patients who had
patients answered yes to this question. In an analogous no initial diagnosis of affective disorder, 11 (20%) had
study, Snyder et al. (31), also using DSM-III criteria, an episode of major depression and four (7%) commit-
obtained MMPI profiles of 26 borderline patients and ted suicide. However, detailed methods and results
19 patients with dysthymic disorder. MMPI depres- from this work have yet to be published.
sion scales were elevated in both groups, but the In contrast, Stone (37) reported that only two of 33
former appeared to experience boredom and emptiness broadly defined borderline patients went on to develop
more intensely than did the latter. Kroll et al. (15) newly diagnosed affective disease; both had frank
obtained comparable MMPI results for their border- manic episodes. He also compared 2-3-year outcomes
line population. of 27 broadly defined borderline patients with 18
Overall, existing research does not confirm hypothe- psychotic patients, most of whom were thought to be
sis I-that major affective disorder is necessary to schizophrenic (38). Global ratings showed improve-
generate other aspects of the borderline syndrome. ment for 20 borderline and four psychotic patients and
Insofar as the affective symptoms of affective and no improvement or worsening in seven and 14, respec-
borderline personality disorder have both similarities tively. Problems of idiosyncratic diagnostic use and
and differences, no strong conclusion can be reached rater bias make these findings difficult to interpret; at
about whether depression arises from the borderline least some of the psychotic patients had bipolar affec-
syndrome (hypothesis II) or the two are independent tive disorder.
disorders (hypothesis III). Two more recent longitudinal studies of borderline
patients used better-defined selection criteria. Pope et
al. (10) prospectively studied patients at McLean Hos-
COURSE pital diagnosed according to DSM-III as having either
bipolar affective disorder or schizophrenia, as well as a
The first studies of the longitudinal course of pa- group of strictly defined borderline patients. They
tients with borderline personality disorder were flawed concluded that over the 4-7-year follow-up period,
with idiosyncratic methods of subject recruitment, borderline patients generally did about as poorly as
making it difficult to generalize the findings (table 3). schizophrenic patients, except for having somewhat
In a 6-8-year follow-up study from the Michael Reese better occupational or academic functioning. Nearly
Hospital, Werble (36) described the outcome of 28 of half of the borderline patients in this study had concur-
an original S 1 broadly defined borderline patients; the rent diagnoses of a major affective disorder. This
rest of the sample either refused to provide informa- group did substantially better than did borderline
tion or could not be found. She concluded that the patients without an affective disorder; their overall
patients showed no substantial change over that time. prognosis was intermediate between the schizophrenic
Gunderson et al. (26) and Carpenter et al. (27) studied and bipolar affective groups. This study confirmed the
the 2-S-year course of an operationally defined sample serious, long-term morbidity associated with a diagno-
of borderline patients and a matched sample of schizo- sis of borderline personality disorder but suggested
phrenic patients at the National Institute of Mental that coexistence of an affective disorder may improve
Health (NIMH). The borderline patients showed little the prognosis. It also offered evidence that borderline
change over time in the type
of dysfunction; or amount personality diagnosis itself is relatively stable over
the only reported difference between the two patient time; however, patients with a concurrent diagnosis of
groups was that, at S years, the former had a better affective disorder can be expected to have the fluctua-
quality of social contacts. Both of these studies indicat- tions in course typical of that disorder.
ed that the borderline disorder was associated with In an even longer-term study, McGlashan (39) re-
substantial morbidity but that, contrary to expecta- ported a 15-year follow-up of patients treated at
tions at the time, the patients did not go on to develop Chestnut Lodge in which borderline patients were
schizophrenia. compared with several other diagnostic groups accord-
Subsequent research on natural history is more ing to DSM-III criteria. McGlashan also found that
directly relevant to questions about the interface of the borderline diagnosis was stable over time but
affective disorder and borderline personality disorder concluded that the overall course for borderline pa-
(table 3). Easily the most striking evidence for an tients closely resembled that for patients with affective
association between the two is research cited by Akis- disorder. The outcome measures used were hospital-
kal (3) in a review paper. He reported on a group of ization, work, social activity, symptoms, and global

280 Am J Psychiatry 142:3, March 1985

 JOHN G. GUNDERSON AND GLEN R. ELLIOTF

TABLE 3. Follow-Up Studies of Patients With Borderline Personality Disorder

Length of
Follow-Up Diagnosis of Most Outcome Relative to
Study N (years) Patients at Follow-Up Comparison Group

Werble (36) 28 6-8 Borderline personality disorder

Gunderson et al. (26) 24 2 Borderline personality disorder Same as schizophrenic patients
Carpenter et al. (27) 24 S Borderline personality disorder Same as schizophrenic patients
Akiskal (3) 100 0.5-3.0 Affective disorder
Stone (37, 38) 27 2-3 Borderline personality disorder Better than psychotic patients
(2 patients had affective disorder)
Ikpe et al. (10) 27 4-7 Borderline personality disorder; Same as schizophrenic patients;
10 of 14 patients with con- borderline patients with con-
current affective disorder current affective disorder
still had it significantly better on some
measures
McGlashan (39) 70 or 78 15 Borderline personality disorder Same as patients with affective
(2-32) disorder; borderline patients
and patients with affective
disorder better than
schizophrenic patients

