Frontolimbic Circuitry in BPD

Prefrontal and limbic neural systems mediate both the processing of and responses to emotional stimuli (Phillips, Drevets, Rauch, & Lane, 2003), and therefore dysfunction in these circuits could mediate a number of core symptoms of BPD (Figure 1). Orbitofrontal dysfunction may underlie the impulsivity, aggression, and mood instability seen in BPD because lesions of this region have been linked with such behaviors in both animals and humans (Mega & Cummings, 1994). Neurobehavioral studies in animals suggest that an interplay between PFC, particularly the OFC, and the amygdala is important for affective style in humans (Davidson, 2002). In fact, extensive reciprocal connections between the OFC and the amygdala are well known, and these regions function as part of an integrated neural system guiding socioemotional decision making and adaptive response selection (Baxter, Parker, Lindner, Izquierdo, & Murray, 2000).

Frontotemporo–limbic and subcortical brain regions relevant to neural models of BPD. Selected functional affiliations of these regions (see text) are briefly noted.

Amygdala dysfunction has been implicated in playing an important role in the development of BPD because this limbic structure is known to be involved in the processing of negative valence stimuli, the establishment of fear conditioning, the production of strong negative emotional states, and vigilance for environmental stimuli with important survival value (Davis & Whalen, 2001; Isenberg et al., 1999; LeDoux, 1996). Findings of amygdala dysfunction in the above-mentioned fMRI studies likely reflect abnormal processing of aversive paradigm stimuli and/or the activation of emotional memories induced by such stimuli (Breiter, Rauch, et al., 1996). Furthermore, increased amygdala activation in BPD subjects may reflect an oversensitization to aversive emotional stimuli or a relative failure of normal habituation. A strong habituation of the BOLD fMRI signal in response to emotional stimuli has been observed in normal controls (Breiter, Etcoff, et al., 1996; LaBar, Gatenby, Gore, LeDoux, & Phelps, 1998), and may be dysfunctional in BPD individuals with long-lasting, intense emotional reactions.

Hippocampal dysfunction is not surprising as well, given the susceptibility of this neural structure to stress in the setting of increased cortisol and excitatory amino acids (McEwen & Magariños, 2001) and the prevalence of early traumatization experienced by those with BPD (Zanarini, 2000). Hippocampal volumes in individuals with BPD have been negatively correlated with the extent and duration of self-reported traumatic experiences, suggesting the presence of stress-induced structural changes in this disorder (Driessen et al., 2000). Because the hippocampus mediates fear responses to contextual features of the environment (Phillips & LeDoux, 1992), dysfunction of this structure may result in a maladaptive overgeneralization of fear responses (e.g., fear of abandonment) to a broad range of interpersonal situations as observed in those with BPD.

Activation of fusiform gyrus (BA 37), a region implicated in complex visual feature detection and recognition of facial expression (George et al., 1993), could be due to backprojections from the amygdala for the purpose of strengthening visual attention to emotionally relevant stimuli, a finding consistent with the hypervigilance for emotional aspects of the environment seen in individuals with BPD. Additional evidence for an emotion-based attentional bias in BPD is provided by Arntz, Appels, and Sieswerda's (2000) emotional Stroop study in which BPD subjects had longer response times than normal controls to negative versus neutral valence words. Interestingly, this finding was not specific to words reflecting BPD themes, but rather seemed to reflect a more general hypervigilance for all negative valence stimuli.

Dysfunction in medial prefrontal areas, in particular anterior cingulate cortex, a region involved in emotional regulation and with important inhibitory connections to the amygdala that mediate the extinction of fear responses (Carmichael & Price, 1995; Devinsky, Morrell, & Vogt, 1995), might underlie the difficulty modulating anxiety and other intense affects seen in individuals with BPD. Medial PFC lesions in humans result in emotional dysfunction and deficits in interpersonal functioning that requires the correct interpretation of the emotional expressions of others (Damasio, Grabowski, Frank, Galaburda, & Damasio, 1994). Although normal individuals tend to have increased activity in the anterior cingulate during the induction of angry affect states (Dougherty et al., 1999; Kimbrell et al., 1999), perhaps reflecting adaptive modulating functions, individuals with BPD have been shown to have decreased function of (De la Fuente et al., 1997; Goyer et al., 1994) and reduced gray matter volume in (Hazlett et al., in press) this area. In addition, anterior cingulate cortex recently has been implicated in the processing of the sensory and affective dimensions of pain (Hofbauer, Rainville, Duncan, & Bushnell, 2001; Olausson et al., 2001), and diminished activity in this region may account for the significantly attenuated pain perception among individuals with BPD and self-injurious behaviors (Bohus et al., 2000).

