Neural Correlates of Negative Emotionality in Borderline Personality Disorder: An Activation- Likelihood-Estimation Meta-Analysis

ARCHIVAL REPORT Neural Correlates of Negative Emotionality in Borderline Personality Disorder: An Activation- Likelihood-Estimation Meta-Analysis Anthony C. Ruocco, Sathya Amirthavasagam, Lois W. Choi-Kain, and Shelley F. McMain Background: Emotional vulnerabilities at the core of borderline personality disorder (BPD) involve a dysfunction of frontolimbic systems subserving negative emotionality. The specific regions identified in individual studies, however, vary widely and provide an incomplete understanding of the functional brain abnormalities that characterize this illness. A quantitative synthesis of functional neuroimaging studies might clarify the neural systems dysfunctions that underlie negative emotionality in BPD. Methods: An electronic search of Medline and PsycInfo databases from 2000 to 2012 identified 18 potential studies, of which 11 met inclusion criteria for the meta-analysis and comprised a pooled sample of 154 BPD patients and 150 healthy control subjects. Contrasts of negative versus neutral emotion conditions were analyzed with an activation-likelihood-estimation meta-analytic approach. Group comparisons were performed on study-reported between-subjects contrasts and independent subtraction analyses based on within-subjects contrasts. Results: Healthy control subjects activated a well-characterized network of brain regions associated with processing negative emotions that included the anterior cingulate cortex and amygdala. Compared with healthy control subjects, BPD patients demonstrated greater activation within the insula and posterior cingulate cortex. Conversely, they showed less activation than control subjects in a network of regions that extended from the amygdala to the subgenual anterior cingulate and dorsolateral prefrontal cortex. Conclusions: Processing of negative emotions in BPD might be subserved by an abnormal reciprocal relationship between limbic structures representing the degree of subjectively experienced negative emotion and anterior brain regions that support the regulation of emotion. Contrary to early studies, BPD patients showed less activation than control subjects in the amygdala under conditions of negative emotionality. Key Words: Anterior cingulate cortex, borderline personality disorder, emotion regulation, functional magnetic resonance imaging, insula, negative emotionality Borderline personality disorder (BPD) is a severe psychiatric illness affecting 1% 2% of the general population and upwards of 20% of psychiatric inpatients (1,2). Emotion dysregulation is a hallmark symptom of BPD, characterized by an unstable expression and more intense subjective experience of negative emotions (3,4). The neural basis of emotion dysregulation in BPD has been the subject of considerable scrutiny among neuroscientists, with the bulk of this research with functional magnetic resonance imaging (fmri) to investigate negative emotionality in BPD (5,6). The processing of negative emotions in healthy individuals is subserved by a network of brain regions comprising the medial/ventral prefrontal cortex (PFC), subcallosal/anterior cingulate cortex (ACC), insular cortex, and amygdala (7,8). Given that individuals with BPD experience difficulties in the regulation of negative emotions, the task of elucidating which neural systems underlie these symptoms is critical for constructing a coherent neurobiological model of emotion dysregulation in BPD. From the Department of Psychology (ACR, SA), University of Toronto Scarborough; Borderline Personality Disorder Clinic and Centralized Assessment Triage and Support Program (SFM), Centre for Addiction and Mental Health, Toronto, Canada; and the Department of Psychiatry (LWC-K), Harvard Medical School, McLean Hospital, Belmont, Massachusetts. Address correspondence to Anthony C. Ruocco, Ph.D., Department of Psychology, University of Toronto Scarborough, 1265 Military Trail, Toronto, Canada; E-mail: anthony.ruocco@gmail.com. Received Jun 21, 2012; revised Jul 17, 2012; accepted Jul 18, 2012. 0006-3223/$36.00 http://dx.doi.org/10.1016/j.biopsych.2012.07.014 Early fmri studies of BPD presumed that biological vulnerabilities to emotional hyperreactivity might have their substrate in heightened neural activity in limbic structures (e.g., amygdala) that were understood to be involved in the subjective experience of negative emotions. Consistent with this hypothesis, Herpertz et al. (9) demonstrated elevated bilateral amygdala activity in six female BPD patients while they passively viewed highly arousing and unpleasant photographs. Following this work, Donegan et al. (10) measured activity within the amygdala in BPD patients while they viewed neutral, happy, sad, and fearful facial expressions and found higher activity in the left amygdala as compared with healthy control (HC) subjects. A series of subsequent investigations used a variety of paradigms to evaluate negative emotionality in BPD, including tasks that asked subjects to recall unresolved life events (11), use scripts to visualize episodes of self-injury (12), and employ a psychological distancing strategy to regulate emotional responses (13,14). The results of these studies revealed functional abnormalities in a network of brain regions extending beyond the amygdala to include the occipital cortex; dorsal ACC; and dorsolateral, orbital and medial PFC. On the basis of these studies, narrative reviews of neuroimaging findings in BPD (15,16) have converged on a model of emotion dysregulation in this illness that implicates a dysfunction of two neural processes: a deficient regulatory control system operating through anterior brain regions (i.e., PFC, ACC) that show reduced engagement in functional neuroimaging studies; and a hyperresponsive subcortical limbic system that reflects heightened activity in specific neural structures (e.g., amygdala, insula) and might be associated with a subjectively more intense experience of negative emotions. According to this model, emotion dysregulation in BPD is thought to result from a failure of top-down frontal control pro- BIOL PSYCHIATRY 2013;73:153 160 2013 Society of Biological Psychiatry

