Alterations in intrinsic functional networks in Parkinson’s disease patients with depression: A resting‐state functional magnetic resonance imaging study

Abstract Aims The aim of this research was to investigate the alterations in functional brain networks and to assess the relationship between depressive impairment and topological network changes in Parkinson's disease (PD) patients with depression (DPD). Methods Twenty‐two DPD patients, 23 PD patients without depression (NDPD), and 25 matched healthy controls (HCs) were enrolled. All participants were examined by resting‐state functional magnetic resonance imaging scans. Graph theoretical analysis and network‐based statistic methods were used to analyze brain network topological properties and abnormal subnetworks, respectively. Results The DPD group showed significantly decreased local efficiency compared with the HC group (P = .008, FDR corrected). In nodal metrics analyses, the degree of the right inferior occipital gyrus (P = .0001, FDR corrected) was positively correlated with the Hamilton Depression Rating Scale scores in the DPD group. Meanwhile, the temporal visual cortex, including the bilateral middle temporal gyri and right inferior temporal gyrus in the HC and NDPD groups and the left posterior cingulate gyrus in the NDPD group, was defined as hub region, but not in the DPD group. Compared with the HC group, the DPD group had extensive weakening of connections between the temporal‐occipital visual cortex and the prefrontal‐limbic network. Conclusions These results suggest that PD depression is associated with disruptions in the topological organization of functional brain networks, mainly involved the temporal‐occipital visual cortex and the posterior cingulate gyrus and may advance our current understanding of the pathophysiological mechanisms underlying DPD.


| INTRODUC TI ON
Parkinson's disease (PD) is a common neurodegenerative disease with a growing incidence in recent decades. Depression is considered one of the most common nonmotor symptoms of PD, with an incidence of up to 40%. 1 In addition, depression is one of the established clinical prodromal markers and may be an independent risk factor for PD, 2,3 often occurring before motor symptoms appear and causing a serious negative impact on PD patients' quality of life. 4 However, the pathophysiology underlying the contribution of depression to PD remains unclear. Thus, exploring the pathological mechanisms of depression in PD may contribute to early identification and effective treatment of PD patients in the clinical prodromal stage of the disease; that is, it might delay the progression of the disease and improve quality of life.
With the rapid development of imaging technology and its continuous application in neuroscience research, strong noninvasive technical support is available for exploring the pathogenesis of neuropsychiatric diseases and realizing the early diagnosis of PD patients with depression (DPD). Graph theory provides a powerful mathematical framework for describing the topological organization of brain networks in terms of nodes (ie, brain regions) and edges (ie, physical or functional connections between brain regions). 5 The latest research has suggested that network topological metric measurements derived using graph theory may be potential biomarkers in PD to evaluate disease progression and to monitor therapeutic effects. 6 By using diffusion tensor imaging, one study showed that despite preserved small-world topology, DPD exhibited higher network efficiency in the fronto-limbic system compared to that in PD patients without depression (NDPD). 7 Gou et al, 8 by using graph theory and network-based statistic (NBS) methods, showed that the integration of the structural brain network was impaired with disrupted connectivity of the limbic and visual systems in de novo and drug-naïve DPD patients. Notably, these results were approximately consistent with those obtained from depressive patients, suggesting that deficits of the regional and connectivity characteristics in the structural networks were primarily found in the frontal brain regions, limbic system, and occipital lobe compared to that in healthy controls (HCs). 9,10 Therefore, it is an urgent problem to explore the characteristic imaging markers of DPD and to realize early diagnosis and treatment. A previous study showed that even though the overall structural organization of the PD connectome remains robust at relatively early disease stages, there is a breakdown in the functional modular organization of the PD connectome. 11 Therefore, characterization of the functional network connectome in DPD patients may help researchers to further explore the mechanisms underlying DPD.
The purpose of this study was to investigate the changes in the topological properties and the functional connectivity of the wholebrain functional network in patients with DPD and to assess whether depressive impairment is correlated with functional topological network changes in DPD patients.

