RESEARCH OPEN ACCESS

Visual tests predict dementia risk inParkinson diseaseLouise-Ann Leyland PhD Fion D Bremner PhD Ribeya Mahmood MSc Sam Hewitt MSc

Marion Durteste MSc Molly RE Cartlidge Michelle M-M Lai FRCOphth Luke E Miller PhD

Ayse P Saygin PhD Pearse A Keane PhD Anette E Schrag PhD and Rimona S Weil PhD

Neurology Clinical Practice February 2020 vol 10 no 1 29-39 doi101212CPJ0000000000000719

Correspondence

Dr Weil

rweiluclacuk

AbstractObjectiveTo assess the role of visual measures and retinal volume to predictthe risk of Parkinson disease (PD) dementia

MethodsIn this cohort study we collected visual cognitive and motor datain people with PD Participants underwent ophthalmic examina-tion retinal imaging using optical coherence tomography andvisual assessment including acuity and contrast sensitivity and high-level visuoperception measures of skew tolerance and biologicalmotion We assessed the risk of PD dementia using a recently de-scribed algorithm that combines age at onset sex depressionmotor scores and baseline cognition

ResultsOne hundred forty-six people were included in the study (112 withPD and 34 age-matched controls) The mean disease duration was41 (plusmn2middot5) years None of these participants had dementia Higher risk of dementia wasassociated with poorer performance in visual measures (acuity ρ = 029 p = 00024 contrastsensitivity ρ = minus037 p lt 00001 skew tolerance ρ = minus025 p = 00073 and biological motionρ = minus026 p = 00054) In addition higher risk of PD dementia was associated with thinnerretinal structure in layers containing dopaminergic cells measured as ganglion cell layer (GCL)and inner plexiform layer (IPL) thinning (ρ = minus029 p = 00021 ρ = minus033 p = 000044) Theserelationships were not seen for the retinal nerve fiber layer that does not contain dopaminergiccells and were not seen in unaffected controls

ConclusionVisual measures and retinal structure in dopaminergic layers were related to risk of PD de-mentia Our findings suggest that visual measures and retinal GCL and IPL volumes may beuseful to predict the risk of dementia in PD

Dementia Research Centre (L-AL RM RSW) Institute of Neurology University College London United Kingdom Neuro-ophthalmology (FDB MM-ML) National Hospital for Neurology andNeurosurgery University College London Hospitals London United Kingdom Institute of Neurology (SH MD MREC) University College London UCL United Kingdom School ofBiomedical Sciences (MREC) Biological Sciences Leeds University United Kingdom ImpAct (LEM) Lyon Neuroscience Research Center France Department of Cognitive Science (APS)University of California San Diego Kavli Institute for Brain and Mind (APS) University of California San Diego Institute of Ophthalmology (PAK) UCL United Kingdom Moorfields EyeHospital (PAK) London United Kingdom Department of Clinical Neuroscience (AES) Institute of Neurology UCL Hampstead Campus London United Kingdom Movement DisordersConsortium (AES RSW) UCL United Kingdom and The Wellcome Centre for Human Neuroimaging (RSW) Institute of Neurology University College London United Kingdom

Funding information and disclosures are provided at the end of the article Full disclosure form information provided by the authors is available with the full text of this article atNeurologyorgcp

The Article Processing Charge was funded by Wellcome Trust

This is an open access article distributed under the terms of the Creative Commons Attribution License 40 (CC BY) which permits unrestricted use distribution and reproduction in anymedium provided the original work is properly cited

Copyright copy 2019 The Author(s) Published by Wolters Kluwer Health Inc on behalf of the American Academy of Neurology 29

Dementia is a debilitating aspect of Parkinson disease (PD)affecting 50 of patients within 10 years of diagnosis withvariability in timing and severity Patients with PD who showvisual deficits including color and higher-order visual changesmay have higher rates of converting to PD dementia or de-velop dementia earlier in their disease course1ndash3 Howeverthis has not yet been examined systematically and whetherearlier stages of visual processing and retinal structure arelinked to the risk of PD dementia is not yet known

Retinal structure can be imaged noninvasively using opticalcoherence tomography (OCT)4 and shows thinning in Alz-heimer disease5 and as a population screen for dementia6ndash8

Retinal structural changes are seen in PD but early studies ofretinal nerve fiber layer (RNFL) thinning in PD were notreplicated9 The location of dopaminergic amacrine cells in theinner plexiform layer (IPL)10 in nonhuman studies and post-mortem findings of phosphorylated alpha-synuclein in the IPLsuggest that deeper layers aremore likely to be affected in PD11

Recent studies show consistent thinning in the IPL and gan-glion cell layer (GCL) in PD12 but whether IPL or GCLthinning is linked to PD dementia is not known

New algorithms have emerged that combine measures suchas age and motor severity with sensitivity to predict cognitivechange in PD13ndash15 Although these algorithms are usefulthey do not allow tracking of disease progression or relatedirectly to changes in the parkinsonian brain

We therefore examined the association between visualmeasures and retinal structure with the risk of dementia in

PD These measures have potential for stratifying high-riskpatients for clinical trials

