ARTICLE OPEN ACCESS

Associations between tau Aβ and corticalthickness with cognition in Alzheimer diseaseRik Ossenkoppele PhD Ruben Smith MD PhD Tomas Ohlsson PhD Olof Strandberg PhD

Niklas Mattsson MD PhD Philip S Insel MSc Sebastian Palmqvist MD PhD and Oskar Hansson MD PhD

Neurologyreg 201992e601-e612 doi101212WNL0000000000006875

Correspondence

Dr Ossenkoppele

rossenkoppelevumcnl

or Dr Hansson

oskarhanssonmedluse

AbstractObjectiveTo examine the cross-sectional associations between regional tau β-amyloid (Aβ) and corticalthickness and neuropsychological function across the preclinical and clinical spectrum ofAlzheimer disease (AD)

MethodsWe included 106 participants from the Swedish Biomarkers for Identifying NeurodegenerativeDisorders Early and Reliably (BioFINDER) study of whom 33 had preclinical AD (Aβ-positivecognitively normal individuals) 25 had prodromal AD (Aβ-positive mild cognitive impair-ment) and 48 had probable AD dementia All underwent [18F]flortaucipir (tau) and structuralMRI (cortical thickness) and 88 of 106 underwent [18F]flutemetamol (Aβ) PET Linearregression models adjusted for age sex and education were performed to examine associationsbetween 7 regions of interest and 7 neuropsychological tests for all 3 imaging modalities

ResultsIn preclinical AD [18F]flortaucipir but not [18F]flutemetamol or cortical thickness was as-sociated with decreased global cognition memory and processing speed (range standardizedβ = 035ndash052 p lt 005 uncorrected for multiple comparisons) In the combined prodromal ADand AD dementia group both increased [18F]flortaucipir uptake and reduced cortical thicknesswere associated with worse performance on a variety of neuropsychological tests (most regionsof interest survived correction for multiple comparisons at p lt 005) while increased [18F]flutemetamol uptake was specifically associated with lower scores on a delayed recall memorytask (p lt 005 uncorrected for multiple comparisons) The strongest effects for both [18F]flortaucipir and cortical thickness on cognition were found in the lateral and medial parietalcortex and lateral temporal cortex The effect of [18F]flutemetamol on cognition was generallyweaker and less region specific

ConclusionOur findings suggest that tau PET is more sensitive than Aβ PET and measures of corticalthickness for detecting early cognitive changes in preclinical AD Furthermore both [18F]flortaucipir PET and cortical thickness show strong cognitive correlates at the clinical stages ofAD

From the Clinical Memory Research Unit (RO RS OS NM PSI SP OH) Lund University Sweden Department of Neurology and Alzheimer Center (RO) VU UniversityMedical Center Amsterdam Neuroscience the Netherlands Department of Radiation Physics (TO) SkaringneUniversity Hospital LundMemory Clinic (OH) SkaringneUniversity HospitalMalmo Sweden and Center for Imaging of Neurodegenerative Diseases (PSI) Department of Veterans Affairs Medical Center San Francisco CA

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 the Swedish Alzheimerrsquos Foundation

This is an open access article distributed under the terms of the Creative Commons Attribution-NonCommercial-NoDerivatives License 40 (CC BY-NC-ND) which permits downloadingand sharing the work provided it is properly cited The work cannot be changed in any way or used commercially without permission from the journal

Copyright copy 2019 The Author(s) Published by Wolters Kluwer Health Inc on behalf of the American Academy of Neurology e601

Alzheimer disease (AD) has a long-lasting preclinical stage inwhich pathogenic proteins (β-amyloid [Aβ] and tau) aggre-gate and subtle structural and functional brain changesemerge without the co-occurrence of significant cognitiveimpairment1ndash3 The accumulation of Aβ is widespread rela-tively early in the disease course while there is an ongoingaccumulation of tau pathology and synaptic and neuronal losswhich corresponds to the clinical progression of the disease4

Advances in the field of neuroimaging now allow in vivowhole-brain mapping of these pathophysiologic (Aβ and taudeposition) and neurodegenerative (atrophy) processes en-abling detailed examination of their relationships with cog-nitive function In line with the neuropathology literature5

PET and MRI studies have shown intimate links betweencognitive performance and both tau tracer retention and brainatrophy in AD-specific regions6ndash9 Although some variabilityexists across studies especially cross-sectionally the cognitivecorrelates for Aβ pathology are generally weaker and lessregion specific10ndash12

Identifying the contribution of the neurobiological hallmarksof AD to cognitive changes may inform design participantselection and monitoring of clinical trials In the presentstudy we performed [18F]flortaucipir PET (tau) [18F]flute-metamol PET (Aβ) structural MRI (cortical thickness) anda neuropsychological test battery in 106 Aβ-positive patientswith normal cognition mild cognitive impairment (MCI) orAD dementia We aimed to assess to what degree these im-aging modalities were related to cognition and whether theserelationships differed by disease stage (ie preclinical orclinical) specific neuropsychological function or region ofinterest (ROI)

MethodsParticipantsFor this cross-sectional study participants were recruitedfrom the Swedish Biomarkers for Identifying Neurodegen-erative Disorders Early and Reliably study (BioFINDERbiofinderse)13 BioFINDER is a prospective study that fo-cuses on identifying key mechanisms and improvement ofdiagnostics in AD and other neurodegenerative disorders Allparticipants underwent [18F]flortaucipir PET [18F]fluteme-tamol PET (Aβ) and structural MRI between June 2014 andNovember 2017 During this period the majority of patientsvisiting the Memory Clinic and Clinical Memory ResearchUnit in Malmo Sweden were invited to participate in the

study All underwent a medical history and complete neuro-logic examination brainMRI and neuropsychological testingWe included 33 Aβ-positive cognitively normal individuals(called preclinical AD3) 25 Aβ-positive patients with MCI(called prodromal AD1415) and 48 Aβ-positive patients withprobable AD dementia16 The preclinical AD groups con-sisted of both cognitively normal research volunteers (n = 17)and patients with subjective cognitive decline17 referred to theMemory Clinic of Skaringne University Hospital in Sweden (n =8) In line with the ldquosyndromal staging of the cognitive con-tinuumrdquo in the recently proposed research framework by theNational Institute on Aging and the Alzheimerrsquos Associa-tion18 cognitively normal participants had a Clinical De-mentia Rating score of 0 a Mini-Mental State Examination(MMSE) score between 27 and 30 and no history of cogni-tive change over time All participants underwent [18F]flor-taucipir PET and structural MRI while [18F]flutemetamolPET was available in 86 of 106 (83) participants A com-parison between the baseline characteristics of the full Bio-FINDER study sample and the subsample with [18F]flortaucipir PET and a flow diagram showing participant se-lection data are available fromDryad (table 1 and figure 1 doiorg105061dryadht92839)

