A voxel-based morphometric MRI study in female patients with borderline personality disorder

Abstract

Subtle prefrontal and limbic structural abnormalities have been reported in borderline personality disorder (BPD). In order to further validate the previously reported findings and to more precisely describe the nature of the structural change we performed a voxel-based morphometric (VBM) study in patients with BPD. Twenty female patients with BPD and 21 female healthy controls were investigated. High-resolution 3-D datasets were acquired and analyzed following an optimized protocol of VBM in the framework of statistical parametric mapping (SPM99). Gray matter volume loss was found in the left amygdala. No other differences in gray or white matter volume or density were found anywhere else in the brain. Our findings support the hypothesis that temporolimbic abnormalities play a role in the pathophysiology of BPD. Prefrontal structural alterations in BPD were not observed in this study.

Introduction

Borderline personality disorder (BPD) is characterized by a pattern of affective instability, impulsiveness, and self-injurious acts, dissociative symptoms, and unstable interpersonal relationships, including impulsive behavior (American Psychiatric Association, 1994). The etiology and pathogenesis of this severe and persistent condition are still unclear. It is commonly assumed that BPD involves developmental or acquired brain dysfunction associated with early traumatic experience (Skodol et al., 2002a). Both genetic predispositions and environmental factors are thought to be important for the etiology of BPD (Skodol et al., 2002b).

There has only recently been a growing interest in the neurobiological alterations underlying personality disorders. In the case of BPD, early studies of structural cerebral alterations resulted in negative findings. CT studies were unable to show abnormal ventricles or brain atrophy in BPD Lucas et al., 1989, Snyder et al., 1983. With the advent of MRI, there were a number of reports of volumetric changes: Lyoo and colleagues reported a marginally significant reduction in overall frontal lobe volumes in BPD (Lyoo et al., 1998), although this finding has been criticized for technical reasons, such as low spatial resolution and lack of correction for head tilt. Recently, Driessen and co-workers employed a high-resolution structural MRI methodology to investigate a large sample of patients with BPD and found a highly significant bilateral volume loss of the hippocampi and marginally significant bilateral volume reductions of the amygdalae, while the overall brain volume was not reduced (Driessen et al., 2000). In particular, hippocampal volume loss has also been reported in different patient samples with posttraumatic stress disorder Bremner et al., 1997, Bremner et al., 1995, Gurvits et al., 1996 and in women with a history of severe sexual abuse in childhood (Stein et al., 1997). Both posttraumatic stress disorder and a history of sexual abuse are known to be common in BPD. In a previous study of our group, Tebartz van Elst and colleagues were able to confirm Driessen's finding of bilateral reduced amygdala and hippocampal volumes using manual volumetry. In addition they found a highly significant volume reduction of the left orbitofrontal cortex and of the right anterior cingulate cortex (Tebartz van Elst et al., in press). This suggests a pattern of hippocampal, amygdala, and prefrontal volume loss which might be a specific feature of BPD. That previous study examined 8 patients with BPD and 8 healthy controls. In an attempt to extend that patient sample, we subsequently recruited 12 further patients and 13 further healthy controls. Thus the resulting sample of the study reported here consists of 20 patients and 21 healthy control subjects and includes a subgroup of 8 patients and 8 healthy controls that was volumetrically investigated in our former study (Tebartz van Elst et al., in press).

Neurochemically, so far only one study employed proton magnetic resonance spectroscopy in BPD: Tebartz van Elst and co-workers found subtle prefrontal pathology, reflected in reduced NAA concentrations, in the dorsolateral prefrontal cortex of patients with BPD (Tebartz van Elst et al., 2001).

Regarding functional investigations, there have been long-standing speculations about electrophysiological alterations in BPD because some features of the condition resemble those observed in patients with temporal lobe epilepsy or complex partial seizures. Accordingly, generalized slow activity in scalp EEG (De la Fuente et al., 1998) and dysfunction of auditory neurointegration as measured by auditory evoked potentials were found (Kutcher et al., 1987). Three studies employing ¹⁸F-labeled desoxyglucose positron emission tomography (FDG-PET) consistently produced evidence of frontal hypometabolism, particularly pronounced in prefrontal cortical areas De la Fuente et al., 1997, Goyer et al., 1994, Soloff et al., 2000. In an alpha-[(11)C]MTrp-PET study of patients with BPD, Leyton and co-workers observed reduced 5-HT synthesis in corticostriatal pathways and this 5-HT synthesis was negatively correlated with impulsivity (Leyton et al., 2001). In a recent study using fMRI with an emotional stimulation paradigm, Herpertz and colleagues found an increased signal in the amygdala bilaterally and in the medial and inferolateral prefrontal cortex (Herpertz et al., 2001).

In summary, most of the functional studies and one neurochemical imaging study report alterations in limbic and prefrontal areas in BPD. Two structural studies found evidence for amygdala and hippocampal volume loss, while two studies showed volume loss in the frontal lobes.

On the background of the results summarized above and in particular our own findings in a previous small sample of patients with BPD (Tebartz van Elst et al., in press), the aim of this study was to further investigate gray matter abnormalities in a larger sample of patients with BPD. More precisely, using the advanced technology of voxel-based morphometry (VBM), we wanted to test the hypothesis that based on our previous findings there is volume reduction and possibly loss of cortical gray matter density in the areas of the mesial temporal lobes (i.e., hippocampus and amygdala) and prefrontal areas (left orbitofrontal and right anterior cingulate cortex). The method of VBM, based on statistical parametric mapping (SPM99, Wellcome Department of Cognitive Neurology, University College, London) has several advantages compared to the manual morphometry used before (Ashburner and Friston, 2001). First, it is an objective, automated method to investigate the entire gray and white matter. Second, following the recently developed protocol of Good and colleagues (2001), it allows the analysis not only of changes in gray and white matter density but also of gray and white matter volume. Since spatial resolution and accuracy of image processing have advanced considerably over the last years, it is now possible to investigate even small cortical or subcortical structures. Finally, VBM has now been used in different patient populations and has been shown to be effective in detecting subtle morphological changes in different neuropsychiatric conditions, such as schizophrenia Hulshoff Pol et al., 2001, Wilke et al., 2001 or intermittent explosive disorder (Woermann et al., 2000).