MethodsParticipantsOne hundred seventeen people with PD were recruited fromour UK center between October 2017 and November 2018Inclusion criteria were early-stage PD (Queen Square BrainBank criteria) within 10 years of diagnosis aged 49ndash82 yearsExclusion criteria were confounding neurologic or psychi-atric disorders a diagnosis of dementia or Mini-Mental StateExamination (MMSE) less than 2513 and ophthalmic dis-ease sufficient to impair visual acuity Two patients wereexcluded because of dementia and 3 were excluded becauseof ophthalmic disease (glaucoma) Therefore the datareported here include 112 people with PD Thirty-five age-matched controls unaffected by neurologic psychiatric orophthalmic disease were additionally recruited from uni-versity databases and unaffected spouses Of these 1 wasexcluded because of developing mild cognitive impairmentwithin 6 months of taking part leaving 34 controls in theanalysis reported here

Clinical and ophthalmic evaluationAll participants were tested on their usual medications andlevodopa equivalent daily dose calculated Symptom severitywas assessed using the Movement Disorders Society UnifiedPD Rating Scale (UPDRS) Olfaction was assessed using theodor identification subset of the Sniffinrsquo sticks test Partic-ipants completed the Hospital Anxiety and Depression Scale

30 Neurology Clinical Practice | Volume 10 Number 1 | February 2020 NeurologyorgCP

and REM Sleep Behavior Disorder Screening Questionnaire(RBDSQ) A comprehensive ophthalmic assessment wasperformed by a consultant ophthalmologist This includedslit-lamp ophthalmic examination and measurement of in-traocular pressures using Goldmann applanation tonometry

Assessments of visual functionVisual measures (figure 1) were all performed before my-driasis Visual acuity was measured binocularly using a log-MAR chart (with eyeglasses if worn) (figure 1A) Contrastsensitivity was measured binocularly (with eyeglasses ifworn) using a Pelli-Robson chart (SSV-281-PC) (sussex-visioncouk) (figure 1C) Color vision was assessed usingthe D15 test and error scores log transformed

Visuoperception was measured using 2 tasks probing distinctaspects of higher-order visuoperception the Cats-and-Dogstest measures tolerance to visual skew (figure 1E) and has beenassociated with PD16 Stimuli were generated as previouslydescribed216 Images were presented centrally subtending 4deg times13deg of visual angle and shown for 280 ms followed by a choicescreen (response time 3800 ms) 90 repetitions (total time 15minutes) To calculate discrimination sensitivity missing trialswere excluded and performance at each level of skew calculatedA sigmoid psychometric curve was fitted and threshold forimage detection was calculated at 75 performance Biologicalmotion measures sensitivity to detect the appearance ofa moving person frommoving dots at the position of the majorjoints of a person17 It is known to be affected in PD18 Stimuliconsisted of point-light walkers (12 white dots on gray back-ground height 7deg) (figure 1G) Control stimuli were generatedusing point-light walkers with dot position and motionscrambled Motion-matched noise dots were added to increasedifficulty using an adaptive staircase procedure17 Stimuli werepresented for 800 ms Participants determined whether theanimation depicted a moving person or scrambled movingdots 225 repetitions total time 15 minutes We used theQUEST Bayesian adaptive method to calculate the maximumnumber of additional noise dots tolerated for performance toreach 82 accuracy Stimuli were generated within MATLABPsychophysics Toolbox 3 implemented on a Dell Latitude3340 in a darkened roomWhether this test relates to the risk ofPD dementia is not known

Retinal structure OCTRetinal imaging was performed on both eyes of each par-ticipant on the same day as other measures were obtainedInner retinal layer structure was measured using high-resolution spectral domain optical coherence tomography(SD-OCT Heidelberg HRASpectralis manufactured in2011) in a dimly lit room after pharmacologic mydriasisaccording to a standard protocol19 Two OCT devices wereused at 1 site (National Hospital for Neurology and Neu-rosurgery) with 4 operators (L-AL RM SH andMREC) with viewing module v 6950 All participantsunderwent macular scan protocols in Infrared-OCT modelaser illumination used for excitation was 486 nm Thisallows simultaneous acquisition of a fundus image witha reflectance wavelength of 816 nm and OCT scan withTruTrack eye-tracking technology to stabilize the retinalimage The volumetric macular protocol of the SpectralisSD-OCT device (NSite application) was used to measuremacular thickness and volume (figure 2A) This uses aninternal fixation source and centers on the participantrsquosfovea The protocol consists of 25 vertical line scans ata resolution of 1536 (scanning angle 20degtimes20deg density240 μm 47 scanss automatic real-time frames 49)Quality control of OCT data was performed in compliancewith international consensus OSCAR 1B OCT qualitycontrol standards (OSCAR-IB)20 and OCT data werecollected and reported according to international APOS-TEL guidelines21

OCT analysisAutomatic layer segmentation was applied via Heidelbergsoftware v11020 with Nsite module included to computethe thickness of the GCL IPL and RNFL Blinded assessors(SH and MREC) verified layer position and manualsegmentation was undertaken when automatic segmentationdeviated from the visible gradient change for that layer(quality checked by LL) All layers were segmented andexported and total volume and total thickness within each ofthe RNFL GCL and IPL were calculated across all 4quadrants and summed for a total layer thickness exported asa 1- 3- 6-mm EDTRS grid Retinal data from 9 individualswere excluded because of ophthalmic disease or poor-qualityscans (table e-1 linkslwwcomCPJA126) One eye fromeach participant was selected for analysis In patients withPD this was the eye contralateral to the most symptomaticside as identified during UPDRS-III For controls andpatients with symmetrical motor signs the selected eye wasselected randomly In 8 patients and in 1 control the se-lected eye could not be used because of ophthalmic diseaseexclusions and the other eye was included for these partic-ipants (table e-1)