Standard protocol approvals registrationsand patient consentsThe regional ethics committee at Lund University the radi-ation protection committee at Skaringne University Hospital andthe Swedish Medical Products Agency approved the studyand informed consent was obtained from all participants ortheir assigned surrogate decision makers

CSF biomarkersAβ status was determined with lumbar CSF sampling per-formed following the Alzheimerrsquos Association flowchart19

Samples were stored in 1-mL polypropylene tubes at minus80degCuntil analysis The samples were analyzed with commerciallyavailable ELISAs (INNOTEST FujirebioInnogeneticsGhent Belgium) to determine the levels of total tau Aβ42and phosphorylated tau Aβ positivity was defined as CSFAβ42 lt650 ngL

13 Board-certified laboratory technicians whowere blinded to clinical data and diagnoses performed allanalyses

Neuropsychological test batteryParticipants underwent a neuropsychological test batterycovering major cognitive functions including global cognition(MMSE) episodic memory (immediate and delayed word listrecall tests from the Alzheimerrsquos Disease Assessment Scale

GlossaryAβ = β-amyloid AD = Alzheimer disease ADAS = Alzheimerrsquos Disease Assessment Scale AQT = A Quick Test for CognitiveSpeed BioFINDER = Biomarkers for Identifying Neurodegenerative Disorders Early and Reliably FDR = false discovery rateMCI = mild cognitive impairment MMSE = Mini-Mental State Examination PVE = partial volume effects ROI = region ofinterest SUVR = standardized uptake value ratio TMT-A = Trail Making Test A

e602 Neurology | Volume 92 Number 6 | February 5 2019 NeurologyorgN

[ADAS] Cognitive Subscale) semantic memory (category[animal] fluency) language (naming condition of ADAS)and processing speed and attention (Trail Making Test A[TMT-A] A Quick Test for Cognitive Speed [AQT] formand color naming) Scores on ADAS tests were inverted sothat higher values indicate better cognitive performance ForTMT-A and AQT scores represent time needed to completethe task thus higher values indicate worse performanceTMT-A and AQT were log transformed to account for theirnonnormal distribution All participants had an MMSEscore but scores were missing for ADAS immediate recall(n = 6) ADAS delayed recall (n = 8) ADAS naming (n =10) TMT-A (n = 9) animal fluency (n = 12) and AQT (n =14) Missing data were imputed with the fully conditionalspecification single-imputation method20 This is an itera-tive Markov chain Monte Carlo method implemented inSPSS software (version 21) that can be used for arbitrarypatterns of missing data For each iteration (we allowed 25iterations in total) and for each variable (we included thedemographic clinical CSF and imaging variables listed intables 1 and 2) the fully conditional specification method

fits a univariate model using all other available variables inthe model as predictors and then imputes missing values forthe variable being fit

Characteristics of participants with missing data are pro-vided in table 2 available from Dryad (doiorg105061dryadht92839) To evaluate patterns of missingness weperformed binary logistic regression models with the de-mographic clinical and imaging variables listed in this tableas predictors and missing cognitive data on each test (yesno) as the dependent variable There were no associationsbetween missing data on neuropsychological tests with anyof the variables In addition the imputed dataset providedresults highly similar to those of the original dataset (datanot shown)

MRI studyT1-weighted MRI was performed on two 30T magneticresonance scanners (Siemens Tim Trio [n = 91 30 withpreclinical AD and 61 with clinical AD] and Siemens Skyra[n = 15 3 with preclinical AD and 12with clinical AD]

Table 1 Demographic and clinical characteristics according to diagnostic group

PreclinicalAD (n = 33) No

ProdromalAD (n = 25) No

AD dementia(n = 48) No

Age y 744 plusmn 73 33 731 plusmn 72 25 715 plusmn 73 48

Sex (malefemale) n 1320 33 169 25 2622 48

Education y 118 plusmn 38 33 127 plusmn 35 25 125 plusmn 38 47

Aβ status positive 100 33 100 25 100 48

APOE laquo4 status positive 656 32 818 22 683 41

CSF Aβ42a 507 plusmn 91 33 435 plusmn 102 24 397 plusmn 110 46

CSF t-taub 449 plusmn 147 33 627 plusmn 186 24 739 plusmn 326 46

CSF p-taub 57 plusmn 14 33 77 plusmn 19 24 89 plusmn 37 46

MMSE scorec 291 plusmn 11 33 256 plusmn 28 25 214 plusmn 49 48

ADAS immediate recall scorede 73 plusmn 14 33 46 plusmn 11 23 39 plusmn 16 44

ADAS delayed recall scorecd 75 plusmn 20 33 37 plusmn 23 23 20 plusmn 22 42

TMT-A scoref 528 plusmn 197 33 687 plusmn 395 22 1048 plusmn 821 42

AQT (color and form) scoree 655 plusmn 155 33 891 plusmn 341 22 1029 plusmn 349 37

ADAS naming scored 112 plusmn 19 32 108 plusmn 19 21 99 plusmn 28 43

Verbal fluency score animals (1 min)f 208 plusmn 57 33 139 plusmn 48 24 110 plusmn 57 38

Abbreviations Aβ = β-amyloid AD = Alzheimer disease ADAS = Alzheimerrsquos Disease Assessment Scale AQT = A Quick Test of Cognitive Speed MMSE = Mini-Mental State Examination p-tau = phosphorylated tau TMT-A = Trail Making Test A t-tau = total tauDifferences in baseline characteristics between groupswere assessedwith analysis of variance with post hoc Bonferroni tests for continuous variables and χ2

and Kruskal-Wallis with post hoc Mann-Whitney U tests for categorical or ordinal data Number indicates the number of participants within each diagnosticgroup with the variable available Note that missing neuropsychological data and education (for 1 patient with AD dementia) were imputed for furtheranalysisa AD dementia lt preclinical AD p lt 005b AD dementia gt preclinical AD p lt 0001 and prodromal AD gt preclinical AD p lt 005c AD dementiaprodromal AD lt preclinical AD p lt 0001 and AD dementia lt prodromal AD p lt 001d Scores are inverted so that higher numbers indicate better performancee AD dementiaprodromal AD lt preclinical AD p lt 001f AD dementia lt prodromal ADpreclinical AD p lt 005