Abnormal activity of ventrolateral PFC (Herpertz et al., 2001), with its direct connections with the basal nucleus of the amygdala, could result in compromised inhibition of amygdala-driven emotional responses thereby undermining top-down control of this limbic region in BPD (Drevets, 1999). Activation of DLPFC (Schmahl, Elzinga, et al., 2003) has been observed in prior imaging studies using personalized scripts in women with abuse-related posttraumatic stress disorder (PTSD; Bremner et al., 1999; Shin et al., 1999), possibly reflecting the role of this prefrontal region in the encoding and retrieval of verbal memories (Tulving, Kapur, Craik, Moscovitch, & Houle, 1994) and/or its emerging role in aspects of emotional processing (Davidson, Jackson, & Kalin, 2000) and emotional–cognitive interactions (Liotti & Mayberg, 2001; Sergerie, Lepage, & Armony, 2005). Additional evidence of DLPFC dysfunction comes from Tebartz van Elst et al.'s (2001) proton magnetic resonance spectroscopy study demonstrating subtle prefrontal neuropathology as indicated by reduced N-acetylaspartate concentrations in this prefrontal region in individuals with BPD.

Interaction of Emotion and Behavior in BPD

As summarized above, functional neuroimaging studies have investigated the neural systems underlying emotional dysregulation in BPD using a variety of neuropsychological probes. Numerous other imaging studies with normal subjects and various patient populations have examined the neural basis of emotion and behavior, but rarely the interaction between these two domains. Our lab has developed an emotional Go/No-Go fMRI paradigm to specifically investigate the functional neuroanatomy underlying the interaction of emotion and inhibitory behavioral control in both normals and individuals with BPD. We believe that this important interaction has relevance to a wide range of human experience and behavior in health and disease. The BPD work is being performed in collaboration with Dr. Otto Kernberg's research group at the Personality Disorders Institute of the Weill Medical College of Cornell University. This approach can shed light on the neural circuits underlying the clinically observed phenomenon of behavioral impulsivity in the context of intense negative affect states so commonly seen in this patient population. The methodological approach of using neuropsychological probes that conjoin emotional stimuli with nonemotional behavioral task demands is an emerging line of investigation within the cognitive neurosciences to further investigate important interactions between emotion, cognition, and behavior (Dolan, 2002; Elliott, Rubinsztein, Sahakian, & Dolan, 2002).

Our fMRI paradigm is a variation on the classic Go/No-Go Task (Drewe, 1975) used extensively in investigational neuropsychology to probe response inhibition, and involves the execution or inhibition of a prepotent motor response triggered by a go or no-go stimulus, respectively. Our subjects are instructed to execute or inhibit a response to visually presented emotional stimuli. By experimentally manipulating the emotional valence of these stimuli that reflect themes salient for individuals with BPD, we are able to investigate the modulation of behavioral task demand (go or no-go) by emotional valence, thereby probing the neural systems subserving the interaction of emotion and behavioral control in this population.

We have hypothesized and detected dysfunction in medial OFC in individuals with BPD while performing this emotional Go/No-Go Task, a finding consistent with this region's presumed role in regulating behavior under changing contingencies and its role in guiding adaptive behavioral responses within socioemotional contexts (Damasio et al., 1994; Elliott et al., 2002; Rolls, 2000). Specifically, we have hypothesized differential neural activity in this region in BPD subjects compared with normals, reflecting the former's greater difficulty modulating behavioral responses under conditions of negative emotion.

Medial OFC is believed to process viscerosomatic afferent information as cues reflecting the personal relevance of environmental contexts, and to integrate such emotional information into decision making and response selection (Bechara, Damasio, & Damasio, 2000; Dolan, 1999). According to Damasio's somatic marker hypothesis (1994), dysfunction of medial OFC results in an impaired capacity to adaptively integrate emotional information into behavioral response selection. This may have particular relevance for individuals with BPD who, while experiencing intense negative affect states, have greater difficulty regulating behavioral responses. In addition, medial OFC appears to be involved in the suppression of context-activated emotional memories irrelevant to current cognitive/behavioral demands (Treyer, Buck, & Schnider, 2003), thereby allowing for more adaptive responses to the immediate environment. Individuals with BPD, however, seem to respond in characteristically inflexible and maladaptive ways based not upon current social contexts, but rather according to implicit emotional memories of past interpersonal experiences, possibly reflecting OFC dysfunction.

Integration With Developmental Perspectives

Evidence of dysfunction in limbic and prefrontal neural systems subserving affect regulation and behavioral control is consistent with developmental theories of BPD. Clarkin, Yeomans, and Kernberg's (1999) psychoanalytic developmental theory of BPD posits both constitutional (i.e., neurobiologic) and early environmental factors giving rise to a psychic structure populated by distorted, unidimensional mental representations of self and others, which leads to maladaptive perceptions of and responses to current interpersonal reality. According to this model, the normal infant's early experiences with caregivers lead to a multitude of distinct and contradictory internalized relationship patterns; for example, the satisfied, calm self linked with an affect of love to the nurturing other, and the needy, frustrated self linked with an affect of hate to the depriving other. Ideally, over the course of development these split-off mental representations are integrated, resulting in a more complex and realistic image of the other who can both satisfy and frustrate, allowing for more adaptive interpersonal functioning. Adaptive integration of these early, distorted representations of self and others, and the intense affects associated with them, likely requires sufficient affect regulation. Individuals who develop BPD appear not to reach this developmental stage, perhaps due to a primary (genetic) neurobiological deficit in those neural systems subserving affect regulation, effects of early trauma on these developing systems, and/or deficits in parental modulation of this emerging self-regulating capacity.

