State of the Art

Radiology

Jeffrey R. Petrella, MD Neuroimaging and Early

R. Edward Coleman, MD
P. Murali Doraiswamy, MD Diagnosis of Alzheimer
Index terms:
Disease: A Look to the Future1
Alzheimer disease, 13.83
Brain, CT, 13.12111
Brain, function Alzheimer disease (AD), a progressive neurodegenerative disorder, is the most
Brain, MR, 13.121411, 13.121412, common cause of dementia in the elderly. Current consensus statements have
13.121416, 13.12144, 13.12145 emphasized the need for early recognition and the fact that a diagnosis of AD can
Brain, radionuclide studies,
13.12162, 13.12163
be made with high accuracy by using clinical, neuropsychologic, and imaging
State of the Art assessments. Magnetic resonance (MR) or computed tomographic (CT) imaging is
recommended for the routine evaluation of AD. Coronal MR images can be useful
Published online before print to document or quantify atrophy of the hippocampus and entorhinal cortex, both
10.1148/radiol.2262011600
of which occur early in the disease process. Both volumetric and subtraction MR
Radiology 2003; 226:315–336
techniques can be used to quantify and monitor dementia progression and rates of
Abbreviations: regional atrophy. MR measures are also increasingly being used to monitor treat-
AD ⫽ Alzheimer disease ment effects in clinical trials of cognitive enhancers and antidementia agents.
FDG ⫽ fluorodeoxyglucose Positron emission tomography (PET) and single photon emission CT offer value in
MCI ⫽ mild cognitive impairment
MMSE ⫽ Mini-Mental State
the differential diagnosis of AD from other cortical and subcortical dementias and
Examination may also offer prognostic value. In addition, PET studies have demonstrated that
NAA ⫽ N-acetylaspartate subtle abnormalities may be apparent in the prodromal stages of AD and in subjects
who carry susceptibility genes. PET ligands are in late-stage development for dem-
1
From the Departments of Radiology onstration of amyloid plaques, and human studies have already begun. Functional
(J.R.P., R.E.C.) and Geriatric Medicine MR– based memory challenge tests are in development as well.
and Psychiatry (P.M.D.), Duke Univer- © RSNA, 2003
sity Medical Center, Duke Hospital
North, Rm 1513, Erwin Rd, Durham,
NC 27710. Received September 28,
2001; revision requested December 3;
revision received February 25, 2002; ac-
cepted March 14. J.R.P. supported by Alzheimer disease (AD) is a progressive neurodegenerative disorder associated with disrup-
grants from the RSNA Research and Ed- tion of neuronal function and gradual deterioration in cognition, function, and behavior
ucation Foundation and the National In- (1). It is the most common cause of dementia in the elderly and affects approximately 2– 4
stitute of Aging (R01AG019728-01).
million individuals in the United States and more than 30 million worldwide. Age is a
P.M.D. supported by grants from the
American Federation of Aging Research strong risk factor, with the disease affecting approximately 8% of individuals over the age
and the National Institutes of Health. of 65 years and 30% over the age of 85 years in developed countries. The progression of AD
Address correspondence to J.R.P. (e- is gradual, and the average patient lives 8 –10 years after onset of symptoms. Thus, with the
mail: [email protected]). growth of the older population in developed nations, the prevalence of AD is expected to
© RSNA, 2003 triple over the next 50 years (2). Female-to-male prevalence is 70%, likely related to
increased life expectancy in women. The annual cost of the disease in the United States
alone, including medical, long-term, and home care, as well as loss in productivity, is
currently estimated at $100 billion (3). Not only is the financial burden substantial, but the
psychologic and emotional burden on patients and their families is even greater.
With the recent availability of several effective pharmaceutical agents for treatment of
AD symptoms, along with several new agents undergoing clinical trials, we are entering a
new age in the treatment of AD. Current consensus statements have emphasized the need
for early recognition; thus, there is an urgent need to develop sensitive markers that may
serve as adjuncts to current clinical and neuropsychologic tests to facilitate detection
and/or monitoring of early brain changes suggestive of AD. Such markers may also
facilitate early-intervention studies to prevent or slow disease progression.
Conventional structural neuroimaging, such as computed tomography (CT) or magnetic
resonance (MR), has long played a supportive role in the diagnosis of memory disorders
and is recommended for the routine evaluation of AD. However, because structural
changes may not be detected at visual inspection until late in the course of the disease,
more contemporary structural imaging techniques, such as serial volumetric imaging and
voxel compression subtraction, have emphasized a quantitative approach capable of

315
 aiding in detection of subtle changes not
readily apparent on routine images ob-
tained at a single time point. Likewise,
functional imaging modalities that dem-
onstrate physiologic changes in the brain,
Radiology

including positron emission tomography

(PET), single photon emission CT (SPECT),
and functional MR imaging, also have
the potential to enable identification of
more subtle pathologic changes earlier
during the disease course and, therefore,
may have equal or greater potential in
comparison with structural imaging mo-
dalities such as CT and conventional MR
imaging.
This article will address the evolving
role of contemporary neuroimaging in
the evaluation of memory disorders, par-
ticularly AD. The intent is to focus pri-
marily on AD; therefore, discussion of
other causes of dementia is presented, in
less detail, for the purpose of differential
diagnosis. We will begin by briefly de-
scribing the current understanding of the
neuropathology and genetics of AD, be-
Figure 1. Two theories of how damage occurs in AD. (a) From outside the cell, amyloid ␤
cause these will have a direct bearing on
peptides secreted by brain cells are normally soluble, and any excess is cleared away. When these
diagnosis and treatment. Second, we will peptides become insoluble, however, they collect in the space between cells. Amyloid fibrils are
discuss pathologic and clinical diagnosis “herded” together by chaperone proteins. The large plaques that form then damage brain cells
of memory disorders, followed by a de- and attract reactive cells, microglia and astrocytes, which cause further damage. (b) From inside
scription of the current role of structural the cell, tau proteins, which normally stabilize microtubules in brain cells, undergo abnormal
neuroimaging evaluation. Third, we will chemical changes and assemble into spirals called paired helical filaments, thus creating tangles
that disrupt cell functions and lead to cell death. (Images courtesy of Dr John Trojanowski and Dr
give the reader perspective on the poten-
Virginia M. Y. Lee, University of Pennsylvania Medical Center, Philadelphia.)
tial role of functional imaging in the
broader scheme of clinical diagnosis and
conventional imaging. Last, we will ad-
Senile Plaques cellular fluids throughout life. Increased
dress what future steps are required for
production of amyloid ␤ protein leads to
the implementation of advanced imag- All patients with AD develop neuritic
precipitation, aggregation, and progres-
ing techniques as useful diagnostic tools plaques, which are extracellular aggrega-
sive deposition in the form of plaques.
for work-up in the patient with demen- tions of amyloid protein intimately asso-
Plaque buildup is thought to incite an
tia. ciated with dystrophic axons and den-
inflammatory and/or oxidative response,
drites (neurites), activated microglia, and
leading to progressive disruption of neu-
reactive astrocytes. Aggregations of amy-
PATHOLOGY AND ETIOLOGY loid protein that lack altered neurites and ronal function and injury in the hip-
glia are known as diffuse plaques and pocampal complex and cerebral cortex
Over the past 2 decades, rapid advances may be found in normal aging, as well as (6).
have occurred in our molecular genetic in degenerative diseases other than AD. Recent evidence suggests that genetic
and cell biologic understanding of the The primary component of these plaques alterations underlying familial AD result
pathogenesis of AD. Pathologically, AD is is the amyloid ␤ protein. This protein is in increased production of amyloid ␤
characterized by damage to the large cor- characteristic of AD and structurally dis- protein (Table 1). Mutations in the gene
tical neurons subserving cognition, ini- tinct from the amyloid-forming proteins for amyloid precursor protein, leading to
tially in the temporal lobes and later in found in systemic amyloidosis. abnormal cleavage, are located on chro-
the remaining neocortex and association Release of amyloid ␤ protein results mosome 21 and are present in patients
areas. Damage is believed to occur owing from abnormal cleavage of the amyloid with the rare early-onset familial form of
to mechanisms outside the neuron, as precursor protein, a glycoprotein found the disease. Cloning of these mutations
well as inside the neuron, and is charac- in high concentrations in the cell mem- has lead to increased production of amy-
terized by the appearance of extracellular brane of virtually all mammalian cells loid ␤ protein in cell cultures, and plasma
senile (amyloid) plaques and intracellular but in especially high concentrations in levels of amyloid ␤ protein are elevated
neurofibrillary tangles (4) (Fig 1). Much neuronal cell membranes. The enzymes in some mutation carriers, even presymp-
work over the past 2 decades has focused responsible for normal and abnormal tomatically. Mutations of the presenilin
on these two pathologic components of cleavage of amyloid precursor protein (␣, 1, or PS-1, and presenilin 2, or PS-2,
the disease, with application of molecu- ␤, ␥ secretases) have been described and genes, on chromosomes 1 and 14 respec-
lar genetic, biochemical, and morpho- targeted for drug therapy. Amyloid pre- tively, are also associated with the early-
logic techniques to study them in great cursor protein is normally secreted by onset familial form of AD. Moreover,
detail (5). both brain and nonbrain cells into extra- postmortem studies in patients with tri-

316 䡠 Radiology 䡠 February 2003 Petrella et al

 (11). Stage I is defined by the appearance
TABLE 1 of neurofibrillary tangle– bearing neu-
Genetic Factors Predisposing to AD: Relationship to Amyloid-␤ Phenotype
rons in the transentorhinal region, with
Age of progression to the entorhinal region and
Chromosome Gene Defect Onset (y) Amyloid-␤ Phenotype hippocampal formation in stage II. These
Radiology

21 ␤ APP mutations 50 Production of total amyloid-␤ peptides stages represent the presymptomatic
or of amyloid-␤ 42 peptides phase of the disease. Further progression
19 Apolipoprotein ⑀4 ⱖ60 Density of amyloid-␤ plaques and leads to stages III and IV, which are char-
polymorphism vascular deposits acterized by severe involvement of the
14 PS-1 mutations 40s and 50s Production of amyloid-␤ 42 peptides
1 PS-2 mutations 50s Production of amyloid-␤ 42 peptides entorhinal region, amygdala, and hip-
pocampal formation. This involvement
Note.—Adapted, with permission, from reference 5. leads to interruption of the limbic loop,
which is responsible for data transfer
back and forth between the neocortex
and the hippocampal formation. In addi-
somy 21 (Down syndrome) and in those severity of dementia (7), whereas the to- tion, there is also involvement of many
who carry susceptibility genes for AD tal density of neurofibrillary tangles does subcortical nuclei with diffuse projec-
have shown that amyloid ␤ protein accu- (8,9). Indeed, some elderly individuals tions to the cortex, among them the cho-
mulation in the cortex is an early invari- without dementia have as much plaque linergic system of the basal forebrain, in-
ant event in AD pathology, sometimes deposition as have those with severe de- cluding the nucleus basalis of Meynert.
occurring decades before symptoms be- mentia. However, the neuritic variety of Patients with disease at these stages may
come apparent (5). Given this evidence, plaque associated with dystrophic axons present with symptoms of mild cognitive
current thinking suggests that AD may and dendrites and reactive glial cells ap- impairment (MCI) or prodromal AD, al-
represent not a single disease entity but pears to be a more specific feature of AD though symptoms may be masked by a
rather a syndrome resulting from differ- than is the neurofibrillary tangle. Fur- high cognitive-reserve capacity. In stage
ent genetic determinants that lead to a thermore, results of a number of in vitro V, there is extension to all major regions
common phenotype with amyloid depo- and in vivo studies have shown amyloid of the cerebral cortex, from association
sition and neuronal degeneration in spe- ␤ protein to be directly toxic to neurons, areas into primary areas. In stage VI, even
cific brain regions (6). leading to phosphorylation of tau pro- the primary sensory areas are severely in-
tein, the principal component of neuro- volved, although the motor area contin-
fibrillary tangles (10). ues to be relatively spared. Stages V and
Neurofibrillary Tangles
VI are generally characterized by severe
Neurofibrillary tangles consist of intra- dementia.
Pathologic Progression
neuronal bundles of paired helical fila-
ments composed of an abnormal micro- The insidious onset and gradual pro-
tubule-associated protein known as tau. gression of symptoms in patients with PATHOLOGIC DIAGNOSIS
Virtually all brains of AD patients con- AD are thought to parallel the progres-
tain neurofibrillary tangles. Unlike the sion of AD-related brain destruction from AD is characterized pathologically by re-
gene for amyloid precursor protein, how- entorhinal cortex to hippocampus to ductions in the number of large cortical
ever, there is no evidence of defects in neocortex (11). In typical cases, there ini- neurons in temporal and frontal cortex
the gene encoding for tau in patients tially is isolated impairment of learning and by the appearance of senile (amy-
with familial forms of AD. In fact, neuro- and short-term memory, without alter- loid) plaques and neurofibrillary tangles
fibrillary tangles, consisting of similar ation in other domains of cognition or (4). In living subjects, only a provisional
forms of modified tau protein, are de- consciousness. This is followed by changes diagnosis of possible or probable AD may
tected in other etiologically distinct dis- in long-term memory, personality, orien- be made, by using clinical, laboratory,
eases such as subacute sclerosing panen- tation, and executive function. After cog- and imaging evidence. Definitive diagno-
cephalitis, Hallervorden-Spatz disease, nitive losses, there are behavioral problems sis of AD is made by means of pathologic
Parkinson dementia complex, and de- (eg, hallucinations, delusions). Finally, there examination of tissue derived from au-
mentia pugilistica. Despite the lack of a is deterioration of language, visuospatial topsy or brain biopsy. For this reason,
direct etiologic connection, the results of skills, and, eventually, motor function, criteria have been created to standardize
numerous studies have demonstrated el- which leads to loss of activities of daily the pathologic diagnosis.
evated levels of tau protein in the cere- living. In the past, the two most extensively
brospinal fluid of patients with AD, com- Three stages in the gradual evolution used pathologic criteria have been the
pared with those levels in age-matched of plaque deposition have been described National Institute on Aging consensus
control subjects. (11). Stage A is characterized by deposi- criteria (1) and the Consortium to Estab-
At present, the molecular and etiologic tion in the basal temporal neocortex, or lish a Registry for Alzheimer’s Disease, or
relationship between plaques and tangles entorhinal cortex. There then is exten- CERAD, criteria (12). The National Insti-
is not fully understood. The bulk of cur- sion through the hippocampal formation tute on Aging criteria are based on sub-
rent data suggests that these two lesions in stage B, eventually leading to deposits ject age and clinical history, taking into
might be formed independently of each in virtually all cortical areas, including account the number of senile plaques per
other. In fact, the distribution and den- the highly myelinated primary areas of high-power microscopy field and the
sity of both diffuse and neuritic senile the neocortex in stage C. presence of neurofibrillary tangles in spe-
plaques have not been consistently There are reported to be six stages in cific regions of the neocortex, hippocam-
shown to correlate with the presence or the evolution of neurofibrillary tangles pus, and subcortical gray matter. The

Volume 226 䡠 Number 2 Neuroimaging and Early Diagnosis of Alzheimer Disease 䡠 317
 CERAD criteria use only the presence and
density of neuritic senile plaques in the
neocortex for establishment of the diag-
nosis of AD. The density is classified as
“sparse,” “moderate,” or “severe” and is
then combined with the age of the sub-
Radiology

ject to yield an age-related plaque score.

