R E S E A R C H A R T I C L E

Linking Vestibular Function and Subcortical Gray

Matter Volume Changes in a Longitudinal Study

ofAging Adults

Dominic Padova,

a,* J. Tilak Ratnanather,b Qian-Li Xue,c,d Susan M. Resnick,e and Yuri Agrawala

a Department of Otolaryngology – Head and Neck Surgery, Johns Hopkins University School of Medicine, Baltimore, MD, USA

b Center for Imaging Science and Institute for Computational Medicine, Department of Biomedical Engineering, Johns Hopkins University, Baltimore, MD, USA

c Department of Medicine Division of Geriatric Medicine and Gerontology, School of Medicine, Johns Hopkins University, Baltimore, Maryland, USA

d Department of Epidemiology, Johns Hopkins Bloomberg School of Public Health, Baltimore, Maryland, USA

e Laboratory of Behavioral Neuroscience, National Institute on Aging, National Institutes of Health, Baltimore, MD, USA

ABSTRACT

Emerging evidence suggests a relationship between impairments of the vestibular (inner ear balance) system and alterations in the

function and the structure of the central nervous system (CNS) in older adults. However, it is unclear whether age-related vestibular

loss is associated with volume loss in brain regions known to receive vestibular input. To address this gap, we investigated the asso-

ciation between vestibular function and the volumes of four structures that process vestibular information (the hippocampus, ento-

rhinal cortex, thalamus, and basal ganglia) in a longitudinal study of 97 healthy, older participants from the Baltimore Longitudinal

Study of Aging. Vestibular testing included cervical vestibular-evoked myogenic potentials (cVEMP) to measure saccular function,

ocular VEMP (oVEMP) to measure utricular function, and video head impulse tests to measure the horizontal semicircular canal ves-

tibulo-ocular re ex (VOR). Participants in the sample had vestibular and brain MRI data for a total of one (18.6%), two (49.5%), and

three (32.0%) visits. Linear mixed-effects regression was used to model regional volume over time as a function of vestibular physi-

ological function, correcting for age, sex, intracranial volume, and intersubject random variation in the baseline levels and rates of

change of volume over time. We found that poorer saccular function, characterized by lower cVEMP amplitude, is associated with

reduced bilateral volumes of the basal ganglia and thalamus at each time point, demonstrated by a 0.0714 cm3 ± 0.0344 (unad-

justed p = 0.038; 95% CI: 0.00397–0.139) lower bilateral-mean volume of the basal ganglia and a 0.0440 cm3 ± 0.0221 (unadjusted

p = 0.046; 95% CI: 0.000727–0.0873) lower bilateral-mean volume of the thalamus for each 1-unit lower cVEMP amplitude. We also

found a relationship between a lower mean VOR gain and lower left hippocampal volume (β = 0.121, unadjusted p = 0.018, 95% CI:

0.0212–0.222). There were no signi cant associations between volume and oVEMP. These  ndings provide insight into the speci c

brain structures that undergo atrophy in the context of age-related loss of peripheral vestibular function.

Keywords: Vestibular, cVEMP, MRI, Volume, Sub-cortical, Longitudinal

Correspondence: Dominic Padova, Department of Otolaryngology – Head and Neck Surgery, Johns Hopkins University School of Medicine, Baltimore, MD, USA,

Email: dpadova1@jhu.edu

Received: November 2, 2020

Accepted: August 24, 2021

DOI: 10.52294/6727e860-95c1-445c-a47a-177d9e699d46

Padova etal. This is an open access article distributed under the terms of the Creative Commons Attribution-NonCommercial-NoDerivs 4.0 IGO License, which permits the copy

and redistribution of the material in any medium or format provided the original work and author are properly credited. In any reproduction of this article there should not be any

suggestion that APERTURE NEURO or this article endorse any speci c organization or products. The use of the APERTURE NEURO logo is not permitted. This notice should be

preserved along with the article’s original URL. Open access logo and text by PLoS, under the Creative Commons Attribution-Share Alike 4.0 Unported license.

: 2021, Volume 1 - 1 - CC-BY: © Padova et al.

: 2021, Volume 1 - 1 -

INTRODUCTION

The vestibular (inner ear balance) system consists of semi-

circular canals, which detect rotational movements of the

head, and otolith organs, which detect linear head move-

ments and the orientation of the head with respect to

gravity [1]. The vestibular sensory system is known to play

a critical role in maintaining stable balance and gaze con-

trol. Additionally, a growing body of literature indicates a

preeminent contribution of the vestibular system to visuo-

spatial cognitive ability [2, 1–5]. Vestibular function declines

with normal aging [6–8], and vestibular dysfunction is as-

sociated with cognitive declines in older adults [2, 9, 10],

including declines in spatial cognitive function [11, 12].

Functional neuroimaging and vestibular stimulation

studies have demonstrated that vestibular information

is transmitted to multiple neural regions, including the

hippocampus, thalamus, basal ganglia, entorhinal cor-

tex, cerebellum, insula, parietal regions, and somatosen-

sory regions [13–25]. In fact, several clinical studies have

shown that lack of vestibular input leads to functional and

structural changes in the brain, either directly or indirectly.

: 2021, Volume 1 - 2 - CC-BY: © Padova et al.