functioning. This study is the first, to our knowledge, ations, especially in late adolescence. In 1983, for
to employ a comparison group of patients with severe example, Gossett et al. (40) noted that some adoles-
neurotic character and other personality disorders. cents who present with character problems-especially
The outcome of patients in this group resembled those those with “excited and expansive behavior alternat-
of the borderline and affective cohorts; patients in all ing with periods of lethargy” (p. 143)-later develop a
three groups fared better on all measures than did the full-blown affective syndrome as young adults.
schizophrenic group. Most of the studies described in this section were
The differences between the McLean and Chestnut devised to test whether the illness of patients with
Lodge studies probably arise from a combination of borderline personality disorder eventually resolves into
characteristics of the latter treatment setting: 1) the some discrete form of affective disorder (hypothesis II),
schizophrenic patients may have been especially unre- and most failed to support that idea. So far, no one has
sponsive to treatment, 2) the borderline patients may done a comparable test for hypothesis I, to see how
have responded to the unusually extensive treatment many depressed patients, especially those with a non-
provided, which included more than 3 years of hospi- melancholic unipolar depression, subsequently mani-
talization and intensive psychotherapy, and 3) the fest other features of the borderline syndrome.
patients with affective disorder, most of whom were
chronically depressed, may have been somewhat less
responsive to treatment compared with the patients FAMILY PREVALENCE
with primarily bipolar affective disorder used in the
McLean study. Thus the McLean study may inadver- Since 1977, there have been six reports on the
tently have emphasized differences in the borderline familial prevalence of psychiatric disorders in relatives
and affective patients, and the Chestnut Lodge study of borderline patients (3, 10, 18, 21, 37, 38, 41-43)
may have minimized them. Alternatively, borderline (table 4). The aim of these studies was to highlight
patients may have especially bad outcomes only for the clusters of specific diagnostic categories within fam-
first S years or so, followed by some remission in their ilies. Most available studies looked at first-degree
subsequent course according to usual measures of relatives (parents, siblings, and children) of identified
clinical outcome. patients. This approach can demonstrate an excess
To date, Akiskal (3) stands virtually alone in identi- occurrence of specific disorders but cannot distinguish
fy ing the affective disorders as a common outcome for between purely genetic effects and environmental fac-
many patients with borderline personality disorder. tors of growing up in a particular setting.
Other investigators have found that the borderline Stone made the groundbreaking studies in research
syndrome is relatively stable over time. It seems most on families of borderline patients (18, 37, 38, 41). His
likely that Akiskal’s result reflects the effects of concur- reports derived from three sequential sample popula-
rent affective disorder. As has been reported (10), such tions-his clinical practice, the New York State Psychi-
patients appear to be stable with respect to the person- atric Institute, and the Westchester Division of Cornell.
ality diagnosis but may have repeated episodes of These studies documented that first-degree relatives of
affective disease. Available evidence in this area best patients with broadly defined borderline personality
supports the suggestion that the two disorders are disorder were at greater risk of having an affective
unrelated (hypothesis III). However, attention still disorder than were relatives of a control group of
should be given to possible developmental consider- patients with psychosis (primarily schizophrenia but

Am J Psychiatry 1 42:3, March 1985 281

BORDERLINE PERSONALITY DISORDER AND AFFECTIVE DISORDER

TABLE 4. Psychopathology in First-Degree Relatives of Borderline Patientsa

Diagnosable First-Degree Relatives

Study Diagnosis of Subjects Diagnosis Percent

Stone (37, 38, 41) Borderline personality organization, including various forms Affective disorder 19.0
of personality disorder (N=23)
Psychotic personality organization, mainly schizophrenia, Affective disorder 13.0
including schizoaffective and psychotic affective disorders Schizophrenia 6.7
(N=23)
Borderline personality, loosely corresponding to DSM-III Affective disorder 27.4
borderline personality disorder (N 13)
Stone et al. (18) Borderline personality, loosely corresponding to DSM-llI Affective disorder 14.1
borderline personality disorder (N39) Schizophrenia 2.2
Psychotic personality organization, mostly schizophrenia, Affective disorder 11.0
including schizoaffective and psychotic affective disorders Schizophrenia 14.4
(N=36)
Askiskal (3) Borderline personality, loosely corresponding to DSM-III Unipolar affective disorder 17.0”
borderline personality disorder (N= 100) Bipolar affective disorder 17.Ob
Schizophrenia 3.Ob
Andrulonis et al. (42) Borderline personality, loosely corresponding to DSM-III Unipolar affective disorder 32.lb
borderline personality disorder (N= 106) Schizophrenia 3.8b
Affective disorder (unipolar, N=32; dysthymic, N= 19; Unipolar affective disorder 455b
bipolar, N=4) (N=SS) Schizophrenia 55b
Schizophrenia (N=SS) Unipolar affective disorder 377b
Schizophrenia 20.Ob
Loranger et al. (21)C Borderline personality disorder (N83) Unipolar affective disorder 6.4
Bipolar affective disorder 0.5
Borderline personality disorder 11.7
Bipolar affective disorder (N= 100) Unipolar affective disorder 7.4
Bipolar affective disorder 2.3
Schizophrenia 0.2
Borderline personality disorder 0.7
Schizophrenia (N= 100) Unipolar affective disorder 2.0
Bipolar affective disorder 0.4
Schizophrenia 2.9
Borderline personality disorder 1.4
Soloff and Millward (43) Borderline personality disorder (N48) Unipolar affective disorder 8.7
Schizophrenia 2.6
Antisocial personality disorder 7.0
Unipolar affective disorder (N=32) Unipolar affective disorder 15.0
Schizophrenia 1.9
Antisocial personality disorder 4.4
Schizophrenia (N=42) Unipolar affective disorder 5.0
Schizophrenia 4.5
Antisocial personality disorder 3.5
Pope et al. (10)c Borderline personality disorder (N33) Affective disorder 7.2
Antisocial, borderline,
histrionic personality disorders 7.7
Borderline personality disorder with affective disorder Affective disorder 10.1
(N=17) Antisocial, borderline,
histrionic personality disorders 5.8
Borderline personality disorder without affective disorder Affective disorder 1.6
(N=16) Antisocial, borderline,
histrionic personality disorders 9.8
Bipolar affective disorder (N=16) Affective disorder 7.5
Antisocial, borderline,
histrionic personality disorders 0.6
Schizophrenia (N31) Affective disorder 0.6
Antisocial, borderline,
histrionic personality disorders 2.2
aAlI diagnostic categories are consistent with DSM-I!! unless otherwise indicated.
bpercent of patients who had at least one relative with the disorder, as opposed to percent of first-degree relatives with the disorder.
CRelatives were assessed by raters who were blind to the diagnosis of the index subject.