NeurologyorgN Neurology | Volume 92 Number 6 | February 5 2019 e603

Siemens Medical Solutions Erlangen Germany) resultingin high-resolution anatomic magnetization-prepared rapidgradient-echo images (Tim Trio repetition time 1950milliseconds echo time 34 milliseconds 1-mm isotropicvoxels and 176 slices Skyra repetition time 1900 milli-seconds echo time 25 milliseconds 1-mm isotropic voxelsand 176 slices) Cortical reconstruction and volumetricsegmentation were performed with the FreeSurfer (version53) image analysis pipelines (surfernmrmghharvardedu) Briefly the magnetization-prepared rapid gradient-echo images underwent correction for intensity homoge-neity21 removal of nonbrain tissue22 and segmentationinto gray matter and white matter with intensity gradientand connectivity among voxels23 Cortical thickness wasmeasured as the distance from the gray matterwhitematter boundary to the corresponding pial surface24

Reconstructed datasets were visually inspected for accu-racy and segmentation errors were corrected Corticalthickness was determined for 7 predefined ROIs includingthe medial and lateral parietal cortex medial and lateraltemporal cortex frontal cortex occipital cortex anda whole-brain composite (including all cortical gray

matter) Detailed composition of each ROI by FreeSurferlabel is shown in table 3 available from Dryad (doiorg105061dryadht92839)

Tau and Aβ PET[18F]flortaucipir and [18F]flutemetamol were synthesized atSkaringne University Hospital Lund and PET scans were per-formed on a GEDiscovery 690 PET scanner (General ElectricMedical Systems Chicago IL) and a Philips Gemini TF 16scanner (Philips Healthcare Amsterdam the Netherlands)respectively [18F]flortaucipir (previously known as [18F]AV1451 or [18F]T807) is a PET tracer with high affinity topaired helical filaments of tau The mean injected dose of[18F]flortaucipir was asymp370 MBq and participants underwenta PET scan during the 80- to 100-minute interval after in-jection Images were motion corrected with the AFNI3dvolreg time averaged and rigidly coregistered to the skull-stripped MRI scan Standardized uptake value ratio (SUVR)images for the interval between 80 and 100 seconds afterinjection were created with inferior cerebellar gray matterused as the reference region25 We additionally performeda correction for partial volume effects (PVE) using the

Table 2 Mean [18F]flortaucipir [18F]flutemetamol and cortical thickness and their correlations

[18F]flortaucipirSUVR

[18F]flutemetamolSUVR

Corticalthickness mm

r Flortaucipir vsflutemetamol

r Flortaucipir vsthickness

r Flutemetamol vsthickness

Preclinical AD

No 33 30 33 30 33 33

Lateral parietal 110 plusmn 012 086 plusmn 016 211 plusmn 013 020 008 003

Medial parietal 108 plusmn 008 091 plusmn 019 212 plusmn 011 035 minus019 010

Lateral temporal 116 plusmn 011 081 plusmn 015 248 plusmn 014 051bd 002 minus002

Medial temporal 111 plusmn 017 065 plusmn 007 289 plusmn 021 053bd minus003 024

Frontal 105 plusmn 007 087 plusmn 017 228 plusmn 013 015 019 minus011

Occipital 110 plusmn 007 073 plusmn 011 186 plusmn 010 030 minus015 minus011

Whole brain 107 plusmn 007 082 plusmn 014 221 plusmn 012 037a 011 005

Prodromal AD andAD dementia

No 73 58 73 58 73 58

Lateral parietal 163 plusmn 055 106 plusmn 014 198 plusmn 014 015 minus054cd minus013

Medial parietal 158 plusmn 054 113 plusmn 017 202 plusmn 013 013 minus039bd 002

Lateral temporal 174 plusmn 046 100 plusmn 015 230 plusmn 018 minus003 minus037bd minus007

Medial temporal 153 plusmn 025 071 plusmn 008 244 plusmn 028 minus003 minus010 004

Frontal 134 plusmn 040 109 plusmn 016 220 plusmn 013 002 minus025a minus006

Occipital 145 plusmn 037 085 plusmn 013 182 plusmn 011 032a minus037bd minus016

Whole brain 145 plusmn 036 100 plusmn 013 210 plusmn 011 003 minus028a minus010

Abbreviations AD = Alzheimer dementia SUVR = standardized uptake value ratioData presented are mean plusmn SD for the imaging modalities and unadjusted correlation coefficients between modalitiesap lt 005 bp lt 001 cp lt 0001 dp lt 005 after correction for multiple comparisons using the false discovery rate

e604 Neurology | Volume 92 Number 6 | February 5 2019 NeurologyorgN

Geomatric Transfer Matrix approach26 and report both un-corrected (main report) and corrected (available from Dryaddoiorg105061dryadht92839) data For [18F]flutemeta-mol the mean injected dose was asymp185 MBq PET imageswere acquired between 90 and 110 minutes after injection (4times 5 minute frames) and SUVR images for this time framewere computed with the pons used as the reference region

FreeSurfer (version 53) parcellation of the T1-weighted MRIscan was applied to the PET data transformed to participantsrsquonative T1 space to extract mean regional SUVR values foreach participant in the same 7 ROIs used for analyses ofcortical thickness [18F]flutemetamol PET data were missingfor 3 participants with preclinical AD 7 with prodromal ADand 8 with AD dementia For 3 participants (1 with preclinicalAD 2 with prodromal AD) with missing CSF data Aβ posi-tivity was determined with [18F]flutemetamol SUVR gt069 ina composite neocortical ROI27

