ARTICLE OPEN ACCESS

Association of specific biotypes in patients withParkinson disease and disease progressionLinbo Wang PhD Wei Cheng PhD Edmund T Rolls DPhil DSc Fuli Dai MSc Weikang Gong PhD

Jingnan DuMSc Wei Zhang BSc ShouyanWang PhD Fengtao Liu MD JianWang MD Peter BrownMD and

Jianfeng Feng PhD

Neurologyreg 202095e1445-e1460 doi101212WNL0000000000010498

Correspondence

Dr Feng

jffengfudaneducn

or Dr Brown

peterbrownndcnoxacuk

or Dr Cheng

wchengfdugmailcom

AbstractObjectiveTo identify biotypes in patients with newly diagnosed Parkinson disease (PD) and to testwhether these biotypes could explain interindividual differences in longitudinal progression

MethodsIn this longitudinal analysis we use a data-driven approach clustering PD patients from theParkinsonrsquos Progression Markers Initiative (n = 314 age 610 plusmn 95 years 341 female 5 yearsof follow-up) Voxel-level neuroanatomic features were estimated with deformation-basedmorphometry (DBM) of T1-weighted MRI Voxels with deformation values that were sig-nificantly correlated (p lt 001) with clinical scores (Movement Disorder Societyndashsponsoredrevision of the Unified Parkinsonrsquos Disease Rating Scale Parts IndashIII and total score tremorscore and postural instability and gait difficulty score) at baseline were selected Then theseneuroanatomic features were subjected to hierarchical cluster analysis Changes in the longi-tudinal progression and neuroanatomic pattern were compared between different biotypes

ResultsTwo neuroanatomic biotypes were identified biotype 1 (n = 114) with subcortical brainvolumes smaller than heathy controls and biotype 2 (n = 200) with subcortical brain volumeslarger than heathy controls Biotype 1 had more severe motor impairment autonomic dys-function and much worse REM sleep behavior disorder than biotype 2 at baseline Althoughdisease durations at the initial visit and follow-up were similar between biotypes patients withPD with smaller subcortical brain volume had poorer prognosis with more rapid decline inseveral clinical domains and in dopamine functional neuroimaging over an average of 5 years

ConclusionRobust neuroanatomic biotypes exist in PD with distinct clinical and neuroanatomic patternsThese biotypes can be detected at diagnosis and predict the course of longitudinal progressionwhich should benefit trial design and evaluation

RELATED ARTICLE

EditorialToward biomarker-basedclinical subtyping ofParkinson disease

Page 461

These authors contributed equally to this work

From the Institute of Science and Technology for Brain-inspired Intelligence (LW WC ER FD WG J D WZ SW JF) Fudan University Key Laboratory of ComputationalNeuroscience and Brain-Inspired Intelligence (LW WC J D WZ SW JF) (Fudan University) Ministry of Education Shanghai China Department of Computer Science (ER JF)University of Warwick Coventry Oxford Centre for Computational Neuroscience (ER) UK Department of Neurology and National Clinical Research Center for Aging and Medicine(FL JW) Huashan Hospital Fudan University Shanghai China and Medical Research Council Brain Network Dynamics Unit (PB) and Nuffield Department of Clinical Neuro-sciences (PB) University of Oxford UK

Go to NeurologyorgN for full disclosures Funding information and disclosures deemed relevant by the authors if any are provided at the end of the article

The Article Processing Charge was funded by Shanghai Municipal Science and Technology Major Project and ZJLab

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 2020 The Author(s) Published by Wolters Kluwer Health Inc on behalf of the American Academy of Neurology e1445

Patients with Parkinson disease (PD) present heterogeneousmotor and nonmotor clinical manifestations and have a variableprognosis12 Although the diagnosis of PD is dependent on thepresence of tremor bradykinesia and rigidity some nonmotorphenomenamdasheg REM sleep behavior disorder (RBD)hyposmia and depressionmdashcan precede motor deficits byseveral years Conversely as the disease progresses nonmotorproblems such as autonomic disturbances sleep disorders andcognitive impairment can dominate the clinical picture in somepatients1 Recent evidence suggests that PD may have severalbiotypes3ndash9 but their identity and neurobiological basis remainpoorly understood2 Assuming that homogeneous groups ofpatients are more likely to share pathologic features recogni-tion of different subcategories of patients with PD may be keyto better understanding underlying biological mechanismspredicting disease profile and progression and eventually de-signing more efficient personalized clinical trials23

Subtypes of PD have previously been defined mainly accordingto clinical symptoms and demographic characteristics3ndash8

However cluster results are only as good as the data that un-derpin them and the depths of phenotypic information used bythese studies were variable resulting in quite heterogeneousand controversial clusters2 In addition these clinical data-driven PD subtype classification systems may suffer from lackreproducibility10 An alternative to subtyping patients with PDon the basis of co-occurring clinical symptoms is to identifyneuroanatomic biotypes by clustering patients according toshared neuroanatomic signatures which can objectively cap-ture different aspects of patient characteristics Studying brainneuroanatomic patterns of PD provides an opportunity to ex-amine biological heterogeneity in vivo11 Data-driven methodsprovide an unbiased approach to detect groups of patients withsimilar profiles across multiple neuroanatomic feature dimen-sions and thus may yield a more refined description of het-erogeneity in PD T1-weighted MRI is an especially suitablemodality to describe brain anatomy with high resolution and toquantify regional brain volumes12 Brain volume may mediatebrain reserve which promotes the resilience of large-scale brainnetworks and helps maintain normal function in the face ofneurodegeneration13ndash16 Previous studies have shown that sub-cortical volume loss reflects clinical measures of disease severityand is related to the development of cognitive impairment17ndash21

These studies raise the intriguing possibility that T1-weightedMRI measures of brain volume could be leveraged to identifybiotypes of PD Critically such PD biotypes defined by brainvolume at diagnosis may predict disease progression which maybe advantageous in helping to determine prognosis and to

identify subgroups for clinical trials Cluster analysis in non-demented PDwith limited sampling of patients showed differentPD cortical thinning subtypes22 These PD subtypes also showeddifferent cortical thinning progression over time but the differ-ence between motor symptoms and the rates of disease pro-gression of the different subtypes was not reported Severity andrate of disease progression are an important issue in PD thera-peutics and identifying progression biotypes of PD at diagnosiswith the use of neuroanatomic patterns may be one way toaddress heterogeneity in PD

In this study we used data-driven clustering approaches toidentify neuroanatomic biotypes in patients with early PD in theParkinsonrsquos Progression Markers Initiative (PPMI ppmi-infoorg) database23 according to the neuroanatomic pattern derivedby deformation-basedmorphometry (DBM)24DBM is based onnonlinear and intensity-based registration procedures that spa-tially normalize the entire brain to a standard template24 DBMdoes not assume the distributions of gray matter or white matterand preserves the entirety of the MRI data PD involves axonaldegeneration and neuronal cell death with the latter beingindexed by gray matter atrophy which is a relatively late event inthe pathogenesis of PD25 Moreover neurodegeneration in PDinitially preferentially affects subcortical regions through a pur-ported disease-spreading process26 A particularly strong aspect ofthe DBM method is that it enables the detection of subcorticalneuroanatomic features27 and previous studies have shown thatDBM can detect morphologic tissue changes in patients withearly-stage PD19 Therefore DBM is particularly suitable for PDbiotype discovery compared to cortical thinning patterns andvoxel-based morphometry We hypothesize that if heterogeneityin clinical symptoms reflects true neuroanatomic biotypes of PDthen such neuroanatomic biotypes should be detectable in earlydisease and might predict the type of symptoms or disease pro-gression that a patient will develop The aims of our study were(1) to identify biotypes of PD with cluster analysis based on abaseline neuroimaging dataset (2) to introduce a practical clinicaltyping method which assigns individual patients to their biotypeand (3) to compare the behavioral assessments and rate of diseaseprogression between different PD biotypes

