The term atypical depression was coined in the 1950s to describe a group of patients particularly responsive to monoamine oxidase inhibitors (MAOIs), in contrast to more typical patients with endogenous or melancholic depression, who responded to electroconvulsive therapy (ECT) and tricyclic antidepressants (TCAs) [14, 15]. Once psychiatric researchers identified this divergence in treatment response, they began to retrospectively characterize clinical features which separated these patients. Central to the descriptions of the patients with this variant of atypical depression was the prominence of anxiety, with associated features of severe fatigue, weight gain, evening worsening, somatic preoccupation, and premenstrual tension [14, 15]. Depression was of “secondary importance” and characterized as “reactive” and “exogenous” [16]. Both the depression and the personalities of these atypical patients were described as hysterical [15, 17]. In its earliest descriptions, atypical depression seemed to characterize patients who did not respond to typical antidepressant treatments because of the prominence of anxiety and personality features.

Further elaboration of the personality styles associated with atypical depression occurred in the 1960s to 1980s. The term “hysteroid dysphoria” was used to signify a type of depressive dysphoria and mood reactivity seen in “histrionic” women with MAOI-responsive symptoms [18–20]. The cardinal feature of hysteroid dysphoria was the sensitivity of mood and exaggerated behavioral reactivity to romantic relationships. At the loss of romantic relationships, these patients would demonstrate anger, hostility, dysphoria, and at times suicidality. In this rejected and dysphoric state, these patients exhibited a tendency to oversleep and overeat, particularly craving chocolate and sweets. Conversely, in responses to positive romantic developments, these patients would present as euphoric or giddy [18]. The clinical phenomenology of hysteroid dysphoria overlaps significantly with that of both atypical depression and BPD [21], but it has failed to be validated as a discrete diagnostic entity [22].

As the field of psychiatry moved away from psychoanalytic conceptualizations emphasizing phenomena such as hysteria, it began to focus heavily on descriptive approaches to defining psychiatric illnesses. During this time, the definition of atypically depressed patients became more focused on distinct symptoms rather than on personality. The diagnostic criteria later adopted in DSM-IV were formulated by the Columbia University mood disorders research group of Klein and collaborators to include: (1) the presence of mood reactivity (i.e., positive mood changes from positive life events) and (2) two or more of the following four symptoms: interpersonal rejection sensitivity, leaden paralysis, significant weight gain or overeating, and oversleeping [23, 24]. This revision of the diagnostic framework for atypical depression moved away from emphasizing anxiety and characterologic features connecting atypical depression to Cluster B disorders, instead focusing on mood reactivity as its cardinal feature.

As this conceptualization of atypical depression has been put to empirical test since its inclusion in the DSM-IV, its status as a valid mood disorder subtype has remained controversial [17]. Studies testing its diagnostic validity and coherence have proposed further revision to these criteria. Studies have challenged the primacy of mood reactivity, which demonstrates weak association to the other four accessory symptoms [17, 25, 26] indicating that anxiety [27, 28] and rejection sensitivity [29] are more empirically robust candidates as an organizing core feature of this distinctive atypical form of depression. Hyperphagia and hypersomnia have been conceptualized as compensatory or self-regulatory homeostatic responses to the depression to which rejection-sensitive individuals are prone [17, 30].

Parker has developed a conceptualization of atypical depression as a “multiaxial” diagnosis which spans both Axis I symptoms and Axis II personality features [31]. Using a “spectrum model,” Parker argues that temperamental and personality characteristics—which include internalizing/anxious, externalizing/irritable, and volatile/self-focused types—interact with life stress to produce anxious, irritable, and hostile depressive phenotypes. Parker specifies atypical depression as a variant of depression co-occurring with “a personality style of sensitivity to rejection predisposing to a set of dysregulated emotional responses and self-consolatory strategies.” According to Parker, inherent in this rejection sensitive personality are features of “(1) feeling abandoned, (2) feeling unable to rely on other people, (3) feeling rejected, (4) feeling lonely, and (5) crying” [31] which overlap with the interpersonal phenomenology of BPD. Parker’s proposed model of the relationship between temperamental or personality features and different depression phenotypes provides a more coherent framework for understanding the interaction between the personality trait of rejection sensitivity and the acute variation on depressive symptoms represented in the atypical depression subtype.

