Participants

Participants were included from the Comprehensive Assessment of Neurodegeneration and Dementia (COMPASS-ND). ^{Reference Chertkow, Borrie and Whitehead13} COMPASS-ND is a Canadian national study investigating those with dementia or other cognitive concerns that is observational in nature. ^{Reference Chertkow, Borrie and Whitehead13} This sub-study of COMPASS-ND (Pro00061456) was approved by the University of Alberta Research Ethics Board (Pro00099707). All participants provided written informed consent for the study. CU participants were from the COMPASS-ND study and CBBRAIN, a neuroimaging study of healthy controls. ^{Reference Treit, Naji and Seres14} For the CBRAIN study, participants underwent cognitive screening using the National Institutes of Health Toolbox Cognitive Battery ^{Reference Denboer, Nicholls, Corte and Chestnut15} which encompasses testing for verbal recall, working memory, processing speed, and episodic memory where participants needed to score within two standard deviations of the mean; participants also could not have a self-reported history of brain injury, neurologic or psychiatric conditions. ^{Reference Solar, Treit and Beaulieu16} Inclusion and exclusion criteria for the PD participants were the same as for COMPASS-ND in general (outlined below, also see ^{Reference Chertkow, Borrie and Whitehead13,Reference Pieruccini-Faria, Black and Masellis17} for additional information). Consecutive patients with PD or PD with mild cognitive impairment (MCI) at the University of Alberta were included. Participants with PD (with or without MCI but without dementia; n = 29, referred to as the PD group) and age-matched CU (n = 27) were included. PD participants were not selected per se but, rather, were subjects that were willing to undergo additional neuroimaging at the University of Alberta that included QSM. This included patients who were on treatment and not on treatment. Patients who had undergone deep brain stimulation prior to study initiation were not included.

Functional impairment on the basis of cognition was necessary for the diagnosis of dementia. A Montreal Cognitive Assessment (MoCA) score <25 in a participant who otherwise met criteria for PD (based on the MDS criteria for MD) ^{Reference Postuma, Poewe and Litvan18} defined PD-MCI, consistent with published criteria. ^{Reference Litvan, Goldman and Troster19} PD participants could be taking PD medications: levodopa equivalent daily dose was calculated. ^{Reference Tomlinson, Stowe, Patel, Rick, Gray and Clarke20} Exclusion criteria have been published and included significant known chronic brain disease, developmental disorders, malignant tumors, current alcohol or drug abuse, lack of a study partner, insufficient proficiency in English or French, total MoCA score of less than 13, inability to comprehend test instructions, and incapable of walking more than 10 m independently. ^{Reference Chertkow, Borrie and Whitehead13,Reference Pieruccini-Faria, Black and Masellis17}

Participant Assessments

Patient assessments include the Movement Disorders Society United Parkinson’s Disease Rating Scale (MDS-UPDRS). Items from Part II (patient’s view of motor aspects of experiences of daily living) and Part III of the MDS-UPDRS (motor examination completed by the clinician) addressing gait difficulties and FOG were used. ^{Reference Goetz, Tilley and Shaftman21} The items included were 2.12 (difficulties with walking and balance), 2.13 (freezing), 3.10 (general gait impairment), and 3.11 (FOG specifically). Hoehn and Yahr scores provide an estimation of overall impairment in an individual with PD. ^{Reference Goetz, Poewe and Rascol22}

