N

1

2

3

4

5

6

7

8

9

10

11

12

13

14

15

A16

httncwcli

17

18

19

20

21

22

23

24

25

26

©27

K28

29

130

31

t32

m33

NYT

0d

TED

PR

OO

F

ARTICLE IN PRESSBA 7198 1–10

Neurobiology of Aging xxx (2008) xxx–xxx

Plaque and tangle imaging and cognition in normal agingand Alzheimer’s disease

Meredith N. Braskie a,b, Andrea D. Klunder c, Kiralee M. Hayashi c, Hillary Protas d,h,Vladimir Kepe d, Karen J. Miller e, S.-C. Huang d, Jorge R. Barrio d, Linda M. Ercoli e,

Prabha Siddarth e, Nagichettiar Satyamurthy d, Jie Liu d, Arthur W. Toga c,Susan Y. Bookheimer b,e,g, Gary W. Small e,f, Paul M. Thompson c,∗

a Helen Wills Neuroscience Institute, University of California Berkeley, CA 94720, USAb Ahmanson-Lovelace Brain Mapping Center, David Geffen School of Medicine at UCLA, CA 90095, USA

c Laboratory of Neuro Imaging, Department of Neurology, David Geffen School of Medicine at UCLA, CA 90095, USAd Department of Molecular and Medical Pharmacology, David Geffen School of Medicine at UCLA, CA 90095, USAe Department of Psychiatry and Biobehavioral Sciences, David Geffen School of Medicine at UCLA, CA 90095, USA

f Center on Aging, David Geffen School of Medicine at UCLA, CA 90095, USAg Department of Psychology, David Geffen School of Medicine at UCLA, CA 90095, USA

h Department of Biomathematics, David Geffen School of Medicine at UCLA, CA 90095, USA

Received 26 March 2008; received in revised form 9 July 2008; accepted 22 September 2008

bstract

Amyloid plaques and tau neurofibrillary tangles, the pathological hallmarks of Alzheimer’s disease (AD), begin accumulating in theealthy human brain decades before clinical dementia symptoms can be detected. There is great interest in how this pathology spreads inhe living brain and its association with cognitive deterioration. Using MRI-derived cortical surface models and four-dimensional animationechniques, we related cognitive ability to positron emission tomography (PET) signal from 2-(1-{6-[(2-[F-18]fluoroethyl)(methyl)amino]-2-aphthyl}ethylidene)malononitrile ([18F]FDDNP), a molecular imaging probe for plaques and tangles. We examined this relationship at eachortical surface point in 23 older adults (10 cognitively intact, 6 with amnestic mild cognitive impairment, 7 with AD). [18F]FDDNP-PET signalas highly correlated with cognitive performance, even in cognitively intact subjects. Animations of [18F]FDDNP signal growth with decreased

ECognition across all subjects (http://www.loni.ucla.edu/∼thompson/FDDNP/video.html) mirrored the classic Braak and Braak trajectory in

ateral temporal, parietal, and frontal cortices. Regions in which cognitive performance was significantly correlated with [18F]FDDNP signalnclude those that deteriorate earliest in AD, suggesting the potential utility of [18F]FDDNP for early diagnosis.

2008 Elsevier Inc. All rights reserved.

t 34

a 35

RReywords: Amyloid; Cerebral cortex; Cognitive aging; Memory; PET

. Introduction

UN

CO

Please cite this article in press as: Braskie, M.N., et al., Plaque and tangleNeurobiol Aging (2008), doi:10.1016/j.neurobiolaging.2008.09.012

Decades may elapse between initial cortical accumula-ions of tau neurofibrillary tangles and amyloid plaques (the

ajor pathologic hallmarks of Alzheimer’s disease (AD)) and

∗ Corresponding author at: Laboratory of Neuro Imaging, Department ofeurology, David Geffen School of Medicine at UCLA, 635 Charles E.oung Drive South, Suite 225E, Los Angeles, CA 90095-7332, USA.el.: +1 310 206 2101; fax: +1 310 206 5518.

E-mail address: thompson@loni.ucla.edu (P.M. Thompson).

t 36

l 37

H 38

t 39

o 40

2sso

197-4580/$ – see front matter © 2008 Elsevier Inc. All rights reserved.oi:10.1016/j.neurobiolaging.2008.09.012

he cognitive changes required for clinical diagnosis (Braaknd Braak, 1991; Thal et al., 2002). Historically, plaques andangles were only detectable post mortem, making the patho-ogical burden difficult to relate to cognitive performance.owever, PET probe advances may facilitate investigation of

his pathology in living humans. In vivo cortical accumulationf the ligand 2-(1-{6-[(2-[F-18]fluoroethyl)(methyl)amino]-

imaging and cognition in normal aging and Alzheimer’s disease.

-naphthyl}ethylidene)malononitrile ([18F]FDDNP) is con- 41

istent with depositions of both plaques and tangles in living 42

ubjects (Agdeppa et al., 2001b; Small et al., 2006), and previ- 43

usly has been verified via subsequent autopsy to co-localize 44

D

INNBA 7198 1–10

2 iology o

w45

a46

h47

m48

i49

c50

F51

s52

u53

a54

b55

256

a57

r58

d59

a60

(61

62

t63

c64

c65

c66

c67

t68

u69

m70

t71

c72

m73

m74

a75

o76

277

278

79

a80

e81

h82

p83

e84

f85

s86

o87

T88

d89

i90

i91

m92

w93

n94

P95

c96

s 97

R 98

l 99

e 100

i 101

( 102

b 103

104

a 105

e 106

q 107

w 108

109

o 110

I 111

L 112

2 113

114

c 115

u 116

W 117

o 118

l 119

r 120

e 121

t 122

a 123

124

t 125

o 126

i 127

A 128

w 129

p 130

o 131

132

f 133

s 134

p 135

w 136

f 137

i 138

t 139

e 140

M 141

c 142

u 143

o 144

A 145

t 146

NC

OR

RE

CTE

ARTICLEM.N. Braskie et al. / Neurob

ith plaques and tangles (Agdeppa et al., 2001b). Digitalutoradiography of AD brain specimens using [18F]FDDNPas likewise demonstrated that the pattern of ligand bindingatches the pattern of plaques and tangles in neighbor-

ng slices (as determined using immunohistochemistry andonfocal fluorescence microscopy) (Agdeppa et al., 2001a).urthermore, in a recent study, global [18F]FDDNP washown to be more accurate than FDG-PET or MRI brain vol-mes at discriminating among clinical diagnoses (Small etl., 2006). In addition to [18F]FDDNP, PET ligands Pitts-urgh Compound B (PIB) (Buckner et al., 2005; Klunk et al.,004; Mintun et al., 2006) and stilbene (SB-13) (Verhoeff etl., 2004) are both reported to visualize plaques alone. Initialeports have focused on validating the imaging probes andifferentiating diagnostic groups by examining signal aver-ged over regions of interest rather than at each cortical pointKlunk et al., 2004; Small et al., 2006; Verhoeff et al., 2004).