Deficits in affect regulation and impulse control during early development may lead to particularly conflictual experiences with caregivers, experiences that become internalized as prototypes for future relationships, eventually reflected in adult psychopathology. In addition, it is possible that these conflict-ridden interpersonal experiences and the resultant internalized mental representations act back upon the neurobiological substrate to produce further dysfunction in neural circuits required for affect/behavioral regulation. In animal models, chronic psychosocial and behavioral stress have been shown to produce neuronal reorganization and atrophy in medial PFC and hippocampus (Magariños, McEwen, Flugge, & Fuchs, 1996; Radley et al., 2004), as well as neuronal growth in the amygdala (McEwen, 2004), in part mediated via glucocorticoid action on these brain regions. Findings of reduced prefrontal and hippocampal volumes in individuals with BPD from human neuroimaging studies (Brambilla et al., 2004; Driessen et al., 2000; Irle, Lange, & Sachsse, 2005; Tebartz van Elst et al., 2003) are consistent with these animal findings, and may reflect early psychosocial stress-induced changes in these areas. For example, Posner et al. (2003) have suggested that the anterior cingulate and the self-regulating functions it subserves may be susceptible to the effects of trauma during early development, possibly accounting for the deficits in effortful control observed in adults with BPD.

Epidemiological data strongly suggest that numerous types of early psychosocial stress (sexual/physical abuse, neglect) are involved in the development of BPD (Zanarini, 2000), perhaps accounting for the high comorbidity of PTSD (50%) seen in BPD patients (McGlashan et al., 2000). The effects of psychosocial stress on the developing brain may be mediated in part by genetic factors. Just as a polymorphism of the serotonin transporter (5-HTT) gene has been shown to moderate the influence of stressful life events on the emergence of depression (Caspi et al., 2003), certain genes may render some individuals more susceptible to developing BPD in response to environmental stressors via functional and structural changes in the relevant neural circuits discussed above. Although the functional consequences of these morphological changes in prefrontal and limbic areas remain unclear, they could result in ongoing difficulties in affect regulation and behavioral control leading to continued psychosocial stress, resulting in a pathological reciprocal interaction between neurobiology and the environment in those with BPD.

An alternative to this psychosocial stress-induced model of BPD comes from Fonagy, Gergely, Jurist, and Target (2002), who propose that the development of effective affect regulation takes place within the context of the moment to moment interactions between infant and caregiver. They propose a social biofeedback model of parental affect mirroring according to which the parent's attuned mirroring of the infant's emotional expressions is vital for the child's development of higher order mental representations of his internal feeling states, eventually providing the basis for the infant's emerging capacity to reason about and to regulate his own affects. According to this theory, insufficient affect mirroring leads to a developmental deficit in the ability to mentalize about emotional states and may result in the relative failure of effective cognitive control over one's affect states observed in individuals with BPD. The emerging neuroimaging evidence of prefrontal dysfunction in BPD, a region involved in the top-down cognitive control of limbic areas, may in some cases reflect this early failure of affect mirroring and provides another developmental model for the effects of early social experience on the development of neural systems required for adaptive affect regulation.

Potential Neural Mechanisms of Psychotherapeutic Change in BPD

Based upon recent neuroscientific findings of dysfunctional neurocircuitry in BPD and these developmental theories of borderline psychopathology, we can begin to conceptualize putative mechanisms of psychotherapeutic action at the neural systems level in the treatment of this disorder. The psychotherapeutic use of language to describe and talk about emotional experience provides and fosters a representational/symbolic means to become aware of, to reason about, and ultimately to modulate affect states. This may be similar to the developmental emergence of mentalization (Fonagy et al., 2002), and may reflect the therapeutic enhancement of dysfunctional prefrontal regions underlying the capacity to mentalize. In addition, the activation of unconscious emotional memories of early interpersonal experiences with significant others within the therapeutic relationship may present the opportunity for enhanced declarative remembering of the past and more adaptive reconsolidation of such pathological memories (Ross, 2003). In fact, evidence from animal models demonstrates the possibility of altering emotional memories via disruption of reconsolidation at the level of the amygdala (Debiec & LeDoux, 2004). Finally, psychodynamically oriented psychotherapies that focus on the patient's transference to the therapist (the patterned and inflexible repertoire of emotional/behavioral responses based on internalized representations of past relationships; Clarkin et al., 1999) may lead to a greater flexibility of responses via alterations of frontolimbic circuitry. For example, changes in the OFC and hippocampus that mediate emotional/behavioral responses to contextual features of the social environment could allow for a greater capacity to process and respond to current interpersonal reality in novel, more adaptive ways rather than in repetitive, inflexible ways based upon past experience. In addition, transference-focused psychotherapies might promote higher order prefrontal processes that mediate the capacity for more complex mental representations of self and others, thereby changing the very quality of the subjective interpersonal world of those with BPD.