The plaque score is integrated with clin-
ical information for the diagnosis of “def-
inite,” “probable,” or “possible” AD.
More recently, the National Institute
on Aging–Reagan Institute criteria, based
on the number of both plaques and tan-
gles in the neocortex and the limbic and
paralimbic regions, were introduced
(13,14). These criteria essentially com-
bined the CERAD criteria, which are
based on plaque frequency, and the stag-
ing criteria of Braak and Braak (15),
which are based on neurofibrillary tangle
distribution and frequency. The possibil-
ity of coexisting pathologic conditions
other than AD is also recognized.
Unfortunately, because these criteria
place varying degrees of dependence on
the number and type of senile plaques
and neurofibrillary tangles in various sec-
tions of the brain, as well as on the pres-
ence of a clinical history of dementia, the
same diagnosis may not be rendered at
autopsy in a given patient (16). In a re-
cent study (17), however, a good correla-
tion was shown between the National In-
stitute on Aging–Reagan Institute criteria
and the previous two criteria. The Na-
tional Institute on Aging–Reagan Insti-
tute criteria offer the possibility of better
diagnostic refinement because of the rec-
ognition that dementia in the elderly
may arise from more than one disorder.
In dementia with Lewy bodies, for exam-
ple, there may be considerable clinical
and pathologic overlap with AD.

CLINICAL DIAGNOSIS

The accuracy of a clinical diagnosis of AD

depends on the stage of disease and the
clinical setting. In an academic memory-
disorders clinic setting, the diagnostic ac-
curacy can exceed 90% (18 –20), while
accuracy may be substantially lower in
the general community setting where Figure 2. NINCDS/ADRDA (National Institute of Neurological and Communicative Disorders
and Stroke/Alzheimer’s Disease and Related Disorders Association) criteria (21) for probable AD.
there is less consistency and reliance on
EEG ⫽ electroencephalogram.
established diagnostic criteria. In this sec-
tion, we will briefly review the clinical
criteria for diagnosing AD and other
common dementias of old age. We will Institute of Neurological and Communi- mentia of the Alzheimer type. Probable
also present current treatment options. cative Disorders and Stroke/Alzheimer’s AD can be categorized as mild (early),
Disease and Related Disorders Associa- moderate (middle), or severe (late) de-
tion (NINCDS/ADRDA) criteria (21) (Fig mentia. Clinical work-up includes a de-
Diagnostic Criteria
2) for possible or probable AD or the Di- tailed history from the patient and/or an
AD.—Clinical criteria used for a provi- agnostic and Statistical Manual of Mental informant, a general physical and mental
sional diagnosis consist of the National Disorders (DSM-IV) criteria (22) for de- status examination, neurologic assess-

318 䡠 Radiology 䡠 February 2003 Petrella et al

 vitamin E, ginkgo biloba, and psychotropic
drugs.
Differential diagnosis with other forms of
dementia.—In developed countries, AD
accounts for approximately 50%–70% of
Radiology

all cases of dementia in the elderly. Diag-

nostic criteria require that AD be distin-
guished from other causes of dementia
such as vascular dementia, frontotempo-
ral dementia, dementia with Lewy bod-
ies, normal-pressure hydrocephalus, de-
pression-related cognitive impairment,
and intracranial mass.
Vascular dementia, which accounts for
about 15% of all dementias in the United
States, can be distinguished from AD by
its more sudden onset and association
with vascular risk factors. Vascular de-
mentia is characterized by a stepwise
course with periods of stability followed
Figure 3. Over the clinical course of AD, patients will demonstrate progressive declines in by sudden decline in cognitive function.
functional ability that correlate with MMSE scores. In the preclinical phase, also called MCI, Patients may experience focal neurologic
patients with MMSE score greater than 23 will demonstrate minimal impairment— generally, deficits after a sudden decline, such as
mild memory loss—while functioning normally and independently. The typical length of the
MCI phase remains undetermined. Onset of mild AD is indicated by MMSE score of 20 –23; these
slurred speech or sensorimotor dysfunc-
patients exhibit gradual alterations in cognition, function, behavior, and mood. Forgetfulness tion. Clinical criteria for diagnosis of
and repetitive questions are hallmarks; daily function begins to become impaired. In moderate probable vascular dementia were created
AD (MMSE score of 10 –19), cognitive impairments progressively deteriorate and now include in 1993 and are known as the National
short-term memory loss (eg, difficulty in recalling recent conversations, forgetting to keep Institute of Neurological Disorders and
appointments, inability to remember recent events). Patients begin to have trouble with verbal Stroke–Association Internationale pour la
fluency; specifically, increasing difficulty in remembering the correct word. Severe stage is
reached when MMSE scores decline to less than 10. At this point, patients exhibit behavioral
Recherche et l’Enseignement en Neuro-
changes that include agitation, delusion, aggression, wandering, and hallucination. Sleep pat- sciences (NINDS-AIREN) criteria (Fig 4)
terns are altered, and the patient eventually becomes completely dependent on others for (24). The term vascular dementia is pre-
dressing, feeding, and bathing. The clinical progression from onset of mild AD to onset of severe ferred over multiinfarct dementia, be-
AD, although variable, is about 10 years. cause the former is an acknowledgment
of the many vascular causes that can con-
tribute to dementia. Control of vascular
ment, and objective tests such as the as the most prominent manifestation, risk factors is the treatment of choice.
Mini-Mental State Examination (MMSE) with a deficit in one or more cognitive Cholinesterase inhibitors are also in-
to help assess cognitive dysfunction. In domains, including language, orientation, creasingly being used to treat vascular de-
addition, tests are also available to quan- or executive functioning (ie, planning, or- mentia.
tify activities of daily living, behavioral ganizing, abstracting). The course of the Frontotemporal dementia comprises a
dysfunction, and caregiver stress. Staging disease is variable but is characterized by group of dementias (eg, Pick disease) and
instruments such as the Clinical Demen- gradual onset with progressive decline in accounts for approximately 5%–10% of
tia Rating Scale and the Global Deterio- memory, with initial sparing of sensory cases of dementia. Frontotemporal de-
ration Scale are frequently used in aca- and motor function. The memory impair- mentia is clinically characterized by be-
demic settings. ment may initially manifest itself as a def- havioral and language disturbances that
Laboratory evaluation is performed to icit in learning or recall of new informa- may precede or overshadow memory def-
exclude medical diseases that may cause icits. There currently is no treatment for
tion, followed by loss of remote memory.
dementia (eg, hypothyroidism, vitamin this condition.
Behavioral and psychologic disturbances
B12 deficiency). Cerebrospinal fluid as- Normal-pressure hydrocephalus is
(depression, delusions, wandering, para-
says of amyloid ␤-42 and tau protein characterized by a classic triad of gait dis-
noia, anxiety, sleep changes) also develop
have been shown to have high sensitivity turbance (25), dementia, and inconti-
in virtually 100% of patients over the
(⬃80%) and moderate specificity for the nence. There is evidence that ventricular
diagnosis of AD but are not widely used course of the disease (Fig 3). Differentiation shunting in a select group of these pa-
in routine practice. Other tests such as from dementia associated with depression tients may improve symptoms, provided
plasma assays of amyloid, cerebrospinal (pseudodementia) may be difficult. either pronounced gait difficulty or the
assays of neurotransmitter metabolites, AD is characterized by cholinergic def- full triad are present; however, there is
and urine-based assays of neural thread icits. There are currently three cholinest- little effect in patients in whom dementia
protein have not been widely accepted as erase inhibitors in wide use as treatments is the first or primary symptom (26,27).
providing diagnostic value. Genetic test- of choice. Such treatments improve or Dementia with Lewy bodies is being
ing is discussed elsewhere in this article. stabilize cognition, and some evidence diagnosed more frequently; study results
Although variants with atypical mani- suggests that early initiation of therapy have demonstrated that it is responsible
festations have been described, the diag- may delay institutionalization (23). Many for approximately 25% of dementias. The
nosis of AD requires memory impairment patients are also treated “off label” with clinical manifestation may be similar to

Volume 226 䡠 Number 2 Neuroimaging and Early Diagnosis of Alzheimer Disease 䡠 319
 that of AD or the dementia associated
with Parkinson disease. Patients typically
present with one of three symptom com-
plexes: detailed visual hallucinations,
Parkinson-like symptoms, or alterations
in alertness and attention (3). Pathologi-
Radiology

cally, the disease is characterized by the

presence of Lewy bodies in various re-
gions of the hippocampal complex, sub-
cortical nuclei, and neocortex, with a
variable number of diffuse amyloid
plaques; however, the neurites (axons
and dendrites) and neurofibrillary tan-
gles seen in AD are not seen in this dis-
ease. In 1996, consensus criteria were de-
veloped for the diagnosis of probable
dementia with Lewy bodies (28) (Fig 5).
Cholinesterase inhibitors are currently
the treatment of choice for this condi-
tion.
MCI.—Increasing evidence now sug-
gests that the early pathogenic process of
AD is protracted and may extend over
decades (29). This preclinical stage of AD
appears to be separated into two stages:
an extended “latent period,” where there
may be no observable symptoms of the
disease, followed by a shorter “prodromal
phase,” where mild symptoms are ob-
served but preclude a clinical diagnosis.
The prodromal phase is characterized by
progressive isolated memory deficit, a
condition now increasingly being termed
in research settings as MCI. In contrast to
MCI, patients with mild AD have more
prominent deficits in one or more other
cognitive domains (eg, language, execu-
tive function), as well as in daily func-
tioning, and their memory deficits are
usually also more pronounced.
Results of neuropsychologic studies of
early AD have demonstrated that specific
patterns of recent memory impairment
help distinguish between early-stage AD,
normal aging, and other dementing con-
ditions (30). Based on these and other
data, research criteria have been devel-
oped for identification of patients with
MCI (“prodromal AD”), and such criteria
are currently being used for selecting pa-
tients for ongoing-intervention trials
(31). Specifically, MCI is characterized by
a prominent and relatively isolated im-
pairment in so-called secondary mem-
ory, or memory for information after
short delays. Patients with MCI may also
have impairments of working memory.
This deficit in patients with MCI is be-
Figure 4. NINDS-AIREN (National Institute of Neurological Disorders and Stroke and Associa-
lieved to be to caused by impaired acqui-
tion Internationale pour la Recherche et l’Enseignement en Neurosciences) criteria (24) for
sition or consolidation of new informa- probable vascular dementia. ACA ⫽ anterior cerebral artery, CVD ⫽ cerebrovascular disease,
tion into permanent memory stores, PCA ⫽ posterior cerebral artery.
presumably related to damage to the me-
dial temporal lobe and/or specific pre-
frontal–temporal lobe circuits (Fig 6).

320 䡠 Radiology 䡠 February 2003 Petrella et al

 There is still some controversy about
the definition of MCI. Some experts be-
lieve that MCI simply reflects a very mild
form of AD and that virtually all patients
with MCI may have pathologic changes
Radiology

of AD (33). Others believe that patients

with MCI represent a more heterogenous
group, with some proportion of subjects
staying stable or even “reversing” back to
normal and with some patients progress-
ing to develop additional deficits and
meeting criteria for AD.
Although the definition of MCI has
differed somewhat from study to study,
there are now several longitudinal inves-
tigations in subjects with MCI. Peterson
(34) has followed up MCI subjects and
demonstrated that, on average, approxi-
mately 40% of carefully characterized
subjects with MCI will progress to meet
criteria for AD over a 4-year period. At
present, there is no reliable test to predict
which subjects will convert or progress,
although all of the aforementioned risk
factors may play a role. Symptoms of for-
getfulness can be ambiguous and easily
confused with the benign forms of mem-
ory change associated with normal aging
(35) (Fig 7). Thus, there is an urgent need
to develop sensitive markers that may
serve as adjuncts to current clinical and
neuropsychologic tests to facilitate detec-
tion and/or monitoring of early brain
Figure 5. Proposed criteria of McKeith et al (28) for probable dementia with Lewy bodies (DLB). changes suggestive of AD. Such markers
may also facilitate early intervention
studies to prevent or slow further neuro-
nal loss.
The criteria for identification and clas-
sification of subjects with MCI are evolv-
ing. Current criteria for identifying sub-
jects with MCI for secondary prevention
trials of AD usually depend on one or
more of the following factors: (a) impaired
memory performance on a normalized
objective verbal memory delayed-recall
test (eg, word list or paragraph); (b) recent
history of symptomatic worsening in
memory, supported by an informant;
(c) normal or near-normal performance
on global cognitive tests (eg, MMSE
score ⬎ 24), as well as on an activities
of daily living scale; (d) global rating of
0.5 (questionable dementia) on the
Figure 6. Model of neural deficits in early AD. Neural connections associated with normal Clinical Dementia Rating Scale, with a
memory function involve frontal and temporal lobes. Somatosensory, visual, and auditory
score of at least 0.5 on the memory
information proceed from primary and association cortex to prefrontal cortex, located in the
posterior frontal lobe. This region plays a major role in executive function (ie, organizing and scale; (e) absence of other factors that may
directing attention), as well as in working memory, acting as a “mental scratchpad” for short-term better explain memory loss; and (f) pres-
information needed to perform a task, such as dialing a phone number. Part of this information ence of other risk factors such as family
may be consolidated, branching to the medial temporal lobe region via the entorhinal cortex history, older age, and presence of the apo-
(medial temporal lobe) and into the hippocampal complex. Projections from the hippocampal lipoprotein ⑀4 allele. While these criteria
complex can transfer long-term information back to prefrontal cortex. This back-and-forth
may seem relatively easy to apply, ques-
pathway between prefrontal cortex and medial temporal lobes is known as the limbic loop and
is considered important for emotional stability, learning and memory function, and regulation of tions still remain with regard to which of
autonomic and endocrine functions. It is precisely these areas that are particularly susceptible to these criteria to use, which memory
the pathologic changes of AD (11). (Adapted, with permission, from reference 32.) test(s) to use, what cutoffs to apply to de-

Volume 226 䡠 Number 2 Neuroimaging and Early Diagnosis of Alzheimer Disease 䡠 321
 termine impairment, and how to deal with
fluctuations in performance.