Statistical analysisDifferences in baseline characteristics between groups wereassessed with analysis of variance with post hoc Bonferronitests for continuous variables and χ2 and Kruskal-Wallis withpost hoc Mann-Whitney U tests for categorical or ordinalvariables Associations between [18F]flortaucipir [18F]flute-metamol and cortical thickness were examined with the(unadjusted) Pearson correlation analysis Next we per-formed linear regression models with an imaging ROI as theindependent variable and a neuropsychological test score asthe dependent variable while adjusting for age sex and edu-cation These covariates were included on the basis of theirestablished strong associations with cognitive performanceWe did not control for MRI scanner type because the 2 Sie-mens scanners had highly similar properties and only 15participants were scanned on the Siemens Skyra The analyseswere performed for each combination of ROI (n = 7) imagingmodality (n = 3) and neuropsychological test (n = 7) and westratified participants on the basis of preclinical (only cogni-tively normal individuals) or clinical (those with MCI and ADdementia combined) stage of AD We repeated these analysesusing nonparametric linear regression models (adjusted forage sex and education) for all 3 imaging markers and usingPVE-corrected data for [18F]flortaucipir We repeated theanalyses for cortical thickness by adding MRI scanner type asan additional covariate To provide more fine-grained regionalinformation we performed the analyses using all 68 (unilat-eral) FreeSurfer parcellations derived from the Desikan-Killiany atlas To assess whether the degree of asymmetry wasrelated to specific neuropsychological functions we repeatedthe aforementioned analyses using a left and right hemispherecomposite of temporoparietal regions Statistical significancewas set at p lt 005 Because results were mostly in the sameand expected direction which reduces the likelihood of false-positive findings we indicated the level of significance bothwithout and with correction for multiple comparisons usingthe Benjamini-Hochberg procedure with a false discovery rate(FDR) Q value of 5

Data availabilityAnonymized data will be shared by request from any qualifiedinvestigator for the sole purpose of replicating procedures andresults presented in the article and as long as data transfer is inagreement with European Union legislation on the generaldata protection regulation

ResultsParticipantsThere were no significant differences in age education andsex between the preclinical prodromal and dementia groupsAs expected patients with AD dementia generally had themost severe biomarker or imaging abnormalities and worsecognitive performance followed by those with prodromal ADand then those with preclinical AD (tables 1 and 2)

Associations between [18F]flortaucipir [18F]flortaucipir and cortical thicknessIn preclinical AD regional correlations between [18F]flor-taucipir and [18F]flutemetamol were observed in the medialand temporal cortical cortex and in the whole-brain compositeROI (range Pearson r = 037ndash053 all p lt 005 table 2) In theprodromal AD and AD dementia group [18F]flortaucipir and[18F]flutemetamol SUVRs were correlated in the occipitalcortex (Pearson r = 032 p lt 005 table 2) There was anassociation between [18F]flortaucipir and cortical thickness inall ROIs (range Pearson r = 025ndash054 all p lt 005) except forthe medial temporal lobe There were no regional associationsbetween [18F]flutemetamol and cortical thickness for any ofthe diagnostic groups (table 2)

Preclinical ADLinear regression models adjusted for age sex and educa-tion in the preclinical AD group showed associationsbetween increased [18F]flortaucipir uptake and worse per-formance on the ADAS immediate recall (lateral and medialparietal cortex lateral temporal cortex and whole brain)ADAS delayed recall (lateral parietal cortex) MMSE (me-dial temporal cortex) and TMT-A (medial parietal cortexlateral temporal cortex frontal cortex and whole-brain)(table 3 and figure 1) None of these associations survivedFDR correction for multiple comparisons There were nosignificant associations between regional cortical thicknessor [18F]flutemetamol uptake and any of the neuro-psychological tests (table 3 and figure 1) Analyses for all 3imaging markers using nonparametric regression modelsand PVE-corrected [18F]flortaucipir data yielded largelysimilar results (tables 4 and 5 available from Dryad doiorg105061dryadht92839) Cortical thickness in the medialparietal and occipital cortex was significantly associated withTMA-A when additionally adjusted for MRI scanner typewhile all other associations remained nonsignificant (table 6available from Dryad) The associations between all 68(unilateral) FreeSurfer parcellations and cognition areshown in table 7 available from Dryad

NeurologyorgN Neurology | Volume 92 Number 6 | February 5 2019 e605

Prodromal AD and AD dementiaTable 3 presents the standardized β coefficients resulting fromlinear regression models in the combined prodromal AD andAD dementia group Regional [18F]flortaucipir SUVR andcortical thickness showed strong associations with cognitionacross all neuropsychological tests (table 4 and figure 2) Forboth [18F]flortaucipir and cortical thickness standardized βcoefficients were generally higher in temporoparietal regionscompared to frontal occipital or whole-brain ROIs Further-more for most neuropsychological tests (especially episodicmemory tasks) the associations with cognition were slightlystronger for [18F]flortaucipir than for cortical thickness (withthe exception of the AQT table 4) For [18F]flutemetamol

increased uptakewas associated with lower scores on the ADASdelayed recall for all ROIs except the occipital cortex (table 4)while occipital [18F]flutemetamol retention was associatedwith worse performance on tasks involving processing speedand attention (ie TMT-A and AQT) For [18F]flortaucipirand cortical thickness several associations with cognition sur-vived FDR correction for multiple comparisons but not for[18F]flutemetamol (table 4)

Results are also provided for the separate diagnostic groups intable 8 (prodromal AD) and table 9 (AD dementia) availablefrom Dryad (doiorg105061dryadht92839) Results for pro-dromal AD andADdementia groupsweremostly comparable to

Table 3 Associations between [18F]flortaucipir [18F]flutemetamol and cortical thickness and cognition in preclinical AD

MMSEscore

ADAS immediaterecall score

ADAS delayedrecall score

TMT-Ascore

AQTscore

ADASnaming score

Animalfluency score

[18F]flortaucipir (n = 33)

Lateral parietal minus008 minus051b minus039a 024 minus008 minus006 minus022

Medial parietal minus019 minus052b minus030 039a 011 002 010

Lateral temporal minus023 minus040a minus026 039a 014 minus005 minus020

Medial temporal minus045a minus022 minus016 035a 027 004 minus009

Frontal minus005 minus036 minus033 039a minus009 minus013 minus021

Occipital minus022 minus028 minus025 032 000 minus016 005

Whole brain minus015 minus042a minus032 039a 001 minus010 minus016

[18F]flutemetamol (n = 30)

Lateral parietal minus011 minus014 004 002 019 minus012 minus004

Medial parietal minus014 minus015 006 minus013 013 minus009 minus007

Lateral temporal minus015 minus012 008 005 024 002 minus008

Medial temporal 001 minus019 007 006 029 minus007 minus007

Frontal minus026 minus013 001 minus013 015 minus003 minus004

Occipital 005 minus013 013 015 025 minus009 minus014

Whole brain minus015 minus014 006 minus003 018 minus005 minus006

Cortical thickness (n = 33)