A competing theory explaining the relationship between temperamental or personality-related factors and mood problems has been advanced by advocates arguing for the expansion of the bipolar spectrum. This theory postulates that cyclothymia is the underlying basis of atypical depression. Perugi et al. [13, 32] proposed that atypical depression may be most accurately located on the “soft” bipolar spectrum and that a cyclothymic diathesis may be the central characteristic that mediates mood lability in personality disorders such as BPD. Despite the failure of empirical investigations to validate mood reactivity as the central characteristic of atypical depression, advocates for the expansion of the bipolar spectrum have continued to presume its centrality to justify a bridge between this atypical variant of depression and bipolar disorder.

The history of the concept of cyclothymia extends to the beginning of the classification of psychiatric diseases [33–36]. Ewald Hecker, a student and close collaborator of Karl Kahlbaum, coined the term in 1898, when he defined cylcothymia as a condition with fluctuations in mood between depressive and excited extremes, lasting as briefly as days in milder subthreshold mood states to longer episodes lasting weeks to months with more severe or full-blown presentations of depression and hypomania. Kahlbaum elaborated on Hecker’s description, delineating the difference between this milder variant of cyclothymia and “circular typical insanity.” According to Kahlbaum, cyclothymia could be considered a “partial disorder of the soul,” with a favorable course even though it could last a lifetime and did not require treatment or hospitalization. In contrast, the more clinically serious form of cyclical insanity involved a course of alternating depressive and manic events, with a marked tendency towards deterioration and confusion [37]. Both Hecker’s and Kahlbaum’s early descriptions of cyclothymia distinguish it as a milder, ambulatory, constitutional feature that can develop into more full-fledged versions of bipolar illness. Kraepelin also positioned cyclothymia as a constitutional state predisposing individuals to frank manic-depressive illness, conceptualizing it on the milder, predisposing side of the bipolar spectrum [33].

Since its introduction into the psychiatric literature, the scope and utility of cyclothymia has remained unclear. In the early twentieth century, its critics argued it was too loosely applied and appeared to be a “wastebasket diagnosis” [34]. The boundaries between the constitutional subthreshold, and full-blown forms of manic-depressive illness remain blurred in the spectrum concept which claims a continuum between cyclothymic temperament, a normal variation on personality, and manic psychosis, a discrete pathological variation on mood. In the mid-twentieth century, Kurt Schneider argued against this spectrum concept of psychiatric disease, rejecting the notion that schizophrenia or manic-depression were rooted in a temperamental or personality-based diathesis [34, 38].

Amid the controversy about defining cyclothymia as either a personality variant predisposing individuals to bipolar disorder or as an episodic mood disorder, cyclothymia was introduced under the affective personalities section in the second edition of the DSM as a personality style alternating between depression, with features of “worry, pessimism, low energy, and a sense of futility,” and elation, with features “ambitious, warmth, enthusiasm, optimism, and high energy” [39]. With the advent of antidepressant medications, a major paradigm shift occurred in understanding affective disorders, emphasizing the distinction between unipolar and bipolar illness over the distinctions between organic endogenous and neurotic reactive mood disorder variants [40, 41]. DSM-II diagnostic criteria were put to empirical test and found to lack reliability, catalyzing a reformulation of more empirically derived diagnostic criteria. The radical shift in therapeutics and diagnostics is reflected in the transition to DSM-III [42], where cyclothymia moved from classification as an affective personality to classification as a mood disorder [34].

Since its reclassification as a subthreshold form of bipolar disorder in the DSM-III, cyclothymia has had little clinical utility as a diagnostic entity [7, 35]. Most patients who present with subthreshold versions of bipolar disorder are commonly diagnosed as bipolar II, bipolar not otherwise specified, or with rapid cycling bipolar disorder. Since conversion into the DSM as a “Bipolar and Related Disorder,” it has remained intact in terms of describing a mood disorder, but has continued to be researched as a personality trait or temperament. Its use continues to vary as (1) a form of bipolar disorder independent of bipolar I and II, (2) a subthreshold and milder version of bipolar disorder, (3) an early variant or forme fruste that eventually evolves into bipolar disorder, and (4) a predisposing factor that influences vulnerability to both disorders [28]. Among the advocates for the concept of bipolar spectrum disorders, Hagop Akiskal argues that cyclothymia is both a lifelong trait and predisposition to bipolar spectrum variants as well as a discrete subsyndromal variant of bipolar disorder [34, 43]. In this dual definition, cyclothymia describes a trait, personality feature, or temperament, as well as a state or disorder, like a mood episode. Similar to atypical depression, cyclothymia has been defined in a way that has crossed diagnostic axes with both mood and personality features.