Neuropsychological evaluationCognitive assessment was in line with recent MovementDisorder Society guidelines22 with 2 assessments percognitive domain General cognitive function was

In this large PD cohort visual

measures across multiple stages of

visual processing and structural

retinal changes are associated with

a higher risk of more rapid

development of PD dementia

NeurologyorgCP Neurology Clinical Practice | Volume 10 Number 1 | February 2020 31

Figure 1 Tests of visual function and relationship with risk of Parkinson disease (PD) dementia

(A) LogMAR visual acuity chart Adapted by permission from BMJ Publishing Group Limited35 (B) Relationship between the risk of PD dementia and visualacuity (C) Pelli-Robson chart for assessing contrast sensitivity (D) Relationship between the risk of PD dementia and contrast sensitivity (E) Cats-and-Dogstest of higher-order visuoperception The task is to identify whether the animal shown is a cat or a dog with differing amounts of skew applied to the image todetermine the level of skew tolerated (F) Relationship between the risk of PD dementia and higher-order visuoperception tested by skew tolerance (G)Biological motion Dots are shown at the position of themajor joints of the body The dotsmove to give the strong percept of a person walking Extra dots areadded and the number of dots tolerated where the participant can still detect a person moving is calculated (H) Relationship between the risk of PDdementia and higher-order visuoperception tested with biological motion Poorer performance in each of these measures is linked with a higher risk of PDdementia

32 Neurology Clinical Practice | Volume 10 Number 1 | February 2020 NeurologyorgCP

assessed with the MMSE and Montreal Cognitive As-sessment Memory was assessed with the RecognitionMemory Test for words and immediate and delayed ver-sions of the Logical Memory task (Wechsler MemoryScale IV) Language was assessed with the Graded Nam-ing Test and letter fluency Visuospatial abilities weretested with Benton Judgment of Line Orientation andHooper Visual Organization Test Executive functionswere measured with the Stroop task from the Delis-KaplanExecutive Function System and category fluency Atten-tion was tested using color naming from the Stroop andDigit Span

Defining dementia risk statusWe defined dementia risk using a recently describedprospectively validated algorithm13 This combinesclinical information on sex age at disease onset years ofeducation UPDRS-III (motor examination) and MMSEto generate a risk score (e-Methods for details linkslwwcomCPJA137) In addition to the continuous scoreswe categorized patients with PD into high vs low risk ofdementia using a median split of these algorithmicscores As this algorithm includes age at diagnosis itcannot be used to stratify controls We repeated ouranalysis with 2 other recently described algorithms oneusing age years of education UPDRS-III RBDSQ

depression and olfaction14 and the other using ageUPDRS axial scores and animal fluency15 (e-Methodsfor further details)

Statistical analysesPerformance was compared between groups using 2-tailedWelch t tests or Mann-Whitney-Wilcoxon tests for non-normally distributed data We used linear regression toexamine the effects of visual measures and retinal struc-ture on dementia risk in PD and controls and Spearmanrank correlation where data were not normally distrib-uted p lt 005 Bonferroni corrected for multiple com-parisons (8 comparisons significance lt00063) wasaccepted as the threshold for statistical significanceSample sizes were based on power analyses performedbefore data collection Analyses were performed in R (r-projectorg) Data were inspected and outliers (beyondmean plusmn 3 times SD) removed Participants with missing datawere omitted for that measure (table e-1 linkslwwcomCPJA126)

Standard protocol approvals registrationsand patient consentsAll participants gave written informed consent and the studywas approved by the Queen Square Research Ethics Com-mittee (15LO0476)

Figure 2 Relationship between the retinal volume and risk of Parkinson disease (PD) dementia

(A) Output of optical coherence tomography retinal imaging with cross section at themacula Retinal layers identified by automatic segmentation are shown(B) Relationship between the risk of PD dementia and RNFL volume (C) Relationship between the risk of PD dementia and GCL volume (D) Relationshipbetween the risk of PD dementia and IPL volume Retinal layers that contain dopaminergic cells (GCL and IPL) show greater thinning linked with PD dementiarisk This relationship is not seen in the RNFL that does not contain dopaminergic cells RNFL = retinal nerve fiber layer GCL = ganglion cell layer IPL = innerplexiform layer

NeurologyorgCP Neurology Clinical Practice | Volume 10 Number 1 | February 2020 33

Data availabilityAnonymized data can be made available by request fromqualified investigators to the senior author of this publication

ResultsDemographicsOne hundred twelve people with PD and 34 unaffectedcontrols took part Age and sex did not differ between PD andcontrols mean age PD 643 plusmn 8 years mean age controls 648plusmn 9 years (t(49) = 03 p = 074) Fifty-six people with PDwere assigned low risk based on a median split of the riskscore13 and 56 were high risk The average disease durationwas 41 plusmn 25 years and did not differ between high- and low-risk patients There was also no difference in motor sleepolfaction or levodopa dose between high- vs low-riskpatients (table 1) (age differed between high- and low-riskgroups as this was strongly linked to age at disease onsetthat defined risk and sex was included in the riskalgorithm)13

