Amygdala Resting State Connectivity Differences Between Bipolar II and Borderline Personality Disorders

Abstract

Background:

Borderline personality disorder (BPD) and Bipolar II Disorder (BD II) have significant clinical overlap, leaving the potential for diagnostic inaccuracies and inadequate treatment recommendations. However, few studies have probed for clinical and neurobiological differences between the two disorders. Clinically, some prior studies have linked BPD with greater impulsivity and more frequent negative affective shifts than BD II, whereas previous neuroimaging studies have highlighted both similar and distinct neural abnormalities in BPD and BD II. Notably, no prior study has specifically targeted cortico-limbic neural differences, which have been hypothesized to underlie these core clinical differences.

Methods:

Individuals with BPD (n = 14) and BD II (n = 15) completed various clinical measures and a RSFC scan at 3T. Whole-brain amygdala resting state functional connectivity (RSFC) were compared between the two groups.

Results:

Relative to the BD II group, BPD participants reported significantly higher levels of impulsivity, trait anxiety, more frequent negative affective shifts, greater interpersonally reactive affective instability, lower overall functioning, and were characterized by lower amygdala-middle frontal gyrus RSFC. Lower amygdala –middle frontal gyrus RSFC was associated with greater impulsivity, trait anxiety, affective shifts, interpersonal affective reactivity, and functional impairment.

Limitations:

The current study consisted of small sample sizes and lacked a control group.

Conclusions:

This preliminary study suggests that amygdala-frontal RSFC may distinguish BPD from BD II. These results may guide future work aimed at identifying neural markers that can help disentangle these two disorders, leading to greater diagnostic accuracy and appropriate treatment implementation.

Introduction

The relationship between Borderline Personality Disorder (BPD) and Bipolar II Disorder (BD II) remains the subject of debate (1). Some researchers have suggested that there is significant overlap between the two disorders (2), whereas others have asserted that the two disorders have distinctly different clinical phenomenologies, courses, etiologies, and responses to treatment (3). Highlighting such differences, these researchers have emphasized that inaccurately diagnosing borderline patients with bipolar disorder may lead to unnecessary and ineffective treatment with pharmacotherapy and may prevent borderline patients from receiving appropriate psychotherapeutic treatments.

Regarding potential differences in the clinical phenomenology of BPD vs. BD II, BPD has been associated with higher levels of impulsivity compared to BD II (4), and affective instability in BPD has been found to be both qualitatively and quantitatively different from that in BD II (5,6). In particular, BD II patients may experience more intense positive affect, whereas BPD patients may have more frequent negative affective shifts as well as overall greater self-reported maladaptive emotion regulation strategies (5–7). Additionally, BPD patients report more interpersonally reactive affective instability compared to those with BD II (6). This suggests that those with BPD may demonstrate greater impairments in self-regulation compared to those with BD II, which may also be reflected by distinct neural substrates.

The neural signatures of these self-regulation deficits likely involve aberrant functioning of cortico-limbic neural circuits, as higher levels of impulsivity and difficulties with regulating negative emotions have been associated with weaker amygdala-prefrontal functional connectivity (8–10). These circuits are also consistently implicated in both BPD and BD II. Specifically, a meta-analysis found that individuals with BPD show stronger amygdala responses and less recruitment of the dorsolateral prefrontal cortex when processing negative emotional information compared to healthy controls (11). Most studies examining the neural correlates of aberrant emotion processing in BD have focused on BD I or combined BD I and BD II groups. However, a study exclusively targeting BD II found that BD II participants in a depressed state relative to healthy controls showed blunted amygdala-ventrolateral prefrontal cortex, amygdala-orbitofrontal cortex, and amygdala-dorsolateral prefrontal cortex functional connectivity during an emotion processing task (12).

Despite the clinical significance of being able to accurately diagnose BPD versus BD II, which can often be difficult due to the common overlap in clinical presentation, very few studies have directly compared similarities and differences in neural functioning between these two disorders. One study comparing the two disorders using an emotional Stroop task found that both BD (bipolar subtype was not specified) and BPD groups exhibited decreased dorsolateral prefrontal cortex activity and increased ventral lateral prefrontal cortex activity compared to healthy controls. However, the two clinical groups also demonstrated differences, with the BD group showing greater dorsomedial prefrontal cortex activity vs. the BPD group, and the BPD group showing less amygdala activity vs. the BD group (13). Together, these studies suggest that both BD II and BPD are characterized by impaired cortico-limbic functioning during processing of negative emotional information and that these disorders may have both overlapping and distinct patterns of cortico-limbic dysregulation.