The term “borderline” was first introduced by Adolph Stern in 1938 [44] to describe a category of patients whose presentations defied classification as either psychotic or neurotic and therefore seemed to occupy a borderline between these major nosological domains. In his characterization of these patients, Stern described narcissism, psychic bleeding (i.e., low perseverance), inordinate hypersensitivity, psychic and body rigidity, constitutional feeling of inferiority, masochism, organic insecurity, projective mechanisms, and difficulties in reality testing. Stern also highlighted the tendency of these patients to have “negative therapeutic reactions,” noting they were “extremely difficult to handle effectively by any psychotherapeutic method” [44]. In this initial description by Stern, borderline patients were distinguished by their poor fit with both prevailing diagnostic categories and existing therapeutic approaches. Parallel to the development of the atypical depression diagnosis, the discovery of the borderline group of patients occurred in the context of a poor response to usual treatments.

The confusion in defining the boundaries and overlaps between this group of borderline patients and those patients with neurotic and psychotic disorder is reflected in the variety of terms used to signify this clinical phenomenon. At first, borderline personality was defined at its psychotic borders, as a less severe variant of schizophrenia, with distinction as a form of ambulatory schizophrenia [45], pseudopsychopathic schizophrenia [46], or pseudoneurotic schizophrenia [47, 48]. While psychiatrists struggled to clarify the defining features of this borderline group of patients, what was notable about their presentations is that they were “stable in their instability” [49]. Robert Knight critically observed the lack of consistency in the various descriptions of these patients, noting the term borderline had become a “wastebasket” for patients whom psychiatrists were not able to classify as either purely psychotic or purely neurotic [50]. Despite Knight’s criticism of the diagnosis, he noted that the failure to identify the particular needs of this patient group contributed to management conflicts and difficulties on inpatient units which arise out of a failure to appreciate the vulnerability these patients had to regress in unstructured environments.

In the 1960s, Otto Kernberg clarified the concept of borderline personality organization, distinguishing aspects of psychological functioning that characterized this group of patients. Kernberg outlined three key features of patients operating at a borderline level of personality, including (1) failed identity formation or identity diffusion, (2) primitive defenses (i.e., splitting and projective identification), and (3) stress-related lapses in reality testing which defined this level of personality functioning occupying the border between its more disturbed psychotic and more healthy neurotic counterparts [51]. The first formal study of borderline patients was published by Roy Grinker shortly thereafter, establishing an empirically derived criterion set which included (1) anger as the dominant affect, (2) impaired interpersonal relationships, (3) lack of self-identity, and (4) depression [52]. Grinker’s description of the borderline syndrome defined the affective border between borderline and depressed patients by incorporating the characteristic interpersonal and identity dysfunction with an emphasis on anger as the dominant affect (rather than depression). In a synthesis of the literature, Gunderson and Singer proposed a more detailed set of criteria [53] which was later operationalized in a reliable structured interview [54]. The adaptation of Kernberg’s concept of identity diffusion, Gunderson’s criteria for BPD, was adopted into the DSM-III.

Since the inclusion of BPD into the DSM-III, the borderline construct has transformed from a type of personality organization with a psychoanalytic explanation [51] to a disorder [55] which in a medicalized paradigm connects it to specific etiological and therapeutic bases. After the publication of the DSM-III, a robust scientific literature has evolved to validate BPD as a reliable and discrete diagnostic entity, by more stringent standards set by Robins and Guze [56] for all psychiatric diagnoses [55, 57–59]. In the transition of the diagnostic criteria from DSM-III to DSM-IV, the affective instability described in BPD was changed from its characterization as mood lability—with shifts between depression, irritability, and anxiety—to mood reactivity marked by intense episodic dysphoria. This distinction in quality of mood instability in BPD both overlapped with the mood reactive, dysphoric quality of atypical depression and also clarified the distinction from bipolar disorder, where the mood instability was distinguished as labile.

