The Neuropsychopharmacology of Personality Disorders

Emil F. Coccaro and Larry J. Siever

The inclusion for the first time of a chapter on the neuropsychopharmacology of the personality disorders in the "ACNP Generation of Progress" series reflects the increasing appreciation of underlying neurobiologic substraits for these disorders and the value of targeted psychopharmacologic treatment. The personality disorders, located on Axis II in DSM-III-R and DSM-IV, consist of constellations of enduring or persistent maladaptive traits and/or symptoms that are characteristic of the way an individual experiences and interacts with his/her environment. In contrast with the Axis I disorders, which are primarily symptom-oriented and wax and wane in severity (often in episodic fashion), the Axis II personality disorders are conceived to be characteristic of an individual throughout his lifetime. This conceptual distinction between the Axis I and II disorders, while heuristically useful, is becoming increasingly blurred because studies into the neuropsychopharmacology of personality disorder suggest that enduring traits reflect underlying biologic variations that may be amenable to alteration with psychopharmacologic treatment. While some have argued that these considerations argue for the abolition of Axis II, the separation of these disorders on Axis II has highlighted the fact that traits considered to be part of an individual's stable personality can be substantially impacted by psychopharmacologic treatment.

While the DSM-III and III-R represented a significant advance over DSM-II in the identification of specific categories of personality disorder, many of these categories were found to be overlapping and have yet to be validated by external validators. By the advent of DSM-IV, validating antecedents and correlates in both biologic and psychosocial arenas have been found for some of the personality disorders, particularly schizotypal and borderline personality disorder. The total number of these disorders has been reduced from 11 to 10, and criteria have been sharpened with no major changes in the constructs (see Introduction to Clical Neuropsychopharmacology).

Because the categories are highly overlapping and describe a range of interpersonal, attitudinal, and more fundamental psychologic traits, neurobiologic and psychopharmacologic approaches to the personality disorders can also be organized around "dimensional" or "cluster" approaches to the personality disorders that may cross categorical boundaries. DSM-III and DSM-III-R was organized for heuristic purposes into clusters: the "odd" cluster, the "dramatic" cluster, and the "anxious" cluster. These clusters also may be mapped into dimensions of cognitive disorganization for the "odd" cluster, impulsivity and affective instability for the "dramatic" cluster, and anxiety for the "anxious" cluster (53). Cognitive organization refers to the capacity of an individual to attend to and select relevant information from the environment, organize it in relation to past experience, and formulate appropriate strategies to interact with the environment. Impairment in this dimension ranges from the chronic psychosis of schizophrenia, where perception of reality is markedly distorted, to the psychotic-like symptoms of schizotypal personality disorder and perhaps even to more subtle deficits in interpretation of social cues. Impulsivity, which is characteristic of borderline, histrionic, and antisocial personality disorders, may be defined as a lowered threshold for motoric action, particularly aggressive behavior, in response to environmental stimuli. Individuals with a tendency toward impulsivity tend to externalize their problems and overreact to environmental events. Affective-related symptoms are another dimension prominent in the "dramatic" cluster personality disorders. Affective instability is a criterion of borderline personality disorder and may be observed in histrionic and narcissistic personality disorders as well. Individuals with affective instability are characterized by rapidly occurring shifts in affect, changing from anger to disappointment to excitement in a matter of hours or minutes. These affective shifts are exquisitely sensitive to shifts in the environment, such as separation or frustration. Finally, anxiety thresholds vary between individuals. Individuals with high anxiety have a greater readiness to anticipate punishment or aversive consequences of their behavior and often show concomitant autonomic arousal associated with their fearfulness. They may inhibit a variety of behaviors they perceive as potentially assertive or competitive, fearing the consequences of their behavior. While a number of psychometric dimensional schema exist for the personality disorders, emerging evidence supports a biologic underpinning to these clinically based dimensions, which also may constitute target symptoms for psychopharmacologic intervention.

In this review, the available evidence regarding the biology of the personality disorders, particularly because they have implications for psychopharmacologic treatment, will be presented in relation to each of these target symptom dimensions, which may overlap with specific disorders. Most pharmacologic studies have focused on patients with specific personality disorders, such as borderline or schizotypal personality disorder, and have evaluated changes in target domains of psychosis, affect, anxiety, or impulsivity. Thus, both syndromal and dimensional considerations will be discussed. It should be noted, however, that there are few placebo-controlled pharmacologic trials in patients with well-characterized personality disorders. Therefore, the conclusions presented in this chapter must be considered tentative at this time. Accordingly, we will highlight trends in the data, both biologic and pharmacologic, and present possible future directions for research in the clinical psychopharmacologic management of patients with personality disorders.

The schizophrenia-related personality disorders of DSM-III and DSM-IV include schizotypal, paranoid, and schizoid personality disorders. Of these, schizotypal personality disorder is the best characterized and the most severe of the schizophrenia-related personality disorders, and it is the most closely related to schizophrenia biologically, phenomenologically, and genetically (53). In addition to psychotic-like symptoms, patients with schizotypal personality disorder also manifest social detachment and other deficit-related symptoms. Initial hypotheses centered around defining similarities between schizotypal personality disorder patients and schizophrenic patients in these domains, whereas more recent studies have pursued more specific correlates (particularly in the psychotic-like and deficit-like symptom complexes) of underlying psychopathologic processes in schizotypal personality disorder to better understand the fundamental pathophysiologic processes of the schizophrenia-related disorders (53).