Relationship between visual measures and PDdementia riskThere was no difference in visual measures between peoplewith PD and unaffected controls apart from poorer highervisual function (Cats-and-Dogs test) in people with PD thatdid not survive correction formultiple comparisons (table 2)However when patients were split into high vs low risk ofPD dementia using a median split high-risk patients showedworse performance in almost all visual measures (table 2)

Table 1 Demographics

ControlsN = 34

PD allN = 112

t or χ2 p (PD vsHC)

Low riskN = 56

High risk N =56

t or χ2 p (low risk vshigh risk)

Age 648 (9) 643 (8) 03 (49) 074 591 (5) 694 (7) minus93(102)

lt00001a

Disease duration mdash 41 (25) mdash mdash 44 (3) 38 (2) 12 (110) 025

Age onset PD mdash 607 (8) mdash mdash 55 (5) 66 (7) minus100(97)

lt00001a

UPDRS total 84 (5) 465 (22) minus167(141)

lt00001a 432 (22) 498 (22) minus16(110)

012

LEDD mdash 4388 (256) mdash mdash 463 (295) 414 (209) 10 (99) 031

Sex FM 1915 5359 076 (1) 038 3620 1739 13 (1) 00003a

RBDSQ 18 (14) 42 (25) minus73 (98) lt00001a 42 (25) 42 (25) 00 (110) 10

Smell test 122 (26) 77 (32) 85 (66) lt00001a 83 (3) 70 (3) 23 (106) 0026

Neuropsychology

Baseline MoCA 287 (13) 279 (19) 27 (82) 00087 284 (15) 274 (22) 28 (99) 00069

Language (GNT) 227 (6) 236 (3) minus09 (39) 039 237 (3) 235 (3) 031(104)

076

Executive (stroop interferencetime)

25 (8) 58 (17) minus16(113)

012 20 (13) 97 (19) minus25 (94) 0016

Attention (digit span forward) 93 (2) 93 (2) 005 (24) 096 96 (2) 91 (2) 11 (84) 027

Memory (logical memory(delayed))

141 (4) 133 (4) 08 (26) 045 139 (4) 127 (4) 13 (85) 018

Visuoperceptual (Hooper) 257 (2) 244 (3) 30 (85) 00036a 254 (3) 234 (3) 34 (109) 000087a

Fluency (category) 220 (5) 213 (6) 07 (60) 052 225 (6) 201 (6) 22 (110) 0027

Abbreviations GNT = Graded Naming Test HC = healthy control LEDD = levodopa equivalent daily dose MoCA = Montreal Cognitive Assessment PD =Parkinson disease RBDSQ = REM Sleep Behavior Disorder Screening Questionnaire UPRDS = Unified Parkinsonrsquos Disease Rating ScaleHigh and low risk refer to risk of developing PD dementia as calculated using a median split for established algorithms13a Indicates significance after Bonferroni correction (p lt 00063)

Poorer visual function across several

levels of visual processing was

associated with a higher risk of

dementia although patients did not

report deficits in visual function

34 Neurology Clinical Practice | Volume 10 Number 1 | February 2020 NeurologyorgCP

with similar differences seen when patients were divided intoquartiles for risk (table e-2 linkslwwcomCPJA126)

Poorer visual function across several levels of visual pro-cessing was associated with a higher risk of dementia (table3) although patients did not report deficits in visual func-tion Visual acuity measured using the logMAR and contrastsensitivity measured using the Pelli-Robson and color vi-sion were all associated with a higher risk of dementia(ρ = 029 p = 00024 ρ = minus037 p lt 00001 ρ = 026 p =00054) (table 3 and figure 1AndashD) Higher-order visuo-perception measured using biological motion showed im-paired performance in higher-risk patients (ρ = minus0026 p =00054) with trend to significance for visual skew (ρ = minus025p = 00073) (table 3 and figure 1EndashF)

Relationship between structural retinalmeasures and PD dementia riskPatients at a higher risk of dementia showed more retinalthinning in dopamine-containing layers GCL (ρ = minus029 p= 00021) and IPL (ρ = minus033 p = 000044) These dif-ferences were not seen in the RNFL that does not containdopaminergic cells (ρ = 0012 p = 090 (table 3 andfigure 2)

Generalizability across other estimates of riskTo ensure that our findings were generalizable across dif-ferent measures of risk we repeated our analysis using 2alternative Parkinson dementia risk algorithms one adaptedfor clinical values14 and found qualitatively the same rela-tionships (table 4)

Table 2 Visual scores in controls and PD and in low vs high risk for dementia in PD