154 BIOL PSYCHIATRY 2013;73:153 160 A.C. Ruocco et al. Table 1. Sample Characteristics of Studies Included in the Meta-Analysis Authors Borderline Patients Healthy Control Subjects n Age % Female Patient Type n Age % Female Beblo et al. (11) 20 31.3 100 Inpatient 21 32.6 100 Guirtart-Masip et al. (66) 10 31.3 50 Outpatient 10 31.2 50 Herpertz et al. (33) 6 26.2 100 Inpatient 6 27.2 100 Koenigsberg et al. (21) 18 32.6 55.6 Outpatient 16 31.8 56.2 Kraus et al. (12) 11 25.6 100 Inpatient 10 25.6 100 Minzenberg et al. (67) 12 30.3 41.7 Outpatient 12 30.7 50 Schnell et al. (22) 14 28 100 Inpatient 14 28.4 100 Schulze et al. (48) 15 27.6 100 Inpatient 15 24.5 100 Silbersweig et al. (61) 16 31.2 93.8 Not applicable 14 23.8 71.4 Smoski et al. (68) 12 30.8 0 Outpatient 12 32.8 0 Wingenfeld et al. (69) 20 29.8 70 Outpatient 20 29.4 70 Total/Mean 154 29.5 73.7 150 28.9 72.5 cesses involved in modulating activity in over-reactive emotiongenerating limbic structures. A precise characterization of the neural systems abnormalities underlying negative emotionality in BPD, however, remains elusive. Integration of findings across individual studies is complicated by considerable variability in sample sizes, gender compositions, and psychiatric comorbidities, which limits the conclusions that might be drawn from any one study. The purpose of the current metaanalysis, therefore, was to quantitatively synthesize individual neuroimaging studies of negative emotionality in BPD so as to identify those neural structures that show the greatest functional abnormalities in this regard. We used an activation-likelihood-estimation (ALE) approach to examine differences in functional activation on a voxel-wise basis between BPD patients and HC subjects (17) and thereby provide a more coherent understanding of the neural systems subserving negative emotionality in BPD. Methods and Materials Study Selection The electronic databases Medline and PsycInfo were searched with the key words borderline with independent matched searches with the key word(s) borderline personality disorder, functional magnetic resonance imaging, fmri, neuroimaging, neural, imaging, emotion, and affect. The asterisk symbol (*) was used to incorporate all possible suffix variations of the search terms in study retrieval. Both English and non-english language articles were considered in the literature search. Articles were considered for inclusion in the meta-analysis if they met the following criteria: 1) publication between 2000 and 2012; 2) research designs that included within-subjects contrasts for BPD patients and/or between-subject contrasts for BPD patients versus HC; and 3) reported stereotactic coordinates (i.e., Talairach, Montreal Neurological Institute [MNI]) compatible with the meta-analysis software. If these three criteria were met, the article was required to meet certain standards. First, patients must have met diagnostic criteria for BPD according to the DSM (third edition or later) with a reliable and valid interview (e.g., Structured Clinical Interview for DSM-IV Axis II Disorders, International Personality Disorder Examination). Second, diagnostic co-occurrence of posttraumatic stress disorder (PTSD) must not have exceeded 50% of the patient sample. We chose to limit the extent of diagnostic comorbidity of PTSD to ensure a more homogeneous set of patients, given the distinct neural responses associated with PTSD when comorbid with BPD (18,19). Third, subjects must have completed a paradigm that included at least two conditions: 1) a negative emotion condition, and 2) a neutral comparison condition. Studies of pain perception and reward processing were excluded. Of the 18 fmri studies initially identified in the literature search, 11 met criteria for inclusion in the meta-analysis. Excluded studies did not meet inclusion criteria for the following reasons: they did not report a negative emotion minus neutral contrast for either within- or between-subjects comparisons (18,20 25); solely region-of-interest (ROI) analyses were reported (10); at least 50% of patients were comorbid for PTSD (10,18); or the study was based on the same sample as a separate report included in the meta-analysis (26). The final combined sample included a total of 154 BPD patients and 150 HC (Table 1). We did not conduct separate ROI analyses (e.g., amygdala), because they were used in too few studies (n 3). Contrast Selection Neuroimaging studies of negative emotionality in BPD employed a number of tasks (e.g., passive viewing, script-driven imagery) and evaluated a variety of contrasts among several task conditions. After examining these characteristics of individual studies, we selected for inclusion in the meta-analysis any neuroimaging study of BPD that investigated negative emotionality broadly defined (Table 2). By adopting this approach, we sought to evaluate as many relevant studies as possible while maintaining a reasonable level of homogeneity in our measurement of negative emotionality. Coordinates based on within-subjects and between-subjects contrasts of negatively valenced minus neutral emotion conditions were extracted and included in the meta-analysis. The inverse contrast was reported very infrequently in primary studies and thus was not further investigated. This approach resulted in four primary analyses: 1) within-subjects contrasts for BPD patients; 2) within-subjects contrasts for HC; 3) between-subjects contrasts for BPD HC; and 4) between-subjects contrasts for HC BPD. Two secondary analyses were conducted with the subtraction analysis method of GingerALE software version 2.1 to aggregate within-subject contrasts reported by individual studies to generate an independent between-subjects contrast (BPD HC and HC BPD). Following this data analytic approach, a total of 11 studies were included in the meta-analysis, 10 of which contributed to study-reported betweensubjects contrasts and 6 of which contributed to within-subjects contrasts. ALE Meta-Analysis The GingerALE 2.1 BrainMap application (27) was used to generate quantitative voxel-wise ALE maps for the contrasts of interest. Input files of study foci were manually created for coordinate-based data in both Talairach and MNI spaces, although the final ALE anal-

A.C. Ruocco et al. BIOL PSYCHIATRY 2013;73:153 160 155 Table 2. Paradigm Descriptions and Contrasts Evaluated in the Meta-Analysis Study Paradigm Between- and Within-Subjects Contrasts Beblo et al. (11) Recollection of resolved vs. unresolved life events BPD HC, unresolved resolved life events Guirtart-Masip et al. (66) Discrimination of negative vs. neutral faces BPD HC, negative neutral faces Herpertz et al. (33) Perception of highly arousing unpleasant vs. neutral BPD HC, negative neutral photographs photographs Koenigsberg et al. (21) Perception of photographs portraying aversive (negative) vs. neutral interpersonal situations HC BPD, negative minus neutral BPD HC, negative minus neutral Kraus et al. (12) Script-driven imagery to induce a negative vs. neutral emotional state BPD HC, negative minus neutral script HC BPD, negative minus neutral script Minzenberg et al. (67) Photographs of faces with angry, fearful and neutral expressions BPD HC, fear minus neutral BPD HC, fear minus neutral BPD HC, neutral minus fear BPD HC, neutral minus fear BPD HC, anger minus neutral BPD HC, anger minus neutral BPD HC, neutral minus anger Schnell et al. (22) Perception of negatively valenced drawings vs. neutral photographs HC, negative minus