MRS shows abnormalities before
symptoms in familial Alzheimer disease
A.K. Godbolt, MRCP; A.D. Waldman, PhD; D.G. MacManus, MSc; J.M. Schott, MD; C. Frost, MSc;
L. Cipolotti, PhD; N.C. Fox, MD; and M.N. Rossor, MD
Abstract—Background: Pathologic change in Alzheimer disease (AD) begins some years before symptoms. MRS has the
potential to detect metabolic abnormalities reflecting this early pathologic change. Presenilin 1 (PS1) and amyloid
precursor protein (APP) mutation carriers have a nearly 100% risk of developing AD and may be studied prior to symptom
onset. Methods: Short echo time proton MR spectra were acquired from a midline posterior cingulate voxel in presymp-
tomatic carriers of PS1 or APP mutations (“presymptomatic mutation carriers” [PMCs]; n ? 7) and age- and sex-matched
control subjects (n ? 6). Ratios of N-acetyl aspartate (NAA), myo-inositol (MI), and choline-containing compounds (Cho) to
creatine (Cr) were measured and NAA/MI calculated. Regression analyses and t tests were performed after log transfor-
mation. Results: PMC and control subjects were matched for age and sex. PMC subjects were 1.7 to 21.6 years (mean 9.8
years) before expected symptom onset, predicted from family-specific mean age at onset. Age did not significantly affect
metabolite ratios. Geometric mean ratios in control subjects were as follows: NAA/Cr ? 1.75, MI/Cr ? 0.59, and NAA/MI ?
2.95. NAA/Cr and NAA/MI were significantly reduced in PMC relative to controls (NAA/Cr mean decrease 10% [95% CI 2
to 18%]; NAA/MI mean decrease 25% [95% CI 3 to 44%]). MI/Cr was increased in PMC, but the differences did not achieve
significance (19% increase [95% CI 1% decrease to 41% increase]; p ? 0.07)). In PMCs, reduction in NAA/MI (p ? 0.001)
and MI/Cr (p ? 0.002) were related to proximity of expected age at onset. Conclusions: Metabolic changes are detectable in
presymptomatic mutation carriers years before expected onset of Alzheimer disease. Their magnitude is related to
proximity of expected age at onset.
Pathologic change in Alzheimer disease (AD) is
thought to start some years before symptoms begin.1
Localized MRS allows in vivo measurement of major
metabolites in defined regions of the brain and thus
has the potential to permit noninvasive detection of
abnormalities that reflect these early pathologic pro-
cesses.2Short echo time proton MRS allows evalua-
tion of N-acetyl aspartate (NAA), myo-inositol (MI),
creatine/phosphocreatine (Cr), choline-containing
compounds (Cho), and glutamine/glutamate (Glx).
Reduction in NAA and increase in MI are consistent
findings in established AD.3
The posterior cingulate shows hypometabolism4
and atrophy5both in presymptomatic AD and
throughout the course of the disease. It is a well-
defined midline structure from which reproducible,
high-quality single-voxel spectra can be acquired and
is therefore an attractive site for MRS in early AD.
With use of a posterior cingulate voxel, increased
MI/Cr has been shown in patients with mild cogni-
tive impairment (MCI),6and increases in MI and
decreases in NAA have been shown in nondemented
patients with Down syndrome.7These results raise
the possibility that posterior cingulate MRS may be
useful in the presymptomatic diagnosis of AD. MCI
patients are, however, a heterogeneous group, in-
cluding individuals with static cognitive impairment;
individuals who subsequently develop other demen-
tias, and patients who are manifesting the earliest
symptoms of AD. Patients possessing mutations in
amyloid precursor protein (APP), presenilin 1 (PS1),
or presenilin 2 (PS2) are destined to develop AD, and
families with a given mutation develop symptoms at
a fairly consistent age (see http://www.molgen.ua.
ac.be/ADMutations/). Such patients may be studied
from a presymptomatic stage and thus provide a
unique opportunity to investigate the very earliest
stages of the disease. In this study, we assessed
whether subjects destined to develop AD on the basis
of carrying mutations in APP or PS1 had metabolite
abnormalities detectable with MRS prior to the onset
those at risk of familial AD (FAD). The study was approved by the
local ethics committee, and all subjects gave informed consent.
Subjects were a subset of those recruited to a study of
From the Dementia Research Centre (A.K.G., A.D.W., J.M.S., C.F., N.C.F., M.N.R.) and NMR Research Unit (D.G.M.), Institute of Neurology, University
College of London, London School of Hygiene and Tropical Medicine (C.F.), University of London, Department of Neuropsychology (L.C.), National Hospital
for Neurology and Neurosurgery, Division of Neuroscience and Psychological Medicine (M.N.R.), Imperial College, and Charing Cross Hospital (A.D.W.),
Supported by MRC program grant G9626876 and a grant from the Alzheimer’s Society (J.M.S.). N.C.F. is a Medical Research Council Senior Clinical Fellow.