MethodsStandard protocol approvals registrationsand patient consentsThe study was approved by the institutional review board ateach PPMI site Written informed consent for research wasobtained from all participants in the study

GlossaryDAT = dopamine transporter DBM = deformation-based morphometry FDR = false discovery rate MDS-UPDRS =Movement Disorder Societyndashsponsored revision of the Unified Parkinsonrsquos Disease Rating Scale PD = Parkinson diseasePIGD = postural instability and gait disorder PPMI = Parkinsonrsquos Progression Markers Initiative RBD = REM sleep behaviordisorder

e1446 Neurology | Volume 95 Number 11 | September 15 2020 NeurologyorgN

Overall designA flowchart of the analysis is shown in figure 1 We began bydesigning and implementing a preprocessing procedure tocontrol for site- and age-related effects in a multisite datasetthat comprised structural MRI scans for 457 participants (n= 314 patients with PD n = 143 healthy controls) A graphicsummary of the participants selection is shown in dataavailable from Dryad (figure e-1 doiorg105061dryadxsj3tx9bf) Patients and controls were matched for age andsex DBM was used to detect the volume of each voxelcompared to the template brain24 Next to select features foruse in clustering we used Spearman rank correlation analysisto identify a low-dimensional representation of neuroana-tomic features that were associated with baseline clinicalsymptoms within patients with PD including MovementDisorder Societyndashsponsored revision of the Unified Par-kinsonrsquos Disease Rating Scale (MDS-UPDRS) Parts I II IIItremor and postural instability and gait disorder (PIGD)scores To capture more neuroanatomic features related toPD correlations were not corrected for multiple compari-sons but the dimensions of selected features were furtherreduced by principal components analysis Then hierarchi-cal clustering was used to discover clusters of patientsaccording to the principal components Finally to validatethe clustering results we investigated differences in follow-up clinical symptoms and neuroanatomic patterns betweensubgroups of patients and we investigated the extent towhich the analysis could reliably discriminate between dif-ferent subgroups of patients using a pattern classificationapproach

Study setting and patientsThe PPMI (ppmi-infoorg) is a landmark observationallongitudinal database consisting of neuroimaging biologicaltests and clinical and behavioral assessments in gt400 pa-tients with de novo PD23 All patients underwent dopaminetransporter (DAT) imaging and the diagnosis was con-firmed by the DAT deficit All clinical features were reas-sessed annually over 5 years so that markers of diseaseprogression could be discerned The neuroimaging data andextensive longitudinal clinical information provide an un-precedented opportunity to identify the neuroanatomicbiotypes of PD and the longitudinal assessment of PD pro-gression of different biotypes in patients with de novo PDThe clinical and behavioral assessments have been describedextensively elsewhere

Recruitment criteria included age gt30 years PD diagnosiswithin the last 2 years baseline Hoehn and Yahr stage I to IIand no anticipated need for symptomatic treatment within 6months of baseline23 Clinical assessments were performed atbaseline at 3-months interval during the first year of partici-pation and then every 6 months thereafter (see data availablefrom Dryad table e-1 doiorg105061dryadxsj3tx9bf)Data from the PPMI database were obtained in May 2018 incompliance with the PPMI Data Use Agreement

Baseline and clinical assessmentsA comprehensive set of clinical assessments were evaluated inPPMI We focused on clinical features that capture major PDsymptoms including both motor and nonmotor symptomsin line with previous studies372829 Details of clinical as-sessments used in this study are presented in PPMI (ppmi-infoorgwp-contentuploads201004PPMI-General-Op-erations-Manualpdf) Derived variable definitions and scorecalculations are available in the PPMI (in the Study_Docs) Alist of abbreviations of clinical assessments is given in dataavailable from Dryad (table e-2 doiorg105061dryadxsj3tx9bf) Features captured include the following

1 Demographics age sex race symptom duration andeducation level

2 Blood biomarkers biochemical tests3 Motor MDS-UPDRS Parts I through III30 tremorPIGD

motor phenotype PIGD tremor subscale and Schwab-England activities of daily living scores

4 Cognitive testing cognitive function (ageeducationadjusted Montreal Cognitive Assessment total score)31

and neuropsychological variables including visuospatialand visuoperceptual functions (Benton Judgment of LineOrientation)32 cognition performance (Symbol DigitModalities Test)33 verbal learning and memory (HopkinsVerbal Learning Test for total recall delayed recallretention and recognition-discrimination)34 semanticmemory (Semantic Verbal Fluency test)34 and workingmemory capacity (letter-number sequencing)35

5 Autonomic testing autonomic dysfunction (Scales forOutcomes in Parkinsonrsquos DiseasendashAutonomic total scoreand its subscores cardiovascular constipation orofacialthermoregulatory sexual pupillomotor and urinary)36

6 Sleep disorders RBD (RBD Screening Questionnairescore)37 average sleep propensity in daily life (EpworthSleepiness Scale score)38

7 Neurobehavior depression (Geriatric Depression Scalescore)39 trait and state anxiety (State-Trait AnxietyInventory score)29 and impulse control disorders andrelated disorders (Questionnaire for Impulsive-Compulsive Disorders in Parkinsonrsquos Disease score)40

8 Olfactory testing impaired olfaction (agesex adjustedUniversity of Pennsylvania Smell Identification Testscore)41

9 Physical activity Physical Activity Scale of the Elderly42

Three activity categories were assessed leisure householdchores and workvolunteering

CSF and SPECT biomarkersA lumbar puncture was conducted for all participants for thecollection of CSF β-Amyloid1ndash42 phosphorylated tau andtotal tau protein were measured by INNO-BIA AlzBio3 im-munoassay (Innogenetics Inc Ghent Belgium) and α-synu-clein concentration was measured by ELISA SPECT with theDAT tracer 123I-ioflupane was acquired at baseline and follow-up visits23

NeurologyorgN Neurology | Volume 95 Number 11 | September 15 2020 e1447

Imaging processingT1-weighted MRI scan acquisition parameters are detailedelsewhere (ppmi-infoorgwp-contentuploads201706PPMI-MRI-Operations-Manual-V7pdf)

The T1-weighted MRI images were preprocessed with theComputational Anatomy Toolbox (CAT 12) (dbmneurouni-jenadecat12) which is an extension of SPM12 to

provide computational anatomy All these images were cor-rected for bias noise and intensity and linearly and thennonlinearly registered to the Montreal Neurological Institute152-2009c template Then the DBM (ie the determinant ofthe jacobian transformation matrix) maps were calculated toestimate the local volume in each voxel (DBM values) Rawimages of lower quality (CAT image quality rating lt75)were excluded The rest images were further visually checked

