Edited by: Tetsuo Kida, National Institute for Physiological Sciences (NIPS), Japan
Reviewed by: Shane McKie, University of Manchester, UK; Martin Driessen, Ev. Hospital Bielefeld, Germany
*Correspondence: Anna Buchheim
This is an open-access article distributed under the terms of the Creative Commons Attribution License (CC BY). The use, distribution and reproduction in other forums is permitted, provided the original author(s) or licensor are credited and that the original publication in this journal is cited, in accordance with accepted academic practice. No use, distribution or reproduction is permitted which does not comply with these terms.
Individuals with borderline personality disorder (BPD) are characterized by emotional instability, impaired emotion regulation and unresolved attachment patterns associated with abusive childhood experiences. We investigated the neural response during the activation of the attachment system in BPD patients compared to healthy controls using functional magnetic resonance imaging (fMRI). Eleven female patients with BPD without posttraumatic stress disorder (PTSD) and 17 healthy female controls matched for age and education were telling stories in the scanner in response to the Adult Attachment Projective Picture System (AAP), an eight-picture set assessment of adult attachment. The picture set includes theoretically-derived attachment scenes, such as separation, death, threat and potential abuse. The picture presentation order is designed to gradually increase the activation of the attachment system. Each picture stimulus was presented for 2 min. Analyses examine group differences in attachment classifications and neural activation patterns over the course of the task. Unresolved attachment was associated with increasing amygdala activation over the course of the attachment task in patients as well as controls. Unresolved controls, but not patients, showed activation in the right dorsolateral prefrontal cortex (DLPFC) and the rostral cingulate zone (RCZ). We interpret this as a neural signature of BPD patients’ inability to exert top-down control under conditions of attachment distress. These findings point to possible neural mechanisms for underlying affective dysregulation in BPD in the context of attachment trauma and fear.
Disturbances in the processing and regulation of emotions are core symptoms of borderline personality disorder (BPD; Leichsenring et al.,
Attachment theory provides a powerful framework for understanding links between close relationships, mental representations of attachment, and psychopathology (Westen et al.,
Insecure attachment patterns demonstrate risk for the maladaptive personality traits underlying BPD (Scott et al.,
Emotional vulnerability in BPD patients may result from a marked sensitivity to emotional stimuli, an impairment of emotion regulation, or both (Gunderson and Lyons-Ruth,
As emotion regulation is dependent on regions exerting cognitive control, like the dorsolateral prefrontal cortex (DLPFC) and medial prefrontal/anterior cingulate cortex (mPFC/ACC; for a review see Ochsner and Gross,
Moreover studies investigated brain activation during processing of autobiographical memory in BPD. One study examining unresolved as compared to resolved life events found, among other regions, increasing activation of amygdala and anterior cingulate cortex (Beblo et al.,
In sum, emotion regulation and their neural correlates are impaired in BPD patients; for example, amygdala responses are prolonged and enhanced in particular during the presentation of emotional stimuli (for a review see Buchheim et al.,
The functional neuroimaging studies summarized above measured brain activation patterns in response to visual stimuli (pictures, faces) or passively presented scripts. Several studies have investigated the neural correlates of “social” attachment (i.e., defined loosely as individuals in intimate relationships) in healthy populations. The main paradigm in these studies is the presentation of pictures of the beloved sexual partner or their own infant by contrasting familiar vs. non-familiar stimuli (Bartels and Zeki,
Buchheim et al. (
The present study investigated neural correlates of attachment narratives in borderline patients using the same paradigm. In this article, we focus on differences between “unresolved” and “resolved” subjects. Our design permitted analysis related to other linguistic features of the AAP, which were reported elsewhere (Buchheim et al.,
Thirteen female BPD inpatients were recruited from a psychiatric hospital (Psychosomatic and Psychiatric Hospital, Bad Wiessee, Germany) and compared to 21 healthy female volunteers recruited for the study by an advertisement in a local newspaper and leaflets distributed in the Hospital of the University of Ulm. The sample of the 17 healthy controls in this study included the 11 subjects of our pilot study (Buchheim et al.,
C Control ( |
B Borderline ( |
C×B effect size | Exact U-test | ||||
---|---|---|---|---|---|---|---|
Clinical scales | |||||||
GSI (SCL-90) | 0.22 | 0.51 | 3.34 | 4.432 | 0.000 | ||
Barrett impulsivity scale | 9.99 | 10.36 | 1.75 | 3.715 | 0.000 | ||
Dissociative experience scale | 3.91 | 16.31 | 1.02 | 3.390 | 0.001 |
Participants were administered with the fMRI-adapted version of the Adult AAP (Buchheim et al.,
AAP responses are classified on the basis of verbatim transcribed narratives. The coding system defines unresolved attachment as failure to contain frightening or threatening narrative material demonstrated, for example, by story elements representing attachment dysregulation, such as death, attack, abuse, or devastation (George et al.,