Disclosure: The authors report no conflicts of interest.
Received June 24, 2005. Accepted in final form November 4, 2005.
Address correspondence and reprint requests to Dr. M.N. Rossor, Dementia Research Centre, Institute of Neurology, Queen Square, London WC1N 3BG, UK;
Copyright © 2006 by AAN Enterprises, Inc.
Each subject had chosen independently to have predictive genetic
testing through an independent geneticist and was found to carry
a mutation in either PS1 or APP. Such subjects are at approach-
ing 100% risk of developing FAD. All these subjects (“presymp-
tomatic mutation carriers” [PMCs]) were aware of their genetic
status but had no complaints of memory problems; an indepen-
dent history from a family member corroborated their asymptom-
atic status. Six of seven subjects were working; one younger
subject was unemployed. The patients in this study were from a
total of six FAD families, each with a known mutation. For each
patient, the expected age at symptom onset was predicted from
the mean age at symptom onset in their affected family members.
The age at onset in each family member with FAD was deter-
mined from interviews with relatives and medical records and was
the age at which symptoms were first reported (table 1). Control
subjects were spouses or family members in whom genetic testing
had excluded PS1 or APP mutations.
PMC subjects underwent a clinical and neuropsychological as-
sessment at the time of MRI. Clinical examination included sepa-
rate interviews with the patient and an informant, assessment
using the Mini-Mental State Examination,8and a full neurologic
examination. Neuropsychological examination included assess-
ment of verbal and visual IQ,9verbal and visual memory,10,11vi-
sion perception,12frontal executive function,13calculation,14and
nominal15and linguistic16ability. All subjects except Subject 4 had
had between one and four previous neuropsychological assess-
ments, the most recent a minimum of 6 months previously. Alter-
native versions of the memory tests were used on repeat
assessments to minimize practice effects.
Single-voxel proton MR spectra were acquired on a 1.5
T Signa 5X system using an automated point-resolved spectros-
copy (PRESS) technique (repetition time ? 2,000 milliseconds,
echo time ? 30 milliseconds, 192 averages; General Electric Med-
ical Systems, Milwaukee, WI). (figure 1) All spectra were acquired
by the same experienced radiographer. A T1-weighted spin echo
sagittal scout was obtained, followed by six axial 5-mm fast inver-
sion recovery images. A 15-mm-thick midline voxel of interest was
placed within three adjacent slices, with the other voxel dimen-
sions sized individually to maximize gray matter cingulate con-
tent (volume range 3.0 to 7.4 mL). For each individual, the
metabolite ratios NAA/Cr, Cho/Cr, and MI/Cr were measured us-
ing on-line software (PROBE-Q) and the ratio NAA/MI was
NAA/Cr, MI/Cr, Cho/Cr, and NAA/MI
were analyzed on a logarithmic scale to ensure invariance of re-
sults to choice of numerator in ratios (e.g., equivalent results will
be obtained using NAA/MI and MI/NAA). Ratios were compared
using t tests. Potential confounding by age was assessed using
linear regression analysis. For mutation carriers, the relationship
of NAA/MI and proximity to onset was analyzed using linear
spouses and two subjects who had had genetic testing ex-
cluding PS1 and APP mutations) were recruited. Mean age
was 36 years (range 27 to 50 years) for mutation carriers
(four women) and 37 years (range 25 to 50 years) for con-
trol subjects (four women). No subjects met criteria for
MCI17or AD18or had diabetes mellitus or other medical
conditions. None was taking cholinesterase inhibitors.
PMCs had mutations in PS1 (delta 4, L250S, L235V,
M139V, M146I [two subjects]) or APP (V717I).
Mutation carriers had a mean Verbal IQ of 100 (range
83 to 122) and mean Performance IQ of 108 (range 84 to
130) (table 2). Only one subject (2.6 years from expected
onset) showed intellectual underfunctioning (Performance
IQ only) defined as a score ?10 points of that predicted by
the National Adult Reading Test.16All subjects scored
above the 5th percentile on tests of recognition memory,
although verbal recognition memory was weak in three
subjects, with one (Subject 2) also scoring below the 5th
percentile on a recall test. Visual memory was well pre-
served in all subjects. In other cognitive domains, the only
abnormality was frontal executive dysfunction in one sub-
ject (Subject 2).
