Demystifying borderline personality: critique of the concept and unorthodox reflections on its natural kinship with the bipolar spectrum


I was invited by the Editor to write this guest editorial on behalf of Acta Psychiatrica Scandinavica to discuss the relationship between borderline personality disorder (BPD) and the bipolar spectrum. This is a complex topic of great theoretical, clinical and public health significance. To do justice to it, I felt it necessary to expand my comments beyond an editorial to permit the expression of personal reflections on BPD.

Is BPD a personality disorder?

Borderline personality disorder continues to clutter the psychiatric literature, especially in North America, without anything new emerging. Like the troubled and tempestuous patients it purports to describe, it splits mental health professionals into those who deify its unique nosologic status, and those who prefer to reclassify it under more familiar psychiatric disorders, e.g. affective, anxious, epileptoid, addictive and/or sociopathic. Research on BPD operates within the axis II operational paradigm (1) and instruments developed to measure it. In this respect, Gunderson et al. (2) have rendered an important service to our field by stimulating a plethora of data-based investigations on BPD. Unfortunately, some of the research inspired by this paradigm rediscovers what is already in the BPD criteria. Such circularity is seen, for instance, in the finding that, suicidality touted as the best prospective validation of BPD, is largely limited to those patients defined by the presence of suicidality in its criteria set (3).

A more recently reported predictor of suicidality is affective instability (4). Nonetheless, this team of researchers (5) continues to downplay the relationship of BPD to the disorders of affect. This is all the more surprising given that their own research found that 100% of BPD patients met criteria for one type of depressive disorder or another (6). Such data do not seem to impress them, because the role of affectivity in BPD is trivialized as being ‘atypical’ (7). A related problem is that their usual frame of reference for affective disorder is depression, with little appreciation of the fact that mood lability and cyclothymia are core characteristics of bipolar II (8–10), and that these traits are the markers of a familial-genetic diathesis for bipolar spectrum disorders (11).

That two-thirds of BPD patients remit from BPD in a relatively short period of a few years (see, for instance, Zanarini et al., this issue: 12), does not seem to make these authors question their basic premise of personality disorder. On the contrary, they conclude that the concept of personality disorder as a lifelong disorder must be changed! Paradoxically, one of the major objections of this school to the affective disorder thesis of BPD has always been that the former is episodic, the latter chronic. Are these authors prepared to embrace the emerging paradigm of affective disorders as chronically fluctuating and relapsing disorders (13–15)?

Definitional inadequacies of BPD

The operational construct of BPD has been oversimplified and overstretched. DSM-IV-TR (1) now even permits remission within a year which, as already pointed out, is an obvious contradiction for a personality construct. Such reformulation appears to be a concession to advocacy group criticism – e.g. by the mothers of BPD girls – that the construct had become a ‘certificate of untreatability’ that, in effect, it condemned the patient to poor prognosis. Another weakness is low discriminatory validity. BPD has an unwieldy heterogeneity and overlaps, not only with personality disorders within its own erratic cluster, but also with the odd and anxious clusters. Zanarini et al. (12) report that such overlap pertains largely to the one-third of non-remitting BPD probands. The conclusion one can draw from this is that core BPD (i.e. BPD that does not change over time), rather than being a specific personality disorder, represents an ensemble of personality disorders. Paradoxically, BPD researchers who fiercely defend the uniqueness of BPD, concede that BPD has no pathognomonic features in the 5-factor personality framework (16).

The significant overlap of BPD with affective and addictive disorders is in part due to the fact that BPD criteria (1), rather than being restricted to personality attributes, mix traits, symptoms and behaviors largely in the domain of affective and impulse control (including, among others, boredom, emptiness, ‘object hunger,’‘abandonment depression,’ mercurial moods, reactive dysphoria, angry outbursts, and impulsive suicidality). To postulate anger as the emotional specialty of BPD, is to ignore that irritability figures prominently in the Stem A criteria for both depressive and manic/hypomanic episodes in the DSM system (1). Those who believe that adding identity, dissociative, and micropsychotic disturbances will change the affective nature of a disorder so-defined must have had insufficient clinical exposure to affectively ill patients. The same can be said of those who opine that chaotic relationships and trauma – emotional, physical, sexual or otherwise – are unique to the childhood history of BPD. To think that impulsivity is one of the hallmarks of BPD and not shared with bipolar spectrum disorders is also a flagrant disregard of clinical facts. The continued insistence that BPD does not lie on the border of any psychiatric disorder is in violation of the literature: BPD is actually among the most ‘promiscuous’ of all the mental disorders, both as symptom and as nosologic construct! Borderline circa 2004 is still an adjective in search of a specific noun (17).