Image Acquisition

Whole brain QSM images were acquired on a 3T (Siemens Prisma) scanner using a 20-channel head coil. Acquisition parameters were as follows: axial multi-gradient echo sequence with seven echoes (first echo at 3.82 ms and subsequent echoes spaced at 5.49 ms), TR = 45.0 ms, flip angle = 17°, 80 slices, 2.0 mm slice thickness for a total coverage of 160 mm, 240 × 218 mm FOV, voxel dimensions 0.9 mm × 0.9 mm × 2.0 mm = 1.62 mm³ without interpolation, parallel imaging (generalized autocalibrating partially parallel acquisitions (GRAPPA) with an acceleration factor of 2), and a scan time of 5.65 minutes. Whole brain sagittal T1-weighted images were acquired using 3D magnetization-prepared rapid gradient echo (MPRAGE) with the following parameters: 192 slices, 1.0 mm slice thickness; 256 mm FOV; image matrix 256 × 256; voxel size of 1.0 mm × 1.0 mm × 1.0 mm = 1.0 mm³ with no subsequent interpolation; TR = 2300 ms; TE = 2.98 ms; TI = 900 ms; flip angle 9°; parallel imaging (GRAPPA with acceleration factor of 2); scan time of 5.35 minutes.

Processing of Susceptibility Maps

After saving raw phase images, QSM was reconstructed using a series of steps including phase unwrapping, background field removal, and then susceptibility inversion. ^{Reference Sun, Walsh and Lebel23} Briefly, brain tissue was extracted using brain extraction tool ^{Reference Smith24} on the magnitude image of the first echo; coil-combined phase was unwrapped using phase region expanding labeler for unwrapping discrete estimates (PRELUDE), ^{Reference Jenkinson25} background field was removed with the regularization enabled sophisticated harmonic artifact reduction on phase data (RESHARP) method, ^{Reference Sun and Wilman26} and dipole inversion via total variation. ^{Reference Bilgic, Pfefferbaum, Rohlfing, Sullivan and Adalsteinsson27}

ROI Analysis

ROIs investigated in our study included the caudate nucleus, globus pallidus, putamen, pulvinar nucleus of the thalamus, substantia nigra, red nucleus, and dentate nucleus. These were selected based on previous studies that have looked at and shown susceptibility changes in PD patients. ^{Reference Chen, Cai and Li11,Reference Cheng, Huang and Liang28}

Using ImageJ (https://imagej.nih.gov/ij/, 1997–2021), ROIs were drawn manually on the susceptibility maps by a single rater (NN) who was blinded to participant group using three separate axial slices. The slice with the largest cross section of the globus pallidus that also enabled visualization of each of the caudate nucleus, putamen, and pulvinar nucleus of the thalamus was selected as the first slice for looking at the aforementioned ROIs. ROIs were drawn bilaterally on the most central part of each area of interest (caudate nucleus, putamen, globus pallidus, pulvinar nucleus of the thalamus) with attention being given to avoid encompassing the edges of a given ROI to avoid inclusion of noise in the susceptibility measurement as much as possible. ROIs were placed bilaterally on a more caudal single slice in axial orientation (second slice) that best delineated the substantia nigra and red nucleus. ROIs for the dentate nucleus were placed bilaterally on an even more caudal single slice (third slice) that delineated this structure best.

The posterior limb of the internal capsule was examined as a reference region for which the ROI was drawn bilaterally on the same slice as the one that was used to outline the caudate, putamen, globus pallidus, and pulvinar nucleus of the thalamus (Figure 1). For each ROI tracing aside from the substantia nigra which was drawn individually for each subject, the size of each individual ROI was kept consistent across all subjects to enable for more consistent susceptibility measurements. Tracing was repeated on data from 10 randomly selected participants after an interval of 1 week to determine intra-rater reliability.

Figure 1: Sample quantitative susceptibility map outlining the regions of interest selected for analysis.

Volumetric Analysis

To determine the presence or absence of atrophy of structures included as ROIs, volumetric analysis was undertaken using T1-weighted images. Automated segmentation of subcortical brain regions and ventricles was performed using the "recon-all" pipeline in FreeSurfer (v6.0). ^{Reference Fischl29} Processing was done on the Canadian Brain Imaging Research Platform, ^{Reference Sherif, Rioux and Rousseau30} a web-based software for distributed computing for neuroimaging research. Volumes of the caudate, putamen, globus pallidus, thalamus, and intracranial volume were estimated. The left and right volumes were summed and reported as a % of intracranial volume (summed volume/intracranial volume × 100%).