Visual measure Controls N = 34 PD N = 112 t (or W) p Value

Acuity (logMAR) minus008 (02) minus0087 (01) 023 (40) 082

Contrast sensitivity (Pelli-Robson) 180 (02) 179 (02) 2104a 032

Color vision (log D15 error score) 026 (04) 028 (04) 1804a 078

Visuoperception Cats-and-Dogs 21 (06) 19 (06) 21 (54) 0045

Visuoperception biological motion 148 (10) 162 (11) 18185a 069

IPL thickness 297 (23) 294 (23) 056 (47) 058

IPL volume 086 (008) 085 (007) 068 (45) 050

GCL thickness 343 (31) 339 (31) 066 (47) 051

GCL volume 10 (01) 10 (01) 056 (43) 058

RNFL thickness 249 (26) 249 (28) minus0064 (50) 095

RNFL volume 093 (01) 092 (01) 014 (45) 089

Visual measure Low risk N = 56 High risk N = 56 t (or W) p Value

Acuity (logMAR) minus012 (01) minus0054 (01) minus28 (105) 00062b

Contrast sensitivity (Pelli-Robson) 184 (02) 174 (02) 2046a 00028b

Color vision (log D15 error score) 019 (04) 037 (05) 1211a 0043

Visuoperception Cats-and-Dogs 20 (05) 17 (06) 26 (109) 0009c

Visuoperception biological motion 191 (13) 133 (9) 20095a 0010c

IPL thickness 301 (20) 288 (25) 31 (101) 00029b

IPL volume 087 (006) 083 (008) 29 (101) 00045b

GCL thickness 345 (27) 332 (34) 23 (100) 0026c

GCL volume 102 (008) 098 (01) 25 (103) 0016c

RNFL thickness 251 (29) 247 (27) 07 (104) 048

RNFL volume 093 (01) 092 (01) 045 (105) 065

Abbreviations GCL = ganglion cell layer IPL = inner plexiform layer PD = Parkinson disease RNFL = retinal nerve fiber layera Mann-Whitney-Wilcoxon test used for non-normal datab Indicates significance after Bonferroni correction (p lt 00063)c Trend to significance after Bonferroni correction

NeurologyorgCP Neurology Clinical Practice | Volume 10 Number 1 | February 2020 35

Relationship between vision anddementia riskin unaffected controlsTo test whether these findings were related to the risk of PDdementia rather than nonspecific effects of aging we exam-ined these relationships in unaffected controlsWe found thatunlike in PD in controls visual measures were not related todementia risk scores or age (table 4)

DiscussionIn this large PDcohort visualmeasures acrossmultiple stages ofvisual processing and structural retinal changes are associatedwith a higher risk of more rapid development of PD dementia

Isolated visual measures such as pentagon copying and colorvision have previously been linked with cognitive changes inPD13 and involvement of visual processing brain regions isassociated with more rapid PD dementia23 Here we showthat vision along the entire visual processing axis is linkedwith the risk of cognitive change in PD This spans fromhigher-order processes involving posterior brain regions tocontrast sensitivity and visual acuity mediated by ophthalmicstructures or primary visual cortex and dopamine-containinglayers of the retina

Our finding that retinal thinning is linked with a higher riskof more rapid PD dementia is consistent with a growingliterature showing retinal involvement in PD Dopamine isa key modulatory neurotransmitter in the retina24 Reduceddopamine innervation is seen around the fovea in PD25 andlower retinal dopaminergic concentrations are found atpostmortem in untreated PD26 OCT initially suggestedRNFL thinning in PD9 but this was not replicated byothers27 potentially due to methodological differences in-cluding sample characteristics lack of appropriate statisti-cal correction and segmentation protocols

As OCT technology has improved it has become clearer thatretinal thinning in PD is restricted to the dopamine-containing layers the GCL and IPL rather than theRNFL12 Nuclei of dopaminergic amacrine interneurons lieadjacent to the IPL24 with axons running horizontally acrossthe IPL and GCL28 and postmortem studies show thatretinal alpha-synuclein accumulates at the interface with theIPL in PD11 Our finding of a link between retinal thinningand dementia risk in these specific layers has importantmechanistic implications for progression of PD dementia

In the wider population RNFL thinning is linked with higherrates of cognitive decline7 and another study found that GC-IPL thinning (but not RNFL) is linked with prevalent de-mentia8 Our finding of a link between retinal thinning andrisk of dementia may not be wholly specific to Parkinsondementia and may also apply to other types of dementiaUltimately this will need to be tested in prospectively fol-lowed cohorts developing different forms of dementia In PDdementia RNFL thinning has previously been shown tocorrelate with MMSE scores29 but this relationship withcognition has not yet been shown in PD without dementiabut at risk of dementia Recent postmortem findings of ret-inal phosphorylated alpha-synuclein strengthen the link be-tween the retina and brain disease in PD as the amount ofphosphorylated alpha-synuclein in the retina correlates withthe density of Lewy-type alpha-synuclein in the brain30 Thisanatomic finding supports the use of retinal structuralmeasures as a window into brain pathology

The reason for selective vulnerability of dopamine-containing GC and IP layers is not known but may relateto common properties of dopamine-containing cellsThese show autonomous spiking behavior with low intrinsicCa2+ buffering31 This may lead to free-radical accumula-tion and increased vulnerability to neurodegenerationWhether retinal layers are affected before cortical regionsin PD dementia can be examined in longitudinal evaluation

We did not find impairment of visual dysfunction in the overallPD group compared with controls apart from skew tolerance(Cats-and-Dogs test) differences were only seen between thehigh- and low-risk individuals Previous studies have reportedvisual dysfunction in PD compared with controls23233 It istherefore possible that previous studies reporting visual deficits

