The study of personality disorders conceived as categorical entities has traditionally been the province of psychoanalytic or behavioral models. However, there is an emerging area of study, spearheaded by the NIMH Research Domain Criteria (RDoC) program (1) that aims to uncover the neurobiological underpinnings of the dimensions that make up mental disorders (2). These new directions in psychiatry neurobiological research converge with new dimensional conceptions of personality disorders (3), and with efforts to identify the neural basis for the traits underlying personality disorders, such as affective dysregulation (affective instability and negative affectivity), disinhibited aggression, anxiety/avoidance, cognitive /perceptual dysregulation, and social detachment/isolation. The extremes of these traits, expressed as symptom dimensions, crystallize to the prototypic personality disorders. For example, borderline personality disorder (BPD) is characterized by affective instability, disinhibition/aggression, and social cognitive/interpersonal impairment. Schizotypal personality disorder (STPD) comprises social isolation/detachment and cognitive/perceptual disorganization, and antisocial personality disorder (ASPD) is characterized by disinhibited aggression and antagonism. The cluster of traits that place an individual at risk for the development of a personality disorder also places them at risk for other psychiatric illnesses, such as depression and anxiety disorders particularly, accounting for the high rate of comorbidity with personality disorders (2).
The study of the neurobiology of personality disorders provides a gateway to understanding relationships between brain and behavior building on individual variation in anxiety threshold, affective regulation, social cognition, and inhibition/aggression and thereby can help us understand the circuitry underlying these critical domains. Several structural and functional brain abnormalities have been identified as the putative biological underpinning of the dimensional traits that underlie personality disorders, particularly the findings of magnetic resonance imaging (MRI), positron emission tomography (PET), and single-photon emission computed tomography (SPECT) neuroimaging studies. For example, the affective dysregulation and disinhibition/aggression of BPD are related to dysfunction in frontolimbic circuits, including limbic structures such as the amygdala and insula as regulated by prefrontal regions including the orbital frontal cortex (OFC). The cognitive impairment of STPD may be related to alterations in prefrontal dopaminergic function, while deficiencies in ventral striatum dopamine systems may be related to detachment/anhedonia. These specific circuits are modulated by neurotransmitters such as serotonin or norepinephrine and neuropeptides, and these modulators may tune the sensitivity and response characteristics of these circuits.
The study of the genetics of personality disorders can identify critical genes that regulate the structure of these circuits and their connectivity as well as the modulators that regulate them. Since personality disorders are thought to evolve from the interaction of genetics and environment throughout the course of development, understanding the neurobiology of these disorders will allow for the characterization of gene-by-environment interactions as well as the mechanisms by which these interactions unfold in the course of development. Environmental influences may also alter the expression of the genome through epigenetic factors, which are also beginning to be investigated in personality disorders. Through identifying genetic variation and its epigenetic regulation, as well as discerning functional aspects of specific neurocircuitry, the molecular mechanisms underlying personality disorders can be characterized.
BPD (4), ASPD (5), and STPD (6) have been the most studied among PDs. In this review, rather than reviewing all of the DSM-IV PDs, we review findings in these three PDs.
BPD is characterized by a pervasive pattern of instability of interpersonal relationships, self-image, and affect, and marked impulsivity beginning by early adulthood and present in a variety of contexts, as indicated by at least five of nine DSM-IV-TR criteria (7). Core dimensions of BPD include affective instability, disinhibition/impulsive aggression, and social cognitive/interpersonal impairment. BPD patients have a high risk of suicide, with a mortality rate of approximately 8%–10% (8).
The neurobiological factors contributing to the genesis of BPD may be conceptualized in relation to core traits of the disorder (affective instability and impulsive aggression).