During the last three decades, the stigmatization of the BPD diagnosis has been reduced by a growing empirical literature suggesting high rates of remission and low rates of relapse over 10 years [60, 61] as well as amenability to treatment by a variety of different manualized psychotherapeutic approaches [8, 55]. Furthermore, family and twin studies of BPD demonstrate a level of heritability between that of major depressive and bipolar disorders, suggesting a significant contribution of genetic and environmental factors to its development. These studies also report evidence that a single latent BPD factor organizes the affective, behavioral, cognitive, and interpersonal symptoms within the diagnosis [62, 63]. These findings confirm the integrity and biological basis of the BPD diagnosis.

In the development of the latest edition of the DSM, fierce controversy ensued over a radical plan to overhaul the existing system of personality disorder classification to further dimensionalize these disorders and to consider core personality features in relevance to both normal functioning and psychiatric disorders rather than exclusively in terms of the pathological variants described as personality disorders [64, 65]. The movement to dimensionalize all psychiatric diagnoses was propelled by a recognition of significant co-occurrence among different diagnoses, difficulty defining a valid cutoff between normal and pathological variants, and tendency towards diagnosis of atypical or “not otherwise specified” presentations due to the failure of existing diagnoses to define clinically prevalent presentations [65]. Two main alternatives considered in revising the personality disorders section of the DSM-V proposed dimensional ratings of personality features using the five-factor model [66] versus a prototype model, using the Shedler-Westen Assessment Procedure (SWAP) [67, 68]. In a study testing these alternative procedures for diagnosing personality disorders against existing DSM-IV diagnostic criteria, clinicians demonstrated significant difficulty translating ratings from the FFM and SWAP into DSM diagnoses, especially in cases presenting with comorbidity [69]. Considering the hard-won empirical basis for the existing diagnostic criteria, comparative inadequacy of empirical support for a new system, and evidence of the significant difficulty clinicians would have in utilizing a dimensionalized system, the DSM-IV diagnostic criteria of all the personality disorders were retained without change. However, one significant change which occurred in the larger DSM-V revision was the elimination of the multiaxial system of diagnosis which suggested a false distinction between mental and medical illnesses as well as between Axis I and II disorders.

Despite the lack of adequate consensus and evidence for a more dimensionalized approach to the diagnosis of personality, there are important clinical and empirical benefits to considering broader underlying factors that determine the patterns of comorbidity we observe in these disorders. As noted throughout, the comorbidity of BPD with a variety of other psychiatric diagnoses is common [70], and multivariate methods have been applied to analyze the underlying structures that may determine the co-occurence of different disorders, suggesting that internalizing dimensions contribute to the pattern of comorbidity with unipolar mood and anxiety disorders, while externalizing dimensions contribute to comorbidity with disinhibitory disorders such as substance-related disorders and antisocial personality [71]. This movement to understand broader factors which underlie the complex comorbidity patterns may ultimately clarify the overlaps between mood and personality features. Furthermore, longitudinal studies have clarified distinctions between the more enduring stable affective and interpersonal traits of BPD—such as intolerance of aloneness and dependency—and more acute, reactive, and impulsive features of the illness such as self-destructive and suicidal acts [60]. This finding suggests a division between temperamental traits and stress-reactive symptom states within the BPD diagnosis. More research is needed to understand the core biological and temperamental traits which may predispose individuals to acute manifestations of BPD as well as other comorbidities that span the previously divided Axis I and II disorders.

The core features of atypical depression overlap completely with defining affective and interpersonal core features of BPD (Table 3.2). Individuals with either diagnosis present with mood reactivity, primarily influenced by interpersonal triggers such as rejection or abandonment. In the literature on atypical depression, this qualifier is often referred to as “paradoxical anhedonia,” alluding to the distinction that these atypically depressed states are exogenously (as opposed to endogenously or biologically) determined. While formally required to meet the DSM diagnostic criteria for atypical depression, mood reactivity fails to show significant relationships to the other criteria within the diagnosis [17, 25]. For borderline patients, mood reactivity also refers to the generation of negative affects to stressful or negative environmental triggers or events. These environmental triggers are primarily interpersonal, but can include other types of life stress and typically precede impulsive or self-destructive acts [73, 74]. Similar to the dysphoric states observed in atypical depression, depressive states in BPD are prone to radical shifts if attachment figures previously seen as rejecting are then experienced as accepting [75, 76]. In contrast to the research findings pointing to the lack of relationship among the criteria for atypical depression, factor analytic studies of BPD suggest that the affective, behavioral, and interpersonal symptom sectors are interrelated [77–79]. Both affective dysregulation and interpersonal hypersensitivity have been hypothesized as core organizing features of BPD [80–82].