Lateral parietal 026 021 003 minus006 minus015 minus025 002

Medial parietal 016 019 minus008 minus031 minus032 minus020 001

Lateral temporal 007 012 minus003 minus019 000 minus031 minus002

Medial temporal minus014 007 009 minus010 018 001 006

Frontal 015 015 minus003 003 001 026 008

Occipital 020 021 013 minus024 minus001 minus018 004

Whole brain 018 018 003 minus013 minus005 minus025 002

Abbreviations AD = Alzheimer dementia ADAS = Alzheimerrsquos Disease Assessment Scale AQT = A Quick Test of Cognitive Speed MMSE = Mini-Mental StateExamination TMT-A = Trail Making Test AData presented are standardized β coefficients derived from linear regression models adjusted for age sex and education TMT-A and AQT were log-transformed before statistical analysisap lt 005 bp lt 001 (both uncorrected for multiple comparisons)

e606 Neurology | Volume 92 Number 6 | February 5 2019 NeurologyorgN

those of the combined group in terms of direction (figure 2) andstrength of the effect sizes but were less (often) significant be-cause of reduced statistical power related to smaller sample sizesRepeating the analysis for [18F]flortaucipir using PVE correcteddata resulted in weaker associations with MMSE and ADASmemory and language tests and yielded comparable associationswith TMT AQT and animal fluency (table 5 available fromDryad) Analyses for all 3 imaging markers using nonparametricregression models (table 10 available from Dryad) yieldedlargely similar results compared to its parametric counterpartreported in table 4 When MRI scanner type was added as an

additional covariate to the model the associations betweencortical thickness and cognition remained virtually the same(table 6 available from Dryad) The associations between all 68(unilateral) FreeSurfer parcellations and cognition are given intable 7 available from Dryad

LateralizationIn preclinical AD associations between [18F]flortaucipir up-take and cognition were more pronounced in the left tem-poroparietal cortex than in the right temporoparietal cortexfor ADAS immediate recall (standardized β coefficient minus057

Figure 1 Associations between [18F]flortaucipir [18F]flutemetamol and cortical thickness and cognition in preclinical AD

(A) Lateral parietal region of interest (ROI) for each imagingmodality vs AlzheimerrsquosDisease Assessment Scale (ADAS) immediate recall scores (B)Whole-brainROI vs Trail Making Test A (TMT-A) (log-transformed for statistical analysis but original data are shown here) βs are standardized and presented with andwithout adjustment for age sex and education SUVR = standardized uptake value ratio

NeurologyorgN Neurology | Volume 92 Number 6 | February 5 2019 e607

vs minus037) and for ADAS delayed recall (minus040 vs minus025)There were no clear asymmetric relationships for tempor-oparietal [18F]flortaucipir uptake with any of the otherneuropsychological tests or for temporoparietal [18F]flute-metamol or cortical thickness (table 11 available from Dryaddoiorg105061dryadht92839) For prodromal AD and ADdementia [18F]flortaucipir uptake and cortical thickness werestrongly related to ADAS naming in the left temporoparietalcortex but not the right temporoparietal cortex (figure 3)There were no clear asymmetric findings for the other neu-ropsychological tests and [18F]flutemetamol showed com-parable results for left vs right temporoparietal cortex (table12 available from Dryad)

DiscussionIn this study we examined the cross-sectional associations be-tween tau Aβ and cortical thickness and neuropsychologicalfunction across Aβ-positive individuals at the preclinical andclinical stages of AD In the preclinical AD group we foundassociations with cognition only for tau PET but not for Aβ PETor cortical thickness In prodromal AD and AD dementia wefound strong relationships between increased tau pathology andreduced cortical thickness with worse performance on a widevariety of neuropsychological tests These relationships weremost pronounced in bilateral temporoparietal regions Aβ pa-thology was specifically related to decreased delayed recall on

Table 4 Associations between [18F]flortaucipir [18F]flutemetamol and cortical thickness and cognition in prodromal ADand AD dementia

MMSEscore

ADAS immediaterecall score

ADAS delayedrecall score

TMT-Ascore

AQTscore

ADASnaming score

Animalfluency score

[18F]flortaucipir (n = 73)

Lateral parietal minus032ad minus012 minus022 055cd 033ad minus020 minus041bd

Medial parietal minus028a minus026 minus030a 047bd 030a minus031a minus048bd

Lateral temporal minus034ad minus031ad minus035bd 035bd 017 minus038bd minus045bd

Medial temporal minus025a minus031ad minus032ad 005 minus002 minus022 minus025a

Frontal minus037bd minus046bd minus033ad 017 008 minus051cd minus053cd

Occipital minus020 minus007 minus027a 048cd 023 005 minus029ad

Whole brain minus036bd minus033ad minus033ad 041bd 022 minus036ad minus050cd

[18F]flutemetamol (n = 58)

Lateral parietal minus021 minus002 minus033a 015 016 minus005 minus002

Medial parietal minus015 minus010 minus035b 005 003 minus009 002

Lateral temporal minus024 minus009 minus037b 001 010 minus007 003

Medial temporal minus026 minus003 minus030a minus004 minus003 minus004 minus001

Frontal minus023 minus009 minus029a 004 006 minus007 003

Occipital minus015 minus005 minus025 027a 024a 008 minus008

Whole brain minus023 minus008 minus033a 009 012 minus005 001

Cortical thickness (n = 73)

Lateral parietal 031ad 000 016 minus036bd minus041cd 000 024

Medial parietal 021 minus008 006 minus034bd minus044cd minus002 016

Lateral temporal 032bd 014 025a minus023 minus030bd 025a 038bd

Medial temporal 040bd 028a 037bd minus007 minus016 016 037bd

Frontal 025a 015 007 minus001 minus007 014 035bd

Occipital 013 minus016 013 minus033bd minus035bd minus010 006

Whole brain 032bd 009 018 minus023 minus029bd 013 034bd

Abbreviations AD = Alzheimer dementia ADAS = Alzheimerrsquos Disease Assessment Scale AQT = A Quick Test of Cognitive Speed MMSE = Mini-Mental StateExamination TMT-A = Trail Making Test AData presented are standardized β coefficients derived from linear regression models adjusted for age sex and education TMT-A and AQT were log-transformed prior to statistical analysisap lt 005 bp lt 001 cp lt 0001 dp lt 005 after correction for multiple comparisons using the false discovery rate

e608 Neurology | Volume 92 Number 6 | February 5 2019 NeurologyorgN

a word list learning task and showed limited regional specificityThese findings suggest that [18F]flortaucipir PET is the mostsensitive of thesemarkers for detecting early changes in cognitivefunction while both [18F]flortaucipir PET and cortical thicknessshow strong cognitive correlates at the clinical stages of AD