Genetic Testing
Genetic testing can be performed by
Radiology

using numerous markers (Table 1), many

of which facilitate a diagnosis of early-
onset dementia. Mutations of the PS-1, Figure 7. Conceptual continuum of memory decline in early AD.
PS-2, and amyloid precursor protein Diagram demonstrates overlap between healthy control subjects and
genes on chromosomes 1, 14, and 21, patients with MCI and between patients with mild AD and those with
respectively, are associated with early on- MCI.
set of the familial—and extremely rare—
form of AD. The only useful marker for
the more common late-onset (sporadic
form) dementia is the apolipoprotein ⑀
allele on chromosome 19. Results of ge-
netic studies have shown that the ⑀4 al-
lele is associated with a high incidence of
AD, whereas the ⑀2 allele may be a pro-
tective factor. The ⑀3 allele is thought to
represent no increased or decreased risk
(36). Because the ⑀4 allele is more preva-
lent in patients with AD, it may be useful
for differential diagnosis in patients with
memory disorders. However, it is cur-
rently not recommended as a predictor of
prognosis in asymptomatic subjects. The
⑀4 allele is found in about 30% of healthy
subjects and is absent in approximately
30%– 40% of patients with AD (37).

ROLE OF STRUCTURAL
NEUROIMAGING
Conventional CT and MR Imaging
Conventional CT and MR imaging are
used routinely in the work-up of patients Figure 8. Coronal T2-weighted fast spin-echo MR images (repetition time msec/echo time msec,
2,700/80) in (a, b) a patient with AD and (c, d) an age-matched control subject. (a, b) The patient
with dementia; however, because of their
with AD has severe bilateral hippocampal atrophy (arrows). Compare this with (c, d) normal
low sensitivity and specificity for the di- hippocampus (arrows) in the control subject. Note that b is a close-up of a, and d is a close-up of
agnosis of AD, they are primarily used as c. (Images courtesy of Daniel P. Barboriak, MD, Duke University Medical Center, Durham, NC.)
an adjunct to help assess the degree of
atrophy and rule out other causes of de-
mentia such as normal-pressure hydro-
cephalus, vascular dementia, or intracra- aqueductal, cerebellar, or dorsal thalamic images in patients with AD (42,43). Early
nial mass. Currently, there is no reason to changes (eg, alcohol-related dementia) investigators used CT to assess cortical,
support CT over MR in the evaluation of (40). At centers where fast spin-echo cerebellar, and hippocampal atrophy on
patients with dementia, except in cases rather than conventional spin-echo im- transverse images; however, the addition
where MR is contraindicated or not aging is used, T2*-weighted or gradient- of the coronal imaging plane with MR
readily available or affordable. echo imaging may be needed to help assess provided better qualitative assessments
Standard MR pulse sequences at most the degree of iron deposition (eg, parkin- of the hippocampal region (Fig 8). Stud-
centers consist of nonenhanced trans- sonian disorders) (41). T1-weighted and ies in which T1-weighted MR imaging
verse spin-echo T1-weighted imaging to T2-weighted sequences may also be used to was used in evaluation of the anterior
help assess the presence and degree of help assess for the presence of ventricular hippocampus have been particularly use-
generalized (eg, AD) or focal (eg, cerebellar dilatation or intracranial mass. Thin-sec- ful. Data from a study (44) with 222 sub-
degenerative diseases, Pick disease, AD) at- tion (ⱕ3-mm-thick) coronal T1-weighted jects (controls and patients with various
rophy (38,39). Transverse dual-echo long- images obtained in a plane orthogonal to forms of dementia) in which a visual rat-
repetition-time imaging, including proton- the long axis of the hippocampus are use- ing scale was used to assess temporal lobe
density and/or fluid-attenuated inversion- ful for assessment of the degree of medial atrophy suggest that sensitivities and
recovery and T2-weighted imaging, is also temporal lobe and hippocampal atrophy. specificities in the area of 85% may be
used in the assessment for periventricular AD.—Medial temporal lobe atrophy, obtained for patients with AD.
and subcortical white matter hyperintensi- particularly of the amygdala, hippocam- Normal-pressure hydrocephalus.—Normal-
ties; cortical or deep gray matter lacunar pus, and parahippocampal gyrus, can be pressure hydrocephalus, an idiopathic
infarcts (vascular dementia); and peri- seen with higher frequency on structural form of communicating hydrocephalus,

322 䡠 Radiology 䡠 February 2003 Petrella et al

 clinical diagnosis (24). This diagnosis
cannot be made on the basis of imaging
findings alone, as white matter hyperin-
tensities or lacunae on MR images do not
necessarily always lead to dementia or
Radiology

notable cognitive impairment. Although

study results have associated cognitive
impairment with MR hyperintensities,
the nature and strength of these associa-
tions have varied from one study to an-
other (44,50 –52). This variability may be
due in part to the confounding effects of
high cognitive reserve or education level,
which may mask a neurocognitive de-
cline. The volume of white matter lesions
does, however, correlate with the pres-
ence of cerebrovascular risk factors (53).
Patients with AD may also have abun-
dant white matter lesions, and there is
often overlap between vascular dementia
and AD, a so-called mixed dementia.
Thus, the presence of white matter le-
sions on T2-weighted MR images should
not be used to exclude a diagnosis of AD
(3). Likewise, several strategically placed
Figure 9. Coronal T1-weighted MR images demonstrate tracings of the hippocampus and infarcts may be sufficient to cause symp-
parahippocampal gyrus in a 75-year-old female control subject (left) and 73-year-old woman with toms of vascular dementia and should
AD (right). Outlines of the amygdala and hippocampus are indicated in the bottom images. not be viewed as unimportant.

is characterized by holoventricular en- Despite the existence of consistent im- Quantitative Imaging of Atrophy
largement out of proportion to sulcal en- aging findings, it should be noted that
largement. These findings are in contrast the diagnosis of normal-pressure hydro- Quantitative measures of generalized
to the findings of generalized atrophy, cephalus is made primarily on the basis cerebral atrophy were developed for both
where the degree of sulcal and ventricu- of clinical findings. Without a prior sus- CT and MR imaging in an attempt to
lar prominence are concordant. Radionu- pected clinical diagnosis, it is unlikely identify pathologic atrophy in early de-
clide cisternography, demonstrating ac- the diagnosis will be made on the basis of mentia, as distinguished from the atro-
tivity within the ventricular system at 24 imaging results alone. phy of “normal” aging. Such approaches
hours, has been used to help diagnose Vascular dementia.—Vascular dementia, included linear and volumetric CT mea-
normal-pressure hydrocephalus and de- or multiinfarct dementia, is often charac- surements of ventricles and subarach-
termine which patients are more likely to terized by multiple areas of high signal in- noid spaces and MR imaging measure-
respond to shunting (45). MR imaging tensity on T2-weighted, proton-density, or ments of cerebrospinal fluid and gray and
can also be useful in this regard. MR im- fluid-attenuated inversion-recovery MR white matter volumes (54 –56). Unfortu-
ages may show the presence of high sig- images. The high signal intensity may be nately, age-related atrophy in individuals
nal intensity in the periventricular and in white matter, basal ganglia, and/or thal- without dementia also occurs, causing
subcortical white matter on T2-weighted, amus. Focal infarcts or lacunae in strategic substantial overlap between the two
proton-density, or fluid-attenuated in- locations can also be associated with vas- groups and limiting the diagnostic utility
version-recovery images; this high signal cular dementia. Clinical onset of cognitive of such measurements. Subsequent work
intensity has been thought to represent deficits is characterized by a stepwise de- focused on the medial temporal lobes,
transependymal edema. These abnormali- cline, as opposed to the more continuous given that the pathologic substrate of AD
ties may, however, represent small-vessel decline in AD. Periods of abrupt decline is known to affect the hippocampus and
ischemic changes, reflecting the relation- may be followed by stable periods. Patients anatomically related areas such as the en-
ship between normal-pressure hydroceph- typically have cerebrovascular risk factors, torhinal cortex at the earliest stages of
alus and decreased blood flow in the including male sex, smoking, hyperten- the disease. Authors of early studies with
periventricular white matter (46). The pres- sion, and/or diabetes. Clinical characteris- CT who focused on this region applied a
ence of a prominent flow void in the cere- tics such as abrupt onset of symptoms of linear approach to measure medial tem-
bral aqueduct has been shown to correlate memory impairment, emotional lability, poral lobe atrophy (57,58) and a stereo-
well with a positive response to shunting somatic complaints, and focal neurologic logical approach to measure hippocam-
(47). More recently, a number of groups deficits, along with a history of stroke or pal volume (59). However, because of the
have demonstrated increased aqueductal transient ischemic attacks, often guide the small size and uneven shape of the hip-
cerebrospinal fluid flow, as measured with clinician in making a diagnosis of vascular pocampus, as well as considerations of
quantitative cine phase-contrast MR imag- dementia. section thickness, the transverse CT ap-
ing, as a good predictor of response to Modern diagnostic criteria emphasize proach turned out not to be well suited
shunting (48,49). the importance of imaging to support a for volumetric assessment.

Volume 226 䡠 Number 2 Neuroimaging and Early Diagnosis of Alzheimer Disease 䡠 323
 Quantitative hippocampal volume anal-
ysis with MR imaging turned out to be
better than that with CT, given the supe-
rior soft-tissue contrast capabilities of MR,
the lack of beam-hardening artifacts, and
Radiology

the multiplanar capability. Initial tech-

niques used linear and area measurements.
Subsequent techniques have used semiau-
tomated computer algorithms to calculate
hippocampal volumes. Authors of a num-
ber of initial studies describing MR-based
volume measurements reported extremely
high sensitivity in distinguishing patients
with AD from elderly control subjects;
however, many of these studies were
flawed by imaging methods that were not
state of the art, by variations in neuroana-
tomic boundary criteria, and by small and
highly selected samples, which made ex-
Figure 10. T1-weighted MR images with
trapolation to diagnosis of AD in a general
color voxel compression mapping overlay in a
setting problematic (60). The most widely 46-year-old man with familial AD. Images were
used volumetric technique is the semi- obtained in (a) transverse, (b) coronal, and
automated tracing-thresholding method, (c) sagittal planes over 30 months during
where a threshold is used to outline the which symptoms developed. Green and blue
high-contrast portions of the anatomic areas demonstrate interval volume loss, partic-
ularly in temporal lobes and in other neocor-
boundary and an interactive manual
tical areas. (Images courtesy of Nick C. Fox,
trace is used to outline the low-contrast MD, MRCP, National Hospital for Neurology
portions (61). Using the tracing-thresh- and Neurosurgery, London, England.)
olding method, Jack et al (62) reported a
sensitivity of 82% and a specificity of
80% in discriminating patients with AD regions in four patients with autosomal
from control subjects in a study of 220 dominant mutations for early onset AD, all
individuals (Fig 9). of whom developed the disease over a 5– 8-
Difficulties with single measurements year follow-up (Fig 10). The same tech-
of generalized and regional atrophy in- nique can be used to monitor the effects of
clude the large overlap in volume mea- therapeutic regimens in slowing the rate of
surements between control and AD pop- brain atrophy. It has been estimated that
ulations. Serial measurements may be detection of a 20% reduction in atrophy
jects who subsequently showed a decline rate would require approximately 200 pa-
more specific than single measurements
had a significantly greater rate of atrophy tients and 200 control subjects in a 1-year
for detecting differences between pa-
than did those whose condition remained placebo-controlled drug trial (68).
tients with AD and control subjects, as
clinically stable. Despite its promise, the lack of availabil-
serial measurements are less affected by
The potential of serial volume measure- ity of a completely automated method for
overlap between populations. Several
ments to aid in prediction of a subsequent segmenting small and irregular structures
groups have evaluated dementia popula-
development of AD and evaluation of the such as the hippocampus, as well as the
tions by using serial volume measure-
efficacy of therapeutic regimens in at-risk time-consuming and labor-intensive na-
ments in various memory-specific brain
populations, particularly those with MCI ture of the current methods, limits wide-
regions, including the medial temporal
lobes. Rates of hippocampal atrophy (65,66), has been evaluated. Fox et al (67) spread use of serial volume measurements
have correlated with cognitive status at used a voxel compression subtraction tech- in clinical practice at this time. Techniques
baseline and with change in cognitive nique to assess serial volume changes in for automated segmentation of structures
status over time. Jack et al (63) studied 22 individual subjects. The technique uses such as the hippocampus are in develop-
subjects with probable AD and 22 age- scaling, interpolation, and a rigid body reg- ment (69) and await more extensive testing
and sex-matched control subjects. The istration method to first bring the later and before they can be used in routine clinical
annualized rate of hippocampal atrophy earlier images into subvoxel alignment. A settings (Fig 11).
and temporal horn enlargement was ap- nonlinear registration algorithm using a
proximately two times greater in patients compressible viscous fluid model is then
with probable AD. More recently, in a applied to track local cerebral losses and ROLE OF FUNCTIONAL
study with 129 patients grouped accord- deformations, which are presented in the NEUROIMAGING
ing to probable AD, MCI, or age-matched form of a color overlay map on the base-
control, Jack et al (64) showed that rates line anatomic image. Fox et al (67) recently Conventional structural neuroimaging
of hippocampal atrophy lie along a con- reported use of voxel compression map- with CT or MR is recommended in the
tinuum from patients with AD to those ping to show progressive regional atrophy routine evaluation of patients with mem-
with MCI to age-matched control subjects. involving the temporoparietal cortex and ory disorders, to exclude treatable causes
Within the control and MCI groups, sub- medial temporal and posterior cingulate such as normal-pressure hydrocephalus

324 䡠 Radiology 䡠 February 2003 Petrella et al

 more subtle pathologic changes earlier
during the disease course.

