fMRI evidence for abnormal resting-state functional connectivity in euthymic bipolar patients

Abstract

Background

Neural substrates of bipolar disorder (BD) have frequently been characterized by dysregulation of fronto-limbic networks that may persist during euthymic periods. Only a few studies have investigated euthymic bipolar patients (BP) functional connectivity at rest. The current study aims to assess resting-state functional connectivity in euthymic BP in order to identify trait abnormalities responsible for enduring mood dysregulation in these patients.

Methods

Medial prefrontal cortex (mPFC) functional connectivity was investigated in 20 euthymic BP and 20 healthy subjects (HS). The functional connectivity maps were compared across groups using a between-group random effect analysis. Additional region of interest (ROI) analysis focused on mPFC–amygdala functional connectivity as well as correlations between the clinical features in euthymic BP was also conducted.

Results

A significant difference between euthymic BP and HS was observed in terms of connectivity between the mPFC and the right dorsolateral prefrontal cortex (dlPFC). A significant negative correlation between the activity of these regions was found in HS but not in euthymic BP. In addition, euthymic BP showed greater connectivity between mPFC and right amygdala compared to HS, which was also correlated with the duration of the disease.

Limitations

The BP group was heterogeneous with respect to the bipolarity subtype and the medication. The robustness of results could be improved with an increased sample size.

Conclusions

Compared to HS, the euthymic BP showed abnormal decoupling (decreased functional connectivity) activity between mPFC–dlPFC and hyperconnectivity (increased functional connectivity) and between mPFC and amygdala. These abnormalities could underlie the pathophysiology of BD, and may deteriorate further in accordance with disease duration.

Introduction

Bipolar disorder (BD) is a severe psychiatric disorder characterized by abnormal emotional and cognitive processing during acute (mania and depression) and inter-critical (euthymic) periods. Euthymic periods are characterized by disabling residual symptoms such as mood instability, increased emotional reactivity as well as cognitive processing deficit (Bora et al., 2009, Henry et al., 2008, Judd et al., 2005, Martínez-Arán et al., 2004). Neurofunctional impairments in euthymic bipolar patients (BP) may underlie the primary pathological process of BD and represent trait characteristics of the disorder. The identification of trait abnormalities in BD is particularly important as it could contribute to early diagnosis of BD thus reducing the latency to adequate treatment thereby improving outcome (Altamura et al., 2010).

Current pathophysiological models of BD highlight the presence of abnormalities in fronto-limbic regions responsible for the emotional dysregulation seen in this disorder (Phillips et al., 2008, Strakowski et al., 2012). During euthymic periods, both increased and decreased activations in prefrontal and limbic regions have been found when either cognitive and/or emotional tasks have been employed (Chen et al., 2011). Thus, persistent cerebral abnormalities during euthymic periods are not completely understood. fMRI paradigms not dependent on the employed task, i.e. resting-state fMRI paradigm, provide a promising approach to the study of brain function in BD and identification of the precise state and trait characteristics of this disorder.

Recent studies suggested that brain activity during resting-state reflects neural activity in the absence of any explicit cognitive demand. Consequently, its evaluation may be of interest in studying pathophysiological characteristics of affective disorders (see Whitfield-Gabrieli and Ford (2012) for review). Among the identified resting-state brain networks, the default mode network (DMN) is composed of a set of brain regions preferentially activated during internally-generated thoughts and self-reflection, and deactivated during cognitively demanding tasks in fMRI (Greicius et al., 2003, Raichle et al., 2001). It encompasses four core brain regions: medial prefrontal cortex (mPFC), posterior cingulate cortex (PCC)/precuneus and lateral parietal cortex (angular gyrus). Additional brain regions such as the hippocampus and parahippocampal cortices, inferior temporal and superior frontal cortices are also connected to these core regions (Buckner et al., 2008, Fox et al., 2005, Greicius et al., 2003). The DMN has frequently been found to be abnormally activated and/or connected in patients with unipolar and bipolar depression (Greicius et al., 2007, Liu et al., 2012a, Nejad et al., 2013), schizophrenia (Whitfield-Gabrieli et al., 2009) and manic periods of BD (Calhoun et al., 2008, Chai et al., 2011, Öngür et al., 2010). In particular, abnormal recruitment of parietal areas and ventral medial prefrontal cortex (mPFC) in manic BP (Öngür et al., 2010) and decreased connectivity in posterior cingulate gyrus in BP with psychotic symptoms (Khadka et al., 2013) have been reported. In bipolar depression enhancement of the local synchronization of spontaneous neural activities in the left mPFC and left inferior parietal lobule has been found (Liu et al., 2012a). However, although DMN functional connectivity may also be impaired during euthymic periods, this has not been extensively explored in BD.

Moreover, abnormal functional connectivity patterns at rest have been observed in cortico-limbic regions in BP but without specificity according to mood state (Anand et al., 2009, Chepenik et al., 2010, Meda et al., 2012, Xu et al., 2014). Indeed, most of the previous resting-state studies reported abnormal cortico-limbic connectivity in BP, thus supporting the emotional dysregulation hypothesis (see Vargas et al. (2013), for review). As far as we know, only two studies specifically recruited BP during euthymic state. These studies, performed in a group of medicated type I bipolar patients (BP) compared to healthy subjects (HS), revealed reduced global functional connectivity in mPFC (Anticevic et al., 2012), increased mPFC–amygdala and ventrolateral PFC (vlPFC)–amygdala connectivity (Anticevic et al., 2012, Torrisi et al., 2013, respectively) and also decreased dorsolateral PFC (dlPFC)–amygdala in BP with a history of psychosis (Anticevic et al., 2012).

In the present study, we employed a seed-based correlation analysis to evaluate resting-state brain activity in euthymic BP compared to HS. This method was preferred to an Independent Component Analysis (ICA) approach because it allowed for direct assessment of the network of areas most strongly functionally connected areas to the seed region. Moreover, seed-analysis allows for the identification of anticorrelations, i.e. networks with inverse temporal relationships or negative activity correlations (e.g. the DMN and the task-positive network) (Chai et al., 2011, Fox et al., 2005). This phenomenon may be crucial for understanding the pathophysiology of psychiatric disorders. The ventral part of the mPFC has been chosen as the seed-region because it has been identified as a core region of the DMN, showing positive functional connectivity with DMN regions and negative functional connectivity (anti-correlations) with task-related regions, such as lateral prefrontal regions (Fox et al., 2005, Greicius et al., 2003). Moreover, the mPFC is widely connected to subcortical and limbic areas, suggesting its import to emotional regulation processes and in affective symptoms of BD. We hypothesized in euthymic BP persistent abnormal resting-state functional connectivity between mPFC and limbic regions as well as between mPFC and lateral prefrontal regions. We also assumed stronger abnormalities in functional connectivity when the disease has further progressed.