Resolved AAP story (Control) | Unresolved AAP story (Control) |
---|---|
“An elder man in the graveyard. The man is standing in front of his mother’s tombstone. As he accidentally visited his hometown he also visited the graveyard and lays down a bunch of flowers to his mother. He is thinking about the past, how things had been when she was still alive, what she had pointed out to him for his live. He is very centered upon the past remembering many things and at the same time he is gathering courage for the future since he knows that life is transient. He is keeping to this task for a while; then he returns to his apartment lost in thought. The next day he is leaving his hometown after having visited some of his old friends and some of mother’s neighbors to talk to”. | “A man is standing besides a grave. His wife has recently died. She died suddenly in a car accident. The man is |
“On a graveyard a man is standing by a grave he had been searching for many years. It’s the grave of his parent, who gave their son up for |
The psychometric properties of the AAP were established in an independent validity study with 144 subjects from a healthy community sample of non-patient subjects. This study demonstrated strong psychometric validity, including meeting psychometric standards for inter-judge reliability, test-retest reliability (after 3 months), discriminant validity, and convergent construct validity with another validated developmental attachment narrative assessment, the AAI (George et al.,
In the present study, two blind, reliable AAP judges independently coded the AAP narratives of the stories that participants told in the scanner. There was unresolved-resolved agreement in 27 out of 28 cases (96%). The resulting inter-rater agreement was κ = 0.93, corresponding to “almost perfect agreement” by Landis and Koch (
Subjects were instructed in the AAP story telling task before they entered the scanner using two non-AAP “neutral” (i.e., not attachment related scenes) pictures. The pictures were same size and drawn in the same style as the stimuli in the AAP set. The goal was for subjects to understand the scope of the probes normally asked during in-person administration. The training procedure was repeated two more times, as needed to achieve this goal. The fMRI acquisition started with the original AAP picture presentation and instructions. During scanning, subjects were visually presented the standard AAP instruction at the introduction of the stimulus (“what led up to that scene, what are the characters thinking or feeling, and what might happen next?”) for 10 s and a fixation cross for 10 s. Afterwards each AAP picture was presented (120 s) along the original order of the measure (described in the “Materials and Methods” Section). Subjects were asked to talk about the picture for at least 2 min. A fixation cross was shown for 15 s after the picture presentation until beginning a new cycle of instruction and the next picture presentation. The total procedure included nine pictures, the two neutral and seven standard AAP attachment scenes.
1.5 Tesla Siemens Magnetom Symphony scanner (Siemens, Erlangen, Germany), image size: 64 × 64 voxels, FOV of 192 mm, slice thickness 4 mm with 1 mm gap, 25 slices covering the whole brain, TE/TR 40 ms/2500 ms, total acquisition time 25 min (=598 volumes, one session). Instructions and pictures were shown with fMRI compatible video-goggles (Resonance Technologies, Northridge, CA, USA). Speech was digitally recorded beginning at the onset of each picture using an fMRI compatible microphone and saved digitally on a computer using Cool Edit Pro (Syntrillium Software Cop. Phoenix, Arizona). Head movement was minimized by using padded earphones fixating the head within the gradient insert coil.
Group differences of the behavioral attachment data were analyzed using the exact Mann-Whitney U-test and the Kruskal-Wallis H-test and (SPSS version 14). We used non-parametric tests because of the non-normal distribution of the dependent variables. Preprocessing and statistical analysis of fMRI data were carried out with SPM2
The effects of interests in this study were narrative story responses to the seven attachment pictures. We calculated the contrast picture presentation for each subject during speech (regressor 2) + picture presentation before the subjects start to speak (regressor 1) vs. baseline (fixation cross), thereby including potential mental processes before the actual speaking phase starts. For each subject, contrasts for single pictures were calculated, that is, seven contrasts for the attachment pictures ordered 1–7.
A within subject repeated measures ANOVA with three groups (resolved healthy controls, unresolved healthy controls and unresolved BPD patients) and seven repeated measurements was calculated as a second level analysis in order to test for effects of group and attachment classification. The AAP is designed to activate the attachment system increasingly from picture 1–7. The contrast of interest within each group was labeled “AAP effect” (–3 –2 –1 0 1 2 3), in accordance with the results of our pilot study (Buchheim et al.,
Borderline patients differed significantly from controls in all clinical scales (Table
Both control and patient groups showed activations in bilateral occipital cortex, bilateral superior medial frontal cortex, bilateral precentral and left inferior frontal gyrus for the main effect of picture presentation and bilateral medial and superior temporal gyrus, bilateral precentral gyrus, medial occipital gyrus and cerebellum for the main effect of speech production.