Psychotic-like symptoms in these personality disorders include magical thinking, ideas of reference, and perceptual distortions and are some of the most discriminating criteria for clinically identified schizotypal personality disorder patients. Because psychotic symptoms in the schizophrenic disorders have been linked to excessive dopaminergic activity, most studies have explored this hypothesis in order to understand the biologic underpinnings of psychosis proneness in these individuals. The "dopamine hypothesis" of schizophrenia was stimulated by (a) the observation that antipsychotic medications had potent dopamine antagonist properties that correlate with their therapeutic efficacy (34) and (b) the observation of psychotogenic effects of dopamine-releasing agents such as amphetamine when administered over long periods of time. While studies measuring the dopamine metabolite homovanillic acid (HVA) in cerebrospinal fluid (CSF) in schizophrenic patients and comparing them with controls have been inconsistent (69), there are suggestions of increases in some paranoid subtypes and decreases associated with deficit-like symptoms (52). However, current hypotheses regarding the dopaminergic system in schizophrenia might suggest hypodopaminergia in frontal cortical areas and hyperdopaminergia in subcortical areas. However, indices of dopaminergic activity such as CSF and plasma HVA depend on contributions from multiple brain regions and the periphery, making interpretations of these measures problematic.

The study of schizotypal patients affords an opportunity to disentangle these two processes. Both CSF and plasma HVA levels are greater in clinically selected schizotypal patients than in normal and other personality disorder controls, specifically associated with the psychotic-like symptoms of schizotypal personality disorder (51, 52). In schizotypal relatives of schizophrenic patients, characterized more by deficit-like symptoms, plasma HVA is reduced compared to relatives of nonschizotypal relatives (Amin et al., 1993, unpublished data).

These considerations imply that the psychotic-like symptoms of the schizophrenia-related personality disorders may be ameliorated by neuroleptics and worsened by psychostimulants such as amphetamine. In one study of borderline/schizotypal patients, amphetamine induced psychotic-like symptoms, particularly in those patients with a schizotypal personality disorder diagnosis (48). In clinical psychopharmacologic trials, neuroleptic treatment has generally been associated with global improvement in patients with borderline and/or schizotypal personality disorder (7). In two relatively large placebo-controlled trials in patients with borderline and/or schizotypal personality disorder, psychotic-like symptoms (as well as symptoms of anxiety) were reduced by treatment with a neuroleptic (thiothixene in ref. 17, haloperidol in ref. 57). The generalizability of the data from one trial may be limited, however, to personality-disordered patients with histories of brief transient psychotic-like symptoms prior to the start of the trial. A smaller study involving only females with severe borderline personality disorder found only modest efficacy for the neuroleptic (trifluroperizine) over placebo. While these data are in general agreement with those of several other studies involving neuroleptic treatment in personality-disordered patients, the most recent study found no efficacy for the neuroleptic (haloperidol) on psychotic-like symptoms in borderline and/or schizotypal personality-disordered patients (56). The authors noted, however, that the patients in their previous study, where haloperidol had been efficacious in treating psychotic-like symptoms, had significantly higher ratings of "psychoticism," "schizotypal symptom severity," and "global impairment" than in the more recent study. In conjunction with the finding that "severity of schizotypal symptoms" was a favorable predictor of response to thiothixine (17), these results suggest that low-dose neuroleptic treatment may be best indicated for moderately to severely impaired patients with prominent histories of psychotic-like schizotypal symptoms.

The social detachment in interpersonal isolation of the schizophrenia spectrum personality disorders may be rooted in biologic processes impairing cortical processing of complex interpersonal cues as well as deficits in attachment behavior. Interpersonal relationships depend on selecting appropriate information and cues from other people and synchronizing one's responses in a reciprocal interaction. These interactional patterns might be set down in infancy and may be disrupted when there is neurologic immaturity on the part of the developing infant (53). Offspring of schizophrenic patients show pandevelopmental immaturities, raising the possibility that the diathesis to the schizophrenic disorders may impair the development of mutually satisfying interactions (53). A variety of studies implicate deficits in information processing, neuropsychological, and psychophysiologic tasks in the schizophrenia-related personality disorders, as in schizophrenia. These are often particularly associated with the deficit-like symptoms of the schizophrenic spectrum personality disorders and have been hypothesized to be related to alterations in brain structure (53a). The cognitive impairment seems to be associated with decreased indices of dopaminergic activity, consistent with the notion that adequate dopaminergic tone, particularly at D1 receptors in frontal cortical areas, may be necessary for the integrity of working memory and other executive cognitive functions. Preliminary evidence supporting this hypothesis comes from psychophysiologic, neuropsychologic, and imaging studies (53a).

Impaired eye movements tracking a smoothly moving target is one of the most consistent findings in the schizophrenia-related disorders. Abnormalities in smooth pursuit tracking are seen in schizophrenic patients and in schizotypal personality-disordered patients. Eye-tracking impairment is specifically associated with the "deficit-like" traits of schizotypal personality disorder (53).

In addition, patients with schizotypal personality disorder show abnormalities and other attentional tasks. These include the Continuous Performance Task (CPT) and the Backward Masking Task (BMT). The CPT is a test of sustained attention that involves identifying target stimuli from a continually presented array. Poor performance on the CPT has been observed in studies of schizotypal volunteers, patients, and offspring of schizophrenic patients (53a) and is correlated with social detachment in offspring of schizophrenic patients (12). The BMT is a visual information processing task which has also been reported to be abnormal in patients with schizotypal personality disorders as well as in volunteers selected because of their schizotypal traits (36). Abnormalities in electroencephalographic (EEG) responses to an unexpected stimuli provide a measure of brain responses to an attentional paradigm. Such evoked potential studies suggest alterations in a positive wave at 300 msec following a stimulus (P300) in schizotypal volunteers and patients similar to those demonstrated in schizophrenic patients. Abnormalities in galvanic skin orienting response and visual reaction time similar to that observed in schizophrenic patients also suggest altered information processing in schizotypal individuals (53a).

Imaging studies suggest that there may be increased ventricular size in the schizophrenia-related personality disorders (5). In one study, lateral ventricles were enlarged (particularly on the left side), and enlargement of the frontal horn and ventricle was associated with impaired performance on the Wisconsin Card Sort Test. In contrast, no such abnormalities have been found in patients with borderline personality disorder (53a). In exploratory analyses, increased ventricular size was associated with reduced concentrations of plasma HVA and deficit-like symptoms (53a), raising the possibilities that frontal cortical impairment may be associated with increased ventricular size and hypodopaminergia in this area. Also consistent with this possibility are magnetic resonance imaging (MRI) studies suggesting reduced frontal size correlated with schizotypal traits in volunteer subjects (44).

