ModelBiomarkersADcascadeLancetNeuro10

8/9/2019 ModelBiomarkersADcascadeLancetNeuro10

1/10www.thelancet.com/neurology Vol 9 January 2010 119

Personal View

Hypothetical model of dynamic biomarkers of the

Alzheimers pathological cascadeClifford R Jack Jr, David S Knopman, William J Jagust, Leslie M Shaw, Paul S Aisen, Michael W Weiner, Ronald C Petersen, John Q Trojanowski

Currently available evidence strongly supports the position that the initiating event in Alzheimers disease (AD) isrelated to abnormal processing of-amyloid (A) peptide, ultimately leading to formation of A plaques in the brain.This process occurs while individuals are still cognitively normal. Biomarkers of brain -amyloidosis are reductionsin CSF A42 and increased amyloid PET tracer retention. After a lag period, which varies from patient to patient,neuronal dysfunction and neurodegeneration become the dominant pathological processes. Biomarkers of neuronalinjury and neurodegeneration are increased CSF tau and structural MRI measures of cerebral atrophy.Neurodegeneration is accompanied by synaptic dysfunction, which is indicated by decreased fluorodeoxyglucoseuptake on PET. We propose a model that relates disease stage to AD biomarkers in which A biomarkers becomeabnormal first, before neurodegenerative biomarkers and cognitive symptoms, and neurodegenerative biomarkersbecome abnormal later, and correlate with clinical symptom severity.

IntroductionAs recently as 2 to 3 decades ago, a compartmentalisedmodel of Alzheimers disease (AD) was widely accepted.The view at that time was that people either had ADpathological changes, in which case they had dementia,or they did not have such changes and were cognitivelynormal. In the meantime, a revised view of the diseasehas been developed, in which both AD pathologicalprocesses and clinical decline occur gradually, withdementia representing the end stage of many years ofaccumulation of these pathological changes. Anadditional feature of the current view of AD is that thesechanges begin to develop decades before the earliest

clinical symptoms occur.Biomarkers, both chemical and imaging, are indicatorsof specific changes that characterise AD in vivo. Evidencesuggests that these AD biomarkers do not reach abnormallevels or peak simultaneously but do so in an orderedmanner. Measurement of these biomarkers in longi-tudinal observational studies is now commonplace,enabling investigators to establish the correct ordering ofthe relevant biomarkers and their relationships to clinicalsymptoms.

For biomarkers of AD to be used eectively for diseasestaging, the time-dependent ordering of biomarkers mustbe thoroughly understood. This is particularly true sincethe introduction of clinical trials of disease-modifying

therapies in which disease biomarkers play an increas-ingly important part both as outcome measures and asinclusion criteria. We will review the five most wellvalidated AD biomarkers. We then propose a hypotheticalmodel of the time-dependent ordering of onset andmaxima of these biomarkers. The purpose of this paper isto oer this model as a conceptual construct within whichresearch studies from dierent disciplines can relate toone another through a common framework. The modelsuggests a series of testable hypotheses from which aclearer picture of the time-dependent trajectories of ADbiomarkers relative to clinical disease stage and to eachother can be derived.

AD clinical features and pathological changesDementia is the clinically observable result of thecumulative burden of multiple pathological insults in thebrain. Most elderly patients with dementia have multiplepathological changes underlying their dementia; however,the most common pathological substrate is AD.1,2

The clinical disease stages of AD have been dividedinto three phases. First is a pre-symptomatic phase inwhich individuals are cognitively normal but some haveAD pathological changes. To some extent, labelling theseindividuals as having pre-symptomatic AD is a hypothesisrather than a statement of fact, because some of theseindividuals will die without ever expressing clinical

symptoms.35

The hypothetical assumption is that anasymptomatic individual with pathological changes thatare indicative of AD would ultimately have becomesymptomatic if he or she lived long enough. Second is aprodromal phase of AD, commonly referred to as mildcognitive impairment (MCI),6 which is characterised bythe onset of the earliest cognitive symptoms (typicallydeficits in episodic memory) that do not meet the criteriafor dementia. The severity of cognitive impairment in theMCI phase of AD varies from the earliest appearance ofmemory dysfunction to more widespread dysfunction inother cognitive domains. The final phase in the evolutionof AD is dementia, defined as impairments in multipledomains that are severe enough to produce loss of

function.Recent recommendations have suggested redefining

research criteria for AD by labelling individuals withmemory impairment plus accompanying biomarkerevidence of AD as having early AD. 7 These investigatorspropose eliminating the distinction between pre-dementia(ie, MCI) and dementia, but this is not uniformly acceptedbecause the label dementia serves a practical role inclinical practice. A clinical diagnosis of dementia is aclear indication to both the patient and family that thepatient has a disorder that precludes independent livingand has a decidedly worse prognosis than do milderforms of cognitive impairment, and implies that he or

Lancet Neurol 2010; 9: 11928

See Round-up page 4

Department of Radiology

(C R Jack Jr MD) and Department

of Neurology (D S Knopman MD,

R C Petersen MD), Mayo Clinic,

Rochester, MN, USA;School of

Public Health and Helen Wills

Neuroscience Institute,

University of California,

Berkeley, CA, USA

(W J Jagust MD); Department of

Pathology and LaboratoryMedicine, and Institute on

Aging, University of

Pennsylvania School of

Medicine, Philadelphia, PA,

USA (L M Shaw PhD,

J Q Trojanowski PhD) ;

Department of Neurosciences,

University of California-

San Diego, La Jolla, CA, USA

(P S Aisen MD); and Veterans

Affairs and University of

California, San Francisco, CA,

USA (M W Weiner MD)

Correspondence to:

Clifford R Jack, Department of

Radiology, Mayo Clinic,

200 First Street SW, Roches ter,

MN 55905, USA

[email protected]


2/10

120 www.thelancet.com/neurology Vol 9 January 2010

Personal View

she is on an inevitable course toward complete loss ofindependence.

The concept of using biomarkers for early diagnosticpurposes has a long history, with many studies showingthat AD biomarkers can be used to predict conversionfrom MCI to AD. These prediction studies show thatindividuals destined to develop AD can be identified earlierin the disease course by use of the MCI designation withthe addition of imaging and CSF biomarkers to enhancediagnostic specificity.813 However, at present, the clinicaldiagnosis of AD requires the presence of dementia. 14

A widely accepted assumption is that AD begins withabnormal processing of amyloid precursor protein (APP),which then leads to excess production or reducedclearance of -amyloid (A) in the cortex.15 All knownforms of autosomal-dominant AD involve genes that

either encode APP itself, or encode protease subunits(PS1 and PS2) that are involved in the cleavage of A fromAPP to generate amyloidogenic A peptides. By unknownmechanisms, but possibly as a result of the toxic eects ofA oligomers,16 one or more forms of A leads to a cascadecharacterised by abnormal tau aggregation, synapticdysfunction, cell death, and brain shrinkage.17

The abnormal protein deposits that characterise ADpathologically are well known: A plaques and neuro-fibrillary tangles (NFTs) formed by hyperphosphorylatedtau. Neurodegeneration is as important as these hallmarkpathological lesions of AD, and manifests as atrophy,neuron loss, and gliosis, which are routinely noted inresearch post-mortem examinations. Although the loss

of synapses also is highly significant for the clinicalmanifestations of AD, this is dicult to assess withoutthe use of labour-intensive morphometric methods, so itis not routinely measured in most AD research centres.Neurodegeneration and NFT deposition are bothneuronal processes and occur in roughly the sametopographic distribution. A plaques are extracellularand occur in a dierent, but to some degree overlapping,topographic distribution from NFT and neurodegen-erative pathological changes.