R E S E A R C H A R T I C L E

: 2021, Volume 1 - 2 -

A pioneering study of subjects with bilateral vestibulop-

athy (BVP) by Brandt etal. [20] demonstrated that lack of

vestibular input leads to bilateral hippocampal atrophy

(16.91% reduction relative to controls) that correlated with

impaired spatial memory and navigation assessed by the

virtual Morris water navigation task (vMWT). Hufner etal.

[26] reported reductions in gray matter volume in the cer-

ebellum due to the schwannoma removal, supramarginal

gyrus on the same side as the lesion, postcentral and su-

perior temporal gyri, and motion-sensitive visual area MT/

V5 in pa tientswith complete unilateral vestibular deaffer-

entation (UVD) compared to healthy controls. Helmchen

etal. [27] demonstrated increases in gray matter volume

in the insula, retroinsular vestibular cortex, and superior

temporal gyrus correlated with improving clinical vestib-

ular assessments in patients recovering from vestibular

neuritis. zu Eulenburg et al. [28] showed that patients

with an acute unilateral vestibular de cit stemming from

vestibular neuritis recovery exhibited gray matter volume

reduction in the left posterior hippocampus, the right su-

perior temporal gyrus, and the right superior frontal gyrus

regardless of the side of vestibular impairment. Göttlich

etal. [29] used voxel-based morphometry (VBM) to  nd

reductions in gray matter volume in the CA3 region of

the hippocampus bilaterally with increasing vestibulop-

athy-related disability in patients with incomplete BVP

compared to healthy age- and gender-matched controls.

Kremmyda etal. [30] found that vestibular loss due to in-

complete BVP leads to gray matter volume reductions

bilaterally in the middle hippocampus and posterior par-

ahippocampus using VBM, to de cits in objective and

subjective evaluations of spatial memory and navigation

performance using the vMWT and questionnaires, and

to increased spatial anxiety. In a cross-sectional VBM

study of patients with persistent postural perceptual diz-

ziness, Wurthmann etal. [31] demonstrated reductions in

gray matter volumes in the left superior temporal gyrus,

left motion-sensitive visual area MT/V5 and bilateral

middle temporal gyrus, cerebellum (bilaterally), left-sided

posterior hippocampus, right precentral gyrus, left anterior

cingulate cortex, left side of left caudate nucleus, and left

dorsolateral prefrontal cortex.

However, little is known about which structures are affect-

ed by subclinical, age-related vestibular loss. Furthermore,

little is known about the speci c vestibular end organs (i.e.

semicircular canal and/or otolith organs) that contribute

to those changes. Bridging these knowledge gaps, novel

cross-sectional studies in healthy, older individuals found

signi cant associations between age-related vestibular

loss and gross gray matter loss of the hippocampus [38,

39] and the entorhinal cortex [39], as well as shape changes

in the hippocampus, amygdala, caudate nucleus, putamen,

thalamus, entorhinal cortex, and entorhinal-transentorhinal

cortical complex [39].

In this study, we examined the relationship between

age-related vestibular function and regional volume of

the hippocampus, entorhinal cortex, thalamus, and basal

ganglia, four structures known to process vestibular in-

formation. Linear mixed-effects regression was used to

evaluate the association between saccular, utricular, and

horizontal semicircular canal vestibulo-ocular re ex (VOR)

functions and MRI-based volumes of 97 healthy, older

participants aged 60 years and older from the Baltimore

Longitudinal Study of Aging (BLSA) over a 5-year fol-

low-up period. To our knowledge, this study is one of the

 rst to investigate the associations over time between

age-related vestibular function and volumes of brain re-

gions that receive vestibular input.

MATERIALS AND METHODS

Data

Data on 97 older adult participants (aged ≥ 60 years)

who had at least one vestibular physiological test and

structural MRI scan on the same visit between 2013 and

2017 were selected from the BLSA [40–42]. All partici-

pants gave written informed consent, and none had any

history of psychiatric disorders or were diagnosed with

a vestibular, ophthalmological, or neurodegenerative

disease. Hearing loss was measured and was included

as a confounding variable in the supplemental analysis.

Vestibular physiologic testing

Vestibular function testing was based on the cervical

vestibular-evoked myogenic potential (cVEMP), ocular

VEMP (oVEMP), and video head impulse test (vHIT), which

measure saccular function, utricular function, and the

horizontal semicircular canal VOR, respectively. cVEMP

and oVEMP were recorded using a commercial electro-

myographic system (software version 14.1, Carefusion

Synergy, Dublin, OH) [32, 33]. Electromyograms (EMGs)

were recorded by disposable, pregelled Ag/AgCl elec-

trodes placed with 40-inch safety lead wires from GN

Otometrics (Schaumburg, IL). cVEMP and oVEMP signals

were ampli ed ×2500 and band-pass  ltered for the 20–

2000 Hz and 3–500 Hz frequency intervals, respectively.

Cervical vestibular-evoked myogenic potential

cVEMPs are short-latency EMGs of the relaxation re-

sponse of the sternocleidomastoid muscles induced by

sound stimuli in the ear or by head-tapping vibrations.