In this study, we compared the voxel-wise spatial rela-ionship between 3D cortical [18F]FDDNP distribution andognitive ability across subjects with diagnoses ranging fromognitively intact to mild AD. This approach empoweredortical signal detection without requiring a priori specifi-ation of regions of interest (ROIs), thus preventing biaseshat could be introduced by variations in ROI sizes. We eval-ated cognition using composite test scores as a continuouseasure, which ensured that participants close to diagnos-

ic boundaries did not obscure results as they might withategorical comparisons. Finally, we mapped cortical grayatter thickness (Thompson et al., 2004) both to separateolecular pathology from the effects of structural atrophy

nd to assess the specificity and independent predictive valuef PET versus MRI measures.

. Methods

.1. Subjects

Twenty-three community-dwelling subjects weressessed with standard neurological and psychologicalxams. Those with a history of stroke, mental illness, seriousead injury, and non-AD diseases that could affect cognitiveerformance were excluded. Although subjects were notxcluded for presence of white matter lesions, which arerequently present in AD, our composite cognitive testcore was not significantly correlated with the occurrencef at least one visible white matter hyperintensity in a2-weighted MRI image (p = 0.868). Seven subjects metiagnostic criteria for AD, 6 for amnestic mild cognitivempairment (MCI) (Petersen, 2004), and 10 were cognitivelyntact controls, although some had typical age-related

emory complaints. Individuals with AD or amnestic MCI

U

Please cite this article in press as: Braskie, M.N., et al., Plaque and tangleNeurobiol Aging (2008), doi:10.1016/j.neurobiolaging.2008.09.012

ere diagnosed using previously published standard diag-ostic criteria (American Psychological Association, 2000;etersen, 2004). Control subjects did not meet diagnosticriteria for MCI or AD. Diagnostic groups did not differ

ccao

PR

OO

F

PRESSf Aging xxx (2008) xxx–xxx

ignificantly in age, education, sex, or Hamilton Depressionating Scale (21 item) (Hamilton, 1960). To maximize the

ikelihood of finding cognitively intact subjects who havearly AD-like pathology, we included several cognitivelyntact subjects who had a family history of dementia6), were apolipoprotein E e4 (APOE4) carriers (4), oroth (2).

All but seven of the subjects included in the current studylso were reported in a study previously published (Smallt al., 2006). Subjects from that study who lacked high-uality T1-weighted MPRAGE MRI scans (Siemens 3T)ere excluded from our study.We obtained informed written consent from all subjects

r their medical proxies, and the study was approved by thenstitutional Review Board of the University of California,os Angeles (UCLA).

.2. Neuropsychological testing

Participants underwent a full battery of neuropsychologi-al tests whose scores were converted to age-adjusted Z scoressing established age-based normative data for each test.e created a composite average cognitive Z score composed

f three tests of episodic memory and three tests of frontalobe function (Table 1). Selected tasks had a wide range ofesponses and high sensitivity to AD-related changes (Jonest al., 2006; Tabert et al., 2006) and assessed a range of men-al function including memory, processing speed, attentionalnd response control, and phonemic fluency.

Among the AD patients, cognitive impairment preventedwo subjects from completing the Stroop C Interference task,ne from completing the Buschke-Fuld Selective Remind-ng task, and one from completing the WMS Verbal Pairedssociates Immediate task. In these cases, we substituted theorst Z score on that task by any study subject for the incom-lete test score. No subject failed to complete more than onef the tasks.

Age-adjusted Z scores on the well-established tests per-ormed by subjects were correlated with the compositecores they composed (Pearson’s r values range = 0.67–0.90;≤ 0.0005 for all tests), suggesting that the composite scoreas a valid indicator of memory ability and frontal lobe

unction in these participants. Averaging Z scores from thendividual tests into a composite score reduced the likelihoodhat an aberrant score on any one task would unduly influ-nce the results of the study. The composite score and theMSE score were also correlated (r = 0.73, p < 0.0001). The

omposite score, however, offered a greater range of val-es than the MMSE, allowing us to distinguish gradationsf cognitive ability even in cognitively intact older adults.

covariance matrix of the age-adjusted Z scores for eachest across all subjects showed that all the neuropsychologi-

imaging and cognition in normal aging and Alzheimer’s disease.

al scores included in the composite score were significantly 147

orrelated with one another (r = 0.42–0.83), suggesting that 148

veraging these scores provided a reasonable single measure 149

f global cognition.

ED

F

ARTICLE IN PRESSNBA 7198 1–10

M.N. Braskie et al. / Neurobiology of Aging xxx (2008) xxx–xxx 3

Table 1Demographics and clinical characteristics.

AD MCI Controls

No. of participantsa 7 6 10Average ageb 77 ± 10.4 73 ± 12.8 73 ± 10.4Males/females 3/4 2/4 3/7Years of educationb 16 ± 2 17 ± 3 17 ± 3Apolipoprotein E �4 carriersc 4 (57%) 2 (33%) 4 (40%)Known family history of dementiad 2 (29%) 5 (83%) 6 (60%)Mini-Mental State Exam scoreb 23 ± 2e 28 ± 1e 29 ± 1Hamilton Depression Rating Scale, 21 itemb 3 ± 3 4 ± 3 2 ± 2Cognitive composite Z scoreb,f −2.2 ± 0.9e −1.0 ± 0.4e 0.7 ± 0.8e

a Two participants with AD and one with MCI were excluded from the cortical thickness comparison because of technical difficulties with their MRI scans.b Listed as mean ± standard deviation.c APOE genotype was not available for two subjects with MCI.d Family history defined as having a parent, grandparent or sibling with dementia.e Significant difference between starred items.

Fuld SeW e functioA

2150

151

r152

(153

e154

0155

2156

2157

158

d159

O160

S161

a162

m163

t164

t165

a166

t167

i168

a169

s170

t171

f172

173

a174

v175

h176

g177

m178

a179

180

a181

t182

t183

T 184

w 185

186

2 187

e 188

o 189

O 190

g 191

m 192

193

i 194

t 195

a 196

a 197

a 198

r 199

( 200

2 201

202

[ 203

( 204

v 205

f 206

K 207

s 208

( 209

a 210

5 211

a 212

2 213

NC

OR

RE

CT

f Mean age-adjusted Z scores for three tests of verbal memory (Buschke-MS verbal paired associates immediate total) and three tests of frontal lobssociation Test: FAS).

.3. MRI protocol

We obtained sagittal T1-weighted magnetization preparedapid acquisition gradient-echo (MPRAGE) volumetric scans3T Siemens Allegra MRI): repetition time (TR) 2300 ms;cho time (TE) 2.93 ms; 160 slices; slice thickness 1 mm/skip.5 mm; in-plane voxel size 1.3 mm × 1.3 mm; field of view56 × 256; flip angle 8◦.

.4. MRI image processing

MRI scans were processed using a sequence of stepsescribed previously (Thompson et al., 2004). We used thexford Centre for Functional MRI of the Brain (FMRIB)oftware Library (FSL) Brain Extraction Tool (Smith, 2002)nd “FSL view” to create and manually refine individual brainasks, which were then applied to exclude non-brain mat-

er. The raw data were scaled and spatially normalized tohe International Consortium for Brain Mapping ICBM53verage brain imaging template with a nine-parameter linearransformation (Collins et al., 1994). Magnetic susceptibil-ty artifacts and image non-uniformities were reduced usingregularized tricubic B-spline approach. We automatically

egmented each resulting image into gray matter, white mat-er, and cerebrospinal fluid using a Gaussian mixture modelor their MRI signal values (Shattuck et al., 2001).

