The Effect of Genotype on Self‐Reported Dysarthria and Dysphagia in Parkinson's Disease: A Parkinson's Progression Marker Initiative Study

ABSTRACT

Objectives

The objective of this study was to examine baseline and longitudinal differences of self‐reported dysarthria and dysphagia in the most common genetic subtypes of Parkinson's disease (PD) using the Parkinson's Progression Marker Initiative (PPMI) dataset.

Methods

This was a retrospective, longitudinal study utilizing data from the PPMI dataset. Dysarthria‐ and dysphagia‐specific questions from the Unified Parkinson's Disease Rating Scale (UPDRS) and Scales of Outcomes in Parkinson's Disease‐Autonomic questionnaire (SCOPA‐AUT) were extracted for people with Parkinson's disease (PwPD) with leucine‐rich repeat kinase 2 gene (LRRK2), GBA, and SNCA genotypes across up to five research visits. Relevant patient (age, sex, etc.) and disease (severity, phenotype, medication, etc.) data were extracted along with scores from the Montreal Cognitive Assessment (MoCA). Linear mixed models (LMMs) were used to analyse longitudinal data between genotypes, as well as to examine the interaction effects between genotypes and tremor dominant (TD) or postural instability/gait determinant (PIGD) phenotypes.

Results

A total of 211 PwPD met inclusion criteria at their baseline visit (LRRK2 n = 115, GBA n = 68, SNCA n = 15). LMMs displayed significant differences in genotypes longitudinally, with significant differences between LRRK2, GBA, and SNCA genotypes at multiple time points in dysarthria and dysphagia self‐reports. LRRK2‐genotyped participants routinely self‐reported lower dysarthria (p < 0.001) and dysphagia severity for UPDRS (p < 0.001) and SCOPA‐AUT (p = 0.007) questions. SNCA‐genotyped participants self‐reported the most severe dysarthria (p = 0.002) and dysphagia symptoms for UPDRS (p < 0.001) and SCOPA‐AUT (p < 0.05) over time. There were no differences between genotypes at baseline, and no effects of motor phenotype at any time point.

Conclusions

This was the first study to examine longitudinally how genotypes in PD specifically impact self‐reported dysarthria and dysphagia severity. Findings from our study suggest different genotypes of PD affect the degree of self‐reporting dysarthria and dysphagia severity. Specifically, LRRK2 genotypes self‐reported lower dysarthria and dysphagia severity, while SNCA genotypes self‐reported the most severe dysarthria and dysphagia of this sample. Importantly, SNCA genotypes self‐report a faster increase in severity over time compared to other genotypes. Substantially more work is needed to investigate the underlying physiological differences manifesting in dysarthria and dysphagia in different genotypes of PD.

WHAT THIS PAPER ADDS

What is already known on this subject

• Limited evidence is available regarding the effects that different genotypes of Parkinson's disease (PD) have on speech, voice, and swallowing. Current evidence suggests differences in voice acoustic signals between idiopathic and genetic Parkinson's, but with no other evidence in other domains such as swallowing. Additionally, there is no current data on how different genotypes perceive or report speech, voice, or swallowing impairment.

What this study adds to the existing knowledge

• This study adds novel, preliminary evidence from a large open‐source dataset that different genotypes of PD may report differential impairment in speech, voice, and swallowing. Importantly, the progression of reported dysarthria and dysphagia over time interacts with these genotypes, suggesting different genotypes experience differential impairment in dysarthria and dysphagia as time goes on. Specifically, SNCA‐genotyped people with PD may report significantly greater dysarthria and dysphagia than other genotypes.

What are the potential or actual clinical implications for this study?

• Clinically, this study highlights how different genotypes of PD experience dysarthria and dysphagia over the course of their disease and provides potentially relevant timelines of when these impairments reach clinically meaningful thresholds. As an example, baseline reports of dysarthria and dysphagia were similar and remained stable up to the third year of reports, where significant differences between genotypes began to emerge. This may provide clinicians with useful information regarding when to expect dysarthria and dysphagia in genotypic PD to manifest and when to best initiate additional assessment and treatment based on patient information.

1. Introduction

Parkinson's disease (PD) is a common neurodegenerative disease characterized by neuronal loss in the brain, specifically in the substantia nigra of the basal ganglia, with the presence of Lewy bodies throughout the central nervous system (Poewe et al. 2017; Sveinbjornsdottir 2016). While motor symptoms in the extremities are often the most salient symptomology of the disease, people with Parkinson's disease (PwPD) often display speech, voice, and swallowing dysfunction, leading to dysphagia (swallowing impairment) and dysarthria (speech/voice impairment) (Poewe et al. 2017; Sveinbjornsdottir 2016; Sapir et al. 2008).

A major hurdle in the screening, assessment, and management of dysphagia and dysarthria in PwPD is the individualistic presentation and manifestation of many disease symptoms at any given time point of the disease (Luo et al. 2019; Mestre et al. 2021; Politis et al. 2010). Despite this individuality in disease manifestation, efforts such as genotyping (i.e., grouping PwPD based on identification of specific genes with a manifestation of Parkinson's) or phenotyping (i.e., grouping PwPD based on common manifestations of clinical symptomology) are considered useful in identifying the most salient disease presentations to improve clinical management, diagnosis, medication efficacy, and progression expectations (De Pablo‐Fernández et al. 2019; Marras, 2015; Thenganatt and Jankovic 2014; Von Coelln and Shulman 2016).

To date, there is limited available evidence examining how genotypes of PD may impact speech, voice, or swallowing function. The majority of available PD research in dysarthria or dysphagia has centred around iPD, and understandably so, given that genetically linked development of PD only accounts for approximately 25% of the risk for developing PD (Day and Mullin 2021). There are currently no studies that have investigated how different genotypes of PD impact swallowing, and only one has investigated dysarthria (Arora et al. 2018), to the best of our knowledge. Consequently, a substantial knowledge gap exists regarding the effects genotypic PD may have on the manifestation of dysarthria and dysphagia in known genetic subtypes of the disease.

Differences have been documented in the underlying pathology between genetic types of PD (PD caused by a known genetic mutation) and idiopathic PD (iPD; PD without a known genetic cause). Exploring how these differences contribute to dysarthria and dysphagia manifestations is needed and important, as genetic and iPD may be individualistic in their manifestation and pathological mechanism. Genetic mutations responsible for a subset of PD may have specific clinical presentations that can manifest similar to iPD or as a separate phenotype altogether (Guadagnolo et al. 2021; Klein et al. 2018). Certain genetic mutations that are known risk factors and causes of PD, such as the leucine‐rich repeat kinase 2 gene (LRRK2), may clinically appear as typical iPD (Berg et al. 2005) yet may also have substantial variability and different neuropathology (i.e., cause of PD symptoms) compared to other PD genotypes (Puschmann 2013). As an example, it is thought that SNCA (alpha synuclein) genotypes are expected to present with Lewy body along with alpha‐synuclein pathology, while Lewy body pathology is rare in many LRRK2 genotype manifestations (Jia et al. 2022). In a larger summary of phenotypic presentation of autosomal dominant forms of PD such as GBA (acid beta glucosidase), SNCA, and LRRK2, Koros et al. (2017) highlight how LRRK2 may present quite similarly to iPD, while GBA is often considered to present in a more severe form. Regardless of any similarities to iPD, it is still clear that the clinical presentation, even within the same mutations of a genotype, will vary (Schneider and Alcalay 2017).