| Neuropsychological assessments
Motor symptom severity in PD patients was measured by the MDS

| MRI data acquisition
All the PD subjects in this study underwent rs-fMRI after discontinuing antiparkinsonian medication for more than 12 hours. The scans were performed on a 3.0T GE SIGNA MR scanner (Signa Excite HD; GE Healthcare) with an 8-channel head coil in the Department of Radiology of Guangdong Provincial People's Hospital. All participants were required to keep their eyes closed, relax, and remain awake during the MRI scans. The rs-fMRI scanning parameters were as follows:

| Data preprocessing
Data preprocessing was performed on the MATLAB-based DPABI, the Data Processing Assistant for rs-fMRI. The first ten volumes of the functional images were discarded. The remaining volumes were slice-time corrected, and realignment was performed to correct the motion between time points. Head motion parameters were computed by estimating the translation in each direction and the angular rotation on each axis for each volume. The subjects with motion (a mean framewise displacement [FD] Jenkinson 14 ) greater than 2*standard deviations (SDs) above the group mean motion were excluded. 15 According to this exclusion criterion, only two subjects were excluded from the HC group (Tables S1 and S2). No significant intergroup differences were found in head motion ( Table 1). The fMRI data were coregistered to the same subject's high-resolution T1-weighted image. The coregistered images were segmented into gray matter, white matter (WM), and cerebrospinal fluid (CSF) and were then spatially normalized to the standard stereotaxic coordinates of the Montreal Neurological Institute space using an echoplanar imaging template and resampled into a voxel size of 3*3*3 mm 3 . Then, WM, CSF, and head motion were removed as nuisance variables; head motion was removed using the Friston 24-parameter model. The generated images were spatially smoothed with a 4-mm full-width at half-maximum Gaussian kernel. A temporal filter (0.01-0.1 Hz) was used to decrease the effect of low-frequency drifts and physiological high-frequency noise. In addition, adequate quality control was conducted on MRI data, which is essential for establishing reliable results. 16

| Brain network construction
The binarized functional brain network was constructed by GRETNA, 17 a graph theoretical network analysis toolbox for imaging connectomics. The functional connectivity matrices (composed of positive correlations), which were used to analyze brain TA B L E 1 Demographic and clinical characteristics and global topological properties of the NDPD, DPD, and HC groups

| Functional brain network analyses
Graph theory was applied to analyze the functional brain network.
For the global network, the global efficiency (Eglob) and characteristic path length (Lp) were used to investigate the integration of the functional brain network, while segregation was assessed by the local efficiency (Eloc), clustering coefficient (Cp), and modularity. For each participant, to assess whether the network had small-world properties, the network measures were normalized to comparable values from random networks (N = 100). Small-world organization was assessed by the normalized Lp (λ), the normalized Cp (γ), and small worldness (σ). The local topological properties (the degree, betweenness centrality, and nodal efficiency) were obtained to measure the regional network organization. To investigate group differences in these networks, we calculated the area under the curve for each network metric, which provides a summarized scalar for topological characterization of brain networks independent of specific threshold selection. The NBS approach, which is a validated, nonparametric statistical approach for controlling familywise errors in connectome analyses, 22 was utilized to further identify functional connections showing differences between each pair of groups.

TA B L E 2
The abbreviations of the 90 brain regions in AAL-90 atlas  that were significant at a corrected level of P = .001 were reported.
The effects of age, gender, and years of education were adjusted for these analyses.

| Population characteristics
The demographic and clinical features of the participants are presented in Table 1. No significant differences were observed in age, sex, years of education, or MMSE scores among the three groups.
In addition, the DPD and NDPD patients had comparable values for disease course, the LED, H&Y stage, and MDS-UPDRS III scores.
However, both the HAMD and HAMA scores in the DPD group were significantly higher than those in the other two groups (P < .001), which is consistent with previous studies revealing that DPD coexisted with anxiety disorder.
Statistical comparisons were performed to detect significant differences in the AUCs of global parameters among the three groups. The DPD group showed a significantly lower integrated Eloc (P = .008, FDR corrected) compared with the HC group. No significant differences (P > .05) were found in Eglob, Lp, Cp, modularity, γ, λ, and σ (Table 1).

| Intergroup differences in local topological properties
Brain regions exhibiting significant intergroup differences in at least one nodal metric (the degree, efficiency, or betweenness) were identified. But there were no brain regions surviving the FDR correction for multiple comparisons.