A main finding in the present study was that regional tau PETwas related to cognitive function already in cognitively normalAβ-positive individuals in the absence of similar effects for

cortical thickness and Aβ pathology This is in line with earlierclinicopathologic528 and imaging studies2930 in preclinicalstages of AD showing intimate links between tau pathologyand cognition The lack of an effect for Aβ pathology could beexplained by the estimated time interval of asymp15 to 20 yearsbetween its pathologic beginnings and symptom onset131

Relationships between reductions in cortical thickness areless affected by this separation in space and time as observedfor Aβ pathology3233 Cortical thickness however is

Figure 2 Associations between [18F]flortaucipir [18F]flutemetamol and cortical thickness and cognition in prodromal ADand AD dementia

(A) Lateral temporal region of interest (ROI) for each imagingmodality vs Alzheimerrsquos Disease Assessment Scale (ADAS) delayed recall scores (B) Whole-brainROI vs performance on animal fluency βs are standardized and presented with and without adjustment for age sex and education AD = Alzheimerdementia SUVR = standardized uptake value ratio

NeurologyorgN Neurology | Volume 92 Number 6 | February 5 2019 e609

characterized by substantial preexisting variation amongindividuals which might reduce the sensitivity for separatingthe earliest AD-related changes from age-related brainchanges Furthermore tau pathology might be present beforeneurodegeneration starts434 and might exert its effects oncognition through both structural35 (ie cortical thicknessreductions) and functional3637 (ie synaptic alterations)pathways This finding has potential ramifications for clinicaltrials in that tau PET might be superior to the other measuresfor identifying individuals at risk for future cognitive declineand for tracking cognitive trajectories during treatment withdisease-modifying drugs

At the clinical stages of AD (ie prodromal AD and ADdementia) regional patterns of both tau pathology and cor-tical thickness showed strong associations with cognitivefunction This finding in a cohort of patients with mostly late-onset AD with memory-predominant presentations is in linewith previous tau PET studies characterized by an on averageyoung age at onset and often atypical presentation6738

Compared to the preclinical stages of AD the effects of tauPET were considerably stronger and were present across allneuropsychological tests and ROIs although there wasa temporoparietal predominance (table 3) The latter is inagreement with the composition of our neuropsychologicaltest battery (which did not include a specific frontal test) incombination with the localization of tau pathology whichtypically strongly occupies temporoparietal regions in ADThe effects of cortical thickness vs tau PET results wereoverall comparable although slightly weaker for cortical

thickness on memory domains and slightly stronger on a taskinvolving processing speed Although our models were set upto test cognition-imaging correlates at a group level and not tomake individual predictions the results open the possibilitythat both tau PET and cortical thickness have potential utilityfor clinical use (ie diagnosis or prognosis) at clinical stages ofthe disease and for patient selection and monitoring of clinicaltrials

In line with the assumptions of a hypothetical biomarkermodel4 the relationships between Aβ pathology and cogni-tion at the clinical stages of AD were modest at best The rateof accumulation of Aβ pathology is thought to attenuate oreven plateau at the time when clinical symptoms haveemerged and at that point Aβ pathology has spread out intonearly the entire neocortex It has therefore been argued thatAβ pathology might be triggering tau pathology to spreadfrom entorhinal cortex to neocortical areas and that this eventis the actual driver of cognitive loss3940 Consequently Aβpathology is often not or only modestly related to cognitiveperformance101241 although there are some exceptions invery young patients with AD42 or atypical variants of AD43 Inthe present study the effects of Aβ pathology were muchweaker compared to those of tau pathology and corticalthickness but we found significant associations between AβPET and episodic memory (delayed recall) in almost allneocortical areas This could be explained by the complexityof this task previous research has shown that cognitivelydemanding tests may be required to capture the subtle effectsof Aβ pathology on cognition44 Furthermore if effects of Aβ

Figure 3 Asymmetric relationships between imaging modalities and object naming

Stronger relationships with Alzheimerrsquos Disease Assessment Scale (ADAS) object naming for left temporoparietal cortex (A) [18F]flortaucipir uptake and (B)cortical thickness compared to the right temporoparietal cortex βs are standardized and presented adjusted for age sex and education

e610 Neurology | Volume 92 Number 6 | February 5 2019 NeurologyorgN

pathology on cognition are shown they typically involvememory function The reason could be that this is usually themost impaired cognitive domain in patients with AD or withAβ pathology initially spreading through posterior parts of thedefault mode network4546 which could be affecting memoryfunction through its strong connections with medial temporallobe structures47 Occipital Aβ load was associated with re-duced processing speed and attention Because the occipitalcortex is among the brain regions affected by Aβ deposition inlate stages this could reflect a more advanced disease stageAlternatively the occipital signal may reflect the presence ofcerebral amyloid angiopathy which has a predilection for theoccipital cortex48 and is associated with mental slowness andattentional deficits49

A strength of this study is that we combined 3 imaging mo-dalities in combination with a neuropsychological test batteryin individuals across the AD spectrum There are also somelimitations First although our neuropsychological batteryincluded tests commonly used in clinical studies and trials wedid not capture all cognitive domains In particular tests forvisuospatial and some executive functions were missingSecond Aβ PET was missing in 18 (17) participants Thirdthe sample size differed between analyses in preclinical (n =33) and clinical (n = 73) stages Interpretation was thereforefocused on effect size (not affected by sample size) and level ofsignificance (affected by sample size) Fourth [18F]flortau-cipir is still a relatively new tracer thus only a few antemor-tem vs postmortem comparisons have been reported and itsbinding properties are not yet fully understood Factors thatcould affect the accuracy of the signal include off-targetbinding (eg to monoamine oxidase B neuromelanin orvascular lesions) and the continued increase in specific tracerbinding after the duration of the PET scan50 Fifth it is pos-sible that some patients had comorbid pathologies such asα-synuclein TAR DNA-binding protein 43 or vascular pa-thology which were not examined The presence of suchcomorbid conditions may potentially attenuate the associa-tions between cognitive scores and biomarkers of tau Aβ andcortical thickness