They measure saccular (and inferior vestibular nerve) func-

tion. The saccule is a vestibular end organ that transduces

linear acceleration, detects the orientation of the head

with respect to gravity, and plays a role in spatial cogni-

tion [32]. Testing followed an established protocol [33–35].

Participants sat on a chair inclined to 30° above the hor-

izontal plane, and quali ed examiners placed EMG elec-

trodes on the sternocleidomastoid and sternoclavicular

junction bilaterally and a ground electrode on the manu-

brium sterni. Sound stimuli of 500 Hz and 125 dB were

administered in bursts of 100 stimuli monoaurally through

: 2021, Volume 1 - 3 - CC-BY: © Padova et al.

R E S E A R C H A R T I C L E

: 2021, Volume 1 - 3 -

vestibular hypofunction [36, 44, 45]. The mean VOR gain

from the left and right sides was used in this study.

Structural MRI acquisition and processing

T1-weighted volumetric MRI scans were acquired in

the sagittal plane using a 3T Philips Achieva scanner at

the National Institute on Aging Clinical Research Unit.

Sequences included a T1 volumetric scan magnetization

prepared rapid acquisition with gradient echo (MPRAGE;

repetition time (TR) = 6.5 ms, echo time (TE) = 3.1 ms,  ip

angle = 8°, image matrix = 256 × 256, 170 slices, voxel size =

1.0 × 1.0 × 1.2 mm slice thickness, FOV = 256 × 240 mm,

sagittal acquisition). A semiautomated quality control pro-

tocol was employed to automatically identify and manu-

ally exclude scans that have segmentation errors de ned

as outliers of each region-speci c sample distribution.

Anatomical labels were obtained using Multi-atlas region

Segmentation using Ensembles of deformable registra-

tion algorithms and parameters [46]. Global and regional

brain volumes were calculated from the binary segmen-

tation output by MUSE by counting the total number of

voxels in the binary label image and multiplying that voxel

count by the spatial size of a voxel in cm3. We corrected

for intracranial volume (ICV) individually estimated at age

70 using the residual volume approach described by Jack

etal. [47]: for each region and each scan, residual volumes

were calculated as the difference, in cm3, of the measured

regional volume from the expected regional volume, given

the ICV for the individual. The basal ganglia volume was

de ned to be the sum of the volumes of the lenticular nu-

cleus, globus pallidus, and the caudate nucleus. Figure1

headphones. Recorded cVEMP amplitudes were correct-

ed for spontaneous background EMG activity collected

10 ms prior to the onset of the sound stimulus. An absent

response was de ned according to previously published

amplitude and latency thresholds [32, 33]. In case of an

absent recording, the response assessment was repeated

for con rmation. For present responses, the higher cVEMP

from the left and right sides was used in this analysis.

Ocular vestibular-evoked myogenic potential

oVEMPs are short-latency EMGs of the excitation re-

sponse of the inferior oblique muscles of the eye elicited

by vibration stimulation of the skull. They measure utric-

ular (and superior vestibular nerve) function [32]. The utri-

cle is the otolith organ that transduces linear acceleration

and detects the orientation of the head with respect to

gravity. Testing followed an established protocol [33–35].

Participants sat on a chair inclined to 30°, and quali ed

examiners placed a noninverting electrode ~3 mm below

the eye centered below the pupil, an inverting electrode

2 cm below the noninverting electrode, and a ground

electrode on the manubrium sterni. To ensure that the

signals recorded from both eyes are symmetric before

stimulation, participants were asked to perform multiple

20° vertical saccades. New electrodes were applied in

place of the old ones if the signal difference exhibited

>25% asymmetry. Participants were asked to maintain

an upgaze of 20° during oVEMP testing and recording.

Head taps were applied to the midline of the face at the

hairline and ~30% of the distance between the inion and

nasion using a re ex hammer (Aesculap model ACO12C,

Center Valley, PA). An absent response was de ned ac-

cording to previously published amplitude and latency

thresholds [32, 33]. In case of an absent recording, the

response assessment was repeated for con rmation. For

present responses, the higher oVEMP from the left and

right sides was used in this analysis.

Video head impulse test

The horizontal VOR was assessed using the vHIT [35–37].

To determine VOR gain, the vHIT was performed using

the EyeSeeCam system (Interacoustics, Eden Prarie, MN)

in the same plane as the right and left horizontal semi-

circular canals [37, 43]. The participant’s head was tilted

downward 30° below the horizontal plane to correctly

position the horizontal canals in the plane of stimulation.

Participants were asked to maintain their gaze on a wall

target ~1.5 m away. A quali ed examiner rotated the

participant’s head 5° –10° quickly (~150°–250° per sec-

ond) parallel to the ground toward the right and left at

least 10 times in both directions, chosen randomly for

unpredictability. The EyeSeeCam system quanti ed eye

and head velocity, and a corresponding VOR gain was

calculated as the unitless ratio of the eye velocity to the

head velocity. A normal eye and head velocity should be

equal, yielding a VOR gain equal to 1.0. A VOR gain <0.8

accompanied by re xation saccades suggests peripheral

Fig. 1. Simpli ed 3D illustration of three subcortical and one cortical regions of

the vestibular network in the left hemisphere of the JHU-MNI-SS brain [48]. The

medial view of the left side of the pial surface of the JHU-MNI-SS template [48]

generated by FreeSurfer and mapped to native space [49] is shown. The three

subcortical (thalamus, hippocampus, basal ganglia, comprised of the putamen,

caudate nucleus, and globus pallidus) and one cortical (entorhinal cortex) struc-

tures were obtained from the JHU-MNI-SS labels and triangulated. Vestibular

information from the semicircular canals (SCC), otoliths, and nuclei is relayed

through the thalamus to the hippocampus, entorhinal cortex, and basal ganglia.