Table 3 Relationship between visual measures and PDdementia risk

Visual measures Rho p Value

Visual measures

Acuity (logMAR) 029 00024a

Contrast sensitivity (Pelli-Robson) minus037 lt00001a

Color vision (log D15 error score) 026 00054a

Visuoperception Cats-and-Dogs minus025 00073b

Visuoperception biological motion minus026 00054a

Retinal structure

IPL thickness minus034 000031a

IPL volume minus033 000044a

GCL thickness minus028 00040a

GCL volume minus029 00021a

RNFL thickness 00079 094

RNFL volume 0012 090

Other measures

MOCA minus038 lt00001a

UPDRS (motor) 028 00030a

UPDRS (total) 024 00096b

Abbreviations GCL = ganglion cell layer IPL = inner plexiform layer MoCA =Montreal Cognitive Assessment RNFL = retinal nerve fiber layer UPDRS =Unified Parkinsonrsquos Disease Rating Scalea Indicates significance after Bonferroni correction (p lt 00063)b Indicates trend to significance

36 Neurology Clinical Practice | Volume 10 Number 1 | February 2020 NeurologyorgCP

in PD overall included higher proportions of high-risk individ-uals or patients with cognitive involvement

LimitationsAlthough our study examines risk based on cross-sectionaldata the algorithms we use are validated using prospectivefollow-up Ultimately however whether measures identifiedhere truly predict the development of dementia in PD willneed to be tested in longitudinal analyses

A further question is whether retinal and visual measuresare affected by the same factors that promote Parkinsondementia such as age The relationship between older ageand poorer vision is well established34 and higher age atonset and age in itself is strongly linked with developmentof dementia in PD1415 Whether some factor such as in-creased amyloid deposition in cortical and retinal struc-tures is responsible for both effects or whether there areseparate processes affecting vision more selectively is not

Table 4 Relationship between vision scores and 2 other risk algorithms

Risk score 215 Risk score 316

R2 p Value R2 p Value

People with Parkinson disease

Visual measures

Acuity (logMAR) 0090 00014a 015 lt00001a

Contrast sensitivity (pelli-Robson) 020 lt00001a 021 lt00001a

Color vision (log D15 error score) 00041 051 0041 0034

Visuoperception Cats-and-Dogs 0092 00011a 012 000013a

Visuoperception biological motion 0097 000085a 024 lt00001a

Retinal structures

IPL thickness 0077 00039a 0068 00068b

IPL volume 0067 00071b 0078 00036a

GCL thickness 0058 0012 0063 00092b

GCL volume 0064 00083b 0087 00021a

RNFL thickness 00018 067 00022 064

RNFL volume 00021 064 00013 071

Unaffected controls

Visual measures

Acuity (logMAR) 00023 079 00062 066

Contrast sensitivity (pelli-Robson) 0099 0070 012 0043

Color vision (log D15 error score) 017 0017 0078 011

Visuoperception Cats-and-Dogs 0016 047 009 0084

Visuoperception biological motion 0059 017 0019 044

Retinal structures

IPL thickness 0061 019 0022 043

IPL volume 0076 014 0033 033

GCL thickness 011 0069 0081 013

GCL volume 011 0068 0080 013

RNFL thickness 012 0061 0035 032

RNFL volume 0094 010 0017 049

Abbreviations GCL = ganglion cell layer IPL = inner plexiform layer RNFL = retinal nerve fiber layera Indicates significance after Bonferroni correction (p lt 00063)b Indicates trend to significance after Bonferroni correction

NeurologyorgCP Neurology Clinical Practice | Volume 10 Number 1 | February 2020 37

yet known Age (or age at onset) is incorporated into allalgorithmic risk scores For this reason we were unable tocorrect for age in our regression analyses However lack ofassociation between visual measures and dementia risk inunaffected controls suggests that this relationship betweenvision and dementia is more specific to the risk of PDdementia and is not purely a result of deteriorating func-tion with age (although we note that the control group issmaller than the group with PD) We will require longi-tudinal assessment of visual and cognitive factors inpatients with PD to determine and validate whether visualmeasures such as those presented here show additionalsensitivity to recently defined algorithms that use moregeneral clinical inputs

Our finding of a link between acuity and PD dementiarisk raises the question of whether higher-order visualchanges are a result of lower acuity or arise independentlydue to cortical changes in at-risk individuals This will needto be tested by examining cortical differences in thesegroups and their timing relative to retinal and acuitydeficits

We show that visual measures and retinal thinning are linkedwith a higher risk of more rapid PD dementia This providesuseful insights into the development of PD dementia im-plicating visual brain regions as being involved at earlystages As Parkinson dementia is becoming better un-derstood noninvasive visual measures such as those used inthis study may have potential to be used alone or in com-bination with other clinical factors as possible biomarkersfor PD dementia and to stratify high-risk patients fordisease-modifying trials This could enable better-poweredtrials and ultimately may pave the way toward new treat-ments for PD dementia

Study fundingRS W is supported by a Wellcome Clinical Research CareerDevelopment Fellowship (201567Z16Z) This work wasfunded by grants from UCLH Biomedical Research CentreGrant (BRC302NSRW101410) and by grants from Na-tional Institute for Health Research (NIHR) and Fight forSight UK

DisclosureL-A Leyland FD Bremner R Mahmood S HewittM Durteste MRE Cartlidge MM-M Lai LE Miller andAP Saygin report no disclosures PA Keane reports per-sonal fees from Topcon Heidelberg Engineering Deep-Mind Optos Novartis Bayer Allergan and Santen AESchrag reports personal fees from Medtronic and AstraZe-neca RS Weil has received personal fees from GE Fulldisclosure form information provided by the authors isavailable with the full text of this article at Neurologyorgcp

Publication historyReceived by Neurology Clinical Practice April 9 2019 Accepted in finalform June 7 2019