Figure 1 Schematic overview of the design of this study

DBM = deformation-based morphometry PCA = principal component analysis UPDRS = Unified Parkinsonrsquos Disease Rating Scale

e1448 Neurology | Volume 95 Number 11 | September 15 2020 NeurologyorgN

Finally the obtained preprocessed volume-based DBM datafrom 314 patients with PD and 145 healthy controls weresmoothed with an 8-mm full width at half-maximum

For volumetric analysis FreeSurfer version 53 was used toderive measures of the volume of subcortical nuclei This is awell-documented and freely available software12

Voxel-level association study and clusteringWe reasoned that a low-dimensional representation of a subsetof neuroanatomic features would best characterize biologicallymeaningful PD biotypes similar to the atrophy subtypesdetected in prodromal Alzheimer disease11 Therefore to selecta set of neuroanatomic features for use in clustering we usedSpearman rank correlation analysis to identify features thatwere significantly correlated (p lt 001) with clinical scores(baseline value) theMDS-UPDRS Parts I to III UPDRS totaltremor and PIGD scores Confounding factors such as age sexyears of education race (categorized as white or other) and siteeffect were regressed out before feature selection

To further exclude undesired background noise principalcomponent analysis was used to extract a lower-dimensionalcomponent space of the selected features (79 principalcomponents were used capturing 90 of the variance) Thenwe used hierarchical clustering to assign participants tonested subgroups with similar pattern We calculated asimilarity matrix describing the correlation distance betweenevery pair of participants and then we used the Ward min-imum variance method to iteratively link pairs of participantsin closest proximity forming progressively larger clusters ina hierarchical tree Calinski-Harabasz criterion values wereused to estimate the optimal number of clusters and theresult suggested 2 clusters as the best choice (see dataavailable from Dryad figure e-2 doiorg105061dryadxsj3tx9bf) Furthermore to validate the clustering we alsoclustered data using k-means clustering The Cohen κagreement rate between hierarchical clustering and k-meansclustering was 068 which is in the substantial range sug-gesting that the patterns identified by the 2 different clus-tering methods were similar (see data available from Dryadfigure e-3)

ClassificationTo further test the clinical relevance of the identified neuro-anatomic features as diagnostic features of biotypes we ap-plied a support vector machine to test how well this coulddiscriminate these 2 biotypes that is classify individuals into 1of these 2 subgroups A 10-fold cross-validation strategy wasused to estimate its accuracy sensitivity and specificity Thedetails of classification are depicted in data available fromDryad (figure e-4 doiorg105061dryadxsj3tx9bf)

Statistical analyses

Demographics and clinical variablesThe t test was used to determine the statistical significance ofcontinuous demographic and clinical variables after the

removal of confounding variables age sex years of educationrace (categorized as white or other) and site effect The χ2 testwas used to test the significance of categorical demographicvariables and phenotype variables Statistical significance wasestablished at p lt 005 (false discovery rate [FDR] correc-tion) and the values were reported as mean (SD) for eachdemographic and clinical variable Missing data were not in-cluded in all analysis

Linear mixed model fitting for disease progressionratesWe estimated rates of progression for each patient with 5years of follow-up Linear mixed models were used to evaluatebaseline and disease progression rates over time in patientsclassified in the subtypes using the lme4 package in R43 Sexage sites race time from baseline (months) biotype andtheir interaction were included as fixed effects Participantintercepts and slopes (rates of progression) were modeled asrandom effects

Data availabilityAll deidentified clinical and imaging data are available on thePPMI website (ppmi-infoorg) and from the correspondingauthor on reasonable request

ResultsBaseline dataset characteristicsA total of 314 early patients with PD were included in thisstudy consisting of 207 (659) men and 107 (341)women On average these patients with PD were 610 plusmn 95years of age with a disease duration (date of enrollment minusthe date of diagnosis) of 69 plusmn 68 months at baseline Themean MDS-UPDRS Parts I through III scores were 56 plusmn 4059 plusmn 42 and 207 plusmn 87 Clinical biological and cognitivecharacteristics of early patients with PD and matched healthycontrols are given in table 1

Neuroanatomic features associated with thesymptoms of PDFigure 2 shows the correlation analysis between deformationvalues and baseline clinical scores We found that a number ofneuroanatomic features correlated with UPDRS scores in-cluding areas spanning the caudate putamen thalamus hip-pocampus supplementary motor area and orbital frontal gyrus(figure 2) These areas are consistent with previous findingsthat subcortical volume loss correlates with motor symptomseverity18 This empirical data-driven approach to feature se-lection identified 23213 voxelDBMvalues that were correlatedwith at least 1 baseline clinical score (figure 2)

Brain neuroanatomic patterns define 2PD biotypesWe then tested whether these neuroanatomic feature setstended to cluster in patient subgroups As illustrated in figure3 and data available from Dryad (figure e-3 doiorg105061dryadxsj3tx9bf) the cluster analysis revealed 2 distinct

NeurologyorgN Neurology | Volume 95 Number 11 | September 15 2020 e1449

Table 1 Demographic clinical and imaging characteristics of controls and PD biotype groups

Demography

Biotype 1 PD Biotype 2 PD t Value p Value

Baseline D (t2 2 t1) Baseline D (t2 2 t1) Baseline D (t2 2 t1) Baseline D (t2 2 t1)

Age at onset y 616 (86) 607 (10) minus0697 0691

Male sex n () 76 (07) 131 (07) 0044 0901

Education history y 156 (28) 155 (32) minus1298 047

White race n () 106 (09) 187 (09) 0044 0901

Symptoms duration mo 73 (75) 67 (63) minus0935 0562

UPDRS total score 3392(139)

237 (208) 3086(126)

139 (175) minus244 minus200 0147 0178

Motor symptoms and signs

UPDRS Part II score 63 (436) 76 (65) 563 (411) 37 (53) minus139 minus306 0455 0029

UPDRS Part III score 2114(894)

107 (144) 2011(863)

71 (118) minus152 minus13 039 038

Tremor score 046 (027) 010 (040) 049 (031) 0 (04) 068 minus046 0692 0876

PIGD score 024 (024) 004 (060) 022 (021) 02 (03) minus111 minus329 0559 002

Schwab-England score 9268(592)

minus111 (13) 9395(546)

minus71 (93) 246 261 0147 0049

TremorPIGD phenotype n () 0347

Tremor dominant 76 (67) 144 (72)

PIGD dominant 17 (15) 36 (18)

Intermediate 21 (18) 20 (10)

Nonmotor symptoms andsigns

UPDRS Part I score 648 (44) 54 (51) 512 (362) 32 (5) minus334 minus150 0017 034

RBD ScreeningQuestionnairescore

507 (299) 093 (285) 369 (23) 089 (308) minus431 minus028 0001 0903

Epworth Sleepiness Scalescore

61 (34) 286 (486) 58 (36) 125 (408) minus086 minus230 0585 0085

Cognitive function

MoCA (adjusted) score 2712 (21) minus141 (466) 2748 (21) minus007 (259) 106 295 0559 0031