One of the most consistent findings is that BPD individuals have a decrease in volume in the anterior cingulate gyrus (ACG) compared with healthy comparison subjects (9–12). Other structural abnormalities in BPD include volume reduction in the hippocampus (13–16) and amygdala (16, 17) and surrounding areas of the temporal lobe (18). It is important to note that some, but not all (19), studies have raised the possibility that the smaller volumes in BPD may relate to comorbidity with PTSD or history of serious trauma for hippocampal volume (16, 20, 21), and the effect of comorbid MDD for amygdala volume remains unclear (22).
There is considerable support for the model of reduced medial prefrontal modulation of limbic structures (especially the amygdala), which appear to be hyperactive in patients with BPD, resulting in dysregulation of emotions and aggression (23, 24). Multiple studies have found altered activation of frontal and prefrontal areas involved in emotion regulation in BPD (25, 26). Decreased activity in the ACG and the orbitofrontal cortex (OFC) is correlated with impulsive aggression in BPD (25). Early PET imaging studies showed decreased activity of OFC and ACG in BPD relative to healthy comparison subjects (25–29). A more recent PET study of laboratory-induced aggression using the Point Subtraction Aggression Paradigm found that BPD patients with impulsive aggression showed increased relative glucose metabolic rate in OFC and amygdala in response to provocation, but not in more dorsal brain regions associated with cognitive control of aggression. In contrast, during aggression provocation, healthy individuals showed increased relative glucose metabolic response in dorsal regions of prefrontal cortex, involved in top-down cognitive control of aggression, and, more broadly, of emotion (30).
Most functional magnetic resonance imaging (fMRI) studies using emotional stimuli have shown similar results of decreased prefrontal activation in BPD, with some exceptions and conflicting results (31–33). Most studies in BPD have shown less activation (or more deactivation) of frontal areas involved in top-down control of emotions, including OFC and ACG, in BPD compared with healthy comparison subjects in response to emotional probes (10, 34–37), although some studies showed heightened prefrontal activation in BPD to emotional pictures (10, 33), to unresolved conflicts (38), and pain (31). Poor connectivity between OFC and amygdala has also been reported in association with aggression (39). Meanwhile, many neuroimaging studies suggest limbic abnormalities such as amygdala hyperactivity in BPD (30, 40). In summary, it seems that in BPD patients, prefrontal brain regions that normally put the brakes on expressions of emotions and more broadly of aggression (e.g., the OFC and ACG) may fail to become activated during emotional provocation, while areas of the limbic system (e.g., the amygdala) appear to hyper-respond to emotional probes.
It is important to note that many of the circuits implicated in BPD (including a model of decreased ACG/OFC response with an associated hyper-response of the amygdala) appear to be implicated in other psychiatric disorders, including major depressive disorder (MDD) (41), bipolar disorder (42),and posttraumatic stress disorder (PTSD) (43), indicating potential lack of specificity.
Neuropeptides are a recent area of interest in the biology of BPD. Oxytocin has anxiolytic and prosocial effects (44) and reduces amygdala activation in response to emotional stimuli in healthy individuals (45). The limited empirical data in BPD suggests oxytocin may decrease subjective anxiety but also decreases cooperative behavior (46, 47) and may be associated with anger dyscontrol (Siever et al., unpublished data). One recent imaging study measured μ-opioid receptor binding, by using the µ-opiate ligand, [11C] carfentanil, in patients with BPD during induction of neutral and sad sustained emotional states. They found greater baseline μ-opioid receptor availability in BPD, suggesting a deficit in endogenous circulating opioids, but also suggesting possible enhancement of endogenous opiate availability during sad mood induction (48), reflecting a compensatory response. These findings are consistent with the lower endogenous opioid levels observed in self-injurers. Endogenous opioids may also be related to self-cutting and interpersonal difficulties in BPD (49, 50). We have found that polymorphisms of the μ-opioid receptor may be associated with affective instability and BPD. These associations also seem exacerbated by trauma, underscoring the interactive effects of genetics and environment (Siever et al., unpublished data).