Clinical symptoms are more closely related to NFTsthan to plaque formation.18,19 However, the most directpathological substrate of clinical symptoms is

neurodegeneration, and most specifically synapse loss.20Recent autopsy data have confirmed that gross cerebralatrophy (indicating the loss of synapses and neurons),and not A or NFT burden, is the most proximatepathological substrate of cognitive impairment in AD.5

Biomarkers of ADBiomarkers are variables (physiological, biochemical,anatomical) that can be measured in vivo and thatindicate specific features of disease-related pathologicalchanges. We have used the term biomarker to denoteboth imaging and biospecimen (ie, CSF) measures. Wewill focus on the five most widely studied biomarkers ofAD pathology, based on the current literature: decreasedCSF A42, increased CSF tau, decreased fluorodeoxyglucoseuptake on PET (FDG-PET), PET amyloid imaging, and

structural MRI measures of cerebral atrophy. Each ofthese five biomarkers is well validated enough to be usedin currently active therapeutic trials and large multisiteobservational studies. Other potential AD biomarkers aresummarised elsewhere,21,22 and are not discussed here.We briefly review the evidence supporting the positionthat each of these biomarkers is an in vivo indicator of aspecific aspect of AD pathology (panel).

Biomarkers of A-plaque depositionBoth CSF A42 and amyloid PET imaging are biomarkersof brain A plaque deposition. Nearly all patients whohave a clinical diagnosis of AD have positive amyloidimaging studies.2325 Excellent correspondence has been

seen between Pittsburgh compound B (PiB) binding andfibrillar A deposition in the brain (or cerebralvasculature) in most,26,27 but not all,28 patients who haveundergone ante-mortem PiB-PET imaging and autopsy.PiB specifically binds to fibrillar A, and not to solubleA or to diuse plaques.26,27 Low concentrations ofCSF A42 correlate with both the clinical diagnosis ofAD and A neuropathology at autopsy.2931 Nearly 100%concordance is present between abnormally low CSFA42 and positive PiB amyloid imaging findings inpatients who have undergone both tests.3235 The evidencetherefore strongly supports the notion that both amyloidimaging and low CSF A42 are valid biomarkers of brainA-plaque load.

Biomarkers of neurodegenerationCSF tau is an indicator of tau pathological changes andassociated neuronal injury. Although phosphotau mightbe a more specific indicator of AD, concentrations ofboth phosphotau and total tau increase in AD.36 IncreasedCSF tau is not specific for AD, but does correlate withclinical disease severity, with higher concentrationsassociated with greater cognitive impairment inindividuals on the normalMCIAD cognitive spectrum.37In general, increases in CSF tau seem to indicateneuronal damage, and are seen in ischaemic andtraumatic brain injury.38,39 In AD, increased tau in the

Panel: Imaging and CSF biomarker categories

in Alzheimers disease

Brain A-plaque deposition

CSF A142 PET A imaging

Neurodegeneration

CSF tau

Fluorodeoxyglucose-PET

Structural MRI

A=-amyloid.


3/10

www.thelancet.com/neurology Vol 9 January 2010 121

Personal View

CSF is thought to occur as a direct result of tauaccumulation in neurons, particularly axons; this disruptsneuronal activity and causes release into the extracellularspace of cytoskeletal elements, including tau, which thenappear in the CSF.40,41 Increased CSF tau correlates withthe presence of NFTs at autopsy.42 Of note, for reasonsthat remain elusive, similar increases in CSF tau are notseen in pure tauopathies such as progressive supranuclearpalsy (PSP) and corticobasal degeneration (CBD), despitethe fact that the brains of patients with CBD often showfar more tau accumulation than do the brains of patientswith AD at autopsy.43,44 This might suggest thatextracellular A plaques have an eect on the clearanceof tau released from degenerating neurons in AD, thatdierent species of pathological tau are involved in ADversus CBD and PSP, or that other factors render tau

more readily diusible and less degradable in AD versusCBD and PSP.45

FDG-PET is used to measure net brain metabolism,which, although including many neural and glialfunctions, largely indicates synaptic activity.46,47 Brainglucose metabolism measured with FDG-PET is highlycorrelated with post-mortem measures of the synapticstructural protein synaptophysin.48 In the context of AD,decreased FDG-PET uptake is an indicator of impairedsynaptic function. FDG-PET studies in patients with ADshow a specific topographic pattern of decreased glucoseuptake in a lateral temporal-parietal and posteriorcingulate, precuneus distribution.49 Correction forcortical atrophy in patients with AD leaves metabolism

still diminished.50

Greater decreases in FDG uptakecorrelate with greater cognitive impairment along thecontinuum from normal cognitive status to MCI to ADdementia.51 Combined imaging and autopsy studiesshow good correlation between the ante-mortemFDG-PET diagnosis of AD and post-mortemconfirmation.52 In cognitively normal elderly individuals,correlations are seen between decreased FDG-PETuptake and both low CSF A and increased CSF tau.53Together, these data indicate that FDG-PET uptake is avalid indicator of the synaptic dysfunction thataccompanies neurodegeneration in AD.

Structural MRI can provide measures of cerebralatrophy, which is caused by dendritic pruning and loss of

synapses and neurons.54

Volumetric or voxel-basedmeasures of brain atrophy show a strong correlationbetween the severity of atrophy and the severity ofcognitive impairment in patients along the continuumfrom normal cognitive status to AD dementia.55 Thus,rates of neuronal and synaptic loss indicated by theprogressive rate of brain atrophy correlate with rates ofcognitive decline.56 Atrophy on MRI is not specific forAD, but the degree of atrophy correlates well with Braakstaging at autopsy.5759 Additionally, the topographicdistribution of atrophy on MRI maps well onto Braaksstaging of NFT pathology in patients who have undergoneante-mortem MRI and post-mortem AD staging.60

Temporal ordering of biomarker abnormalitiesA crucial element of biomarker-based staging of AD isthe notion of temporal ordering of dierent biomarkers.The model that we propose, which relates pathologicalstage to AD biomarkers, is based on a largely biphasicview of disease progression.61,62 In this model, biomarkersof A deposition become abnormal early, beforeneurodegeneration and clinical symptoms occur.Biomarkers of neuronal injury, dysfunction, andneurodegeneration become abnormal later in the disease.Cognitive symptoms are directly related to biomarkers ofneurodegeneration rather than biomarkers of Adeposition. We examine evidence to support thesetime-dependent assumptions, beginning with evidencethat biomarker abnormalities typically precede clinicalsymptoms.