Benton Judgment of LineOrientation score

125 (22) minus096 (251) 13 (2) minus045 (209) 219 145 0147 034

HVLT immediate totalrecallscore

247 (54) minus171 (582) 245 (47) 075 (552) minus097 284 0562 0049

HVLT discriminationrecognition score

103 (16) 041 (199) 103 (15) 066 (211) 023 094 0901 0595

HVLT score 086 (019) 0008 (0316) 087 (019) 002 (025) 009 05 0948 0876

Letter-numbersequencingscore

104 (26) minus119 (266) 108 (27) minus053 (255) 063 144 0698 034

Semantic fluencyscore

474 (11) minus15 (1079) 491 (109) 071 (1046) 115 140 054 034

Symbol digit modalitiesscore

407 (91) minus477 (1089) 424 (96) minus1811 (943) 126 189 0473 0204

Continued

e1450 Neurology | Volume 95 Number 11 | September 15 2020 NeurologyorgN

Table 1 Demographic clinical and imaging characteristics of controls and PD biotype groups (continued)

Demography

Biotype 1 PD Biotype 2 PD t Value p Value

Baseline D (t2 2 t1) Baseline D (t2 2 t1) Baseline D (t2 2 t1) Baseline D (t2 2 t1)

SCOPAndashAutonomicquestionnaire score

Gastrointestinal 238 (202) 231 (219) 2 (21) 148 (254) minus132 minus262 047 0049

Urinary question 514 (555) 147 (915) 397 (251) 119 (26) minus231 minus019 0147 0903

Cardiovascular 061 (103) 044 (133) 036 (061) 040 (106) minus261 minus029 0135 0903

Thermoregulatory 124 (154) 097 (187) 109 (13) 066 (162) minus094 minus117 0562 0453

Pupillomotor 045 (072) 021 (087) 039 (068) 024 (067) minus064 019 0698 0903

Sexual 47 (64) 165 (687) 32 (57) 18117 (668) minus225 minus006 0147 095

Total score 145 (103) 727 (1336) 110 (76) 579 (899) minus335 minus099 0017 0591

Geriatric Depression Scalescore

26 (25) 05 (248) 21 (22) 068 (286) minus215 031 0147 0903

STAI score 333 (98) 123 (862) 315 (91) 094 (864) minus184 minus022 0315 0903

STAI state subscore 341 (107) minus0043 (951) 321 (98) minus068 (1052) minus157 minus06 0381 0876

Impulse control disorders(QUIP scores)

GMB 002 (013) 0029 (0168) 001 (007) minus001 (009) minus103 minus151 0559 034

Sex 004 (021) 009 (03) 002 (012) 004 (027) minus16 minus045 0371 0876

Eat 005 (022) 003 (034) 001 (01) 0041 (024) minus222 042 0147 0876

Buy 002 (013) 003 (017) 001 (007) minus001 (009) minus103 minus151 0559 034

Current short 028 (073) 017 (099) 017 (048) 011 (079) minus161 minus014 0371 0917

Other 015 (045) 0 (064) 013 (038) 004 (057) minus034 047 0868 0876

Olfaction UPSIT scorepercentile

213 (880) 23 (130) 128 047

Brain imaging

Right caudate 2 (069) 197 (051) minus033 0868

Left caudate 192 (068) 200 (052) 135 047

Right putamen 085 (039) 083 (034) 010 0948

Left putamen 077 (037) 082 (034) 172 0347

CSF biomarkers

α-Synuclein 143102(56496)

150634(68983)

088 0583

Aβ42 8805(32285)

90831(42908)

042 0848

T-tau 16512(534)

16931(5783)

049 0805

P-tau 1435(476)

15 (545) 079 0628

T-tauAβ42 0203(0095)

0203(0101)

003 0976

P-tauAβ42 0017(001)

0018(001)

039 0859

Continued

NeurologyorgN Neurology | Volume 95 Number 11 | September 15 2020 e1451

clusters of patients with PD with similar disease duration (p =035 table 1) These 2 clusters comprised 3631 (114 pa-tients) and 6369 (200 patients) of the 314 patients withPD Table 1 shows the demographic characteristics of the 2biotype groups There was no significant difference in age sexeducation symptom duration or ratio of PIGD- and tremor-dominant patients between the 2 biotype groups (table 1 anddata available from Dryad table e-3)

Neuroanatomic pattern of the 2 biotypesTo illustrate the neuroanatomic pattern in the different PDbiotypes the DBM values of the 2 biotypes were comparedwith those of controls Compared to normal controls indi-viduals with both biotypes showed significant differences insubcortical regions Figure 4 shows the comparison betweeneach biotype and healthy controls (p lt 0005 FDRcorrection)

Those with biotype 1 had pronounced differences com-pared to normal controls in almost the whole brain in-dicating more severe atrophy in brain areas in early PDThose with biotype 2 had predominant differences in thesubcortical regions In addition we found other distinctpatterns that differentiated the 2 biotypes For examplecompared to controls DBM values in the putamen cau-date pallidum lingual gyrus temporal cortex insulaamygdala hippocampus and orbital frontal cortex whichregulate motor-related cognitive and emotion-relatedbehavior were significantly lower in patients with biotype1 which were characterized in part by increased motor andnonmotor symptom scores In contrast patients withbiotype 2 had significantly higher DBM values in thebrainstem putamen caudate occipital lobe lingual gyrusolfactory posterior cingulate cortex and white matterareas compared to controls Additional volumetric analy-ses with FreeSurfer showed that patients with biotype 1had significantly lower subcortical volumes within thethalamus caudate putamen pallidum accumbens amyg-dala and hippocampus (p lt 005 FDR correction)

compared to patients with biotype 2 (see data availablefrom Dryad table e-4 doiorg105061dryadxsj3tx9bf)

Neuroanatomic patterns for diagnosingPD biotypesWe reasoned that clustering could be used for training theclassifiers for the diagnosis of PD biotypes solely on the basisof structural MRI measures To this end we trained classifiersfor predicting the PD biotype in individual patients Ten-foldcross-validation was used to assess performance and signifi-cance Support vector machine classifiers (using gaussiankernel functions) yielded overall accuracy rates of 841(sensitivity 071 specificity 089 area under the curve 090data available from Dryad figure e-6 doiorg105061dryadxsj3tx9bf) for the clusters characterized above

Baseline differences in symptomsbetween biotypesTable 1 shows that at baseline patients with biotype 1 hadworse mentation behavior and mood (higher MDS-UPDRSPart I score) and very much worse RBD Screening Ques-tionnaire score than those with biotype 2 There was alsoevidence of more severe autonomic function in biotype 1(Scales for Outcomes in Parkinsonrsquos Disease total score)(table 1)

Disease progression in the PD biotypesThe PPMI patients were followed up for 5 years The samplesize of the progression analysis is shown in data available fromDryad (table e-1 doiorg105061dryadxsj3tx9bf) Resultsfrom a linear mixed model showed that the patients withbiotype 1 had significantly greater progression in all MDS-UPDRS scores with the exception of the tremor score (figure5 and table 2) In addition those with biotype 1 tended todevelop more severe cognitive impairment (Hopkins VerbalLearning Test immediate total recall letter-number se-quencing and Symbol Digit Modalities) (table 2 and dataavailable from Dryad figure e- 8) More rapid progressioncould also be seen in the patients with biotype 1 in activities of