Schizotypal personality disorder patients or volunteers, like schizophrenic patients, also demonstrate impaired performance on tests sensitive to prefrontal function, including the Wisconsin Card Sort Test (WCST) (44, 53a). On the other hand, performance on the verbal fluency test and Wechsler Adult Intelligence Scale (WAIS) vocabulary and block design that does not significantly differ from normal controls suggests that cortical impairment is not global and may be more selective for brain circuits including frontal and perhaps temporal regions.

Poor performance on the WCST as reflected in increased preseverative errors, as well as poor performance on the Trails B Test, tends to be associated with reduced concentrations of plasma HVA, the primary metabolite of dopamine in the brain (53a). Furthermore, increased ventricular size tends also to be associated with decreased concentrations of plasma HVA. In personality disorder patients, plasma HVA shows a trend to be negatively related to the deficit-like symptoms; that is, reduced concentrations of plasma HVA are related to increased social withdrawal and constricted affect (53a). A significant correlation between deficit-like symptoms and plasma HVA has been reported in relatives of schizophrenic patients with schizotypal traits (Amin et al., 1993, unpublished data). These results contrast with clinical studies suggesting that increases in plasma HVA are correlated with the psychotic-like symptoms of schizotypal personality disorder.

Together, these findings suggest that schizotypal personality-disordered patients, particularly those with deficit-like symptoms, are characterized by impairment on a variety of cortical processing tasks, increased ventricular size, and reduced indices of dopaminergic activity. Because the dopamine system is implicated in working memory via D1 receptors in frontal cortex, it is tempting to speculate that the cortical dysfunction (in areas such as prefrontal cortex) associated with reduced dopaminergic function may contribute to the "core" deficit-related psychopathology of the schizophrenia spectrum. Other factors must modify this core diathesis in the direction of chronic psychosis for true schizophrenia to emerge. Schizotypal patients with prominent deficit-like symptoms may, however, represent the most common expression of a genetic susceptibility to a neurodevelopmental lesion which results in cortical malfunction.

If the theory discussed above is correct, deficits in cortical function and associated social deficits might be improved with administration of agents that enhance dopaminergic activity. Preliminary data from our laboratory suggests that amphetamine may improve cognitive performance on tests sensitive to prefrontal function (such as the WCST) in schizotypal patients (53a). Therapeutic trials with dopamine reuptake inhibitors such as mazindole, psychostimulants, L-DOPA, or monoamine oxidase (MAO) inhibitors might be warranted. Also intriguing is the possibility that selective D1 agonists, when they become clinically available, might more selectively enhance cognitive function in such patients. If, as correlational analyses hint, social deficits are also related to these underlying cognitive impairments, these strategies may improve the interpersonal functioning of the withdrawn, schizophrenia spectrum personality-disordered patients.

Impulsivity is a defining feature of borderline personality disorder and is prominent as well in antisocial personality disorder and, to a lesser degree, in histrionic and narcissistic personality disorders. Impulsive personality disorder patients are likely to act without reflecting, particularly when it comes to the expression of aggression. They are more easily irritated and are more likely to engage in assaultive behavior, substance abuse, self-damaging acts, and promiscuity. Their impulsive traits may account for the instability of relationships in these patients and their tendencies toward dramatic presentations of themselves. In this cluster of personality disorders, the impulsivity may be differently expressed in the various disorders. For example, in the borderline personality disordered patient, impulsivity is coupled with affective instability such that the patient often reacts with impulsive or aggressive action to the dysphoria engendered by a loss or separation. Such patients frequently present as depressed or feeling abandoned following the dissolution of a relationship. In contrast, antisocial personality disordered patients are much less likely to have affective instability, and their antisocial and aggressive behaviors are fairly persistent and more likely to result in their being seen in a forensic setting rather than in a psychiatric clinic.

Impulsivity as a personality trait appears to be partially heritable. Studies of twins reared apart suggest that impulsivity in healthy, nonpsychiatric populations may be heritable (8). Twin studies of patients with borderline personality disorder also suggest that impulsivity may be partially inherited, although the diagnosis of borderline personality disorder itself is not (66). While individuals with borderline personality disorder may aggregate in families, the core features of impulsivity and affective instability appear to independently aggregate in relatives of borderline patients and may thus combine to provide the susceptibility to borderline personality disorder (55). The presence of a heritable substrait for impulsivity raises the possibility that biologic correlates of impulsivity might be identified.

Neuromodulators that play a role in stimulating and inhibiting external behavior are likely candidates for biologic systems that underlie impulsivity and/or aggression. The serotonergic system, which serves as a behavioral inhibitory system, has been increasingly implicated in the biology of these behavioral traits. In addition, impulsivity and/or aggression in personality-disordered patients may be associated with the presence of epileptiform disorders, attentional disorders, and elevated levels of circulating testosterone and/or endorphins.

Abnormalities of the serotonin are perhaps the most well-documented findings in relation to impulsive aggression and personality disorder patients. Studies in the rodent suggest that lesions of serotonergic neurons result in disinhibited aggression (6). Agents that enhance serotonergic activity can reverse this aggressive behavior as well as inhibit spontaneous or induced aggression. In primates, individual differences in serotonin activity as indexed by concentrations of CSF 5-hydroxyindoleacetic acid (5-HIAA) appear to be heritable and associated with aggressive, dominant behavior (20!popup(ch148ref20)).

Reductions in indices of central 5-HT function are also seen in humans with impulsive aggression, whether directed towards the self (e.g., suicide attempt) or against others (3). Reduced indices of presynaptic serotonergic activity including CSF concentrations of 5-HIAA have been reported in depressed patients who have made suicide attempts or who have engaged in parasuicidal behavior (9). Furthermore, serotonin and its metabolites are decreased in postmortem studies of the brains of suicide victims regardless of diagnosis (9).