also some psychotic affective disorders) and intro- der; another-possibly overlapping-17% had at least
duced the idea of a genetic affiliation between border- one relative with bipolar affective disorder.
line personality and affective disorder. From a some- These initial reports provided incentive for several
what different viewpoint, Akiskal (3) reported that methodologically superior studies. Andrulonis et al.
17% of his sample of borderline patients had at least (42) initially reported that first-degree relatives of a
one first-degree relative with unipolar affective disor- sample of patients with strictly defined borderline

282 Am J Psychiatry 142:3, March 1985

 JOHN G. GUNDERSON AND GLEN R. ELLIOTT

disorder appeared to have a greater-than-expected affective disorder in the first-degree relatives of border-
prevalence of affective disorder. However, Andrulonis line patients resembled the results of Loranger et al.
and Vogel (unpublished paper) subsequently conclud- (21) and Soloff and Millward (43). Comparison within
ed that the rates observed were comparable to those the group shows that the presence of an affective
for relatives of a schizophrenic comparison group and disorder in the identified patient is associated with a
much less than those for relatives of a sample of fivefold increase in the prevalence of affective disorder
patients with affective disorders. They tried explicitly in the family. This suggests that research results might
to exclude borderline patients who had a concurrent be greatly affected by selection criteria changing the
affective disorder, and their relative success in doing so percentage of affective disorder in samples of border-
may account for some of the differences between their line patients. Pope et al. (10) also found more people
results and those of the earlier studies. with personality disorders among the relatives of pa-
Loranger et al. (21) were the first to use raters blind tients with borderline personality, with a trend toward
to the diagnosis of the patient to assess relatives; for a greater prevalence among relatives of borderline
comparison, they used patients with bipolar affective patients without affective illness.
disorder and patients with schizophrenia as controls Taken together, these studies are informative but
(table 4). Due to problems of obtaining information incomplete. Relatives of borderline patients have an
about relatives, the investigators used the same criteria increased prevalence of personality disorders corn-
but a lower threshold for diagnosing relatives than for pared with those of patients with affective disorder or
diagnosing the index subjects. With respect to affective schizophrenia, but neither the overall magnitude nor
disorder and schizophrenia, first-degree relatives of the distribution among types of personality disorders is
borderline patients more closely resembled relatives of known. Also, it is unclear how these relatives would
bipolar patients than they did relatives of schizophren- differ from those of patients with nonborderline per-
ic patients, although their prevalence of bipolar disease sonality disorders. It seems likely that much of the
was closer to that of the latter group. Most strikingly, observed increase in prevalence of affective disorders
they were at least 10 times more likely than relatives of in families of borderline patients results from the
either comparison group to have been treated at some concurrent presence of affective disorder in the border-
time for a “borderline-like” disorder. line patients, but what types of affective disorder? The
Soloff and Millward (43) also studied first-degree data suggest that bipolar disorder probably is not
relatives of borderline patients and of control subjects involved. Much of the variability among studies results
with unipolar depression and schizophrenia (table 4). from idiosyncracies in sample populations, diagnostic
They too found a higher prevalence of affective disor- criteria, and methodological approaches. The studies
der in families of borderline patients compared with do not support the idea that either disorder arises from
those of schizophrenic patients but much less than the other (hypotheses I and II). The results of Pope et
with families of unipolar depressed patients. Looking al. (10) are consistent with their being mutually inde-
at antisocial personality disorder, they found that the pendent (hypothesis III).
families of borderline patients had almost twice the
prevalence of either comparison group. Unfortunately,
they did not assess patients with other types of person- PSYCHODYNAMIC FACTORS
ality disorder, including borderline personality disor-
der, so no comparisons can be made with the results of Major depression is diagnosed in DSM-III on the
Loranger et al. (21). Within the borderline sample, basis of specific signs and symptoms, without regard to
however, they identified an index group of patients their etiology. Thus, even though some borderline
with schizotypal personality disorder. There was no patients meet such criteria, they may differ from other
evidence of an increased prevalence of schizophrenia in depressed patients on the basis of such important
first-degree relatives of these patients either; the preva- psychodynamic factors as underlying meanings, cir-
lence of affective disorder was comparable to that of cumstances that lead to depression, and the effect of
the entire borderline group. This argues against the the depression on the person’s ongoing self- and world
idea that studies of borderline patients with schizoty- view. Data for this area of investigation derive largely
pal features may artificially lower the prevalence of from the clinical observations and theoretical infer-
affective disorder in relatives. ences drawn from psychotherapy. Therefore, the con-
Finally, Pope et al. (10) assessed relatives of several clusions here must be understood in the context of
groups of carefully diagnosed patients, including bor- possible observer and sampling biases present in such
derline patients with or without concurrent affective work.
disorder, patients with schizophrenia, and patients There have been two somewhat varying psychody-
with bipolar affective disorder (table 4). Raters blindly namic formulations of depression in borderline pa-
assessed the relatives for schizophrenia, affective disor- tients. The first and most widely known is “abandon-
der, and three personality disorders. Again, relatives of ment depression,” proposed by Masterson (5). He
borderline patients had an increased prevalence of posited that traumatic separations during the rap-
affective illness-but only if the patient also had an prochement subphase of early development create a
affective disorder. Without this distinction, the rate of basic depressive nexus in borderline patients that is

