Identifying Brain Connectomic Alterations Specific to Mild Cognitive Impairment and Depression Co-morbid with Parkinson’s Disease
Sinan Zhao1, Peipeng Liang2,3,4, and Gopikrishna Deshpande1,5,6

1AU MRI Research Center, Department of Electrical and Computer Engineering, Auburn University, Auburn, AL, United States, 2Department of Radiology, Xuanwu Hospital, Capital Medical University, Beijing, China, People's Republic of, 3Beijing Key Lab of MRI and Brain Informatics, Beijing, China, People's Republic of, 4Key Laboratory for Neurodegenerative Diseases, Ministry of Education, Beijing, China, People's Republic of, 5Department of Psychology, Auburn University, Auburn, AL, United States, 6Alabama Advanced Imaging Consortium, Auburn University and University of Alabama Birmingham, Auburn, AL, United States

### Synopsis

Resting state fMRI has been used to investigate connectomic alterations in Parkinson’s disease (PD). These studies used conventional connectivity analysis where in connectivity is assumed to be stationary over time. However, recent work suggests that temporal variability of connectivity is sensitive to human behavior in health and disease. Therefore, we estimated static functional connectivity (SFC), dynamic FC (DFC) from: PD, PD subjects with mild-cognitive-impairment (PDMCI), Depressed PD subjects with MCI (DPDMCI) and Normal Controls (NC). We hypothesized that increased disease burden would lead to reduced strength of SFC and the variability of DFC. We provide evidence to support this hypothesis.

### Introduction

Parkinson’s disease (PD) is often comorbid with other disorders such as depression and/or cognitive impairment. Understanding the neural basis of such comorbidity is important because it often leads to worse disease progression. Functional magnetic resonance imaging (fMRI) based resting state connectivity analysis has been previously used to investigate connectomic alterations in PD comorbid with cognitive impairment [1,2] and depression [2], but not both. Specifically, these studies used conventional connectivity analysis where in connectivity is assumed to be stationary over time and one value of connectivity is derived for the whole experiment. However, recent work suggests that connectivity varies over time and the temporal variability of connectivity is sensitive to human behavior in health and disease [4]. Therefore, in addition to conventional static functional connectivity (SFC), we also estimated dynamic FC (DFC) from resting state fMRI data acquired from the following populations: Parkinson’s disease (PD), PD subjects with mild cognitive impairment (PDMCI), Depressed PD subjects with mild cognitive impairment (DPDMCI) and matched Normal Control (NC). Based on previous studies, we hypothesized that increased disease burden would lead to reduced strength of connectivity. Further, since pathological states are associated with lower flexibility/adaptability, we surmised that variability of DFC will decrease with increased disease burden.

### Methods

Resting state fMRI data was collected from 22 subjects in each of the following groups: NC, PD, PDMCI and DPDMCI. Data acquisition was performed on a SIEMENS Trio 3T scanner using an echo-planar imaging (EPI) sequence with TR = 2000ms, TE = 30ms, FOV = 24cm, FA = 90°, resolution = 64 × 64 matrix, slices = 33, thickness = 4mm, gap = 1mm. After standard preprocessing, fMRI voxel time series were obtained from 264 spherical regions (5mm radius). These regions were defined through meta-analyses of task fMRI data and resting state functional connectivity data [5] to provide a comprehensive sampling of functional regions. Hidden neural signals underlying each voxel time series were obtained using blind hemodynamic deconvolution [6]. Subsequently, mean latent neural time series for these 264 ROIs were calculated. Pearson’s correlation was used to calculate SFC while DFC was estimated using sliding-windowed correlation analysis with variable length windows determined by the Dickey-Fuller test [4]. Significant group differences were obtained (controlled for age, gender, education, disease duration, unified Parkinson's disease rating scale score, geriatric depression scale score, Hamilton depression rating scale, Montreal cognitive assessment score and Levodopa equivalent dose) upon performing pair-wise t-tests between the four groups. Let $T_{1,...,L}$ be the paths that are common among the following comparisons: NC > PD, NC > PDMCI, NC > DPDMCI. Let $T_{1,...,L,L+1,...,M}$ be the paths that are common among the following comparisons: NC > PD, NC > PDMCI. If paths $T_{L+1,...,M}$ also satisfy PD > PDMCI, then they would be specific to MCI with comorbid PD. Finally, let $T_{1,...,L,L+1,...,M,M+1,...,N}$ be the paths that are common among the following comparisons: NC > PD, NC > DPDMCI. If paths $T_{M+1,...,N}$ also satisfy PDMCI > DPDMCI, then they would be specific to depression with comorbid PDMC.

### Results and Discussion

As for Static FC, paths specific to PD, $T_{1,...,L}$ (p < 0.0001 uncorrected) were found to be altered between regions in the visual network (VN), somatosensory network (SN), and frontal-parietal network (FPN) (Fig.1a). Paths specific to MCI with comorbid PD were found between SN (also within SN), default mode network (DMN), and VN (Fig.2a), while changes caused by depression with comorbid PDMCI were more specific to paths between somatosensory, visual and auditory networks (Fig.3a). Similar changes within and between networks were also obtained by the variance of DFC (Fig.1b-3b), suggesting that adaptability and flexible engagement within and between the above networks decrease when MCI and depression are comorbid with PD. Further research is needed to disentangle precise neural mechanisms underlying these findings.

### Acknowledgements

No acknowledgement found.

### References

Reference: [1] Baggio, et al, Hum Brain Mapp, 36(1): 199-212, 2015 [2] Amboni, et al, J Neurol, 262: 425-434, 2015 [3] Luo, et al, J Neurol Neurosurg Psychiatry, 85:675-683, 2014 [4] Jia, et al, Brain Connectivity, vol. 4(9), pp. 741-759, 2014 [5] Power, et al, Neuron, 72(4): 665–678, 2011. [6] Wu, et al, Med Image Anal 17:365–374, 2013.

### Figures

Fig.1 Paths specific to PD. (A) SFC, (B) DFC

Fig.2 Paths specific to MCI with comorbid PD. (A) SFC, (B) DFC

Fig.3 Paths specific to Depression with comorbid PDMCI. (A) SFC, (B) DFC

Proc. Intl. Soc. Mag. Reson. Med. 24 (2016)
1224