neutral BPD, negative minus neutral Schulze et al. (48) Viewing of negative vs. neutral photographs HC BPD, negative minus neutral BPD HC, negative minus neutral Silbersweig et al. (61) Emotional lexical go/no-go task HC BPD, negative minus neutral for no-go trials BPD HC, negative minus neutral for no-go trials Smoski et al. (68) Emotional oddball task containing neutral and negative photographs HC BPD, negative minus neutral BPD HC, negative minus neutral Wingenfeld et al. (69) Emotional Stroop containing words that were neutral, generally negative, or related to a past negative life event BPD, borderline personality disorder; HC, healthy control subjects. HC BPD, negative minus neutral HC, negative minus neutral ysis was performed in Talairach space. Any coordinates originally reported in MNI space were converted to Talairach space with the icbm2tal algorithm software (28). From this foci dataset, ALE values were computed for each voxel in the brain. These values in turn are used to calculate a three-dimensional Gaussian distribution of the ALE statistic at each voxel, blurring the coordinate data to correct for potential spatial uncertainty (29). This Gaussian distribution is defined by a predetermined full-width half-maximum outlined by Eickhoff et al. (27). The method used by GingerALE to create ALE maps is a form of random effects analysis, comparing above-chance clustering between experiments instead of between foci (i.e., fixed effects analysis) (29). A threshold is then computed for the ALE map on the basis of a false discovery rate algorithm (30). The resulting thresholded ALE maps then undergo a cluster analysis based on a user-determined minimum cluster size. These clusters are representative of areas with a high likelihood of activation and are assigned anatomical labels at their weighted centers by the Talairach Daemon (27). An independent subtraction analysis was also conducted on the basis of within-subjects contrasts. For this analysis, ALE maps for BPD and HC and a pooled map (a dataset comprising both BPD and control foci) were contrasted to observe statistically significant differences in convergence. A permutation test was performed during the subtraction analysis with 5000 simulations to determine ALE significance at each individual voxel (29). In total, five ALE maps were created for the purposes of this study. Each map was thresholded with a false discovery rate-corrected threshold of p.05 and a minimum cluster threshold of 100 mm 3. Results Within-Subjects Contrasts HC. OurALEanalysisofHCyieldedsevenclustersofactivationthat were dependent on five studies reporting negative neutral contrasts. The right ACC (Brodmann area [BA] 32) contained three independent sites of activation that comprised the dorsal/midcingulate cortex and perigenual and subgenual ACC (Table 3). Significant foci of activation Table 3. Within-Subjects Contrasts of Negative Emotion Neutral Conditions for Healthy Control Subjects Cluster Size (mm 3 ) Talairach Coordinate of Weighted Center x y z Hemisphere Anatomical Region BA 968 5 35 26 Right Anterior cingulate 32 824 1 32 2 Right Anterior cingulate 25 424 30 3 15 Right Amygdala 248 10 29 27 Left Cingulate gyrus 32 128 25 6 10 Left Amygdala 112 13 34 9 Right Anterior cingulate 32 BA, Brodmann area.

156 BIOL PSYCHIATRY 2013;73:153 160 A.C. Ruocco et al. Figure 1. Activation-likelihood-estimation contrast maps of negative emotion neutral for control subjects (top row), borderline personality disorder (middle row) and borderline personality disorder control subjects (bottom row). Maps are based on a false discovery rate-corrected threshold of p.05 and a minimum cluster threshold of 100 mm 3. Areas showing higher activation are in red; lower activation in blue. were also found bilaterally in the amygdala (Figure 1). Table 4 provides the weighted center and cluster size for each ALE-based activation site. Borderline Personality Disorder. A total of five studies reported the within-subjects contrast of negative emotion minus neutral for BPD patients. Our ALE analysis of these studies elicited eight clusters of activation (Table 2). The largest clusters encompassed the dorsal ACC, predominantly on the left side. There were also significant clusters of activation isolated to the right medial (BA 9) and dorsolateral PFC (BA 8), left superior temporal gyrus (BA 38), and left posterior cingulate (BA 30) and a single cluster encompassing both the anterior culmen and posterior declive of the cerebellum (Figure 1). Between-Subjects Contrasts Study-Reported Between-Subjects Contrasts. Ten studies reported between-groups contrasts for negative neutral experimental conditions for BPD patients as compared with control subjects. These contrasts produced areas of activation consistent with those observed in within-subjects contrasts for the respective groups. The ALE-based contrast of BPD patients control subjects revealed greater activation for BPD patients bilaterally within the posterior cingulate gyrus, right insula, and left inferior frontal gyrus (Table 5). Conversely, BPD patients showed less activation than control subjects in the bilateral dorsolateral PFC, right subgenual and dorsal ACC, right amygdala, and left superior temporal gyrus. Independent Subtraction Analysis. We conducted a separate between-groups contrast of negative neutral task conditions on the basis of the lower-level within-subjects contrasts reported separately for BPD and HC in individual studies. With the GingerALE 2.1 subtraction analysis feature, we found that BPD patients showed less activation within a single pronounced cluster bilaterally in the subgenual ACC (BA 25), which also corresponded with the results of study-reported between-subjects contrasts. No significant areas of activation were revealed in the contrast of BPD HC. Discussion Difficulties in the regulation of negative emotions represent a hallmark feature of BPD. Neuroimaging has increasingly been used to understand the neural systems dysfunctions that underlie negative emotionality in BPD, with narrative reviews of this literature suggesting abnormally heightened activity in limbic structures and reduced activation of anterior brain regions during negative emotion processing in this illness (31). Findings across individual studies are highly discrepant, however, perhaps due to variations in meth- Table 4. Within-Subjects Contrasts of Negative Emotion Neutral Conditions for Borderline Personality Disorder Cluster Size (mm 3 ) Talairach Coordinates of Weighted Center x y z Hemisphere Anatomical Region BA 656 13 31 26 Left Cingulate gyrus 32 344 11 27 26 Right Anterior cingulate 32 336 15 41 14 Right Medial frontal gyrus 9 320 44 11 31 Left Superior temporal gyrus 38 288 41 51 20 Left Anterior culmen and posterior declive 152 20 54 14 Left Posterior cingulate 30 112 53 7 37 Right Middle frontal gyrus 8 BA, Brodmann area.