While differences in the types of task and stimuli used to probe emotion processing deficits in these two disorders may lead to heterogeneous results, researchers have circumvented these issues by using resting state functional magnetic resonance imaging (fMRI) to examine cortico-limbic functioning. This allows for the examination of pervasive cortico-limbic alterations that may be independent of symptom provocation. Resting state studies comparing individuals with BPD vs. healthy controls have found that individuals with BPD show lower amygdala-medial prefrontal cortex, amygdala-ventral anterior cingulate cortex, and amygdala-orbitofrontal cortex RSFC (8, 14). One of these studies also found that lower amygdala-medial prefrontal cortex RSFC was associated with higher levels of self-reported impulsivity and emotion dysregulation (14). However, others have failed to find reduced amygdala-medial prefrontal cortex RSFC in BPD (15), and one study has reported increased connectivity between the amygdala and anterior cingulate cortex (16).

With respect to BD II, mirroring the task-based fMRI literature, most resting state studies have focused on BP I disorder exclusively or combined BP I and BD II disorder groups, despite documented clinical and neurobiological differences between the two disorder subtypes (e.g., 17, 18). Of the few studies focusing on BD II disorder, one study specifically probing amygdala RSFC failed to find amygdala-prefrontal RSFC differences compared to healthy controls (19), while other studies have targeted other large-scale networks outside of the amygdala (20–22). Specifically, these studies have found that individuals with BD II show greater temporo-insular network connectivity, but decreased cerebellar-central executive network connectivity (20), cerebellar-default mode network connectivity (22), and medial prefrontal voxel-mirrored homotopic connectivity (21). To date, only one RSFC study has compared individuals with BPD vs. BD disorder, and found that individuals with BD (bipolar subtype was not specified) showed increased connectivity between a number of large networks compared to those with BPD, including social salience (dorsal anterior cingula cortex, anterior insula)-frontoparietal (dorsolateral prefrontal cortex, lateral parietal cortex) networks, social salience-precuneus networks, social salience-ventromedial prefrontal cortex networks, and default mode (medial prefrontal cortex, posterior cingulate cortex)-precuneus networks (23). However, this study did not assess group differences in amygdala-based networks.

Given the sparse number of studies directly comparing BPD vs. BD II disorder, the goal of this study was to replicate and extend prior work by assessing differences in clinical phenomenology as well as probing potential group differences in amygdala RSFC. Prior work has shown that individuals with BPD exhibit greater levels of impulsivity and have more frequent affective shifts (4–6), and disruptions in cortico-limbic neural circuits have been demonstrated to underlie these deficits (8–10). Thus, we hypothesized that individuals with BPD would show weaker amygdala-frontal cortex RSFC compared to the BD II group. Additionally, we predicted that symptomatology characteristic of BPD (e.g. greater impulsivity) would correlate with reduced amygdala-frontal cortex RSFC.

Methods

Participants and Procedures

Participants were 29 females between the ages of 18 and 45, assessed as part of an fMRI study examining neural correlates of affective instability in BPD and BD II. Participants were recruited through Craig’s List and referrals from clinicians affiliated with McLean Hospital. After providing informed written consent to a protocol approved by the Partners Health Care IRB, participants were evaluated using the Revised Diagnostic Interview for Borderlines (DIB-R) (24), the borderline module of the Diagnostic Interview for DSM-IV Personality Disorders (DIPD-IV) (25), and the Structured Clinical Interview for DSM-IV Disorders (SCID-IV) (26). To be eligible for the study, BD II participants had to meet criteria for BD II disorder as assessed by the bipolar module of the SCID-IV, and BPD participants had to have a DIB-R score of ≥ 8 and meet DSM-IV criteria for BPD. Exclusion criteria included: 1) lifetime diagnosis of organic mental disorder or psychotic disorder (schizophrenia, schizoaffective disorder, or mood disorder with psychotic features); 2) diagnosis of substance dependence or alcohol dependence; 3) current alcohol use exceeding 14 drinks per week and/or binge alcohol use; 4) substance abuse within the last 6 months; and 5) meeting diagnostic criteria for both BPD and BD. After establishing study eligibility during a screening session, participants were invited for a separate imaging session, which took place at the McLean Imaging Center.