| Differences in hub regions among the three groups
The hubs were identified as nodes with both degree and betweenness centrality that were one SD above the network averages. 23 In the HC group, five hub regions were defined in the temporal visual area, including the bilateral fusiform gyri (FFG), bilateral middle temporal gyri (MTG), and right inferior temporal gyrus (ITG.R).
Compared with the HC group, the NDPD group lacked the bilateral FFG as hub regions, while the left posterior cingulate gyrus (PCG.L) was a hub region. However, no hub regions were screened out in the DPD group (Tables 3 and S5).  Table S3 present the results of the correlation analysis between the degree and HAMD scores in the DPD group. A significantly and specifically positive correlation was observed in the right inferior occipital gyrus (IOG.R; r = .848, P = .0001, FDR corrected) that was not present in the NDPD group (Table S4). There was no correlation between HAMD scores and the other network topological properties surviving the FDR correction for multiple comparisons.

| Functional connectivity characteristics
In the NBS analysis, the primary test statistic thresholds were set  Figure 3, Table S6). Compared with the HC group, we localized a connected subnetwork involving more extensive brain regions with 21 nodes and 29 edges that were significantly decreased in the DPD group, which contained an extensive portion of the prefrontal (eg, the bilateral median cingulate and paracingulate gyri) and temporal-occipital (eg, the amygdala, bilateral FFG, and occipital lobe) lobes in addition to the same brain regions included in the different subnetwork between the NDPD and HC groups ( Figure 3, Table S7).
Most of these regions were the components of the prefrontal-limbic network and temporal-occipital visual cortex. In contrast, the DPD group did not show any alterations in functional connectivity compared with the NDPD group.

| D ISCUSS I ON
In the present study, we comprehensively examined whole-brain resting-state networks for functional changes in DPD patients using graph theoretical and network-based analyses. Our findings revealed that the brain functional network of DPD patients preserved the small-world property, but the Eloc was significantly reduced compared with the HCs. In nodal metrics analyses, the degree of the occipital visual cortex in DPD group was found to be positively associated with the depressive symptoms. Meanwhile, there was a reorganization of the network's hubs in the temporal visual cortex and PCG.L in the DPD group. Moreover, different functional connectivity trends were observed in the NDPD and DPD groups vs the HC group.
A small-world network involves a combination of a high Cp (a measure of local network connectivity) and a short characteristic Lp (a measure of global network connectivity), reflecting a highly effective topological organization combining regional specialization and efficient global information transfer to realize efficient transmission of information. [24][25][26] Here, we found that the functional brain networks of the DPD, NDPD and HC individuals exhibited a small-world architecture, which is consistent with previous findings suggesting the preservation of a small-world architecture in the presence of pathology. 21 Despite the common small-world topology, the Eloc showed significantly smaller values over a wide range of sparsity in Note: The abbreviations of the 90 brain regions are given in Table 2. Functional magnetic resonance studies have revealed that degeneration of nigrostriatal neurons in PD may be associated with largescale network reorganization and that levodopa tended to normalize the disrupted network topology in PD patients. 29 Although all PD subjects underwent rs-fMRI after discontinuing antiparkinsonian medication for more than 12 hours, the interference of drugs could not be completely excluded.