Of perhaps greater concern is that the operational construct of BPD may not coincide with what an authority of the stature of Kernberg means by borderline personality organization (18). The latter denotes a vulnerable psychic structure, rather than a formal nosologic entity. As such, it refers to patients with certain specific defensive operations which place their functioning at a ‘stably unstable’ level between the classic neuroses and psychoses.

An illustration of the diagnostic confusion that could result from competing concepts of BPD occurred in 1996 at a Harvard affiliated hospital grand rounds (19), where a female senior house-staff physician with a few months of atypical affective symptoms – but in the absence of unequivocal past history of BPD – was diagnosed as BPD on the basis of her impulsive sexual infidelity when faced with a maturational task (engaged to be married). Gunderson's team (19) thereby inferred that BPD must have been present all along – but latent prior to the maturational challenge to becoming a woman. My comment to such broad use of the BPD label is to paraphrase a well-known manifesto – all young women unite, you have nothing to lose but your borderline diagnoses.

Redefining the affective border

Initially, the borderline concept developed as a dilute form of psychosis, and its main usefulness was to exclude such patients from falling apart on the couch. Stone can be credited for having been the first to make a persuasive argument about the need to shift from borderline as a subschizophrenic to a subaffective disorder (20).

Independently, in a Memphis study (13, 17), we examined the same question in a prospective design. We had started off with the hypothesis that BPD patients would be on the border of schizophrenic disorders, as defined in the framework of the Danish adoption study of schizophrenia (21). Contrary to our expectations, a greater percentage of BPD emerged as dysthymic, cyclothymic, and bipolar II than schizotypal and schizophrenia spectrum. Moreover, in family history, BPD probands were closest to the affective, especially the bipolar, comparison group. This familial-genetic bipolar link was reinforced by antidepressant associated switches into irritable-angry hypomanic and mixed states in 20%. ‘Paradoxical’ disinhibition on antidepressants in borderline patients has also been observed by others (22, 23). The affective – predominantly bipolar spectrum – affinities of BPD revealed in the Memphis study is typically dismissed by borderline researchers on the mistaken assumption that it was conducted in a mood clinic; in reality, these were consecutive BPD patients in the larger mental health clinic. Moreover, to further discredit the affective thesis of BPD, these researchers (7) contrasted the ‘empty depressions’ of BPD with that of ‘guilt depressions’ in classical affective disorder. Thus, unstable, hostile, and labile moods – the unrelenting tension and irritability with superimposed paroxysms of rage – are relegated to the characterologic realm. The thrust of this argument appears based on a misapprehension that only classical depressive disorder is a ‘true’ affective disorder. They do not seem to realize that bipolar II is highly prevalent, is a hybrid of volatile affective temperament and affective states, and highly suicidal (10, 24). To consider BPD as the root cause of much of impulsive suicidal behavior can lead to dangerous complacency and to the self-serving belief that it is the ‘specialty’ of these patients and, therefore, beyond the doctor's professional responsibility.

Given the high suicidality of BPD and bipolar-II patients, it is a matter of urgent public health to intensify research efforts to better understand the reasons for the overlap of the two disorders. The problem with much of the borderline literature is that it uses rigorous measures of BPD deriving from the axis II paradigm (1), but not those of bipolar-II (25). I would therefore summarize those studies that have assessed the relationship between BPD, atypical depression, cyclothymic temperament, and bipolar-II with the requisite methodologic rigor. In the Pisa-San Diego collaboration (26), two-thirds of depressive patients with atypical features simultaneously met criteria for bipolar II and antecedent cyclothymic temperament; both cluster B (borderline-histrionic) and cluster C (sensitive-avoidant) personality disorders were prevalent. Deltito et al. (27), based on the conservative formal definition (1) of bipolar disorder (types I and II), reported a rate of 44% among BPD at Westchester-Cornell; taking the most liberal definition of bipolarity (including bipolar I, bipolar II, pharmacologic-hypomania, cyclothymic temperament, and family history for bipolar disorder), 81% of BPD patients could be considered lying within the bipolar spectrum. In the NIMH Collaborative Study of Depression (8), trait mood lability emerged as the best predictor (86% specificity) of which ‘unipolar’ major depressives would, over a prospective observation period of 11 years, switch to bipolar II; in the predictive statistical model used, this trait measure superceded borderline features. That mood lability is not pathognomonic for BPD, and occurs in bipolar II as well, has been replicated in a cross-sectional Franco-American study (9). Likewise, in a large US study of SADS-derived constructs of cyclothymic and labile personalities, both were highly correlated with each other and associated primarily with intermittent affective disorders (28). A Canadian study (29), too, has shown significant operational overlap between the cyclothymic temperament and BPD. Finally, in a New Zealand study (30), BPD was characterized by high novelty seeking and harm avoidance, and low self-directedness and low cooperativeness – the same profile has been found for the cyclothymic temperament (31).