We found that tau PET is a more sensitive marker fordetecting the earliest cognitive changes in AD than Aβ PETand cortical thickness measures In clinical stages both re-gional tau PET and cortical thickness measures showed strongcognitive correlates while Aβ PET showed weaker and lessregion-specific associations with cognition Future studiesshould include longitudinal assessments of both neuro-psychological scores and neuroimaging to determine whichmodality is most suited for predicting and monitoring cog-nitive changes over time Furthermore the overlapping andcomplementary effects of tau and brain atrophy on cognitionshould be further investigated7 and ROIs have to be refinedIdentifying the contribution of these neurobiological hall-marks of AD to cognitive changes could eventually improvethe diagnostic and prognostic process in AD and may informdesign participant selection and monitoring of clinical trials

Author contributionsRO contributed to study design analysis and interpretationof data writing and revising the manuscript and performedstatistical analysis TO and OS contributed to the acquisi-tion and analysis of data RS NM PSI and SP criticallyrevised the manuscript for intellectual content OH con-tributed to study concept and design acquisition of patientdata from the Swedish BioFINDER study interpretation ofdata critical revision of the manuscript for intellectual con-tent and supervised the study

AcknowledgmentThe authors thank Colin Groot for support with the figures

Study fundingWork at Lund University was supported by the EuropeanResearch Council the Swedish Research Council theMarianne and Marcus Wallenberg foundation the StrategicResearch Area MultiPark (Multidisciplinary Research inParkinsonrsquos disease) at Lund University the Swedish BrainFoundation the Swedish Alzheimer Foundation The Par-kinson Foundation of Sweden The Parkinson ResearchFoundation the Skaringne University Hospital Foundation andthe Swedish federal government under the ALF agreementDoses of 18F-flutemetamol injection were sponsored by GEHealthcare and the precursor of [18F]flortaucipir was pro-vided by AVID Radiopharmaceuticals

DisclosureR Ossenkoppele R Smith T Ohlsson O Strandberg NMattsson P Insel and S Palmqvist report no disclosuresrelevant to the manuscript O Hansson has acquired researchsupport (for the institution) from Roche GE HealthcareBiogen AVID Radiopharmaceuticals Fujirebio and Euro-immun In the past 2 years he has received consultancyspeaker fees (paid to the institution) from Lilly Rocha andFujirebio Go to NeurologyorgN for full disclosures

Publication historyReceived by Neurology March 21 2018 Accepted in final formOctober 3 2018

References1 JansenWJ Ossenkoppele R Knol DL et al Prevalence of cerebral amyloid pathology

in persons without dementia a meta-analysis JAMA 20153131924ndash19382 Jack CR Jr Wiste HJ Weigand SD et al Age-specific and sex-specific prevalence of

cerebral beta-amyloidosis tauopathy and neurodegeneration in cognitively un-impaired individuals aged 50-95 years a cross-sectional study Lancet Neurol 201716435ndash444

3 Sperling RA Aisen PS Beckett LA et al Toward defining the preclinical stages ofAlzheimerrsquos disease recommendations from the National Institute on Aging-Alz-heimerrsquos Association workgroups on diagnostic guidelines for Alzheimerrsquos diseaseAlzheimers Dement 20117280ndash292

4 Jack CR Jr Holtzman DM Biomarker modeling of Alzheimerrsquos disease Neuron2013801347ndash1358

5 Nelson PT Alafuzoff I Bigio EH et al Correlation of Alzheimer disease neuro-pathologic changes with cognitive status a review of the literature J Neuropathol ExpNeurol 201271362ndash381

6 Ossenkoppele R Schonhaut DR Scholl M et al Tau PET patterns mirror clinical andneuroanatomical variability in Alzheimerrsquos disease Brain 20161391551ndash1567

7 Bejanin A Schonhaut DR La Joie R et al Tau pathology and neurodegenerationcontribute to cognitive impairment in Alzheimerrsquos disease Brain 20171403286ndash3300

8 Jack CR Jr Wiste HJ Weigand SD et al Defining imaging biomarker cut points forbrain aging and Alzheimerrsquos disease Alzheimers Dement 201713205ndash216

NeurologyorgN Neurology | Volume 92 Number 6 | February 5 2019 e611

9 Zhang X Mormino EC Sun N et al Bayesian model reveals latent atrophy factorswith dissociable cognitive trajectories in Alzheimerrsquos disease Proc Natl Acad Sci USA2016113E6535ndashE6544

10 Jansen WJ Ossenkoppele R Tijms BM et al Association of cerebral amyloid-betaaggregation with cognitive functioning in persons without dementia JAMAPsychiatry2018584ndash95

11 Ossenkoppele R van der Flier WM Verfaillie SC et al Long-term effects of amyloidhypometabolism and atrophy on neuropsychological functions Neurology 2014821768ndash1775

12 Hedden T OhH Younger AP Patel TAMeta-analysis of amyloid-cognition relationsin cognitively normal older adults Neurology 2013801341ndash1348

13 Palmqvist S Zetterberg H Blennow K et al Accuracy of brain amyloid detection inclinical practice using cerebrospinal fluid beta-amyloid 42 a cross-validation studyagainst amyloid positron emission tomography JAMA Neurol 2014711282ndash1289

14 Albert MS DeKosky ST Dickson D et al The diagnosis of mild cognitive impairmentdue to Alzheimerrsquos disease recommendations from the National Institute on Aging-Alzheimerrsquos Association workgroups on diagnostic guidelines for Alzheimerrsquos diseaseAlzheimers Dement 20117270ndash279

15 Dubois B Feldman HH Jacova C et al Advancing research diagnostic criteria forAlzheimerrsquos disease the IWG-2 criteria Lancet Neurol 201413614ndash629

16 McKhann GM Knopman DS Chertkow H et al The diagnosis of dementia due toAlzheimerrsquos disease recommendations from the National Institute on Aging-Alz-heimerrsquos Association workgroups on diagnostic guidelines for Alzheimerrsquos diseaseAlzheimers Dement 20117263ndash269

17 Jessen F Amariglio RE van Boxtel M et al A conceptual framework for research onsubjective cognitive decline in preclinical Alzheimerrsquos disease Alzheimers Dement201410844ndash852

18 Jack CR Jr Bennett DA Blennow K et al NIA-AA Research Framework towarda biological definition of Alzheimerrsquos disease Alzheimers Dement 201814535ndash562

19 Blennow K Hampel H Weiner M Zetterberg H Cerebrospinal fluid and plasmabiomarkers in Alzheimer disease Nat Rev Neurol 20106131ndash144

20 Spratt M Carpenter J Sterne JA et al Strategies for multiple imputation in longi-tudinal studies Am J Epidemiol 2010172478ndash487