A three-dimensional cortical surface model was extractedutomatically from each subject’s scan as described pre-iously (MacDonald et al., 2000). Three-dimensionalemispheric reconstructions were created, onto which a sin-le trained researcher, blind to [18F]FDDNP-PET results,anually traced neuroanatomical landmarks. Inter-rater reli-

bility has been reported previously (Sowell et al., 2000).

U

Please cite this article in press as: Braskie, M.N., et al., Plaque and tangleNeurobiol Aging (2008), doi:10.1016/j.neurobiolaging.2008.09.012

We created an inter-subject 3D average cortical model,s detailed previously (Thompson et al., 2003), by flatteninghe cortex and gyral landmarks into two-dimensional space,hen warping all landmarks into alignment across subjects.

f(

PR

OOlective Reminding total recall, WMS logical memory II delayed total, and

n (Stroop C interference, WAIS-R digit symbol, and Controlled Oral Word

his method allows voxel by voxel averaging of these imagesithin each delineated region across subjects.After extracting gray matter volumes (Shattuck et al.,

001) and spatially registering them to the hemispheric mod-ls, we calculated cortical gray matter thickness at each pointf the brain surface (Sapiro, 2001; Thompson et al., 2004).ur approach defines thickness as the distance from the innerray–white boundary to the closest point on the outer grayatter surface.After supersampling the image data to create 0.33 mm

sotropic voxels, a 3D Eikonal equation was applied onlyo gray matter voxels, and a smoothing kernel was used toverage gray matter thickness values within a 15-mm spheret each cortical surface point (Hayashi et al., 2002; Sowell etl., 2004). Test–retest reliability of cortical thickness usingepeated scanning and analysis has been reported previouslySowell et al., 2004).

.5. [18F]FDDNP-PET scanning

Each subject received a bolus injection of 320–550 MBq18F]FDDNP, prepared at very high specific activities>37 GBq/mmol) (Liu et al., 2007), through an in-dwellingenous catheter. Consecutive dynamic PET scans were per-ormed for 2 h on an EXACT HR+ tomograph (Siemens-CTI,noxville, TN) while participants lay supine. Sixty-three

lices were collected parallel to the orbito-meatal line2.24 mm plane separation). The scans were decay correctednd reconstructed using filtered back-projection (Hann filter,.5 mm FWHM) with correction for scatter and measuredttenuation.

.6. [18F]FDDNP-PET image analysis

imaging and cognition in normal aging and Alzheimer’s disease.

We performed Logan graphical analysis (using PET 214

rames 30–125 min) to create distribution volume ratio 215

DVR) parametric images of relative [18F]FDDNP-PET 216

D P

RO

OF

IN PRESSNBA 7198 1–10

4 iology of Aging xxx (2008) xxx–xxx

b217

i218

l219

d220

d221

e222

223

a224

d225

v226

a227

s228

c229

230

v231

m232

a233

r234

2235

2236

237

b238

p239

t240

p241

q242

a243

a244

i245

o246

w247

c248

c249

2250

c251

252

i253

d254

o255

t256

a257

t258

a259

e260

2261

262

t263

[264

b265

f

Fig. 1. Increasing [18F]FDDNP and declining cognition. Cortical maps showthat as [18F]FDDNP signal (DVR) increased at each cortical surface point(across all subjects), cognitive performance decreased (see Section 2 forcalculation of composite test scores). Regions of significant correlation(p < 0.05), color-coded in red, follow the profile of tangle and plaque accu- Q2

mulation characteristic of mild AD, as established in large-scale post mortemmapping studies (Braak and Braak, 1991). Corresponding maps did not showregions where both [18F]FDDNP signal and cognition increased together (notstv

3 266

3 267

p 268

269

w 270

c 271

w 272

r 273

p 274

f 275

m 276

o 277

a 278

w 279

c 280

a 281

NC

OR

RE

CTE

ARTICLEM.N. Braskie et al. / Neurob

inding (manifested as PET signal). In these parametricmages, the value at each point represented the slope of theinear portion of the Logan plot (or DVR), calculated as theistribution volume of the tracer at that point divided by theistribution volume in the cerebellar reference region (Logant al., 1996), where little [18F]FDDNP binding was expected.

PET images were co-registered to the MRI images usingmutual information-based rigid body transformation. As

escribed previously (Protas et al., 2005), we assigned PETalues to each cortical surface vertex by computing the aver-ge value of the DVR image in a kernel of radius 7 mmurrounding each cortical mesh point, while excluding extra-ortical voxels.

For each subject, we also obtained average [18F]FDDNPalues within ROIs that included bilateral frontal, parietal,edial temporal, and lateral temporal brain regions as well

s posterior cingulate gyrus. These were traced on the co-egistered MRI scans as described previously (Small et al.,006).

.7. Statistical analysis

Statistical maps were generated indicating the correlationetween [18F]FDDNP-PET signal at each cortical surfaceoint and each subject’s composite test score. Different sta-istical models were fitted at each surface vertex, as detailedreviously (Thompson et al., 2004), including linear anduadratic terms in the model, and retaining only terms withsignificant fit. The resulting significance values associ-

ting [18F]FDDNP signal and cognitive performance werendicated by a color code plotted at each surface pointn the average cortex. After obtaining average ROI valuese additionally performed non-parametric Spearman’s rank

orrelations between regional [18F]FDDNP values and theomposite cognitive scores.

.8. Permutation testing and multiple comparisonsorrection

The significance map was corrected for multiple compar-sons by permutation testing using a threshold of p < 0.05 toefine a suprathreshold region. The area of this suprathresh-ld region was compared with a null distribution of statisticshat occurred by chance when test scores were randomlyssigned to subjects in 100,000 random simulations. Permu-ation testing has been used widely in imaging and providesglobal p value for the observed pattern of effects (Bullmoret al., 1999; Thompson et al., 2003).

.9. Hypotheses

We used one-sided hypothesis testing, predicting a priori

U

Please cite this article in press as: Braskie, M.N., et al., Plaque and tangleNeurobiol Aging (2008), doi:10.1016/j.neurobiolaging.2008.09.012

hat subjects with lower composite scores would show greater18F]FDDNP signal, particularly in those cortical regions thatoth degenerate early in Alzheimer’s disease and are criticalor cognitive performance.

c

dH

hown here; maps are blue with p > 0.05 at all voxels). (For interpretation ofhe references to color in this figure legend, the reader is referred to the webersion of the article.)

