The altered multiscale dynamics of spontaneous brain activity in depression with Parkinson’s disease

Parkinson’s disease (PD), in addition to manifesting with the cardinal motor symptoms (tremor, rigidity, bradykinesia) and other frequent motor features such as freezing of gait, is often accompanied with non-motor features including cognitive and affective disturbances. More than 30% of people with PD suffer from depression [1, 2], and mounting evidence has showed that the depression in PD (DPD) is oftentimes accompanied with apathy and anhedonia [3], and associated with elevated risk of dementia, more severe motor dysfunction and diminished quality of life in this population [4, 5]. It is thus important to better understand the underlying neurophysiological mechanisms of DPD, which can ultimately help the management of DPD and optimize the design of therapeutic strategies for DPD.

Depression is a mood disorder that is linked to abnormal regulation within the cortical networks of the brain. Evidence suggests that compared to those in non-depression PD (NDPD), people in DPD would loss dopamine transporter availability in the striatum and limbic brain regions, but not the serotonergic systems, of which the changes are oftentimes observed in people with depression only [3, 6]. Many neuroimaging studies have characterized the resting-state brain activities in DPD. By characterizing the amplitude of low-frequency fluctuation (ALFF) of functional MRI (fMRI) resting-state blood-oxygen-level-dependent (BOLD) signal, for example, it was observed that as compared to non-depression PD (NDPD) cohort, those with DPD had higher ALFF in the left orbitofrontal [7] and right subgenual cingulate cortex [8]. In another study, Luo and colleagues characterized the functional connectivity (FC) between cortical networks and observed that in DPD, the FC within the prefrontal-limbic network is reduced as compared to NDPD [7].

However, these techniques are based upon a single-scale measure (e.g., ALFF) or focus on the relationship between networks (e.g., FC); the interaction and communication between neurons or functional networks at resting state are over multiple scales of time, ranging from millisecond (e.g., the time to transmit neural impulses) to hours or days (e.g., circadian rhythms). For example, multiple neural circuits are associated with the regulation of mood, including the frontal-subcortical, basal ganglic-thalamic-cortical circuit [9‐12]. The homeostatic mechanisms underlying neuronal/synaptic activity in depression range from minutes (e.g., dynamic enhancement/reduction of the postsynaptic response) to hours (e.g., strengthening/weakening of synaptic connections and neurochemical alterations including synaptic insertion and removal of neurotransmitters’ receptors) [13]. The complex multiscale dynamics of resting-state BOLD fluctuation thus contain critical information to the neuro-physiological procedures pertaining to the regulation of mental and physical functions. Several studies have measured the complexity of the multiscale dynamics in resting-state BOLD fluctuations using well-established non-linear techniques, such as multiscale entropy (MSE), and have linked such complexity to important functions [14‐17]. Zhou et al., for example, observed that in older adults, lower complexity of the brain functional networks (e.g., motor and attention networks) was associated with slower walking speeds [14]. However, the complexity of resting-state BOLD fluctuations in DPD is still not well-characterized.

In this study, we primarily aimed to characterize the complexity of resting-state BOLD fluctuation in DPD, to identify the regions with significantly lower BOLD complexity in DPD, and to examine the potential factors that are pertaining to decreased complexity in DPD. Specifically, we hypothesize that (1) compared to those with NDPD, the whole-brain BOLD complexity (i.e., complexity averaged across all the brain regions) would be lower in participants with DPD;(2) the complexity of regions pertaining to mood regulation (e.g., prefrontal lobe, anterior cingulate cortex) would be lower; and (3) lower BOLD complexity of the identified region would be associated with less connections/interactions between this region and other regions, which was measured by global FC (i.e., the functional connectivity of the region of interest (ROI) to all the other brain regions).

All the 30 participants completed this study. One participant in NDPD group was excluded in the analysis because of the poor imaging quality due to the excessive head motion during MRI scan. Table 1 showed the demographic and clinical information of the 29 participants (10 in DPD and 19 in NDPD group). ANOVA models showed no significant differences in age, UPDRS-III score, course of PD, and cognitive function (i.e., MMSE score) between the DPD and NDPD groups. As expected, the score of HAMD in DPD group was significantly higher as compared to the NDPD group (F = 67.1, p < 0.0001).

To our knowledge, this pilot study is the first to characterize the multiscale dynamics of spontaneous brain activities in people with DPD, and to explore the underlying mechanism that may contribute to such altered dynamics in this cohort. It is observed that (1) in DPD, the resting-state BOLD complexity is significantly diminished, particularly in several regions pertaining to the regulation of mood; (2) and the depression-related loss of BOLD complexity within those regions is cross-sectionally associated with the reduction of the functional connections of the region. These thus indicated that the complexity of the spontaneous brain activities may provide unique insight into the regulation of mood in those with PD, and may inform the management and therapeutic strategies for DPD with the focus on the complexity of spontaneous brain activity.