Am J Psychiatry 142:3, March 1985 283

BORDERLINE PERSONALITY DISORDER AND AFFECTIVE DISORDER

then defended against with rage, primitive psychologi- patients. This finding confirms the central hypothesis
cal defenses, action, and withdrawal. This depression of the first formulation and is consistent with the
is thought to underlie many of the behaviors that predictions of the second one. This and other studies
typify the borderline syndrome and to endure as a (49, 50) also implicate early parental loss as especially
chronic part of the character, breaking through only at prevalent in borderline patients. Such work suggests
times into symptomatic expression. The content of this that identifiable pathogenic developmental processes
depressive experience-emptiness, despair, and hope- may help to differentiate depression in borderline
lessness-has been compared with Spitz’s concept of patients from primary depression.
anaclitic depression (25, 44). Masterson’s formulation Clearly, the above material supports the idea that
was essentially reiterated in later accounts by Harto- borderline personality disorder can produce depres-
collis (34) and Rinsley (8). sion (hypothesis II), but it is not inconsistent with
A second formulation emphasizes the pervasive other hypotheses. For example, the traumatic separa-
sense of “inner badness” that many borderline patients tions emphasized in the first formulation and the
experience. It too conceives of depression as a part of warded-off rage postulated in the second might be-
the borderline character structure, this time as a result come pathological only with a preexisting biophysiolo-
of an enduring self-image as evil and destructive. This gical vulnerability. In this regard, Donald Klein (1, 2)
formulation gives a central dissociated
role to anger: has suggested that the usually tolerable frustrations of
depression is both a reaction to and an introjected early life may permanently affect some people because
expression of this affective state (45, 46). Kernberg (4) of biological or physiological factors that mold their
viewed this self-accusation as the product of a primi- responses. If true, this would be consistent with the
tive superego and posited that the depression of bor- idea that the borderline syndrome arises from a pri-
derline patients characteristically is accompanied by mary affective disorder (hypothesis I).
ego disorganization. This formulation emphasizes the
central importance of anger toward needed others:
anger that is experienced as dangerous is redirected BIOLOGICAL FACTORS
inward and finds its introjected expression in depres-
sion. At present there is no independent biological hy-
According to the first formulation, borderline pa- pothesis for borderline personality disorder. The clini-
tients use anger and other types of “acting out” to cal tests suggested as markers of depression include a
defend against an anaclitic-like depression, but not nonsuppressing dexamethasone suppression test (DST)
always successfully. According to the second, their result, blunted thyrotropin (TSH) response, and al-
depression is characterized by self-condemnation that tered sleep EEGs. None is sensitive or specific enough
is thought to represent a defensive reaction to aggres- for depression to be an ideal research tool for studying
sive impulses toward others. In both these formula- a depressive component in other disorders (Si), but
tions, the depressive affect is symptomatic only inter- they do represent a theoretically attractive way of
mittently and is intimately connected to the current examining the overlap between borderline and affec-
state of object relationships. However, the personality tive disorders.
structure contains a lasting vulnerability such as poor Carroll et al. (19) reported that of 21 subjects with
self-image or the need to avoid the experience of DSM-III diagnoses of borderline personality disorder,
depression. 62% had an abnormal DST. Nearly all of the non-
Few investigators have attempted to categorize de- suppressors had a concurrent diagnosis of major de-
pressions according to differentiating psychological pression; almost none of the suppressors had that
features (47). This is true even for such large sub- diagnosis. Sternbach et al. (52) studied 13 women who
groups as chronic versus reactive and melancholic completely met and 1 1 who were one criterion short of
versus nonmelancholic depressions. One exception is meeting a DSM-III diagnosis of borderline personality
that both of the above formulations differentiate the disorder. All 24 women had unspecified “depressive
depression of borderline patients from those depres- features,” and 17 met DSM-III criteria for major
sions characterized by guilt, remorse, or a sense of depression. A nonsuppressing DST was found in 65%
failure, i.e., those depressions believed to be character- of patients with major depression and 29% of the
istic of a healthier personality and a more integrated remainder, compared with 7% of 14 healthy control
superego. Yet, even in this instance, the difference may subjects. In contrast, Soloff et al. (53) studied seven
be mainly one of degree, insofar as healthier depressed patients with borderline, five with schizotypal, and
individuals have only partial aspects of the self under seven with mixed personality disorders diagnosed ac-
attack, but borderline patients experience attacks on cording to DSM-III. Fourteen of these 1 9 patients also
the entire self. met criteria for major depression. Soloff et al. found a
A second inroad into this area is offered by a recent total of three abnormal DSTs (16%) in the 19 patients,
empirical study by Soloff and Millward (48), which all in subjects who also had been diagnosed as de-
indicates that a pathological sensitivity to separation pressed. These results are difficult to reconcile with
experiences is twice as common among borderline each other. If, as many have suggested, the DST is a
depressed patients as among nonborderline depressed nonspecific indicator of severity of illness, they may, in