Given the potential clinical variability of genetic subtypes of PD, similar to iPD, describing a patient's individual clinical presentation as a phenotype may be of clinical interest. More research is needed to continue elucidating the role existing motor phenotypes (such as tremor dominant [TD], non‐tremor dominant [NTD], or postural instability/gait dominant [PIGD]) play in further defining the individualistic manifestation of genetic PD. Trinh et al.'s (2018) findings suggest SNCA genotypes may have more NTD symptomology, such as impaired cognitive function, and others have summarized SNCA genotypes may also experience substantial motor involvement and fluctuations as well, while LRRK2 genotypes often have less NTD phenotypic involvement (Jia et al. 2022).

In terms of how these phenotypes may intersect with communication and swallowing, a recent review (Thijs and Dumican 2023) highlighted that current available data supports the hypothesis that TD phenotypes of PD tend to present with less severe voice and swallowing issues. However, this work is limited primarily to iPD. Though more research is warranted, these findings underscore that although certain forms of genotypic PD may manifest similarly to iPD, phenotyping based on clinical symptoms may uncover specific differences between genotypes. Exploring the potential variability in presentation, including in communication and swallowing, may provide important context in terms of manifestation and progression of communication and swallowing symptoms for clinicians to provide individualized and patient‐centred care.

A starting point of investigation may therefore be examining how patients from different genotypes, and their associated phenotypes, of PD self‐report the presentation and subsequent changes of speech, voice, and swallowing. A viable resource to perform this investigation is using self‐reported dysarthria and dysphagia scores from the Parkinson's Progression Marker Initiative (PPMI) dataset. These scores have been used in previous studies utilizing PPMI data in iPD cohorts (Polychronis et al. 2019a; Polychronis et al. 2019b; Watts and Zhang 2022). However, this has not been previously investigated in genotypes of PD. Prior research suggests that utilizing self‐reported dysarthria and dysphagia from the Unified Parkinson's Disease Rating Scale (UPDRS), as well as other speech, voice, or swallowing questionnaires has shown clinical utility as a screening approach (Dumican et al. 2024).

As such, there were three overarching questions this study sought to examine regarding how genotypes of PD differ in self‐reported dysarthria and dysphagia: (1) at baseline assessments and (2) longitudinally, at subsequent yearly visits, as well as (3) how these self‐reports were impacted by categorized tremor phenotypes and their interactions with genotypes at each yearly visit. These questions were addressed using a free and publicly available data set from the Parkinson's Progression Markers Initiative (PPMI) database.

Based on the limited extant literature available, we hypothesized that (1) LRRK2 genotypes may self‐report less dysarthria and dysphagia compared to other genotypes, (2) self‐reported dysphagia and dysarthria would decline longitudinally, with LRRK2 genotypes self‐reporting significantly less decline longitudinally than other genotypes, and (3) non‐tremor phenotype patients would self‐report significantly worse dysarthria and dysphagia compared to other phenotypes both within and across genotypes.

2. Methodology

2.1. Data Source

Data were sourced from the PPMI study database (https://www.ppmi‐info.org/access‐data‐specimens/download‐data/). The PPMI database contains longitudinal patient and disease‐specific data during an 8‐year study period. This data is broadly organized into ‘sporadicʼʼ or idiopathic PD and genetic PD. Data was extracted from participants in the genetic cohort who received MDS‐UPDRS (Goetz et al. 2008) assessments for dysarthria (UPDRS question 2.1; ‘Speech’) and dysphagia (UPDRS question 2.3; ‘Chewing/Swallowing’) for up to 6 yearly visits beyond baseline (i.e., conducted at baseline, 12 months, 24 months, 36 months, etc.). We also extracted questions 1 and 3 of the Scales of Outcomes in Parkinson's Disease‐Autonomic questionnaire (SCOPA‐AUT) (Visser et al. 2004), as these questions specifically ask about swallowing function and potential dysphagia that may be interpreted outside the scope of autonomic gastrointestinal function. Participant answers to these questions may therefore add additional context to dysphagia reports. Though dysarthria reports are a major component of this study's analysis, there were no other self‐reported speech, voice, or dysarthria‐focused measures collected throughout the PPMI. Given that the goal of this study is to examine self‐reported dysarthria and dysphagia, and not clinician‐reported severity, only UPDRS question 2.1 was deemed relevant and included.

A potential limitation of utilizing the PPMI is the documented high attrition rates beyond year 6 of the PPMI study (https://www.ppmi‐info.org/study‐design/study‐cohorts/). In another study that utilized the PPMI, Watts and Zhang (2022) utilized year 6 as their inclusion criteria, with participants needing complete data up to visit year 6. We utilized a similar cutoff in data extraction of the studies' measures to ensure that the sample size was robust to patient attrition during the longitudinal portions of our analysis.

2.2. Participants

As a part of the PPMI data collecting procedures, each patient underwent additional comprehensive assessments at regular intervals in addition to the MDS UPDRS assessments, though not all data were collected at every visit interval (i.e., 3 months, 6 months, 18 months, etc.). Participants were staged using the Hoehn and Yahr (H&Y), medication status and amount of levodopa or equivalent medication dosage (LEDD), a confirmed diagnosis of PD during study participation (i.e., participants who completed up to six consecutive yearly visits but were ultimately not diagnosed with PD, but another neurological condition, were excluded), and genetic testing that accompanied clinical assessments that confirmed a genetic subtype of PD (i.e., participants for this study could not have had idiopathic or sporadic PD).

2.3. Measures

2.3.1. Dysarthria and Dysphagia Questions

The dependent variables of interest for this study included in our analyses were UPDRS questions 2.1 and 2.3 (UPDRS 2.1; UPDRS 2.3) and SCOPA‐AUT questions 1 and 3 (SCOPA‐AUT Q1; SCOPA‐AUT Q3). The specifics of these questions, including their scoring, can be found in the Appendix A. UPDRS 2.1 and 2.3 have been utilized in PPMI‐specific research previously (Watts and Zhang 2022), as has the SCOPA‐AUT (Polychronis et al. 2019). Of note, UPDRS questions 2.1 (Speech) and 2.3 (Chewing/Swallowing) were extracted from Section 2 of the UPDRS from the PPMI, whereby, according to PPMI documentation, these sections were only assessed in the ‘ON’ state of the patient. Data for Section 3  of the UPDRS may be administered either in the ‘ON’ or ‘OFF’ state in the PPMI, but this was not the case for Section 2. A specific listing of all possible answers and corresponding scores can be found in the Appendix A.

There is mixed guidance regarding the measurement of questions such as 2.1 and 2.3 from the UPDRS. Clinically based studies of the UPDRS continue to recommend sum scores and reports in mean and standard deviation of UPDRS sections (Goetz et al. 2023), as do voice‐based studies (Pu et al. 2021). However, it has also been suggested that questions from the UPDRS are more closely ordinal or categorical in nature, albeit with cautious interpretation (Hendricks and Khasawneh 2021). Given varying interpretations, we report UPDRS question scores in both mean and standard deviation as well as in median and IQR.

Additionally, research has highlighted a lack of consensus on exactly what UPDRS question 2.1 in particular is designed to measure: speech function, voice function, or both, given that elements of reflecting on voice‐related domains such as volume are included in the question. However, available research suggests that other self‐reported measures of voice, such as the Voice Handicap Index (VHI) and Voice Related Quality of Life (V‐RQOL), do not align with UPDRS scores (Dumican et al. 2024, Midi et al. 2007). Therefore, we chose to refer to UPDRS 2.1 and any reported impairment as “dysarthria” rather than solely ‘speech’, as per the question title.