The criterion of interpersonal rejection sensitivity, describing pattern of anxious hypervigilance and angry reactivity towards real or perceived rejection [83], stands alone in terms of relationally based criterion in all of the mood disorder diagnoses, but is a core feature of a number of disorders (e.g., social anxiety, avoidant personality disorder) including BPD [84]. Rejection sensitivity was first incorporated into the diagnostic definition of atypical depression with two important distinguishing features: (1) a trait-like quality with persistence outside the time frame of active depressive episodes and (2) a degree of relevant functional impairment, described in terms of “stormy relationships.” Notably, this definition of rejection sensitivity was introduced into the atypical depression diagnostic criteria by a group of mood researchers at Columbia University, where the rejection sensitivity concept was later operationalized and studied as a psychological concept separate from a diagnostic entity [83]. In its development, rejection sensitivity was associated with a pattern of behavioral reactivity, such as leaving work early and substance use [6].

In BPD, interpersonal rejection may be a prototypical trigger to both affective (anxious and angry) and behavioral (substance abuse) components of the disorder. The specific sensitivity of individuals with BPD to rejection has been documented both empirically and phenomenologically [84–86]. Some experts have theorized IRS as a core trait of individuals with BPD, advocating for its inclusion into a revised diagnostic description of BPD [87]. While mood reactivity is the required criteria for the diagnosis of atypical depression, more recent research suggests IRS as a more defining clinical feature both atypical depression and BPD [31].

The reversed neurovegetative symptoms (i.e., hypersomnia and hyperphagia) and leaden paralysis, while not diagnostic of BPD, are highly characteristic of behavioral and psychosomatic aspects of individuals with the disorder. Impulsive symptoms in BPD are often conceptualized in terms of solutions to or avoidance of emotional distress. Oversleeping and overeating function as behavioral responses, which serve to distract from or enable avoidance of painful emotional intensity for BPD patients. These symptoms have also been characterized as compensatory and reactive in atypical depression [17]. Leaden paralysis, defined as the tendency to feel heavy, weighed down, and paralyzed for at least 1 h a day at least three times weekly, has been re-characterized within the atypical depression literature as lethargy and fatigue [17]. Recent research suggests it may not be a feature that distinguishes atypically depressed patients from those with other forms of depression. Chronic fatigue has been associated with BPD [4], and both may be connected with beliefs about negative emotions as unacceptable [88].

Very little scientific literature exists to clarify the relationship between these two diagnostic entities, despite the clear overlaps in clinical features. In studying the types of depression in BPD, Soloff and collaborators reported that 16 of 39 inpatients with BPD met criteria for atypical depression (41 %), and 25 of 29 (64.1 %) met criteria for hysteroid depression [89]. While a majority of borderline patients met criteria for atypical variants of depressive disorders, no one form of depression seemed to capture the depressive features in BPD accurately [89]. In a much larger study by Posternack and Zimmerman of 579 psychiatric outpatients with major depression, subjects with atypical depression demonstrated higher ratings on the traits for all DSM-IV Personality Disorders when compared with patients with non-atypical depression, but only traits scores for histrionic and avoidant personality were higher at a statistically significant level for patients with atypical depression compared to non-atypical depression. Scores on borderline and narcissistic traits were reported as higher in the atypically depressed subjects compared to the non-atypically depressed subjects, but these differences were not significant once Bonferroni correction was applied [25]. Lastly, a study by Perugi and colleagues assessed 107 ambulatory (partial hospital and outpatient) subjects with atypical depression [32], finding a large proportion of these subjects met criteria for various personality disorders including histrionic (25.2 %), narcissistic (18.7 %), obsessive-compulsive (38.3 %), dependent (46.7 %), avoidant (60.7 %), and borderline (37.4 %) personality disorders [32]. Like in the Posternack and Zimmerman study, avoidant personality disorder, not BPD, was most common among atypically depressed patients. Interpersonal rejection sensitivity and separation anxiety are two features that avoidant, dependent, and borderline personality disorders have in common with atypical depression [13].