SPECT Imaging
Radiology

Cerebral SPECT, which is based on

brain uptake of a technetium 99m– based
lipid-soluble radionuclide such as ethyl
cysteinate dimer or hexamethylpropyl-
ene amine oxime, is a widely available
technique for evaluation of brain perfu-
sion with a rotating gamma camera. Al-
though the technique does not yield ab-
solute values for cerebral blood flow, it
does yield semiquantitative or relative
values for comparison within and be-
tween patients when normalization by
cerebellar activity is used (70).
Patients with AD have typically dem-
onstrated a relative paucity of activity in
the temporoparietal regions, compared
with the activity in control subjects. Re-
sults of several studies have also shown
that the magnitude of the perfusion ab-
normality correlates with the severity of
cognitive impairment (71). In a prospec-
tive study with histologic confirmation
of over 200 dementia cases and 119 con-
trol cases (72), the technique was re-
ported to enable differentiation of pa-
tients with AD from healthy control
subjects with a high degree of sensitivity
and specificity (89% and 80%, respec-
tively). Moreover, the accuracy of a clin-
ical diagnosis of AD is improved with the
aid of this technique, especially in mild
cases of dementia, or “possible AD.” In a
study with histologic confirmation in 70
patients with dementia and 85 control
subjects (73), a positive SPECT scan in-
creased the pretest probability from 84%
to 92% in patients with a clinical diagno-
sis of “probable AD” and from 67% to
84% in patients with “possible AD.” A
negative SPECT scan decreased the prob-
ability from 84% to 70% in patients with
“probable AD” and from 67% to 52% in
patients with “possible AD.”
SPECT imaging also aids in the differ-
Figure 11. Three-dimensional model– based segmentation. Screenshot demonstrates elastic
model– based segmentation of limbic structures, including amygdalae (green and yellow) and ential diagnosis of patients with demen-
hippocampi (blue and red). (a) Segmentation is shown in three-dimensional surface rendering tia. Numerous studies have been per-
(bottom left) and two-dimensional overlay on T1-weighted MR images in three orthogonal planes formed in patients with various forms of
(top left and right, bottom right). (b) Sagittal T1-weighted MR image demonstrates outline of dementia, including AD, frontotemporal
hippocampus in green. (c) Hippocampus (yellow) and neighboring brain structures (blue, green, dementia, and vascular dementia, with
and red) are shown on three-dimensional volume rendering of the brain. (Images courtesy of
demonstration of unique uptake pat-
Guido Gerig, PhD, University of North Carolina, Chapel Hill.)
terns. The presence of a frontal or ante-
rior perfusion deficit has been associated
with frontotemporal dementia and not
or intracranial mass. Functional imaging Lewy bodies, and depression. Moreover, with AD. In one study (74) in which a
modalities, including SPECT and PET, because structural changes occur late in simple decision rule based on discrimi-
also offer value in the differential diagno- the course of the disease, functional im- nant analysis of SPECT data was applied,
sis of dementia, particularly in distin- aging modalities, including SPECT, PET, 20 patients with probable AD and 20
guishing AD from vascular dementia, and functional MR imaging techniques, with probable frontotemporal dementia
frontotemporal dementia, dementia with may have greater potential in identifying were evaluated. One hundred percent of

Volume 226 䡠 Number 2 Neuroimaging and Early Diagnosis of Alzheimer Disease 䡠 325
 patients with frontotemporal dementia
and 90% of patients with AD were cor-
rectly classified. In another study (75) with
16 patients with frontotemporal dementia,
71 patients with other forms of dementia,
Radiology

and 28 control subjects, an anterior-to-pos-

terior ratio was successfully used to classify
patients with frontotemporal dementia
from those with other forms of dementia
and from control subjects with a sensitivity
of 87.5% and a specificity of 78.6%. The
effect of different SPECT uptake patterns at
baseline on modification of the differential
diagnosis was evaluated in a study (76)
with 363 patients followed up for a median
of 3 years. Patients were classified into dis-
ease groups on the basis of clinical criteria.
A bilateral posterior perfusion abnormality
was associated with AD, whereas a bilateral
anterior abnormality was associated with
frontotemporal dementia. A patchy uptake
pattern significantly increased the odds of
a patient having vascular dementia.
It has been suggested that dementia
with Lewy bodies may demonstrate oc-
cipital hypoperfusion, differentiating it
from AD (77). Recently, Lobotesis et al
(78) studied 23 subjects with a clinical
diagnosis of dementia with Lewy bodies
that had been based on consensus crite-
ria. SPECT measures of occipital and me-
dial temporal perfusion enabled correct
classification of 69% of all subjects, with
65% sensitivity and 87% specificity for
dementia with Lewy bodies versus AD
and controls.
Thus, results of numerous studies to
date have demonstrated that SPECT im-
aging offers additional diagnostic value
in the work-up of patients with demen- Figure 12. FDG PET images demonstrate typical findings in (a– c) three patients with AD and
tia. The question remains as to the cost- (d) one patient with frontotemporal dementia. (a) Note bilateral parieto-occipital hypometabo-
effectiveness of this additional diagnostic lism (arrows) in a 77-year-old woman with AD. (b) Bilateral parietal hypometabolism (arrows) is
also noted in a 62-year-old woman with AD. (c) Changes may be asymmetric, as demonstrated in
value at this time. McMahon et al (79)
a 53-year-old woman with AD, with the metabolic defect (arrow) primarily on the left. (d) As
recently examined this question by uti- shown in a patient with frontotemporal dementia, PET may help differentiate frontotemporal
lizing a decision model designed to cal- dementia from AD by demonstrating metabolic deficits (arrows) in the frontal and anterior
culate quality-adjusted life-years that ac- temporal lobes (not shown).
crue to hypothetical cohorts of patients
at the time of presentation to an AD cen-
ter. This model depended on the sensitiv-
ity and specificity of the standard diag- ity at baseline examination and subse- complex in asymptomatic subjects with
nostic work-up, the effectiveness of quent cognitive decline in MMSE score. the PS-1 gene mutation, as compared
current treatments, and the severity of The investigators in that study used vi- with that in control subjects and in the
disease. The authors concluded that ad- sual inspection alone for evaluation of anterior and posterior cingulate gyrus,
dition of SPECT imaging to the diagnos- SPECT scans. Authors of another study posterior parietal lobe, and anterior fron-
tic regimen was not cost-effective given (81), using a quantitative singular-value tal lobe regions. The authors suggested
currently available therapies (79). decomposition analysis of SPECT images that a distributed brain network pertain-
The use of a temporoparietal hypoper- in patients with MCI, demonstrated re- ing to memory might be selectively af-
fusion pattern on SPECT images to help gional decreases in uptake in the medial fected in the earliest stages of the disease.
predict the clinical onset of AD in at-risk temporal lobe and cingulate regions that
populations has not been extensively were most prominent among patients
PET Imaging
studied. One study (80) in which 36 pa- with MCI who subsequently converted to
tients with MCI were followed up for 3 AD over the 1- to 2-year follow-up. The PET has been used to study the brain’s
years failed to show a correlation be- same group (82) also demonstrated re- metabolic uptake of fluorine 18 (18F)-la-
tween the presence of a SPECT abnormal- duced perfusion in the hippocampal beled fluorodeoxyglucose (FDG) and blood

326 䡠 Radiology 䡠 February 2003 Petrella et al

 mal temporoparietal uptake patterns at
PET, including in asymptomatic mem-
bers of families in which a familial form
of early-onset AD is found (86). Further-
more, asymptomatic subjects with the
Radiology

apolipoprotein ⑀4 allele, a genetic risk

factor for the more common late-onset
form of AD, have significantly lower tem-
poroparietal metabolic activity than those
without the allele (87).
The prognostic value of PET for pre-
dicting cognitive decline has also been
evaluated. Drzezga et al (89) performed
FDG PET in 15 patients with MCI at the
time of protocol entry and at 1-year fol-
low-up. In six of these patients, MCI pro-
gressed to a diagnosis of probable AD at 1
year; all six patients showed greater met-
abolic reduction in the posterior associa-
tion cortex on the initial PET scan, com-
pared with that on PET scans in those in
whom the condition did not convert.
These patients also showed a progressive
metabolic reduction in the frontal asso-
Figure 13. PET images comparing temporal lobe uptake of [18F]FDDNP (see text), an amyloid- ciation cortex over the 1-year follow-up.
binding radiotracer, and FDG, a marker of glucose metabolism, in a patient with AD (left) and a Perhaps the most compelling evidence
control subject (right). Note increased uptake and retention of [18F]FDDNP (arrowheads) in to date for the prognostic value of FDG
temporal lobes of the patient with AD, compared with those in control subject. The patient with
PET for prediction of cognitive decline is
AD still demonstrates typical findings of decreased temporal (arrows) and parietal (not shown)
FDG uptake. provided by a recent multinational con-
sortium study (90) conducted at eight ac-
ademic institutions, with autopsy data in
138 of 284 patients evaluated for symp-
flow in patients with dementia. PET has matter and cortical regions. Dementia toms of dementia. Mean age was 66 years
been performed at baseline, as well as dur- with Lewy bodies demonstrates a bilat- and mean MMSE score was 24, with a
ing the performance of various cognitive eral temporoparietal deficit similar to mean follow-up of 3.2 years (range, 2.0 –
tasks. Baseline PET studies have been suc- that seen in AD, but the deficit also in- 9.4 years). All scans were interpreted as
cessful in enabling differentiation of AD volves the occipital lobes and cerebel- positive or negative on the basis of visual
from other forms of dementia on the basis lum, areas typically spared in AD (85). assessment of the metabolic distribu-
of the pattern of FDG uptake. Specifically, Several investigators have demon- tions. In the group with longitudinal
deficits in temporoparietal metabolism are strated that FDG PET offers diagnostic clinical follow-up (n ⫽ 146), progressive
typically seen in patients with AD and not value even in mild or very mild stages of dementia was predicted with a high de-
in patients with other forms of dementia or AD, although sensitivity and specificity gree of sensitivity (93%) and moderate
in age-matched control subjects (Fig 12a, are greater in moderate to severe cases. In specificity (73%). Among the group with
12b). Asymmetry of the metabolic deficits a study (84) with 129 cognitively im- a pathologic diagnosis (n ⫽ 138), PET
is not uncommon (Fig 12c). There typically paired patients, the overall sensitivity for enabled correct identification of the pres-
is sparing of the basal ganglia, thalamus, detecting temporoparietal hypometabo- ence of AD with similar sensitivity (94%)
cerebellum, and primary sensorimotor cor- lism in patients with probable AD was and specificity (73%). Because accurate
tex (83). The magnitude of these deficits 94%. In patients with mild disease, the diagnosis is most clinically relevant early
correlates well with the degree of cognitive sensitivity was 87%, whereas in patients in the course of the disease, a subset of
impairment. with moderate to severe disease, the sen- patients with pathologic confirmation
The ability to differentiate AD from sitivity was 96%. In another study (86) who were documented to have question-
frontotemporal dementia, depression, with 23 patients with isolated mild mem- able or mild dementia at the time of the
vascular dementia, and dementia with ory impairment or MCI in whom a diag- initial PET scan were analyzed separately.
Lewy bodies was demonstrated in a nosis of probable AD could not be made The overall sensitivity (95%) and speci-
blinded evaluation of PET studies (84). at the time the initial PET scan was ob- ficity (71%) of PET in this group were
The frontotemporal dementias demon- tained, eight patients subsequently pro- similar.
strate decreased FDG uptake in the fron- gressed to AD over a 3.3-year follow-up. In a number of blood flow and glucose
tal, anterior temporal, and medial tem- All patients demonstrated changes simi- metabolism PET activation studies ob-
poral cortices (Fig 12d). Patients with lar to those on scans in patients with tained during the performance of mem-
depression demonstrate a normal scan probable AD; these changes, although ory tasks, decreased responses have been
easily distinguishable from the pattern less apparent, were still statistically sig- demonstrated in the brains of patients
on scans in patients with AD. Patients nificant. with AD, compared with responses in
with vascular dementia usually demon- Recent studies in asymptomatic at-risk corresponding regions in control subjects
strate patchy defects in central white populations have demonstrated abnor- (91–94). These data suggest there may be

Volume 226 䡠 Number 2 Neuroimaging and Early Diagnosis of Alzheimer Disease 䡠 327
Radiology

Figure 14. Transgenic mouse model of AD (tg2576). (a) Section from entorhinal cortex demonstrates in vivo labeling of amyloid plaques.
Fluorescent labeling of plaques is seen after intravenous injection of (trans,trans)-1-bromo-2,5-bis-(3-hydroxycarbonyl-4-hydroxy)styrylbenzene, or
BSB, an amyloid ␤ binding agent that crosses the blood-brain barrier. (Original magnification, ⫻400.) (b) The same section was also immunostained
with amyloid ␤–specific antiserum 2332, as demonstrated in this light microscopy image. (Original magnification, ⫻400.) (c) Digital overlay of a
and b reveals high specificity of BSB plaque labeling. (Original magnification, ⫻400.) (Adapted and reprinted, with permission, from reference 100.)