2.3.2. Person‐ and Disease‐Centred Measurements

The independent variables of interest for this study included the confirmed genotype of each participant, their categorized motor phenotype, and the year of each visit for longitudinal examination. Clinical motor phenotype was determined based on previously published guidelines (Stebbins et al. 2013) as a calculation of several UPDRS questions to derive either a TD or NTD classification, with the NTD classification consisting of either PIGD or indeterminate phenotypes. This method of calculation is included in the PPMI dataset as its own, pre‐filled, and coded variable within the MDS‐UPDRS scoring. Briefly, 11 tremor‐based items and 5 PIGD items from the MDS‐UPDRS are separately averaged together to calculate a tremor score and PIGD score. The Tremor score is then divided by the PIGD score to derive a ratio. If this ratio is ≥ 1.15, then a given participant is denoted as TD. For the purposes of the PPMI, participants are classified broadly as NTD if this ratio is below 1.15. Motor phenotype was assessed and determined based on the participant's ‘ON’ period, when medications were most effective for them. Based on these guidelines, we categorized motor phenotypes into TD and NTD.

There are many different ways to potentially phenotype PwPD, such as by motor, cognitive, or autonomic manifestations, and any may also employ several classification subgroups, including an indeterminate grouping, though implementing some type of motor classification revolving around tremor (TD or NTD) is most common (Mestre et al. 2021). We chose to implement the most often utilized, dichotomous, tremor‐based phenotyping technique as an initial investigation point and to facilitate ease of potential clinical applicability. It is also established that, over time, motor phenotypes fluctuate within a patient (Mestre et al. 2021), and the differences in clinical presentation used to classify a phenotype decrease over time, whereby different phenotypes appear more similar as the disease goes on (Luo et al. 2019). However, we believe that by utilizing a repeated measures design, we were able to capture what an individual's phenotype currently was at each time point. This therefore gives an accurate description of how phenotype affects them compared to other genotypes at that moment in time. This may reflect how the classification of a given phenotype at a given point in time, or across points in time, is associated with changes in impairment.

We also included several extracted variables that were expected to have a potential effect on dysphagia and dysarthria self‐reports; these included age at each extracted time point, sex, H&Y stage, cognitive function measured via the Montreal Cognitive Assessment (MoCA) (Nasreddine et al. 2005), and medication dose as described by LEDD. We also extracted the site from which PPMI data was collected, which was included as a random effect during modelling (see Statistical Analysis). These variables were included as covariates during statistical analysis to control for their potential effects on the dependent variables but were not analysed separately to assess their contributions to any models.

2.4. Statistical Analysis

SPSS (version 28.0; IBM Corp.) was used for all statistical analyses. Continuous variables, including demographic or disease‐specific information such as age, time since diagnosis, and so forth, were expressed in mean and standard deviation (mean ± SD); categorical variables, including sex and motor phenotype, were expressed in frequency distributions (%); and the H&Y stage was expressed as median and interquartile range (IQR) for descriptive purposes. The UPDRS and SCOPA‐AUT question scores were expressed in both mean and standard deviation as well as median and IQR for descriptive purposes. Linear mixed models (LMMs) were used because of their robustness in handling unbalanced designs where there may be more participants in one genetic subgroup than another, where sample size attrition occurs at later time points, their ability to covary specified variables rather than performing a multiple linear regression, and potential missing data from participants that is missing at random (MAR) or missing completely at random (MCAR) using listwise rather than casewise deletion (Harel and McAllister 2019).

Statistical hypothesis testing was performed using a single omnibus repeated measures linear mixed‐effect model (LMM) for each dependent variable of interest, which included UPDRS 2.1, UPDRS 2.3, SCOPA AUT Q1, and SCOPA AUT Q3 scores through year 6, for a total of four LMMs. For each LMM, visit year was used as a repeated variable with a compound symmetry structure. Visit year, genotype, and phenotype were used as fixed effects to examine their contributions individually to the models, as well as to assess any interactions between these variables for longitudinally focused research questions. In total, this provided three 1‐way, three 2‐way, and one 3‐way interactions per model. Previously mentioned covariates of importance were included in each LMM and included age at each extracted time point, sex, H&Y stage, cognitive function measured via the Montreal Cognitive Assessment (MoCA) (Nasreddine et al. 2005), and medication dose as described by LEDD. For each LMM, research PPMI data collection sites were included as a random effect with random intercepts and an unstructured covariance structure.

All models are reported with test statistics and p values. Post‐hoc pairwise comparisons with Bonferroni corrections were applied for multiple comparisons across all models and reported with test statistics (F statistic), p values and confidence intervals where significant results were present, and Hedge's g was calculated to provide relevant effect sizes for additional interpretation of these findings with unequal group sizes. The α level for detecting a significant effect was set to 0.05 prior to Bonferroni corrections. When interpreting any significant results, mean differences between grouping levels for each dependent variable (i.e., UPDRS 2.1) were denoted using the Δ (delta) symbol as mean Δ. For longitudinal, repeated measures analyses, the visit year timepoints (Y) at which the dependent variables were measured are denoted as Y1 for year 1, Y2 for year 2, and so forth.

2.5. Preliminary Data Screening

During data extraction, several categories of genotypes were identified from the PPMI dataset. However, many of these genotypes occurred rarely (i.e., with only one participant). The most common genetic subtypes included in the PPMI dataset were LRRK2, GBA, and SNCA genotypes, which comprised 91% of the genetic participants that met our criteria. As such, LRRK2, GBA, and SNCA subtypes were carried forward as the genetic subsets of interest, and the remaining 9% of data (21 participants) comprising these subsets were excluded from analysis. To ensure there were no differences in the number of testing visits in one group compared to another, preliminary crosstabulations were performed on genotype by visit year, which found no associations (χ² = 31.01, p = 0.15). This suggested that each genotype group participated in an approximately equal number of visits relative to its group size. Additionally, during the data screening and visualization process, it was revealed that at year 6 (Y6) there were only two SNCA participants remaining in the study. Despite the ability of LMMs to handle unbalanced samples, we questioned the ability to draw meaning from a sample with only two participants in comparison to larger groups. As such, we examined and reported data only out to year 5 (Y5), rather than Y6 as planned.

3. Results

A total of 211 PwPD who were enrolled in the PPMI study had valid measures of dysphagia and dysarthria self‐reports from the UPDRS and SCOPA‐AUT at their initial baseline visit. The baseline group n of valid responses for each genotype was n = 128 for LRRK2, n = 68 for GBA, and n = 15 for SNCA, respectively. Each genotype group experienced attrition at each time point except during the time from the first baseline visit to the first yearly visit (Y1), where the number of LRRK2 and GBA participants increased, likely due to awaiting genetic testing results for classification, and one SNCA participant dropped out. The n of each group at each time point is presented in Figure 1. The mean age for the entire sample at the time of baseline was 61.94 (±10.16) years, and mean time since diagnosis was 2.95 (±2.44) years, suggesting that this sample of genotypic PD was relatively young and early post‐diagnosis. A full description of demographic and disease‐specific information by genotype can also be found in Table 1. At baseline, the entire sample self‐reported minimal dysarthria or dysphagia severity across all measures. UPDRS 2.1 displayed an average of 0.56 (±0.77) scale points and UPDRS 2.3 displayed an average of 0.27 (±0.63) scale points. A full description of baseline dysphagia and dysarthria by genetic subgroup can be found in Table 2. Overall model summaries of variables and interactions, as well as random effect variance, can be found in Tables 3, 4, 5, 6.

FIGURE 1.

Number of participants per visit year.

TABLE 1.

Demographic and disease‐specific information at baseline visit.

