Grey and white matter correlates of dysphagia in progressive supranuclear palsy

Abstract

Background:

Dysphagia is a common symptom of progressive supranuclear palsy (PSP) often leading to aspiration pneumonia and death.

Objectives:

To examine how impairments of the oral and pharyngeal phases of the swallow and airway incursion during liquid swallows relate to grey and white matter integrity.

Methods:

Thirty-eight PSP participants underwent videofluorographic swallowing assessment and structural and diffusion tensor head MRI. Penalized linear regression models assessed relationships between swallowing metrics and regional grey matter volumes and white matter fractional anisotropy and mean diffusivity.

Results:

Oral phase impairments were associated with reduced superior parietal volumes, and abnormal diffusivity in parietal and sensorimotor white matter, posterior limb of the internal capsule, and superior longitudinal fasciculus. Pharyngeal phase impairments were associated with disruption to medial frontal lobe, corticospinal tract and cerebral peduncle. No regions were predictive of airway incursion.

Conclusions:

Differential patterns of neuroanatomical impairment corresponded to oral and pharyngeal phase swallowing impairments.

Introduction

Progressive supranuclear palsy (PSP) is a neurodegenerative tauopathy^{1, 2} characterized by oculomotor dysfunction, postural instability and falls, akinesia, and cognitive dysfunction³. Bulbar symptoms often present early, can be severe^{4, 5} and are an independent predictor of survival^{4, 6, 7}, presumably because physiologic swallowing deficits lead to increased risk of aspiration pneumonia, the most common cause of death in PSP⁸. Previous reports of dysphagia in PSP have consistently documented predominant oral relative to pharyngeal phase impairments. Commonly observed oral phase impairments include back and forth rocking, or festinating movements of the tongue to propel the bolus toward the pharynx^{5, 9–11}, reduced tongue base retraction^{9, 11} and delayed onset of the pharyngeal swallow. Severity of pharyngeal phase impairments and degree of airway incursion are associated with PSP disease severity^11–15. Primary brainstem and supranuclear control centers support deglutition, with the oral phase under voluntary control with greater influence of supranuclear control centers¹⁶ and the pharyngeal and esophageal phases driven by a medullary central pattern generator^{17, 18}. Fiber tracts including corticobulbar tract and corpus callosum subserve the functionality of this network^19–21. However, little is known about the neuroanatomical underpinnings of dysphagia in PSP.

We aimed to examine gray matter volumes and white matter tract integrity in relation to oral and pharyngeal phase dysphagia and airway incursion. It was predicted that oral phase impairment would be associated with high-level cortical swallowing network, while pharyngeal phase impairments would associate with brainstem pathways. Airway incursion is not distinct from oral and pharyngeal phase impairments but is instead of unique interest because of its potential role in predicting survival. Because airway incursion may result from many physiologic impairments, no a priori predictions were made about neuroimaging correlates.

Methods

Participants

Thirty-eight participants with PSP were consecutively recruited by the Neurodegenerative Research Group (NRG) from the Department of Neurology, Mayo Clinic Rochester, MN, between 12/12/2017 and 3/3/2020. To be included patients had to meet criteria for possible or probable PSP according to the Movement Disorders Society clinical criteria for PSP³. Exclusion criteria included concurrent illnesses/conditions that could account for symptoms or confound MRI, and if MRI was contraindicated. All participants underwent a neurological and dysphagia examination¹¹ and MRI. The dysphagia evaluation was done as a component of a research protocol and subjects were enrolled regardless of whether they had any swallowing complaints. Twenty-four participants reported dysphagia, all but one indicating dysphagia for liquids was worse than solids. The study was approved by the Mayo Clinic IRB and all patients provided written consent for participation in the study.

Dysphagia examination

The videofluorographic swallowing assessment utilized the Modified Barium Swallow Impairment Profile (MBSImP) protocol²². Consistencies included thin and nectar thick barium (liquid), E-Z paste (puree), (Varibar) and Lorna Doone cookie with paste coating (solid). The volume and order of recorded swallows have been previously published¹¹. The MBSImP yields ratings for 17 individual components of swallowing function, as well as total sum scores for Oral and Pharyngeal components. The total oral and pharyngeal sum scores (maximum=17 and 31 points, respectively) were the primary outcomes for this study. The Penetration-Aspiration Scale (PAS)²³ was scored for thin liquid swallows. For the purpose of participant description, the Functional Oral Intake Scale (FOIS)²⁴ was scored to characterize the degree to which oral diets were modified or nonoral nutrition was required.