References1 Anang JB Gagnon JF Bertrand JA et al Predictors of dementia in Parkinson disease

a prospective cohort study Neurology 2014831253ndash12602 Weil RS Pappa K Schade RN et al The Cats-and-Dogs test a tool to identify

visuoperceptual deficits in Parkinsonrsquos disease Mov Disord 2017321789ndash17903 Williams-Gray CH Mason SL Evans JR et al The CamPaIGN study of Parkinsonrsquos

disease 10-year outlook in an incident population-based cohort J Neurol NeurosurgPsychiatry 2013841258ndash1264

Appendix Authors

Name Location Contribution

Louise-AnnLeyland PhD

University CollegeLondon London

Designed andconceptualized the studycollected data analyzeddata and drafted themanuscript for intellectualcontent

Fion D BremnerPhD

National Hospital forNeurology ampNeurosurgery London

Collected and interpreteddata and reviewed themanuscript for intellectualcontent

RibeyaMahmood MSc

University CollegeLondon London

Collected and analyzeddata and reviewed themanuscript for intellectualcontent

Sam Hewitt MSc University CollegeLondon London

Collected and analyzeddata and reviewed themanuscript for intellectualcontent

Marion DurtesteMSc

University CollegeLondon London

Collected and analyzeddata and reviewed themanuscript for intellectualcontent

Molly RECartlidge

University CollegeLondon London

Collected and analyzeddata and reviewed themanuscript for intellectualcontent

Michelle M-MLai FRCOphth

National Hospital forNeurology ampNeurosurgery London

Collected and interpreteddata and reviewed themanuscript for intellectualcontent

Luke E MillerPhD

Lyon NeuroscienceResearch Center France

Designed aspects of thestudy interpreted thedata and reviewed themanuscript for intellectualcontent

Ayse P SayginPhD

University of CaliforniaSan Diego

Designed aspects of thestudy interpreted thedata and reviewed themanuscript for intellectualcontent

Pearse A KeanePhD

University CollegeLondon London

Designed aspects of thestudy interpreted thedata and reviewed themanuscript for intellectualcontent

Anette E SchragPhD

University CollegeLondon London

Interpreted the data andreviewed the manuscriptfor intellectual content

Rimona S WeilPhD

University CollegeLondon London

Designed andconceptualized the studyanalyzed the datainterpreted the data anddrafted revised andedited the manuscriptfor intellectualcontent

38 Neurology Clinical Practice | Volume 10 Number 1 | February 2020 NeurologyorgCP

4 Bodis-Wollner I Miri S Glazman S Venturing into the no-manrsquos land of the retina inParkinsonrsquos disease Mov Disord 20142915ndash22

5 Cheung CY Ong YT Hilal S et al Retinal ganglion cell analysis using high-definitionoptical coherence tomography in patients with mild cognitive impairment and Alz-heimerrsquos disease J Alzheimers Dis 20154545ndash56

6 Cheung CY Chan VT Mok VC Chen C Wong TY Potential retinal biomarkers fordementia what is new Curr Opin Neurol 20193282ndash91

7 Ko F Muthy ZA Gallacher J et al Association of retinal nerve fiber layer thinningwith current and future cognitive decline a study using optical coherence tomogra-phy JAMA Neurology 2018751198ndash1205

8 Mutlu U Colijn JM Ikram MA et al Association of retinal neurodegeneration onoptical coherence tomography with dementia a population-based study JAMANeurol 2018751256ndash1263

9 Inzelberg R Ramirez JA Nisipeanu P Ophir A Retinal nerve fiber layer thinning inParkinson disease Vision Res 2004442793ndash2797

10 Balasubramanian R Gan L Development of retinal amacrine cells and their dendriticstratification Current Ophthalmol Rep 20142100ndash106

11 BeachTGCarew J SerranoG et al Phosphorylated α-synuclein-immunoreactive retinalneuronal elements in Parkinsonrsquos disease subjects Neurosci Lett 201457134ndash38

12 Zivkovic M Dayanir V Stamenovic J et al Retinal ganglion cellinner plexiform layerthickness in patients with Parkinsonrsquos disease Folia Neuropathol 201755168ndash173

13 Liu G Locascio JJ Corvol JC et al Prediction of cognition in Parkinsonrsquos disease witha clinical-genetic score a longitudinal analysis of nine cohorts Lancet Neurol 201716620ndash629

14 Schrag A Siddiqui UF Anastasiou Z Weintraub D Schott JM Clinical variables andbiomarkers in prediction of cognitive impairment in patients with newly diagnosedParkinsonrsquos disease a cohort study Lancet Neurol 20171666ndash75

15 Velseboer DC de Bie RM Wieske L et al Development and external validation ofa prognostic model in newly diagnosed Parkinson disease Neurology 201686986ndash993

16 Weil RS Schwarzkopf DS Bahrami B et al Assessing cognitive dysfunction in Par-kinsonrsquos disease an online tool to detect visuo-perceptual deficits Mov Disord 201833544ndash553

17 Saygin AP Superior temporal and premotor brain areas necessary for biologicalmotion perception Brain 20071302452ndash2461

18 Jaywant A Shiffrar M Roy S Cronin-Golomb A Impaired perception of biologicalmotion in Parkinsonrsquos disease Neuropsychology 201630720ndash730