. Results

.1. [18F]FDDNP-PET signal and cognitiveerformance

[18F]FDDNP-PET signal was significantly higher acrossidespread cortical regions in subjects with poorer neuropsy-

hological test performance (Fig. 1). Strong correlationsere seen in the entorhinal, orbitofrontal, and lateral tempo-

al cortices, temporoparietal and perisylvian language areas,arietal association cortices, and much of the dorsolateral pre-rontal cortex. Correlated regions were bilateral, except in theedial wall, where above threshold correlations were visible

nly in the left hemisphere in restricted frontal pole regions,lthough the difference in signal between hemispheresas not significant. As expected, primary sensorimotor

ortices (e.g., central and pre-central gyri) did not shown association between [18F]FDDNP signal variation and

imaging and cognition in normal aging and Alzheimer’s disease.

ognition. 282

Notably, cingulate and paralimbic belts did not show 283

etectable correlations on the medial hemispheric surface. 284

owever, significant correlations were found in the medial 285

TED

PR

OO

F

ARTICLE IN PRESSNBA 7198 1–10

M.N. Braskie et al. / Neurobiology of Aging xxx (2008) xxx–xxx 5

Fig. 2. [18F]FDDNP and cognition in cognitively intact subjects. Frontalregions of the right lateral cortical surface showed elevated [18F]FDDNPsignal in healthy normal subjects with poorer cognitive performance whencorrelation analyses were restricted to cognitively intact subjects (controlsonly). Cortical maps show similar correlations between [18F]FDDNP signaland composite cognitive score as in Fig. 1, but in a more restricted region.Dorsolateral prefrontal regions are implicated in executive function, whichis among the functions required for optimal performance on the neuropsy-chological tests. Additional correlations were found in parietal associationareas; the left lateral hemisphere did not show broad regions with correla-tions (shown on the right panels here). Corroborating these results, mapso 18

s

a286

t287

288

o289

a290

o291

(292

i293

294

[295

t296

a297

t298

r299

p300

o301

a302

w303

(304

Fig. 3. Time-lapse films calibrating maps of [18F]FDDNP signal versus cog-nition. Projected mean [18F]FDDNP signal (DVR) can be calculated forvarious cognition scores based on the relationship of [18F]FDDNP signalwith cognition in the subjects studied. Here we show projected signal forsubjects who scored (A) high normal (2 standard deviations above age-normal), (B) at age-normal levels (Z score = 0), (C) low normal (2 standarddeviations lower than age-normal), and (D) at impaired levels (4 standarddeviations below age-normal). Red colors denote regions in which greaterpredicted [18F]FDDNP signal is associated with lower cognitive Z scoresat each cortical point based on a nonlinear spatially varying model. Theparameterization of disease stages is based on cross-sectional data, but it isplausible that a comparable trajectory would be followed for each cognitives(i

305

w 306

t 307

[ 308

o 309

j 310

v 311

d 312

M 313

v 314

o 315

( 316

ing two standard deviations from the mean lies within the 317

UN

CO

RR

ECf regions in which [ F]FDDNP signal and cognition in cognitively intact

ubjects increased together showed no effects, as expected (not shown).

nd lateral temporal cortices, where plaques and tangles arehought to be deposited first (Braak and Braak, 1997).

To determine whether these patterns could have beenbserved by chance, permutation tests were conducted tossign global p values to the maps. At a voxel-level thresholdf p = 0.05, the corrected significance values were p = 0.004left hemisphere) and p = 0.008 (right hemisphere), suggest-ng that the results were unlikely to be due to chance.

When only cognitively intact subjects were considered,18F]FDDNP signal was higher in restricted right frontal cor-ices, and in some parietal association areas in those havinglower composite cognitive score (Fig. 2). At a voxel-level

hreshold of p = 0.05, the corrected significance value in theight hemisphere was p = 0.031 after permutation testing waserformed. In the left hemisphere, correlations were detectednly in isolated anterior prefrontal regions on the medial wall

Please cite this article in press as: Braskie, M.N., et al., Plaque and tangleNeurobiol Aging (2008), doi:10.1016/j.neurobiolaging.2008.09.012

nd in some of the occipital lobe medial surface, but theseere not significant after correction for multiple comparisons

p = 0.129).

9ns

tage in an individual subject, albeit with variable timing across individuals.For interpretation of the references to color in this figure legend, the readers referred to the web version of the article.)

Given that the estimation of a nonlinear model relationas significant both pointwise and overall after permuta-

ion testing, it is possible to predict the cortical profile of18F]FDDNP signal that would be expected at each levelf cognitive deterioration. Based on the 65,536 fitted tra-ectories for [18F]FDDNP signal at the cortical surfaceertices across all subjects, we created an image of pre-icted [18F]FDDNP signal for each composite score value.aps of mean [18F]FDDNP signal were generated for indi-

iduals at each end of the normal range and in the middlef the normal range (maps corresponding to Z = +2, 0, −2)Fig. 3). These Z scores were selected because a person scor-

imaging and cognition in normal aging and Alzheimer’s disease.

5th percentile for normal performance. Notably, there is a 318

euroanatomical spread in the areas of higher [18F]FDDNP 319

ignal even within the normal range, with greatest signal 320

D P

RO

F

ARTICLE IN PRESSNBA 7198 1–10

6 M.N. Braskie et al. / Neurobiology of Aging xxx (2008) xxx–xxx

F tionshii terior ch

i321

e322

s323

m324

o325

a326

i327

h328

329

w330

t331

t332

r333

334

o335

c336

a337

p338

p339

w340

w341

c342

t343

c344

a345

i346

[347

(348

l349

i350

3351

352

[353

t 354

signal with cortical thickness. A map of the pointwise rela- 355

tionship between cortical thickness and [18F]FDDNP signal 356

(Fig. 5) showed that such correlations were low and non- 357

Fig. 5. PET-MRI correlation. (A) Maps of the pointwise significancebetween [18F]FDDNP signal and cortical thickness are shown in millimeters,in 20 subjects. (B) Maps of the pointwise correlation (Pearson’s r value) areshown for the same relationship. These maps show that there is no detectableresidual correlation between PET and cortical thickness, and therefore thePET effects observed in Figs. 1 and 2 are not confounded by the effects of

NC

OR

RE

CTE

ig. 4. ROI analysis. Scatterplot graphs and linear regressions show the relan (A) frontal lobe, (B) parietal lobe, (C) lateral temporal lobe, and (D) posemispheres was pooled for these analyses.

n the medial temporal, entorhinal, orbitofrontal, and lat-ral temporal cortices. The increase in frontal [18F]FDDNPignal is characteristic of subjects that are outside the nor-al range of cognition (see map corresponding to Z = −4,

r scores 4 standard deviations below the mean). Thedvancement of pathology can be seen in the accompany-ng Supplementary Online Data, observable on the internet atttp://www.loni.ucla.edu/∼thompson/FDDNP/video.html.

In this film, frames corresponding to Z scores +2 to −4ere estimated from the statistical model fitted at each cor-

ical surface vertex, and were concatenated at 30 frames/so create a digital animation of the path of pathology withespect to different levels of cognitive performance.

To supplement our cortical surface map results with thosebtained using an ROI analysis, we performed Spearman rankorrelations between composite cognitive scores and aver-ge [18F]FDDNP signal DVR values in several ROIs: frontal,arietal, medial temporal and lateral temporal regions, and theosterior cingulate gyrus. As hypothesized, when all subjectsere included, average [18F]FDDNP values in each regionere significantly lower in those having higher composite

ognitive scores (p < 0.05). Graphs in Fig. 4 demonstratehese relationships in four of the ROIs. In contrast, when onlyognitively intact subjects were considered using the ROInalysis (rather than the voxel-wise approach), the compos-te cognitive score was significantly correlated with average18F]FDDNP signal only in the posterior cingulate gyrusrho = −0.76; p = 0.01) (not shown). The posterior cingu-ate gyrus is a region thought to show metabolic changesn subjects at risk for AD (Small et al., 2000).