Neuroendocrine challenge studies also suggest that central serotonergic activity is reduced in personality disorder patients with impulsive aggression. Prolactin responses to fenfluramine (both d,l- and d-stereoisomer forms) are blunted in personality disorder patients, specifically borderline (11) and antisocial (39) personality disorder patients. In three studies, a negative correlation has been demonstrated between the prolactin response to d,lfenfluramine (11), m-CPP (38), and buspirone (10) and inventories of irritability and aggression in patients with personality disorder. These studies provide support for the hypothesis that decreased serotonin activity is associated with aggression and impulsive behavior in patients with personality disorders. They raise the possibility, reviewed later, that selective serotonin reuptake inhibitors might ameliorate impulsive aggression in such patients.

In contrast to serotonin, the noradrenergic system may play a facilitatory role in promoting impulsivity and aggression. In preclinical studies, the noradrenergic system, with nerve cell bodies located in the locus coeruleus, plays a major role in the regulation of arousal and responsiveness to the environment. While increased activity of the locus coeruleus is associated with reactivity to novel and particularly threatening stimuli (2), decreased locus coeruleus activity has been documented during self-restitutive or vegetative activity such as eating, self-grooming, and sleeping. Heightened noradrenergic activity may be associated with increased irritable aggression (62), whereas a social withdrawal, such as is observed in separation of an infant primate from its mother, seems to be associated with decreased noradrenergic activity (35). Heightened noradrenergic activity may then be related to enhanced sensitivity or reactivity to stimuli in the environment, whereas reductions in noradrenergic activity appear to be more associated with withdrawal from the environment and restitutive functions.

These clinical observations are consistent with studies of noradrenergic abnormalities in the clinical arena. For example, reductions in the responsiveness of the noradrenergic system as indicated by responses to noradrenergic challenges such as clonidine are observed in patients with major depressive disorder, particularly those with endogenous depression (50). A hyporeactive noradrenergic system may thus be related to the disengagement from the environment observed in the patient with autonomous or endogenous depression, particularly with symptoms of withdrawal, psychomotor disturbance, and reduced concentration.

Conversely, increased engagement and reactivity to the environment seems to be associated with enhanced noradrenergic activity in clinical studies. For example, gamblers evidence increased arousal and enhanced noradrenergic activity associated with risk-taking, boredom, susceptibility, and sensation-seeking (47). Similarly, an augmented growth-hormone response to clonidine is positively correlated with scores on the irritability subscale of the Buss–Durkee Hostility Inventory (10). In a larger overlapping sample of personality-disordered patients, presynaptic measures of noradrenergic activity such as plasma norepinephrine were positively correlated with risk-taking and total scores on the Barratt Impulsivity Scale (BIS), and growth-hormone responses to clonidine were associated with irritability and verbal hostility on the BDHI (67). Thus, a heightened responsiveness of the noradrenergic system might contribute to the increased sensitivity and reactivity of personality-disordered patients with impulsivity and affective instability such as may be observed in those with "dramatic cluster" diagnoses. Agents that stabilize or reduce noradrenergic activity might be expected to partially improve irritability and reactivity in these patients.

An etiologic role for epileptiform activity in the limbic system in patients with personality disorder is suggested by some preliminary data related to (a) the presence of certain EEG abnormalities in patients with borderline personality disorder and (b) the response of dyscontrol behaviors to anticonvulsant medications (14; see below). Attention-deficit hyperactivity disorders (ADHDs) are associated with impulsivity and may persist into adulthood. The decrease in impulsivity and irritability in adults with ADHD in response to treatment with stimulants (70) suggests that some irritable impulsive patients may have an underlying adult ADHD syndrome. Finally, there are limited data which suggest that elevated testosterone (free, not total) and endorphin levels may play a contributing role in aggressive (1) and self-injurious behaviors (27), respectively.

Our current understanding of these traits and behaviors suggests a number of potential treatment strategies. Evidence of a reduction in central 5-hydroxytryptamine (5-HT) neurotransmission suggests that impulsive aggressive behavior may be treated with agents which enhance 5-HT in the brain. Open-label clinical trials with fluoxetine in patients with a primary Axis II diagnosis of borderline personality disorder (7) provide evidence that these agents may be effective in reducing impulsive aggression. Effective doses in studies to date have ranged from 20 to 80 mg p.o. q.d., and positive effects have been reported to occur within the first week or within the first month of treatment (8) studies. Unfortunately, none of these studies were placebo-controlled, and the clinical presentations of the patients were varied. However, because impulsive aggressive behavior is intermittent and sensitive to environmental provocation, placebo-controlled trials of adequate duration and careful behavioral monitoring are required. Future studies need to assess the effect of 5-HT uptake inhibitors in patients who do not have comorbid disorders that are also responsive to these medications (e.g., major depression, obsessive–compulsive disorder). It is currently difficult to assess whether gains in impulsive aggressive behavior associated with treatment with fluoxetine have been due to a primary effect of the drug or occurred through a more primary effect on comorbid diagnoses such as depression or obsessive–compulsive disorder. Studies with some of the newer selective 5-HT uptake inhibitors have yet to be reported.

Other agents with putative anti-aggressive properties include lithium, beta-adrenergic antagonists, carbamazepine (and other anticonvulsants), neuroleptics, stimulants, and opiate antagonists. While not all of these agents have been studied specifically in personality-disordered patients, specifically, all have demonstrated some efficacy in disorders associated with impulsive aggressive behavior.