A.C. Ruocco et al. BIOL PSYCHIATRY 2013;73:153 160 157 Table 5. Between-Subjects Contrasts of Negative Emotion Neutral for BPD and HC Contrast Cluster Size (mm 3 ) Talairach Coordinates of Weighted Center x y z Hemisphere Anatomical Region BA BPD HC 792 1 2 30 Left/Right Posterior cingulate 24 456 51 35 20 Right Insula 13 256 50 9 8 Left DLPFC 44 HC BPD 1104 49 15 6 Right Superior temporal gyrus 22 728 39 7 49 Right DLPFC 6 688 43 15 50 Left DLPFC 4 672 1 33 2 Right Anterior cingulate 25 456 63 26 2 Right Superior temporal gyrus 22 384 2 35 26 Right Anterior cingulate 32 144 26 1 11 Right Amygdala BA, Brodmann area; BPD, borderline personality disorder; DLPFC, dorsolateral prefrontal cortex; HC, healthy control subjects. odology (i.e., activation tasks employed, aural vs. visual task presentation) and sample characteristics (i.e., diagnostic comorbidity, clinical severity of patients). The purpose of the present study, therefore, was to quantitatively synthesize individual neuroimaging studies of negative emotionality in BPD with the aim of identifying more consistent patterns of functional activation in this patient group. We examined the contrast of negative neutral task conditions in control subjects and found expected patterns of activation that encompassed the amygdala, midcingulate cortex, and perigenual and subgenual ACC. These findings are consistent with established brain regions known to be involved in negative emotion processing in healthy individuals (7). The BPD patients also activated the ACC, particularly within the dorsal region, as well as the medial and dorsolateral PFC, superior temporal gyrus, posterior cingulate, and cerebellum. These widespread areas of activation suggest that BPD patients might engage a more diffuse network of neural structures associated with negative emotion processing. Compared with control subjects, BPD patients had significantly reduced areas of activation in a complex of brain regions extending from the amygdala to the superior temporal cortex, ACC, and dorsolateral PFC. The amygdala has been reported as showing excessive activity in BPD patients relative to control subjects (32), particularly in early work that used ROI analyses to measure the extent of neural activation in this structure (10). Inspection of the few individual studies that did detect increased amygdala activation in BPD patients suggested that these findings could be related to differences in sample characteristics and task methodology. For example, Herpertz et al. (33) studied only six BPD inpatients with a fixedeffects statistical analysis, and Donegan et al. (10) subtracted crosshair fixation (rather than a neutral emotion condition) from negative facial emotion slides, and nearly all BPD patients had a current major mood disorder including bipolar I disorder. Although the amygdala represents a neural structure susceptible to inhomogeneities in the magnetic field that could cause signal dropout and reduce the sensitivity of the blood oxygen level dependent signal in echo-planar fmri imaging (34), there is no reason to suspect that BPD patients would be systematically more prone to this issue than control subjects, which as a group demonstrated consistent amygdala activation in response to negative emotions in this study. Additionally, a reduction of neural activity in the amygdala is more consistent with structural brain imaging studies, which show a robust 13% reduction in the volume of this structure bilaterally in BPD patients (35). The results of this ALE meta-analysis identified the right insular cortex as showing heightened activity in BPD patients as compared with control subjects. The insula is considered a critical structure involved in representing the degree of subjectively experienced negative emotion (36), especially disgust (37). This structure has efferent connections to many regions that showed significantly reduced activation in BPD patients, including the ACC, superior temporal cortex, and amygdala (38). The insula also has an efferent connection to the inferior frontal gyrus, which showed greater activation in BPD patients, and both afferent and efferent projections to regions of the PFC that had significantly reduced activity in BPD patients (38). Given its extensive interconnections with several neural regions responsible for the subjective representation and regulation of emotion, the insula might play a crucial role in modulating negative emotions in patients with this illness. Increased activation within the insula itself might reflect a more intense subjective experience of negative emotion in these patients, and through its connections with anterior brain regions, it might be involved in modulating the efficiency of regulatory processes involved in the top-down control of emotion. The subgenual ACC (BA 25) showed the most robust reduction in activity for BPD patients relative to control subjects. Indeed, this was the only area implicated in both study-reported between-subjects contrasts and independent subtraction ALE analyses on the basis of within-subjects comparisons of negative neutral task conditions. The subgenual ACC is thought to be a key component of a neural network that along with the amygdala, superior temporal cortex, and mid- and posterior cingulate cortex and other regions is involved in the evaluative, expressive, and experiential aspects of emotion (39). In particular, activity in the subgenual ACC corresponds closely with that observed in the amygdala under conditions of high emotional conflict, and activity in both regions is strongly linked to trait levels of negative emotionality, particularly anxiety (40). Therefore, reduced activation in the subgenual ACC and amygdala in BPD might signify a diminished capacity for the regulation of emotions, especially under conditions of high negative emotionality that might require cognitive control to reduce the subjectively experienced intensity of emotion (41). Because BPD patients show significant gray matter volume decreases in the ACC (42,43), it will be important for future work to statistically account for partial volume averaging effects in this region to determine whether these patients indeed show signs of reduced hemodynamic activity in this area. Nevertheless, these findings are noteworthy, because this region is thought to play a critical role in the etiology of major depressive disorder (MDD). Mayberg et al. (44) have consistently associated hyperactivity in this