Demographic and disease specific information	Genotype
	LRRK2 (n = 128)	GBA (n = 68)	SNCA (n = 15)
Mean (±SD)
Age (in years)	63.52 (±8.80)	61.79 (±10.34)	51 (±12.34)
Disease duration (in years)	3.01 (±2.53)	2.72 (±2.24)	3.48 (±2.52)
LEDD	455.49 (±402.37)	443.49 (±416.14)	697.69 (±459.74)
MoCA	26.16 (±2.92)	26.69 (±2.20)	25.39 (±5.38)
Percentage (%)
Sex (Male/Female)	47/53	53/47	39/61
Motor phenotype (TD/PIGD)	41/59	48/52	33/67
Median (IQR)
Hoehn and Yahr	2(1)	2(1)	2(1)

Abbreviations: IQR, interquartile range; LEDD, Levadopa equivalent dosage; MoCA, Montreal Cognitive Assessment; PIGD, postural instability/gait dominant; TD, tremor dominant.

TABLE 2.

Dysarthria and dysphagia self‐report scores by genotype at baseline.

Self‐report measure	Genotype
	LRRK2 (n = 128)	GBA (n = 68)	SNCA (n = 15)
Mean (±SD)
UPDRS 2.1 (‘Speech’)	0.46 (±0.71)	0.72 (±0.87)	0.60 (±0.74)
UPDRS 2.3 (‘Eating and Swallowing’)	0.22 (±0.56)	0.39 (±0.75)	0.20 (±0.56)
SCOPA‐AUT Q1	0.33 (±0.67)	0.33 (±0.53)	0.20 (±0.41)
SCOPA‐AUT Q3	0.23 (±0.51)	0.33 (±0.56)	0.33 (±0.62)
Median (IQR)
UPDRS 2.1 (‘Speech’)	0 (1)	1 (1)	0 (1)
UPDRS 2.3 (‘Eating and Swallowing’)	0 (0)	0 (1)	0 (0)
SCOPA‐AUT Q1	0 (0)	0 (1)	0 (0)
SCOPA‐AUT Q3	0 (0)	0 (1)	0 (1)

TABLE 3.

Final model for LMM examining UPDRS 2.1.

Fixed effects
	F statistic (numerator Df, error Df)	p value
Intercept	149.36 (1, 67.33)	< 0.001
Visit year	8.30 (5, 1001.02)	< 0.001
Genotype	13.42 (2, 266.39)	< 0.001
Phenotype	2.39 (1, 1139.45)	0.12
Visit year X genotype	3.04 (11, 1001.98)	< 0.001
Visit year X phenotype	0.61 (5, 1010.95)	0.79
Genotype X phenotype	0.31 (2, 1148.86)	0.74
Visit year X genotype X phenotype	0.99 (5, 1016.99)	0.45

Random effects
	Variance	Standard error
PPMI site	0.04	0.03

TABLE 4.

Final model for LMM examining UPDRS 2.3.

Fixed effects
	F statistic (numerator Df, error Df)	p value
Intercept	48.03 (1, 1202)	< 0.001
Visit year	2.30 (5, 1202)	0.009
Genotype	4.23 (2, 1202)	0.02
Phenotype	0.13 (1, 1202)	0.72
Visit year X genotype	2.56 (11, 1202)	< 0.001
Visit year X phenotype	1.28 (5, 1202)	0.24
Genotype X phenotype	0.26 (2, 1202)	0.77
Visit year X genotype X Phenotype	0.74 (5, 1202)	0.69

Random effects
	Variance	Standard error
PPMI site	0.00	0.00

TABLE 5.

Final model for LMM examining SCOPA‐AUT Q1.

Fixed effects
	F statistic (numerator Df, error Df)	p value
Intercept	35.13 (1, 105.09)	< 0.001
Visit year	2.38 (5, 1013.06)	0.007
Genotype	1.15 (2, 238.99)	0.32
Phenotype	2.53 (1, 1116.67)	0.11
Visit year X genotype	1.71 (11, 1013.28)	0.036
Visit year X phenotype	0.55 (5, 1025.09)	0.84
Genotype X phenotype	0.53 (2, 1163.69)	0.59
Visit year X genotype X phenotype	0.91 (5, 1031.93)	0.53

Random effects
	Variance	Standard error
PPMI site	0.007	0.009

TABLE 6.

Final model for LMM examining SCOPA‐AUT Q3.

Fixed effects
	F statistic (numerator Df, error Df)	p value
Intercept	41.63 (1, 1195)	< 0.001
Visit year	1.68 (5, 1195)	0.07
Genotype	1.33 (2, 1195)	0.27
Phenotype	1.59 (1, 1195)	0.21
Visit year X genotype	0.97 (11, 1195)	0.49
Visit year X phenotype	0.89 (5, 1195)	0.54
Genotype X phenotype	0.14 (2, 1195)	0.87
Visit year X genotype X phenotype	0.72 (5, 1195)	0.72

Random effects
	Variance	Standard error
PPMI site	0.00	0.00

3.1. Question 1: Differences in Dysphagia and Dysarthria Self‐Reports Between Genotypes at Baseline

There were no significant effects of genotype at baseline for either UPDRS or SCOPA‐AUT questions (p > 0.05), suggesting all genotypes at their initial study visit self‐reported similar levels of impairment across dysarthria and dysphagia domains.

3.2. Question 2: Longitudinal Self‐Reports of Dysarthria and Dysphagia Between Genotypes

3.2.1. Dysarthria: UPDRS 2.1

Repeated measures LMMs displayed an overall effect of visit year for UPDRS 2.1 (F = 8.30, p < 0.001) with a significant interaction effect of genotype by visit year (F = 3.04, p < 0.001), suggesting that differences in genotype significantly contributed to our longitudinal model. As such, we report on the interaction between these two terms rather than visit year alone. Inspection of pairwise comparisons suggests significant differences between at least two genotypes in every visit year. Overall, LRRK2‐genotyped participants displayed significantly (p < 0.05) lower dysarthria severity self‐report scores than other genotypes at visit years Y2–Y5. GBA‐genotyped participants only reported significantly different UPDRS 2.1 scores than SNCA genotyped participants at Y5, where GBA reported significantly lower scores (Mean Δ = −1.04, p = 0.002, 95% CI = −1.75 to −0.33). Effect size calculations using Hedge's g = 1.04, suggesting a large effect of genotype difference between GBA and SNCA at Y5. Figure 2 illustrates changes in UPDRS 2.1 over time, by genotype.

FIGURE 2.

UPDRS 2.1 longitudinal scores by genotype.

3.2.2. Dysphagia: UPDRS 2.3

There was a significant overall effect of visit year on UPDRS 2.3 scores (F = 2.3, p = 0.009) but with a significant interaction effect with genotype (F = 2.56, p < 0.001) similar to UPDRS 2.1 results. Inspection of pairwise comparisons for the visit year and genotype interaction displayed SNCA genotyped participants reported significantly greater UPDRS 2.3 scores compared to LRRK2 (Mean Δ = 0.99, p = 0.002, 95% CI = 0.29 to 1.68) and GBA (Mean Δ = 1.11, p < 0.001, 95% CI = 0.41 to 1.82) in Y3 and LRRK2 (Mean Δ = 1.08, p < 0.001, 95% CI = 0.54 to 1.61) and GBA (Mean Δ = 1.19, p < 0.001, 95% CI = 0.62 to 1.77) in Y5. Effect size calculations using Hedge's g for differences at Y3 between SNCA and LRRK2 (g = 1.50) and GBA (g = 1.77) and Y5 (g = 2.38 and g = 1.34, respectively) suggest a large effect of genotype differences in Y3 and Y5. Figure 3 demonstrates the changes in UPDRS 2.3 over time, by genotype.

FIGURE 3.

UPDRS 2.3 longitudinal scores by genotype.

