Differential functional dysconnectivity of caudate nucleus subdivisions in Parkinson’s disease

Demographic results

As shown in Table 1, significantly lower MoCA (p < 0.001) and higher HAMD scores (p < 0.001) were detected in patients with PD compared to HEs. No significant difference was found in age and sex between HEs and patients with PD.

VBM results of GM

Whole brain VBM analysis found that no significant GM atrophy was detected in PD patients compared to HEs. Specifically, the GM volumes in the subdivisions of CN were further extracted and compared between the two groups. It was found that no significant difference was detected between HE and PD groups in the CN head (464.34 ± 53.19 for HE vs. 463.79 ± 56.09, p = 0.97), body (412.75 ± 54.73 for HE vs. 413.38 ± 48.57 for PD, p = 0.96), and tail (39.15 ± 4.84 for HE vs. 39.77 ± 5.58 for PD, p = 0.628), respectively.

Specific subdivision networks of the CN in the HE

CN subdivision networks in HEs were identified in the whole CN network (Figure 1A and Supplementary Table 1). Compared to the other CN subdivisions, the head showed a specific higher connectivity with the anterior cingulate cortex (ACC) extending into medial prefrontal cortex (MPFC) and the CN head, and the precuneus (PCu) extending into the posterior cingulate cortex (PCC) bilaterally (see Table 2 and Figure 1B). The body showed specific higher connectivity with the CN body extending into the thalamus, the dorsolateral prefrontal cortex (DLPFC) extending into the anterior prefrontal cortex (APFC), and the angular gyrus (AG), bilaterally (see Table 2 and Figure 2C). The tail showed higher connectivity with the hippocampus and occipital cortex bilaterally (see Table 2 and Figure 1D). Altered FC in the CN subdivision networks were revealed in PD OFF-mediation and ON-medication (see Supplementary Figure 1).

Figure 1. Intrinsic connectivity networks of caudate nucleus (CN) subdivisions in healthy elderly (HE). (A) Whole CN network; (B) CN head network; (C) CN body network; (D) CN tail network. Results were illustrated at an uncorrected voxel-wise height threshold of p < 0.001 combined with an FWE-corrected cluster-wise threshold of p < 0.001

Altered functional connectivity and dopamine modulation in CN subdivision networks in PD

These regions in the specific CN subdivision network were defined as regions of interest (ROIs) for the confirmation analyses. All FC values in each ROI were extracted and analyzed. Patients with PD OFF-medication exhibited a significantly decreased connectivity within the CN head network compared to HEs (t = -3.217, p = 0.002 for the ACC/MPFC/CN head cluster and t = -3.016, p = 0.004 for the PCu/PCC cluster). When L-dopa was administered to patients with PD, a significantly improved connectivity was revealed in the CN head/ACC/MPFC cluster and this improvement returned to the normal level (ON-medication vs. OFF-medication, t = 5.605, p < 0.001; ON-medication vs. HE, t = 1.217, p = 0.228) (see Figure 2A).

Figure 2. Functional connectivity (FC) changes in caudate nucleus (CN) subdivisions between PD and healthy elderly (HE), and between PD OFF-medication and ON-medication. (A) CN head network; (B) CN body network; (C) CN tail network. Results were illustrated at an uncorrected voxel-wise height threshold of p < 0.001 combined with an FWE-corrected cluster-wise threshold of p < 0.001. Bar graphs demonstrated FC values in the subdivision networks in HE and patients with PD ON-medication and OFF-medication; ^**, p < 0.005; ^***, p < 0.001. Abbreviations: FC, functional connectivity; PD-OFF, PD OFF-medication; PD-ON, PD ON-medication.

Within the CN body network, significantly decreased connectivity was revealed in the DLPFC/APFC cluster (t = -3.080, p = 0.003) and the left AG (t = -3.157, p = 0.002) in PD patients OFF-medication compared to HEs. A significant improvement in the DLPFC/APFC cluster was found after administration of L-dopa (ON-medication vs. OFF-medication, t = 2.97, p = 0.006; ON-medication vs. HE, t = -0.343, p = 0.733) (see Figure 2B).