Demographic and Clinical Measures

Prior to completing neuroimaging, participants were assessed with a series of clinical instruments, including the Affective Lability Interview for Borderline Personality Disorder (ALI-BPD), the Impulsivity-Venturesomeness-Empathy Questionnaire (IVE Questionnaire), the Barratt Impulsivity Scale (BIS-II), the Beck Depression Inventory (BDI-II), the Montgomery-Asberg Depression Rating Scale (MADRS), the Young Mania Rating Scale (YMRS), and the Spielberger State-Trait Anxiety Inventory (STAI). In addition, study staff provided a Global Assessment of Functioning Score (GAF) for each participant (27) and rated the socioeconomic status of each subject using the Hollingshead-Redlich Scale (28). The ALI-BPD is a 9-item clinician-clinician administered instrument designed to measure different dimensions of affective instability over the last week and been shown to differentiate BPD from BD (5). The instrument asks the interviewer to rate affective changes in each dimension in terms of both frequency and intensity, including the percentage of affective changes that are interpersonally reactive on a 5-point scale ranging from 0 (0–10% of the time) to 4 (90–100% of the time). This instrument has been shown to differentiate BPD from BD. The IVE Questionnaire is a 63-item self-report questionnaire that provides discrete measures of impulsivity, venturesomeness, and empathy (29). The BIS-II is a 45-item self-report measure that is commonly used to measure impulsivity which includes 3 subscales: Cognitive Impulsivity, Motor Impulsivity, and Non-Planning Impulsivity (30). The BDI-II is a 21-item self-report measure that assesses depression severity over the last two weeks. It has been widely used to measure depression and has been used in both clinical and non-clinical samples (31). The MADRS is a 10-item clinician-rated scale that assesses a range of depressive symptoms and has shown to be an efficient and practical measure of depression (32). The YMRS is an 11-item clinician administered measure of manic symptom severity (33). The STAI is a self-report measure widely used to measure anxiety in adults; it includes two 20-item scales, one measuring the more temporary condition of “state” anxiety and the other measuring the more general and long-standing quality of “trait anxiety” (34).

Imaging Acquisition

Each participant completed a six minute and forty-seven second resting state scan. Imaging datasets were acquired on a Siemens Tim Trio 3.0 Tesla MRI scanner with a 32-channel head coil. Resting state functional data were acquired using a T2*-weighted gradient-echo, echoplanar pulse sequence. We collected 41 contiguous coronal slices (TR = 2.5 s, TE = 30 ms; FOV = 224 × 224 mm, matrix = 64 × 64, slice thickness = 3.5 mm). A high-resolution T1-weighted anatomical was also collected and served as an anatomical map for the functional images (TR = 20 s, TE = 2.15 ms, FOV = 240 mm; matrix = 166 × 256 mm; flip angle = 12°, slice thickness = 1.3 mm).

fMRI Analysis

The first 5 seconds of each participants resting state functional data were dropped to allow for magnetic field stabilization. The resting state functional data were preprocessed in SPM12, according to the following steps: slice-time correction, realignment, normalization to Montreal Neurological Institute (MNI) space and resampling to 2 × 2 × 2 mm voxels and smoothing with a 6-mm kernel. The artifact detection tool box was used to identify outlier data points, which were defined as volumes that exceeded a global mean intensity of 3 standard deviations away from the mean intensity across functional runs, or a composite threshold of 0.5 mm framewise displacement.

Additional pre-processing steps and whole-brain resting state functional connectivity analyses were conducted using the CONN toolbox (35). Physiological noise from white matter and cerebrospinal fluid was estimated and regressed out for each participant using the CompCor method (36). In a first-level general regression model, detrending, modeling of outlier images along with the three translation and three rotation parameters, plus one composite motion parameter indexing the maximum scan-to-scan movement and the ComCor corrections were conducted simultaneously. Next, a temporal band-pass filter 0.008–0.09 Hz was applied to the time series. The time series was extracted from a right and left amygdala seed region. The amygdala seed regions were taken from the Harvard-Oxford maximum likelihood probabilistic atlas provided in the CONN toolbox (37–39). The time series of the left and right amygdala were separately correlated with each voxel in the brain. The whole-brain correlation maps (r) were subsequently normalized using a Fischer’s z transformation. Finally, these normalized correlation maps were used to calculate all group-level statistics.