U

Please cite this article in press as: Braskie, M.N., et al., Plaque and tangleNeurobiol Aging (2008), doi:10.1016/j.neurobiolaging.2008.09.012

.2. Correlations with cortical thickness

In order to determine the independent predictive value of18F]FDDNP versus MRI-derived measures, we examined

aiFca

Op between the composite cognitive scores and average [18F]FDDNP signalingulate gyrus across all subjects. Average [18F]FDDNP signal from both

he separate relationships of both cognition and [18F]FDDNP

imaging and cognition in normal aging and Alzheimer’s disease.

trophy. Thickness and PET signal are orthogonal (not correlated) exceptn very small, scattered regions not overlapping with the main effects inigs. 1 and 2. The left frontal pole effects cover no more than 5% of theortex, so they are likely false positives (5% false positives are expected inny null map thresholded at p = 0.05).

ED

PR

OO

F

ARTICLE IN PRESSNBA 7198 1–10

M.N. Braskie et al. / Neurobiology of Aging xxx (2008) xxx–xxx 7

Fig. 6. Maps of post mortem amyloid load and cortical thickness. (A) Images, which were adapted from Fig. 1 of a previously published paper (Braak and Braak,1991) (copyright Springer-Verlag, 1991), with kind permission of the authors, Springer Science, and Business Media, display the classic post mortem amyloiddeposition pattern. Images displayed in (B) also have been previously published (Thompson et al., 2003) (copyright 2003 by the Society for Neuroscience),and show thinner average gray matter in AD patients (mean MMSE = 18) compared with controls at an initial scan (top right) and a follow-up scan 1.5 yearsl ). The pp se regio[ from th

s358

m359

c360

e361

i362

fi363

t364

[365

t366

a367

(368

369

r370

i371

T372

m373

374

n375

r376

4377

378

[379

i 380

n 381

r 382

a 383

g 384

i 385

n 386

m 387

B 388

[ 389

r 390

fi 391

392

w 393

t 394

w 395

p 396

p 397

p 398

v 399

r 400

NC

OR

RE

CT

ater, when the mean MMSE for AD patients dropped to 13 (bottom rightreviously in patients with mild to moderate AD (upper panels) mirrors tho18F]FDDNP signal in Fig. 1 of the current study [adapted with permission

ignificant throughout the cortex. Note that there is no knownethod to ascribe a global significance value to a map of

orrelations between two spatially varying signals, as thexchangeability assumption, required for permutation test-ng, does not apply, and there is no associated Gaussianeld theory yet developed to give such a global p value for

wo correlated random processes. Interestingly, the pattern of18F]FDDNP signal in this study did match the pattern of cor-ical thinning found previously in a population having moredvanced AD than the subjects in the current study (Fig. 6B)Thompson et al., 2003).

Because reduced cortical gray matter thickness has beeneported to be associated with AD, MCI, and poorer cognitionn prior studies (Apostolova et al., 2006; Singh et al., 2006;hompson et al., 2004), we also correlated cortical thicknesseasures with cognitive performance in our sample.As with [18F]FDDNP signal, cognitive performance was

ot associated with cortical thickness in this sample (cor-ected p > 0.05; both brain hemispheres).

U

Please cite this article in press as: Braskie, M.N., et al., Plaque and tangleNeurobiol Aging (2008), doi:10.1016/j.neurobiolaging.2008.09.012

. Discussion

Cognitive performance was significantly correlated with18F]FDDNP signal in right frontal and parietal regions

p 401

ots

attern of cortical thinning and post mortem amyloid deposition displayedns demonstrating a significant correlation between cognitive function and

e authors and publishers].

n cognitively intact subjects, suggesting that some cog-itive aging that is considered age-normal actually mayeflect pathological brain changes, particularly in subjectst risk for AD. That is not to say that plaque and tan-le accumulations are a result of aging or are presentn all older adults, but rather that in some people diag-osed as having normal cognitive aging, plaques and tanglesay be associated with their subtle cognitive decline.oth the lack of significant positive correlations between

18F]FDDNP signal and composite cognitive scores and theigor of permutation testing supported the validity of ourndings.

It is important to note that even though [18F]FDDNP signalas elevated in the medial temporal lobe in some cogni-

ively intact adults, it was primarily in the frontal cortexhere [18F]FDDNP signal distinguished those controls whoerformed better on certain cognitive tasks from those whoerformed worse. Permutation testing without a restricted ariori search region appropriately applies a somewhat conser-ative correction for false positives, so the correlation in theight hemisphere but not the left may reflect limited statisticalower rather than true hemispheric specificity. The specificity

imaging and cognition in normal aging and Alzheimer’s disease.

f the relationship between [18F]FDDNP signal and cogni- 402

ion for right frontal cortex will be tested specifically in future 403

tudies having larger sample sizes. 404

D

INNBA 7198 1–10

8 iology o

405

s406

t407

C408

w409

o410

t411

b412

1413

i414

t415

416

c417

n418

b419

i420

s421

[422

p423

r424

i425

r426

t427

m428

s429

v430

n431

e432

s433

n434

b435

w436

t437

a438

439

v440

f441

r442

e443

t444

e445

N446

d447

b448

H449

n450

t451

a452

s453

t454

d455

c456

c457

g458

f459

a460

b 461

2 462

p 463

t 464

t 465

466

c 467

i 468

b 469

i 470

t 471

i 472

s 473

n 474

m 475

l 476

c 477

p 478

j 479

e 480

c 481

n 482

m 483

T 484

b 485

n 486

f 487

t 488

d 489

b 490

M 491

f 492

d 493

494

i 495

t 496

a 497

l 498

t 499

t 500

n 501

d 502

c 503

o 504

a 505

v 506

e 507

s 508

n 509

t 510

a 511

n 512

NC

OR

RE

CTE

ARTICLEM.N. Braskie et al. / Neurob

The correlations we observed in the cognitively intactubjects alone were maintained and expanded when cogni-ively impaired subjects were also included in the sample.ognitive performance across all subjects was correlatedith [18F]FDDNP signal in inferior and lateral temporal,rbitofrontal, dorsolateral prefrontal, and parietal associa-ion cortices—regions with the greatest plaque and tangleurden in histopathological studies of AD (Braak and Braak,991). The anatomical agreement is striking between thesen vivo maps and the well-established post mortem maps forhe staging of AD (Braak and Braak, 1991).

ROI analyses yielded significant relationships between theomposite cognitive scores and the average [18F]FDDNP sig-al in all regions examined when all subjects were included,ut only in the posterior cingulate gyrus when cognitivelyntact subjects alone were considered. Given the reducedample size and the restricted range of variability in the18F]FDDNP signal within the control group, it is not sur-rising that the controls alone did not demonstrate significantelationships between [18F]FDDNP signal and the compos-te cognitive score in several of the ROIs. When there is aestricted range of variability in the [18F]FDDNP signal (ashere is within the control group), a voxel-wise approach

ay provide advantages over an ROI approach (in whichignal is averaged across significant and non-significantoxels) for detecting relationships between [18F]FDDNP sig-al and cognition. It is interesting to note, however, thatven using these averaged ROIs to determine [18F]FDDNPignal, the graphs in Fig. 4 demonstrate that there doot appear to be outlying data points in the relationshipsetween [18F]FDDNP signal and composite cognitive scoresithin the controls, lending support to the significant rela-

ionships we found using a voxel-wise statistical mappingpproach.