Within the CN tail network, the reduced connectivity in the bilateral hippocampus (t = -2.983, p = 0.004 for the left; t = -2.576, p = 0.012 for the right) and bilateral cuneus (t = -3.340, p = 0.001) was exhibited in PD OFF-medication compared to HE (see Figure 2C). No significant changes were identified in the CN tail network in patients with PD when administered L-dopa.

Exploratory analysis of dopamine effect in PD

Three voxel-wise exploratory analyses were conducted within each subdivision network to refine the regions affected most by the dopamine administration within each CN subdivision network. The left MPFC (MNI coordinate: x = -9, y = 21, z = -9; t = 3.46) in the CN head network was significantly modulated by dopamine treatment in PD. The mean FC strength was significantly negatively correlated with the motivation measured by UPDRS part I in patients with PD (see Figure 3A). The left DLFPC (MNI coordinate: x = -45, y = 21, z = 36; t = 4.53) in the CN body network was significantly modulated by dopamine treatment in PD. Mean functional connectivity strength was significantly positively correlated with MoCA score in patients with PD (see Figure 3B). No voxels showed significant dopamine effect in the CN tail network.

Figure 3. Increased FC in PD ON-medication. (A) Increased FC in the MPFC in CN head network; (B) Increased FC in the DLPFC in CN body network. Bar graphs showed the extracted FC values in HE and PD. Scatterplots showed the relationship between FC values and neuropsychological measurements. Abbreviations: FC, functional connectivity; MPFC, medial prefrontal cortex; DLPFC, dorsolateral prefrontal cortex; MoCA, Montreal Cognitive Assessment; HE, healthy elderly; PD-OFF, PD OFF-medication; PD-ON, PD ON-medication. ^***, p < 0.001.

Participants

In this study, 34 early stage PD patients (18 males, 57.11 ± 11.85 years old vs. 16 females, 64.13 ± 9.27 years old) with Hoehn and Yahr (H&Y) stage of I or II, meeting the UK Bank criteria for the diagnosis of PD [21], were recruited. All PD participants were examined in their ON-medication and OFF-medication states (at least after a 12-hour withdrawal of anti-Parkinson medication). Patients received L-dopa 30 minutes after receiving 10 mg domperidone on an empty stomach in the morning. L-dopa dosage was determined by the regular effective dosage of the patient. ON-medication examinations were administered an hour later when L-dopa reaches its peak plasma dose. The HE comprised 26 volunteers (10 males, 56.12 ± 10.07 years old), who were matched for age and sex ratio. This study was approved by the Research Ethics Committee of Beijing Tiantan Hospital, Capital Medical University. Written informed consent was obtained from each participant.

Neuropsychological assessments

The clinical characteristics were assessed ON-medication except UPDRS part-III which was evaluated both ON- and OFF-medication. The disease stage was scored using the H&Y stage score and disease severity was captured by the UPDRS part-III. Assessments of the neuropsychological state included the Montreal Cognitive Assessment (MoCA) and the Hamilton Depression Rating Scale (HAMD-24) [22].

MRI data acquisition

MRI data were acquired using a Siemens 3 Tesla Prisma MRI system (Siemens, Erlangen, Germany) by using a 32-channel head coil. Foam padding and headphones were used to limit head motion and reduce scanner noise. Participants were instructed to keep still and remain motionless. High-resolution structural images were acquired using 3D T1-weighted magnetization-prepared rapid gradient echo (MPRAGE). Scan parameters were as follows: TR = 2300 ms, TE = 2.3 ms, flip angle = 8°, matrix size =256 × 256, 192 1 mm sagittal slices. Resting-state functional MR imaging (rs-fMRI) data were obtained by using an echo-planar imaging sequence that lasted 8 minutes 20 seconds (200 volumes) with the parameters: TR = 2500 ms, TE = 30 ms, flip angle = 90°, FOV = 200 mm × 200 mm, matrix size =70 × 70, 43 slices, and slice thickness = 3 mm.