284 Am] Psychiatry 142:3, March 1985

 JOHN G. GUNDERSON AND GLEN R. ELLIOTT

part, reflect such differences among the patient popula- classes. Antipsychotics have been used for many years
tions. in varying doses; several investigators currently advo-
Blunting of the usual TSH response to thyrotropin- cate low doses (61, 62).
releasing hormone (TRH) is even less established as a Donald Klein’s controversial category of “hysteroid
marker of depression than is the DST. However, two dysphoria” (1, 2) is a type of recurrent, atypical
groups have extended its use to borderline patients. depression thought to occur in some borderline pa-
Garbutt et al. (54) found a blunted response in seven tients that may be particularly responsive to mono-
(47%) of 15 borderline subjects diagnosed by DSM-III amine oxidase inhibitors (MAOIs). Initial data suggest
criteria, compared with no blunting in any of a group that MAOIs are useful primarily in decreasing the
of matched control subjects. Five of the patients with a chronic feelings of emptiness or boredom and impul-
blunted response had concurrent diagnoses of alcohol- sive actions, including those which are self-damaging,
ism or depression, but two had only the borderline and that they are less effective for treating other long-
diagnosis. Sternbach et al. (52) also measured TSH term borderline features (63).
blunting; 41% of the depressed and 29% of the In retrospective reviews of drug responses of well-
nondepressed borderline women they studied showed defined hospitalized borderline patients, both Soloff
blunting. These results are somewhat higher than the (58) and Cole et al. (60) concluded independently that
25%-30% typically reported in pure depression (55). drug response was heterogeneous. Some patients
Sleep EEGs in borderline patients have been more seemed to respond to antidepressants, others to anti-
promising than either the DSTor the TSH response to psychotics. Soloff (58) found that either class of drugs
TRH. Akiskal (3) first noted that REM latencies in a was superior to no drug therapy; 63% of medicated
sample of eight borderline patients resembled those of patients showed improvement, compared with 19% of
a depressed control group. His finding was confirmed unmedicated patients. He also found no differences in
by McNamara et al. (56), who studied 10 borderline outcome of patients given antipsychotics, antidepres-
patients who also had substantial depressive symptoms sants, or a combination of the two.
during testing. Bell et al. (57) compared the sleep Cole et al. (60) found that borderline patients with a
architecture of a group of eight inpatients with DSM- concurrent affective disorder were more responsive to
III diagnoses of both major depression and borderline antidepressants and antipsychotics than were those
disorder with that of 1 1 nonborderline depressed who had only borderline personality disorder. Howev-
patients and found no differences between the groups. er, such patients did not have the response rates
The cause of decreased REM latency in depression expected for purely depressed patients. In a study using
remains unknown, but, inasmuch as it is specific for some of the same patient population, Pope et al. (10)
depression, it suggests that the affective disorder of found that among patients having only borderline
borderline patients resembles that of patients with personality disorder, only 37% were treated with
other types of depression. However, all of these subject drugs and none showed a clear response. However,
populations were selected in ways that probably as- 77% of borderline patients with an affective disorder
sured a high prevalence of both borderline and affec- received medications, and about half definitely re-
tive disorder, so the reports do little to distinguish sponded. Responses in this latter group were equally
among the various hypotheses. distributed among patients receiving antipsychotics,
The use of biological markers may be of some utility antidepressants, or lithium. This result is in accord
in studying the interface between affective disorders with the report by Charney et al. (13) that fewer
and borderline personality, but more progress needs to nonmelancholic depressed borderline patients re-
be made in finding sensitive and specific markers for sponded to drug treatment than did those without a
depression. In addition, when such tests are used in personality disorder (36% versus 76%, respectively).
studies of borderline patients, care must be taken to Similarly, Gaviria et al. (14) found that bipolar pa-
distinguish between subjects with and without concur- tients with borderline personality disorder had a poor-
rent affective disorders and to use control groups of er response to lithium maintenance than did those
patients with other personality disorders. Still, given without a personality disorder. There is no informa-
the current state of such tests, they are unlikely to be tion about whether individual patients respond equal-
definitive in clarifying interrelationships between these ly well to any of the drug classes, but that does not
two disorders in the near future. seem likely.
Much remains to be learned about the use of
medication in borderline personality disorder. With
DRUG RESPONSE respect to the overlap between affective disorder and
borderline disorder, antidepressants appear to be use-
Although borderline patients account for 8%-20% ful at times, especially when an affective disorder is
of hospital admissions and frequently receive medica- present. But even when they are useful, the response
tions (58-60), controlled studies of their drug respon- rate appears to be lower than expected. Such findings
sivity have just recently begun. Perhaps as a result of do not fit especially well with any of the three hypothe-
this lack of needed data, competing advice ranges from ses about the relationship between affective and bor-
using no drugs at all to using each of the major drug derline personality disorders.