As of this date, lithium carbonate is the only agent which has been shown in a blind placebo-controlled study to diminish the frequency of impulsive aggressive behaviors in individuals with probable antisocial personality disorder. In a landmark study, a 3-month treatment trial of lithium carbonate was associated with a cessation of impulsive aggressive behavior in a group of prison inmates (49). Blind crossover to placebo, at the end of 3 months of treatment, resulted in a return to the baseline frequency of these behaviors. While these individuals would most likely be diagnosed with antisocial personality disorder, only acts of impulsive aggressive behavior were affected by the lithium treatment. These results suggest that other aspects of antisocial personality disorder related to "conning" or "disregard for the rights of others" may not be affected by lithium. The mechanism of lithium carbonate's anti-aggressive action has been proposed to be secondary to its putative ability to increase 5-HT neurotransmission. However, lithium treatment enhanced prolactin responses to intravenous tryptophan, a measure of serotonin responsiveness, only at 1 week in normal controls and not at all in depressed patients, raising questions regarding serotonergic mechanisms (42). It is also possible that lithium's effect of dampening catecholaminergic function (4) plays a role in its anti-aggressive properties observed in antisocial personality-disordered patients.

Beta-adrenergic antagonists (e.g., propranolol, pindolol) have also been shown to have anti-aggressive properties in patients with a variety of diagnoses (although no studies have been reported in patients with primary diagnoses of personality disorder). In general, high doses of these agents are necessary to produce an anti-aggressive effect. The mechanism of action for this effect is also unclear, although antagonism at beta-noradrenergic sites and possible agonism (especially at the higher doses usually employed) at 5-HT1 receptors raise the possibility that these agents may work through both noradrenergic and serotonergic mechanisms. It is also possible, however, that these agents also work through a peripheral mechanism, either in concert with a central mechanism or exclusive of one. Consistent with this possibility, nadolol, a beta-adrenergic antagonist which does not cross the blood–brain barrier, has also been reported to have anti-aggressive efficacy in violent individuals (73).

Carbamazepine treatment has been associated with a striking, and significant, reduction in the severity of episodic dyscontrol in a study of a small group of well-diagnosed female patients with DSM-III borderline personality disorder (14). In this sample, episodic dyscontrol (i.e., worst episode) was rated as moderate or severe in only 10% of cases, whereas placebo treatment was associated with episodic dyscontrol of moderate or severe intensity in 60% of cases. In addition to carbamazepine, diphenylhydantoin has been reported to decrease "anger," "irritability," "impatience," and "anxiety" in neurotic (DSM-II diagnosed) psychiatric outpatients (60). While it is not known how many of these patients would meet DSM-III criteria for a personality disorder, it is noteworthy that most patients entered into this trial were characterized by prominent histories of hostility and/or anxiety. The mechanism of action for the anti-aggressive effect of these anticonvulsive agents is unknown, although it may involve stabilization of limbic neuronal discharges. Abnormal EEG patterns, similar to those seen in patients with partial complex seizures, have been reported for a small proportion of borderline personality-disordered patients in some, but not all (13), studies. It is possible that limbic excitability may underlie both the etiology and treatment responsiveness of some borderline personality-disordered patients to anticonvulsants. It is also possible that these anticonvulsive agents work partially through the central 5-HT system. Enhancement of the prolactin responses to intravenous tryptophan challenge during carbamazepine treatment has been reported, suggesting that carbamazepine may enhance 5-HT activity in humans (15).

The efficacy of neuroleptics in the treatment of impulsive aggressive behavior is likely to be associated with its effect on dopamine receptors. Low-dose high-potency neuroleptic agents have been shown to have modest efficacy in several placebo-controlled studies (17, 57) involving the use of neuroleptics in the treatment of patients with personality disorder. In the most recent study of patients with borderline and/or schizotypal personality disorder, haloperidol's main effects were on observed measures of "hostile belligerence" and impulsive aggressive behaviors (56). In a placebo-controlled trial, treatment with the neuroleptic flupenthixol was associated with a significant reduction in suicidal behavior in personality-disordered patients with histories of recurrent suicidal behavior (37). While these patients were not formally diagnosed as personality-disordered, most of these patients would have met DSM-III criteria for borderline personality disorder.

Finally, there are some limited, but noteworthy, data regarding the effect of stimulants and opiate antagonists on impulsivity and aggression. In a placebo-controlled study of adult patients with attention deficit disorder (ADD), the stimulant pemoline was shown to significantly decrease ratings of both "impulsivity" and "hot or explosive temper" (70). Reductions in these symptoms occurred in the context of similar improvement in ratings of "attention difficulties" and "hyperactivity," but not of "affective liability." While patients with "borderline" or schizotypal personality disorders were excluded from study, at least 27% met Research Diagnostic Criteria for antisocial personality disorder (definite or probable). In addition, all patients had at least two or three of the following six symptoms (i.e., in addition to either motor hyperactivity or attentional deficits persisting from childhood): impulsivity, hot or explosive temper, affective liability, impaired interpersonal relationships, stress intolerance, and inability to complete tasks. It is noteworthy that the first four of these criteria are found in the DSM-III criteria for borderline personality disorder. Because patients with borderline personality disorder were excluded, this suggests that "impulsivity" and/or "hot or explosive temper" may respond to stimulant medication in the absence of a full personality-disorder diagnosis. These findings were generally replicated in another sample with the stimulant methylphenidate (71). Hence, it is possible that stimulant medication may be beneficial in treating impulsivity and irritability/anger when it occurs in the context of adult ADD. Studies with opiate antagonists in the treatment of self-injurious behavior in a variety of psychiatric patients, mostly with mental retardation, have met with mixed success (27) but may offer another avenue for research for pharmacologic intervention in very difficult patients with borderline personality disorder who cut or otherwise mutilate themselves.