21 Sled JG Zijdenbos AP Evans AC A nonparametric method for automatic correctionof intensity nonuniformity in MRI data IEEE Trans Med Imaging 19981787ndash97

22 Segonne F Dale AM Busa E et al A hybrid approach to the skull stripping problem inMRI Neuroimage 2004221060ndash1075

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

24 Fischl B Dale AM Measuring the thickness of the human cerebral cortex frommagnetic resonance images Proc Natl Acad Sci USA 20009711050ndash11055

25 Maass A Landau S Baker SL et al Comparison of multiple tau-PET measures asbiomarkers in aging and Alzheimerrsquos disease Neuroimage 2017157448ndash463

26 Rousset OG Ma Y Evans AC Correction for partial volume effects in PET principleand validation J Nucl Med 199839904ndash911

27 Mattsson N Smith R Strandberg O et al Comparing (18)F-AV-1451 with CSF t-tauand p-tau for diagnosis of Alzheimer disease Neurology 201890e388ndashe395

28 Mitchell TW Mufson EJ Schneider JA et al Parahippocampal tau pathology inhealthy aging mild cognitive impairment and early Alzheimerrsquos disease Ann Neurol200251182ndash189

29 Johnson KA Schultz A Betensky RA et al Tau positron emission tomographicimaging in aging and early Alzheimer disease Ann Neurol 201679110ndash119

30 Scholl M Lockhart SN Schonhaut DR et al PET imaging of tau deposition in theaging human brain Neuron 201689971ndash982

31 Jack CR Jr Wiste HJ Weigand SD et al Age-specific population frequencies ofcerebral beta-amyloidosis and neurodegeneration among people with normal cogni-tive function aged 50-89 years a cross-sectional study Lancet Neurol 201413997ndash1005

32 Iaccarino L Tammewar G Ayakta N et al Local and distant relationships betweenamyloid tau and neurodegeneration in Alzheimerrsquos disease Neuroimage Clin 201817452ndash464

33 LaPoint MR Chhatwal JP Sepulcre J Johnson KA Sperling RA Schultz AP Theassociation between tau PET and retrospective cortical thinning in clinically normalelderly Neuroimage 2017157612ndash622

34 Braak H Del Tredici K The preclinical phase of the pathological process underlyingsporadic Alzheimerrsquos disease Brain 20151382814ndash2833

35 Spillantini MG Goedert M Tau protein pathology in neurodegenerative diseasesTrends Neurosci 199821428ndash433

36 Terry RD Masliah E Salmon DP et al Physical basis of cognitive alterations inAlzheimerrsquos disease synapse loss is the major correlate of cognitive impairment AnnNeurol 199130572ndash580

37 Spires-Jones TL Hyman BT The intersection of amyloid beta and tau at synapses inAlzheimerrsquos disease Neuron 201482756ndash771

38 Phillips JS Das SR McMillan CT et al Tau PET imaging predicts cognition inatypical variants of Alzheimerrsquos disease Hum Brain Mapp 201839691ndash708

39 Giannakopoulos P Herrmann FR Bussiere T et al Tangle and neuron numbers butnot amyloid load predict cognitive status in Alzheimerrsquos disease Neurology 2003601495ndash1500

40 Jacobs HIL Hedden T Schultz AP et al Structural tract alterations predict down-stream tau accumulation in amyloid-positive older individuals Nat Neurosci 201821424ndash431

41 Baker JE Lim YY Pietrzak RH et al Cognitive impairment and decline in cognitivelynormal older adults with high amyloid-beta a meta-analysis Alzheimers Dement(Amst) 20176108ndash121

42 Ossenkoppele R Zwan MD Tolboom N et al Amyloid burden and metabolicfunction in early-onset Alzheimerrsquos disease parietal lobe involvement Brain 20121352115ndash2125

43 Frings L Hellwig S Spehl TS et al Asymmetries of amyloid-beta burden and neu-ronal dysfunction are positively correlated in Alzheimerrsquos disease Brain 20151383089ndash3099

44 Rentz DM Amariglio RE Becker JA et al Face-name associative memory perfor-mance is related to amyloid burden in normal elderly Neuropsychologia 2011492776ndash2783

45 Palmqvist S Scholl M Strandberg O et al Earliest accumulation of beta-amyloidoccurs within the default-mode network and concurrently affects brain connectivityNat Commun 201781214

46 Buckner RL Snyder AZ Shannon BJ et al Molecular structural and functionalcharacterization of Alzheimerrsquos disease evidence for a relationship between defaultactivity amyloid and memory J Neurosci 2005257709ndash7717

47 Buckner RL Andrews-Hanna JR Schacter DL The brainrsquos default network anatomyfunction and relevance to disease Ann NY Acad Sci 200811241ndash38

48 Johnson KA Gregas M Becker JA et al Imaging of amyloid burden and distributionin cerebral amyloid angiopathy Ann Neurol 200762229ndash234

49 van der Flier WM Skoog I Schneider JA et al Vascular cognitive impairment NatRev Dis Primers 2018418003

50 Sander K Lashley T Gami P et al Characterization of tau positron emission tomog-raphy tracer [(18)F]AV-1451 binding to postmortem tissue in Alzheimerrsquos diseaseprimary tauopathies and other dementias Alzheimers Dement 2016121116ndash1124

e612 Neurology | Volume 92 Number 6 | February 5 2019 NeurologyorgN

DOI 101212WNL0000000000006875201992e601-e612 Published Online before print January 9 2019Neurology Rik Ossenkoppele Ruben Smith Tomas Ohlsson et al

disease and cortical thickness with cognition in AlzheimerβAssociations between tau A

This information is current as of January 9 2019

ServicesUpdated Information amp

httpnneurologyorgcontent926e601fullincluding high resolution figures can be found at

References httpnneurologyorgcontent926e601fullref-list-1

This article cites 50 articles 8 of which you can access for free at

Citations httpnneurologyorgcontent926e601fullotherarticles

This article has been cited by 6 HighWire-hosted articles

Subspecialty Collections

httpnneurologyorgcgicollectionpetPET

httpnneurologyorgcgicollectionmriMRI

httpnneurologyorgcgicollectionmci_mild_cognitive_impairmentMCI (mild cognitive impairment)

httpnneurologyorgcgicollectionalzheimers_diseaseAlzheimers disease

httpnneurologyorgcgicollectionall_neuropsychology_behaviorAll NeuropsychologyBehaviorfollowing 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 2019 The Author(s) Published by

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