Biomarker abnormalities precede clinical symptomsApproximately 2040% of cognitively normal elderlypeople have evidence of significant brain A-plaquedeposition, either from amyloid imaging or CSF A42concentrations.37,6366 These data on A imaging and CSFbiomarkers are in agreement with autopsy studies thatalso show an AD pathological burden sucient to meetcriteria for a diagnosis of AD in roughly the sameproportion of cognitively normal elderly individuals.35These data support the principle that the presence ofA-plaque deposition alone, even in substantial quantities,is not sucient to produce dementia, and that abnormalitiesin biomarkers of A deposition precede clinical/cognitive

symptoms.6771

This principle is clearly illustrated by datafrom the individual in figure 1B who was cognitivelynormal with no evidence of atrophy on MRI, but had a

0

05

10

15

20

25

30A B C

Figure: Illustration of biomarker staging of Alzheimers disease

Three elderly individuals are placed in order from left to right by use of our proposed biomarker staging scheme.

(A) A cognitively normal individual with no evidence of A on PET amyloid imaging with PiBand no evidence of

atrophy on MRI. (B) A cognitively normal individual who has no evidence of neurodegenerative atrophy on MRI,

but has significant A deposition on PET amyloid imaging. (B) An individual who has dementia and a clinical

diagnosis of Alzheimers disease, a positive PET amyloid imaging study, and neurodegenerative atrophy on MRI.

A=-amyloid. PiB=Pittsburgh compound B.


4/10


Personal View

highly abnormal PiB study. Calculated rates from serialPiB imaging studies indicate that A-plaque accumulationin individuals destined to become demented might beginas much as two decades before the manifestation of clinicalsymptoms.62 We note that both A deposition and NFTscan be present in individuals with no symptoms. However,the presence of NFTs in asymptomatic individuals tends tobe confined to the entorhinal cortex, Braak stage III,whereas NFTs in symptomatic individuals are far morewidespread.35,72 By contrast, A-plaque deposition can bewidespread in clinically asymptomatic individuals.

There is strong evidence that MRI, FDG-PET, and CSFtau biomarkers are already abnormal in patients who arein the MCI phase of AD.37,51,7375 Abnormalities in neuro-degenerative AD biomarkers also precede the appearanceof the first cognitive symptoms. Of the three

neurodegenerative biomarkers, evidence that FDG-PETabnormalities precede any cognitive symptoms inindividuals who later progress to AD is probably thestrongest.76,77 However, rates of atrophy on MRI dobecome abnormal in cognitively normal individuals wholater progress to AD.7880 Thus, the available data stronglysupport the conclusion that abnormalities in both Aand neurodegenerative biomarkers precede clinicalsymptoms.

A biomarker abnormalities precede neurodegenerativebiomarker abnormalitiesThe rate of MRI atrophy on serial imaging studies isgreatest in patients with a clinical diagnosis of AD, least

in cognitively normal individuals, and intermediate inthose with a clinical diagnosis of MCI. By contrast, ratesof change in PiB retention over time are just slightlygreater than zero for all these three clinical groups, anddo not dier by clinical group.62 Thus, in patients who arerapidly declining clinically (ie, patients with AD) MRIrates map well onto simultaneous cognitive deterioration,

whereas rates of change in PiB do not.62,81 Similarly, CSFA does not change significantly over time in patientswith AD. Rates of brain atrophy correlate well withpathological indices of NFTs and other neurodegenerativechanges, but do not correlate with severity of Adeposition at autopsy.82 Cognitively normal individualswith positive A imaging studies might have normalstructural MRI studies, implying that a substantial Aload can accumulate with no immediate eect on grossbrain structure or cognition (figure 1).83 In a modellingstudy inferring cause and eect, Mormino andcolleagues84 found that the direct substrate of memoryimpairment was hippocampal atrophy on MRI, and notA deposition as measured by PiB imaging. Frisoni andcolleagues85 also placed amyloid deposition before MRIchanges in the sequence of events. These findings

support the conclusion that an abnormality in biomarkersof A-plaque deposition is an early event that nears aplateau before the appearance of both atrophy on MRIand cognitive symptoms, and remains relativelystatic thereafter. By contrast, abnormalities in neuro-degenerative biomarkers on MRI accelerate as symptomsappear, and then parallel cognitive decline.

Neurodegenerative biomarkers are temporally orderedAvailable evidence suggests that FDG-PET changesmight precede MRI changes.77,86,87 Up to this point, wehave discussed the temporal ordering of AD biomarkersfrom the perspective of which biomarker becomesabnormal earlier during the progression of AD. However,

the order in which the dynamic range of biomarkersapproaches its maximum is also relevant to the discussionof biomarker ordering. MRI and CSF tau correlate wellwith cognition if individuals who span the entire cognitivespectrum (controls, MCI, and AD) are combined.However, among patients with MCI or AD alone,correlations with measures of general cognition arestrong with structural MRI, but are not significant withCSF tau.88 These data are consistent with studiesindicating that CSF tau does not change appreciably overtime in cognitively impaired patients.89,90 Furthermore,although both MRI and CSF tau are predictive of futureconversion from MCI to AD, the predictive power ofstructural MRI is greater.91 These findings imply that the

correlations between cognition and CSF tau weaken aspatients progress into the mid and late stages of theclinical AD spectrum. Conversely, structural MRImeasures of atrophy retain a highly significant correlationwith observed clinical impairment in both the MCI anddementia phases of AD. Moreover, rates of atrophy onMRI are significantly greater in patients with AD than incognitively normal elderly individuals.92 This body ofliterature implies that MRI atrophy is a later event in ADprogression, preceded by abnormalities in CSF tau andFDG-PET, and that MRI retains a closer correlation withcognitive symptoms later in disease progression thandoes CSF tau.

Abnormal

NormalCognitively normal MCI

Clinical disease stage

Biomarkerm

agnitude

Dementia

ATau-meditated neuronal injury and dysfunction

Brain structureMemory

Clinical function

Figure:Dynamic biomarkers of the Alzheimers pathological cascade

A is identified by CSF A42 or PET amyloid imaging. Tau-mediated neuronal injury and dysfunction is identified by

CSF tau or fluorodeoxyglucose-PET. Brain structure is measured by use of structural MRI. A=-amyloid. MCI=mild

cognitive impairment.


5/10


Personal View

Use of biomarkers to stage AD in vivoAutopsy studies have been, and will continue to be,essential in uncovering the biological basis of the clinicalsymptoms in AD. However, by definition, autopsy studiesare unable to provide clinicalhistological correlationsduring life, when pathological changes actually occur,resulting in an inability to isolate relationships betweentime-dependent histological changes and clinical/cognitive consequences. This point underlies the value ofusing biomarkers in the staging of disease.

On the basis of the evidence presented above, wepropose the use of specific AD biomarkers for diseasestaging in vivo. The disease model and biomarker stagingare shown in figure 2, which embodies the following setof principles. First, the biomarkers become abnormal ina temporally ordered manner as the disease progresses.

Second, A-plaque biomarkers are dynamic early in thedisease, before the appearance of clinical symptoms, andhave largely reached a plateau by the time clinicalsymptoms appear. Third, biomarkers of neuronal injury,dysfunction, and degeneration are dynamic later in thedisease and correlate with clinical symptom severity.Fourth, MRI is the last biomarker to become abnormal;however, MRI retains a closer relationship with cognitiveperformance later into the disease than other bio-markers.88,91 Fifth, none of the biomarkers is static; ratesof change in each biomarker change over time and followa non-linear time course, which we hypothesise to besigmoid shaped. Non-linearity has been clearly shown inMRI studies, in which atrophy rates accelerate as patients

approach clinical dementia.93,94

A sigmoid shape as afunction of time implies that the maximum eect of eachbiomarker varies over the course of disease progression.95Comprehensive biomarker-based staging of disease in anindividual at a given point in time should be possiblefrom measures of the magnitude and slope (ie, rate ofchange) of several dierent biomarkers (figure 3). Sixth,anatomical information from imaging biomarkersprovides crucial disease-staging information. Similar toNFT accumulation, cerebral atrophy, for example, beginsin medial temporal limbic areas and spreads from thereto adjacent limbic and paralimbic areas and later to theisocortical association cortex.96 Therefore, at a giventimepoint, dierent brain areas will be at dierent stages.