19 Cetin EN Bir LS Sarac G Yaldızkaya F Yaylalı V Optic disc and retinal nerve fibrelayer changes in Parkinsonrsquos disease Neuroophthalmology 20133720ndash23

20 Tewarie P Balk L Costello F et al TheOSCAR-IB consensus criteria for retinal OCTquality assessment PloS one 20127e34823

21 Cruz-Herranz A Balk LJ Oberwahrenbrock T et al The APOSTEL recom-mendations for reporting quantitative optical coherence tomography studies Neu-rology 2016862303ndash2309

22 Litvan I Goldman JG Troster AI et al Diagnostic criteria for mild cognitive im-pairment in Parkinsonrsquos disease movement disorder society task force guidelinesMov Disord 201227349ndash356

23 Toledo JB Gopal P Raible K et al Pathological α-synuclein distribution in subjects withcoincident Alzheimerrsquos and Lewy body pathology Acta Neuropathol 2016131393ndash409

24 Frederick JM Rayborn ME Laties AM Lam DM Hollyfield JG Dopaminergicneurons in the human retina J Comp Neurol 198221065ndash79

25 Nguyen-Legros J Functional neuroarchitecture of the retina hypothesis on thedysfunction of retinal dopaminergic circuitry in Parkinsonrsquos disease Surg Radiol Anat198810137ndash144

26 Harnois C Di Paolo T Decreased dopamine in the retinas of patients with Parkin-sonrsquos disease Invest Ophthalmol Vis Sci 1990312473ndash2475

27 Tsironi EE Dastiridou A Katsanos A et al Perimetric and retinal nerve fiber layerfindings in patients with Parkinsonrsquos disease BMC Ophthalmol 20121254

28 Archibald NK Clarke MP Mosimann UP Burn DJ Retinal thickness in Parkinsonrsquosdisease Parkinsonism Relat Disord 201117431ndash436

29 Moreno-Ramos T Benito-Leon J Villarejo A Bermejo-Pareja F Retinal nerve fiberlayer thinning in dementia associated with Parkinsonrsquos disease dementia with Lewybodies and Alzheimerrsquos disease J Alzheimerrsquos Dis 201334659ndash664

30 Ortuntildeo-Lizaran I Beach TG Serrano GE Walker DG Adler CH Cuenca N Phos-phorylated α‐synuclein in the retina is a biomarker of Parkinsonrsquos disease pathologyseverity Movement Disorders 2018331315ndash1324

31 Surmeier DJ Obeso JA Halliday GM Selective neuronal vulnerability in Parkinsondisease Nat Rev Neurosci 201718101ndash113

32 Levin BE Llabre MM Reisman S et al Visuospatial impairment in Parkinsonrsquosdisease Neurology 199141365ndash369

33 Regan D Neima D Low-contrast letter charts in early diabetic retinopathy ocularhypertension glaucoma and Parkinsonrsquos disease Br J Ophthalmol 198468885ndash889

34 McKendrick AM Chan YM Nguyen BN Spatial vision in older adults perceptualchanges and neural bases Ophthalmic and Physiol Opt201838(4)363ndash375

35 Laidlaw D Abbott A Rosser DA Development of a clinically feasible logMAR al-ternative to the Snellen chart performance of the ldquocompact reduced logMARrdquo visualacuity chart in amblyopic children Br J Ophthalmol 2003871232ndash1234

Practical ImplicationsNeurologyreg Clinical Practice is committed to providing clinical insights helpful to neurologists in everyday practice Each FullCase includes a ldquoPractical Implicationsrdquo statement a pearl of wisdom for the practicing clinician

NeurologyorgCP Neurology Clinical Practice | Volume 10 Number 1 | February 2020 39

DOI 101212CPJ000000000000071920201029-39 Published Online before print September 18 2019Neurol Clin Pract

Louise-Ann Leyland Fion D Bremner Ribeya Mahmood et al Visual tests predict dementia risk in Parkinson disease

This information is current as of September 18 2019

ServicesUpdated Information amp

httpcpneurologyorgcontent10129fullhtmlincluding high resolution figures can be found at

References httpcpneurologyorgcontent10129fullhtmlref-list-1

This article cites 35 articles 4 of which you can access for free at

Citations httpcpneurologyorgcontent10129fullhtmlotherarticles

This article has been cited by 2 HighWire-hosted articles

Subspecialty Collections

httpcpneurologyorgcgicollectionvisual_processingVisual processing

httpcpneurologyorgcgicollectionretinaRetina

mhttpcpneurologyorgcgicollectionparkinsons_disease_parkinsonisParkinsons diseaseParkinsonism

tiahttpcpneurologyorgcgicollectionparkinsons_disease_with_demenParkinsons disease with dementiafollowing collection(s) This article along with others on similar topics appears in the

Permissions amp Licensing

httpcpneurologyorgmiscaboutxhtmlpermissionsits entirety can be found online atInformation about reproducing this article in parts (figurestables) or in

Reprints

httpcpneurologyorgmiscaddirxhtmlreprintsusInformation about ordering reprints can be found online

reserved Print ISSN 2163-0402 Online ISSN 2163-0933Published by Wolters Kluwer Health Inc on behalf of the American Academy of Neurology All rightssince 2011 it is now a bimonthly with 6 issues per year Copyright Copyright copy 2019 The Author(s)

is an official journal of the American Academy of Neurology Published continuouslyNeurol Clin Pract