Am] Psychiatry 142:3, March 1985 285

BORDERLINE PERSONALITY DISORDER AND AFFECTIVE DISORDER

DISCUSSION ing biological or psychological vulnerability to depres-

sion.
Borderline and affective disorders have been under Hypothesis Ill, which suggests that affective disorder
intense study during the past decade. That interest has and borderline personality disorder are independent
heightened awareness of them and broadened their but can coexist, has somewhat more support (table 5).
role in differential diagnoses. It seems clear now that Both family prevalence and longitudinal studies sug-
both diagnostic categories include heterogeneous sub- gest that the disorders can follow independent courses
groups whose differentiating features remain to be in the same person. Thus, families of borderline pa-
elucidated. As a byproduct of the work in this area, the tients have a higher prevalence of personality disorders
longstanding controversy about the boundaries of whether or not the identified patients have a concur-
borderline personality disorder has shifted from rent affective disorder; but only the families of patients
schizophrenia to the affective disorders, and the long- with an affective disorder also have a loading for
standing interest in the relationship between personal- affective disease. However, prevalence studies repeat-
ity and affective syndromes has surfaced with fresh edly have shown that borderline personality disorder
vigor and rapidly accumulating information. The easi- and affective disorder coexist in patients more fre-
est summary of the data just reviewed is that more quently than expected. Also, the affective symptoms of
research is needed and that an assessment of the patients with both disorders ought to be more respon-
available literature can only capture the state of a field sive to drugs than they appear to be.
still in motion. Still, we believe that the existing We propose a fourth hypothesis that we think fits
research strongly suggests the inadequacy of several of the data better than does any of the other three. Both
the existing hypotheses about the relationship between borderline personality and affective disorder are diag-
borderline personality disorder and affective disorder. nosed with signs and symptoms arising from many
We suggest that a different viewpoint is needed. sources. Hypothesis IV suggests that the observed
Hypothesis I suggests that a primary depression can concurrence of affective and borderline symptoms
produce behaviors that result in a diagnosis of border- results from that heterogeneity. For either disorder,
line personality disorder. This hypothesis receives little individuals may start with a biophysiological vulnera-
support (table 5). The stability of the borderline diag- bility that increases their risk of being psychologically
nosis over time argues against it, as does the fact that impaired in early development. Such early traumas
antidepressants are not the treatment of choice for may create vulnerability to either or both disorders,
even the majority of such patients. Also, borderline but the actual presentation varies as a function of later
patients have an increased frequency of relatives with physiological and psychological reactions to environ-
personality disorders compared with patients with ment and temperament. The key to the overlap and
affective disorder, suggesting that the disorders can run dissimilarities between these two disorders, then, may
independently in families. Furthermore, many border- be a constellation of innate and external factors that
line patients never display a symptom picture of major are inconsequential individually but combine to shape
depression. Those affective disorders with which the depression, chronic dysphoria, or borderline behav-
borderline syndrome seems to have the greatest over- ior-alone or in any possible combination.
lap-unipolar nonmelancholic depressions-have the Hypothesis IV is supported by all of the evidence of
weakest evidence for biogenetic determinants (S 1 , 64) heterogenity within this patient population (table 5).
and are relatively unresponsive to drugs (51). None- As noted, the actual overlap between borderline and
theless, some borderline patients do have a concurrent affective disorders remains to be established, but all
major affective disorder that can respond to the ex- existing data suggest that is is higher than expected.
pected medications. Both the phenomenology and drug response are espe-
Hypothesis II proposes that depression is secondary cially consistent with the concept of a variety of
to the primary processes of the borderline syndrome; etiologies leading to a common endpoint. In the family
again, it has no compelling support (table 5). Arguing prevalence studies, distinguishing borderline patients
against it is the evidence that, even for borderline with a full affective syndrome from those without one
patients without depression, there is some family load- simplified and clarified the data-again supporting the
ing for affective disorder. Also, such patients may have existence of subgroups that may have interactive ef-
biological markers like those of depressed patients. fects.
However, there are differences in the depression that Patients at the interface of affective and borderline
borderline patients experience and that of the more personality disorders often present troubling and per-
typical affective disorders, including qualitative aspects sistent problems for the clinician. These patients in-
of symptom presentation and underlying psychody- dude not only those who fit contemporary criteria for
namic etiologies. In addition, borderline patients ap- these disorders but some who fulfill criteria for other
pear to be less responsive to antidepressant medica- diagnoses such as chronic dysphoria, hysteroid dys-
tions than patients with major depressions. Perhaps, phoria, cyclothymia, and other personality disorders.
for at least some borderline patients, depression results Their problems tend to be chronic, with repeated
from an interaction between the consequences of cer- exacerbations, and they often show no or only a
tain behavioral traits and the existence of an underly- partial response to any of the available forms of

286 Am] Psychiatry 142:3, March 1985

 JOHN G. GUNDERSON AND GLEN R. ELLIOTT

TABLE 5. Evaluations of Four Hypotheses About the Interface Between Borderline Personality Disorder and Affective Disorders

Evaluative Strategy
Family Psychodynamic Biological Drug
Hypothesis Incidence Phenomenology Course Prevalence Factors Factors Response

I. Affective disorder is the primary + - - - - 0 + - -

problem; it can produce signs and

symptoms of borderline personality
disorder as well
II. Borderline character pathology is the + + - - + 0 +

primary problem; it can produce

signs and symptoms of affective
disorder as well
III. Affective disorder and borderline - - - + + + - + -

personality disorder are unrelated

but can coexist in the same person
because they are so common
IV. Both diagnoses rely on signs and + + + + + + 0 + + +

symptoms that can arise from a

variety of sources; some of the
resulting heterogeneous population
have both syndromes
aScores represent the consensus of J.G.G. and GRE. about whether the evidence supports or refutes each hypothesis: - - means strongly refutes, - means
somewhat refutes, 0 means equivocal, + means supports, and + + means strongly supports.