In any given area of an individuals brain there might bean amyloid phase, a neuronal dysfunction phase, etc, andthis will occur in an anatomical order (figure 4). Thisimplies an advantage for imaging biomarkers over fluidbiomarkers, because imaging can resolve the dierentphases of the disease both temporally and anatomically.Finally, a feature of this biomarker model of AD is thepresence of a lag phase of unknown duration betweenA-plaque formation and the neurodegenerative cascade.Between-patient variation in the lag period between Adeposition and the neurodegenerative cascade is probablyan indication of dierences in A processing, in theeects of abnormal A processing on neuronal injury,

and dierences in brain resilience or cognitive reserve.97

Other pathological changes, particularly cerebrovascular,-synuclein, and TAR DNA-binding protein 43proteinopathy mechanisms, also contribute significantlyto between-individual variations in clinical diseaseexpression.98

Clinical trialsOur proposed model has implications for clinical trials.For example, it is rational to select patients for inclusionin trials of anti-A therapies on the basis of biomarkerevidence of the presence of A in the brain by use ofeither amyloid PET imaging or CSF A42. Althoughbiomarkers of neurodegeneration correlate with clinical

Abnormal

Normal

M

N

ASA

SN

SM

t

Clinical disease stage

Biomarkermagnitude

ANeuronal injury and dysfunction

MRI

Figure: Staging Alzheimers disease with dynamic biomarkers

Disease stage based on biomarkers is described by the magnitude of the biomarker abnormality and its rate of

change at a given point in time (t). This is illustrated by the following terms: A(t)=magnitude of the A plaque

biomarker at time t; SA(t)=slope of the A plaque function at time t; N(t)=tau-mediated neuron injury at time t;

SN(t)=slope of N(t); M(t)=MRI at time t; SM(t)=slope of M(t). A=-amyloid.

Abnormal

Normal

Disease stage

Biomarkermagnitude

FDG-PET

Posterior cingulateLateral temporal

FrontalStructural MRI

Medial temporalLateral temporal

Frontal

Figure: Anatomical imaging information

For simplicity in other figures, imaging biomarkers have been shown as individual curves. However, anatomical

variation exists in the time courses within each imaging mode. For example, in FDG-PET, one would expect

abnormalities to appear in the following order: precuneus/posterior cingulate, lateral temporal, and frontal lobe

much later. Similarly, in structural MRI, one would expect abnormalities to appear in t he following order: medial

temporal, lateral temporal, and frontal lo be later. FGD=fluorodeoxyglucose.


6/10


Personal View

and pathological severity, they are not specific for AD andthus should not take precedence over A biomarkers asinclusion criteria for patients in anti-amyloid therapeutictrials (although an A biomarker might be combinedwith MRI volumetrics to provide an indication of diseasestage). Conversely, change in A load over time has littlerelation to change in cognition in natural history studies.In addition, evidence of therapeutic plaque removal inpatients who already have dementia does not seem tocorrelate with a change in the trajectory of cognitivedeterioration (at least, not in every case examined).99Measures of the neurodegenerative portion of the cascade(eg, CSF tau, FDG-PET, or structural MRI) should

therefore be used in therapeutic trials as evidence oftherapeutic modification of the neurodegenerative aspectof the AD pathological process. Therapeutic modificationof the slopes of clinical outcome measures is a commonoutcome metric used in clinical trials. A consideration inthe use of biomarkers as co-primary outcome measuresis the fact that the slopes (rates of change over time) ofdierent biomarkers vary over the course of the disease(figure 3). Ultimately, a combination of biomarkers mightbe needed in clinical trials to select appropriateparticipants and to follow disease progression.

CaveatsWe have attempted to integrate the five most thoroughlyvalidated biomarkers of AD pathology into a model thatis supported by currently available data. We have used a

best-fit approach, realising that every publishedobservation cannot neatly fit into our (or any other) singleidealised model of disease progression. The proposedbiomarker model represents a model of typical diseaseprogression. We certainly do not preclude the existenceof individual deviations from this generic model.

Although we have used the available evidence tosupport our model relating biomarkers to disease stagein AD, we recognise that some of the proposedrelationships take the form of hypotheses rather thanstatements of fact. For example, although A imagingand CSF A are denoted as occurring simultaneously(figures 2, 3, and 5), it might be that one precedes orplateaus earlier than the other. The same logic applies to

the timing relationships between FDG-PET and CSF tau.We have superimposed these curves in the current modelnot because there is good evidence to suggest that thecurves should be identical, but because there is not goodevidence at present to show that they are dierent. Wealso acknowledge that temporal ordering among theneurodegenerative biomarkers might ultimately dierfrom that in our proposed model.100 For example, recentdata from Fagan and colleagues68 suggest that MRIatrophy might precede increases in CSF tau.

We recognise that well-validated biomarkers do notcurrently exist for some important features of the disease.This includes reliable chemical biomarkers of specifictoxic oligomeric forms of soluble A and imaging

measures of soluble A or diuse plaques. We aretherefore not in a position to include in our modelimportant mechanistic features such as the role of toxicA oligomeric species and the timing of their appearance.The absence of PET ligands that specifically measure theburden of NFTs and other tau abnormalities alsoconstitutes a serious gap in our current armamentariumof imaging biomarkers. Another shortcoming is theabsence of a widely accepted biomarker for microglialactivation. PET imaging ligands and the magneticresonance spectroscopy metabolite myo-inositol havebeen proposed as potential biomarkers of glialactivation;101,102 however, neither is widely used for this

Abnormal

A APOE4 non-carrier

Normal

Age

65 years

Biomarkermagnitude

Amyloid

Neuron injury and dysfunctionMRI

Abnormal

B APOE4 carrier

Normal

65 years

Biomarkermagnitude

Amyloid

Neuron injury and dysfunctionMRI

Abnormal

C Other factors

Normal

Biomarke

rmagnitude

Amyloid

C

C0

C+

Figure:Modulators of biomarker temporal relationships

(A,B) Relative to a fixed age (here, 65 years), the hypoth esised effect ofAPOE4 is to shift -amyloid plaque

deposition and the neurodegenerative cascade both to an earlier age compared with 4 non-carriers. (C) The

hypothesised effect of the presence of different diseases and genes on cognition: C=cognition in the presence of

comorbidities (eg, Lewy bodies or vascular disease) or risk amplification genes; C+=cognition in patients with

enhanced cognitive reserve or protective genes; Co=cognition in individuals without comorbidity or enhanced

cognitive reserve.