Twin studies of BPD show substantial heritability scores of 0.65 to 0.76 (51–53). One study suggests that a highly heritable factor underlies BPD symptom domains and is closely related to affective instability, and that there is a strong genetic correlation between BPD traits and neuroticism, and an inverse relationship with conscientiousness and agreeableness (54). Identifying the neurobiology underlying the dimensions that make up mental disorders, such as BPD, is a goal of the NIMH RDoC program. The resulting data will allow to better understand the illness and to identify new therapeutics both for BPD and the other disorders for which BPD individuals are predisposed (55). Candidate genes for impulsive aggression and emotional dysregulation include those that regulate the activity of neuromodulators such as serotonin and catecholamines, as well as neuropeptides (56, 57).
DSM-IV defines ASPD as a pervasive pattern of disregard for and violation of the rights of others that has been occurring since the age of 15 years, as indicated by at least three of seven criteria (7). As defined in the new alternative method to diagnose PDs, to be seen in Section III of DSM−5, the diagnosis of ASPD is characterized by impairments in personality (self and interpersonal) functioning and the presence of pathological personality traits, including disinhibition (characterized by irresponsibility, impulsivity, and risk-taking) and antagonism (characterized by manipulativeness, deceitfulness, callousness, and hostility).
ASPD is characterized by two types of aggression: impulsive or reactive aggression and instrumental aggression. Impulsive aggression, which is more retaliatory and impulsive and occurs in response to a perceived threat, is believed to be the core dimension underlying ASPD, as is also seen in all the axis II Cluster B PDs. Instrumental aggression, which is controlled/planned and serves an instrumental, goal-directed end, is also found in ASPD but is characteristic of psychopathy (58–60).
The data strongly support a disruption of amygdala and prefrontal cortex function—specifically in the OFC, ACG, and dorsolateral prefrontal cortex—in individuals with psychopathic traits and/or antisocial behavior. However, it is important to distinguish between ASPD and psychopathy. Psychopathy is a construct characterized by severe deficits in emotional processing (reduced guilt, empathy, and attachment to significant others; callous and unemotional traits) and increased risk for antisocial behavior (61, 62). Despite its overlap with ASPD, psychopathy is a distinct disorder: while most individuals who are diagnosed with psychopathy will also meet criteria for ASPD, only about 10% of those with ASPD meet criteria for psychopathy (63). There is a paucity of studies focusing on the neurobiology of ASPD specifically and separate from psychopathy.
Laakso et al. (64) observed reductions in volume of the dorsolateral, medial frontal, and orbitofrontal cortices in subjects with ASPD. However, after controlling for substance use and education, they concluded that the observed volume deficits were related more to alcoholism or differences in education rather than the diagnosis of ASPD. Other research does suggest reduced prefrontal volumes in ASPD, after controlling for effects of substance use (60, 65–67). ASPD subjects have also been reported to have smaller temporal lobes (68, 69), smaller whole brain volumes (68), larger putamen volumes (68), larger occipital (66) and parietal lobes (66), larger cerebellum volumes (66), decreased volumes in specific areas of the cingulate cortex, insula, and postcentral gyri (66), and cortical thinning in medial frontal cortices (70). However, other studies (71) found no differences in gray matter volumes between offenders with ASPD without psychopathy and healthy comparison subjects. Based on animal models, reactive aggression is part of a progressive response to threat mediated by a threat system that involves the amygdala, the hypothalamus, and the periaqueductal gray. This system is regulated by medial, orbital, and inferior frontal cortices (59, 72). According to this model, individuals with high reactive aggression should show increased amygdala responses to emotional provocation and reduced frontal emotional regulatory activity (59). In support of this model, multiple studies have reported decreased activity in the frontal lobes in individuals with antisocial and violent behavior, particularly in the OFC, ACC, and dorsolateral prefrontal cortex (73–79). Raine et al. (80) observed that impulsive murderers had lower left and right prefrontal metabolism with PET, higher right hemisphere subcortical metabolism, and lower right hemisphere prefrontal/subcortical ratios. Goethals et al. showed that patients with BPD or ASPD who had impulsive behavior had low perfusion in the right prefrontal and temporal cortex, but they found no differences in brain perfusion between BPD and ASPD patients (81). The data also suggest decreased serotonergic responsiveness in ASPD compared with healthy volunteers in OFC, adjacent ventral medial frontal cortex, and cingulate cortex (27).