Given that affective temperaments are absent in the DSM-IV-TR axis II framework (1), clinicians faced with affectively ill patients are forced to make a choice between the dramatic and anxious clusters. Furthermore, structured interviewing tends to misclassify subthreshold affective disturbances in bipolar probands as dramatic cluster (32). No wonder that such subthreshold symptoms dominate the course of major affective disorders (14, 15).

It appears highly implausible that nature would create entirely distinct patterns of affective dysregulation, one for BPD and another one for affective disorder (33). Both depressive and bipolar disorders pursue extremely variable patterns of episodic, fluctuating, protracted and non-remitting courses (13–15). Stated more tersely, the affective universe is dimensional, leading to constant shifting between trait and state. Mood dysregulation occurring over hours or within a day or two [presumably pathognomonic for BPD (1)] is in continuum with bipolar II defined by more sustained hypomania. That is why BPD does not have a unique profile distinct from affective, especially bipolar spectrum disorders as currently conceptualized – whether it pertains to clinical, biological and psychopharmacological data (17, 33, 34).

Borderline as the dysphoric facet of cyclothymia

The bipolar nature of the extreme emotional reactivity in BPD patients is often missed, because dysphoria pervades their short-lived periods of excitement. Recent data from a French national collaborative study (10) has shown that the notion of hypomania as positive (‘sunny’) euphoric traits and behaviors represents just one facet of the bipolar spectrum. There also exists a more pervasive irritable-tempestuous side to bipolarity, the form most likely to arise from a cyclothymic baseline, representing an unstable variant of bipolar II which we have characterized as BP-II1/2.

How is such an emotional diathesis to be understood? Temperamentally (35), they are sensitive, reactive, excitable, labile, hostile, impulsive (and sometimes litigious and violence-prone). These individuals respond to trivial interpersonal stress with emotional avalanches, followed by depressive dips with atypical features. A similar affective syndrome, characterized by anhedonia, emptiness, restlessness, irritability, explosive anger, tension and psychotic anxiety, has been described among female prisoners in the UK (36). In the Pisa-San Diego Collaborative study (37), among atypical major depressive patients, logistic regression revealed that cyclothymic temperament accounted for much of the relationship between atypicality and BPD. The cyclothymic-sensitive disposition seems to represent the common denominator in the complex syndromic pattern of anxiety, mood, and impulse control disorders. Such data, in turn, support the hypothesis that atypical depression, borderline personality, cyclothymia, and bipolar II represent overlapping manifestations of a common underlying diathesis.

Patients with this diathesis are often mislabeled as ‘psychopathic.’ This characterization misses the core emotionality of cyclothymic and BPD patients who can be observed in both probands and in their biologic kin (11, 38). As discussed in greater detail elsewhere (39), interpersonal sensitivity, mood reactivity, and lability are more germane to the origin of both BPD and cyclothymia. It is their innate – and, to some extent, developmentally acquired – intense reactivity to others that creates their turbulent relationships, their maladroit behavior, and marginal social existence – and indeed their entire biography. The tragedy of BPD patients is that their impulsive drive thrusts them into the theater of human interactions, which coupled with their negative affectivity and interpersonal sensitivity, accentuates and thereby validates their sense of being rejected emotionally, maltreated, and abused. The more unfortunate among them do actually get sexually abused by psychopathic or alcoholic kin.

BPD as a counter-transference diagnosis

Borderline personality disorder is at best a confusing concept, and at worst, a counter-transference diagnosis (39, 40), which robs the patient of the opportunity of much needed treatment. Most researchers exclude BPD patients from their study samples in order to avoid complications in the conduct of research. Senior clinicians tend to avoid these patients too, yet have little qualms about assigning them to psychiatric trainees as ‘excellent teaching cases’. Affective reconceptualization may substantially reduce the stigma associated with this diagnosis and therapists’ countertransference (41), because now the patient is likely to be viewed as ill, rather than someone who engages in emotional exploitation of others through anger and suicide threats. The suicide risk in BPD – a potentially fatal consequence of the intense affective dysregulation and the associated impulsivity – should be clinically managed as rigorously as in any patient with serious mood disorder. As switching on antidepressants may aggravate illness course and suicidality (42), atypical antipsychotics, anti-convulsants, and mood stabilizers, including divalproex or lamotrigine (43, 44), should be the mainstay of pharmacotherapy. BPD patients ‘refractory’ to garden-variety pharmacotherapy and psychotherapy show such a sustained remission on lamotrigine (43), including allegations of having been abused, that one is tempted to postulate that their illness is due to ‘lamotrigine deficiency’. The emerging pharmacotherapy of BPD not only can help reduce affective instability and impulsivity, but also often attenuates the disagreeableness of BPD patients, thereby making therapy possible. The judicious use of such medications is a difficult art, but given the suicide risk, one of vital public health priority.