The relationship of plaques and tangles to AD is contro-ersial. Both must accompany specific cognitive impairmentor a definitive AD diagnosis (McKhann et al., 1984). Someesearchers believe that plaques or tangles cause the dis-ase (Binder et al., 2005; Selkoe, 2001); others contend thathese merely tend to co-occur with other more causative dis-ase processes (Castellani et al., 2006; Watson et al., 2005).eurofibrillary tangle density correlates more strongly withisease severity and neuronal death than does total plaqueurden (Berg et al., 1998; Giannakopoulos et al., 2003).owever, even in studies in which total plaque load didot correlate with disease severity, a simple comparison ofotal plaque load in AD subjects versus controls (rather than

correlation with severity of dementia) showed that ADubjects had on average more extensive plaques than con-rols (Bouras et al., 1994; Gomez-Isla et al., 1996). Theseata suggest that both plaques and tangles are good indi-ators of disease processes, regardless of whether they are

U

Please cite this article in press as: Braskie, M.N., et al., Plaque and tangleNeurobiol Aging (2008), doi:10.1016/j.neurobiolaging.2008.09.012

ausative factors in AD. Although we are unable to distin-uish [18F]FDDNP signal associated with amyloid plaquesrom that associated with tau neurofibrillary tangles in vivo,

recent study compared [18F]FDDNP-PET scanning and

t 513

d

f

PR

OO

F

PRESSf Aging xxx (2008) xxx–xxx

rain autopsy assessment in the same patient (Small et al.,006). In that study, [18F]FDDNP signal in the medial tem-oral lobe was mainly associated with tau pathology whereashat in other areas of the brain was overwhelmingly relatedo amyloid plaque deposition.

We did not find a significant correlation between corti-al thickness and either [18F]FDDNP signal or cognitionn the current study. However, the significant correlationsetween [18F]FDDNP signal distribution and cognitive abil-ty in the current study were evident in the same regionshat showed cortical thinning related to more advanced ADn our prior studies (Fig. 5B) (Thompson et al., 2003). Asuch, the pattern of cortical [18F]FDDNP signal in this cog-itively intact and mildly affected population very closelyatches the topography of cortical thinning known to appear

ater, albeit with a substantial time-lag. Previous studies thatonsidered cognition and cortical thickness either focusedrimarily on cognitively impaired subjects or had larger sub-ect samples than we had in the current study (Apostolovat al., 2006; Lerch et al., 2005; Thompson et al., 2004). Inontrast, nearly half of the subjects in the current study wereormal controls who would not be expected to show anythingore than the most subtle pathology-related cortical thinning.his early distribution of plaques and tangles may thereforee followed by MRI-detectable cortical thinning only wheneuronal damage has become more extensive than is typicallyound in cognitively intact older adults. Our results suggesthat plaque and tangle deposition occurs early and precedesetectable changes in cortical structure, so [18F]FDDNP maye more sensitive to early cognitive changes than structuralRI measures are, and therefore may offer greater power

or disease detection, at least during the early stages of theisease process.

Without partial volume correction, PET measures arenfluenced by cortical atrophy, which reduces the gray mat-er volume emitting radioisotope signals, resulting in signalttenuation. However, partial volume correction is arguablyess critical for interpretation of [18F]FDDNP-PET scanshan for metabolic or perfusion PET images, as the diseaseends to elevate [18F]FDDNP signal and reduce cortical thick-ess. Therefore, any atrophic effect works against finding aisease-associated PET signal increase, and PET increasesannot reasonably be attributed to cortical thinning; usef uncorrected values is, therefore, a slightly conservativepproach. It also avoids the risk of overcorrecting the signalalues, which could occur if the partial volume model was notxactly correct, and ensures that any observed [18F]FDDNPignal increase can be interpreted as related to the ligand andot to structural atrophy. Future empirical estimation of par-ial volume models for different gray/white matter fractionsnd local cortical geometries may increase the signal-to-oise ratio for detecting correlations with cognition withhis ligand, so the current approach should be considered as

imaging and cognition in normal aging and Alzheimer’s disease.

eliberately conservative. 514

Eighteen of our 23 subjects were APOE4+, had a known 515

amily history of dementia, or both. Because our subjects 516

ED

INNBA 7198 1–10

iology o

w517

h518

a519

a520

f521

i522

s523

l524

s525

C526

527

P528

A529

l530

B531

i532

N533

i534

A535

536

M537

B538

a539

m540

H541

(542

M543

J544

t545

C546

R547

R548

b549

O550

J551

T552

R553

v554

f555

o556

f557

A558

M559

560

D561

r562

G563

t564

H565

F

R 566

A 567

568

569

570

571

A 572

Q1 573

574

575

A 576

577

A 578

579

580

581

B 582

583

584

585

586

587

B 588

589

590

B 591

592

593

594

595

B 596

597

B 598

599

B 600

601

602

603

604

B 605

606

607

608

C 609

610

611

C 612

613

614

G 615

616

617

618

G 619

620

621

622

H 623

624

H 625

626

NC

OR

RE

CT

ARTICLEM.N. Braskie et al. / Neurob

ere at high risk for AD and were highly educated, those whoad lower memory ability on certain tasks than their same-ge peers were more likely than the general population to beffected by a pathological condition. Including participantsrom backgrounds representative of the general populationn future studies may help to further elucidate these relation-hips. Finally, the results presented here are cross-sectional;ongitudinal follow-up is needed to determine which controlubjects will eventually develop AD.

onflicts of interest

The University of California, Los Angeles, owns a U.S.atent (6,274,119) entitled “Methods for Labeling Beta-myloid Plaques and Neurofibrillary Tangles,” which is

icensed to Siemens. Drs. Small, Huang, Satyamurthy, andarrio are among the inventors and each receives royalties

n regard to the application of the FDDNP-PET radioligand.one of the other authors have real or perceived conflicts of

nterest.

cknowledgements

National Institute on Aging, the National Library ofedicine, the National Institute for Biomedical Imaging andioengineering, the National Center for Research Resources,nd the National Institute for Child Health and Develop-ent (AG016570, LM05639, EB01651, RR019771, andD050735) to PMT; National Institutes of Health grants

P01-AG024831, AG13308, P50 AG 16570, MH/AG58156,H52453, AG10123, and M01-RR00865) to GWS and

RB; the Department of Energy (DE-FC03-87-ER60615),he General Clinical Research Centers Program, the RotaryART Fund, the Alzheimer’s Association, the Fran anday Stark Foundation Fund for Alzheimer’s Diseaseesearch, the Ahmanson Foundation, the Larry L. Hill-lom Foundation, the Lovelace Foundation, the Judithlenick Elgart Fund for Research on Brain Aging, the

ohn D. French Foundation for Alzheimer’s Research, theamkin Foundation, and the National Center for Researchesources grants (RR13642 and RR021813) to AWT; Indi-idual National Research Service Award (F31 NS45425)rom the National Institutes of Health/National Institutef Neurological Disorders and Stroke and a scholarshiprom ARCS Foundation, Inc./The John Douglas Frenchlzheimer Foundation (with the Erteszek Foundation) toNB.We are indebted to Ms. Andrea Kaplan, Ms. Deborah

orsey, and Ms. Teresann Crowe-Lear for their help inecruiting volunteers and coordinating the study, and to Ms.