The red arrow points toward the ventral lateral nucleus, a putative sub eld of the

thalamus that receives vestibular input [39]. CAWorks (www.cis.jhu.edu/software/

caworks) was used for visualization.

: 2021, Volume 1 - 4 - CC-BY: © Padova et al.

R E S E A R C H A R T I C L E

: 2021, Volume 1 - 4 -

where agei,j of subject i at visit j is centered at age 70,

sexi is a binary indicator variable for the sex of subject

i, and ICVi denotes baseline ICV, comprised of bilateral

cerebral volumes, cerebellum, brainstem, and cerebro-

spinal  uid, individually estimated at 70 years of age.

Age, sex, and ICV were treated as  xed effects. We

assumed linear dependence of regional volume with re-

spect to age. We assumed uai, u0i are zero-mean Gaussian

distributed with unknown variances and covariances and,

respectively, represent the intrasubject random effect,

which captures correlations between measurements

within subject i over time, and the between-subject het-

erogeneity in terms of individual deviation in regional

volume from its sample mean at age 70. The covariance

structure of random effects was taken to be unstructured.

We assumed measurement noise εi,j is independently

and identically distributed zero-mean Gaussian with un-

known common variance. The unknown  xed-effects

{βv, βo, βc, βa, βs, βi, μ} and random-effects covariance ma-

trix parameters were estimated via maximum likelihood.

All effects were considered signi cant at the p < 0.05 level.

A Benjamini-Hochberg procedure was used to control the

false discovery rate (FDR) of the comparisons made in this

study [50]. FDR q-values indicate the expected proportion

of rejected null hypotheses that are false. We considered

an FDR threshold of 0.05 and also 0.10, given that these

were initial analyses testing speci c hypotheses based

on prior work. These statistics were performed using the

xtmixed function in Stata 15 (College Station, TX).

RESULTS

Baseline and longitudinal characteristics

Table 1 shows the baseline characteristics for the study

sample from the BLSA. Eighteen participants had one

vestibular function on regional brain volume. The model

included a by-subject random intercept and a random

slope on age to account for intersubject heterogeneity

in the baseline level of and rate of change in regional

brain volume over time.

The null hypothesis in Eq. (1) predicts regional volume

voli,j, for participant i,i = 1,…,N for observation j,j = 1,…,ni.

The alternate hypotheses predict volume using mean

VOR gain termed VORi,j in Eq. (2), oVEMPi,j in Eq. (3), and

cVEMPi,j in Eq. (4) as continuous independent variables,

provides a visualization of the relative locations of the

thalamus, hippocampus, basal ganglia, and entorhinal

cortex in a three-dimensional hemibrain.

Mixed-effects modeling

Linear mixed-effects regression was used to model the

evolution of regional brain volume over time while al-

lowing a time-independent cross-sectional effect of

observation with both vestibular and MRI data available,

48 participants had two observations, and 31 participants

had three observations. In this sample, there was only a

single case of participant dropout, due to death, after two

visits. At baseline, this participant had low cVEMP, missing

oVEMP and VOR, and an average hippocampal volume.

Association between vestibular function

andregional brain volumes based

in mixed-effects models

Table 2 reports the association between vestibular func-

tion and regional brain volume. We found signi cant

relationships in which each 1-unit lower cVEMP ampli-

tude was associated with a 0.0714 cm3 ± 0.0344 (unad-

justed p = 0.038, 95% con dence interval (CI): 0.00397–

0.139) lower bilateral-mean volume of the basal ganglia

and a 0.0440 cm3 ± 0.0221 (unadjusted p= 0.046, 95%

CI: 0.000727–0.0873) lower bilateral-mean volume of

the thalamus. This means every 1-unit decrease in cVEMP

amplitude is associated with an average reduction of

~0.4% in the basal ganglia and ~0.31% in the thala-

mus. We examined the left and right hemispheres sep-

arately and found signi cant relationships for the left

thalamus (β = 0.0232, unadjusted p = 0.043, 95% CI:

0.000684–0.0457) and for both the left (β = 0.0376, un-

adjusted p = 0.048, 95% CI: 0.000385–0.0747) and right

(β =0.0359, unadjusted p = 0.045, 95% CI: 0.000727–

0.071) basal ganglia. The effect was borderline for the

right thalamus (β = 0.0231, unadjusted p = 0.065, 95%

CI: −0.00148 to 0.0477). We also found a signi cant rela-

tionship between a lower mean VOR gain and lower left

hippocampal volume (β = 0.121, unadjusted p = 0.018,

95% CI: 0.0212–0.222). In other words, every 1-unit de-

crease in mean VOR gain is associated with an average

H0vo:

i, j

ai agu

)

i, j

a0i

exs

ICβ++V

=++ε

)

(

µi, j (1)

=β+β+β++

()

µεu+

Hv:ol ++β

()

uVORexsageICV

i, jv i, jaai i, jsiii0ii, j(2)

=β++

(

β+β++

()

µεu+

Hv:ol oVEMPexsuag

)

eICV

i, jaoi, jai i, jii 0ii i, j(3)

=β++

(

β+β:++

()

µεu+

Hvol cVEMPexsuag

)

eICV

i, jc ai, jai

+β

si, jii 0ii i, j(4)

: 2021, Volume 1 - 5 - CC-BY: © Padova et al.