Clinical Gait Assessment

The original FOG questionnaire (FOGQ) developed by Giladi and colleagues was used. ^{Reference Giladi, Shabtai, Simon, Biran, Tal and Korczyn31} This is a self-reported questionnaire with items addressing different aspects of gait which include the following: need for assistance during the individual’s worst state of gait; if gait difficulties are affecting activities of daily living; if feet feel glued to the floor while walking or turning (freezing); the duration of the freezing episode; the duration of the typical start hesitation (freezing at time of gait initiation); and the duration of turning hesitation (freezing when turning). ^{Reference Giladi, Shabtai, Simon, Biran, Tal and Korczyn31}

The PD group was classified into two different ways with respect to gait. First, participants were categorized as gait impaired if FOGQ total score ≥ 1 (n = 23) or not have gait impairment (n = 6). Second, participants were divided into those with FOG (FOG+, n = 12) and those without (FOG−, n = 17) where the FOG+ group consisted of the participants who scored ≥1 on items 3–6 inclusive. This approach was chosen as other studies have used only item 3 from the FOGQ ^{Reference Naduthota, Honnedevasthana and Lenka8,Reference Canu, Agosta and Sarasso32} ; however, items 4−6 also encompass other pertinent measures of freezing (i.e., duration of freezing episode, start hesitation, turning hesitation) which would help reflect the degree of FOG.

Quantitative Gait Assessment

Quantitative gait measures were determined as previously described in the ON state (on medication). ^{Reference Pieruccini-Faria, Black and Masellis17,Reference Cullen, Montero-Odasso and Bherer33} In brief, participants were instructed to walk on a 6 m electronic walkway (Zeno®- Protokinetics™) at their self-selected pace, starting 1 m from the walkway and ending 1 m beyond the walkway to minimize recording of accelerations and decelerations. Three trials were performed, and gait parameters were derived using Zeno®- Protokinetics™ software (stride velocity, cadence, stride length, step time, stride time variability, and double support time).

Statistical Analyses

GraphPad Prism (version 9.0.2; GraphPad Software, LLC) was used for statistical analyses. To compare groups with respect to participant demographics, all continuous variables were compared using two-sided t-tests and all categorical variables were compared using Fisher’s exact test, with the exception of Hoehn Yahr score distribution between the groups, which was compared with Chi-squared test.

Intra-class correlation coefficient was used for 10 participants who had repeat ROI tracing (test, re-test, blinded to participant group in all instances, conducted by NN) to determine intra-rater reliability. To compare QSM values for each ROI between the left and right sides in CU and PD, unpaired t-tests were used. As there was no significant difference between left and right sides for any ROIs in CU and PD (p > 0.05), the average of left and right-side values was used for the remainder of analysis.

To test for whether the data followed a normal distribution or not, the D’Agostino and Pearson test was used. To compare QSM values for each ROI between CU and PD, unpaired t-tests were used for data that was normally distributed; in cases where data were not normally distributed, the Mann–Whitney nonparametric test was used instead. Pearson correlation coefficient was used to look for correlation between age and QSM values for each ROI in CU and PD; to look for correlation between sex and QSM values for each ROI in CU and PD; and to look for correlation between UPDRS-III total score and QSM values for each ROI in PD. In cases where data were not normally distributed, the Spearman correlation nonparametric test was used instead.

Unpaired t-tests were used to compare QSM values for each ROI between gait impaired and non-gait impaired PD participants. Unpaired t-tests were used to compare each of the quantitative gait measures between gait impaired and non-gait impaired PD participants. As above, in cases where data were not normally distributed, the Mann–Whitney nonparametric test was used instead. Unpaired t-tests were used to compare ROI volumes between gait impaired and non-gait impaired, and FOG+ versus FOG−. A p-value of 0.05 (two-tailed) was deemed statistically significant.