Some of the studies suggest that at least part of the neural abnormalities found in ASPD subjects may not be specific to this disorder but rather associated with aggressive traits that are associated with a tendency to violent behavior. For example, Barkataki et al. (82) found that both violent ASPD subjects and violent schizophrenia patients, but not nonviolent schizophrenia patients, showed reduced thalamic activity in association with modulation of inhibition in a go/no-go task. However, another study by the same group suggests that, although there are neural alterations related to violence found both in violent schizophrenic and violent ASPD patients in occipital and temporal regions, there are interesting differences specific to ASPD and schizophrenia, respectively. Specifically, they found that the violent ASPD subjects showed attenuated thalamic-striatal activity during later periods in a “threat of electric shock” task, whereas in the violent schizophrenic subjects there was hyperactivation in the same areas (83). This suggests that although there is a shared biological deficit, violent behaviors may arise from different mechanisms according to the specific disorder.
Family, twin, and adoption studies suggest that antisocial spectrum disorders and psychopathy are heritable (84, 85). In the last decade, considerable scientific energy has been focused on identifying specific genetic factors involved in the development of aggressive behavior, as a trait observed in antisocial spectrum disorders and psychopathy. However, behavioral genetics has yet to elucidate specific genetic pathways that lead to the genesis of the disorders, or develop molecular genetic tests that may inform diagnosis or treatment (86). It has been suggested that examining gene-by-environmental interactions, performing detailed whole genome association studies, functional imaging studies of genetic variants, and examining the role of epigenetics may provide valuable new targets for research (87). One of the challenges of the existing research is the heterogeneity of the phenotypes analyzed in different studies, including individuals with ASPD with or without psychopathy, psychopathy with or without ASPD, antisocial behavior, conduct disorder, oppositional defiant disorder, disruptive behavior disorder, criminals, violent offenders, or aggressive individuals, with only a handful of studies focusing on ASPD specifically (86).
Several genome-wide linkage and association studies have suggested possible genomic locations in chromosomes 1, 2, 3, 4, 9, 11, 12, 13, 14, 17, 19, and 20 for antisocial spectrum disorders, but they must be interpreted with caution, since very few findings reach genome-wide significance, and even fewer have been replicated (86). Of note, only one of these studies specifically included subjects with a diagnosis of ASPD, and found several regions of interest in the genome (88).
The most widely studied genes in antisocial spectrum disorders have been those related to serotonergic and dopaminergic systems, including catechol-O-methyltransferase (COMT), monoamine oxidase A (MAOA), dopamine beta hydroxylase (DBH), trytophan hydroxylase 1 and 2 (TPH 1 and 2), dopamine receptor D2 (DRD2), dopamine receptor D4 (DRD4), serotonin receptor 1B (5HTR1B), serotonin receptor 2A (5HTR2A), serotonin transporter (5HTT) and dopamine transporter (DAT). Other targets include androgen receptors (AR), based on the gender differences in frequencies of antisocial spectrum disorders, and novel sites such as SNAP25, which was identified as a region of interest in genome-wide studies (86). Currently, the strongest evidence available points to the MAOA and 5HTT genes in antisocial spectrum disorders (86).
In sum, there is compelling evidence that genes involved in the serotonergic system are implicated in impulsive aggression, but the findings regarding genetic factors related to antisocial personality disorder are not as robust.
Schizotypal personality disorder (STPD) is part of the schizophrenia spectrum and is characterized by the presence of attenuated symptoms typically present in chronic schizophrenia. STPD is defined by DSM-IV-TR as “a pervasive pattern of social and interpersonal deficits marked by acute discomfort with, and reduced capacity for, close relationships as well as by cognitive or perceptual distortions and eccentricities of behavior, beginning by early adulthood and present in a variety of contexts”, and requires five or more of its nine criteria (7).