158 BIOL PSYCHIATRY 2013;73:153 160 A.C. Ruocco et al. region during negative emotion processing tasks with MDD, particularly treatment-resistant depression. Interestingly, BPD patients show less activity in the subgenual ACC as compared with healthy individuals, a pattern of activity that lies in stark contrast to findings of greater activity in this region for patients with MDD. These findings, therefore, could suggest that reduced activity in this region might represent a distinct neural marker for BPD that could be used to distinguish this illness from patients with mood disorders (and no personality disorder). Indeed, these distinguishing neural underpinnings for BPD as compared with MDD might refute the notion that BPD could be better understood as a variant of a mood disorder rather than a bona fide and distinct psychiatric illness (45,46). Given that BPD is often comorbid with MDD, future research should directly contrast BPD patients with MDD versus MDD patients without a personality disorder to evaluate the specificity of these findings. The PFC has been the focus of many studies of BPD, because of its identified role in effortful emotion regulation (47,48) and response inhibition (49,50) in this illness. We found bilaterally reduced activation within the dorsolateral PFC in BPD patients relative to control subjects associated with negative emotion processing. Although the dorsolateral PFC is better known for its role in working memory and executive functioning (51), it might also play an important role in emotion regulation by way of its interaction with the amygdala, possibly by way of effortful cognitive strategies to modulate emotional experiences (52). Attenuated activity in the dorsolateral PFC in BPD likely reflects a diminished capacity for cognitive control in the modulation of subjectively experienced negative emotion. Increased activity in the left inferior frontal gyrus for BPD patients might also suggest a disruption of frontal systems involved in cognitive control, because this region is commonly associated with response inhibition (53). Heightened activity in this cortical region for BPD during negative emotion processing could denote a deficiency in inhibitory mechanisms involved in the modulation of emotion. These results also support neurocognitive findings of dorsolateral PFC and inferior frontal cortex dysfunction in BPD such as on tests of decision-making, cognitive flexibility, and motor response control (54). Taken together, the results of this study are consistent with the theory that BPD patients might show hyperarousal in response to negatively valenced emotional stimuli, which might be subserved, at least in part, by heightened activity in the insular cortex. This hyperarousal might interfere with attentional disengagement from emotionally salient stimuli via efferent projections from the insula to anterior brain regions (ACC, dorsolateral PFC) that show significantly reduced functional activation in BPD. The BPD patients might therefore show a diminished capacity to efficiently modulate the intensity of subjectively experienced negative emotions, perhaps as a result of aberrant reciprocal connections between the insula and PFC. This neural network might also underlie well-documented deficits in emotion perception in BPD, especially those involved in the recognition of facial expressions of anger and disgust that are particularly affected in this illness (55). The neural systems dysfunctions revealed in this study are also consistent with those subserving the capacity for mental state attribution (56), which is thought to be disrupted in BPD (57), and could underlie interpersonal deficits in this illness. Another potentially important implication of these findings is that neurobiologically based deficiencies in the regulation of negative emotions in BPD might conceivably be ameliorated by psychological treatments aimed at improving the topdown regulatory capacities of these patients (22,58). Several limitations should be considered as they relate to the current meta-analysis. First, there are a number of methodological and statistical concerns relating to the use of ALE for coordinatebased meta-analysis of neuroimaging data (for a review, see Eickhoff et al. [59]). Many concerns related to the assessment of spatial association between experiments and correcting for family-wise error and cluster-level significance level are addressed in the most recent version of the GingerALE BrainMap application (version 2.1), which was employed in the current meta-analysis. Second, to address the issue of variability across individual studies in the tasks used to evaluate negative emotion processing in BPD, we decided to use a broad definition of negative emotion paradigms (i.e., passive viewing, script-driven imagery), so as to synthesize as many individual studies on this topic as possible while also maintaining a reasonably narrow assessment of negative emotionality. In so doing, we acknowledge that our evaluation of this construct accommodates multiple aspects of negative emotionality (e.g., visual and aural stimuli, images of faces and scenes) and different types of instructional sets for emotional stimuli (e.g., passive viewing, recognition of emotions) that could be associated with distinct neural responses and might be obscured by pooling individual studies. Nevertheless, the concordance of results across studies using these diverse paradigms and instructional sets is striking and will help guide future work in this area. This study might also have limited generalizability, given the restrictions of the study selection criteria. The experimental paradigms examined in this meta-analysis might also not be generalizable to the real-life stressors experienced by BPD patients, although there are preliminary findings based on more ecologically valid social stressors relevant to BPD that are consistent with these results (60). Third, we were unable to separately evaluate the dimensions of valence and arousal for the emotion paradigms, because most studies reported results collapsing across (negative) emotional valence rather than arousal. Fourth, there are other potentially important brain regions not identified in this meta-analysis that require greater attention, because they might underlie aspects of emotion dysregulation in BPD. For example, the ventromedial/orbitofrontal PFC typically shows decreased activation in neuroimaging studies of BPD patients across a variety of task conditions (12,48,61 63) and might have abnormally reduced connectivity with the amygdala in these patients (64). Accordingly, future work should emphasize the complexity of the interactions among brain circuits subserving negative emotionality in BPD and confirm the results of the current study using multiple methods (e.g., structural and functional neuroimaging). Fifth, there was significant comorbidity of BPD and mood disorders in individual studies, which might call into question the specificity of these findings to BPD. There were, however, several areas of brain activation revealed in this study that diverged considerably from those typically seen in patients with mood disorders (e.g., lower activation in BA 25 as compared with MDD). It could be argued that the results of this study are perhaps more generalizable to the majority of BPD patients who show considerable diagnostic comorbidity, especially with mood disorders. Nevertheless, future work is needed to directly evaluate the specificity of these findings for BPD as opposed to other related disorders (e.g., mood disorders, other personality disorders) and understand how trauma might contribute to the neural substrates underlying affective processing in BPD (18,19). Finally, the limited number of studies in this area that used a consistent negative emotion paradigm restricted our capability to perform moderator analyses of potentially important task and sample characteristics. Given the characteristics of the studies included in this meta-analysis, it would be prudent to consider these findings as more representative of younger (approximately 30 years old) women with moderate to severe BPD. Future work should determine the extent to which these findings are applicable to men with

A.C. Ruocco et al. BIOL PSYCHIATRY 2013;73:153 160 159 BPD and the typically younger first-presentation patients (65) who are less-impacted by confounds relating to psychotropic medication use and other course-of-illness factors. These limitations notwithstanding, the results of the current study provide the first quantitative synthesis of functional neuroimaging findings in BPD and suggest that a dysfunction of specific frontolimbic systems might underlie negative emotionality in this illness. The authors report no biomedical financial interests or potential conflicts of interest. 1. Leichsenring F, Leibing E, Kruse J, New AS, Leweke F (2011): Borderline personality disorder. Lancet 377:74 84. 2. Lenzenweger MF, Lane MC, Loranger AW, Kessler RC (2007): DSM-IV personality disorders in the National Comorbidity Survey Replication. Biol Psychiatry 62:553 564. 3. Trull TJ, Solhan MB, Tragesser SL, Jahng S, Wood PK, Piasecki TM, et al. (2008): Affective instability: Measuring a core feature of borderline personality disorder with ecological momentary assessment. J Abnorm Psychol 117:647 661. 4. Henry C, Mitropoulou V, New AS, Koenigsberg HW, Silverman J, Siever LJ (2001): Affective instability and impulsivity in borderline personality and bipolar II disorders: Similarities and differences. J Psychiatr Res 35: 307 312. 5. Johnson PA, Hurley RA, Benkelfat C, Herpertz SC, Taber KH (2003): Understanding emotion regulation in borderline personality disorder: Contributions of neuroimaging. J Neuropsychiatry Clin Neurosci 15:397 402. 6. Rosenthal MZ, Gratz KL, Kosson DS, Cheavens JS, Lejuez CW, Lynch TR (2008): Borderline personality disorder and emotional responding: A review of the research literature. Clin Psychol Rev 28:75 91. 7. Phan KL, Wager T, Taylor SF, Liberzon I (2002): Functional neuroanatomy of emotion: A meta-analysis of emotion activation studies in PET and fmri. Neuroimage 16:331 348. 8. Vytal K, Hamann S (2010): Neuroimaging support for discrete neural correlates of basic emotions: A voxel-based meta-analysis. J Cogn Neurosci 22:2864 2885. 9. Herpertz SC, Dietrich TM, Wenning B, Krings T, Erberich SG, Willmes K, et al. (2001): Evidence of abnormal amygdala functioning in borderline personality disorder: A functional MRI study. Biol Psychiatry 50:292 298. 10. Donegan NH, Sanislow CA, Blumberg HP, Fulbright RK, Lacadie C, Skudlarski P, et al. (2003): Amygdala hyperreactivity in borderline personality disorder: Implications for emotional dysregulation. Biol Psychiatry 54: 1284 1293. 