U

Please cite this article in press as: Braskie, M.N., et al., Plaque and tangleNeurobiol Aging (2008), doi:10.1016/j.neurobiolaging.2008.09.012

wendolyn Byrd for her help with scheduling of volun-eers and collection of their personal data. We also thank Dr.eiko Braak for allowing us to include his images as part ofig. 6.

627

J

PR

OO

F

PRESSf Aging xxx (2008) xxx–xxx 9

eferences

gdeppa, E.D., Kepe, V., Liu, J., Flores-Torres, S., Satyamurthy, N., Petric,A., Cole, G.M., Small, G.W., Huang, S.C., Barrio, J.R., 2001a. Bindingcharacteristics of radiofluorinated 6-dialkylamino-2-naphthylethylidenederivatives as positron emission tomography imaging probes for beta-amyloid plaques in Alzheimer’s disease. J. Neurosci. 21, RC189.

gdeppa, E.D., Kepe, V., Shoghi-Jadid, K., Satyamurthy, N., Small, G.W.,Petric, A., Vinters, H.V., Huang, S.-C., Barrio, J.R., 2001b. In vivo andin vitro labeling of plaques and tangles in the brain of an Alzheimer’sdisease patient: a case study. J. Nuclear Med. (Suppl 42), 65P.

merican Psychological Association, 2000. Diagnosis and Statistical Man-ual of Mental Disorders DSM-IV-TR (Text Revision). Washington, DC.

postolova, L.G., Lu, P.H., Rogers, S., Dutton, R.A., Hayashi, K.M.,Toga, A.W., Cummings, J.L., Thompson, P.M., 2006. 3D mapping ofmini-mental state examination performance in clinical and preclinicalAlzheimer disease. Alzheimer Dis. Assoc. Disord. 20, 224–231.

erg, L., McKeel Jr., D.W., Miller, J.P., Storandt, M., Rubin, E.H., Morris,J.C., Baty, J., Coats, M., Norton, J., Goate, A.M., Price, J.L., Gearing,M., Mirra, S.S., Saunders, A.M., 1998. Clinicopathologic studies in cog-nitively healthy aging and Alzheimer’s disease: relation of histologicmarkers to dementia severity, age, sex, and apolipoprotein E genotype.Arch. Neurol. 55, 326–335.

inder, L.I., Guillozet-Bongaarts, A.L., Garcia-Sierra, F., Berry, R.W., 2005.Tau, tangles, and Alzheimer’s disease. Biochim. Biophys. Acta 1739,216–223.

ouras, C., Hof, P.R., Giannakopoulos, P., Michel, J.P., Morrison, J.H., 1994.Regional distribution of neurofibrillary tangles and senile plaques inthe cerebral cortex of elderly patients: a quantitative evaluation of aone-year autopsy population from a geriatric hospital. Cereb. Cortex 4,138–150.

raak, H., Braak, E., 1991. Neuropathological stageing of Alzheimer-relatedchanges. Acta Neuropathol. (Berl) 82, 239–259.

raak, H., Braak, E., 1997. Frequency of stages of Alzheimer-related lesionsin different age categories. Neurobiol. Aging 18, 351–357.

uckner, R.L., Snyder, A.Z., Shannon, B.J., LaRossa, G., Sachs, R., Fotenos,A.F., Sheline, Y.I., Klunk, W.E., Mathis, C.A., Morris, J.C., Mintun,M.A., 2005. Molecular, structural, and functional characterization ofAlzheimer’s disease: evidence for a relationship between default activity,amyloid, and memory. J. Neurosci. 25, 7709–7717.

ullmore, E.T., Suckling, J., Overmeyer, S., Rabe-Hesketh, S., Taylor, E.,Brammer, M.J., 1999. Global, voxel, and cluster tests, by theory and per-mutation, for a difference between two groups of structural MR imagesof the brain. IEEE Trans. Med. Imag. 18, 32–42.

astellani, R.J., Lee, H.G., Perry, G., Smith, M.A., 2006. Antioxidant pro-tection and neurodegenerative disease: the role of amyloid-beta and tau.Am. J. Alzheimers Dis. Other Demen. 21, 126–130.

ollins, D.L., Neelin, P., Peters, T.M., Evans, A.C., 1994. Automatic 3Dintersubject registration of MR volumetric data in standardized Talairachspace. J. Comput. Assist. Tomogr. 18, 192–205.

iannakopoulos, P., Herrmann, F.R., Bussiere, T., Bouras, C., Kovari, E.,Perl, D.P., Morrison, J.H., Gold, G., Hof, P.R., 2003. Tangle and neuronnumbers, but not amyloid load, predict cognitive status in Alzheimer’sdisease. Neurology 60, 1495–1500.

omez-Isla, T., Price, J.L., McKeel Jr., D.W., Morris, J.C., Growdon, J.H.,Hyman, B.T., 1996. Profound loss of layer II entorhinal cortex neu-rons occurs in very mild Alzheimer’s disease. J. Neurosci. 16, 4491–4500.

amilton, M., 1960. A rating scale for depression. J. Neurol. Neurosurg.Psychiat. 23, 56–62.

ayashi K.M., Thompson, P.M., Mega, M.S., Zoumalan, C.I., 2002.Medial hemispheric surface gyral pattern delineation in 3D: sur-face curve protocol http://www.loni.ucla.edu/∼khayashi/Public/Medial

imaging and cognition in normal aging and Alzheimer’s disease.

suface/MedialLinesProtocol. 628

ones, S., Laukka, E.J., Backman, L., 2006. Differential verbal fluency 629

deficits in the preclinical stages of Alzheimer’s disease and vascular 630

dementia. Cortex 42, 347–355. 631

D

INNBA 7198 1–10

1 iology o

K632

633

634

635

636

637

L638

639

640

L641

642

643

644

645

L646

647

648

M649

650

651

M652

653

654

655

656

M657

658

659

660

P661

662

P663

664

665

666

S667

668

S669

670

671

S672

673

674

S675

676

S 677

678

679

680

681

682

683

S 684

685

686

687

688

S 689

690

S 691

692

693

S 694

695

696

T 697

698

699

700

701

T 702

703

704

T 705

706

707

708

T 709

710

711

712

713

V 714

715

716

717

TE

ARTICLE0 M.N. Braskie et al. / Neurob

lunk, W.E., Engler, H., Nordberg, A., Wang, Y., Blomqvist, G., Holt, D.P.,Bergstrom, M., Savitcheva, I., Huang, G.F., Estrada, S., Ausen, B., Deb-nath, M.L., Barletta, J., Price, J.C., Sandell, J., Lopresti, B.J., Wall, A.,Koivisto, P., Antoni, G., Mathis, C.A., Langstrom, B., 2004. Imagingbrain amyloid in Alzheimer’s disease with Pittsburgh Compound-B.Ann. Neurol. 55, 306–319.