The goal of the neuroimaging analysis was to examine the AAP effect (increasing attachment activation during the task) in control and borderline subjects. The first analysis examined the AAP effect for the control group. Three regions showed increasing activation during the task: right amygdala, right DLPFC and mPFC in the rostral cingulate zone (RCZ; Table
Region | BA | |||||
---|---|---|---|---|---|---|
Amygdala | R | 4.51 | 24 | −3 | −27 | |
Dorsolateral prefrontal cortex | 46 | R | 3.54 | 48 | 24 | 15 |
Medial prefrontal cortex | 9 | 4.09 | −3 | 33 | 33 | |
Superior temporal sulcus | 39 | L | 4.24 | −51 | −60 | 21 |
Amygdala | R | 4.60 | 24 | 3 | −27 | |
Dorsolateral prefrontal cortex | 46 | R | 3.71 | 48 | 27 | 15 |
Superior frontal gyrus | 6 | R | 4.26 | 21 | 15 | 63 |
Amygdala | R | 4.42 | 24 | 3 | −24 | |
Superior temporal sulcus | 39 | R | 3.87 | 57 | −51 | 12 |
Medial prefrontal cortex | 8 | R | 3.85 | 3 | 21 | 45 |
Amygdala | R | 3.66 | 21 | −6 | −21 | |
Anterior cingulate cortex | 32 | R | 3.70 | 3 | 21 | 33 |
Cingulate gyrus | 24 | R | 3.58 | 3 | −18 | 42 |
Superior temporal sulcus | 39 | R | 3.99 | 60 | −57 | 15 |
Medial prefrontal cortex | 8 | L | 2.92*,# | −6 | 36 | 39 |
Medial prefrontal cortex | 8 | L | 2.83* | −6 | 36 | 39 |
Amygdala | R | 3.15*,# | 27 | 3 | −21 | |
Amygdala | R | 3.41*,# | 27 | 3 | −24 | |
Dorsolateral prefrontal cortex | 46 | R | 2.51* | 48 | 27 | 15 |
Amygdala | R | 2.88*,# | 21 | −6 | −21 | |
Dorsolateral prefrontal cortex | 46 | R | 3.11*,# | 48 | 24 | 15 |
Medial prefrontal cortex | 8 | L | 2.44* | −6 | 36 | 39 |
The results of these analyses are shown in Table
In order to check that our results are not solely due to
The present study examined the neural correlates of attachment dysregulation in a group of BPD patients compared to controls. This study used a paradigm that evaluated neural response patterns while subjects told attachment stories in response to AAP stimuli in the fMRI scanner. The fMRI analysis model followed the logic of the design of the attachment measure. According to this logic, the picture presentation sequence increasingly activates the attachment distress. We labeled this increasing activation over the course of task as “AAP effect.” Due to the fact that almost all BDP patients were classified as unresolved, we investigated only three classification groups: resolved controls, unresolved controls and unresolved BDP patients. There were three main imaging finding. First, all unresolved subjects (borderline and controls), but not the resolved controls, showed the AAP effect reflected in an increasing amygdala activation. Second, all controls (resolved and unresolved), but not the unresolved patients, showed the AAP effect in relation to increasing activation of the RCZ. Third, only the unresolved controls showed an AAP effect of increasing activation of the right DLPFC. This effect was not found in the resolved controls or the unresolved patients. We now discuss these results in the context of attachment research.
The predominant unresolved classification in the BPD patients was consistent with previous research (Fonagy et al.,
As shown in Figure
Increased amygdala activation in BPD patients has been found in a variety of passive stimulation paradigms and these findings are interpreted as heightened emotional sensitivity to aversive stimuli (Davis and Whalen,
The RCZ showed the AAP effect in both control groups, resolved or unresolved. Studies in humans showed that the RCZ (the posterior MFC border zone between the medial areas BA8, BA6 and BA32’ with some extension into BA24’) is involved in monitoring for unfavorable outcomes, performance and conflict monitoring, and decision uncertainty (Carter et al.,
The AAP effect in the DLPFC was found only in the unresolved control group; it was not observed in the resolved controls or the unresolved patients. The right DLPFC has been described as being involved in cognitive control (Duncan and Owen,
BPD patients were showing the highest proportion of unresolved attachment patterns in this sample. The fMRI findings suggest that they were neither able to recruit the DLPFC (cognitive control) nor the RCZ (conflict monitoring) while being emotionally overwhelmed, as reflected in the enhanced amygdala activation. However this finding contrasts with some published neuroimaging studies of affective dysregulation in BPD, which used passive stimulation. Studies on functional correlates of response inhibition have demonstrated further evidence for functional impairments of prefrontal areas, particularly of the DLPFC, the rostral ACC, and the orbitofrontal cortex (OFC). Minzenberg et al. (
There are several limitations to our study that need to be taken into account when interpreting our findings. First, the number of resolved BPD patients (
Second, we did not include a clinical control group. This leads to questions as to whether our results may also be present in patients with other psychiatric disorders and not specific to BPD.