disease-related functional failure in these MR Perfusion Imaging those of cerebral SPECT, with sensitivities
regions. In a number of other studies, in the 87%–95% range for patients with
Dynamic susceptibility-contrast MR
additional areas of activation outside mild or moderate AD and specificities of
perfusion imaging, a method that uses
those seen in control subjects have been 88%–95% in control subjects (104,107).
rapid T2*-weighted imaging of the brain
demonstrated during performance of the To our knowledge, studies have yet to be
during intravenous injection of a bolus of
same task, indicating compensatory re- performed in patients with various forms
paramagnetic contrast material, is an-
cruitment of additional brain regions and of dementia, patients with MCI, and pre-
functional reallocation of brain resources other functional technique that has been symptomatic patients at genetic risk for
(95–97). used to study patients with dementia. future development of AD.
The majority of PET studies in the mem- This technique enables measurement of
ory impairment population have used FDG several hemodynamic parameters, in-
cluding relative cerebral blood volume, Functional MR Imaging
as a radiolabeled tracer. Most recently, a
new radiolabeled tracer, known as 2-(1- flow, and transit time (101,102). Unlike Functional MR imaging has proved to
{6-[(2-[18F]fluoroethyl)(methyl)amino]- PET perfusion imaging, MR perfusion im- be a powerful research technique to aid
2naphthyl}ethylidene)malononitrile, or aging takes less time to perform (about 1 in identifying regions of brain activated
[18F]FDDNP, has been developed. This minute), involves no radiation exposure, by particular stimuli and tasks. With this
tracer is reported (98) to target the amy- and is readily available on any MR unit technique, regional brain activity is mea-
loid ␤ senile plaques and neurofibrillary capable of echo-planar imaging. sured on the basis of local changes in
tangles in AD. An initial study in an 82- Because this technique is relatively deoxyhemoglobin concentration in re-
year-old woman with pathologically new, only a few studies have been per- sponse to various stimuli and tasks (109).
proved AD showed longer tracer reten- formed in patients with dementia. Nev- In brief, rapid T2*-sensitive imaging, usu-
tion times in the hippocampus, which ertheless, it has been shown that MR per- ally gradient-echo echo-planar imaging,
corresponded to pathologically con- fusion measurements of cerebral blood is performed during presentation of a
firmed areas of plaque and tangle depo- volume closely parallel changes in cere- stimulus or performance of a specific task
sition seen at postmortem examination bral metabolic rate of glucose consump- and during rest periods. A voxel-by-voxel
(98). In a subsequent study with the same tion as determined with PET in patients statistical comparison is then performed
tracer (99), nine patients with AD and with dementia and may be a lower cost with images obtained during the stimu-
seven control subjects demonstrated alternative to PET (103). Harris et al lus/task periods versus those obtained
greater accumulation and slower clear- (104,105) originally reported a 17% aver- during the rest periods, creating a statis-
ance of the probe in brain areas worst age reduction in temporoparietal cerebral tical-activation map that can be “thresh-
affected by plaque and tangle deposition. blood volume, with 88.5% accuracy for olded” and presented as a color overlay
Retention time of the tracer in brain re- correct categorization of 13 patients with on anatomic T1-weighted images. Be-
gions known to be affected by AD is cor- AD and 13 control subjects. Maas et al cause functional MR imaging does not
related with lower memory performance (106) examined 16 patients with proba- use ionizing radiation, it can be applied
scores and is significantly greater in pa- ble AD and 16 control subjects and found repeatedly in the same patients without
tients with AD than in control subjects the technique to be 81% sensitive and risk. Although not currently used in the
(Fig 13). Additional ligands are in the 88% specific. More recently, the sensitiv- routine work-up of patients with mem-
preclinical development phase (100) (Fig ity and specificity of MR perfusion imag- ory disorders, the technique offers con-
14). ing have been shown to be superior to siderable potential in early identification

328 䡠 Radiology 䡠 February 2003 Petrella et al

 of patients with prodromal dementia. We

hypotheses; only one memory load

will first discuss functional MR imaging

Restricted to hippocampal formation

condition; no direct monitoring of
Supports compensatory-recruitment

Supports compensatory-recruitment

Supports compensatory-recruitment

Supports compensatory-recruitment
studies of normal memory. This will be

recruitment hypothesis; did not

Does not support compensatory-

Does not support compensatory-

followed by discussion of studies in pa-

compensatory-recruitment
tients with memory impairment.

hypothesis with regard to

and subregions; refutes

recruitment hypothesis
Functional MR imaging activation pat-
Radiology

Comments

monitor performance
subject responses terns during various memory tasks have
been studied in healthy volunteers. Two

hippocampus
main groups of structures involved in

hypothesis

hypothesis

hypothesis
working memory and secondary memory
include the prefrontal cortex and the me-
dial temporal lobe. Hence, these regions
have been the focus of prior functional
MR imaging studies of normal memory.
In prior functional MR imaging studies
In comparison with controls, AD group and
Greater extent and intensity of activation in

gyri in AD group, compared with control

right to left) in intraparietal sulcus region
formation; other 2 of MCI group showed

Equal activation in prefrontal cortex during

cortex and inferior and middle temporal
Reversal of hemispheric dominance (from
(109 –114), the prefrontal cortex has con-
apolipoprotein ⑀4-positive group in left

left hemisphere correlated with greater

activation during retrieval in AD group

greater number of regions activated in

middle and posterior inferior temporal

In AD group, greater atrophy associated

encoding but weak and diffuse or no

Decreased activation in high-risk group

Additional activation in right prefrontal
with greater activation in left inferior
frontal, temporal, and parietal lobes;

sistently been found to be involved in

activation throughout hippocampal
of MCI group showed diminished

relative to that in low-risk group in

learning and recall. In these studies, spe-
cific regions or hemispheres have been
decrease only in subiculum

localized on the basis of the cognitive

despite equal performance

Findings

paradigm used. Activation in the prefron-

cognitive decline at 2 y

tal cortex is greater with increasing mem-

3

ory loads (109) and is more pronounced

during initial, as opposed to repeated,
frontal gyrus

in AD group

attempts at learning (113). Increased

activation during learning is associated

regions
group

with more successful subsequent recall

(113,114).
1
3

Authors of functional MR imaging

studies have also documented the role of
Auditory, learning and recalling

Visual, letter fluency and object

Visual, faces: sex discrimination

Auditory, word pairs (category

the medial temporal lobe in learning and

Visual, novel picture encoding

recall. The medial temporal lobe struc-

example) with semantic

semantic or phonologic
Auditory, word pair with

tures most often implicated in these stud-

unrelated word pairs

ies are the hippocampus and the parahip-

and recognition
Task

pocampal cortex (115). Another area in

the medial temporal lobe, the fusiform
Visual saccade

gyrus (lateral temporo-occipital cortex),

decision

decision

naming

is involved in the processing of faces

(116).
There have been a number of recent
functional MR imaging studies related to
16 Apolipoprotein ⑀4-positive and

early AD, the results of which are sum-

with MCI, 4 patients with mild
14 apolipoprotein ⑀4-negative

apolipoprotein ⑀4 carrier) and

10 Control subjects, 18 patients

marized in Table 2. In the majority of

4 Control subjects, 12 patients

16 Control subjects, 8 patients

Control subjects, patients with

9 Control subjects, 9 patients

these studies, diminished intensity and/or

14 high-risk (family history,

extent of activation has been demon-

12 low-risk subjects

strated in the frontal and temporal re-

Sample

gions in patients with AD, as compared

with mild AD

with mild AD

with those in control subjects, when per-

formance has been controlled for. Studies
Functional MR Imaging Studies of AD

mild AD
with AD
subjects

in which “at-risk” populations were ex-

amined have shown mixed results with
AD

increased (117) or decreased (123,124)

activation in the at-risk group, as com-
pared with activation in a control group.
Bookheimer et al (117), 2000

Authors of the studies in which increased

Thulborn et al (120), 2000
Johnson et al (119), 2000

Corkin et al (122), 1997

Saykin et al (121), 1999

activation was seen in the AD or at-risk

Smith et al (123), 1999
Small et al (118), 2000
Study and Year

groups explain their findings in terms of

a compensatory-recruitment hypothesis,
in which greater cognitive effort is re-
quired to perform the same task as com-
TABLE 2

pared with that required by a control

group (Fig 15). This hypothesis presumes
there is enough healthy neural tissue
present to accommodate the task. In the

Volume 226 䡠 Number 2 Neuroimaging and Early Diagnosis of Alzheimer Disease 䡠 329
 presence of substantial neuronal loss, as
may be seen when clinical manifesta-
tions of memory loss are present, de-
creases in activation may result (117).
Such decreases may be seen only in tasks
that make sufficiently high demands to
Radiology

exceed the subject’s compensatory re-

serve.
In summary, results of previous func-
tional MR imaging studies of cognitively
impaired or at-risk patients have shown
increases or decreases in the intensity
and extent of activation, as compared
with findings in control groups. The dis-
crepancies between increased or de-
creased activation may have to do with
either differing task demands or different
levels of compensatory reserve. Studies in
which results supported or refuted the
compensatory hypothesis have not used
tasks with varied levels of difficulty in the
same subjects; therefore, it is difficult to
assess whether task difficulty met or was
exceeded by cognitive reserve. Depend-
ing on the level of task difficulty and the
patient population, varying degrees of
activation have been seen. Moreover,
many of these studies did not distinguish
between compensatory reserve—that is,
increased activation in regions known to
activate in control subjects—and com-
pensatory recruitment—namely, addi-
Figure 15. Brain activation maps obtained during memory-activa-
tional regions of activation outside those tion task in patients at genetic risk for AD (carriers of apolipoprotein
regions known to activate in control sub- [Apo] ⑀4 allele) compared with control subjects (apolipoprotein ⑀3
jects (117). carriers). Three-dimensional renditions of the brain surface are shown
in gray, and colored areas indicate regions of significantly increased
MR signal intensity during performance of memory task as compared
MR Spectroscopy with that during resting periods. Activation is seen in temporal and
frontal regions in both groups; however, both the extent and inten-
Proton MR spectroscopy is an applica- sity of activation are greater among the genetic risk group, which
tion of MR that allows noninvasive as- suggests compensatory brain function. (Adapted and reprinted, with
sessment of a number of local metabolite permission, from reference 117.)
levels in brain tissue (125). Proton MR
spectroscopy allows in vivo assessment of
N-acetylaspartate (NAA), glutamine and
glutamate, ␥-aminobutyric acid, myo-ino- techniques, known as chemical shift im- most studies (134,138 –142), although
sitol, glycine, mobile choline moieties, cre- aging, are also available (127). myo-inositol levels may be confounded
atine and phosphocreatine, lipids, and lac- In several in vitro studies (128 –131) on by a number of other coexisting disease
tate. NAA is present primarily in neurons postmortem brains, decreases in NAA lev- processes (143–146) (Fig 16).
within the central nervous system but not els have been demonstrated in patients Metabolite levels have correlated with
in glial cells or other nonneuronal tissue. with AD, as compared with levels in con- cognitive scores and dementia severity
Although the exact metabolism of NAA re- trols. Moreover, there has been a positive (138), and several groups have suggested
mains unclear, NAA is generally thought to correlation between the magnitude of that metabolite levels may be used as a
represent a marker of neuronal function NAA decreases and the severity of neuro- diagnostic tool to help differentiate pa-
(125,126). The NAA level is decreased in pathologic findings (eg, counts of amy- tients with AD from those with other
cases of neuronal loss or damage yet may loid plaques and neurofibrillary tangles) forms of dementia and age-matched con-
return to normal levels during recovery. (128,131). Similarly, in vivo studies (132– trol subjects. Shonk et al (139) deter-
Elevated myo-inositol levels may mark gli- 135) have also demonstrated decreases in mined that changes in the myo-inositol–
osis, membrane dysfunction, and/or cy- NAA in patients with AD in both the to-NAA ratio would help distinguish
toskeletal abnormalities. Elevated choline temporal and parietal lobes. In some patients with AD from control subjects
levels may reflect cellular proliferation, as studies (136,137), higher choline levels with a sensitivity of 83% and a specificity
in neoplasia, or myelin breakdown. Most have been demonstrated in the brains of of 98% and that changes in the myo-ino-
studies in the dementia population have patients with AD; however, these results sitol–to– creatine and phosphocreatine
used a single-voxel approach (125); how- have been inconsistent. myo-Inositol lev- ratio would help distinguish patients
ever, multivoxel spectroscopic imaging els have been shown to be elevated in with AD from elderly patients with other

330 䡠 Radiology 䡠 February 2003 Petrella et al

 in patients with MCI to be between those
of cognitively normal subjects and pa-
tients with AD. Patients who are positive
for the apolipoprotein ⑀4 allele tend to
demonstrate similar findings prior to the
onset of dementia or notable anatomic
Radiology

changes (unpublished data, 2002). Lon-

gitudinal evaluation may play an impor-
tant role in determining prognosis and
evaluating treatment in such at-risk
groups of patients. Also, MR spectro-
scopic measures have been shown to be
predictors of cognitive scores at 1-year
follow-up (147). In addition, at least two
controlled studies have used MR spec-
troscopy to monitor treatment effects in
AD. Decreases in the choline-to– creatine
and phosphocreatine ratio have been
demonstrated in response to xanome-
line, a muscarinic agonist; and decreases
in the myo-inositol–to-NAA ratio have
been seen in response to donepezil, a
cholinesterase inhibitor (148,149). Do-
raiswamy et al (125) recently published a
comprehensive review of the role of MR
spectroscopy in dementia drug develop-
ment and have also reviewed all the cen-
tral nervous system drug trials in which
MR spectroscopic results were used as an
outcome measure. They concluded that
MR spectroscopic measures of NAA,
when combined with hippocampal volu-
metry, could provide highly useful surro-
gate markers of AD progression in trials
of neuroprotective agents.