3.2.3. Dysphagia: SCOPA‐AUT Q1

There was a significant overall effect of visit year on SCOPA‐AUT Q1 scores (F = 2.38, p = 0.007), but with a significant interaction effect with genotype (F = 1.71, p = 0.036). Inspection of pairwise comparisons for the visit year and genotype interaction displayed differences between SNCA and LRRK2 and GBA genotyped participants only at Y5. At both time points, SNCA displayed significantly greater SCOPA‐AUT Q1 reports than LRRK2 (Mean Δ = 1.28, p = 0.006, 95% CI = 0.29 to 2.28) and GBA (Mean Δ = 1.32, p = 0.009, 95% CI = 0.25 to 2.40). Effect size calculations using Hedge's g for differences between SNCA and LRRK2 (g = 1.91) and GBA (g = 2.03) suggest large effects of phenotype on SCOPA‐AUT Q1 self‐reports. Figure 4 demonstrates the changes in SCOPA‐QUT Q1 over time.

FIGURE 4.

SCOPA‐AUT Q1 longitudinal scores by genotype.

3.2.4. Dysphagia: SCOPA‐AUT Q3

There were no significant overall effects of visit year (F = 1.68, p = 0.07) or interaction effects with genotype (F = 0.97, p = 0.49). Figure 5 demonstrates the changes in SCOPA‐AUT Q3 over time.

FIGURE 5.

SCOPA‐AUT Q3 longitudinal scores by genotype.

3.3. Question 3: Effects of Motor Phenotype on Longitudinal Dysphagia and Dysarthria Self‐Reports

There were no significant differences or interactions displayed by any of the omnibus LMMs for effects of motor phenotype on dysphagia or dysarthria self‐reports. As such, pairwise comparisons are not reported.

4. Discussion

The goals of this project were threefold: (1) examine the differences between the most common genotyped participants in the PPMI dataset in self‐reports of dysphagia and dysarthria, (2) investigate the change in severity over time of self‐reports of dysphagia and dysarthria in genotyped participants, and (3) assess how motor phenotypes within each genotype affect longitudinal dysphagia and dysarthria reports. Our findings highlight common, yet novel threads within each question. Namely, PwPD with different genotypes self‐report similar levels of dysarthria and dysphagia during their initial, baseline visits. However, these self‐reports do change significantly over time while controlling for age, sex, and disease‐centred factors, and genotype is a significant factor in how dysphagia and dysarthria may manifest in these self‐reports. Finally, when utilizing a dichotomous motor phenotype of either tremor or non‐tremor classification of TD or PIGD, there were no omnibus, significant effects of phenotype on how dysphagia or dysarthria self‐reports change over time in this sample. Our findings in the context of the current literature are explored below.

4.1. TD and NTD Phenotypes Effects on Genotypes and Self‐Reports of Dysarthria and Dysphagia

The lack of significant differences between phenotypes within and across genotypes in this study goes against our original hypothesis, as we hypothesized the PIGD or NTD phenotype would display significant differences, as well as interact with the genotypes included in this study. The reasons for a lack of effect of phenotype on these outcomes could be due to several factors. As an example, while we did not specifically examine changes in the distribution of phenotypes in each genotype over time, our baseline descriptive statistics highlight an unbalanced makeup of PIGD and TD categorizations in LRRK2 and SNCA genotypes.

This is different in clinical presentation from other samples that have examined idiopathic PD participants from the PPMI. As an example, Watts and Zhang (2022) report that at the baseline assessment for their sample, approximately 73% were classified as TD and 16% as PIGD, and at Y6, this had shifted to approximately 47% classified as TD and 39% as PIGD, respectively. Conversely, our sample of genotypic PD participants had a higher distribution of PIGD phenotypes at baseline and then throughout every single timepoint in every genotype. In fact, our current sample distribution of TD/PIGD phenotypes began much closer together, with PIGD becoming the predominant phenotype, in opposition to how early‐stage, drug‐naïve iPD started in Watts and Zhang's (2022) sample.

Consideration should also be given to the differences in motor and non‐motor burden that potentially exist between and across genotypes (discussed further below). As an example, though the evidence is mixed, a recent systematic review (Thijs and Dumican 2023) found a trend towards NTD phenotypes, such as PIGD, presenting with worse voice and swallowing outcomes. However, this was limited in scope to iPD. It is also possible that the questionnaires included in this particular study are not yet sensitive enough to detect changes in dysarthria and dysphagia between phenotypes within genetic subtypes of PwPD. As an example, Dumican and Watts (2020b) found significant differences between self‐reported speech and swallowing impairment in TD and NTD phenotypes of iPD utilizing in‐depth speech and swallowing questionnaires. Conversely, recent evidence has shown that UPDRS question 2.3 was significantly associated with the Sydney Swallow Questionnaire, an in‐depth dysphagia questionnaire, albeit in iPD (Dumican et al. 2024. However, as can be seen, the majority of this available research has been conducted in iPD and, therefore, makes it inappropriate to draw definitive conclusions about the effectiveness of phenotyping for identifying dysarthria or dysphagia manifestation in genotypes of PD, and substantially more research is required.

Another potential reason for a lack of significant findings in this sample may be that the dichotomous phenotypes implemented in this study did not effectively describe the motor phenotypes within these genotypes and therefore do not adequately capture speech or swallowing differences. It is documented that phenotypes fluctuate, and the phenotypes implemented here may not have adequately captured those fluctuations (Mestre et al. 2021). It may be beneficial in the future to explore additional phenotypes outside of motor‐ or tremor‐based types, such as cognitive‐ or autonomic‐based phenotypes (Mestre et al. 2021). Substantially more research is required to elucidate how similar or different phenotypes affect dysarthria or dysphagia and their underlying mechanisms.

Finally, our statistical choice was to include all fixed effects into an omnibus repeated measures design as a means to reduce the statistical tests run and therefore Type I error. It may be that a more focused design to explicitly examine phenotypes rather than as an addition to genotypes may elucidate more explicit and in‐depth findings beyond the scope of this paper. Despite these postulated differences, our findings here are in line with those of Watts and Zhang (2022), where no significant differences over time were seen in the evolution of dysarthria and dysphagia longitudinally in different motor phenotypes.

4.2. Genetic PD Subtypes and Dysarthria

To our knowledge, this is the first examination of how self‐reported dysphagia and dysarthria manifest longitudinally in a large sample of genetically subtyped PD. The findings of the current study suggest that LRRK2‐genotyped participants reported the least severe dysarthria and dysphagia out of the available genetic subtypes. These results present novel findings that add to our current knowledge base regarding how different genotypes of PD self‐report manifestations of dysarthria, in particular. In the landscape of the current literature, the most relevant context is in physiological voice measures. One of the only other studies to our knowledge that examined communication differences considering genotypes of PD is from Arora et al. (2018). They found significant differences in voicing parameters for LRRK2‐associated PD compared to iPD not associated with any genetic marker in various voice acoustic signal‐to‐noise metrics. In that particular study, parameters of noise in the acoustic voice signal of LRRK2‐associated participants were higher compared to participants with iPD.

The substantial differences in overarching goals and methodology between this study and Arora et al. (2018) make a direct comparison difficult. Nonetheless, both studies included LRRK2‐associated PwPD with similar MoCA, LEDD, age, time since diagnosis, and H&Y staging. Ultimately, the present study highlights that LRRK2 genotyped patients self‐report having relatively low levels of dysarthria impairment, while physiological findings suggest they present with greater acoustic impairment. The potential lack of agreement between self‐reported dysarthria in this study and physiological measures of voice is not unexpected, as previous studies in PwPD have shown disagreement between self‐report and acoustic measures (Kwan and Whitehill 2011). However, these findings do highlight the need for further investigations on how genotyped PD participants differ from iPD from a physiological perspective as well as self‐reports, both at baseline (such as at the time of PD diagnosis or referral) and longitudinally.