7/10


Personal View

purpose at present. Thus, our biomarker model of diseaseis just thata model of the stages of disease that can beassessed with currently validated biomarkers, and not acomprehensive model of all pathological processes in AD.In this context, we acknowledge that all biomarkerinformation about disease is limited by the inevitablefilter imposed by detection sensitivity and measurementprecision. Clearly, an in vivo measure is unlikely to be assensitive to the underlying pathology as a detailed autopsyexamination would be.

An observation that does not fit our model is that ofBraak and Braak,72 who concluded that stage III(entorhinal) NFT changes precede isocortical A changes,leading to the conclusion that NFT accumulation is theinitiating event in AD.103 However, the followingobservations contradict this conclusion: (1) the genetics

data in early-onset AD, which implicate disordered Ametabolism as the initiating event; (2) the fact that thefinal pathological picture is identical between late-onsetand early-onset AD cases; and (3) the fact that the majorgenetic risk factor of late-onset AD (APOE 4) isimplicated in disordered tracking of A peptide.104

Observations that will need to be incorporated intofuture revisions of this model relate to evidence ofdisordered glucose uptake in cognitively normal youngand middle-aged APOE4 carriers up to decades beforedisease-related clinical symptoms would be expected toappear.105,106 No consensus exists on the interpretation ofthese observations at this point. These findings couldbe neurodevelopmental in origin or early indicators

of AD.107,108

ConclusionsOur main objective was to provide a framework forhypothesis testing that relates temporal changes in ADbiomarkers to clinical disease stage and to each other.The temporal relationships among the biomarkers andwith clinical disease stage constitute an array of testablehypotheses. For example, carriers ofAPOE4 have anearlier age of onset of dementia than non-carriers, 109and we hypothesise that APOE 4 carriers will have aleftward (earlier in time) shift of both the A-plaqueand neurodegenerative biomarker cascades relative tonon-carriers (figure 5).110 We also hypothesise that

modifiers of the relationship between A pathologicalchanges and their downstream eect on cognitionmight alter the lag time between A-plaque depositionand cognitive decline (figure 5). For example, thecognitive decline curve might shift closer to the Acurve (to the left) in individuals with comorbidities (eg,vascular disease), whereas the cognitive decline curvewould shift away from the amyloid curve (to the right)in individuals with enhanced cognitive reserve.97Similarly, as yet undiscovered neuroprotective genesmight shift the cognitive decline curve to the right,away from the amyloid curve, whereas genes thatamplify the eect of A dysmetabolism on the

neurodegenerative cascade might shift the cognitivedecline curve closer to the amyloid curve. For example,recent genome-wide association studies have identifiednew susceptibility loci involving CR1, CLU, andPICALMgenes.111,112 Clusterin (apolipoprotein J) seems

to regulate the toxicity and solubility of A and mightmodify its clearance at the bloodbrain barrier.111 CR1might also modify A clearance.112 PICALM might berelated to AD through a role in altering synaptic vesiclecycling or APP endocytosis.111 Finally, we anticipate thatother diagnostic modalities (eg, functional MRIconnectivity)113115 will be added to this biomarker stagingmodel as they mature.

Contributors

All authors contributed equally to the preparation of this paper.

Conflicts of interest

CRJ is an investigator in clinical trials sponsored by Pfizer and Baxter,and consults for Elan and Lilly. DSK has served on a data and safetymonitoring board for Lilly, has been a consultant for GlaxoSmithKline,and receives grant funding for clinical trials by Baxter, Elan, and

Forest. WJJ has consulted for Genentech, Synarc, ElanPharmaceuticals, Ceregene, Schering Plough, Merck & Co, and Lilly.LMS has consulted for Pfizer. PSA has consulted for Elan, Wyeth,Eisai, Neurochem, Schering-Plough, Bristol-Myers-Squibb, Lilly,Neurophase, Merck, Roche, Amgen, Genentech, Abbott, Pfizer,Novartis, and Medivation, has stock options with Medivation andNeurophase, and has received grant support from Pfizer, Baxter, andNeuro-Hitech. MWW has been on the scientific advisory boards for theAlzheimers Study Group, Bayer Schering Pharma, Lilly, CoMentis,Neurochem, Eisai, Avid, Aegis, Genentech, Allergan,Bristol-Myers Squibb, Forest, Pfizer, McKinsey, Mitsubishi, andNovartis, has received travel funding from Nestl, honoraria fromBOLT International, commercial research support from Merck andAvid, and has stock options in Synarc and Elan. RCP has been a chair,member of the safety monitoring committee, and consultant for Elan,has chaired a data monitoring committee for Wyeth, and has been aconsultant for GE Healthcare. JQT has received lecture honoraria fromWyeth, Pfizer, and Takeda.

Acknowledgments

CRJ receives research support from the National Institute on Aging(US National Institutes of Health grant AG11378). DSK and RCP receiveNational Institute on Aging funding (AG16574 and MCSA AG06786).JQT receives funding from the Alzheimers Disease Center (AG10124).We thank Nick C Fox and Prashanthi Vemuri for advice, andDenise A Reyes for the figures.

References1 Schneider JA, Arvanitakis Z, Bang W, Bennett DA. Mixed brain

pathologies account for most dementia cases in community-dwellingolder persons. Neurology2007; 69: 2197204.

2 White L, Small BJ, Petrovitch H, et al. Recent clinical-pathologicresearch on the causes of dementia in late life: update from theHonolulu-Asia Aging Study.J Geriatr Psychiatry Neurol2005;18: 22427.

Search strategy and selection criteria

References for this paper were identified through searches ofPubMed between 1984 and October, 2009, with

combinations of the search terms Alzheimers disease,

dementia, MCI, imaging, PET, PiB, amyloid

imaging, MRI, and biomarker. Articles were also

identified through searches of the authors own files. Only

papers published in English were reviewed.


8/10


Personal View

3 Knopman DS, Parisi JE, Salviati A, et al. Neuropathology ofcognitively normal elderly.J Neuropathol Exp Neurol2003;62: 108795.

4 Price JL, Morris JC. Tangles and plaques in nondemented aging andpreclinical Alzheimers disease. Ann Neurol1999; 45: 35868.

5 Savva GM, Wharton SB, Ince PG, Forster G, Matthews FE, Brayne C.Age, neuropathology, and dementia. N Engl J Med2009; 360: 230209.

6 Petersen RC. Mild cognitive impairment as a diagnostic entity.J Intern Med2004; 256: 18394.

7 Dubois B, Feldman HH, Jacova C, et al. Research criteria for thediagnosis of Alzheimers disease: revising the NINCDS-ADRDAcriteria. Lancet Neurol2007; 6: 73446.

8 Chetelat G, Desgranges B, de la Sayette V, Viader F, Eustache F,Baron JC. Mild cognitive impairment: can FDG-PET predict whois to rapidly convert to Alzheimers disease? Neurology2003;60: 137477.

9 Jack CR Jr, Petersen RC, Xu YC, et al. Prediction of AD withMRI-based hippocampal volume in mild cognitive impairment.Neurology1999; 52: 1397403.

10 Yuan Y, Gu ZX, Wei WS. Fluorodeoxyglucose-positron-emissiontomography, single-photon emission tomography, and structural

MR imaging for prediction of rapid conversion to Alzheimerdisease in patients with mild cognitive impairment; a meta-analysis.AJNRAm J Neuroradiol2009; 30: 40410.