Voxel-based morphometry (VBM) analysis

To determine the structural abnormalities in patients, we performed a VBM analysis for all anatomical images of PD and HE using SPM12 (http://www.fil.ion.ucl.ac.uk/spm) and CAT12 (http://www.neuro.uni-jena.de/cat/). T1-weighted images were segmented using the unified segmentation model into gray matter (GM), white matter (WM), and cerebrospinal fluid (CSF) based on tissue probability maps (TPMs) in MNI space. The spatially normalized GM maps were modulated by the Jacobian determinant of the deformation field and corrected for individual brain sizes. The modulated, normalized GM images (voxel size 1.5 × 1.5 × 1.5 mm³) were smoothed with an 8-mm full width at half maximum isotropic Gaussian kernel.

Rs-functional connectivity analysis

Rs-fMRI data were preprocessed using SPM12. Seed-to-voxel correlation analysis was carried out by Data Processing and Analysis for (Resting-State) Brain Imaging (DPABI) V4.3 [23]. The first ten functional images were discarded to minimize the fluctuation of the MRI signal in the initial stage of scanning. The remaining 190 images of each subject were first corrected for slice timing to reduce the within-scan acquisition time differences between slices and then realigned to eliminate the influence of head motion during the experiment. All subjects included in the present study exhibited head motion less than 1.5 mm in any of the x, y, or z directions and less than 1.5^° of any angular dimensions. Next, the realigned images were co-registered to T1 images, spatially normalized into MNI space using transformations from segmentation, and resampled voxel size into 3 × 3 × 3 mm³. Subsequently, the functional images were smoothed with a 4-mm FWHM isotropic Gaussian kernel. After preprocessing, images were then band-pass filtered to 0.008 ~ 0.09 Hz to reduce noise. Further denoising steps included regression of six motion parameters and their first-order derivatives, regression of WM and CSF signals following the CompCor strategy [24] and a linear detrending. The seed-regions of the bilateral caudate head, body, and tail were defined based on WFU_PickAtlas [25] (Figure 4). The correlation coefficients between the seed voxels and all other brain voxels were computed to generate correlation maps. For group analyses, the correlation coefficients were converted to z-value using Fisher's r-to-z transformation [26].

Figure 4. Anatomical seeds of caudate nucleus (CN) subdivisions. Red color represents CN head; Green color represents CN body; Violet color represents CN tail.

Identifying the CN subdivision networks

The three individual connectivity maps (obtained from the CN head, body, and tail) of each HE were entered into a factorial model in SPM12 with age and sex as nuisance variates. The whole CN network was first identified (defined as the regions connected with any of the three CN subdivisions with F-contrast in SPM). Functional connectivity maps specific to each seed region were obtained by comparing to the other two CN subdivisions.

Confirmatory analysis of changes in CN subdivision network and dopamine effect in PD

These regions within each specific CN subdivision network were defined as regions of interest (ROIs) for the confirmatory analyses. Two sample t-tests were performed to explore the dysconnectivity in PD compared to HE. Paired t-test was conducted to examine the dopamine modulation within each subdivision network in PD ON-medication compared to OFF-medication.

Exploratory analysis for dopamine modulation in the CN subdivision networks

An exploratory analysis was performed additionally to ascertain which voxels were affected by the dopamine administration within each CN subdivision network. Partial correlation analyses were examined between the mean functional connectivity values in these voxels and clinical performances controlled for nuisance variables.

Statistical analysis

Two-sample t-test was performed for the demographic, clinical measurements, and extracted CN subdivision volume between PD patients and HE using SPSS 22. GM volume and functional connectivity results were reported based on an uncorrected voxel-wise height threshold of p < 0.001 combined with an FWE-corrected cluster-wise threshold of p < 0.001. Brain regions were localized with xjView (http://www.alivelearn.net/xjview). Correlations between these alterations and neuropsychological performances were examined. Results for confirmatory ROIs analysis and correlations were corrected for multiple comparisons using the Bonferroni correction (p < 0.05).