Table 1 Demographic clinical and imaging characteristics of controls and PD biotype groups (continued)

Demography

Biotype 1 PD Biotype 2 PD t Value p Value

Baseline D (t2 2 t1) Baseline D (t2 2 t1) Baseline D (t2 2 t1) Baseline D (t2 2 t1)

P-tauT-tau 008 (001) 009 (001) 093 0562

Aβ42α-synuclein 065 (019) 064 (023) minus030 0878

T-tauα-synuclein 012 (003) 012 (003) minus096 0562

P-tauα-synuclein 001(0003)

001(0002)

minus018 0905

Abbreviations Aβ = β-amyloid HVLT = Hopkins Verbal Learning Test MoCA = Montreal Cognitive Assessment P-tau = phosphorylated tau PD = Parkinsondisease PIGD = postural instability and gait disorder QUIP = Questionnaire for Impulsive-Compulsive Disorders in Parkinsonrsquos Disease RBD = REM sleepbehavior disorder SCOPA = Scales for Outcomes in Parkinsonrsquos Disease STAI = State-Trait Anxiety Inventory T-tau = total tau UPDRS = Unified ParkinsonrsquosDisease Rating Scale UPSIT = University of Pennsylvania Smell Identification TestThe t values and p values are for the comparison between patients with biotype 1 and biotype 2 PD Baseline scores are presented as well as the change overfollow-up in the case of clinical measures given as D (t2 minus t1) Here t1 is score at baseline and t2 is the score after 5 years follow-up All values in parenthesesare SDs unless otherwise specified The p values are false discovery rate corrected for multiple comparisons Significant p lt 005

e1452 Neurology | Volume 95 Number 11 | September 15 2020 NeurologyorgN

daily living (Schwab-England activities of daily living) (table 2and data available from Dryad figure e-7)

On dopaminergic SPECT scanning patients with biotype1 had worse denervation of both left caudate and left

putamen after an average of 4 years of follow-up The rightcaudate and right putamen showed no significant differ-ence in denervation between the 2 biotypes (see dataavailable from Dryad figure e-13 doiorg105061dryadxsj3tx9bf)

Figure 2 Correlation between DBM values and symptom scores

Warm color indicates a positive correlation between symptom scores and deformation-based morphometry (DBM) values cold color indicates a negativecorrelation between symptom scores and DBM values (A) Correlation between DBM values and Movement Disorder Societyndashsponsored revision of theUnified Parkinsonrsquos Disease Rating Scale (MDS-UPDRS) Part I score (B) Correlation between DBM values and MDS-UPDRS Part II score (C) Correlationbetween DBM values and MDS-UPDRS III score (D) Correlation between DBM values and MDS-UPDRS total score (E) Correlation between DBM values andtremor score (F) Correlation between DBM values and postural instability and gait disorder (PIGD) score In total 8855 2826 4201 6064 6994 and 1380deformation values correlated with UPDRS Part I UPDRS Part II UPDRS Part III total UPDRS PIGD and tremor scores respectively (p lt 001 uncorrected)

NeurologyorgN Neurology | Volume 95 Number 11 | September 15 2020 e1453

To summarize patients with biotype 1 had higher baselineMDS-UPDRS Part I score worse baseline sleep problems andautonomic dysfunction and faster progression of most motorsymptoms cognitive impairment and activities of daily livingcompared to patients with biotype 2 (all p lt 005 FDR cor-rection) (figure 5 and tables 1 and 2)

DiscussionIn this study we identified 2 neuroanatomic biotypes in pa-tients with PD with otherwise similar demographics using anunbiased data-driven clustering approach applied to the PPMIcohort Neuroanatomic biotypes differed in symptomatologyeven at presentation before treatment and thereafter pro-gressed at different rates Themost striking baseline differencewas the much higher RBD Screening Questionnaire score inbiotype 1

Our results provide support for 2 different neuroanatomicphenotypes within patients with PD Compared to healthy

controls the neuroanatomic differences were more wide-spread in patients with biotype 1 spanning across almostall of the brain Compared to patients with biotype 1 pa-tients with biotype 2 had less widespread differences atbaseline

Patients with biotype 2 had larger subcortical volume(higher DBM values) than controls on average suggestingthey may contain more cells (including dopaminergic cells)and synapses increasing the ability to support maintenanceof function despite declines in brain volume Accordinglypatients with biotype 2 had a slower disease progressionrate3 In contrast patients with biotype 1 had less brainreserve and did not compensate as well as those with biotype1 as PD progressed over time resulting in a faster diseaseprogression rate Indeed there was evidence that this wasalready the case at the time of presentation given the worsesymptom severity in several domains in biotype 1 than inbiotype 2 In summary we hypothesize that the differentrates of symptom progression relate to different brainreserves1316

Figure 3 Dendrogram of the final hierarchical cluster solution of patients with PD in the PPMI patients

Matrix under the dendrogramshows the distance ofthe neuroanatomic patterns between different pa-tients with Parkinson disease (PD) Each element (irow and j column) in the matrix indicates the dif-ference of the neuroanatomic patterns between thei participant and j participant PPMI = ParkinsonrsquosProgression Markers Initiative

e1454 Neurology | Volume 95 Number 11 | September 15 2020 NeurologyorgN

Brain reserve describes the differences in brain volume andstructure that may support maintenance of function despitepathology1344 Gross or regional brain volume reflects thequantity of neurons neuronal integrity and synaptic densitieswhich determine the ability of the brain to engage in compen-satory activity1344 Prior works have highlighted a link betweenbrain volume and markers of functional reserve in patients withPD and other neurodegenerative diseases14ndash16 For examplethere is a relationship between brain gray matter volume andthe magnitude of network-level integration16 Basal forebrain

volume can predict future psychosis in early PD and highercholinergic nucleus 4 gray matter density is associated with alower risk of reporting psychotic symptoms17 PDwith cognitiveimpairment shows lower gray matter volume in the nucleusbasalis of Meynert20 Compared to tremor-dominant patientspatients with PIGD had lower gray matter volumes in the globuspallidus and amygdala and have worse prognosis with a morerapid decline4546 In line with these observations our resultssuggest that a larger subcortical volume helps limit the nega-tive impact of PD pathology during disease progression as

Figure 4 Different neuroanatomic patterns in the 2 PD biotypes compared with healthy controls

Warm color indicates higher deformation-basedmorphometry (DBM) values in patients with Parkinson disease (PD) cold color indicates lower DBMvalues (plt 0005 false discovery rate correction) (A) The t values of comparison of DBM values between biotype 1 and controls (B) The t values of comparison of DBMvalues between biotype 2 and controls

NeurologyorgN Neurology | Volume 95 Number 11 | September 15 2020 e1455

Figure 5 Longitudinal changes in outcomes of interest in different biotypes of PD among the PPMI patients with 5 years offollow-up