11 Mattsson N, Zetterberg H, Hansson O, et al. CSF biomarkers andincipient Alzheimer disease in patients with mild cognitiveimpairment.JAMA 2009; 302: 38593.

12 Jagust WJ, Haan MN, Eberling JL, Wolfe N, Reed BR. Functionalimaging predicts cognitive decline in Alzheimers disease.

J Neuroimaging1996; 6: 15660.

13 Visser PJ, Verhey F, Knol DL, et al. Prevalence and prognostic valueof CSF markers of Alzheimers disease pathology in patients withsubjective cognitive impairment or mild cognitive impairment inthe DESCRIPA study: a prospective cohort study. Lancet Neurol2009; 8: 61927.

14 McKhann G, Drachman D, Folstein M, Katzman R, Price D,Stadlan EM. Clinical diagnosis of Alzheimers disease: report of theNINCDS-ADRDA Work Group under the auspices of Departmentof Health and Human Services Task Force on Alzheimers Disease.Neurology1984; 34: 93944.

15 Hardy J, Selkoe DJ. The amyloid hypothesis of Alzheimers disease:progress and problems on the road to therapeutics. Science 2002;297: 35356.

16 Klein WL, Stine WB Jr, Teplow DB. Small assemblies of unmodifiedamyloid beta-protein are the proximate neurotoxin in Alzheimersdisease. Neurobiol Aging2004; 25: 56980.

17 Oddo S, Caccamo A, Tran L, et al. Temporal profile of amyloid- (A)oligomerization in an in vivo model of Alzheimer disease. A linkbetween A and tau pathology.J Biol Chem2006; 281: 1599604.

18 Gomez-Isla T, Hollister R, West H, et al. Neuronal loss correlateswith but exceeds neurofibrillary tangles in Alzheimers disease.Ann Neurol1997; 41: 1724.

19 Bennett DA, Schneider JA, Wilson RS, Bienias JL, Arnold SE.Neurofibrillary tangles mediate the association of amyloid loadwith clinical Alzheimer disease and level of cognitive function.Arch Neurol2004; 61: 37884.

20 Terry RD, Masliah E, Salmon DP, et al. Physical basis of cognitivealterations in Alzheimers disease: synapse loss is the majorcorrelate of cognitive impairment. Ann Neurol1991; 30: 57280.

21 Hampel H, Burger K, Teipel SJ, Bokde AL, Zetterberg H,Blennow K. Core candidate neurochemical and imaging biomarkersof Alzheimers disease. Alzheimers Dement2008; 4: 3848.

22 Shaw LM, Korecka M, Clark CM, Lee VM, Trojanowski JQ.Biomarkers of neurodegeneration for diagnosis and monitoringtherapeutics. Nat Rev Drug Discov2007; 6: 295303.

23 Klunk WE, Engler H, Nordberg A, et al. Imaging brain amyloid inAlzheimers disease with Pittsburgh compound-B. Ann Neurol2004;55: 30619.

24 Rowe CC, Ng S, Ackermann U, et al. Imaging beta-amyloid burdenin aging and dementia. Neurology2007; 68: 171825.

25 Edison P, Archer HA, Hinz R, et al. Amyloid, hypometabolism, andcognition in Alzheimer disease: an [C]PIB and [F]FDG PETstudy. Neurology2007; 68: 50108.

26 Ikonomovic MD, Klunk WE, Abrahamson EE, et al. Post-mortemcorrelates of in vivo PiB-PET amyloid imaging in a typical case ofAlzheimers disease. Brain 2008; 131: 163045.

27 Bacskai BJ, Frosch MP, Freeman SH, et al. Molecular imaging withPittsburgh compound B confirmed at autopsy: a case report.Arch Neurol2007; 64: 43134.

28 Rosen RF, Ciliax BJ, Wingo TS, et al. Deficient high-anity bindingof Pittsburgh compound B in a case of Alzheimers disease.Acta Neuropathol2009; published online Aug 19. DOI:10.1007/s00401-009-0583-3.

29 Clark CM, Xie S, Chittams J, et al. Cerebrospinal fluid tau andbeta-amyloid: how well do these biomarkers reflectautopsy-confirmed dementia diagnoses? Arch Neurol2003;60: 1696702.

30 Strozyk D, Blennow K, White LR, Launer LJ. CSF A 42 levelscorrelate with amyloid-neuropathology in a population-basedautopsy study. Neurology2003; 60: 65256.

31 Schoonenboom NS, van der Flier WM, Blankenstein MA, et al. CSFand MRI markers independently contribute to the diagnosis ofAlzheimers disease. Neurobiol Aging2008; 29: 66975.

32 Fagan AM, Mintun MA, Mach RH, et al. Inverse relation between

in vivo amyloid imaging load and cerebrospinal fluid A42 inhumans. Ann Neurol2006; 59: 51219.

33 Jagust WJ, Landau SM, Shaw LM, et al. Relationships betweenbiomarkers in aging and dementia. Neurology2009; 73: 119399.

34 Grimmer T, Riemenschneider M, Forstl H, et al. Beta amyloid inAlzheimers disease: increased deposition in brain is reflected inreduced concentration in cerebrospinal fluid. Biol Psychiatry2009;65: 92734.

35 Tolboom N, van der Flier WM, Yaqub M, et al. Relationship ofcerebrospinal fluid markers to C-PiB and F-FDDNP binding.J Nucl Med2009; 50: 146470.

36 Buerger K, Ewers M, Pirttila T, et al. CSF phosphorylated tauprotein correlates with neocortical neurofibrillary pathology inAlzheimers disease. Brain 2006; 129: 303541.

37 Shaw LM, Vanderstichele H, Knapik-Czajka M, et al.Cerebrospinal fluid biomarker signature in Alzheimersdisease neuroimaging initiative subjects. Ann Neurol2009;65: 40313.

38 Hesse C, Rosengren L, Andreasen N, et al. Transient increase intotal tau but not phospho-tau in human cerebrospinal fluid afteracute stroke. Neurosci Lett2001; 297: 18790.

39 Ost M, Nylen K, Csajbok L, et al. Initial CSF total tau correlates with1-year outcome in patients with traumatic brain injury. Neurology2006; 67: 160004.

40 Arai H, Terajima M, Miura M, et al. Tau in cerebrospinal fluid:a potential diagnostic marker in Alzheimers disease. Ann Neurol1995; 38: 64952.

41 Blennow K, Wallin A, Agren H, Spenger C, Siegfried J,Vanmechelen E. Tau protein in cerebrospinal fluid: a biochemicalmarker for axonal degeneration in Alzheimer disease?Mol Chem Neuropathol1995; 26: 23145.

42 Tapiola T, Overmyer M, Lehtovirta M, et al. The level ofcerebrospinal fluid tau correlates with neurofibrillary tangles inAlzheimers disease. Neuroreport1997; 8: 396163.

43 Bian H, Van Swieten JC, Leight S, et al. CSF biomarkers infrontotemporal lobar degeneration with known pathology.Neurology2008; 70: 182735.

44 Grossman M, Farmer J, Leight S, et al. Cerebrospinal fluid profilein frontotemporal dementia and Alzheimers disease. Ann Neurol2005; 57: 72129.