Statistics

Wilcoxon Rank Sum Tests or chi-square analyses were used to evaluate potential between-group baseline demographic variables and clinical history variables. Wilcoxon’s Rank Sum Tests were used to assess between-group differences in: 1) mean YMSRS scores; 2) mean MADRS scores; 3) mean STAI-State Anxiety and STAI-Trait Anxiety scores; 4) mean scores on the impulsivity, venturesomeness, and empathy scales of the IVE Questionnaire; 5) mean overall and individual subscale scores of the BIS-11; and 6) ALI borderline frequency and intensity subscale scores and scores on the ALI item measuring reactivity of mood. Cohen’s d values were computed to evaluate effect sizes of putative group differences. With respect to the imaging analyses, two independent sample t-tests were conducted to compare differences between the group with BPD versus the BD II group on right and left amygdala – whole-brain resting state functional connectivity. All RSFC results were considered significant if they passed a voxel threshold p < 0.005 cluster corrected to a family-wise error rate of p < 0.05. Additionally, we conducted correlations between group differences in amygdala RSFC and clinical measures.

Results

Demographics and Clinical Characteristics

Fourteen participants met study criteria for BPD and fifteen participants met study criteria for BD II. A participant from the BPD group was dropped from all imaging analyses due to a brain abnormality. Table 1 summarizes demographic and clinical characteristics for each study group. There were no significant differences between groups with respect to age, education, or socioeconomic status. The BD II group had a significantly higher mean global assessment of functioning score than the BPD group, but both groups had mean scores within the range specifying “moderate symptoms” or “moderate impairments in social, occupational, or school functioning.” Several participants in each group were taking psychiatric medications (stimulants, antidepressants, mood stabilizers, or anxiolytics), but groups did not differ with respect to medication usage. Significantly more participants in the BPD group had a history of psychiatric hospitalization (64.3% vs. 13.3%, Chi squared = 9.12, p < 0.01). For the BPD group, past history of major depression (n=1), post-traumatic stress disorder (n=1), eating disorder (n=1), panic disorder without agoraphobia (n=1), and panic disorder with agoraphobia (n=2) were reported. In the BD II group, past history of obsessive-compulsive disorder (n=1), eating disorder NOS (n=1), binge eating disorder (n=1), body dysmorphic disorder (n=1), and agoraphobia without panic (n=1) were reported.

Table 1:

Demographic and Clinical Variables

Variable	BD II		BPD		P Value
	Mean	SD	Mean	SD
Age	28.13	7.19	26.43	6.65	0.40
SES	2.20	0.676	2.43	1.22	0.89
GAF	57.93	4.30	51.71	3.93	<0.001
Education	15.79	1.25	14.93	1.77	0.71
	% (N)		% (N)
Treatment
Ever Psychiatrically Hospitalized	13.3 (2)		64.3 (9)		<0.01
Ever Therapy	100 (15)		92.9 (12)		0.13
Current Medication (%)
Stimulant	13.3 (2)		21.4 (3)		0.94
Antidepressant	26.6 (4)		57.1 (8)		0.10
Mood Stabilizer	53.3 (8)		35.7 (5)		0.34
Antipsychotic	6.7 (1)		0 (0)		0.33
Anxiolytic	13.3 (2)		14.2 (2)		0.94

Clinical Measures

As detailed in Table 2, significant group differences were found on the IVE-Impulsiveness subscale, the STAI-Trait Anxiety, ALI-Frequency subscale, the ALI-Reactivity subscale, and GAF scores. Mean scores on the impulsivity scale of the IVE were almost twice as high for subjects with BPD compared to subjects with BD II (14.3 vs. 8.7; p=0.001; Cohen’s d value: 1.47). Additionally, individuals with BPD reported ~18% higher trait anxiety levels (STAI-Trait Anxiety) compared to those with BD II (57.5 vs. 47.9; p=0.014; Cohen’s d value: 1.00). Moreover, those with BPD endorsed ~39% more frequent affective shifts (ALI-Frequency; 8.2 vs. 5.5; p=0.03; Cohen’s d value: 0.70) and reported twice as high interpersonal affective reactivity (ALI-Reactivity; 2.5 vs 1.3; p=0.006; Cohen’s d value: 1.05). There was also a non-significant trend for individuals with BPD to report more severe depressive symptoms (BDI-II, p = 0.08, Cohen’s d value: −0.74). However, there were no significant group differences on the YMRS, MADRS, HAM-A, STAI-State Anxiety, BIS-II, IVE-Empathy, IVE-Venturesomeness, and ALI-Intensity scales.