erch, J.P., Pruessner, J.C., Zijdenbos, A., Hampel, H., Teipel, S.J., Evans,A.C., 2005. Focal decline of cortical thickness in Alzheimer’s diseaseidentified by computational neuroanatomy. Cereb. Cortex 15, 995–1001.

iu, J., Kepe, V., Zabjek, A., Petric, A., Padgett, H.C., Satyamurthy, N.,Barrio, J.R., 2007. High-yield, automated radiosynthesis of 2-(1-{6-[(2-[18F]fluoroethyl)(methyl)amino]-2-naphthyl}ethylidene)malononitrile([18F]FDDNP) ready for animal or human administration. Mol. Imag.Biol. 9, 6–16.

ogan, J., Fowler, J.S., Volkow, N.D., Wang, G.J., Ding, Y.S., Alexoff, D.L.,1996. Distribution volume ratios without blood sampling from graphicalanalysis of PET data. J. Cereb. Blood Flow Metab. 16, 834–840.

acDonald, D., Kabani, N., Avis, D., Evans, A.C., 2000. Automated 3Dextraction of inner and outer surfaces of cerebral cortex from MRI.Neuroimage 12, 340–356.

cKhann, G., Drachman, D., Folstein, M., Katzman, R., Price, D., Stad-lan, E.M., 1984. Clinical diagnosis of Alzheimer’s disease: report of theNINCDS-ADRDA Work Group under the auspices of Department ofHealth and Human Services Task Force on Alzheimer’s disease. Neu-rology 34, 939–944.

intun, M.A., Larossa, G.N., Sheline, Y.I., Dence, C.S., Lee, S.Y., Mach,R.H., Klunk, W.E., Mathis, C.A., DeKosky, S.T., Morris, J.C., 2006.[11C]PIB in a nondemented population: potential antecedent marker ofAlzheimer disease. Neurology 67, 446–452.

etersen, R.C., 2004. Mild cognitive impairment as a diagnostic entity. J.Intern. Med. 256, 183–194.

rotas, H., Thompson, P.M., Hayashi, K.M., Huang, S.C., 2005. Corticalbrain surface mapping for studying partial volume effects in brain FDG-PET images. In: Proceedings of the 52nd Annual Meeting of the Societyfor Nuclear Medicine (SNM), Toronto, Canada, June 18–22, 2005.

apiro, G., 2001. Geometric Partial Differential Equations and Image Anal-ysis. Cambridge University Press.

elkoe, D.J., 2001. Alzheimer’s disease results from the cerebral accumu-lation and cytotoxicity of amyloid beta-protein. J. Alzheimers Dis. 3,75–80.

hattuck, D.W., Sandor-Leahy, S.R., Schaper, K.A., Rottenberg, D.A.,

UN

CO

RR

EC

Please cite this article in press as: Braskie, M.N., et al., Plaque and tangleNeurobiol Aging (2008), doi:10.1016/j.neurobiolaging.2008.09.012

Leahy, R.M., 2001. Magnetic resonance image tissue classification usinga partial volume model. Neuroimage 13, 856–876.

ingh, V., Chertkow, H., Lerch, J.P., Evans, A.C., Dorr, A.E., Kabani, N.J.,2006. Spatial patterns of cortical thinning in mild cognitive impairmentand Alzheimer’s disease. Brain 129, 2885–2893.

W

PR

OO

F

PRESSf Aging xxx (2008) xxx–xxx

mall, G.W., Ercoli, L.M., Silverman, D.H., Huang, S.C., Komo, S.,Bookheimer, S.Y., Lavretsky, H., Miller, K., Siddarth, P., Rasgon, N.L.,Mazziotta, J.C., Saxena, S., Wu, H.M., Mega, M.S., Cummings, J.L.,Saunders, A.M., Pericak-Vance, M.A., Roses, A.D., Barrio, J.R., Phelps,M.E., 2000. Cerebral metabolic and cognitive decline in persons atgenetic risk for Alzheimer’s disease. Proc. Natl. Acad. Sci. U.S.A. 97,6037–6042.

mall, G.W., Kepe, V., Ercoli, L.M., Siddarth, P., Bookheimer, S.Y., Miller,K.J., Lavretsky, H., Burggren, A.C., Cole, G.M., Vinters, H.V., Thomp-son, P.M., Huang, S.C., Satyamurthy, N., Phelps, M.E., Barrio, J.R.,2006. PET of brain amyloid and tau in mild cognitive impairment. N.Engl. J. Med. 355, 2652–2663.

mith, S.M., 2002. Fast robust automated brain extraction. Hum. BrainMapp. 17, 143–155.

owell E.R., Thompson, P.M., Mega, M.S., Zoumalan, C.I., Lindshield, C.J.,Rex, D.E., 2000. Gyral pattern delineation in 3D: surface curve protocol.www.loni.ucla.edu/∼esowell/new sulcvar.html.

owell, E.R., Thompson, P.M., Leonard, C.M., Welcome, S.E., Kan, E.,Toga, A.W., 2004. Longitudinal mapping of cortical thickness and braingrowth in normal children. J. Neurosci. 24 (38), 8223–8231.

abert, M.H., Manly, J.J., Liu, X., Pelton, G.H., Rosenblum, S., Jacobs,M., Zamora, D., Goodkind, M., Bell, K., Stern, Y., Devanand, D.P.,2006. Neuropsychological prediction of conversion to Alzheimer dis-ease in patients with mild cognitive impairment. Arch. Gen. Psychiat.63, 916–924.

hal, D.R., Rub, U., Orantes, M., Braak, H., 2002. Phases of A beta-deposition in the human brain and its relevance for the developmentof AD. Neurology 58, 1791–1800.

hompson, P.M., Hayashi, K.M., de Zubicaray, G., Janke, A.L., Rose, S.E.,Semple, J., Herman, D., Hong, M.S., Dittmer, S.S., Doddrell, D.M.,Toga, A.W., 2003. Dynamics of gray matter loss in Alzheimer’s disease.J. Neurosci. 23, 994–1005.

hompson, P.M., Hayashi, K.M., Sowell, E.R., Gogtay, N., Giedd, J.N.,Rapoport, J.L., de Zubicaray, G.I., Janke, A.L., Rose, S.E., Semple, J.,Doddrell, D.M., Wang, Y., van Erp, T.G., Cannon, T.D., Toga, A.W.,2004. Mapping cortical change in Alzheimer’s disease, brain develop-ment, and schizophrenia. Neuroimage 23 (Suppl 1), S2–18.

erhoeff, N.P., Wilson, A.A., Takeshita, S., Trop, L., Hussey, D., Singh, K.,Kung, H.F., Kung, M.P., Houle, S., 2004. In-vivo imaging of Alzheimerdisease beta-amyloid with [11C]SB-13 PET. Am. J. Geriatr. Psychiat.12, 584–595.

imaging and cognition in normal aging and Alzheimer’s disease.

atson, D., Castano, E., Kokjohn, T.A., Kuo, Y.M., Lyubchenko, Y., Pinsky, 718

D., Connolly Jr., E.S., Esh, C., Luehrs, D.C., Stine, W.B., Rowse, L.M., 719

Emmerling, M.R., Roher, A.E., 2005. Physicochemical characteristics 720

of soluble oligomeric A beta and their pathologic role in Alzheimer’s 721

disease. Neurol. Res. 27, 869–881. 722