Neuroimaging analyses

All participants underwent a standardized MRI protocol on a 3T Siemens Prisma scanner, that included a T1-weighted 3D magnetization prepared rapid gradient echo (MPRAGE) sequence²⁵, and a single-shot echo-planar diffusion sequence with 114 gradient directions²⁶. Region-level data was generated for each modality. The Mayo Clinic Adult Lifespan Template (MCALT; https://www.nitrc.org/projects/mcalt) and Deep Brain Stimulation Intrinsic Template²⁷ atlas were used to assess grey matter volume in subcortical and cortical regions-of-interest (ROIs). Midbrain and cerebellar dentate volume were measured using in-house developed atlases²⁸. Total intracranial volume was also measured. Each DTI scan was preprocessed as previously described²⁶. Diffusion tensors were estimated using nonlinear least squares fitting and used to calculate Fractional Anisotropy (FA) and Mean Diffusivity (MD) images in dipy²⁹. The Johns Hopkins University (JHU) “Eve” atlas was used to generate ROI-level data. Voxels with MD > 0.002 mm2/s were masked out to exclude cerebrospinal fluid.

Statistical analysis

Spearman correlations were assessed between the primary swallowing outcomes and cognition (Montreal Cognitive Assessment Battery, MoCA) and an adjusted-PSP Rating Scale calculated by removing scores from the two questions related to dysphagia. We used penalized linear regression (PLM) using the glmnet package (v. 3.0–2) for the R-environment for statistical computing (v 3.6.2) for model selection predicting either oral or pharyngeal total sum of MBSImP and the PAS based on MRI and DTI measures as predictors³⁰. We included MoCA as a covariate to account for confounds of cognitive impairment. The predictors were standardized to have mean=0 and SD=1 so that the coefficients were all on the same scale. Each PLM model was fit with an elastic net penalty with tuning parameter alpha=0.80. Our final models were chosen using a penalty that minimized leave-one-out cross-validation error. This penalized linear regression approach avoids problems with overfitting that arise with model selection³¹. It can be assumed to provide better out-of-sample prediction and estimates that are more generalizable.

Results

Demographic features, neurological data and dysphagia scores are shown in Table 1. MoCA correlated with oral total score (spearman=−0.46, p=0.004), but not pharyngeal total score or PAS. No correlations were identified between adjusted-PSP Rating Scale and the primary swallowing outcomes.

Table 1:

Participant characteristics

	All (n=38)	PSP-RS (n=17)	PSP-P (n=7)	PSP-SL (n=4)	PSP-CBS (n=3)	PSP-PI (n=2)	PSP-PGF (n=2)	PSP-F (n=2)	PSP-OM (n=1)
No. Female	17 (45%)	7 (41%)	1 (14%)	2 (50%)	1 (33%)	2 (100%)	1 (50%)	2 (100%)	1 (100%)
No. APOE carrier	13 (39%)	5 (33%)	3 (50%)	0 (0%)	3 (100%)	0 (0%)	1 (50%)	0 (0%)	1 (100%)
Education, yr	16 (14, 16)	16 (14, 16)	16 (16, 17)	16 (15, 16)	12 (12, 14)	14 (13, 15)	13 (12, 14)	19 (19, 19)	NA
Age at scan, yr	70 (66, 75)	71 (67, 74)	69 (67, 74)	70 (66, 75)	63 (61, 66)	74 (73, 76)	76 (75, 77)	67 (61, 73)	65 (65, 65)
Disease duration, yr	5.0 (3.1, 7.8)	4.7 (3.0, 5.6)	8.0 (5.3, 9.2)	9.7 (8.1, 10.7)	2.7 (2.2, 3.1)	2.9 (2.9, 3.0)	3.6 (3.4, 3.8)	6.9 (6.1, 7.7)	7.0 (7.0, 7.0)
Age onset, yr	64 (59, 69)	65 (61, 67)	63 (60, 67)	62 (58, 66)	61 (58, 64)	72 (70, 73)	72 (71, 74)	60 (54, 66)	58 (58, 58)
Montreal Cognitive Assessment Battery	22 (19, 26)	22 (16, 24)	26 (22, 28)	19 (18, 22)	26 (24, 26)	24 (23, 25)	22 (22, 23)	5 (5, 5)	24 (24, 24)
PSP Rating Scale-adjusted	35 (27, 51)	42 (29, 52)	40 (28, 46)	44 (37, 54)	21 (20, 26)	28 (28, 28)	34 (31, 36)	50 (38, 63)	24 (24, 24)
MDS-UPDRS III	50 (30, 64)	56 (37, 64)	59 (43, 62)	54 (42, 71)	26 (26, 28)	32 (26, 38)	48 (36, 59)	49 (32, 66)	6 (6, 6)
PSP Saccadic Impairment Scale	3 (2, 4)	3 (3, 4)	2 (1, 3)	2 (2, 3)	1 (1, 2)	2 (1, 3)	0 (0, 1)	2 (2, 3)	4 (4, 4)
Test for Upper Limb Apraxia	10 (8, 11)	10 (8, 10)	11 (11, 12)	5 (5, 8)	9 (6, 10)	8 (6, 9)	11 (10, 12)	6 (3, 8)	12 (12, 12)
Frontal Assessment Battery	13 (10, 14)	13 (7, 14)	14 (12, 16)	10 (9, 10)	14 (13, 16)	10 (10, 11)	14 (13, 14)	8 (4, 12)	12 (12, 12)
Oral total sum (Max possible = 17)	8 (5, 9)	8 (7, 9)	5 (4, 6)	12 (10, 12)	8 (5, 8)	6 (4, 8)	5 (4, 6)	11 (10, 12)	6 (6, 6)
Pharyngeal total sum (Max possible = 31)	3 (1, 5)	2 (1, 4)	4 (2, 6)	6 (4, 7)	1 (0, 3)	3 (2, 4)	4 (2, 6)	2 (1, 2)	4 (4, 4)
Penetration-Aspiration Scale	2 (1, 2)	2 (1, 3)	1 (1, 2)	2 (2, 2)	2 (2, 5)	2 (1, 2)	4 (3, 4)	2 (1, 2)	1 (1, 1)
Functional Oral Intake Scale	7 (7, 7)	7 (6, 7)	7 (7, 7)	7 (6, 7)	7 (7, 7)	7 (7, 7)	6 (6, 6)	6 (6, 6)	7 (7, 7)