treatment. Research is needed to identify the relevant I 1. Spitzer RL, Endicort J, Robins E: Research Diagnostic Criteria
(RDC) for a Selected Group of Functional Disorders, 2nd ed.
subgroups within this heterogeneous group. The evi-
New York, New York State Psychiatric Institute, Biometrics
dence reviewed in this paper suggests that important Research, 1975
factors in such analyses will include the prominence 12. Gunderson JG, Kolb JE, Austin V: The diagnostic interview for
and chronicity of affective symptoms, individual varia- borderline patients. Am J Psychiatry 138:896-903, 1981
tions in psychological and biological factors, longitudi- 13. Charney DS, Nelson JC, Quinlan DM: Personality traits and
disorder in depression. Am J Psychiatry 138:1601-1604, 1981
nal course, family prevalence, and drug responsivity.
14. Gaviria M, Flaherty J, Val E: A comparison of bipolar patients
This conclusion is not critical of the existing boundary; with and without a borderline personality disorder. Psychiatr J
rather, it opposes any conceptualization that calls for University of Ottawa 7:190-195, 1982
sharply discontinuous categorical concepts of these is. Kroll J, Sines L, Martin K, et al: Borderline personality disorder:
construct validity of the concept. Arch Gen Psychiatry
two diagnostic groups. As such, this review points
38:1021-1026, 1981
toward the broader need to accept the present divi- 16. Friedman RC, Aronoff MS, Clarkin JF, et al: History of suicidal
sions between axis I and II as paradigmatic, not actual. behavior in depressed borderline patients. Am J Psychiatry
140:1023-1026, 1983
17. Hirschfeld RMA, Klerman GL, Clayton PJ, et al: Assessing
REFERENCES personality: effects of the depressive state on trait measurement.
Am J Psychiatry 140:695-699, 1983
1. Klein DF: Psychopharmacology and the borderline patient, in
18. Stone MH, Kahn E, FIye B: Psychiatrically ill relatives of
Borderline States in Psychiatry. Edited by Mack JD. New York,
Grune & Stratton, 1975
borderline patients: a family study. Psychiatr Q 53:71-84, 1981
19. Carroll BJ, Greden JT, Feinberg M, et al: Neuroendocrine
2. Klein DF: Psychopharmacological treatment and delineation of
evaluation of depression in borderline patients. Psychiatr Clin
borderline disorders, in Borderline Personality Disorders. Edit-
North Am 4:89-99, 1981
ed by Hartocollis P. New York, International Universities Press,
1977
20. Weissman MM, Myers JK, Thompson D: Depression and its
treatment in a US urban community-1975-1976. Arch Gen
3. Akiskal HS: Subaffective disorders: dysthymic, cyclothymic and
Psychiatry 38:417-421, 1981
bipolar II disorders in the “borderline” realm. Psychiatr Clin
21. Loranger AW, Oldham JM, Tulis EH: Familial transmission of
North Am 4:25-46, 1981
4. Kernberg 0: Borderline personality organization. J Am Psy- DSM-III borderline personality disorder. Arch Gen Psychiatry
choanal Assoc 15:641-685, 1967
39:795-799, 1982
22. The “borderline” personality. Medical World News, April 25,
S. Masterson JF: Treatment of the Borderline Adolescent: A
1983, p 51
Developmental Approach. New York, Wiley-Interscience, l972
6. Grinker RR: Diagnosis of borderlines: a discussion. Schizophr 23. Mezzich JE, Coffman GA, Goodpastor SM: A format for DSM-
Bull 5:47-52, 1979 III diagnostic formulation: experience with 1,1 1 1 consecutive
7. Gunderson JG: Empirical studies of the borderline diagnosis, in patients. Am J Psychiatry 139:591-596, 1982
Psychiatry 1982: The American Psychiatric Association Annual 24. Spitzer RL, Skodol AE, Williams JB, et al: Supervising intake
Review. Edited by Grinspoon L. Washington, DC, American diagnoses. Arch Gen Psychiatry 39:1299-1305, 1982
Psychiatric Press, 1982 25. Grinker RR Sr, Werble B, Drye RC: The Borderline Syndrome:
8. Rinsley DB: Borderline and Other Self Disorders. New York, A Behavioral Study of Ego-Functions. New York, Basic Books,
J ason Aronson, 1982 1968
9. Kernberg 0: Two reviews of the literature on borderlines: an 26. Gunderson JG, Carpenter WT Jr, Strauss JS: Borderline and
assessment. Schizophr Bull 5:53-58, 1979 schizophrenic patients: a comparative study. Am J Psychiatry
10. Pope HG, Jonas JM, Hudson JI, et al: The validity of DSM-III 132:1259-1264, 1975
borderline personality disorder. Arch Gen Psychiatry 40:23-30, 27. Carpenter W1’ Gunderson JG, Strauss JS: Considerations of the
1983 borderline syndrome: a longitudinal comparative study of bor-