Diffusion-weighted MR Imaging
Diffusion-weighted MR imaging is a
technique that is sensitive to the micro-
scopic motion of water molecules in tis-
sue. Its primary applications have been in
the evaluation of acute cerebral ischemia.
The technique has also been applied to
the study of patients with AD (150 –152).
Most recently, Kantarci et al (153) stud-
ied various brain regions including the
medial temporal lobes in a group of 19
patients with MCI, 21 with AD, and 55
Figure 16. Multivoxel MR spectroscopic imaging in (a) a 62-year-old healthy volunteer and age-matched control subjects. They used
(b) an 80-year-old patient with AD. Sample spectra are shown from right temporal lobe (bottom an index of mean diffusibility: the appar-
left), left insula (top right) and left thalamus (bottom right). Note the increase in myo-inositol ent diffusion coefficient. Confirming and
(Ino) and choline (Cho) levels and the decrease in NAA level (short arrows in b) in the right expanding on the results of previous
temporal lobe of the patient with AD, compared with those levels in the healthy volunteer,
work, Kantarci et al demonstrated statis-
suggesting the presence of gliosis, increased membrane turnover, and neuronal loss in AD.
(P)Cre ⫽ creatine and phosphocreatine. (Images courtesy of H. Cecil Charles, PhD, Duke Univer- tically significant differences in mean dif-
sity Medical Center, Durham, NC.) fusibility between the AD group and the
control group in a number of brain re-
gions, most notably the hippocampus,
and in temporal, cingulate, and parietal
forms of dementia, with a sensitivity of and only an adjunct to clinical evalua- white matter. Although statistically sig-
82% and a specificity of 64%. However, tion at this time. nificant differences were found only for
lack of specificity in distinguishing AD With regard to patients at risk for AD, the hippocampus, values in the MCI
from other forms of dementia causes MR the results of one study (140) have shown group were between those in the AD
spectroscopy to remain a research tool NAA and myo-inositol metabolite levels group and the control group for all re-

Volume 226 䡠 Number 2 Neuroimaging and Early Diagnosis of Alzheimer Disease 䡠 331
 gions studied. The authors postulated determine the utility of other neuroim- Acknowledgments: The authors thank Haris
that pathologic destruction of cell mem- aging modalities in the work-up of the Sair, BS, and Sarah Hart, BS, for their assistance
in preparing tables and figures, and Christine
branes, with subsequent loss of myelin patient with dementia (157). Hulette, MD, for consultation on neuropatho-
and axonal processes due to wallerian de- PET is currently used primarily as an logic criteria.
generation, lessens the restriction of adjunct to clinical diagnosis, especially in
Radiology

water diffusion in the hippocampus and differentiating AD from vascular demen- References
afferent white matter tracts. Such patho- tia and other focal dementias such as 1. Khachaturian ZS. Diagnosis of Alzhei-
logic changes are manifested by an in- frontal lobe dementia. Given the grow- mer’s disease. Arch Neurol 1985; 42:1097–
creased apparent diffusion coefficient in ing evidence, PET will likely come to the 1105.
these regions. The value of diffusion- 2. Carr DB, Goate A, Phil D, Morris JC.
forefront both as a diagnostic tool and as
Current concepts in the pathogenesis of
weighted MR imaging as a diagnostic tool a prognostic tool. This will be especially Alzheimer’s disease. Am J Med 1997;
is still in question, however. Kantarci et true if amyloid imaging becomes stan- 103(suppl 3A):3S–10S.
al found that at a fixed specificity of 80%, dardized and widely available. Reim- 3. Small GW, Rabins PV, Barry PP, et al.
the sensitivity was only 57% for distin- Diagnosis and treatment of Alzheimer
bursement issues remain a problem with
disease and related disorders: consensus
guishing patients with AD from control PET, and it is hoped scientific advances statement of the American Association
subjects by using the hippocampal appar- will be recognized by third-party payers. for Geriatric Psychiatry, the Alzheimer’s
ent diffusion coefficient. For determination of preclinical or early- Association, and the American Geriatrics
Diffusion tensor MR imaging is a Society. JAMA 1997; 278:1363–1371.
stage disease, imaging will play an in- 4. Jellinger K. Morphology of Alzheimer
more advanced application of diffusion- creasingly greater role in the future, espe- disease and related disorders. In: Maurer
weighted imaging; diffusion tensor imag- cially as new agents to delay the onset of K, Riederer P, Beckmann H, eds. Alzhei-
ing enables measurement of the direc- dementia become available. As clinical mer disease: epidemiology, neuropa-
tionality or asymmetry of microscopic thology and clinics. New York, NY:
trial results confirm the efficacy of new Springer-Verlag, 1990; 61–77.
water movement in tissue (154). Such therapeutic agents and as combination 5. Selkoe DJ. Alzheimer’s disease: geno-
asymmetric diffusion, known as anisot- therapies are used, earlier and more accu- types, phenotypes, and treatments. Sci-
ropy, is seen in normal white matter, ow- rate diagnoses will become more crucial. ence 1997; 275:630 –631.
ing to the integrity of white matter tracts 6. Selkoe DJ. The pathophysiology of Alz-
The at-risk patient with MCI may also be heimer’s disease. In: Scinto LFM, Daffner
that preferentially allows diffusion of wa-
considered for treatment options that KR, eds. The early diagnosis of Alzhei-
ter parallel, rather than perpendicular, to
might alter the rate of progression to de- mer’s disease. Totowa, NJ: Humana,
the tracts. Primary applications of this 2000; 83–104.
mentia. No specific neuroimaging proto-
technique are in the evaluation of cere- 7. Greenberg SM, Rebeck GW, Vonsattel JP,
col is recommended in this group at Gomez-Isla T, Hyman BT. Apolipopro-
bral white matter tracts (155). Recently,
present (159), although more sensitive tein E epsilon 4 and cerebral hemor-
the technique has been applied in hu-
functional imaging techniques, as well as rhage associated with amyloid angiopa-
man studies of AD (156). So far, results thy. Ann Neurol 1995; 38:254 –259.
subtraction MR imaging, seem to be log-
demonstrate a statistically significant re- 8. Bierer LM, Hof PR, Purohit DP, et al.
ical candidates. FDG PET has demon- Neocortical neurofibrillary tangles corre-
duction in white matter integrity through-
strated the highest prognostic utility for late with dementia severity in Alzhei-
out the brain, with relative sparing of the
providing a diagnosis of AD 2–3 years mer’s disease. Arch Neurol 1995; 52:81–
motor tracts, reflecting the known patho- 88.
before the full dementia-related symp-
logic and clinical findings in AD. 9. Arriagada PV, Growdon JH, Hedley-
toms manifest (160). The role of SPECT in Whyte ET, Hyman BT. Neurofibrillary
this regard remains less promising. Fur- tangles but not senile plaques parallel
ther longitudinal studies must be devel- duration and severity of Alzheimer’s dis-
PREDICTION AND EARLY ease. Neurology 1992; 42:631–639.
DIAGNOSIS: THE FUTURE oped to examine the predictive power of
10. Geula C. Pathological diagnosis of Alz-
these functional techniques in order to heimer’s disease. In: Scinto LFM, Daffner
For differential diagnosis in the patient determine which patients will ultimately KR, eds. The early diagnosis of Alzhei-
develop AD. Most likely, the answer will mer’s disease. Totowa, NJ: Humana,
with dementia, during the middle or late 2000; 65–82.
stages of the disease the current clinical be a battery of tests that combine func-
11. Braak H, Braak E, Neuropathological
criteria have a high sensitivity and spec- tional and structural imaging with clini- stages of Alzheimer’s disease. In: de Leon
ificity for determination of AD. Thus, the cal, genetic, and other laboratory data MJ, ed. An atlas of Alzheimer’s disease.
(161). New York, NY: Parthenon, 1999; 57–74.
future role of imaging in patients with
The major goals in treating AD cur- 12. Mirra SS, Heyman A, McKeel DW, et al.
dementia at these stages will likely re- Standardization of the neuropathologic
main to help exclude other causes of de- rently are to recognize the disease early in assessment of Alzheimer’s disease. Neu-
mentia. The American Academy of Neu- order to initiate appropriate therapy and rology 1991; 41:479 –486.
rology Quality Standards Subcommittee delay functional and cognitive losses. In 13. Hyman BT, Trojanowski JQ. Consensus
recommendations for the postmortem
recently published evidence-based practice addition, as powerful antiamyloid thera-
diagnosis of Alzheimer disease from the
parameters for the diagnosis, manage- pies are developed, there will be a need to National Institute on Aging and the Re-
ment, and early detection of dementia monitor brain changes and treatment ef- agan Institute Working Group on diag-
(157–159). Current imaging recommen- ficacy at the earliest stages of the disease, nostic criteria for the neuropathological
assessment of Alzheimer disease. J Neu-
dations for the initial evaluation of pa- perhaps even in prodromal patients. It is
ropathol Exp Neurol 1997; 56:1095–
tients with dementia include nonen- hoped that the widespread availability of 1097.
hanced CT or MR imaging; however, newer MR and PET markers will supple- 14. Consensus recommendations for the
because of insufficient data on validity, ment the strengths of the currently avail- postmortem diagnosis of Alzheimer’s
disease: the National Institute on Aging,
no other imaging procedure is recom- able structural and functional imaging
and Reagan Institute Working Group on
mended in these guidelines. The subcom- techniques in helping to achieve these Diagnostic Criteria for the Neuropatho-
mittee recommended further research to goals. logical Assessment of Alzheimer’s Dis-

332 䡠 Radiology 䡠 February 2003 Petrella et al

 ease. Neurobiol Aging 1997; 18(suppl 4): WR. Linguistic ability in early life and cephalus with and without shunting.
S1–S2. cognitive function and Alzheimer’s dis- J Neurol Neurosurg Psychiatry 1978; 41:
15. Braak H, Braak E. Neuropathological ease in late life: findings from the Nun 961–971.
staging of Alzheimer-related changes. Study. JAMA 1996; 275:528 –532. 46. Bradley WG. Normal pressure hydro-
Acta Neuropathol 1991; 82:239 –259. 30. Butters N, Dellis DC, Lucas JA. Clinical cephalus: new concepts on etiology and
16. Nagy Z, Esiri MM, Joachim C, et al. assessment of memory disorders in am- diagnosis. AJNR Am J Neuroradiol 2000;
Radiology

Comparison of pathological diagnostic nesia and dementia. Annu Rev Psychol 21:1586 –1590.
criteria for Alzheimer disease. Alzheimer 1995; 46:493–523. 47. Bradley WG Jr, Kortman KE, Burgoyne
Dis Assoc Disord 1998; 12:182–189. 31. Petersen RC, Smith GE, Waring SC, B. Flowing cerebrospinal fluid in normal
17. Newell KL, Hyman BT, Growdon JH, Ivnik RJ, Tangalos EG, Kokmen E. Mild and hydrocephalic states: appearance on
Hedley-Whyte ET. Application of the cognitive impairment: clinical charac- MR images. Radiology 1986; 159:611–
National Institute on Aging (NIA)-Rea- terization and outcome. Arch Neurol 616.
gan Institute criteria for the neuropatho- 1999; 56:303–308. 48. Bradley WG Jr, Scalzo D, Queralt J, Nitz
logical diagnosis of Alzheimer disease. 32. Peterson RC. Disorders of memory. In: WN, Atkinson DJ, Wong P. Normal-
J Neuropathol Exp Neurol 1999; 58:1147– Samuels MA, Feske S, eds. Office practice pressure hydrocephalus: evaluation with
1155. of neurology. New York, NY: Churchill cerebrospinal fluid flow measurements
18. Cummings JL, Vinters HV, Cole GM, Livingstone, 1996; 728 –736. at MR imaging. Radiology 1996; 198:523–
Khachaturian ZS. Alzheimer’s disease: 33. Morris JC, Storandt M, McKeel DW Jr, et 529.
etiologies, pathophysiology, cognitive al. Cerebral amyloid deposition and dif- 49. Egeler-Peerdeman SM, Barkhof F, Wal-
reserve, and treatment opportunities. fuse plaques in “normal” aging: evi- chenbach R, Valk J. Cine phase-contrast
Neurology 1998; 51(suppl 1):S2–S17, dence for presymptomatic and very MR imaging in normal pressure hydro-
discussion S65–S67. mild Alzheimer’s disease. Neurology cephalus patients: relation to surgical
19. Larson EB, Edwards JK, O’Meara E, 1996; 46:707–719. outcome. Acta Neurochir Suppl 1998;
Nochlin D, Sumi SM. Neuropathologic 34. Petersen RC. Mild cognitive impair- 71: 340 –342.
diagnostic outcomes from a cohort of ment: transition between aging and Alz- 50. DeCarli C, Murphy DG, Tranh M, et al.
outpatients with suspected dementia. J heimer’s disease. Neurologia 2000; 15: The effect of white matter hyperinten-
Gerontol A Biol Sci Med Sci 1996; 93–101. sity volume on brain structure, cogni-
51(suppl 6):M313–M318. 35. Berg L, Hughes CP, Coben LA, Danziger tive performance, and cerebral metabo-
20. Rasmusson DX, Brandt J, Steele C, He- WL, Martin RL, Knesevich J. Mild senile lism of glucose in 51 healthy adults.
dreen JC, Troncoso JC, Folstein MF. Ac- dementia of Alzheimer type: research di- Neurology 1995; 45:2077–2084.
curacy of clinical diagnosis of Alzheimer agnostic criteria, recruitment, and de- 51. Mirsen TR, Lee DH, Wong CJ, et al. Clin-
disease and clinical features of patients scription of a study population. J Neurol ical correlates of white-matter changes
with non-Alzheimer disease neuropa- Neurosurg Psychiatry 1982; 45:962–968. on magnetic resonance imaging scans of
thology. Alzheimer Dis Assoc Disord 36. Corder EH, Saunders AM, Strittmatter the brain. Arch Neurol 1991; 48:1015–
1996; 10:180 –188. WJ, et al. Gene dose of apolipoprotein E 1021.
21. McKhann G, Drachman D, Folstein M, type 4 allele and the risk of Alzheimer’s 52. Wahlund LO, Basun H, Almkvist O,
Katzman R, Price D, Stadlan EM. Clinical disease in late onset families. Science Andersson-Lundman G, Julin P, Saaf J.
diagnosis of Alzheimer’s disease: report 1993; 261:921–923. White matter hyperintensities in de-
of the NINCDS-ADRDA Work Group un- 37. Mayeux R, Saunders AM, Shea S, et al. mentia: does it matter? Magn Reson Im-
der the auspices of Department of Utility of the apolipoprotein E genotype aging 1994; 12:387–394.
Health and Human Services Task Force in the diagnosis of Alzheimer’s disease: 53. Breteler MM, van Swieten JC, Bots ML,
on Alzheimer’s Disease. Neurology Alzheimer’s Disease Centers Consortium et al. Cerebral white matter lesions, vas-
1984; 34:939 –944. on Apolipoprotein E and Alzheimer’s cular risk factors, and cognitive function
22. Diagnostic and statistical manual of Disease. N Engl J Med 1998; 338:506 – in a population-based study: the Rotter-
mental disorders. 4th ed. Washington, 511. dam Study. Neurology 1994; 44:1246 –
DC: American Psychiatric Association, 38. Huang YP, Tuason MY, Wu T, Plaitakis 1252.
1994. A. MRI and CT features of cerebellar de- 54. de Leon MJ, George AE, Reisberg B, et al.
23. Winblad B, Wimo A. Assessing the soci- generation. J Formos Med Assoc 1993; Alzheimer’s disease: longitudinal CT
etal impact of acetylcholinesterase in- 92:494 –508. studies of ventricular change. AJR Am J
hibitor therapies. Alzheimer Dis Assoc 39. Duara R, Barker W, Luis CA. Frontotem- Roentgenol 1989; 152:1257–1262.
Disord 1999; 13(suppl 2):S9 –S19. poral dementia and Alzheimer’s disease: 55. DeCarli C, Kaye JA, Horwitz B, Rapoport
24. Roman GC, Tatemichi TK, Erkinjuntti T, differential diagnosis. Dement Geriatr SI. Critical analysis of the use of com-
et al. Vascular dementia: diagnostic cri- Cogn Disord 1999; 10:37–42. puter-assisted transverse axial tomogra-
teria for research studies: report of the 40. Charness ME. Brain lesions in alcohol- phy to study human brain in aging and
NINDS-AIREN International Workshop. ics. Alcohol Clin Exp Res 1993; 17:2–11. dementia of the Alzheimer type. Neurol-
Neurology 1993; 43:250 –260. 41. Drayer BP. Imaging of the aging brain. ogy 1990; 40:872–883.
25. Adams RD, Victor M, Ropper A. Adams II. Pathologic conditions. Radiology 56. Rusinek H, de Leon MJ, George AE, et al.
& Victor’s principles of neurology. 6th 1988; 166:797–806. Alzheimer disease: measuring loss of ce-
ed. New York, NY: McGraw-Hill, 1996; 42. Seab JP, Jagust WJ, Wong ST, Roos MS, rebral gray matter with MR imaging. Ra-
121. Reed BR, Budinger TF. Quantitative diology 1991; 178:109 –114.
26. Caruso R, Cervoni L, Vitale AM, Salvati NMR measurements of hippocampal at- 57. Doraiswamy PM, McDonald WM,
M. Idiopathic normal-pressure hydro- rophy in Alzheimer’s disease. Magn Re- Patterson L, et al. Interuncal distance as
cephalus in adults: result of shunting son Med 1988; 8:200 –208. a measure of hippocampal atrophy: nor-
correlated with clinical findings in 18 43. Kesslak JP, Nalcioglu O, Cotman CW. mative data on axial MR imaging. AJNR
patients and review of the literature. Quantification of magnetic resonance Am J Neuroradiol 1993; 14:141–143.
Neurosurg Rev 1997; 20:104 –107. scans for hippocampal and parahip- 58. Early B, Escalona PR, Boyko OB, et al.
27. Black PM, Ojemann RG, Tzouras A. CSF pocampal atrophy in Alzheimer’s dis- Interuncal distance measurements in
shunts for dementia, incontinence, and ease. Neurology 1991; 41:51–54. healthy volunteers and in patients with
gait disturbance. Clin Neurosurg 1985; 44. O’Brien JT, Desmond P, Ames D, Alzheimer disease. AJNR Am J Neurora-
32:632–651. Schweitzer I, Chiu E, Tress B. Temporal diol 1993; 14:907–910.
28. McKeith IG, Galasko D, Kosaka K, et al. lobe magnetic resonance imaging can 59. de Leon MJ, George AE, Stylopoulos LA,
Consensus guidelines for the clinical differentiate Alzheimer’s disease from Smith G, Miller DC. Early marker for
and pathologic diagnosis of dementia normal aging, depression, vascular de- Alzheimer’s disease: the atrophic hip-
with Lewy bodies (DLB): report of the mentia and other causes of cognitive im- pocampus. Lancet 1989; 2:672–673.
consortium on DLB international work- pairment. Psychol Med 1997; 27:1267– 60. Jack CR. Structural imaging approaches
shop. Neurology 1996; 47:1113–1124. 1275. to Alzheimer’s disease. In: Daffner S, ed.
29. Snowdon DA, Kemper SJ, Mortimer JA, 45. Hughes CP, Siegel BA, Coxe WS, et al. Early diagnosis of Alzheimer’s disease.
Greiner LH, Wekstein DR, Markesbery Adult idiopathic communicating hydro- Totowa, NJ: Humana, 2000.

