Journal of Neurology Research, ISSN 1923-2845 print, 1923-2853 online, Open Access
Article copyright, the authors; Journal compilation copyright, J Neurol Res and Elmer Press Inc
Journal website https://www.neurores.org

Original Article

Volume 14, Number 2, December 2024, pages 49-58


Assessing Internal Reproducibility Within a Parkinson’s Disease Cohort by Leveraging an Independent Larger Dataset

Figures

Figure 1.
Figure 1. Principal component analysis of the discovery cohort (n = 179) based on demographic, motor, and nonmotor characteristics demonstrated two PD subtypes (after imputing missing data and controlling for disease duration). (a) Scatter plot depicting separation of subjects across the first two principal components. (b) Heatmap of clinical characteristics for each cluster identified. The greater red hue indicates greater prevalence, greater blue hue indicates lower prevalence. Traits with a statistically significant difference between clusters, defined as Bonferroni corrected P < 0.05, are indicated with an asterix (*). PD: Parkinson’s disease; RBD: REM-sleep behavior disorder; MAO: monoamine oxidase; PIGD: postural instability and gait disorder; GBA: glucocerebrosidase; LRRK2: leucine-rich repeat kinase 2; H&Y: Hoehn and Yahr scale; PC: principal component.
Figure 2.
Figure 2. Principal component analysis of the PPMI cohort (n = 368) based on demographic, motor, and nonmotor characteristics demonstrated three PD subtypes. (a) Scatter plot depicting separation of subjects across the first two principal components. (b) Heatmap of clinical characteristics for each cluster identified, with data scaled using z-scores. The greater red hue indicates more severe impairment or greater number of years; greater blue hue indicates less severe impairment or fewer number of years. Traits with a statistically significant difference between clusters as determined via ANOVA for continuous data and Chi-square for categorical data, defined as Bonferroni corrected P < 0.05, are indicated with an asterix (*). PD: Parkinson’s disease; PPMI: Parkinson’s Progression Markers Initiative; RBD: REM-sleep behavior disorder; UPSIT: University of Pennsylvania Smell Identification Test; MDS-UPDRS: Movement Disorder Society - Unified Parkinson's Disease Rating Scale; MoCA: Montreal Cognitive Assessment; GDS: Geriatric Depression Scale; STAI: State-Trait Anxiety Inventory; SCOPA: Scales for Outcome in Parkinson’s Disease; ESS: Epworth Sleepiness Scale; QUIP: Questionnaire for Impulsive-Compulsive Disorders in Parkinson’s Disease; PC: principal component; ANOVA: analysis of variance; REM: rapid eye movement.
Figure 3.
Figure 3. Cluster membership of subjects in Discovery cohort remained largely unchanged after re-clustering based on a limited selection of traits. Pink represents subjects assigned to cluster 1 in both analyses. Blue represents subjects assigned to cluster 2 in both analyses. Purple indicates subjects originally assigned to cluster 1 who were reassigned to cluster 2. Black indicates subjects originally assigned to cluster 2 who were reassigned to cluster 1. PC: principal component.

Tables

Table 1. Demographic Characteristics for PPMI and Discovery Cohorts at Baseline
 
PPMI cohortDiscovery cohort
PPMI: Parkinson’s Progression Markers Initiative; F: female; M: male; SD: standard deviation.
Gender count (%)M: 242 (66%)/F: 126 (34%)M: 113 (63%)/F: 66 (37%)
Age at symptom onset (mean ± SD)61.9 ± 8.859.9 ± 9.4
Average current age (mean ± SD)61.9 ± 8.868.7 ± 8.5
Disease duration (mean ± SD)0.54 ± 0.55 years8.8 ± 5.1 years
Family history of PD count (%)Yes: 93 (25%)/no: 275 (75%)Yes: 105 (59%)/no: 74 (41%)

 

Table 2. Comparison of Discovery Cohort Clusters
 
Cluster 1 (n = 92)Cluster 2 (n = 87)Uncorrected P valueBonferroni corrected P value
Characterization of clinical clusters of the Discovery cohort (n = 179) was based on principal component analysis (PCA) and nonhierarchical k-means clustering analysis. Uncorrected and Bonferoni corrected P values are also shown (traits with P < 0.05 were considered to be significantly different between the two clusters) from Chi-square analysis, unless otherwise specified (tunpaired t-test, WWilcoxon test). aVariables that were collected in both the PPMI and Discovery cohorts. The other traits were determined by chart review (MSMD subset) or by interview (NYU subset). Tremor dominant (TD) and Postural Instability and Gait Disorder (PGID) phenotypes were defined based on Movement Disorder Society-Unified Parkinson’s disease rating scale (MDS-UPDRS) score [13]. F: female; M: male; MAOi: monoamine oxidase inhibitor; RBD: REM-sleep behavior disorder; REM: rapid eye movement; SD: standard deviation.
Gendera count (%)M: 63 (68%)M: 50 (57%)0.1581
F: 29 (32%)F: 37 (43%)
Age at diagnosisa (year)t (mean ± SD)58.5 ± 9.661.4 ± 9.00.03340.735
Disease durationa (year)W (mean ± SD)8.7 ± 5.48.9 ± 4.90.6091
+GBA mutation count (%)13 (14%)15 (17%)0.1891
TDa count (%)38 (41%)22 (25%)0.2521
PIGD count (%)24 (26%)32 (37%)0.2581
Hoehn & Yahr scale (score count)1:141:100.4441
1.5:181.5:8
2:522:44
2.5:42.5:10
3:33:12
4:14:2
5:05:1
Wearing off count (%)29 (32%)40 (46%)0.3491
Dyskinesias count (%)23 (25%)36 (41%)0.0280.62
Freezing of gait count (%)16 (17%)39 (45%)0.1791
Depressiona count (%)12 (13%)71 (82%)< 0.001< 0.001
Anxietya count (%)7 (7.6%)73 (84%)< 0.001< 0.001
Hallucinations count (%)6 (6.5%)21 (24%)0.3021
Cognitive impairmenta count (%)0 (0%)6 (6.9%)0.3871
Orthostatic hypotensiona count (%)20 (22%)45 (52%)0.0150.33
Constipationa count (%)46 (50%)71 (82%)0.0591
Urinary symptomsa count (%)39 (42%)37 (43%)0.3141
RBDa count (%)32 (35%)46 (53%)0.3671
Aspirin count (%)18 (19%)21 (26%)0.151
Dopaminergic agent count (%)74 (80%)74 (85%)0.2651
MAOi count (%)22 (24%)10 (11%)0.3211