R E S E A R C H A R T I C L E

: 2021, Volume 1 - 5 -

with a 0.0778 cm3 ± 0.00834 (q << 0.0001 FDR correct-

ed, 95% CI: (−0.0941, −0.0614)), or −0.55% per year, re-

duction in bilateral-mean volume and a 0.0392 cm3 ±

0.0041 (q << 0.0001 FDR corrected, 95% CI: (−0.0472,

−0.0311)) reduction in left-side volume (−0.27%/year).

For the left hippocampus VOR model, every 1-year

increase in age was associated with a 0.0239 cm3 ±

0.00271 (q << 0.0001 FDR corrected, 95% CI: (−0.0292,

−0.0186)) reduction in volume (−0.33%/year). Sex pre-

dicted entorhinal cortex volume over time in each

model (q << 0.0001 FDR corrected) and right thalamus

volumes in the VOR model (β = 0.248, unadjusted p =

0.045, 95% CI: (0.00539, 0.49)). Despite failing to reject

the null hypothesis, sex showed strong trends for the

bilateral-mean (β= 0.485, unadjusted p = 0.052) and left

(β = 0.23, unadjusted p = 0.073) thalamus VOR models

and was not a predictor of volume trajectories for the

hippocampus, basal ganglia, or the thalamus cVEMP or

oVEMP models.

Because age is included as a covariate, the volume

growth trajectory related to vestibular function rep-

resents effects above and beyond those attributable to

normal age-related gray matter volume loss. This means,

in addition to the age-related volume reduction rate of

0.19% per year in the basal ganglia and 0.55% per year in

the thalamus, every 1-unit decrease in cVEMP amplitude

is associated with an average reduction of ~0.4% in the

basal ganglia and ~0.31% in the thalamus. In other words,

the cVEMP effects are 2.05 times larger in magnitude and

0.57 times smaller in magnitude than the age-related

volume reductions in the basal ganglia and thalamus, re-

spectively. The VOR effects are 5.08 times larger in mag-

nitude than the age-related volume reduction rate in the

left hippocampus.

DISCUSSION

These  ndings provide some of the  rst evidence for as-

sociations between lower cVEMP amplitude, indicative

of poorer saccular function, and signi cantly reduced

volumes of the thalamus and basal ganglia and lower

VOR gain (poorer canal function) and reduced left hip-

pocampal volumes over time in healthy, older adults

aged 60 years and older. These results are compatible

with clinical studies in patients with vestibular loss that

demonstrated that loss of peripheral vestibular inputs is

Table 1. Baseline characteristics of the study sample (n = 97). Regional volumes

are given in cm3. Observation count denotes the number of participants with at

least one visit where both vestibular and MRI data were available. SD, standard

deviation.

Characteristic Overall

N97

Age (years), mean (SD) 76 (8.39)

Sex (% female) 58.76

Region, mean (SD)

Intracranial volume 1391.39 (147.32)

Hippocampus 7.23 (0.81)

Thalamus 14.2 (1.41)

Basal ganglia 17.9 (1.99)

Entorhinal cortex 4.29 (0.605)

Observation count, n (%)

1 18 (18.6)

2 48 (49.5)

3 31 (32.0)

reduction of 0.121 cm3, or ~1.68%, in the left hippo-

campus. We found no signi cant relationships between

cVEMP and volume of the hippocampus or the ento-

rhinal cortex, no signi cant associations between VOR

and brain volume aside from the left hippocampus, and

no signi cant associations between oVEMP and region-

al volume. After controlling the FDR at the threshold

of 0.10, lower cVEMP amplitude was associated with

lower bilateral-mean volume of the thalamus (q = 0.092

FDR corrected) and with lower bilateral-mean volume

of the basal ganglia (q = 0.092 FDR corrected). In ad-

ditional analyses, we added hearing – represented by

the four-frequency pure tone average from the better

ear – to the regression models. The addition of hearing

to the models reduced the sample size from 97 to 84

participants and resulted in the signi cant associations

previously observed becoming marginally signi cant,

although with similar effect sizes (see Supplementary

Table S1).

The “aging” effect was signi cant for all models (q <<

0.0001 FDR corrected), except for the left (β = −0.0169,

unadjusted p = 0.104), right (β = −0.0167, unadjusted p=

0.064), and bilateral-mean (β = −0.0348, unadjusted p=

0.064) basal ganglia cVEMP models, which showed strong

trends of volume reductions. For the thalamus cVEMP

model, every 1-year increase in age was associated

Table 2. Vestibular predictors of regional volume (cm3) under the alternate hypothesis with mean ± standard error (unadjusted p-value). *p < 0.05.