Our findings also add information about dysarthria reports to the broad presentation of these genotypes. As discussed above, most notably, LRRK2‐genotyped patients reported the least severe dysarthria compared to GBA or SNCA genotypes longitudinally, which confirms our original hypothesis. However, we also found that SNCA‐genotyped patients reported the highest level of dysarthria compared to LRRK2 or GBA, which was a part of our initial hypothesis. SNCA‐genotyped patients reported greater dysarthria at Y2‐Y5 compared to LRRK2 or GBA.

The underlying mechanism of significantly greater dysarthria reports in SNCA genotypes is unclear as, to our knowledge, there is no available data assessing speech or voice function in SNCA‐genotyped PwPD. However, current clinical findings highlight that SNCA‐genotyped patients may present with more aggressive forms of PD (Guadagnolo et al. 2021) and with greater non‐motor symptom burden (Trinh et al. 2018 Our findings in the context of these expectations therefore support a potential connection between non‐motor phenotypes and increased dysarthria. Though our omnibus mixed models for investigating phenotype were non‐significant, our data suggested that PIGD‐phenotyped SNCA patients tended to report greater dysarthria severity than other similarly phenotyped LRRK2 or GBA patients.

This provides insight into the level of burden that non‐motor symptoms, at least in primarily PIGD phenotypes, have on reported dysarthria but should be interpreted with caution given the lack of overall, omnibus effects. The reason for self‐reported dysarthria to trend higher in PIGD or other NTD phenotypes of PD is uncertain. Adding to this uncertainty is conflicting evidence on the inclusion of dysarthria as a motor or non‐motor symptom (Moustafa et al. 2016; Schapira et al. 2017). It is plausible to identify dysarthria and therefore speech and/or voice function as a motor symptom but one that is differentially affected (i.e., more severe) in non‐motor phenotypes such as PIGD. This theory is supported by evidence that non‐tremor phenotypes of PD exhibit lower striatal dopamine as well as substantially higher levels of pathological alpha‐synuclein in both motor and non‐motor circuits, potentially resulting in a higher motor burden than tremor‐dominant phenotypes (Huertas et al. 2017; Zuo et al. 2017). Evidence connecting this to the voice literature has shown NTD phenotypes present with physiologically increased aperiodicity and transglottal airflow during voicing (Burk and Watts 2019) as well as self‐reporting higher levels of speech and voice impairment compared to tremor‐dominant PwPD (Dumican and Watts 2020b. Therefore, it stands to reason that SNCA‐genotyped patients manifesting with NTD phenotypes would self‐report greater speech and/or voice impairment due to the potential for the significant presence and effect of alpha‐synuclein proteins throughout motor and non‐motor pathways in NTD phenotypes.

4.3. Genetic Subtypes and Dysphagia

To our knowledge, this is the first investigation into how self‐reported symptoms of dysphagia vary between genotypes of PD. Our findings suggest that, over time, SNCA genotyped patients tend to self‐report significantly more impaired levels of dysphagia compared to the other genotypes included in this specific study (LRRK2 and GBA genotypes, respectively). Potential mechanisms contributing to these eventual shifts in greater self‐reports of impairment are unclear. However, it may be attributed to similar physiological mechanisms outlined above relating to dysarthria. This is relevant when considering that major components of oropharyngeal swallowing require use of similar muscles and structures for speech and voice, such as laryngeal sensorimotor function, although their control mechanisms and end functions may be goal dependent (i.e., vocal fold adduction for voice vs. vocal fold adduction for cough; reflexive vs. volitional movements, etc.) (Ludlow 2015).

It is also possible that significant differences in dysphagia self‐reporting in SNCA genotyped PwPD may also be attributed to more localized pathology, rather than solely to neuropathology affecting downstream sensorimotor function. As an example, it has been documented that PwPD post‐mortem display significantly elevated levels of α‐synuclein aggregates in pharyngeal nerves compared to healthy controls (Mu et al. 2013), and SNCA genotypes of PD presented with significantly elevated α‐synuclein in biopsies of the enteric nervous system (Chung et al. 2019). This may provide preliminary support that genotypes of PD differentially express their pathological mechanisms, which may over time more severely affect speech, voice, and/or swallowing function in specific genotypes. However, this should be interpreted cautiously given that we did not include any biospecimen data to correlate with dysarthria or dysphagia self‐reports. Future studies may wish to consider exploring and including biospecimen and/or imaging data.

Adding further complexity to these postulations about this sample are the findings that these differences in self‐reports by patients between genotypes do not express themselves broadly until the third year of follow‐up. This suggests that there may not be an immediate effect of genotype on dysarthria or dysphagia self‐report, but rather an increase in impairment to a particular threshold that is reached several years into the disease to ultimately report increased impairment. This potential theory is supported in the current literature, as an example, where both voice and swallowing quality of life are shown to decrease as the disease progresses and show plateaus in the early stages of PD prior to showing significant increases in impairment being reported (Van Hooren et al. 2016).

Stemming from this, it is also possible that altered swallowing physiology is present in SNCA genotypes (as well as others) prior to their third year of follow‐up, as found in this study. The literature supports the theory that PwPD may be unreliable in reporting dysphagia symptoms using commonly available screening questions, where reports of no impairment do not align with findings of sequelae such as aspiration (Nienstedt et al. 2019). Conversely, additional evidence supports the theory that more in‐depth screenings or questionnaires may help predict dysphagia findings (Dumican and Watts 2020a, Given the conflicting evidence available, these results highlight the ongoing need to provide robust swallowing screening and assessment to PwPD to assess and manage dysphagia as soon as possible and potentially extend this to specific genotypes of PD as well. Substantially more research in more robust, controlled, prospective studies should be conducted to further investigate the manifestation and assessment of dysphagia in genotypic PD.

4.4. Other Factors Impacting Dysarthria and Dysphagia

Considering the dysarthria and dysphagia self‐report severity in SNCA genotypes in this study, the found differences to other genotypes may also be attributed to age, cognitive function, or socioeconomic status. The SNCA sample in our study had an average age of 51 and an average disease duration of almost 3.5 years, suggesting participants from the SNCA sample may have a mixture of early‐onset and late‐onset PD. A recent position statement by a taskforce on early‐onset PD has recommended that early‐onset be classified when the age of onset of PD is prior to the age of 50 (Mehanna et al. 2022). In relation to dysarthria, the current literature suggests that early‐onset PD may present with worse performance in certain subdomains of motor speech function compared to late‐onset PD (Rusz et al. 2021). Given that disease duration at baseline was similar in the SNCA participants compared to other LRRK2 and GBA genotypes, it appears unlikely that a longer disease course is affecting these reports. Pathological studies of early‐onset PD without genetic mutations (e.g., no SNCA) suggest PD disease duration in early‐onset PD tends to be longer and is coupled with slower progression (Ferguson et al. 2016). This may suggest that the genetic manifestation of SNCA‐type PD leads to more severe progression of dysarthria and dysphagia symptoms compared to other genotypes.

Other factors to consider in the self‐report of worse dysarthria and dysphagia severity in SNCA participants may be cognitive function and socioeconomic factors. SNCA participants in our sample started at baseline with worse MoCA scores compared to other genotypes. Although these scores were still relatively close, SNCA MoCA scores were notably just below the threshold for mild cognitive impairment. However, over time, an informal examination of our data suggests steep declines in the MoCA across the 5 years of data used in this study. LRRK2‐ and GBA‐genotyped participants remained relatively constant in MoCA scores, maintaining no lower than a 25 out to Y5. SNCA genotyped patients, however, start to experience a substantial decline in MoCA scores from baseline, with the final participants in our sample at Y5 with an average of 19.86 on the MoCA. It is reasonable to consider that this sharp decline in MoCA scores from Baseline‐Y5 of this study suggests a potential effect on self‐reporting ability. However, previous work has also found that in iPD participants, as MoCA scores get worse, the self‐report of voice and swallowing impairment increases (Dumican et al. 2024). This may support that cognitive decline accompanies a worsening in communication and swallowing function, and patients may still be able to identify these symptoms. However, instrumental assessments such as acoustic and aerodynamic measures of voice must be implemented to further understand this relationship.