Volume 226 䡠 Number 2 Neuroimaging and Early Diagnosis of Alzheimer Disease 䡠 333
 61. Jack CR Jr, Bentley MD, Twomey CK, ease and subcortical white matter de- 91. Pietrini P, Alexander GE, Furey ML, et al.
Zinsmeister AR. MR imaging–based vol- mentia by an anterior-to-posterior rCBF- Cerebral metabolic response to passive
ume measurements of the hippocampal SPET ratio. Dement Geriatr Cogn Disord audiovisual stimulation in patients with
formation and anterior temporal lobe: 2000; 11:275–285. Alzheimer’s disease and healthy volun-
validation studies. Radiology 1990; 176: 76. Talbot PR, Lloyd JJ, Snowden JS, Neary teers assessed by PET. J Nucl Med 2000;
205–209. D, Testa HJ. A clinical role for 99mTc- 41:575–583.
62. Jack CR Jr, Petersen RC, Xu YC, et al. HMPAO SPECT in the investigation of 92. Rapoport SI. Functional brain imaging
Radiology

Medial temporal atrophy on MRI in nor- dementia? J Neurol Neurosurg Psychia- in the resting state and during activation
mal aging and very mild Alzheimer’s dis- try 1998; 64:306 –313. in Alzheimer’s disease: implications for
ease. Neurology 1997; 49:786 –794. 77. Ishii K, Yamaji S, Kitagaki H, Imamura T, disease mechanisms involving oxidative
63. Jack CR Jr, Petersen RC, Xu YC, et al. Hirono N, Mori E. Regional cerebral phosphorylation. Ann N Y Acad Sci
Rate of medial temporal lobe atrophy in blood flow difference between dementia 1999; 893:138 –153.
typical aging and Alzheimer’s disease. with Lewy bodies and AD. Neurology 93. Riddle W, O’Carroll RE, Dougall N, et al.
Neurology 1998; 51:993–999. 1999; 53:413–416. A single photon emission computerised
64. Jack CR Jr, Petersen RC, Xu Y, et al. Rates 78. Lobotesis K, Fenwick JD, Phipps A, et al. tomography study of regional brain
of hippocampal atrophy correlate with Occipital hypoperfusion on SPECT in function underlying verbal memory in
change in clinical status in aging and dementia with Lewy bodies but not AD. patients with Alzheimer-type dementia.
AD. Neurology 2000; 55:484 –489. Neurology 2001; 56:643–649. Br J Psychiatry 1993; 163:166 –172.
65. Killiany RJ, Gomez-Isla T, Moss M, et al. 79. McMahon PM, Araki SS, Neumann PJ, 94. Kessler J, Herholz K, Grond M, Heiss
Use of structural magnetic resonance Harris GJ, Gazelle GS. Cost-effectiveness WD. Impaired metabolic activation in
imaging to predict who will get Alzhei- of functional imaging tests in the diag- Alzheimer’s disease: a PET study during
mer’s disease. Ann Neurol 2000; 47:430 – nosis of Alzheimer disease. Radiology continuous visual recognition. Neuro-
439. 2000; 217:58 –68. psychologia 1991; 29:229 –243.
66. Jack CR Jr, Petersen RC, Xu YC, et al. 80. McKelvey R, Bergman H, Stern J, Rush C, 95. Backman L, Andersson JL, Nyberg L,
Prediction of AD with MRI-based hip- Zahirney G, Chertkow H. Lack of prog- Winblad B, Nordberg A, Almkvist O.
pocampal volume in mild cognitive im- nostic significance of SPECT abnormali- Brain regions associated with episodic
pairment. Neurology 1999; 52:1397– ties in non-demented elderly subjects retrieval in normal aging and Alzhei-
1403. with memory loss. Can J Neurol Sci mer’s disease. Neurology 1999; 52:1861–
67. Fox NC, Crum WR, Scahill RI, Stevens 1999; 26:23–28. 1870.
JM, Janssen JC, Rossor MN. Imaging of 81. Johnson KA, Jones K, Holman BL, et al. 96. Woodard JL, Grafton ST, Votaw JR,
onset and progression of Alzheimer’s Preclinical prediction of Alzheimer’s dis- Green RC, Dobraski ME, Hoffman JM.
disease with voxel-compression map- ease using SPECT. Neurology 1998; 50: Compensatory recruitment of neural re-
ping of serial magnetic resonance im- 1563–1571. sources during overt rehearsal of word
ages. Lancet 2001; 358:201–205. 82. Johnson KA, Lopera F, Jones K, et al. lists in Alzheimer’s disease. Neuropsy-
68. Fox NC, Cousens S, Scahill R, Harvey RJ, Presenilin-1-associated abnormalities in chology 1998; 12:491–504.
Rossor MN. Using serial registered brain regional cerebral perfusion. Neurology 97. Becker JT, Mintun MA, Aleva K, Wise-
magnetic resonance imaging to measure 2001; 56:1545–1551. man MB, Nichols T, DeKosky ST. Com-
disease progression in Alzheimer dis- 83. Devanand DP, Jacobs DM, Tang MX, et pensatory reallocation of brain resources
ease: power calculations and estimates al. The course of psychopathologic fea- supporting verbal episodic memory in
of sample size to detect treatment ef- tures in mild to moderate Alzheimer dis- Alzheimer’s disease. Neurology 1996;
fects. Arch Neurol 2000; 57:339 –344. ease. Arch Gen Psychiatry 1997; 54:257– 46: 692–700.
69. Kelemen A, Szekely G, Gerig G. Elastic 263. 98. Agdeppa ED, Kepe V, Shoghi-Jadid K, et
model-based segmentation of 3-D neu- 84. Salmon E, Sadzot B, Maquet P, et al. al. In vivo and in vitro labeling of
roradiological data sets. IEEE Trans Med Differential diagnosis of Alzheimer’s dis- plaques and tangles in the brain of an
Imaging 1999; 18:828 –839. ease with PET. J Nucl Med 1994; 35:391– Alzheimer’s disease patient: a case study
70. Pickut BA, Dierckx RA, Dobbeleir A, et 398. (abstr). J Nucl Med 2001; 42(suppl 1):
al. Validation of the cerebellum as a ref- 85. Okamura N, Arai H, Higuchi M, et al. 65P.
erence region for SPECT quantification [18F]FDG-PET study in dementia with 99. Shoghi-Jadid K, Small GW, Agdeppa ED,
in patients suffering from dementia of Lewy bodies and Alzheimer’s disease. et al. Localization of neurofibrillary tan-
the Alzheimer type. Psychiatry Res 1999; Prog Neuropsychopharmacol Biol Psy- gles and beta-amyloid plaques in the
90:103–112. chiatry 2001; 25:447–456. brains of living patients with Alzheimer
71. Rodriguez G, Vitali P, Calvini P, et al. 86. Minoshima S, Giordani B, Berent S, Frey disease. Am J Geriatr Psychiatry 2002;
Hippocampal perfusion in mild Alzhei- KA, Foster NL, Kuhl DE. Metabolic re- 10:24 –35.
mer’s disease. Psychiatry Res 2000; 100: duction in the posterior cingulate cortex 100. Skovronsky DM, Zhang B, Kung MP,
65–74. in very early Alzheimer’s disease. Ann Kung HF, Trojanowski JQ, Lee VM. In
72. Jobst KA, Barnetson LP, Shepstone BJ. Neurol 1997; 42:85–94. vivo detection of amyloid plaques in a
Accurate prediction of histologically 87. Kennedy AM, Frackowiak RS, Newman mouse model of Alzheimer’s disease.
confirmed Alzheimer’s disease and the SK, et al. Deficits in cerebral glucose me- Proc Natl Acad Sci U S A 2000; 97:7609 –
differential diagnosis of dementia: the tabolism demonstrated by positron 7614.
use of NINCDS-ADRDA and DSM-III-R emission tomography in individuals at 101. Rosen BR, Belliveau JW, Chien D. Perfu-
criteria, SPECT, x-ray CT, and Apo E4 in risk of familial Alzheimer’s disease. Neu- sion imaging by nuclear magnetic reso-
medial temporal lobe dementias: Oxford rosci Lett 1995; 186:17–20. nance. Magn Reson Q 1989; 5:263–281.
Project to Investigate Memory and Ag- 88. Small GW, Mazziotta JC, Collins MT, et 102. Rempp KA, Brix G, Wenz F, Becker CR,
ing. Int Psychogeriatr 1998; 10:271–302. al. Apolipoprotein E type 4 allele and Guckel F, Lorenz WJ. Quantification of
73. Jagust W, Thisted R, Devous MD Sr, et al. cerebral glucose metabolism in relatives regional cerebral blood flow and volume
SPECT perfusion imaging in the diagno- at risk for familial Alzheimer disease. with dynamic susceptibility contrast-en-
sis of Alzheimer’s disease: a clinical- JAMA 1995; 273:942–947. hanced MR imaging. Radiology 1994;
pathologic study. Neurology 2001; 56: 89. Drzezga A, Lautenschlage YU, Meno- 193: 637–641.
950 –956. shima S, et al. Glucose metabolic change 103. Gonzalez RG, Fischman AJ, Guimaraes
74. Charpentier P, Lavenu I, Defebvre L, et associated with conversion from mild AR, et al. Functional MR in the evalua-
al. Alzheimer’s disease and frontotem- cognitive impairment to Alzheimer’s tion of dementia: correlation of abnor-
poral dementia are differentiated by disease: a follow-up PET study (abstr). mal dynamic cerebral blood volume
discriminant analysis applied to Tc J Nucl Med 2001; 42(suppl 1):60P. measurements with changes in cerebral
HmPAO SPECT data. J Neurol Neurosurg 90. Silverman DH, Small GW, Chang CY, et metabolism on positron emission to-
Psychiatry 2000; 69:661–663. al. Positron emission tomography in mography with fludeoxyglucose F18.
75. Sjogren M, Gustafson L, Wikkelso C, evaluation of dementia: regional brain AJNR Am J Neuroradiol 1995; 16:1763–
Wallin A. Frontotemporal dementia can metabolism and long-term outcome. 1770.
be distinguished from Alzheimer’s dis- JAMA 2001; 286:2120 –2127. 104. Harris GJ, Lewis RF, Satlin A, et al. Dy-