Vestibular Variable Hippocampus Entorhinal Cortex Basal Ganglia Thalamus

Mean VOR gain 0.168 ± 0.0988

(p = 0.089)

0.0562 ± 0.169

(p = 0.740)

0.213 ± 0.183

(p = 0.244)

0.159 ± 0.129

(p = 0.217)

Best oVEMP amplitude (µV) −0.00161 ± 0.00228

(p = 0.479)

0.00116 ± 0.00337

(p = 0.730)

−0.00247 ± 0.00393

(p = 0.530)

−0.00223 ± 0.00286

(p = 0.436)

Best corrected cVEMP

amplitude

0.00119 ± 0.0174

(p = 0.945)

−0.0114 ± 0.0291

(p = 0.693)

0.0714 ± 0.0344

(p = 0.038*)

0.0440 ± 0.0221

(p = 0.046*)

: 2021, Volume 1 - 6 - CC-BY: © Padova et al.

R E S E A R C H A R T I C L E

: 2021, Volume 1 - 6 -

sample after two visits due to death. Additionally, there

was no difference in the proportion of participants with

normal versus low cVEMP amplitude (≤25th percentile)

at baseline who had intermittently missing follow-up

observations (and in all cases all available data were

used in analyses).

We note limitations of this study. The relationship

between vestibular function and brainstem and cere-

bellar structures were not studied, as these structures

have been more challenging to parcellate due to their

complex anatomy, and robust measures to analyze the

brainstem and cerebellum in all BLSA participants are

currently being developed. Additionally, we also did

not investigate the potential mediation or modi cation

of the effects of age-related vestibular loss by interven-

ing brain structures such as the cerebellum, brainstem,

or hypothalamus. Moreover, the generalizability of our

results is limited by the age range studied and the ten-

dency of the BLSA participants to have higher levels of

education and socioeconomic status than the broader

adult population. Whether these  ndings can also be

detected in younger adults with reduced vestibular func-

tion is unclear.

Future work will be needed to further clarify the rela-

tionships between vestibular function and the structure

of regions of the limbic system, temporoparietal junc-

tion, and frontal cortex – all of which receive vestibular

input – and their temporal sequence of effects relative

to each other. Additionally, subsequent studies will

need to investigate the direction of causal in uence

between vestibular loss and regional size and shape

changes, while correcting for potential confounding

factors such as hearing or vision loss. Structural equa-

tion modeling with longitudinal data in which vestib-

ular measurements precede structural measurements

can help tease out the direct or indirect relationships

between vestibular function and brain structures.

Cortical thickness of cortical structures, such as the

entorhinal cortex, insula, and prefrontal cortex, can

provide a sensitive measure of cortical integrity com-

plementary to shape changes. Changepoint analysis of

longitudinal data can identify nonlinearities in the tra-

jectories of structural change. By temporally ordering

the set of changepoints for each structure measure,

the sequence of changes in the vestibular network

can be revealed. Future studies will be needed to test

whether there are modulating effects on downstream

structures via intervening brain structures such as the

brainstem, hypothalamus, and cerebellum.

CONCLUSIONS

This study examined the association between reduced

vestibular function and regional brain volume over time

in aging adults. To our knowledge, this study is one of the

 rst to demonstrate signi cant relationships between

associated with gray matter reductions in the hippocam-

pus [20, 28–31] and basal ganglia [31]. cVEMP measures

the physiological function of the saccule, the vestibular

end organ that detects linear acceleration and gravita-

tional cues, and preserves spatial orientation [1, 32]. VOR

gain measures the physiological function of the hori-

zontal semicircular canal, the vestibular end organ that

detects angular acceleration in the transverse plane [1].

Prior work has shown that abnormal VOR gain [11] and

lower cVEMP amplitude are signi cantly associated with

poorer spatial cognitive performance in healthy, older

adults [1, 11, 12]. Importantly, this work expands the un-

derstanding of the role of the peripheral vestibular sys-

tem in anatomical alterations.

We note that the effect sizes of the relationship be-

tween vestibular measures and changes in gray matter

volumes over time are small. These small, though signif-

icant, effect sizes may re ect the indirect links between

the peripheral vestibular system and the thalamus and

basal ganglia. Vestibular inputs are known to traverse

through brain structures such as the cerebellum, brain-

stem, and hypothalamus, where additional afferent in-

puts are integrated (e.g. visual, auditory, autonomic),

thereby modulating and attenuating the direct impact

of vestibular loss on central structural changes [51, 52].

Whether these changes in gray matter volumes repre-

sent loss of neurons remains unclear in the absence of

additional physiological or tissue studies. As such, the

extent to which these effect sizes re ect underlying neu-

ronal loss is unclear. Notably, the aging effects described

are consistent with the age-related cerebral volume re-

duction rates of ~0.5% per year found in previous longi-

tudinal studies in normal older controls [53, 54].