Years of education may also be a contributing factor to self‐report symptoms. With respect to the PPMI, educational categories are split out into below 13 years of education (at maximum equivalent to a high school degree), from 13 to 23 years of education (ranging from at least some college through to some level of graduate degree), and then beyond 23 years, with the number of self‐reported years of education also available. In the total sample, an informal exploration of education suggested approximately 43 participants below 13 years of education, and three of these belonged to the SNCA group at Y5. Given the attrition experienced in the SNCA group in particular to Y5, it is possible that educational level may have played a role in participants' ability to self‐report symptomology accurately. However, education level was not a specific variable of interest in the particular study and, therefore, should be considered with caution. Future studies may wish to examine how geographic location, education, and cognitive function interact in how PwPD self‐report communication and swallowing difficulties.

Additional factors in the remaining sample of SNCA participants at Y5 also highlight the significant functional decline compared to other genotypes. Informal examination of clinical characteristics of this sample at Y5 showed substantially higher LEDD in SNCA participants, at almost 400 units greater than GBA and 500 units greater than LRRK2 participants, indicating much larger dosages of medication for managing PD symptoms. Though we entered both MoCA scores and LEDD units as covariates in our model, these substantial declines in cognitive and functional status provide additional context as to why SNCA genotyped patients may be experiencing greater communication and swallowing impairment.

5. Limitations

This study is not without limitations, and as such, caution should be used in the interpretation of the findings presented in this study. First, this study utilized a large, open‐sourced data set from the PPMI that utilized data from participants at multiple different sites over time. While taking every step to try and account for both demographic and disease‐specific confounders, it is still possible that factors such as geographic location or socioeconomic status may have influenced the participants in ways not readily apparent or within the scope of this study. Although we included PPMI site as a random effect in our model to account for variance across sites, there is likely still variability not captured (e.g., at the researcher or clinician level) that may have influenced data collection or patient responses. Data regarding the reliability of measures, including dysarthria and dysphagia, but also for PD‐centred data, such as, for example, judging tremor or gait, are not available and may affect aspects of phenotyping.

Second, these dysarthria and dysphagia reports were dependent on the self‐reports of PwPD, and the questions specifically targeting communication and swallowing were limited. We recommend that future studies employ specifically designed speech, voice, and swallowing instruments that may better capture impairment, as well as specific aspects of impairment that may affect quality of life. To account for the self‐report nature of these questions, we utilized many potential demographic and disease‐specific measures as covariates in our model to control potential confounding factors in these self‐reports, including medication, cognitive status, and disease severity. However, this data still relies on the self‐reporting and perceived severity of dysarthria and dysphagia symptoms, and as such, these findings cannot be extrapolated to infer instrumental findings or diagnoses of dysarthria or dysphagia. It may be that other genotypes in this study that self‐report less impairment do so due to impaired self‐perception of their deficits. This highlights the need for further studies to examine how instrumental assessments of speech, voice, and swallowing differ between genotypes of PD with greater methodological rigour. Another aspect of dysarthria and dysphagia that could not be accounted for was if these patients were receiving speech, voice, or swallowing‐related therapy. This clinical‐level piece of data was not available through the PPMI and therefore may introduce an additional confounding factor across and even within participants. Additionally, factors including socioeconomic status, cultural factors, or race may have impacted participants’ willingness to report symptom severity accurately or comfortably. While out of the scope of the current study, this should be considered a limitation, and we urge future studies to investigate and address potential systemic influences on patient symptom reporting.

Finally, this sample is also limited only to the genotypes that could be feasibly included in this study due to sample size. Therefore, the findings presented in this study can only be generalized to the genotypes included. A substantial limitation within this consideration is we did not take into account more specific genotypic mutations within each genotype. There are a myriad of specific point or multi‐point mutations in SNCA and LRRK2 genes that may alter the clinical manifestation of PD. While an important consideration for future studies to further improve individualistic dysarthria and dysphagia profiling, this approach was not feasible for the scope of the current study. This should be considered a limitation in the generalizability of our findings.

An important additional limitation stemming from this sample is the unbalanced nature of the group sizes throughout the analysis, as well as the attrition that occurred in the longitudinal retention of study participants. While out of our immediate control given the use of a secondary, open‐source dataset, as well as the use of statistical methods that are designed to handle varying group sample sizes, this is not a limitation to be overlooked and therefore requires cautious interpretation of the results presented. This is particularly the case for SNCA genotyped participants with a much smaller sample than the other genotypes and inherently limits generalizability for this subset specifically. More equal, robust, and prospective groups of varying genotypes are necessary to elucidate these findings further.

6. Conclusions

Self‐reported dysarthria and dysphagia measures that are widely used in the clinical assessment of PwPD display significant differences between genotypes of PD at different timepoints through the progression of the disease. LRRK2‐genotyped PwPD routinely self‐reported less severe dysarthria and dysphagia than other genotypes in this study, with SNCA‐genotyped PwPD self‐reporting the most significantly impaired dysarthria and dysphagia symptoms across all three genotypes examined. Baseline presentations of self‐reported dysarthria and dysphagia were not different between any genotypes, suggesting an inherent time‐based effect of self‐reporting symptoms. There were no effects of motor phenotype between or within genotypes in this sample. Substantially more investigations into physiological differences between genotypes for both dysarthria and dysphagia are warranted. Findings here support the importance of early screening to measure and detect patient‐perceived dysarthria and dysphagia for faster assessment, diagnosis, and treatment.

Conflicts of Interest

The authors declare no conflicts of interest.

Appendix A. Listing and Scoring of Dysarthria and Dysphagia Questions Included for Analysis

UPDRS Question 2.1 – Speech

Over the past week, have you had problems with your speech?

• 0: Normal: Not at all (no problems).

• 1: Slight: My speech is soft, slurred or uneven, but it does not cause others to ask me to repeat myself.

• 2: Mild: My speech causes people to ask me to occasionally repeat myself, but not every day.

• 3: Moderate: My speech is unclear enough that others ask me to repeat myself every day even though most of my speech is understood.

• 4: Severe: Most or all of my speech cannot be understood.

UPDRS Question 2.3 – Chewing and Swallowing

Over the past week, have you usually had problems swallowing pills or eating meals? Do you need your pills cut or crushed or your meals to be made soft, chopped, or blended to avoid choking?

• 0: Normal: No problems.

• 1: Slight: I am aware of slowness in my chewing or increased effort at swallowing, but I do not choke or need to have my food specially prepared.

• 2: Mild: I need to have my pills cut or my food specially prepared because of chewing or swallowing problems, but I have not choked over the past week.

• 3: Moderate: I choked at least once in the past week.

• 4: Severe: Because of chewing and swallowing problems, I need a feeding tube.