Table 2:

Clinical Phenomenology Scores in Bipolar II and Borderline Subjects

	BD II		BPD
Variable	Mean	SD		SD	Z-Score	P-Value	Cohen’s d
YMRS	2.5	2.2	2.4	1.9	0.13	0.91	0.09
MADRS	9.0	5.8	12.2	10.2	0.62	0.56	−0.40
Beck Depression Inventory II	14.3	8.9	22.9	13.9	1.73	0.08	−0.74
State Trait Anxiety Scale (STAI-1, state)	36.8	9.7	42.6	13.0	1.29	N.S.	−0.51
State Trait Anxiety Scale (STAI-2, trait)	47.9	10.6	57.5	8.5	2.43	0.014	−1.00
ALI Borderline Frequency Scale	5.5	3.8	8.2	3.9	2.12	0.03	−0.70
ALI Borderline Intensity Scale	6.2	3.9	6.9	3.2	0.99	0.29	−0.20
ALI Reactivity Item	1.3	1.0	2.5	1.1	2.83	0.006	−1.20
IVE Impulsivity Scale	8.7	3.8	14.3	3.7	3.31	0.001	−1.53
IVE Venturesomeness Scale	9.7	3.2	10.2	3.5	0.29	0.35	−0.14
IVE Empathy Scale	13.3	3.7	13.9	3.9	0.79	0.34	−0.16
BIS-11 Total	69.7	12.9	75.6	12.8	0.85	0.11	−0.46
BIS-11 Attention	18.5	3.9	19.9	4.1	0.61	0.18	−0.35
BIS-11 Motor	23.3	5.8	25.1	6.0	0.79	0.21	−0.31
BIS 11 Non-Planning Score	27.9	5.4	30.6	5.3	1.46	0.09	−0.51

Finally, owing to the fact that the mean Global Assessment of Functioning (GAF) score was significantly higher for the BD II group relative to the BPD group, four separate hierarchical regression analyses were run to test whether the factor group (dummy coded) predicted STAI-Trait Anxiety, ALI borderline frequency subscale and, ALI reactivity scores and IVE impulsivity when accounting for GAF scores. To this end, GAF scores were entered in the first step, followed by group in the second steps. For the STAI-Trait Anxiety (ΔR²=0.229, ΔF(1,26)=7.91, p<0.0090), ALI reactivity scores (ΔR²=0.145, ΔF(1,26)=5.27, p<0.030), and the IVE impulsivity (ΔR²=0.214, ΔF(1,26)=9.08, p<0.006), the adjusted group difference was significant. For the ALI borderline frequency subscale, the adjusted group difference failed to reach significance (ΔR²=0.114, ΔF(1,26)=3.34, p=0.077).

Group-Level fMRI Results

Relative to the BD II disorder group, BPD participants showed lower right amygdala-right middle frontal gyrus connectivity (54, 24, 28, 306 voxels, Z = 4.04, FWE p = 0.000045, Cohen’s d value = 3.47) and right amygdala-left middle frontal gyrus (−44, 10, 32, 132 voxels, Z = 4.03, p = 0.032, Cohen’s d value = 2.25; See Figure 1). There were no other significant group differences in right amygdala whole-brain connectivity and there were no significant group differences in left amygdala-whole-brain connectivity. Given that there were significant group differences in GAF, STAI-Trait Anxiety, ALI frequency, ALI reactivity, and IVE-Impulsiveness, we conducted two follow-up hierarchical regression analyses to test whether the factor group (dummy coded) predicted right amygdala – right middle frontal gyrus and right amygdala – left middle frontal gyrus RSFC when accounting for these clinical differences. The clinical measures were entered in a first step, followed by group in the second step. Group differences in right amygdala – right middle frontal gyrus RSFC (ΔR² = 0.292, ΔF = (1,21) = 31.54, p < 0.001) and right amygdala – left middle frontal gyrus RSFC (ΔR² = 0.450, ΔF = (1,21) = 33.07, p < 0.001) remained even after controlling for clinical differences.

Figure 1.

Individuals with bipolar – II disorder (BP –II) showed greater right (R) amygdala-right middle frontal gyrus (MFG) connectivity and right amygdala – left middle frontal gyrus compared to those with borderline personality disorder (BPD).