The neurobiological factors underlying the genesis of STPD may be conceptualized in relation to each of the core traits of the disorder (psychotic-like symptoms and cognitive organization disturbances). In this way, disturbances in cognitive organization and information processing may contribute to the detachment, desynchrony with the environment, and cognitive/perceptual distortions of STPD and other schizophrenia spectrum personality disorders (57).
Psychotic-like symptomatology is characteristic of STPD patients. Like in schizophrenia, increased dopaminergic neurotransmission is associated with more prominent psychotic symptoms, and the dimension of psychotic-like perceptual distortions has been correlated with measures of dopaminergic activity. The fact that STPD patients have less prominent psychotic symptoms than patients with schizophrenia is believed to be due to better buffered subcortical dopaminergic activity (6, 57). The results of functional and structural imaging and neuroendocrine challenge studies support this hypothesis. This better buffering system may result in less responsiveness to stress by subcortical dopaminergic systems, which may protect against psychosis (6, 57, 89). It has been suggested that dopaminergic activity can be relatively increased or decreased, depending on the predominance of psychosis-like (hypervigilance and stereotypic cognitions/behaviors) or deficit-like (deficits in working memory, cognitive processing, and hedonic tone) symptoms (6).
Cognitive impairment dimension
Research data suggest that patients with STPD suffer cognitive impairment, likely related to structural brain abnormalities, especially in the temporal cortex, similar to those seen in patients with schizophrenia. Despite these similarities, STPD patients differ from schizophrenia patients in that they have less impaired executive function—likely due to greater reserves in prefrontal function (6, 57). Specifically, patients with STPD have increased ventricular volumes and frontal-temporal volume reductions similar but milder than those seen in patients with schizophrenia, with sparing of some key regions (90).
Specific cognitive dimensions found to be impaired in STPD include attention, visual, and auditory working memory; verbal learning; and memory. Although STPD individuals perform poorly on executive function tasks, the more generalized intellectual deficits found in schizophrenia are not observed in STPD (57, 91). These cognitive deficits may contribute to the impairments in social rapport and inability to read social cues seen in STPD patients. Actually, deficits in working memory have been correlated with interpersonal impairment (92).
Decreased dopaminergic and noradrenergic activity in the prefrontal cortex may contribute to the cognitive impairment in STPD. This is consistent with functional studies showing decreases in frontal activation during executive functioning tasks in STPD subjects. However, unlike schizophrenic patients and normal subjects, STPD subjects appear to activate other compensatory regions during executive function tasks (93). STPD subjects also suffer deficits in information processing, reflected in physiological impairments seen in the schizophrenia spectrum. These include deficits in prepulse inhibition (PPI) of the acoustic startle response, in the startle blink paradigm, in the P50 evoked potential paradigm, or in smooth pursuit eye movement among others (See Siever and Davis  for a review).
In summary, STPD subjects show cognitive and physiological impairments that seem to be partially caused by reduced prefrontal dopaminergic function, which may be partially compensated by activation in other brain areas not used by healthy comparison subjects.
STPD is partly heritable (94), and its genetic factors overlap with those for schizophrenia and other schizophrenia spectrum disorders (95, 96). It has been suggested that positive and negative symptoms of STPD represent two distinct heritable dimensions. Thus, in disorders of the schizophrenia spectrum, a set of genetic factors expressed as social and cognitive deficits (“spectrum phenotype”) might be transmitted independently from a second genetic factor set related to psychosis (“psychotic phenotype”) (6). Dopaminergic candidate genes, including the dopamine D4 receptor and the dopamine β-hydroxylase gene, have been found to be associated with psychosis-like symptomatology (6, 57). A polymorphism of catechol-O-methyltransferease (COMT), which metabolizes dopamine and regulates its activity in the frontal cortex, has been associated with working memory deficits and other cognitive deficits both in subjects with schizophrenia and in subjects with STPD (97, 98). In a large cohort of young healthy individuals, Stefanis et al. showed an association between common variants in G-protein signaling 4 (RGS4) and d-amino acid oxidase (DAAO) genes with negative schizotypal personality traits; dysbindin (DTNBP1) variants were associated with positive and paranoid schizotypy measures (99, 100). Finally, preliminary results from our group, using the custom Consortium on the Genetics of Schizophrenia (COGS) 1,536-SNP chip, showed a strong association between polymorphisms in ERBB4, NRG1, and genes involved in glutamate, dopamine, GABA, and serotonin receptor signaling, as well as cell signal transduction, with categorical clinical diagnosis (STPD versus healthy comparison subjects) and dimensional quantitative phenotypes of STPD, including cognitive impairment, interpersonal deficits, and paranoia (Siever and Roussos, unpublished data).