SCOPA‐AUT Question 1:

In the past month, have you had difficulty swallowing or have you choked?

 never sometimes regularly often

SCOPA‐AUT Question 3:

In the past month, has food ever become stuck in your throat?

 never sometimes regularly often

Appendix B. SPSS Code Used for All Linear Mixed Models

LMM Code for UPDRS 2.1

• MIXED NP2SPCH BY YEAR subgroupcode td_pigd_on WITH age_at_visit SEX NHY_ON

• LEDD moca

• /CRITERIA = DFMETHOD(SATTERTHWAITE) CIN(95) MXITER(100) MXSTEP(10) SCORING(1)

• SINGULAR(0.000000000001) HCONVERGE(0.00000001, RELATIVE) LCONVERGE(0, ABSOLUTE) PCONVERGE(0,

• ABSOLUTE)

• /FIXED = YEAR subgroupcode td_pigd_on YEAR*subgroupcode YEAR*td_pigd_on subgroupcode*td_pigd_on

• YEAR*subgroupcode*td_pigd_on | SSTYPE(3)

• /METHOD = REML

• /PRINT = DESCRIPTIVES G SOLUTION TESTCOV

• /RANDOM = INTERCEPT | SUBJECT(subgroupsite) COVTYPE(UN) SOLUTION

• /REPEATED = YEAR | SUBJECT(PATNO*subgroupsite) COVTYPE(CS)

• /EMMEANS = TABLES(OVERALL)

• /EMMEANS = TABLES(YEAR) COMPARE ADJ(BONFERRONI)

• /EMMEANS = TABLES(subgroupcode) COMPARE ADJ(BONFERRONI)

• /EMMEANS = TABLES(YEAR*td_pigd_on) COMPARE (td_pigd_on) ADJ(BONFERRONI)

• /EMMEANS = TABLES(YEAR*subgroupcode) COMPARE (subgroupcode) ADJ(BONFERRONI)

• /EMMEANS = TABLES(td_pigd_on*subgroupcode) COMPARE (subgroupcode) ADJ(BONFERRONI)

• /EMMEANS = TABLES(YEAR*td_pigd_on*subgroupcode) COMPARE (subgroupcode) ADJ(BONFERRONI).

LMM Code for UPDRS 2.3

• MIXED NP2SWAL BY YEAR subgroupcode td_pigd_on WITH age_at_visit SEX NHY_ON

• LEDD moca

• /CRITERIA = DFMETHOD(SATTERTHWAITE) CIN(95) MXITER(100) MXSTEP(10) SCORING(1)

• SINGULAR(0.000000000001) HCONVERGE(0.00000001, RELATIVE) LCONVERGE(0, ABSOLUTE) PCONVERGE(0,

• ABSOLUTE)

• /FIXED = YEAR subgroupcode td_pigd_on YEAR*subgroupcode YEAR*td_pigd_on subgroupcode*td_pigd_on

• YEAR*subgroupcode*td_pigd_on | SSTYPE(3)

• /METHOD = REML

• /PRINT = DESCRIPTIVES G SOLUTION TESTCOV

• /RANDOM = INTERCEPT | SUBJECT(subgroupsite) COVTYPE(UN) SOLUTION

• /REPEATED = YEAR | SUBJECT(PATNO*subgroupsite) COVTYPE(CS)

• /EMMEANS = TABLES(OVERALL)

• /EMMEANS = TABLES(YEAR) COMPARE ADJ(BONFERRONI)

• /EMMEANS = TABLES(subgroupcode) COMPARE ADJ(BONFERRONI)

• /EMMEANS = TABLES(YEAR*td_pigd_on) COMPARE (td_pigd_on) ADJ(BONFERRONI)

• /EMMEANS = TABLES(YEAR*subgroupcode) COMPARE (subgroupcode) ADJ(BONFERRONI)

• /EMMEANS = TABLES(td_pigd_on*subgroupcode) COMPARE (subgroupcode) ADJ(BONFERRONI)

• /EMMEANS = TABLES(YEAR*td_pigd_on*subgroupcode) COMPARE (subgroupcode) ADJ(BONFERRONI).

LMM Code for SCOPA‐AUT Q1

• MIXED SCAU1 BY YEAR subgroupcode td_pigd_on WITH age_at_visit SEX NHY_ON

• LEDD moca

• /CRITERIA = DFMETHOD(SATTERTHWAITE) CIN(95) MXITER(100) MXSTEP(10) SCORING(1)

• SINGULAR(0.000000000001) HCONVERGE(0.00000001, RELATIVE) LCONVERGE(0, ABSOLUTE) PCONVERGE(0,

• ABSOLUTE)

• /FIXED = YEAR subgroupcode td_pigd_on YEAR*subgroupcode YEAR*td_pigd_on subgroupcode*td_pigd_on

• YEAR*subgroupcode*td_pigd_on | SSTYPE(3)

• /METHOD = REML

• /PRINT = DESCRIPTIVES G SOLUTION TESTCOV

• /RANDOM = INTERCEPT | SUBJECT(subgroupsite) COVTYPE(UN) SOLUTION

• /REPEATED = YEAR | SUBJECT(PATNO*subgroupsite) COVTYPE(CS)

• /EMMEANS = TABLES(OVERALL)

• /EMMEANS = TABLES(YEAR) COMPARE ADJ(BONFERRONI)

• /EMMEANS = TABLES(subgroupcode) COMPARE ADJ(BONFERRONI)

• /EMMEANS = TABLES(YEAR*td_pigd_on) COMPARE (td_pigd_on) ADJ(BONFERRONI)

• /EMMEANS = TABLES(YEAR*subgroupcode) COMPARE (subgroupcode) ADJ(BONFERRONI)

• /EMMEANS = TABLES(td_pigd_on*subgroupcode) COMPARE (subgroupcode) ADJ(BONFERRONI)

• /EMMEANS = TABLES(YEAR*td_pigd_on*subgroupcode) COMPARE (subgroupcode) ADJ(BONFERRONI).

LMM Code for SCOPA‐AUT Q3

• MIXED SCAU3 BY YEAR subgroupcode td_pigd_on WITH age_at_visit SEX NHY_ON

• LEDD moca

• /CRITERIA = DFMETHOD(SATTERTHWAITE) CIN(95) MXITER(100) MXSTEP(10) SCORING(1)

• SINGULAR(0.000000000001) HCONVERGE(0.00000001, RELATIVE) LCONVERGE(0, ABSOLUTE) PCONVERGE(0,

• ABSOLUTE)

• /FIXED = YEAR subgroupcode td_pigd_on YEAR*subgroupcode YEAR*td_pigd_on subgroupcode*td_pigd_on

• YEAR*subgroupcode*td_pigd_on | SSTYPE(3)

• /METHOD = REML

• /PRINT = DESCRIPTIVES G SOLUTION TESTCOV

• /RANDOM = INTERCEPT | SUBJECT(subgroupsite) COVTYPE(UN) SOLUTION

• /REPEATED = YEAR | SUBJECT(PATNO*subgroupsite) COVTYPE(CS)

• /EMMEANS = TABLES(OVERALL)

• /EMMEANS = TABLES(YEAR) COMPARE ADJ(BONFERRONI)

• /EMMEANS = TABLES(subgroupcode) COMPARE ADJ(BONFERRONI)

• /EMMEANS = TABLES(YEAR*td_pigd_on) COMPARE (td_pigd_on) ADJ(BONFERRONI)

• /EMMEANS = TABLES(YEAR*subgroupcode) COMPARE (subgroupcode) ADJ(BONFERRONI)

• /EMMEANS = TABLES(td_pigd_on*subgroupcode) COMPARE (subgroupcode) ADJ(BONFERRONI)

• /EMMEANS = TABLES(YEAR*td_pigd_on*subgroupcode) COMPARE (subgroupcode) ADJ(BONFERRONI).

Dumican, M. , Reyers, T. , Malczewski, A. , and and Thijs, Z. (2025) The Effect of Genotype on Self‐Reported Dysarthria and Dysphagia in Parkinson's Disease: A Parkinson's Progression Marker Initiative Study. International Journal of Language & Communication Disorders, 60, e70124. 10.1111/1460-6984.70124

Data Availability Statement

Data used in the preparation of this article were obtained from the Parkinson's Progression Markers Initiative (PPMI) database (https://www.ppmi‐info.org/access‐data‐specimens/download‐data), RRID:SCR_006431. For up‐to‐date information on the study, visit http://www.ppmi‐info.org.