A previous cross-sectional study in healthy, older

adults demonstrated that saccular sensory loss, de ned

as lower cVEMP amplitude, was associated with signif-

icant gross volume loss of the hippocampus [38]. We

did not observe the same relationship between cVEMP

amplitude and hippocampal volume in this longitudi-

nal analysis. A possible explanation for this discrepancy

between the cross-sectional and longitudinal  ndings

is that the hippocampal volume reduction may have

largely occurred prior to the start of this longitudinal

period, such that the relationship was signi cant in the

baseline cross-sectional analysis but not in the longitu-

dinal analysis. In support of this explanation, one-way

analysis of covariance (ANCOVA) controlling for age,

sex, and ICV suggests the baseline bilateral-mean hip-

pocampal volume in participants with impaired cVEMP

at baseline is 0.34 cm3 smaller compared to those

withunimpaired cVEMP at baseline (p = 0.014, 95% CI:

(−0.608, −0.0717)). Another possible explanation for the

discrepancy between our cross-sectional and longitudi-

nal  ndings is a systematic difference between the two

cohorts, due to nonrandom, biased loss to longitudinal

follow-up. However, this is unlikely, given that there was

only a single participant who dropped out of this study

R E S E A R C H A R T I C L E

: 2021, Volume 1 - 7 -

16. Leong ATL, Gu Y, Chan YS, Zheng H, Dong CM, Chan RW, etal. Optogenetic

fMRI interrogation of brain-wide central vestibular pathways. Proceedings

of the National Academy of Sciences USA 2019;116(20):10122–29.

17. Stiles L, Smith PF. The vestibular–basal ganglia connection: balancing

motor control. Brain Research 2015;1597:180–88.

18. Fasold O, von Brevern M, Kuhberg M, Ploner CJ, Villringer A, Lempert T,

etal. Human vestibular cortex as identi ed with caloric stimulation in func-

tional magnetic resonance imaging. Neuroimage 2002;17(3):1384–93.

19. Dieterich M, Bense S, Lutz S, Drzezga A, Stephan T, Bartenstein P, et al.

Dominance for vestibular cortical function in the non-dominant hemisphere.

Cerebral Cortex 2003;13(9):994–1007.

20. Brandt T, Schautzer F, Hamilton DA, Brüning R, Markowitsch HJ, Kalla R,

et al. Vestibular loss causes hippocampal atrophy and impaired spatial

memory in humans. Brain: A Journal of Neurology 2005;128(11):2732–41.

21. Potegal M, Copack P, de Jong J, Krauthamer G, Gilman S. Vestibular input

to the caudate nucleus. Experimental Neurology 1971;32(3):448–65.

22. Suzuki M, Kitano H, Ito R, Kitanishi T, Yazawa Y, Ogawa T, etal. Cortical and

subcortical vestibular response to caloric stimulation detected by function-

al magnetic resonance imaging. Brain Research Cognitive Brain Research

2001;12(3):441–49.

23. Rancz EA, Moya J, Drawitsch F, Brichta AM, Canals S, Margrie TW. Widespread

vestibular activation of the rodent cortex. The Journal of Neuroscience

2015;35(15):5926–34.

24. Vitte E, Derosier C, Caritu Y, Berthoz A, Hasboun D, Soulié D. Activation of

the hippocampal formation by vestibular stimulation: a functional magnetic

resonance imaging study. Experimental Brain Research 1996;112(3):523–26.

25. Lopez C, Blanke O, Mast FW. The human vestibular cortex revealed by coor-

dinate-based activation likelihood estimation meta-analysis. Neuroscience

2012;212:159–79.

26. Hufner K, Stephan T, Hamilton D, Kalla R, Glasauer S, Strupp M, etal. Gray-

matter atrophy after chronic complete unilateral vestibular deafferentation.

Annals of the New York Academy of Sciences 2009;1164(1):383–85.

27. Helmchen C, Klinkenstein J, Machner B, Rambold H, Mohr C, Sander T.

Structural changes in the human brain following vestibular neuritis indicate

central vestibular compensation. Annals of the New York Academy of Sciences

2009;1164(1):104–15.

28. zu Eulenburg P, Stoeter P, Dieterich M. Voxel‐based morphometry de-

picts central compensation after vestibular neuritis. Annals of Neurology

2010;68(2):241–49.

29. Göttlich M, Jandl NM, Sprenger A, Wojak JF, Münte TF, Krämer UM, etal.

Hippocampal gray matter volume in bilateral vestibular failure. Human Brain

Mapping 2016;37(5):1998–2006.

30. Kremmyda O, Hüfner K, Flanagin VL, Hamilton DA, Linn J, Strupp M, etal.

Beyond dizziness: virtual navigation, spatial anxiety and hippocampal volume

in bilateral vestibulopathy. Frontiers in Human Neuroscience 2016;10:139.

31. Wurthmann S, Naegel S, Steinberg BS, Theysohn N, Diener HC,

Kleinschnitz C, etal. Cerebral gray matter changes in persistent postural

perceptual dizziness. Journal of Psychosomatic Research 2017;103:95–101.

32. Li C, Zuniga MG, Nguyen KD, Carey JP, Agrawal Y. How to interpret latencies

of cervical and ocular vestibular-evoked myogenic potentials: our experi-

ence in  fty-three participants. Clinical Otolaryngology 2014;39(5):297–301.

33. Nguyen KD, Welgampola MS, Carey JP, Nguyen KD, Welgampola MS,

Carey JP. Test-retest reliability and age-related characteristics of the oc-

ular and cervical vestibular evoked myogenic potential tests. Otology &

Neurotology 2010;31(5):793–802.