In summary, several genetic variants have been associated with STPD traits and/or dimensional quantitative phenotypes of STPD, including cognitive impairment symptoms, opening promising avenues for research and pharmacological targets.
The focus of research into the neurobiology of psychiatric disorders, including the personality disorders, has increasingly shifted from categories to dimensions of psychopathology and their underpinnings in neurocircuitry and genetics (as exemplified by the RDoC initiative) (1). The prototypic personality disorders are characterized by the extremes of traits such as affective dysregulation (affective instability and negative affectivity), disinhibited aggression, anxiety/avoidance, cognitive/perceptual dysregulation, and social detachment/isolation. There is growing evidence about the neurobiology of personality dimensions, as exemplified by the association between affective instability and disinhibition/impulsive aggression in BPD with alterations in serotonergic genes and abnormal brain circuitry involving limbic structures such as the amygdala as regulated by prefrontal regions including the OFC. However, despite these promising findings, the current evidence on the neurobiology of personality dimensions is limited by the inconsistency and low specificity of the neural circuits and genetic factors involved.
Despite the significant increase in research into the neurobiological underpinnings of personality disorders, two areas are still in need of significant advances, and are thus not covered in depth in this review.
The first is the genetic basis of personality disorders. As is the case in many psychiatric disorders, although their heritability is moderate to high, the genetic underpinnings of personality disorders remain mostly unknown. While several studies have utilized techniques such as genome-wide association to discover genes in other axis I and axis II disorders, few genome-wide association studies (GWAS) have focused on personality disorders. To date, GWAS in personality disorders have been mostly limited to assessing genome-wide linkage for the five basic personality traits as assessed with the NEO Five-Factor Inventory, and not for specific personality disorders. A recent meta-analysis of GWAS (101) indicated strong signals of linkage for extraversion, agreeableness, and openness. However, a meta-analytic review (102) of the relationships between the five-factor traits and DSM-IV-TR personality disorders suggested that openness to experience has zero to little relationship to DSM personality disorders and does not account for personality disorder symptomatology. As such, the field is still in need of more conclusive GWAS data to advance the understanding of the genetics of personality disorders.
The second area in which the field is lacking empirical data is for pharmacological interventions aimed at personality disorders. While there is a wealth of largely unsuccessful efforts at identifying pharmacologic probes for specific symptoms of personality disorders, as reviewed elsewhere (103–105), few studies have been aimed at testing neurobiologically informed treatments for personality disorders, and there are no FDA-approved medications for any personality disorders. This topic is therefore not covered in this review (104).
Of note, there are some important limitations of the reviewed studies including limited sample sizes, differences in comorbidity and clinical heterogeneity of the patients included, differences in the characteristics of the comparison subjects, differences in subject’s handedness across studies, gender differences, and differences in medications and ongoing psychotherapeutic interventions. All of these factors may confound the results by affecting brain structure and function.
Therefore, further investigation is required into the neural circuits underlying the personality disorder dimensions and their modulation by neuropeptides and neurotransmitters. A better understanding of the neurobiology of personality disorders may help us identify pharmacological and psychosocial interventions, as well as predictors of response.