Mounting evidence has shown that the complexity, which characterize the multiscale dynamics in the spontaneous behavior or output (e.g., the BOLD fluctuation here) of a given physiologic system during resting state, rather than the single-scale measures, is associated with the capacity of this system to adapt to stress and thus is related to its functional performance. In neurodegenerative disorders, the quantity and/or quality of the regulatory elements in a physiologic system (e.g., the clusters in the brain) and their structural and functional connections to other systems are often altered [27], which manifests as a loss of the physiologic complexity in the dynamics of the system’s spontaneous behavior (e.g., BOLD fluctuations) [28]. Therefore, the degree of complexity in a given physiologic system is determined by both the inputs from other systems (external factors) and the capacity of the regulatory elements within the region (internal factors) to integrate and process these inputs. We here observed that the degree of complexity in 8 of the 18 identified regions was associated with the level of its global FC, which is consistent with the previous studies [28]. Wang and colleagues, for example, demonstrated that greater complexity of BOLD fluctuations in brain functional networks was correlated with greater functional connectivity within the same network in a relatively healthy cohort [23]. Therefore, the results in this study confirmed the findings in previous studies, that is, the complexity metric captures the capacity of information processing within the brain and lower complexity represents less transition or interaction between brain regions, potentially contributing to the depression in PD. On the other hand, it is interesting that no significant correlation between complexity and global FC was observed in the other 10 regions, suggesting that the degree of complexity in deed depends upon both external factors (i.e., global FC) and the internal factors. It is worthwhile to characterize these internal factors and examine its contribution to the degree of resting-state BOLD complexity in future studies.

In this study, we observed that the resting-state BOLD complexity in right frontal lobe, bilateral SMA, bilateral cingulate regions, right THA, and left temporal lobe is lower in DPD group as compared to NDPD group. These findings were in line with previous studies, which linked the neuro- and bio-physiologic alterations within these regions to depression in both people with and without PD. Meanwhile, multiple neural circuits pertaining to the mood regulation have been proposed in previous studies, such as the frontal-subcortical circuit and basal ganglic-thalamic-cortical circuit [9]. The observations from the view of multiscale dynamics of resting-state BOLD fluctuation here may indicate a neural circuit pertaining to DPD, that is, the SMA-right fronto-limbic circuit. Specifically, the activation in frontal lobe, particularly in the right side, is believed to be associated with the processing of emotional inputs and the responses to such emotional stimulus [24‐26], and the altered activation, hypoperfusion, and hypometabolism in this region are linked to depressive symptoms [27, 28], as well as DPD [29, 30]. The thalamus is also critical to the regulation of emotion processing and the cingulate regions (i.e., ACC and MCC) are the bridges for the transmission of the information related to emotion [31]. The decreased activation within thalamus, as well as the reduced structural and functional connections of cingulate regions to thalamus, has been linked to the presence of DPD [12, 32‐34]. The STG played an important role in the regulation of socio-emotional function, and the alteration in this region (e.g., reduced FC) is observed in DPD [35, 36]. To date, limited evidence has shown the association between SMA, a region primarily related to motor function, and persistent antidepressant effect of medication [37]. We here thus provide novel evidence suggesting that the alteration of the activities within SMA may contribute to DPD. Taken together, the subtle functional and structural changes within this circuit pertaining to DPD may manifest as the loss of resting-state BOLD complexity, which would serve as a target of future’s therapeutic strategies for DPD.

Regarding to the implementation to improve the functionalities by intervening the physiological complexity, studies have shown that the complexity is not an obligation consequence due to aging or disease, but can be restored and improved using appropriate interventions, and the increase of complexity has been linked improved functional performance. In our previous study, repetitive transcranial magnetic stimulation (rTMS) targeted on the bilateral primary motor cortex and cerebellum could relieve motor symptoms in Parkinsonian syndrome, positively correlated with motor network resting-state complexity, and a trend of MSE increase was observed [38]. Therefore, future studies are warranted to examine if the increase of BOLD complexity within this identified circuit as induced by the intervention is associated with the improvement in depressive symptoms.

Though the participants were of high homogeneity (e.g., naïve to anti-depression medications), the sample size in this pilot study is relatively small and the number of participants was not matched between groups, which may limit the power of statistical analyses (e.g., within each cohort, several significant correlations between complexity and global FC observed in the entire population were no longer there). Meanwhile, the participants were at relatively early stage of PD. Future studies with larger sample size of participants over a boarder range of disease severity are thus needed to confirm the observations in this study and explore the influence of disease severity on the resting-state BOLD complexity.