45 Ballatore C, Lee VM, Trojanowski JQ. Tau-mediatedneurodegeneration in Alzheimers disease and related disorders.Nat Rev Neurosci 2007; 8: 66372.

46 Schwartz WJ, Smith CB, Davidsen L, et al. Metabolic mapping offunctional activity in the hypothalamo-neurohypophysial system ofthe rat. Science 1979; 205: 72325.

47 Attwell D, Laughlin SB. An energy budget for signaling inthe grey matter of the brain.J Cereb Blood Flow Metab 2001;21: 113345.

48 Rocher AB, Chapon F, Blaizot X, Baron JC, Chavoix C. Resting-statebrain glucose utilization as measured by PET is directly related toregional synaptophysin levels: a study in baboons. Neuroimage 2003;20: 189498.


9/10


Personal View

49 Jagust W, Reed B, Mungas D, Ellis W, Decarli C. What doesfluorodeoxyglucose PET imaging add to a clinical diagnosis ofdementia? Neurology2007; 69: 87177.

50 Ibanez V, Pietrini P, Alexander GE, et al. Regional glucosemetabolic abnormalities are not the result of atrophy in Alzheimersdisease. Neurology1998; 50: 158593.

51 Minoshima S, Giordani B, Berent S, Frey KA, Foster NL, Kuhl DE.Metabolic reduction in the posterior cingulate cortex in very earlyAlzheimers disease. Ann Neurol1997; 42: 8594.

52 Homan JM, Welsh-Bohmer KA, Hanson M, et al. FDG PETimaging in patients with pathologically verified dementia.

J Nucl Med2000; 41: 192028.

53 Petrie EC, Cross DJ, Galasko D,et al. Preclinical evidence ofAlzheimer changes: convergent cerebrospinal fluid biomarker andfluorodeoxyglucose positron emission tomography findings.Arch Neurol2009; 66: 63237.

54 Bobinski M, de Leon MJ, Wegiel J, et al. The histological validation ofpost mortem magnetic resonance imaging-determined hippocampalvolume in Alzheimers disease. Neuroscience 2000; 95: 72125.

55 Jack CR Jr, Petersen RC, OBrien PC, Tangalos EG. MR-basedhippocampal volumetry in the diagnosis of Alzheimers disease.

Neurology1992; 42: 18388.56 Fox NC, Scahill RI, Crum WR, Rossor MN. Correlation betweenrates of brain atrophy and cognitive decline in AD. Neurology1999;52: 168789.

57 Jack CR Jr, Dickson DW, Parisi JE, et al. Antemortem MRI findingscorrelate with hippocampal neuropathology in typical aging anddementia. Neurology2002; 58: 75057.

58 Silbert LC, Quinn JF, Moore MM, et al. Changes in premorbidbrain volume predict Alzheimers disease pathology. Neurology2003; 61: 48792.

59 Braak H, Braak E. Neuropathological stageing of Alzheimer-relatedchanges. Acta Neuropathol1991; 82: 23959.

60 Whitwell JL, Josephs KA, Murray ME, et al. MRI correlates ofneurofibrillary tangle pathology at autopsy: a voxel-basedmorphometry study. Neurology2008; 71: 74349.

61 Ingelsson M, Fukumoto H, Newell KL, et al. Early A accumulationand progressive synaptic loss, gliosis, and tangle formation in ADbrain. Neurology2004; 62: 92531.

62 Jack CR Jr, Lowe VJ, Weigand SD, et al. Serial PIB and MRI innormal, mild cognitive impairment and Alzheimers disease:implications for sequence of pathological events in Alzheimersdisease. Brain 2009; 132: 135565.

63 Mintun MA, Larossa GN, Sheline YI, et al. [C]PIB in anondemented population: potential antecedent marker ofAlzheimer disease. Neurology2006; 67: 44652.

64 Aizenstein HJ, Nebes RD, Saxton JA, et al. Frequent amyloiddeposition without significant cognitive impairment among theelderly. Arch Neurol2008; 65: 150917.

65 Peskind ER, Li G, Shofer J, et al. Age and apolipoprotein E*4 alleleeects on cerebrospinal fluid beta-amyloid 42 in adults with normalcognition. Arch Neurol2006; 63: 93639.

66 Bouwman FH, Schoonenboom NS, Verwey NA, et al. CSFbiomarker levels in early and late onset Alzheimers disease.Neurobiol Aging2009; 30: 1895901.

67 Fagan AM, Roe CM, Xiong C, Mintun MA, Morris JC,Holtzman DM. Cerebrospinal fluid tau/beta-amyloid(42) ratio as a

prediction of cognitive decline in nondemented older adults.Arch Neurol2007; 64: 34349.

68 Fagan AM, Head D, Shah AR, et al. Decreased cerebrospinal fluidA(42) correlates with brain atrophy in cognitively normal elderly.Ann Neurol2009; 65: 17683.

69 Li G, Sokal I, Quinn JF, et al. CSF tau/A42 ratio for increased riskof mild cognitive impairment: a follow-up study. Neurology2007;69: 63139.

70 Gustafson DR, Skoog I, Rosengren L, Zetterberg H, Blennow K.Cerebrospinal fluid beta-amyloid 1-42 concentration may predictcognitive decline in older women.J Neurol Neurosurg Psychiatry2007; 78: 46164.

71 Stomrud E, Hansson O, Blennow K, Minthon L, Londos E.Cerebrospinal fluid biomarkers predict decline in subjectivecognitive function over 3 years in healthy elderly.Dement Geriatr Cogn Disord2007; 24: 11824.

72 Braak H, Braak E. Frequency of stages of Alzheimer-related lesionsin dierent age categories. Neurobiol Aging1997; 18: 35157.

73 Chetelat G, Desgranges B, De La Sayette V, Viader F, Eustache F,

Baron JC. Mapping gray matter loss with voxel-basedmorphometry in mild cognitive impairment. Neuroreport2002;13: 193943.

74 Ewers M, Buerger K, Teipel SJ, et al. Multicenter assessment ofCSF-phosphorylated tau for the prediction of conversion of MCI.Neurology2007; 69: 220512.

75 Sluimer JD, Bouwman FH, Vrenken H, et al P. Whole-brainatrophy rate and CSF biomarker levels in MCI and AD:a longitudinal study. Neurobiol Aging2008; published online Aug 7.DOI:10.1016/j.neurobiolaging.2008.06.016.

76 de Leon MJ, Convit A, Wolf OT, et al. Prediction of cognitive declinein normal elderly subjects with 2-[F]fluoro-2-deoxy-D-glucose/poitron-emission tomography (FDG/PET). Proc Natl Acad Sci USA2001; 98: 1096671.

77 Jagust W, Gitcho A, Sun F, Kuczynski B, Mungas D, Haan M. Brainimaging evidence of preclinical Alzheimers disease in normalaging. Ann Neurol2006; 59: 67381.

78 Fox NC, Warrington EK, Rossor MN. Serial magnetic resonance

imaging of cerebral atrophy in preclinical Alzheimers disease.Lancet1999; 353: 2125.

79 Jack CR Jr, Shiung MM, Gunter JL, et al. Comparison of dierentMRI brain atrophy rate measures with clinical disease progressionin AD. Neurology2004; 62: 591600.

80 Kaye JA, Swihart T, Howieson D, et al. Volume loss of thehippocampus and temporal lobe in healthy elderly personsdestined to develop dementia. Neurology1997; 48: 1297304.