 

Table 3. Comparison of Demographic and Clinical Characteristics of the Three Clusters From the PPMI Cohort
 
Cluster 1 (n = 71)Cluster 2 (n = 142)Cluster 3 (n = 155)Bonferroni corrected P value
Characterization of clinical clusters of the PPMI cohort (n = 368) was based on principal component analysis (PCA) and nonhierarchical k-means clustering analysis. Uncorrected and Bonferoni corrected P values are also shown (traits with P < 0.05 were considered to be significantly different between the two clusters) from ANOVA analysis, unless otherwise specified (cChi-square). aVariables that were collected in both the PPMI and Discovery cohorts. bData that were collected from screening visit instead of baseline visit, due to data availability. Tremor dominant (TD) and Postural Instability and Gait Disorder (PGID) phenotypes were defined based on Movement Disorder Society-Unified Parkinson’s disease rating scale (MDS-UPDRS) score [13]. Depression was defined as Geriatric Depression scale (GDS) score ≥ 10. Anxiety was defined as state-Trait Anxiety Inventory (STAI) score > 40. Cognitive impairment was defined as Montreal Cognitive Assessment (MOCA) score ≤ 25. Orthostatic hypotension, constipation, and urinary symptoms each was defined as Outcomes in Parkinson’s Disease-Autonomic Dysfunction (SCOPA-AUT) subscore > 0. REM behavior disorder was defined as REM Sleep Behavior Disorder Screening Questionnaire (RBDSQ) score ≥ 5. Impulse control disorder was defined as MDS-UPDRS subscore > 0. Daytime sleepiness was defined as Epworth Sleepiness Scale > 0. Cognitive impairment was defined as Montreal Cognitive Assessment (MOCA) score ≤ 25. PPMI: Parkinson’s Progression Markers Initiative; F: female; M: male; MAOi: monoamine oxidase inhibitor; RBD: REM-sleep behavior disorder; UPSIT: University of Pennsylvania Smell Identification Test; SD: standard deviation.
Genderc counta (%)M: 37 (52%)M: 90 (63%)M: 115 (74%)0.069
F: 34 (48%)F: 52 (37%)F: 40 (26%)
Family historyc counta (%)Y: 20 (28%)Y: 37 (26%)Y: 36 (23%)1
N: 51 (72%)N: 105 (74%)N: 119 (77%)
Age at diagnosis (years) (mean ± SD)61.1 ± 8.757.4 ± 7.666.4 ± 7.5< 0.001
Disease duration (months) (mean ± SD)5.8 ± 5.96.0 ± 6.37.2 ± 7.01
Motor symptoms (MDS-UPDRS part 3) (mean ± SD)20.9 ± 7.515.0 ± 5.426.5 ± 8.0< 0.001
PIGD (mean ± SD)1.5 ± 1.40.64 ± 0.731.4 ± 1.21
TD (mean ± SD)5.2 ± 3.44.5 ± 3.06.5 ± 3.70.008
Hoehn & Yahra scalec (score count)1:351:1051:18< 0.001
2:362:372:135
3:2
Depression (GDS)a (mean ± SD)5.9 ± 3.01.2 ± 1.31.7 ± 1.2< 0.001
Anxiety (STAI)a (mean ± SD)90.4 ± 17.757.9 ± 12.059.6 ± 12.5< 0.001
Cognitive function (MOCA)a, b (mean ± SD)27.1 ± 2.628.0 ± 1.926.5 ± 2.40.101
Orthostatic hypotension (SCOPA-AUT)a (mean ± SD)0.99 ± 1.20.28 ± 0.540.41 ± 0.67< 0.001
Constipation (SCOPA-AUT)a (mean ± SD)1.9 ± 1.60.47 ± 0.781.2 ± 1.10.277
Urinary symptoms (SCOPA-AUT)a (mean ± SD)6.2 ± 6.63.2 ± 2.25.0 ± 4.91
Impulse control disorder (mean ± SD)0.77 ± 1.020.15 ± 0.360.14 ± 0.40< 0.001
Daytime sleepiness (mean ± SD)6.8 ± 4.05.1 ± 3.26.1 ± 3.61
RBD (RBDSQ)a (mean ± SD)5.3 ± 2.93.1 ± 2.04.5 ± 2.81
Smell function (UPSIT) (mean ± SD)21.2 ± 8.225.2 ± 7.720.0 ± 7.90.319