Exacerbation of aggressive behavior may represent an unwanted side effect at some psychopharmacologic agents in personality-disordered patients. The tricyclic antidepressants (e.g., amitriptyline), MAO inhibitors (e.g., tranylcypromine), and alprazolam, all of which may be presented to patients with personality disorder, may increase aggressive behavior. Evidence of potential enhancement of impulsive aggressive behavior during treatment with tricyclic antidepressants in personality-disordered patients was first reported by Soloff et al. (58). Soloff and colleagues noted that borderline personality-disordered patients who had no global improvement during treatment with amitriptyline (i.e., "nonresponders," defined as having a global assessment improvement score after medication below the average for the entire sample) demonstrated increases in ratings of impulsive and aggressive behavior compared to patients who similarly had no global improvement during placebo treatment. This increase in impulsive aggressive behavior occurred in spite of isolated improvements in measures of depression and psychoticism in the amitriptyline "nonresponders" compared to the placebo "nonresponders." These findings suggest a dissociation between antidepressant and anti-aggressive responses to amitriptyline in a subset of patients with borderline personality disorder. Amitriptyline "nonresponders" in this study had significantly higher pretreatment ratings of hostility, psychoticism, and negativism than did amitriptyline "responders" (58), raising the possibility that personality-disordered patients with these characteristics may be at more risk for increased impulsive aggressive behavior during treatment with tricyclic antidepressants. Increases in impulsive and/or aggressive behaviors have previously been reported in depressives during treatment with tricyclic agents (43). While tricyclic antidepressants may enhance serotonergic actively, they can also enhance noradrenergic activity. Given indications of a positive relationship between noradrenergic system function and hostility and/or impulsive aggression, it is possible that tricyclic antidepressants may augment these behaviors by enhancing noradrenergic system function. Increased aggression in patients with borderline personality disorder has also been reported during treatment with MAO inhibitors (14), although this phenomenon is not well-characterized.

An increase in the severity of episodic dyscontrol in borderline personality-disordered patients treated with alprazolam was first noted by Cowdry and Gardner (14) in their placebo-controlled multiple-drug crossover study. The effect was so robust that the treatment arm involving alprazolam was discontinued after the entry of only 12 patients into the trial. This phenomenon is currently understood to represent another example of how benzodiazepine-like agents can disinhibit some individuals (18). Accordingly, caution should probably be taken in prescribing this agent to personality-disordered patients with prominent histories of impulsive aggression.

Depressive symptoms can appear in any patient with a personality disorder. However, affective-related traits are a common part of the diagnostic picture for several of the personality disorders in the dramatic and the anxious cluster. Affective instability characterizes one of the DSM-III criteria for borderline personality disorder, while exaggerated displays of emotion and rapidly shifting emotions characterize two of the DSM-III criteria for histrionic personality disorder. Rejection sensitivity, a trait often associated with depressive symptoms, characterizes two of the DSM-III criteria for dependent personality disorder and one for avoidant personality disorder. The clinical picture of a depressive syndrome in a patient with personality disorder offers a diagnostic challenge that is often difficult to resolve. Specifically, some personality-disordered patients presenting with depressive symptoms appear to have a type of "environmentally sensitive" affective disturbance that is part of their usual personality profile, whereas others appear to have a more sustained and discrete depressive syndrome indistinguishable from major depression.

Familial relationships have been noted between some personality-disordered patients (i.e., those with borderline personality disorder) and major mood disorder. However, recent data suggest that the relationship may be secondary to comorbid major depression in the personality-disordered probands. In recent studies (55) the familial relationship between borderline personality disorder and depression was attributed to the presence of major mood disorder in the proband itself. Family members of borderline personality-disordered patients without histories of depression had the same morbid risk of depression as did other personality-disordered patients without histories of depression. In contrast, family members of patients with borderline personality disorder had an elevated morbid risk of impulsive personality disorder traits and affective personality disorder traits (55). Individuals with impulsive personality disorder traits were characterized as having at least three of the following symptoms chronically: physical fighting with others, not associated with alcohol; nonpremeditated stealing (e.g., shoplifting); problems with drinking or drugs; binge eating; problems with gambling; sexual promiscuity; self-damaging acts (e.g., wrist-slashing etc.); and irrational angry outbursts, or overreaction to minor events, not associated with alcohol. Individuals with affective personality disorder traits were characterized as having either (a) a chronic dysphoria (e.g., depression, anxiety) or (b) fluctuations in mood not associated with highly severe mood disturbances, psychomotor agitation or retardation, psychotic features, or extreme guilt. In addition, these individuals had at least one of the following symptoms chronically: easily disappointed or self-pitying; low self-esteem; pessimistic attitude; and absence of satisfactory intimate relationships (55). These results suggest that a familial relationship exists between borderline personality disorder and personality disorder traits consistent more with the core features of borderline personality disorder (i.e., impulsivity, affective liability) than with major syndromal mood disorder.

Abnormally high rates of nonsuppression of plasma cortisol following dexamethasone has been reported in some, but not all, studies of patients with borderline personality disorder (28. Over all studies, rates of dexamethasone nonsuppression ranged from 9% to 73%. This difference in rates of dexamethasone nonsuppression appeared to be attributable to the presence of a comorbid diagnosis of major depression in most of these studies. In the studies where comorbidity of major depression is low, the rates of dexamethasone nonsuppression are correspondingly low. The same conclusion can be drawn from data from studies utilizing thyrotropin-releasing hormone (TRH) stimulation testing (33) in patients with borderline personality or other personality disorders (24). Two studies have reported elevated rates of blunted thyrotropin-stimulating hormone (TSH) in small samples of patients with DSM-III borderline personality disorder. Blunted TSH responses to TRH were 38–47% in borderline patients compared to 0–9% in healthy controls. The comorbidity of major depression or alcoholism (either of which can be associated with a blunted TSH response to TRH) in these samples, however, was between 80% and 100%, making a specific association between blunted TSH responses to TRH and borderline personality disorder unlikely. The most recent study investigating personality disorders of all types found no difference in the rate of TSH blunting or in the absolute magnitude of TSH responses to TRH in personality-disordered patients compared to healthy controls (24). These data suggest that abnormalities in TSH responses to TRH stimulation in patients with personality disorder are attributable to the presence of a comorbid diagnosis of depression or alcoholism and not to the personality disorder itself.