11. Beblo T, Driessen M, Mertens M, Wingenfeld K, Piefke M, Rullkoetter N, et al. (2006): Functional MRI correlates of the recall of unresolved life events in borderline personality disorder. Psychol Med 36:845 856. 12. Kraus A, Valerius G, Seifritz E, Ruf M, Bremner JD, Bohus M, et al. (2010): Script-driven imagery of self-injurious behavior in patients with borderline personality disorder: A pilot FMRI study. Acta Psychiatr Scand 121: 41 51. 13. Koenigsberg HW, Fan J, Ochsner KN, Liu X, Guise KG, Pizzarello S, et al. (2009): Neural correlates of the use of psychological distancing to regulate responses to negative social cues: A study of patients with borderline personality disorder. Biol Psychiatry 66:854 863. 14. Ruocco AC, Medaglia JD, Ayaz H, Chute DL (2010): Abnormal prefrontal cortical response during affective processing in borderline personality disorder. Psychiatry Res 182:117 122. 15. Mauchnik J, Schmahl C (2010): The latest neuroimaging findings in borderline personality disorder. Curr Psychiatry Rep 12:46 55. 16. New AS, Goodman M, Triebwasser J, Siever LJ (2008): Recent advances in the biological study of personality disorders. Psychiatr Clin North Am 31:441 461, vii. 17. Turkeltaub PE, Eden GF, Jones KM, Zeffiro TA (2002): Meta-analysis of the functional neuroanatomy of single-word reading: method and validation. Neuroimage 16:765 780. 18. Driessen M, Beblo T, Mertens M, Piefke M, Rullkoetter N, Silva-Saavedra A, et al. (2004): Posttraumatic stress disorder and fmri activation patterns of traumatic memory in patients with borderline personality disorder. Biol Psychiatry 55:603 611. 19. Kraus A, Esposito F, Seifritz E, Di Salle F, Ruf M, Valerius G, et al. (2009): Amygdala deactivation as a neural correlate of pain processing in patients with borderline personality disorder and co-occurrent posttraumatic stress disorder. Biol Psychiatry 65:819 822. 20. Lang S, Kotchoubey B, Frick C, Spitzer C, Grabe HJ, Barnow S (2012): Cognitive reappraisal in trauma-exposed women with borderline personality disorder. Neuroimage 59:1727 1734. 21. Koenigsberg HW, Siever LJ, Lee H, Pizzarello S, New AS, Goodman M, et al. (2009): Neural correlates of emotion processing in borderline personality disorder. Psychiatry Res 172:192 199. 22. Schnell K, Herpertz SC (2007): Effects of dialectic-behavioral-therapy on the neural correlates of affective hyperarousal in borderline personality disorder. J Psychiatr Res 41:837 847. 23. Krause-Utz A, Oei NY, Niedtfeld I, Bohus M, Spinhoven P, Schmahl C, et al. (2012): Influence of emotional distraction on working memory performance in borderline personality disorder. Psychol Med 1 12. 24. Kamphausen S, Schroder P, Maier S, Bader K, Feige B, Kaller CP, et al. (2012): Medial prefrontal dysfunction and prolonged amygdala response during instructed fear processing in borderline personality disorder [published online ahead of print March 9]. World J Biol Psychiatry. 25. Hazlett EA, Zhang J, New AS, Zelmanova Y, Goldstein KE, Haznedar MM, et al. (2012): Potentiated amygdala response to repeated emotional pictures in borderline personality disorder [published online ahead of print May 3]. Biol Psychiatry. 26. Driessen M, Wingenfeld K, Rullkoetter N, Mensebach C, Woermann FG, Mertens M, et al. (2009): One-year functional magnetic resonance imaging follow-up study of neural activation during the recall of unresolved negative life events in borderline personality disorder. Psychol Med 39: 507 516. 27. Eickhoff SB, Laird AR, Grefkes C, Wang LE, Zilles K, Fox PT (2009): Coordinate-based activation likelihood estimation meta-analysis of neuroimaging data: A random-effects approach based on empirical estimates of spatial uncertainty. Hum Brain Mapp 30:2907 2926. 28. Lancaster JL, Tordesillas-Gutierrez D, Martinez M, Salinas F, Evans A, ZilleS K, et al. (2007): Bias between MNI and talairach coordinates analyzed using the ICBM-152 brain template. Hum Brain Mapp 28:1194 1205. 29. Laird AR, Eickhoff SB, Fox PM, Uecker AM, Ray KL, Saenz JJ Jr, et al. (2011): The BrainMap strategy for standardization, sharing, and meta-analysis of neuroimaging data. BMC research notes 4:349. 30. Genovese CR, Lazar NA, Nichols T (2002): Thresholding of statistical maps in functional neuroimaging using the false discovery rate. Neuroimage 15:870 878. 31. McCloskey MS, Phan KL, Coccaro EF (2005): Neuroimaging and personality disorders. Curr Psychiatry Rep 7:65 72. 32. Wolf FM (1986): Meta-Analysis: Quantitative Methods for Research Synthesis. London: Sage. 33. Herpertz SC, Dietrich TM, Wenning B, Krings T, Erberich SG, Willmes K, et al. (2001): Evidence of abnormal amygdala functioning in borderline personality disorder: A functional MRI study. Biol Psychiatry 50:292 298. 34. Deichmann R, Josephs O, Hutton C, Corfield DR, Turner R (2002): Compensation of susceptibility-induced BOLD sensitivity losses in echoplanar fmri imaging. Neuroimage 15:120 135. 35. Ruocco AC, Amirthavasagam S, Zakzanis KK (2012): Amygdala and hippocampal volume reductions as candidate endophenotypes for borderline personality disorder: A meta-analysis of magnetic resonance imaging studies. Psychiatry Res 201:245 252. 36. Diekhof EK, Geier K, Falkai P, Gruber O (2011): Fear is only as deep as the mind allows: A coordinate-based meta-analysis of neuroimaging studies on the regulation of negative affect. Neuroimage 58:275 285. 37. Phillips ML, Williams LM, Heining M, Herba CM, Russell T, Andrew C, et al. (2004): Differential neural responses to overt and covert presentations of facial expressions of fear and disgust. Neuroimage 21:1484 1496. 38. Augustine JR (1996): Circuitry and functional aspects of the insular lobe in primates including humans. Brain Res Rev 22:229 244. 39. Drevets WC, Savitz J, Trimble M (2008): The subgenual anterior cingulate cortex in mood disorders. CNS Spectr 13:663 681. 40. Haas BW, Omura K, Constable RT, Canli T (2007): Emotional conflict and neuroticism: Personality-dependent activation in the amygdala and subgenual anterior cingulate. Behav Neurosci 121:249 256. 41. di Pellegrino G, Ciaramelli E, Ladavas E (2007): The regulation of cognitive control following rostral anterior cingulate cortex lesion in humans. J Cogn Neurosci 19:275 286. 42. Tebartz van Elst L, Hesslinger B, Thiel T, Geiger E, Haegele K, Lemieux L, et al. (2003): Frontolimbic brain abnormalities in patients with border-