Asterisks indicate the statistical significance of the comparison between 2 biotypes in the clinical variables at the time of different follow up (AndashF p lt 005 plt 001 p lt 0005 uncorrected for multiple comparisons) The Parkinsonrsquos Progression Markers Initiative (PPMI) data contain 1 baseline set of data and 12follow-up sets of data over 5 years PD = Parkinson disease PIGD = postural instability and gait disorder UPDRS = Unified Parkinsonrsquos Disease Rating Scale

e1456 Neurology | Volume 95 Number 11 | September 15 2020 NeurologyorgN

Table 2 Longitudinal motor and nonmotor score estimated β coefficients in different biotypes of PD with covariatecorrection in the PPMI patients with 5 years of follow-up

Biotype 1 Biotype 2 t Value p Value

Total UPDRS score 532 (039) 34 (03) 392 lt0001

Motor symptoms and signs

UPDRS Part II score 149 (012) 081 (009) 45 lt0001

UPDRS Part III score 26 (025) 173 (019) 276 0025

Tremor score 0022 (0008) 0012 (0006) 099 0485

PIGD score 0111 (0011) 0051 (0009) 42 0001

Schwab-England score minus255 (024) minus166 (018) minus293 0022

Nonmotor symptoms and signs

UPDRS Part I score 122 (01) 084 (008) 291 0022

RBD Screening Questionnaire score 0254 (0057) 0225 (0043) 042 0717

Epworth Sleepiness Scale score 0572 (0087) 0384 (0066) 171 0223

Cognitive function

MOCA (adjusted) score minus0268 (007) minus009 (0053) minus202 0121

Benton Judgment of Line Orientationscore

minus0097 (0042) minus0023 (0032) minus139 0318

HVLT immediate total recall score minus0255 (0101) 011 (0077) minus286 0024

HVLT discrimination recognitionscore

0061 (0041) 009 (0031) minus058 0662

HVLT score minus0013 (0005) 0001 (0004) minus209 0114

Letter-number sequencing score minus0258 (0053) minus009 (004) minus254 0044

Semantic fluency score minus0399 (0198) 0016 (015) minus167 0226

Symbol digit modalities score minus1228 (0199) minus04 (015) minus332 0008

SCOPAndashAutonomic questionnairescore

Gastrointestinal 0409 (0048) 028 (0036) 215 0106

Urinary question 0271 (0087) 0217 (0065) 049 0708

Cardiovascular 0096 (0021) 0085 (0016) 043 0717

Thermoregulatory 018 (0035) 0133 (0027) 107 0485

Pupillomotor 0023 (0015) 0042 (0011) minus103 0485

Sexual 0302 (0128) 0431 (0097) minus08 0582

Total score 1269 (019) 1194 (0143) 032 0776

Geriatric Depression Scalescore

017 (0055) 0112 (0041) 085 0572

STAI score 0249 (0343) 0305 (0259) minus013 0897

STAI state subscore minus0041 (0191) 0098 (0144) minus058 0662

QUIP scores

GMB 0005 (0003) 0 (0002) 144 0314

Sex 0012 (0005) 0008 (0004) 067 0662

Continued

NeurologyorgN Neurology | Volume 95 Number 11 | September 15 2020 e1457

represented by brain atrophy Physical exercise has been shownto increase brain volume in older adults47 Therefore interven-tions that increase physical activity before or in the early courseof PDmay contribute to brain reserve and help slow the rates ofdisease progression

At baseline patients with biotype 1 had evidence of worsebehavioral autonomic andmotor impairment and above all ofworse RBD symptomatology These findings are consistentwith past observations showing that motor dysfunction is as-sociated with cognitive decline autonomic dysfunction andRBD2848 PD with RBD is also associated with faster motorprogression and a higher risk of cognitive decline28 In line withthese studies we found that patients with biotype 1 had muchworse RBD symptomatology and more cognitive decline RBDmay be a useful marker for early subtyping of PD at baseline28

Patients with biotype 1 had significantly higher scores in severalmotor disease symptoms after only 1 year of follow-up butthere was no difference in the progression of tremor betweenthe 2 biotypes This may reflect that rest tremor may be moreclosely related to degeneration of nondopaminergic ratherthan dopaminergic systems49 In most cases there is a sub-stantial asymmetry of clinical symptoms from disease onsetand patients with unilateral disease showed a significant dif-ference in striatal uptake between the ipsilateral and contra-lateral sides in both the caudate and putamen nuclei49

Differences in longitudinal denervation between the left andright caudate and putamen between the 2 biotypes (see dataavailable from Dryad figure e-12 doiorg105061dryadxsj3tx9bf)may reflect different disease severity Consistent withprevious findings that CSF biomarkers are not useful bio-markers of PD progression50 our results did not show signif-icant differences in CSF biomarker levels between the 2biotypes at baseline (table 1)

While the 2 identified neuroanatomic biotypes showed groupdifferences in terms of symptom severity and longitudinalprogression there is overlap in the neuroanatomic features atthe individual level between the 2 biotypes It is plausible thatsome patients were a combination of gt1 biotype and this

would not be captured by our approach of discretizing bio-types Future research should further investigate more refinedbiotype definitions based on continuous membership proba-bility values through longitudinal studies in larger cohortsTherefore we regard the 2 biotypes identified here as just aninitial solution to the problem of diagnostic heterogeneity in asubtyping process that relies primarily on neuroanatomicfeatures correlated with clinical scores It is likely that cohortlimitations (the PPMI patients on the whole have a higherlevel of education are younger and have less baseline dis-ability than the general PD population23) the size of ourcluster-discovery dataset and the subjectivity of clinical-symptom assessments were also limiting factors For thesereasons a novel cohort with longitudinal clinical data will beuseful for validating the present findings

We show that neuroanatomic biotypes can be defined thatrobustly predict different rates of progression suggesting thatthese reflect true biotypes of PD Given that PPMI recruitedearly patients from multiple sites our results should still bemostly generalizable to early PD in real clinical practice wherethe findings can be used to inform estimates of prognosisThese results might also have implications for clinical trialdesign in early PD in the future The existence of neuroana-tomic biotypes that show specific trajectories of clinical scoredecline may require biotype-specific outcome measures tai-lored to the expected rate of decline in different domains

We have robustly identified 2 different neuroanatomic bio-types among patients with early PD using a data-drivenclustering approach These biotypes showed distinct neuro-anatomic patterns symptoms and rates of progression Rec-ognition of this heterogeneity is an important step towardprecision medicine for PD

Study fundingThis study is sponsored by the Michael J Fox Foundation forParkinsonrsquos Research J Feng is supported by the 111 Project(No B18015) the key project of Shanghai Science amp Technol-ogy (No 16JC1420402) National Key RampD Program of China

Table 2 Longitudinal motor and nonmotor score estimated β coefficients in different biotypes of PD with covariatecorrection in the PPMI patients with 5 years of follow-up (continued)