81 Engler H, Forsberg A, Almkvist O, et al. Two-year follow-up ofamyloid deposition in patients with Alzheimers disease. Brain2006; 129: 285666.

82 Josephs KA, Whitwell JL, Ahmed Z, et al. Beta-amyloid burden isnot associated with rates of brain atrophy. Ann Neurol2008;63: 20412.

83 Jack CR Jr, Lowe VJ, Senjem ML, et al. C PiB and structural MRIprovide complementary information in imaging of Alzheimersdisease and amnestic mild cognitive impairment. Brain 2008;131: 66580.

84 Mormino EC, Kluth JT, Madison CM, et al. Episodic memory loss isrelated to hippocampal-mediated beta-amyloid deposition in elderlysubjects. Brain 2009; 132: 131023.

85 Frisoni GB, Lorenzi M, Caroli A, Kemppainen N, Nagren K,Rinne JO. In vivo mapping of amyloid toxicity in Alzheimerdisease. Neurology2009; 72: 150411.

86 De Santi S, de Leon MJ, Rusinek H, et al. Hippocampal formationglucose metabolism and volume losses in MCI and AD.Neurobiol Aging2001; 22: 52939.

87 Reiman EM, Uecker A, Caselli RJ, et al. Hippocampal volumes incognitively normal persons at genetic risk for Alzheimers disease.Ann Neurol1998; 44: 28891.

88 Vemuri P, Wiste HJ, Weigand SD, et al. MRI and CSFbiomarkers in normal, MCI, and AD subjects: diagnosticdiscrimination and cognitive correlations. Neurology2009;73: 28793.

89 Sunderland T, Wolozin B, Galasko D, et al. Longitudinal stabilityof CSF tau levels in Alzheimer patients. Biol Psychiatry1999;46: 75055.

90 Bouwman FH, van der Flier WM, Schoonenboom NS, et al.Longitudinal changes of CSF biomarkers in memory clinic patients.Neurology2007; 69: 100611.

91 Vemuri P, Wiste HJ, Weigand SD, et al. MRI and CSF biomarkersin normal, MCI, and AD subjects: predicting future clinical change.Neurology2009; 73: 294301.

92 Fox NC, Freeborough PA. Brain atrophy progression measuredfrom registered serial MRI: validation and application toAlzheimers disease.J Magn Reson Imaging1997; 7: 106975.

93 Chan D, Janssen JC, Whitwell JL, et al. Change in rates of cerebralatrophy over time in early-onset Alzheimers disease: longitudinalMRI study. Lancet2003; 362: 112122.

94 Ridha BH, Barnes J, Bartlett JW, et al. Tracking atrophy progressionin familial Alzheimers disease: a serial MRI study. Lancet Neurol2006; 5: 82834.


10/10

128 www thelancet com/neurology Vol 9 January 2010

Personal View

95 Wahlund LO, Blennow K. Cerebrospinal fluid biomarkers fordisease stage and intensity in cognitively impaired patients.Neurosci Lett2003; 339: 99102.

96 Whitwell JL, Przybelski SA, Weigand SD, et al. 3D maps frommultiple MRI illustrate changing atrophy patterns as subjectsprogress from mild cognitive impairment to Alzheimers disease.Brain 2007; 130: 177786.

97 Stern Y. Cognitive reserve and Alzheimer disease.Alzheimer Dis Assoc Disord2006; 20 (suppl 2): S6974.

98 Nelson PT, Abner EL, Schmitt FA, et al. Modeling the associationbetween 43 dierent clinical and pathological variables and theseverity of cognitive impairment in a large autopsy cohort of elderlypersons. Brain Pathol2008; published online Nov 19. DOI:10.1111/j.1750-3639.2008.00244.x.

99 Holmes C, Boche D, Wilkinson D, et al. Long-term eects of A42immunisation in Alzheimers disease: follow-up of a randomised,placebo-controlled phase I trial. Lancet2008; 372: 21623.

100 Petersen RC. Alzheimers disease: progress in prediction.Lancet Neurol2010; 9: 45.

101 Edison P, Archer HA, Gerhard A, et al. Microglia, amyloid, andcognition in Alzheimers disease: an [C](R)PK11195-PET and

[C]PIB-PET study. Neurobiol Dis 2008; 32: 41219.102 Ross BD, Bluml S, Cowan R, Danielsen E, Farrow N, Tan J. In vivoMR spectroscopy of human dementia. Neuroimaging Clin N Am1998; 8: 80922.

103 Duyckaerts C, Hauw JJ. Prevalence, incidence and duration ofBraaks stages in the general population: can we know?Neurobiol Aging1997; 18: 36269, 8992.

104 Corder EH, Saunders AM, Strittmatter WJ, et al. Gene dose ofapolipoprotein E type 4 allele and the risk of Alzheimers disease inlate onset families. Science 1993; 261: 92123.

105 Reiman EM, Chen K, Alexander GE, et al. Functional brainabnormalities in young adults at genetic risk for late-onsetAlzheimers dementia. Proc Natl Acad Sci USA 2004; 101: 28489.

106 Scarmeas N, Habeck CG, Stern Y, Anderson KE. APOEgenotypeand cerebral blood flow in healthy young individuals.JAMA 2003;290: 158182.

107 Ghebremedhin E, Schultz C, Braak E, Braak H. High frequency ofapolipoprotein E 4 allele in young individuals with very mildAlzheimers disease-related neurofibrillary changes. Exp Neurol1998; 153: 15255.

108 Reiman EM, Chen K, Liu X, et al. Fibrillar amyloid-beta burden incognitively normal people at 3 levels of genetic risk for Alzheimersdisease. Proc Natl Acad Sci USA 2009; 106: 682025.

109 Roses AD. Apolipoprotein E aects the rate of Alzheimer diseaseexpression: beta-amyloid burden is a secondary consequencedependent on APOEgenotype and duration of disease.J Neuropathol Exp Neurol1994; 53: 42937.

110 Drzezga A, Grimmer T, Henriksen G, et al. Eect ofAPOEgenotype on amyloid plaque load and gray matter volume inAlzheimer disease. Neurology2009; 72: 148794.

111 Harold D, Abraham R, Hollingworth P, et al. Genome-wideassociation study identifies variants at CLUand PICALMassociatedwith Alzheimers disease. Nat Genet2009; 41: 108893.

112 Lambert JC, Heath S, Even G, et al. Genome-wide association study

identifies variants at CLUand CR1 associated with Alzheimersdisease. Nat Genet2009; 41: 109499.

113 Buckner RL, Snyder AZ, Shannon BJ, et al. Molecular, structural,and functional characterization of Alzheimers disease: evidence fora relationship between default activity, amyloid, and memory.J Neurosci 2005; 25: 770917.

114 Sperling RA, Laviolette PS, OKeefe K, et al. Amyloid deposition isassociated with impaired default network function in older personswithout dementia. Neuron 2009; 63: 17888.

115 Buckner RL, Sepulcre J, Talukdar T, et al. Cortical hubs revealed byintrinsic functional connectivity: mapping, assessment of stability,and relation to Alzheimers disease.J Neurosci 2009; 29: 186073.

Documents

ModelBiomarkersADcascadeLancetNeuro10