Am J Psychiatry I 42:3, March 1985 287

BORDERLINE PERSONALITY DISORDER AND AFFECTIVE DISORDER

derline and schizophrenic patients, in Borderline Personality 47. Val E, Flaherty JA, Gaviria F: Psychological Aspects in Affective
Disorders. Edited by Hartocollis P. New York, International Disorders: Psychopathology and Treatment. Chicago, Year
Universities Press, 1977 Book Medical Publications, 1982
28. KrollJ, Carley K, Sines L, et al: Are there borderlines in Britain? 48. Soloff PH, Millward JW: Developmental histories of borderline
a cross-validation of US findings. Arch Gen Psychiatry 39:60- patients. Compr Psychiatry 24:574-588, 1983
63, 1982 49. Bradley SJ: The relationship of early maternal separation to
29. Gunderson JG: Characteristics of borderlines, in Borderline borderline personality in children and adolescents: a pilot study.
Personality Disorders. Edited by Hartocollis P. New York, Am J Psychiatry 136:424-426, 1979
International Universities Press, I 977 so. Walsh F: Family study 1976: 14 new borderline cases, in The
30. Gunderson JG, Kolb JE: Discriminating features of borderline Borderline Patient. Edited by Grinker RR Sr, Werble B. New
patients. Am J Psychiatry 135:792-796, 1978 York, Jason Aronson, 1977
31. Snyder 5, Pitxs WM, Goodpaster WA, et al: MMPI profile of SI . Baldessarini RJ: Biomedical Aspects of Depression and Its
DSM-III borderline personality disorder. Am J Psychiatry Treatment. Washington, DC, American Psychiatric Press, 1983
139:1046-1048, 1982 52. Sternbach HA, Fleming J, Extein I, et al: The dexamethasone
32. Soloff PH, Ulrich RF: The diagnostic interview for borderlines: suppression and thyrotropin-releasing hormone tests in de-
a replication study. Arch Gen Psychiatry 38:686-692, 1981 pressed borderline patients. Psychoneuroendocrinol 8:459-
33. Cary GL: The borderline condition: a structural dynamic 462, 1983
viewpoint. Psychoanal Rev 49:33-54, 1972 53. Soloff PH, George A, Nathan RS: The dexamethasone suppres-
34. Hartocollis P: Affects in borderline disorders, in Borderline sion test in patients with borderline personality disorder. Am J
Personality Disorders. Edited by Hartocollis P. New York, Psychiatry 139:1621-1623, 1982
International Universities Press, 1977 54. Garburt JC, Loosen PT, Tipermas A, et al: The TRH test in
35. Conte HR, Plutchik R, Karasu TB, et al: A self-report border- patients with borderline personality disorder. Psychiatry Res
line scale: discriminative validity and preliminary norms. J Nerv 9:107-113, 1983
Ment Dis 168:428-435, 1980 55. Loosen PT, Prange AJ Jr: Serum thyrotropin response to
36. Werble B: Second follow-up study of borderline patients. Arch thyrotropin-releasing hormone in psychiatric patients: a review.
Gets Psychiatry 23:3-8, 1970 Am J Psychiatry 139:405-414, 1982
37. Stone MH: The Borderline Syndromes: Constitution, Personal- 56. McNamara E, Reynolds CF III, Soloff PH, et al: EEG sleep
ity and Adaptation. New York, McGraw-Hill, 1980 evaluation of depression in borderline patients. Am J Psychiatry
38. Stone MH: Contemporary shift of the borderline concept from 141:182-186, 1984
a subschizophrenic disorder to a subaffective disorder. Psychiatr 57. Bell J, Lycaki H, Jones D, et al: Effect of preexisting borderline
Clin North Am 2:577-594, 1979 personality disorder on clinical and EEG sleep correlates of
39. McGlashan TH: The borderline syndrome, II: is it a variant of depression. Psychiatry Res 9:115-123, 1983
schizophrenia or affective disorder? Arch Gen Psychiatry 58. Soloff PH: Pharmacotherapy of borderline disorders. Compr
40:1319-1323, 1983 Psychiatry 22:535-543, 1981
40. Gosseet JT, Lewis JM, Barnhard FD: To Find a Way: The 59. Cole JO, Sunderland P III: The drug treatment of borderline
Outcome of Hospital Treatment of Disturbed Adolescents. New patients, in Psychiatry 1982: The American Psychiatric Associa-
York, Brunner/Mazel, 1983 tion Annual Review. Edited by Grinspoon L. Washington, DC,
41. Stone MH: The borderline syndrome: evolution of the term, American Psychiatric Press, 1982
genetic aspects, and prognosis. Am J Psychother 3 1:345-365, 60. Cole JO, Salomon M, Gunderson JG, et al: Drug therapy in
1977 borderline patients. Compr Psychiatry 25:249-262, 1984
42. Andrulonis PA, Glueck BL, Stroebel CF, et al: Organic brain 61. Brinkley JR, Beitman BD, Friedel RO: Low dose neuroleptic
dysfunction and the borderline syndrome. Psychiatr Clin North regimes in the treatment of borderline patients. Arch Gen
Am 4:61-66, 1981 Psychiatry 36:319-326, 1979
43. Soloff PH, Millward JW: Psychiatric disorders in the families of 62. Leone NF: Response of borderline patients to loxapine and
borderline patients. Arch Gen Psychiatry 40:37-44, 1983 chlorpromazine. J Clin Psychiatry 43:148-150, 1982
44. Masterson JF: Psychotherapy of the Borderline Adult. New 63. Liebowitz MR, Klein DG: Interrelationship of hysteroid dys-
York, Brunner/Mazel, 1976 phoria and borderline personality disorder. Psychiatr Clin
45. Buie DH, Adler G: Definitive treatment of the borderline North Am 4:67-87, 1981
personality. mt J Psychoanal Psychother 9:51-87, 1982 64. Gershon ES, Bunney WE, Leckman JF, et al: The inheritance of
46. Gunderson JG: Borderline Personality Disorder. Washington, affective disorders: a review of data and hypotheses. Behav
DC, American Psychiatric Press, 1984 Genet 6:227-261, 1976

288 Am J Psychiatry 142:3, March 1985