Data shown as median (inter-quartile range) or N (%). PSP-RS = PSP with Richardson’s syndrome; PSP-P = PSP with predominant parkinsonism; PSP-SL = PSP with predominant speech/language disorder; PSP-CBS =PSP with predominant corticobasal syndrome; PSP-PI = PSP with predominant postural instability; PSP-PGF = PSP with progressive gait freezing; PSP-F = PSP with predominant frontal presentation; PSP-OM = PSP with predominant ocular motor impairment; MDS-UPDRS III = Movement Disorder Society Sponsored Revision of the Unified Parkinson’s Disease Rating Scale. All participants exhibited at least mild oral phase impairments (oral total sum≥2) and all but 5 exhibited at least minimal pharyngeal phase impairments (pharyngeal total sum≥1). Nineteen (50%) of the participants experienced laryngeal penetration of liquids without tracheal aspiration (PAS=2–5) while only 3 experienced tracheal aspiration (PAS=6–8). No participants were tube dependent for any of their nutrition (FOIS≤3) and only 9 (24%) had any restriction on the consistency or preparation of their oral intake (FOIS=4–6).

Greater impairment in oral phase was associated with relatively smaller superior parietal volume, lower FA in superior parietal WM, postcentral WM, and posterior limb of the internal capsule, and higher MD in precentral WM, angular gyrus WM, and superior longitudinal fasciculus (Figure 1 and Supplementary Figure 1). Greater impairment in oral phase was also associated with relative sparing of superior cerebellar peduncle (Figure 1 and Supplementary Figure 1).

Figure 1: Relationship between neuroimaging measures and swallowing metrics.

Three-dimensional renders illustrate the penalized elastic-net model estimates for the regions found to be predictive of oral or pharyngeal total sum scores. Rows one and two illustrate the results of the MRI volume models with superior parietal lobe predictive of oral total sum (model estimate=−0.242) and superior medial frontal lobe predictive of pharyngeal total score (−0.183). The DTI results are shown in rows three, four and five. Yellow-red color scales depict regions where more abnormal diffusivity was associated with more abnormal swallowing function, while green-blue color scales depict regions where relatively more spared diffusivity was associated with more abnormal swallowing function. For the FA oral optimal model, lower FA in superior parietal white matter (WM) (−1.45), postcentral WM (−0.10) and posterior limb of the internal capsule (−0.03) was associated with worse oral score, while greater FA in superior cerebellar peduncle (0.92) was associated with worse oral score. For the MD oral optimal model, greater MD in the precentral WM (0.23), angular WM (0.56) and superior longitudinal fasciculus (0.53) was associated with worse oral score, while lower MD in superior cerebellar peduncle (−0.85) was associated with worse oral score. For the MD pharyngeal optimal model, higher MD in the corticospinal tract (0.85) and cerebral peduncle (2.69) was associated with worse pharyngeal score, while lower MD in the posterior thalamic radiation (−1.54), inferior fronto-occipital fasciculus (−1.49) and the corpus callosum (−0.36 for splenium and −0.02 for genu) was associated with worse pharyngeal score. Renders were created with MRIcroGL.