Biotype 1 Biotype 2 t Value p Value

Eat 0006 (0005) 001 (0004) minus064 0662

Buy 0005 (0003) 0 (0002) 144 0314

Other 0027 (001) 0014 (0008) 1 0485

Current short 0056 (0016) 0032 (0012) 117 0446

Abbreviations GMB = Gambling HVLT = Hopkins Verbal Learning Test PD = Parkinson disease PIGD = postural instability and gait disorder PPMI =Parkinsonrsquos Progression Markers Initiative QUIP = Questionnaire for Impulsive-Compulsive Disorders in Parkinsonrsquos Disease RBD = REM sleep behaviordisorder SCOPA = Scales for Outcomes in Parkinsonrsquos Disease STAI = State-Trait Anxiety Inventory UPDRS = Unified Parkinsonrsquos Disease Rating ScaleAnnual changes were estimated from linear mixedmodels Age sex race and sites were included as covariates Values in parentheses are SDs The p valuesare false discovery rate corrected for multiple comparisons Significant p lt 005

e1458 Neurology | Volume 95 Number 11 | September 15 2020 NeurologyorgN

(No 2018YFC1312900) National Natural Science Foundationof China (NSFC 91630314) Shanghai Municipal Science andTechnology Major Project (No 2018SHZDZX01) and ZJLabW Cheng is supported by grants from the National NaturalSciences Foundation of China (No 81701773 11771010) andsponsored by Shanghai Sailing Program (No 17YF1426200)WCheng is also sponsored by Natural Science Foundation ofShanghai (No 18ZR1404400) Dr Brown is supported by theMedical Research Council [MC_UU_120241] and the Na-tional Institute for Health Research (NIHR) Oxford BiomedicalResearchCentre (BRC) J Zhang is supported by grants from theNational Natural Science Foundation of China (No 61573107)and sponsored by the Natural Science Foundation of Shanghai(No 17ZR1444200) J Wang is sponsored by the Ministry ofScience and Technology of China (grants 2016YFC13065002016YFC1306504) and National Nature Science Foundation ofChina (grant 81771372) F Liu is sponsored by National NatureScience Foundation of China (grant 81701250)

DisclosureThe authors report no disclosures relevant to the manuscriptGo to NeurologyorgN for full disclosures

Publication historyReceived by Neurology November 6 2019 Accepted in final formMarch 18 2020

References1 Schapira AHV Chaudhuri KR Jenner P Non-motor features of Parkinson disease

Nat Rev Neurosci 201718435ndash4502 Fereshtehnejad SM Postuma RB Subtypes of Parkinsonrsquos disease what do they tell

us about disease progression Curr Neurol Neurosci Rep 201717343 Fereshtehnejad SM Zeighami Y Dagher A Postuma RB Clinical criteria for sub-

typing Parkinsonrsquos disease biomarkers and longitudinal progression Brain 20171401959ndash1976

4 Fereshtehnejad SM Romenets SR Anang JBM Latreille V Gagnon JF Postuma RBNew clinical subtypes of Parkinson disease and their longitudinal progression JAMANeurol 201572863ndash873

5 Lewis SJG Foltynie T Blackwell AD Robbins TW Owen AM Barker RA Het-erogeneity of Parkinsonrsquos disease in the early clinical stages using a data drivenapproach J Neurol Neurosurg Psychiatry 200576343ndash348

6 Xu X Chan P Clinical subtypes of Parkinsonrsquos disease based on nonmotor symptomsa cluster analysis Mov Disord 201833S725

7 Zhang X Chou J Liang J et al Data-driven subtyping of Parkinsonrsquos disease usinglongitudinal clinical records a cohort study Sci Rep 20199797

8 De Pablo-Fernandez E Lees AJ Holton JL Warner TT Prognosis and neuropath-ologic correlation of clinical subtypes of Parkinson disease JAMA Neurol 201976470ndash479

9 Erro R Picillo M Vitale C et al Clinical clusters and dopaminergic dysfunction in de-novo Parkinson disease Parkinsonism Relat Disord 201628137ndash140

10 Mestre TA Eberly S Tanner C et al Reproducibility of data-driven Parkinsonrsquosdisease subtypes for clinical research Parkinsonism Relat Disord 201856102ndash106

11 Ten Kate M Dicks E Visser PJ et al Atrophy subtypes in prodromal Alzheimerrsquosdisease are associated with cognitive decline Brain 20181413443ndash3456

12 Fischl B Salat DH Busa E et al Whole brain segmentation automated labeling ofneuroanatomical structures in the human brain Neuron 200233341ndash355

13 Gregory S Long JD Kloppel S et al Operationalizing compensation over time inneurodegenerative disease Brain 20171401158ndash1165

14 Guo LH Alexopoulos P Wagenpfeil S Kurz A Perneczky R Alzheimerrsquos DiseaseNeuroimaging Initiative Brain size and the compensation of Alzheimerrsquos diseasesymptoms a longitudinal cohort study Alzheimers Dement 20139580ndash586

15 Gregory S Long JD Kloppel S et al Testing a longitudinal compensation model inpremanifest Huntingtonrsquos disease Brain 20181412156ndash2166

16 Shine JM Bell PT Matar E et al Dopamine depletion alters macroscopic networkdynamics in Parkinsonrsquos disease Brain 20191421024ndash1034

17 Barrett MJ Blair JC Sperling SA Smolkin ME Druzgal TJ Baseline symptoms andbasal forebrain volume predict future psychosis in early Parkinson disease Neurology201890e1618ndashe1626

18 Wilson H Niccolini F Pellicano C Politis M Cortical thinning across Parkinsonrsquosdisease stages and clinical correlates J Neurol Sci 201939831ndash38

19 Borghammer P Oslashstergaard K Cumming P et al A deformation-based morphometrystudy of patients with early-stage Parkinsonrsquos disease Eur J Neurol 201017314ndash320

20 Schulz J Pagano G Fernandez Bonfante JA Wilson H Politis M Nucleus basalis ofMeynert degeneration precedes and predicts cognitive impairment in Parkinsonrsquosdisease Brain 20181411501ndash1516

21 Zeighami Y Fereshtehnejad SM Dadar M et al A clinical-anatomical signature ofParkinsonrsquos disease identified with partial least squares and magnetic resonance im-aging Neuroimage 201919069ndash78

22 Uribe C Segura B Baggio HC et al Progression of Parkinsonrsquos disease patientsrsquosubtypes based on cortical thinning 4-year follow-up Parkinsonism Relat Disord201964286ndash292

23 Parkinson ProgressionMarker Initiative The Parkinson ProgressionMarker Initiative(PPMI) Prog Neurobiol 201195629ndash635

Appendix Authors

Name Location Contribution

Linbo WangPhD

FudanUniversityShanghai China

Data curation software formalanalysis investigation visualizationmethodology writingmdashoriginaldraft writingndashreview and editing

Wei ChengPhD

FudanUniversityShanghai China

Resources data curationconceptualization formal analysissupervision investigationmethodology projectadministration writingndashreview andediting

Edmund TRolls DPhilDSc

University ofWarwick UK

Conceptualization interpreted thedata methodology revised themanuscript for intellectual content

Fuli Dai MSc FudanUniversityShanghai China

Resources data curationmethodology

WeikangGong PhD

FudanUniversityShanghai China

Resources data curation softwaremethodology

Jingnan DuMSc

FudanUniversityShanghai China

Resources data curation softwaremethodology

Wei ZhangBSc

FudanUniversityShanghai China

Data curation softwaremethodology

ShouyanWang PhD

FudanUniversityShanghai China

Interpreted the data revised themanuscript for intellectual content

Appendix (continued)