Correlations Between RSFC and Clinical Measures

Given that there were group differences in ALI-frequency, ALI-reactivity, and IVE-Impulsiveness, STAI-Trait Anxiety, GAF, and amygdala-middle frontal gyrus RSFC, we examined whether individual differences in these clinical measures and amygdala-middle frontal gyrus RSFC were correlated (See Figure 2). Across the BPD and BP-II clinical groups, decreased right amygdala-middle frontal gyrus RSFC was associated with greater levels of impulsivity (r = −.54, p = 0.003, Pearson correlation), interpersonal affective reactivity (r = −.59, p = 0.001, Spearman’s correlation), and trait anxiety (r = −.54, p = 0.003, Pearson correlation) as well as lower overall general functioning (r = .46, p = 0.01, Pearson correlation). There was a non-significant trend with decreased right amygdala-right middle frontal gyrus RSFC also being associated with more frequent affective shifts (r = −.37, p = 0.05, Pearson correlation). Additionally, decreased right amygdala-left middle frontal gyrus RSFC was associated with increased interpersonal affective reactivity (r =−.49, p = 0.008, Spearman’s correlation). Associations between right amygdala-right middle frontal gyrus and impulsivity, interpersonal affective reactivity, as well as trait anxiety survived Bonferroni correction for multiple comparisons (10 tests, corrected p < 0.005). All other correlations did not survive multiple comparison corrections.

Figure 2.

Across the bipolar – II (BP – II) and borderline personality disorder (BPD) groups, decreased right (R) amygdala – right middle frontal gyrus (MFG) was associated with increased impulsivity (IVE Impulsiveness-- Impulsivity Venturesomeness), trait anxiety (STAI-2, State-Trait Anxiety Inventory), affective shifts (ALI – Frequency, Affective Liability Interview), and interpersonal affective reactivity (ALI – Reactivity), but decreased global assessment of functioning (GAF). Additionally, decreased right (R) amygdala – left middle frontal gyrus (MFG) was linked to higher interpersonally affective reactivity (ALI – Reactivity).

Discussion

Our primary aim was to extend the limited studies comparing the clinical phenomenology and underlying neurobiology of BPD versus BD II. Consistent with prior work (4–6), we found that individuals with BPD reported higher levels of impulsivity, more interpersonally reactive affective instability, and more frequent affective shifts. Additionally, we found that individuals with BPD reported higher levels of trait anxiety compared to the BD II group. While no prior studies have specifically compared trait anxiety differences between the two groups, one study reported that having a BPD diagnosis was associated with higher levels of anxiety-related symptoms on the Hamilton Rating Scale for Depression compared to having just a BD II diagnosis (40). Notably, all clinical phenomenological differences between the two disorders survived when controlling for group differences in GAF scores, except for frequency of affective shifts, which dropped to a trend level.

With respect to the neuroimaging results, the BPD group showed weaker amygdala-middle frontal gyrus RSFC compared to the BD II group. Additionally, decreased amygdala-middle frontal gyrus was associated with higher levels of impulsivity, interpersonal affective reactivity, and trait anxiety, as well as lower overall general functioning. These group differences in amygdala RSFC also survived when accounting for differences in clinical self-report measures and GAF scores, indicating that the weaker fronto-limbic connectivity was not just an epiphenomenon of group differences in clinical profile (or stated differently, the imaging data had incremental predictive validity). These results are consistent with prior work in clinical and non-clinical populations showing that weaker amygdala-frontal cortex connectivity is associated with higher levels of impulsivity and greater self-reported difficulties in regulating negative emotions (8–10). These findings are also consistent with reports (including from the current sample) that individuals with BPD are characterized by greater impulsivity and more difficulties with emotion regulation (4, 7). Given the clinical differences between these disorders and studies demonstrating that cortico-limbic deficits underlie these clinical facets, it is likely that they may be characterized by weaker amygdala-frontal cortex RSFC in BPD versus BD II.

However, results of the current study must be interpreted within the context of several limitations. First, the study was designed as a pilot study and thus represents an initial investigation of RSFC differences between the two clinical groups; accordingly, sample sizes were small and the design did not include a healthy control group. Second, most participants in both study groups were taking psychiatric medication, including mood stabilizers and anxiolytics. In order to ensure ecological validity, we intentionally did not exclude participants taking psychiatric medications. Third, each study group contained some comorbidity, which could potentially have influenced between-group comparisons. Fourth, with respect to the clinical phenomenological differences, we did not correct for multiple comparisons. Despite these limitations, study findings provide a preliminary examination of potential amygdala-based RSFC differences, which may guide future studies aiming to differentiate the two disorders to ensure more timely and accurate clinical diagnoses and subsequent implementation of appropriate treatments.