Other strategies designed to find biologic correlates of mood disorder in personality-disordered patients have met with mixed results. Growth-hormone responses to clonidine challenge, an index of central alpha-2 noradrenergic receptor sensitivity, in personality-disordered patients tend to be similar to those seen in healthy volunteers (10). This is in contrast to the reduction in growth-hormone responses to clonidine that are observed in depressed patients in either the acute-depressed or the remittedeuthymic state (54). Moreover, growth-hormone responses to clonidine in personality-disordered patients and healthy volunteers were correlated with irritability, not depression (10). Similarly, prolactin responses to d,l-fenfluramine challenge do not seem to discriminate personality-disordered patients with current (or past) history of depression from those who have never met criteria for depression (11); again, correlations between prolactin responses to fenfluramine challenge were seen with irritability and assaultiveness, but not with depression.

It is possible that mood dysregulation or instability is related to a dysfunction of more than one neurotransmitter system including both norepinephrine or serotonin (53). For example, amphetamine challenge is associated with an improvement in global function (including mood) in a subgroup of patients with borderline personality disorder (48). Moreover, behavioral responses to amphetamine challenge may correlate with trait indices of affective liability (23). Neurochemical events triggered by amphetamine stimulation may more accurately reflect the biologic substrate associated with affective liability and dysregulation than stimulation by more selective neurotransmitter agents. This hypothesis also may explain why treatment of mood dysfunction with tricyclic antidepressants is often not successful in patients with personality disorder.

Finally, it is possible that cholinergic sensitivity may play an important role in the mood dysregulation seen in many patients with personality disorder. An extensive body of evidence suggests that increased cholinergic receptor responsiveness may be associated with major depressive disorder (21). Physostigmine, an acetylcholinesterase inhibitor, and arecoline, a muscarinic agonist, both of which increase the signal at postsynaptic cholinergic receptors, induce a behavioral syndrome resembling depression in animals (21), acute and remitted depressed patients, and normal volunteers (21). Furthermore, these agents appear to have antimanic properties. These results raise the possibility that enhanced cholinergic receptor sensitivity may be associated with a broader sensitivity to dysphoria. In fact, in normal males, dysphoric responses to physostigmine correlate with traits of irritability and emotional liability (16).

Consistent with these studies, studies of rapid eye movement (REM) latency, partially modulated by cholinergic activity, suggest that borderline personality-disordered patients share disturbances in REM regulation with major depressive disorder. Borderline personality-disordered patients have been reported to have both decreased and more variable REM latency when compared to normal controls, and in a preliminary study they demonstrated exaggerated reduction in REM latency in response to a cholinomimetic agent. Preliminary studies suggest that personality-disordered patients with prominent affective instability, particularly borderline personality-disordered patients, have a significantly greater dysphoric response to the acetyl cholinesterase physostigmine than do patients with other personality disorder (59). The extent of the dysphoric response in this study as measured by the Profile of Mood States (POMS) was correlated with baseline affective instability. Therefore, it is possible that enhanced responsiveness of cholinergic receptors may play a role in the susceptibility to affective shifts observed in Cluster B personality disorders, such as borderline or histrionic personality disorder.

Clinical trials designed to specifically treat depression in patients with personality disorders have tested the efficacy of standard psychotropic medications in patients with borderline and/or schizotypal personality disorder. In general, most medications which have been tested demonstrate some efficacy on measures related to state depression. These medications include chlorpromazine and loxapine (29), thiothixine (17), haloperidol (56, 57), amitriptyline (57), imipramine (32)), phenelzine (19, 40), and tranylcypromine (14). In most studies, improvements in state depression in personality-disordered patients have been minimal to modest in studies involving neuroleptics and tricyclic antidepressants and modestly robust in some studies involving MAO inhibitors.

Of the MAO inhibitors, phenelzine has been the most extensively evaluated of the MAO inhibitors. In the pre-DSM-III era, phenelzine was reported to be globally efficacious in treating a subset of patients with pseudoneurotic schizophrenia (19). This diagnostic categorization describes patients characterized by the triad of pananxiety, panphobias, and chaotic sexuality but without psychosis or major depression. With regard to DSM-III Axis II disorders, this picture comes closest to that of patients falling into the Dramatic and/or Anxiety Cluster of personality disorders. A retrospective study comparing imipramine with phenelzine found that the response rate of atypical depressives with comorbid borderline personality disorder treated with phenelzine was nearly three times as great (89% versus 31%) as the response rate achieved with imipramine. In contrast, the efficacy of phenelzine in atypical depressives without borderline personality disorder was equal to that of imipramine (56% versus 57%;) (40). Despite this earlier literature, phenelzine was reported to be remarkably ineffective in treating state depression in the only prospective placebo-controlled study of phenelzine treatment in patients with borderline and/or schizotypal personality disorder (56). In this study, phenelzine was ineffective as an antidepressant in treating both the typical and atypical signs and symptoms of depression. Two reasons for the negative finding in the study of Soloff and colleagues may relate to differences in the patient selection and in the maximal phenelzine dose. First, patients in this study were inpatients with primary entry diagnoses of borderline or schizotypal personality disorder (less than half with a diagnosis of atypical depression), whereas patients in the previous studies were outpatients with a primary entry diagnosis of atypical depression. Second, the phenelzine dose administered in the Soloff study was probably lower than that administered in either of the previous studies. The average phenelzine dose in this study was 60 mg p.o. q.d., whereas that in the previous studies was higher (i.e., nonresponding patients received 90 mg p.o. q.d. of phenelzine after 4 weeks of treatment). In addition, treatment was 6 weeks or longer in the previous studies, and this may also have affected the response rates. The relatively high frequency of adverse events occurring in the personality-disordered patients prevented phenelzine doses from being raised much above 60 mg p.o. q.d.; only 18% of patients received doses as high as 75 mg p.o. q.d. In spite of the differences between the studies, the absence of efficacy for phenelzine in a well-defined sample of hospitalized personality-disordered patients suggests that phenelzine may not be an efficacious treatment in moderately to severely impaired patients with personality disorder.