160 BIOL PSYCHIATRY 2013;73:153 160 A.C. Ruocco et al. line personality disorder: A volumetric magnetic resonance imaging study. Biol Psychiatry 54:163 171. 43. Hazlett EA, New AS, Newmark R, Haznedar MM, Lo JN, Speiser LJ, et al. (2005): Reduced anterior and posterior cingulate gray matter in borderline personality disorder. Biol Psychiatry 58:614 623. 44. Mayberg HS, Lozano AM, Voon V, McNeely HE, Seminowicz D, Hamani C, et al. (2005): Deep brain stimulation for treatment-resistant depression. Neuron 45:651 660. 45. McGlashan TH (1983): The borderline syndrome. II. Is it a variant of schizophrenia or affective disorder? Arch Gen Psychiatry 40:1319 1323. 46. Sass H, Jünemann K (2003): Affective disorders, personality and personality disorders. Acta Psychiatr Scand 108:34 40. 47. Ruocco AC, Medaglia JD, Ayaz H, Chute DL (2010): Abnormal prefrontal cortical response during affective processing in borderline personality disorder. Psychiatry Res 182:117 122. 48. Schulze L, Domes G, Kruger A, Berger C, Fleischer M, Prehn K, et al. (2011): Neuronal correlates of cognitive reappraisal in borderline patients with affective instability. Biol Psychiatry 69:564 573. 49. Ruocco AC, McCloskey MS, Lee R, Coccaro EF (2009): Indices of orbitofrontal and prefrontal function in Cluster B and Cluster C personality disorders. Psychiatry Res 170:282 285. 50. Ruocco AC, Laporte L, Russell J, Guttman H, Paris J (2012): Response inhibition deficits in unaffected first-degree relatives of patients with borderline personality disorder. Neuropsychology 26:473 482. 51. Wagner AD, Maril A, Bjork RA, Schacter DL (2001): Prefrontal contributions to executive control: fmri evidence for functional distinctions within lateral Prefrontal cortex. Neuroimage 14:1337 1347. 52. Banks SJ, Eddy KT, Angstadt M, Nathan PJ, Phan KL (2007): Amygdalafrontal connectivity during emotion regulation. Soc Cogn Affect Neurosci 2:303 312. 53. Swick D, Ashley V, Turken U (2008): Left inferior frontal gyrus is critical for response inhibition. Bmc Neurosci 9:102. 54. Ruocco AC (2005): The neuropsychology of borderline personality disorder: A meta-analysis and review. Psychiatry Res 137:191 202. 55. Domes G, Schulze L, Herpertz SC (2009): Emotion recognition in borderline personality disorder-a review of the literature. J Pers Disord 23:6 19. 56. Ochsner KN, Knierim K, Ludlow DH, Hanelin J, Ramachandran T, Glover G, et al. (2004): Reflecting upon feelings: An fmri study of neural systems supporting the attribution of emotion to self and other. J Cogn Neurosci 16:1746 1772. 57. Fonagy P, Bateman A (2008): The development of borderline personality disorder a mentalizing model. J Pers Disord 22:4 21. 58. McMain SF, Guimond T, Streiner DL, Cardish RJ, Links PS (2012): Dialectical behavior therapy compared with general psychiatric management for borderline personality disorder: Clinical outcomes and functioning over a 2-year follow-up. AJ Psychiatry 169:650 661. 59. Eickhoff SB, Bzdok D, Laird AR, Kurth F, Fox PT (2012): Activation likelihood estimation meta-analysis revisited. Neuroimage 59:2349 2361. 60. Ruocco AC, Medaglia JD, Tinker JR, Ayaz H, Forman EM, Newman CF, et al. (2010): Medial prefrontal cortex hyperactivation during social exclusion in borderline personality disorder. Psychiatry Res 181:233 236. 61. Silbersweig D, Clarkin JF, Goldstein M, Kernberg OF, Tuescher O, Levy KN, et al. (2007): Failure of frontolimbic inhibitory function in the context of negative emotion in borderline personality disorder. AJ Psychiatry 164:1832 1841. 62. Schmahl CG, Vermetten E, Elzinga BM, Bremner JD (2004): A positron emission tomography study of memories of childhood abuse in borderline personality disorder. Biol Psychiatry 55:759 765. 63. Wolf RC, Thomann PA, Sambataro F, Vasic N, Schmid M, Wolf ND (2012): Orbitofrontal cortex and impulsivity in borderline personality disorder: An MRI study of baseline brain perfusion [published online ahead of print March 11]. Eur Arch Psychiatry Clin Neurosci. 64. New AS, Hazlett EA, Buchsbaum MS, Goodman M, Mitelman SA, Newmark R, et al. (2007): Amygdala-prefrontal disconnection in borderline personality disorder. Neuropsychopharmacology 32:1629 1640. 65. Whittle S, Chanen AM, Fornito A, McGorry PD, Pantelis C, Yucel M (2009): Anterior cingulate volume in adolescents with first-presentation borderline personality disorder. Psychiatry Res 172:155 160. 66. Guitart-Masip M, Pascual JC, Carmona S, Hoekzema E, Berge D, Perez V, et al. (2009): Neural correlates of impaired emotional discrimination in borderline personality disorder: An fmri study. Prog Neuropsychopharmacol Biol Psychiatry 33:1537 1545. 67. Minzenberg MJ, Fan J, New AS, Tang CY, Siever LJ (2007): Fronto-limbic dysfunction in response to facial emotion in borderline personality disorder: An event-related fmri study. Psychiatry Res 155:231 243. 68. Smoski MJ, Salsman N, Wang L, Smith V, Lynch TR, Dager SR, et al. (2011): Functional imaging of emotion reactivity in opiate-dependent borderline personality disorder. Personal Disord 2:230 241. 69. Wingenfeld K, Rullkoetter N, Mensebach C, Beblo T, Mertens M, Kreisel S, et al. (2009): Neural correlates of the individual emotional Stroop in borderline personality disorder. Psychoneuroendocrinology 34:571 586.