TA B L E 3 Hub regions in the HC, NDPD and DPD groups
The nodal degree is the sum of all binary/weighted edges of one node, which measures the single nodal connectivity to the rest of the nodes in a network. Regarding the significantly positive correlation between depressive symptoms and the betweenness centrality of IOG.R, which is the striate cortex regions of the visual cortex, previous studies have reported conclusions consistent with our findings. Some rs-fMRI studies have shown that the value of regional homogeneity in the occipital lobe was decreased 30 and that the synchrony of interhemispheric resting-state functional connectivity was impaired in the occipital lobe in DPD patients. 31 In particular, a graph theory study showed that the node topological properties of the bilateral lingual and bilateral inferior occipital regions were significantly associated with Geriatric Depressive Scale (GDS) scores in DPD patients. 8 Moreover, studies also found that patients with DPD exhibited hyperperfusion and accelerated brainwave activity in the occipital cortex. 32,33 Kim et al 32 discovered that perfusion in the occipital lobe was increased and regional cerebral blood flow (rCBF) in the occipital cluster was positively related to the GDS scores in DPD patients. A quantitative electroencephalogram study showed the higher amplitude in beta in occipital lobe areas in rapid eye moment relative to nonrapid eye moment 2 were significantly different in DPD and NDPD patients. 33 Increased cerebral blood flow and brainwave activity in these areas may accelerate effective information transmission. The mechanism remains unclear but may be related to the reduced γ-aminobutyric acid (GABA) levels, while GABAergic neurons are mostly distributed in the occipital cortex of depressed patients. Prior research has suggested that occipital levels of GABA were significantly lower in recovered depressed patients than in healthy controls. 34 Inverse correlations between the GABA levels and rCBF in the brain have also been suggested by the previous literature. 35,36 Further study revealed that increased baseline activation in the occipital cortex predicted antidepressant response. 37 Accordingly, we considered that changes of topological attributes in occipital visual regions may serve as useful targets for biomarkers in assessing the severity and progression of DPD.
Additionally, the extrastriate cortex regions of the visual cortex (bilateral MTG and ITG.R) were identified as network hubs in the HC and NDPD groups, but not in the DPD group, which were consistent with a recent MRI study reporting that the node topological properties of the temporal visual area were significantly associated with depressive symptoms in PD patients. 8 Previous studies have shown that neurodegenerative diseases target brain regions that are especially highly correlated in healthy subjects. These regions are the network hubs, which play a crucial role in the network, as they interact with many brain regions. 38 One study showed that depression was correlated with impaired color vision in PD patients through clinical observation

F I G U R E 2
The results of the correlation analysis between the degree and HAMD scores in the DPD group. The abbreviations of the 90 brain regions are given in Table 2 and inferred that the visual system is crucial in the depressive pathology in PD patients. 39 We speculate that PD depression might disrupt the visual functional network and reorganize the network's hubs.
In NBS analysis, compared with HCs, the functional connections between the regions of the temporal-occipital visual cortex and the prefrontal-limbic network were significantly weakened in DPD patients, which did not appear in the comparison between NDPD patients and HCs. The most common cognitive feature of depression is a processing bias toward negatively affective stimuli, which is associated with brain activation in visual areas in- first-episode treatment-naive depression showed increased network homogeneity in the PCC. One study in unmedicated major depressive disorder patients reported higher functional connectivities between the PCC and lateral orbitofrontal cortex and between the angular and middle frontal gyri. 44 A previous study also observed that functional connectivity in the right PCC was increased in DPD patients and negatively correlated with depression scores. 45 Given these findings, we speculate that DPD and primary depression may share a similar pathological change and that the PCC plays a key hub role in the pathogenesis of primary depression, as well as in DPD patients.

F I G U R E 3
A, The abnormal subnetworks constituted in NDPD group compared with HC group; B, the abnormal subnetworks constituted in DPD group compared with HC group. There were no significant differences between NDPD compared with DPD. Light blue, the nodes in the frontal lobe; blue, the nodes in the prefrontal lobe; green, the nodes in the parietal lobe; purple, the nodes in the temporal lobe; red, the nodes in the occipital lobe. The abbreviations of the 90 brain regions are given in Table 2 Some limitations of our study should be noted. First, most of the subjects had regularly taken antiparkinsonian drugs, and the interference of drugs on the results was not completely eliminated.
Second, this study had a cross-sectional design and a relatively small sample size. Third, although the AAL atlas is still widely applied to investigate human brain organizations, a previous study suggested that the topological organization of brain networks was affected by various parcellation strategies. 46 Furthermore, this study adopted a single imaging technology to only preliminarily explore the possible relationship between changes in the brain functional network and DPD. Therefore, longitudinal follow-up observations are needed, drug-naïve PD patients should be recruited, and different parcellation schemes should be used in conjunction in future large-scale studies using multimodal MRI techniques to further explore the pathogenesis of DPD.
In conclusion, our study explored the functional brain network in DPD patients based on rs-fMRI. Our results revealed that PD depression is associated with disruptions in the topological organization of functional brain networks, mainly involved the temporal-occipital visual cortex and the posterior cingulate gyrus. These findings may advance our current understanding of pathophysiological mechanism underlying DPD.

CO N FLI C T O F I NTE R E S T
The authors have no conflict of interest to disclose.

DATA AVA I L A B I L I T Y S TAT E M E N T
The data that support the findings of this study are available in the supplementary material of this article.