Unlike phenelzine, tranylcypromine treatment was associated with a significant antidepressant/mood-enhancing effect in a small group of treatment-resistant females with borderline personality disorder (14). In this prospective comparison of tranylcypromine to carbamazepine, it was reported that trifluoperazine, alprazolam, and placebo treatment with tranylcypromine resulted in significantly greater improvement in self-rated assessments of mood (14). While this is yet to be replicated, it is possible that tranylcypromine's mood-enhancing effect was partly due to its amphetamine-like structure and pharmacologic effects. This hypothesis is supported by preliminary data demonstrating that some patients with borderline personality disorder have a marked mood-enhancing response to amphetamine challenge (48).

5-HT uptake inhibitors may also treat the depression seen in patients with personality disorder, specifically borderline personality disorder. However, all published data at this time are based on open-label trials (7). All three reports available suggest that fluoxetine is efficacious in treating state depression as well as other behavioral (e.g., obsessive–compulsive, impulsive) features which can respond to these agents.

Clinical trials designed to specifically treat affective liability have been few. The first placebo-controlled trial in this area compared chlorpromazine to imipramine in a group of patients with "emotionally unstable character disorder" (25). Affective liability in these patients was characterized as rapid autonomous (rather than reactive) shifts in mood. In addition, these patients displayed chronic personality traits such as difficulty with authority, job instability, and problems in interpersonal relations. Both chlorpromazine and imipramine were associated with positive outcomes compared to placebo, although there was clear superiority for chlorpromazine in these patients. Notably, the clinical responses of a subset of imipramine-treated patients was characterized by an increase in anger during the trial (see above discussion regarding tricyclic antidepressants and impulsive aggression). A second placebo-controlled trial with lithium as the experimental agent reported a decrease in the variability of mood in another group of emotionally unstable character-disordered patients (46). A more recent placebo-controlled study of lithium treatment in borderline personality disorder reported a trend for lithium over desipramine in terms of decreasing anger, but did not examine their data in terms of lithium's potential efficacy on mood instability. While limited, these data raise the possibility that other antimanic/anticonvulsant agents (e.g., carbamazepine) could be efficacious in reducing the affective instability of patients with borderline personality disorder, although one uncontrolled study of patients with emotionally unstable character disorder demonstrated no efficacy for the anticonvulsant, diphenylhydantoin (26). Finally, the efficacy of anticholinergic agents in treating affective instability remains to be tested.

The anxiety-related personality disorders include avoidant, dependent, obsessive–compulsive, and passive aggressive (now deleted from DSM-IV). All of these disorders share behavioral traits that may be related to a heightened susceptibility to anxiety and efforts to ward off potential anxiety. For example, patients with avoidant personality disorder have excessive anticipatory anxiety regarding the prospect of future rejection and thus avoid social interactions, while individuals with dependent personality disorder may submit to the wishes of others on whom they depend for fear of rejection or conflict. While there has been almost no systematic investigation into these specific personality disorders, there have been some studies of related or comorbid conditions such as social phobia. For example, the comorbidity of avoidant personality disorder and social phobia may be as high as 90% (72).

Family studies suggest familial transmission of "anxious" personality traits and disorders (45), while twin studies suggest significant heritability of social anxiety (65). Longitudinal studies of children suggest a stable dimension of fearfulness and inhibition (22). Thus, it is likely that at least social anxiety has a genetic basis with some longitudinal stability.

Otherwise, there is limited biologic data regarding social phobia/avoidant personality disorder. Clinical studies of hypothalamic–pituitary function suggest that neither the hypothalamic–adrenal (68) nor the hypothalamic–thyroid axis (63) of these patients differ from that of healthy volunteers. Preliminary results suggest enhanced responsiveness to serotonergic probes but not to adrenergic probes (64), suggesting potential increases of serotonin function associated with enhanced behavioral inhibition, but not noradrenergic system dysfunction. While lactate infusion is a potent stimulus of panic attacks in patients with panic disorder, it produces only inconsistent responses in patients with social phobia (31). One study has suggested decreased dopamine metabolism in social phobic patients (41), but this finding will require replication.

Little has been published which relates to the pharmacologic treatment of patients with anxiety-related personality disorders. Placebo-controlled, double-blind data are published documenting the efficacy of MAO inhibitors in the treatment of social phobia (30). Otherwise, there has been at least one open-label report suggesting that fluoxetine may be beneficial in the treatment of avoidant personality disorder/social phobia (61).

Research into the biologic and neuropsychopharmacologic correlates of behavioral and personality traits has produced many important insights which can aid the psychopharmacologist in the treatment of patients with personality disorders. First, there may be a heritable, biologic component to some selected personality and behavioral traits. Second, the efficacy of various psychotropic agents in treating the more extreme manifestations of these traits may be related to the specific psychopharmacologic properties of the agents (e.g., serotonin-enhancing, catecholamine-dampening properties of lithium) in question (7, 56). Third, symptoms referable to different personality dimensions respond differentially to psychotropic agents within groups of patients with similar diagnoses (17, 56, 57) and within individuals (14). This latter phenomenon reflects the great biologic and clinical heterogeneity present in patients with DSM-III personality disorders.

These insights represent a clear advance in the psychopharmacologic conceptualization of the personality disorders over the past few decades. Despite this, no definitive psychopharmacologic recommendations for specific groups of patients with personality disorders can be made at this time. Many more placebo-controlled clinical trials in well-defined (both clinically and biologically) patients with personality disorder will be required before such recommendations can be made in this area. A first step might be more placebo-controlled studies of agents which have already shown some promise in preliminary placebo-controlled trials (e.g., lithium, carbamazepine, tranylcypromine) of patients with personality disorder.

published 2000