Name Location Contribution

Fengtao LiuMD

FudanUniversityShanghai China

Interpreted the data revised themanuscript for intellectual content

Jian WangMD

FudanUniversityShanghai China

Interpreted the data revised themanuscript for intellectual content

Peter BrownMD

University ofOxford UK

Supervision investigationmethodology projectadministration writingndashreview andediting

JianfengFeng PhD

FudanUniversityShanghai China

Conceptualization resourcessupervision funding acquisitioninvestigation methodology projectadministration writingndashreview andediting

NeurologyorgN Neurology | Volume 95 Number 11 | September 15 2020 e1459

24 Chung MK Worsley KJ Paus T et al A unified statistical approach to deformation-based morphometry Neuroimage 200114595ndash606

25 Lanskey JH McColgan P Schrag AE et al Can neuroimaging predict dementia inParkinsonrsquos disease Brain 20181412545ndash2560

26 Pandya S Zeighami Y Freeze B et al Predictive model of spread of Parkinsonrsquospathology using network diffusion Neuroimage 2019192178ndash194

27 Scanlon C Mueller SG Tosun D et al Impact of methodologic choice for automaticdetection of different aspects of brain atrophy by using temporal lobe epilepsy as amodel AJNR Am J Neuroradiol 2011321669ndash1676

28 Pagano G De Micco R Yousaf T Wilson H Chandra A Politis M REM behaviordisorder predicts motor progression and cognitive decline in Parkinson diseaseNeurology 201891e894ndashe905

29 Kendall PC Finch AJ Jr Auerbach SM Hooke JF Mikulka PJ The state-trait anxietyinventory a systematic evaluation J Consult Clin Psychol 197644406ndash412

30 Goetz CG Tilley BC Shaftman SR et al Movement Disorder Society-sponsoredrevision of the Unified Parkinsonrsquos Disease Rating Scale (MDS-UPDRS) scale pre-sentation and clinimetric testing results Mov Disord 2008232129ndash2170

31 Nasreddine ZS Phillips NA Bedirian V et al The Montreal Cognitive AssessmentMoCA a brief screening tool for mild cognitive impairment J Am Geriatr Soc 200553695ndash699

32 Benton AL Varney NR Hamsher KD Visuospatial judgment a clinical test ArchNeurol 197835364ndash367

33 Forn C Belloch V Bustamante JC et al A symbol digit modalities test version suitablefor functional MRI studies Neurosci Lett 200945611ndash14

34 Shapiro AM Benedict RH Schretlen D Brandt J Construct and concurrent validity ofthe Hopkins Verbal Learning Test-Revised Clin Neuropsychol 199913348ndash358

35 Christensen BK Girard TA Bagby RM Wechsler Adult Intelligence Scale-ThirdEdition short form for index and IQ scores in a psychiatric population Psychol Assess200719236ndash240

36 Visser M Marinus J Stiggelbout AM Van Hilten JJ Assessment of autonomic dys-function in Parkinsonrsquos disease the SCOPA-AUT Mov Disord 2004191306ndash1312

37 Stiasny-Kolster K Mayer G Schafer S Moller JC Heinzel-Gutenbrunner M OertelWH The REM Sleep Behavior Disorder Screening Questionnaire a new diagnosticinstrument Mov Disord 2007222386ndash2393

38 Johns MW A new method for measuring daytime sleepiness the Epworth SleepinessScale Sleep 199114540ndash545

39 Yesavage JA Geriatric Depression Scale Psychopharmacol Bull 198824709ndash71140 Weintraub D Hoops S Shea JA et al Validation of the questionnaire for impulsive-

compulsive disorders in Parkinsonrsquos disease Mov Disord 2009241461ndash146741 Doty RL Shaman P Kimmelman CP Dann MS University of Pennsylvania Smell

Identification Test a rapid quantitative olfactory function test for the clinic Laryn-goscope 198494176ndash178

42 Washburn RA Smith KW Jette AM Janney CA The Physical Activity Scale for theElderly (PASE) development and evaluation J Clin Epidemiol 199346153ndash162

43 Lamprianou I Application of single-level and multi-level Rasch models using the lme4package J Appl Meas 20131479ndash90

44 Gregory S Long JD Tabrizi SJ Rees G Measuring compensation in neuro-degeneration using MRI Curr Opin Neurol 201730380ndash387

45 Rosenberg-Katz K Herman T Jacob Y Kliper E Giladi N Hausdorff JM Subcorticalvolumes differ in Parkinsonrsquos disease motor subtypes new insights into the patho-physiology of disparate symptoms Front Hum Neurosci 201610356

46 Rosenberg-Katz K Herman T Jacob Y Giladi N Hendler T Hausdorff JM Graymatter atrophy distinguishes between Parkinson disease motor subtypes Neurology2013801476ndash1484

47 Erickson KI Voss MW Prakash RS et al Exercise training increases size of hippo-campus and improves memory Proc Natl Acad Sci USA 20111083017ndash3022

48 Wang YX Zhao J Li DK et al Associations between cognitive impairment and motordysfunction in Parkinsonrsquos disease Brain Behav 20177e00719

49 Djaldetti R Ziv I Melamed E The mystery of motor asymmetry in Parkinsonrsquosdisease Lancet Neurol 20065796ndash802

50 Parnetti L Gaetani L Eusebi P et al CSF and blood biomarkers for Parkinsonrsquosdisease Lancet Neurol 201918573ndash586

e1460 Neurology | Volume 95 Number 11 | September 15 2020 NeurologyorgN

DOI 101212WNL0000000000010498202095e1445-e1460 Published Online before print August 14 2020Neurology

Linbo Wang Wei Cheng Edmund T Rolls et al progression

Association of specific biotypes in patients with Parkinson disease and disease

This information is current as of August 14 2020

ServicesUpdated Information amp

httpnneurologyorgcontent9511e1445fullincluding high resolution figures can be found at

References httpnneurologyorgcontent9511e1445fullref-list-1

This article cites 50 articles 6 of which you can access for free at

Citations httpnneurologyorgcontent9511e1445fullotherarticles

This article has been cited by 1 HighWire-hosted articles

Subspecialty Collections

httpnneurologyorgcgicollectionvolumetric_mriVolumetric MRI

httpnneurologyorgcgicollectionprognosisPrognosis

httpnneurologyorgcgicollectionparkinsons_disease_parkinsonismParkinsons diseaseParkinsonism

httpnneurologyorgcgicollectioncohort_studiesCohort studies

httpnneurologyorgcgicollectionbasal_gangliaBasal gangliafollowing collection(s) This article along with others on similar topics appears in the

Permissions amp Licensing

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

Reprints

httpnneurologyorgsubscribersadvertiseInformation about ordering reprints can be found online

ISSN 0028-3878 Online ISSN 1526-632XWolters Kluwer Health Inc on behalf of the American Academy of Neurology All rights reserved Print1951 it is now a weekly with 48 issues per year Copyright Copyright copy 2020 The Author(s) Published by

reg is the official journal of the American Academy of Neurology Published continuously sinceNeurology