Third, 5 out of 11 of the BPD subjects were under low dose medication. This could be a confounding factor in comparing patients and controls, although our control analyses of medication-free patients only speaks against this assumption.
Fourth, this study aimed to characterize differences between healthy participants and BPD patients using an attachment paradigm along the AAP system focusing on attachment related pictures only. Since this study has identified differences between controls and patients, it is rather difficult to interpret the data univocally in the absence of an adequate control stimulus set, which is the focus of a recent study with healthy controls (Labek et al.,
Fifth, all results were derived from the three group analyses. However, this might have inflated statistical results. Therefore, we also calculated separate analyses for those contrasts where only two groups were involved. All results remain significant, showing that increased degrees of freedom did not generally inflate our results. Only mPFC activation in the contrast comparing resolved controls and unresolved patients does not surpass additional small volume correction.
Finally, overt speech in the scanner always is accompanied with head movements. The head movement in this study was less than 2 mm, and we took steps to eliminate residual influences. These included dropping subjects, including movement parameters as a covariate of no interest, and modeling the onset of every spoken word.
Unresolved controls, but not patients, showed activation in the right DLPFC and the RCZ. We interpreted this as a neural signature of BPD patients’ inability to exert top-down control under conditions of attachment distress. These findings point to possible neural mechanisms for underlying affective dysregulation in BPD in the context of attachment trauma and fear. We found that both increased emotional sensitivity, as well as impaired emotion regulation, may have contributed to affective dysregulation in unresolved BPD patients. Increased emotional sensitivity, as evidenced by an AAP effect in the right amygdala, might be explained in large part by their attachment pattern. An alteration of the cognitive control system (RCZ and right DLPFC) is found when unresolved attachment and diagnosis of BDP are present simultaneously. Modulation of BPD patients’ responses in attachment situations during treatment and psychotherapy to patterns similar to those described in the control group might be an important indication of their increasing capacity to regulate attachment distress and to show sufficient cognitive control (Perez et al.,
The study was conceptualized by HW, AB, SE, CG, HK and MS. The study setup and data collection were organized and conducted by HW, AB, SE, MS, HK and PM. fMRI analyses was performed by HW and SE. Coding of attachment interviews were conducted by AB and CG. SE, HW and DP performed the statistical data analysis and contributed substantially to the result interpretation. HW, AB, SE, CG, and DP provided important intellectual contribution in commenting and revising the manuscript. AB, HW, SE, and CG wrote the manuscript and edited its final version.
The authors declare that the research was conducted in the absence of any commercial or financial relationships that could be construed as a potential conflict of interest.
We thank Edgar Schilly and Marco Jahn, Department of Psychiatry, University of Ulm, for technical assistance in fMRI data analysis, Kathrin Brändle, Department of Psychiatry and retain Department of Diagnostic Radiology, University of Ulm for assistance in fMRI-acquisition, Dipl.-Psych. Claudia Simons and Justice Krampen, Department of Psychosomatic Medicine and Psychotherapy, University of Ulm for translating the German transcripts into English. We thank Dipl.-Psych Dagmar Pape, Munich, for administering the Adult Attachment Interviews in the hospital and Prof. Dr. Fabienne Becker-Stoll, Munich, a certified judge, for classifying the Adult Attachment Interviews.
Adult Attachment Interview
anterior cingulate cortex
Adult Attachment Projective Picture System
Brodmann area
Borderline personality disorder
controls
dorsolateral prefrontal cortex
International Personality Disorder Examination
medial prefrontal cortex
orbitofrontal cortex
patients
prefrontal cortex
posttraumatic stress disorder
rostral cingulate zone
Structured Clinical Interview for DSM-IV.
1PM has been trained and certified for reliable diagnosis in the SCID-rating by Prof. Wittchen, Munich, Germany and by A. Loranger MD, New York, NY, USA regarding IPDE. In seven patients an experienced Master’s level psychologist conducted a second SCID interview. Agreement between the raters was κ = 1.0 for both BPD and lifetime depressive episode.
2