Greater impairment in pharyngeal phase was associated with relatively smaller superior medial frontal lobe (SMFL) volume and higher MD in corticospinal tract and cerebral peduncle (Figure 1 and Supplementary Figure 1). Greater impairment in pharyngeal phase was also associated with relative sparing of posterior thalamic radiation, genu and splenium of the corpus callosum, and inferior fronto-occipital fasciculus.

No significant associations were observed between PAS score and volume, FA or MD.

Discussion

It was predicted that disruptions to the oral phase of the swallow, posited to be under volitional control, would be associated with neuroimaging evidence implicating brain structures involved in the high-level swallowing network³². Indeed, oral phase impairments were associated with abnormalities in the parietal and sensorimotor cortex, as well as with degeneration of posterior limb of the internal capsule and WM association tracts of the parietal and frontal lobes, consistent with this network^{32, 33}. The current findings support the association between these cortical motor control centers and pathways with the volitional components of oral phase swallowing, as well as the vulnerability of this network among patient with PSP. Oral phase impairments were also correlated with cognitive impairment, highlighting cognitive contributions to the oral phase of the swallow. We corrected our neuroimaging analyses to account for this potential confound.

For pharyngeal phase impairments, the predicted pattern of association with disruption to the corticospinal tract and cerebral peduncle was observed despite the relatively low frequency and severity of pharyngeal phase impairments. Not predicted was association with smaller volume of the SMFL, which is not explained by current models of neurologic control of swallowing. Even models that acknowledge volitional modulation to patterned pharyngeal responses do not localize that function to the SMFL. Should this finding be replicated, particularly across populations, such models will have to be updated to account for the contribution of the frontal lobe to pharyngeal phase function.

The models for both oral and pharyngeal phase identified regions which were inversely related to swallowing impairments (i.e. more intact diffusivity associated with worse swallow function); the superior cerebellar peduncle for oral phase and supratentorial regions for pharyngeal phase. These somewhat counterintuitive results may reflect heterogeneity in the cohort and suggest that oral phase impairments are likely in patients with cortical abnormalities but relative sparing of infratentorial regions, while pharyngeal phase impairments are likely in patients with greater infratentorial and less severe supratentorial involvement. Indeed, we have previously shown that different clinical variants of PSP have different patterns of involvement of supratentorial and infratentorial regions^{26, 34}. We did observe heterogeneity in swallow function across PSP variants but, unfortunately, we had too few patients with each variant to statistically analyze the contribution of variant. Future studies with larger cohorts will be needed to explore the neuroimaging correlates of swallowing function across PSP variants to help clarify the role of regional heterogeneity.

No significant neuroimaging associations were observed for airway incursion. One explanation is that the frequency and severity of airway incursion was insufficient to reveal the relevant associations. Another explanation is that airway incursion is not related to distinct neuroanatomical disruption but instead may associate with any and all, and therefore no, specific ROIs. This is consistent with what is known about the diversity of physiologic impairments in the oral and pharyngeal phases that may contribute to airway incursion³⁵. That no single pattern of neuroanatomical degeneration predicts airway incursion reinforces the need for careful clinical dysphagia assessment by a qualified speech-language pathologist.

The current study has several limitations. The measures of swallowing function were judged perceptually based on videofluorography, and temporal measures may reveal different associations. Some of the cortical ROIs were relatively large limiting our ability to pinpoint specific areas related to swallowing and we were unable to assess some of the smaller brainstem nuclei and corticobulbar tracts. Issues such as crossing fibers may also have influenced the FA measures. We did not assess radial and axial diffusivity since they are redundant and noise, such as crossing fibers, can lead to unstable estimates. Furthermore, the biological explanation for abnormalities in these metrics is uncertain^{36, 37}.

Conclusions

Dysphagia is a well-established symptom of PSP, with survival implications. The current findings provide evidence for distinct neuroimaging patterns associated with oral versus pharyngeal phase swallowing impairments in PSP. Understanding the pathophysiology of swallowing in PSP is an important first step to allow the development of neuroimaging biomarkers that can potentially track abnormalities in swallowing and predict functional decline and ultimately mortality and may inform the development of treatments for PSP.