34. Li C, Layman AJ, Carey JP, Agrawal Y. Epidemiology of vestibular evoked

myogenic potentials: data from the Baltimore Longitudinal Study of Aging.

Clinical Neurophysiology 2015;126(11):2207–15.

35. Harun A, Oh ES, Bigelow RT, Studenski S, Agrawal Y. Vestibular impairment

in dementia. Otology & Neurotology 2016;37(8):1137–42.

36. Agrawal Y, Davalos-Bichara M, Zuniga MG, Carey JP. Head impulse test

abnormalities and in uence on gait speed and falls in older individuals.

Otology & Neurotology 2013;34(9):1729–35.

37. Agrawal Y, Schubert MC, Migliaccio AA, Zee DS, Schneider E, Lehnen N,

et al. Evaluation of quantitative head impulse testing using search coils

versus video-oculography in older individuals. Otology & Neurotology

2014;35(2):283–88.

38. Kamil RJ, Jacob A, Ratnanather JT, Resnick SM, Agrawal Y. Vestibular func-

tion and hippocampal volume in the baltimore longitudinal study of aging

(BLSA). Otology & Neurotology 2018;39(6):765–71.

39. Jacob A, Tward DJ, Resnick S, Smith PF, Lopez C, Rebello E, etal. Vestibular

function and cortical and sub-cortical alterations in an aging population.

Heliyon 2020;6(8):e04728.

40. Shock NW, Gerontology Research Center. Normal human aging: the

Baltimore longitudinal study of aging. Baltimore, MD: U.S. Dept. of Health

and Human Services, Public Health Service, National Institutes of Health,

vestibular loss – saccular and semicircular canal sensory

loss in particular – and gray matter volume loss of the

thalamus, basal ganglia, and left hippocampus, three

vestibular subcortical structures that receive peripher-

al vestibular input. Future work will need to determine

the timing and sequence of the relationships between

vestibular function and neuromorphological alterations.

ACKNOWLEDGMENTS

This work was supported in part by the National Institute

on Aging [grant number R01 AG057667], National

Institute on Deafness and Other Communication

Disorders [grant number R03 DC015583], and National

Institutes of Health [grant number P41 EB015909] and by

the Intramural Research Program, National Institute on

Aging, National Institutes of Health.

CONFLICTS OF INTEREST

The authors report no con icts of interest.

REFERENCES

1. Smith PF. The growing evidence for the importance of the otoliths in spatial

memory. Frontiers in Neural Circuits 2019;13:66.

2. Bigelow RT, Agrawal Y. Vestibular involvement in cognition: visuospatial

ability, attention, executive function, and memory. Journal of Vestibular

Research 2015;25(2):73–89.

3. Cullen KE. The neural encoding of self-generated and externally applied

movement: implications for the perception of self-motion and spatial

memory. Frontiers in Integrative Neuroscience 2014;7:108.

4. Smith PF, Zheng Y. From ear to uncertainty: vestibular contributions to cog-

nitive function. Frontiers in Integrative Neuroscience 2013;7:84.

5. Yoder RM, Taube JS. The vestibular contribution to the head direction sig-

nal and navigation. Frontiers in Integrative Neuroscience 2014;8:32.

6. Agrawal Y, Carey JP, Della Santina CC, Schubert MC, Minor LB. Disorders

of balance and vestibular function in US adults: data from the National

Health and Nutrition Examination Survey, 2001–2004. Archives of Internal

Medicine 2009;169(10):938–44.

7. Paige GD. Senescence of human visual-vestibular interactions. Journal of

Vestibular Research 1992;2(2):133–51.

8. Zalewski CK. Aging of the human vestibular system. Seminars in Hearing

2015;36(3):175–96.

9. Semenov YR, Bigelow RT, Xue Q, du Lac S, Agrawal Y. Association between

vestibular and cognitive function in U.S. adults: data from the National

Health and Nutrition Examination Survey. The Journals of Gerontology:

Series A: Biological Sciences and Medical Sciences 2016;71(2):243–50.

10. Bigelow RT, Semenov YR, Trevino C, Ferrucci L, Resnick SM, Simonsick EM,

etal. Association between visuospatial ability and vestibular function in the

Baltimore Longitudinal Study of Aging. Journal of the American Geriatrics

Society 2015;63(9):1837–44.

11. Xie Y, Bigelow RT, Frankenthaler SF, Studenski SA, Moffat SD, AgrawalY.

Vestibular loss in older adults is associated with impaired spatial navigation:

data from the triangle completion task. Frontiers in Neurology 2017;8:173.

12. Anson ER, Ehrenburg MR, Wei EX, Bakar D, Simonsick E, Agrawal Y.

Saccular function is associated with both angular and distance errors on the

triangle completion test. Clinical Neurophysiology 2019;130(11):2137–43.

13. Bilkey DK. Space and context in the temporal cortex. Hippocampus

2007;17(9):813–25.

14. Lopez C, Blanke O. The thalamocortical vestibular system in animals and

humans. Brain Research Reviews 2011;67(1–2):119–46.

15. Hitier M, Besnard S, Smith PF. Vestibular pathways involved in cognition.

Frontiers in Integrative Neuroscience 2014;8:59.