As BPD patients – in view of their negative affectivity – often harbor malevolent object representations of significant others (45), clinicians must not automatically assume that their parents are ‘monsters’ or ‘abusive’. Actually, the parents are as likely to be abused and traumatized by the outrageous behavior of such offspring. Therapists who write off parents as pathogenic do not seem to appreciate that parental guidance is often crucial to BPD patients’ mastery of their intense emotionality and maturational tasks. Indeed, a recent study has shown that parental overinvolvement tends to favor good outcome in BPD (46). Nonetheless, Kurt Schneider's (47) wise admonition about the labile personality, ‘on their bad days, keep out of their way as far as possible,’ should not be forgotten. Such brief ‘emotional vacations’ is good advice for parents and the BPD patients themselves. Among the latter, intermittent distance from their parents should help in the cause of reducing amygdalar over-reactivity (48, 49); hospitalization itself may work through such a mechanism. In Stone's follow-up study (50), the ultimate fate of BPD was positive in some instances where patients were steered to – or discovered by themselves – that they did best in jobs with relatively minimal intimate contact with people. Others may act out their emotionality in the theater or in an acting career. Psychodrama, therefore, may represent a rational therapeutic outlet.

My disagreement with the BPD literature is less about what needs to be done to help these patients [Zanarini, for instance, has conducted credible research in this domain (reviewed in 44)], but under what diagnostic rubric such care should optimally be delivered. Bleuler's original description of borderline schizophrenia as persons with ‘irritable, odd, moody and withdrawn’ (51) was, it would appear in retrospect, an error of overinclusion. The BPD of contemporary psychiatry has been highjacked from the affective domain where it must be returned to make caring possible. No healer can sustain an indefinite therapeutic relationship with a disagreeable person (52); not even a loving mother can cope with such a daughter (or son) with equanimity (53), unless there is hope that such behavior is the expression of an emotional storm, albeit protracted, which lends itself to rational interventions – and that it will pass with as little destruction on its path as possible. Like affective disease, it is BPD which destroys the person – and not vice versa. The alternative is to believe that it is the person who is ‘bad’ or ‘evil’– an untenable position which would divorce psychiatry and psychoanalysis from medicine and align them with either the punitive judicial system or religion.

The ultimate fate of BPD

To recapitulate, the affectivity of BPD patients emphasized by affective disorder specialists is not due to the bias of the clinical setting in which such specialists operate – rather, by the very nature of how it has been defined (1), more often than not BPD is affective spectrum disorder in a youthful patient with ‘primitive’ defenses. Just as an affectively ill individual with ‘mature’ defenses along the lines of intellectualization should not be diagnosed as ‘borderline obsessoid personality disorder,’ one with primitive defenses should not be labeled BPD. Indeed, current data indicate that the latter type of defenses attenuate with maturation, and what remains is the basic dysregulation in affect (54). The claim that such attenuation of course occurs only in BPD (55), but not in bipolar spectrum disorder has no basis in fact: the bipolar spectrum is less prevalent in older persons (56).

This author has been blamed for being an ‘affectophile’ who tends to overdiagnose affective disorders (57). The real problem, however, is that BPD criteria encroach so much upon the phenomenology of affective dysregulation, that in the end these criteria lose their discriminatory power from affective spectrum disorders. The proposal to re-label BPD as ‘emotional dysregulation disorder’ (53) will be a step in the right direction, but I understand it has the resistance of BPD experts, who despite their avowed dissatisfaction with BPD on both terminologic and discriminatory validation standpoints, continue using it (55, 58).

Axis II with its borderline construct was a concession to the neo-kraepelinian reconceptualization of the traditional mental disorders on axis I. A quarter of a century later, if one were to judge BPD on the basis of the validating principles of this paradigm (59), BPD (as currently operationalized on axis II) will in the main emerge as essentially similar to the bipolar spectrum in phenomenology, course, laboratory findings, family history and treatment response (13, 27, 33, 34). The question of diagnostic overlap with other axis I disorders (17) is beyond the scope of this commentary. Suffice it to say that such overlap pertains largely to anxiety, eating, addictive, and impulse control disorders, all of which are well-known ‘comorbid’ features of bipolar II (26).

To paraphrase what Michael Stone told me a quarter of a century earlier (60), BPD as a nosologic construct, like some of the patients it describes, will eventually self-destruct itself. This is certainly true of the DSM-IV construct (61, 62). On the other hand, I am on record (17) for having endorsed the Kernberg psychostructural approach as a more sensible conceptual and clinical framework for BPD patients – who could then be diagnosed on axis I (e.g. mood and anxiety disorders) and a new putative independent axis VI for psychodynamic formulation. Psychoanalytic understanding and descriptive nosology are complimentary to one another – they should not be collapsed into one another. The latter is the fatal mistake of those who gave birth to BPD on axis II.