334 䡠 Radiology 䡠 February 2003 Petrella et al

 namic susceptibility contrast MR imag- of normal aging and Alzheimer disease. MRS: two cases. Magn Reson Imaging
ing of regional cerebral blood volume in Neuroimage 2000; 11:179 –187. 1993; 11:1209 –1215.
Alzheimer disease: a promising alterna- 120. Thulborn KR, Martin C, Voyvodic JT. 134. Miller EK, Li L, Desimone R. Activity of
tive to nuclear medicine. AJNR Am J Functional MR imaging using a visually neurons in anterior inferior temporal
Neuroradiol 1998; 19:1727–1732. guided saccade paradigm for comparing cortex during a short-term memory
105. Harris GJ, Lewis RF, Satlin A, et al. Dy- activation patterns in patients with task. J Neurosci 1993; 13:1460 –1478.
Radiology

namic susceptibility contrast MRI of re- probable Alzheimer’s disease and in cog- 135. Frederick BB, Satlin A, Yurgelun-Todd
gional cerebral blood volume in Alzhei- nitively able elderly volunteers. AJNR DA, Renshaw PF. In vivo proton mag-
mer’s disease. Am J Psychiatry 1996; Am J Neuroradiol 2000; 21:524 –531. netic resonance spectroscopy of Alzhei-
153: 721–724. 121. Saykin AJ, Flashman LA, Frutiger SA, et mer’s disease in the parietal and tempo-
106. Maas LC, Harris GJ, Satlin A, English al. Neuroanatomic substrates of seman- ral lobes. Biol Psychiatry 1997; 42:147–
CD, Lewis RF, Renshaw PF. Regional ce- tic memory impairment in Alzheimer’s 150.
rebral blood volume measured by dy- disease: patterns of functional MRI acti- 136. MacKay S, Ezekiel F, Di Sclafani V, et al.
namic susceptibility contrast MR imag- vation. J Int Neuropsychol Soc 1999; 5: Alzheimer disease and subcortical isch-
ing in Alzheimer’s disease: a principal 377–392. emic vascular dementia: evaluation by
components analysis. J Magn Reson Im- 122. Corkin S, Kennedy AM, Bucci J, et al. combining MR imaging segmentation
aging 1997; 7:215–219. Relation between recognition perfor- and H-1 MR spectroscopic imaging. Ra-
107. Bozzao A, Floris R, Baviera ME, Apruzz- mance and fMRI data in Alzheimer’s dis- diology 1996; 198:537–545.
ese A, Simonetti G. Diffusion and perfu- ease and older normal subjects. Soc 137. Constans JM, Meyerhoff DJ, Gerson J, et
sion MR imaging in cases of Alzheimer’s Neuro 1997; 23:193–195. al. H-1 MR spectroscopic imaging of
disease: correlations with cortical atro- 123. Smith CD, Andersen AH, Kryscio RJ, et white matter signal hyperintensities:
phy and lesion load. AJNR Am J Neuro- al. Altered brain activation in cogni- Alzheimer disease and ischemic vascular
radiol 2001; 22:1030 –1036. tively intact individuals at high risk for dementia. Radiology 1995; 197:517–
108. Ogawa S, Lee TM, Kay AR, Tank DW. Alzheimer’s disease. Neurology 1999; 523.
Brain magnetic resonance imaging with 53: 1391–1396. 138. Ernst T, Chang L, Melchor R, Mehringer
contrast dependent on blood oxygen- 124. Small GW, Ercoli LM, Silverman DH, et CM. Frontotemporal dementia and
ation. Proc Natl Acad Sci U S A 1990; al. Cerebral metabolic and cognitive de- early Alzheimer disease: differentiation
87:9868 –9872. cline in persons at genetic risk for Alz- with frontal lobe H-1 MR spectroscopy.
109. Courtney SM, Ungerleider LG, Keil K, heimer’s disease. Proc Natl Acad Sci Radiology 1997; 203:829 –836.
Haxby JV. Transient and sustained activ- U S A 2000; 97:6037–6042. 139. Shonk TK, Moats RA, Gifford P, et al.
ity in a distributed neural system for 125. Doraiswamy PM, Chen JG, Charles HC. Probable Alzheimer disease: diagnosis
human working memory. Nature 1997; Brain magnetic resonance spectroscopy: with proton MR spectroscopy. Radiol-
386:608 –611. role in assessing outcomes in Alzhei- ogy 1995; 195:65–72.
110. Ojemann JG, Buckner RL, Corbetta M, mer’s disease. CNS Drugs 2000; 14:457– 140. Parnetti L, Lowenthal DT, Presciutti O,
et al. H-MRS, MRI-based hippocampal
Raichle ME. Imaging studies of memory 472.
volumetry and Tc-HMPAO-SPECT in
and attention. Neurosurg Clin N Am 126. Chen JG, Charles HC, Barboriak DP, Do-
normal aging, age-associated memory
1997; 8:307–309. raiswamy PM. Magnetic resonance spec-
impairment, and probable Alzheimer’s
111. Gabrieli JD, Poldrack RA, Desmond JE. troscopy in Alzheimer’s disease: focus
disease. J Am Geriatr Soc 1996; 44:133–
The role of left prefrontal cortex in lan- on N-acetylaspartate. Acta Neurol Scand
138.
guage and memory. Proc Natl Acad Sci Suppl 2000; 176:20 –26.
141. Moats RA, Ernst T, Shonk TK, Ross BD.
U S A 1998; 95:906 –913. 127. Lazeyras F, Charles HC, Tupler LA, Erick-
Abnormal cerebral metabolite concen-
112. Wagner AD, Desmond JE, Glover GH, son R, Boyko OB, Krishnan KR. Meta-
trations in patients with probable Alz-
Gabrieli JD. Prefrontal cortex and recog- bolic brain mapping in Alzheimer’s dis-
heimer disease. Magn Reson Med 1994;
nition memory: functional-MRI evi- ease using proton magnetic resonance
32:110 –115.
dence for context-dependent retrieval spectroscopy. Psychiatry Res 1998; 82: 142. Parnetti L, Tarducci R, Presciutti O, et al.
processes. Brain 1998; 121:1985–2002. 95–106. Proton magnetic resonance spectros-
113. Gabrieli JD, Poldrack RA, Desmond JE. 128. Klunk WE, Panchalingam K, Moossy J, copy can differentiate Alzheimer’s dis-
The role of left prefrontal cortex in lan- McClure RJ, Pettegrew JW. –acetyl-L-as- ease from normal aging. Mech Ageing
guage and memory. Proc Natl Acad Sci partate and other amino acid metabo- Dev 1997; 97:9 –14.
U S A 1998; 95:906 –913. lites in Alzheimer’s disease brain: a pre- 143. Kreis R, Ross BD, Farrow NA, Ackerman
114. Buckner RL, Kelley WM, Petersen SE. liminary proton nuclear magnetic Z. Metabolic disorders of the brain in
Frontal cortex contributes to human resonance study. Neurology 1992; 42: chronic hepatic encephalopathy de-
memory formation. Nat Neurosci 1999; 1578 –1585. tected with 1H MR spectroscopy. Radi-
2:311–314. 129. Klunk WE, Panchalingam K, McClure ology 1992; 182:19 –27.
115. Cohen NJ, Ryan J, Hunt C, Romine L, RJ, Stanley JA, Pettegrew JW. Metabolic 144. Haussinger D, Laubenberger J, vom
Wszalek T, Nash C. Hippocampal system alterations in postmortem Alzheimer’s Dahl S, et al. Proton magnetic resonance
and declarative (relational) memory: sum- disease brain are exaggerated by Apo-E4. spectroscopy studies on human brain
marizing the data from functional neuro- Neurobiol Aging 1998; 19:511–515. myo-inositol in hypoosmolarity and he-
imaging studies. Hippocampus 1999; 9: 130. Kwo-On-Yuen PF, Newmark RD, patic encephalopathy. Gastroenterology
83–98. Budinger TF, Kaye JA, Ball MJ, Jagust WJ. 1994; 107:1475–1480.
116. Puce A, Allison T, Gore JC, McCarthy G. Brain N-acetyl-L-aspartic acid in Alzhei- 145. Kreis R, Ross BD. Cerebral metabolic dis-
Face-sensitive regions in human extra- mer’s disease: a proton magnetic reso- turbances in patients with subacute and
striate cortex studied by functional MRI. nance spectroscopy study. Brain Res chronic diabetes mellitus: detection
J Neurophysiol 1995; 74:1192–1199. 1994; 667:167–174. with proton MR spectroscopy. Radiol-
117. Bookheimer SY, Strojwas MH, Cohen 131. Mohanakrishnan P, Fowler AH, Vonsat- ogy 1992; 184:123–130.
MS, et al. Patterns of brain activation in tel JP, et al. An in vitro 1H nuclear mag- 146. Kruse B, Hanefeld F, Christen HJ, et al.
people at risk for Alzheimer’s disease. netic resonance study of the temporopa- Alterations of brain metabolites in meta-
N Engl J Med 2000; 343:450 –456. rietal cortex of Alzheimer brains. Exp chromatic leukodystrophy as detected
118. Small SA, Nava AS, Perera GM, Delapaz Brain Res 1995; 102:503–510. by localized proton magnetic resonance
R, Stern Y. Evaluating the function of 132. Jessen F, Block W, Traber F, et al. Proton spectroscopy in-vivo. J Neurol 1993;
hippocampal subregions with high-res- MR spectroscopy detects a relative de- 241: 68 –74.
olution MRI in Alzheimer’s disease and crease of N-acetylaspartate in the medial 147. Doraiswamy PM, Charles HC, Krishnan
aging. Microsc Res Tech 2000; 51:101– temporal lobe of patients with AD. Neu- KR. Prediction of cognitive decline in
108. rology 2000; 55:684 –688. early Alzheimer’s disease (letter). Lancet
119. Johnson SC, Saykin AJ, Baxter LC, et al. 133. Longo R, Giorgini A, Magnaldi S, Pasca- 1998; 352:1678.
The relationship between fMRI activa- zio L, Ricci C. Alzheimer’s disease histo- 148. Satlin A, Bodick N, Offen WW, Renshaw
tion and cerebral atrophy: comparison logically proven studied by MRI and PF. Brain proton magnetic resonance

Volume 226 䡠 Number 2 Neuroimaging and Early Diagnosis of Alzheimer Disease 䡠 335
 spectroscopy (H-MRS) in Alzheimer’s diffusion in cerebral white matter in Alz- Neurology. Neurology 2001; 56:1143–
disease: changes after treatment with heimer’s disease. Gerontology 1997; 43: 1153.
xanomeline, an M1 selective cholinergic 343–351. 158. Doody RS, Stevens JC, Beck C, et al.
agonist. Am J Psychiatry 1997; 154:1459 – 153. Kantarci K, Jack CR, Xu YC, et al. Mild Practice parameter: management of de-
1461. cognitive impairment and Alzheimer mentia (an evidence-based review)—re-
149. Waldman AD, McConnell JR, Rai GS, disease: regional diffusivity of water. Ra- port of the Quality Standards Subcom-
Radiology

Chaudry M, Grant DS, Martin PA. Auto- diology 2001; 219:101–107. mittee of the American Academy of
mated proton MRS of the brain at 1.0 T: 154. Pierpaoli C, Jezzard P, Basser PJ, Barnett Neurology. Neurology 2001; 56:1154 –
reproducibility and clinical utility in A, Di Chiro G. Diffusion tensor MR im- 1166.
Alzheimer’s disease. Presented at the aging of the human brain. Radiology 159. Petersen RC, Stevens JC, Ganguli M,
Sixth Meeting of the International Soci- 1996; 201:637–648. Tangalos EG, Cummings JL, DeKosky
ety for Magnetic Resonance in Medi- 155. Melhem ER, Mori S, Mukundan G, Kraut ST. Practice parameter: early detection
cine, Sydney, Australia, April 18 –24, MA, Pomper MG, van Zijl PC. Diffusion of dementia: mild cognitive impairment
1998. tensor MR imaging of the brain and (an evidence-based review)—report of
150. Sandson TA, Felician O, Edelman RR, white matter tractography. AJR Am J the Quality Standards Subcommittee of
Warach S. Diffusion-weighted magnetic Roentgenol 2002; 178:3–16. the American Academy of Neurology.
resonance imaging in Alzheimer’s dis- 156. Rose SE, Chen F, Chalk JB, et al. Loss of Neurology 2001; 56:1133–1142.
ease. Dement Geriatr Cogn Disord 1999; connectivity in Alzheimer’s disease: an 160. Silverman PHS, Chang CY, Cummings
10:166 –171. evaluation of white matter tract integ- JL, et al. Prognostic value of regional
151. Hanyu H, Sakurai H, Iwamoto T, rity with colour coded MR diffusion ten- brain metabolism in evaluation of de-
Takasaki M, Shindo H, Abe K. Diffusion- sor imaging. J Neurol Neurosurg Psychi- mentia (abstr). J Nucl Med 1999;
weighted MR imaging of the hippocam- atry 2000; 69:528 –530. 40(suppl 1):71P.
pus and temporal white matter in Alz- 157. Knopman DS, DeKosky ST, Cummings 161. George AE, Cha S. What role does func-
heimer’s disease. J Neurol Sci 1998; 156: JL, et al. Practice parameter: diagnosis of tional MR imaging play in the diagnosis
195–200. dementia (an evidence-based review)— or prediction of future-onset Alzhei-
152. Hanyu H, Shindo H, Kakizaki D, Abe K, report of the Quality Standards Subcom- mer’s disease? AJNR Am J Neuroradiol
Iwamoto T, Takasaki M. Increased water mittee of the American Academy of 2001; 22:1017–1018.

336 䡠 Radiology 䡠 February 2003 Petrella et al