Characterization of brain functional connectivity in treatment-resistant depression

Highlights

• Gender and lateralization dependence of functional connectivity (FC) in TRD

• The degree varied between brain sides, males and females, and normal and TRD groups.

• The degree can be used to identify regions with abnormal FC in TRD and help better understand the mechanisms underlying TRD

• The degree may have the potential to help identify an effective DBS target in TRD

Abstract

Objective

To characterize the functional connectivity (FC) of target brain regions for deep brain stimulation (DBS) in patients with treatment-resistant depression (TRD), and to evaluate its gender and brain lateralization dependence.

Methods

Thirty-one TRD patients and twenty-nine healthy control (HC) subjects participated. FC of subcallosal cingulate gyrus (SCG), ventral caudate (VCa), nucleus accumbens (NAc), lateral habenula (LHb), and inferior thalamic peduncle (ITP) were evaluated using resting-state fMRI. FC was characterized by calculating the nodal ‘degree’, a major feature of the graph theory.

Results

The degree measures of the left and right VCa, the left LHb, and the left ITP were significantly greater in the TRD than in the HC group. The degree was greater in females with TRD in all these regions except the right LHb. Finally, the left hemisphere was generally more affected by depression and presented significant degrees in LHb and ITP regions of the patients.

Conclusion

Our findings demonstrate the ability of degree to characterize brain FC and identify the regions with abnormal activities in TRD patients. This implies that the degree may have the potential to be used as an important graph-theoretical feature to further investigate the mechanisms underlying TRD, and consequently along with other diagnostic markers, to assist in the determination of the appropriate target region for DBS treatment in TRD patients.

Introduction

Major depression is one of the most common psychiatric diseases associated with other medical and mental illnesses. The prevalence of major depressive disorder (MDD) is considerable; ~4.7% of the world's population (Ferrari et al., 2013). Approximately, one-third of MDD patients do not respond to common therapies, known as treatment-resistant depression (TRD) patients (Berlim and Turecki, 2007; Dandekar et al., 2018; Daskalakis and Sun, 2017). Recently, deep-brain stimulation (DBS) has been developed and used to treat TRD (Lozano et al., 2008; Kennedy et al., 2011). The brain regions suggested for DBS treatment are inclusive of the subcallosal cingulate gyrus (SCG) (Lozano et al., 2008; Puigdemont et al., 2012) ventral caudate (VCa) (Malone et al., 2009), inferior thalamic peduncle (ITP) (Jiménez et al., 2005), lateral habenula (LHb) (Sartorius and Henn, 2007; Hoyer et al., 2012), and nucleus accumbens (NAc) (Bewernick et al., 2010). The therapeutic responses to stimulation in each of these brain regions were observed only in 50–60% of TRD subjects (Bewernick et al., 2012; Lozano et al., 2008). These findings may suggest that the success of DBS therapy can be associated with alterations in the structural and functional connections among DBS targets, which is subject-dependent and has remained a major clinical challenge.

Based on the literature, brain structural and functional connectivity (FC) between the DBS regions and other major areas are typically affected by depression. Increased FC between the striatum and dorsolateral prefrontal cortex (He et al., 2019; Vasic et al., 2015), LHb with key brainstem structures (i.e., ventral tegmental area, substantia nigra, pons) as well as the anterior and posterior cingulate cortices, precuneus, thalamus, and sensorimotor cortex. (Ely et al., 2016) have been reported in TRD patients. In contrast, a decreased hypothalamic FC with SCG was observed in these patients (Sudheimer et al., 2015). Thus, it was expected that a suitable stimulation of a DBS target should modulate the abnormal activities in the network of various DBS targets. NAc-DBS reduced abnormal hyperactivity of SCG and orbital frontal cortex (OFC) in TRD (Bewernick et al., 2010). Similar network alterations were obtained by SCG-DBS, which improved the hypoactivity of NAc (Lozano et al., 2008; Kennedy et al., 2011). Interestingly, SCG hyperactivity was also modulated by VCa-DBS. These studies have shown that DBS targets are connected and functionally related. However, the literature demonstrated that these connections are subjective and treatment success depends on the personalized refinement of DBS, particularly a precise selection of DBS target for each TRD patient. In this regard, Accolla et al. (2016) found stronger connectivity of SCG to the medial prefrontal cortex (mPFC) in TRD subjects treated with SCG-DBS compared to those who were non-responders (Accolla et al., 2016). Similarly, another study showed structural connections from SCG to the medial frontal cortex, rostral and dorsal cingulate cortex, and subcortical nuclei in all DBS responders, while these connections were not consistently observed in non-responders (Riva-Posse et al., 2014). Therefore, approximately 40% of patients did not respond appropriately to DBS probably due to a lack of structural and functional connectivity between the different DBS areas in the brain networks (Accolla et al., 2016; Fox et al., 2014; Padmanabhan et al., 2019; Riva-Posse et al., 2014). This indicates that the evaluation of brain structural and FC can be used as a suitable tool to address the clinical challenge of determining the appropriate DBS brain region for each patient. Also, the reflection of the neuronal activity of a brain region is directly related to the cerebral blood flow (CBF) in that region, which can be measured by functional neuroimaging techniques such as PET and arterial spin labeling (ASL) (He et al., 2019). Earlier studies have shown that FC has a metabolic basis that is coupled with CBF and the rate of metabolism. Thus, a brain region with greater FC has a higher metabolism rate (Soddu et al., 2016; Riedl et al., 2014; Li et al., 2012). Therefore, in this study, we used functional magnetic imaging (fMRI) to extract FC between brain regions based on the blood‑oxygen-level-dependent (BOLD) signal at rest.

Graph-theoretical network analysis is an appropriate method that allows to determine the organization of brain connectivity, to characterize the topological properties of brain networks, and to examine the complexity of brain networks globally and locally. Globally, the general properties of the whole brain are examined; whereas, in the nodal or local networks, the graph features are calculated for specific regions of the brain (Wang et al., 2010; Bullmore and Sporns, 2009).

Several metrics are developed to measure the ‘centrality’ of a brain network, including the clustering coefficient, nodal degree centrality, betweenness centrality, node neighbor's degree, and closeness centrality (Bullmore and Sporns, 2009). These features have been used to investigate specific changes in the topological properties of the brain graph. Among them, nodal degree is the most popular measure of centrality since it is directly related to functional connectivity, in contrast to other measures that provide redundant information (Oldham et al., 2019; Bullmore and Sporns, 2009). Furthermore, nodal degree is shown to have a high correlation with other centrality metrics (Oldham et al., 2019; Rueda et al., 2017). Nodal degree is expressed as ‘degree’ throughout the manuscript. For the sake of sensitivity, the degree can be measured locally considering merely the nodes of desired networks. Given that degree quantifies FC, it can be concluded that a brain region with a greater degree has a greater CBF rate and metabolism, and as a result higher neuronal activity.

Importantly, earlier studies reported hemispheric asymmetry for the emotional process (Canli, 1999; Alves et al., 2008). There is also indirect evidence suggesting that the underlying mechanisms of TRD might be different between males and females. This may explain the greater prevalence of TRD among females (Kuehner, 2017; Albert, 2015). Overall, the information about gender and brain lateralization dependence of brain abnormal activities in TRD subjects is limited, ambiguous, and controversial.

We aimed to use the degree to compare the FC of five target brain regions of DBS, as stated above, between both brain hemispheres of females and males in both the TRD and control subjects. We expected the outcomes to improve our understanding of the mechanisms underlying TRD. In particular, we expect the degree along with other major clinical measures to assist with more accurate decision-making in determining the precise target brain region for DBS treatment in each TRD patient.