Three subjects had a mismatch between Verbal and
Performance IQ, with Performance IQ being more than 15
points higher. Whereas a lower Verbal than Performance
IQ may represent decline in the verbal domain, it is nota-
ble that in these subjects the Verbal IQs remained compa-
rable with their National Adult Reading Test predicted
scores. Therefore, we consider a more likely explanation in
these cases to be differential practice effects on Verbal and
Performance IQ resulting in a greater increase in Perfor-
mance than Verbal IQ on repeat assessment.9
Metabolite ratios are given in figure 2 and summarized
in table 3. PMCs had significantly lower NAA/Cr and NAA/
MI, with a borderline increase in MI/Cr (p ? 0.07). Cho/Cr
was not significantly different in PMC and control sub-
jects. Metabolite ratios in the whole group were not related
Seven PMCs and six control subjects (four
Table 1 Age at onset in family members
Mean age at
onset in family, ySD, y
Lowest recorded age
at onset, y
Highest recorded age
at onset, yn*
6 and 7F48.32.146503
*Number of family members with familial Alzheimer disease in whom age at onset is known. Age at onset was determined from infor-
mation from family members and medical records and represents the age at which symptoms were first reported.
Figure 1. Sample spectra from a presymptomatic mutation
carrier (A) and a control subject (B). MI ? myo-inositol;
Cr ? creatine/phosphocreatine; NAA ? N-acetyl aspartate.
March (1 of 2) 2006
to age (NAA/Cr p ? 0.54, MI/Cr p ? 0.52, NAA/Cr p ?
0.49), and the differences between PMC and control sub-
jects were not materially altered after adjustment for age.
Therefore, results shown are not age adjusted. In the PMC
group, proximity to expected age at onset was associated
with a lower NAA/MI ratio (p ? 0.001; figure 3). Both NAA
and MI contributed to this effect, with a stronger relation-
ship for increase in MI/Cr (p ? 0.002) than for decrease in
NAA (p ? 0.034).
strate that metabolite changes in the posterior cin-
gulate are detectable by in vivo proton MRS at a very
early preclinical stage of AD. All of the PMC subjects
were considered by themselves and their families to
have no memory or other cognitive problems and
thus had not presented to medical services other
than as research subjects. All but one (Subject 5) was
employed, with no recent change in employment. In
this practical clinical sense, they were therefore
However, it could be argued that some subjects
did already have subtle evidence of decline on neuro-
In this study of PMCs, we demon-
psychological assessment, although only in Subject 2
was this noted in the conclusion of the examining
neuropsychologist at the time of assessment, and
even then low mood and anxiety were thought to
have confounded the results. In most respects, sub-
jects performed very well on neuropsychological as-
sessment, with four subjects having scores higher
than the 75th percentile on a test of visual recogni-
tion memory and three on a test of verbal recognition
Two of the three subjects who performed rela-
tively poorly (i.e., between the 5th and 10th percen-
tile) on verbal recognition memory tests, in fact, had
perfectly satisfactory scores on the more sensitive
recall memory test, suggesting that anxiety may
have played a part in their performance. Only Sub-
ject 2 had significant decline in expected Perfor-
mance IQ as estimated using the National Adult
Reading Test, a score below the 5th percentile on one
test of recall memory, and poor performance on some
tasks sensitive to executive dysfunction. This might
suggest that this patient was already experiencing
early AD, although in the absence of a memory com-
plaint or impairment of activities of daily living, nei-
ther a diagnosis of MCI nor a diagnosis of AD could
Our subjects had a range of mutations in PS1 and
APP, suggesting that these findings are not likely to
be mutation specific. Only one subject had an APP
mutation (Subject 2), and although the same pattern
of metabolite changes was seen, further APP muta-
tion carriers need to be studied. Very rare cases of
nonpenetrance of PS1 and APP mutations have been
reported,19and we can therefore not be absolutely
certain that all subjects will eventually be diagnosed
with AD. However, all subjects remain under review,
and any nonpenetrant cases would be expected to
reduce the magnitude of the differences between the
groups. Previous studies of familial and sporadic AD
suggest that regional atrophy may begin around 3.5
years before symptoms,20so it is perhaps not surpris-
ing that our results suggest that metabolic change
may occur even earlier. The two PMC subjects who
Table 2 Neuropsychological scores for mutation carriers
expected onset*Age, y
(max 30) VIQPIQ
PA trial 1
PA trial 2
1 1.7 35.530101 122100 101 25–50
2 2.6 49.92798 84 94 96 5–1010–2510–25
59.629.2 30839986 88
6 20.2 28.130 107122 102104
*Expected onset estimated from mean age at onset in the subject’s family.
MMSE ? Mini-Mental State Examination8; VIQ ? Verbal IQ9; PIQ ? Performance IQ9; NART ? National Adult Reading Test;
RMW ? Recognition Memory Test for Words10; RMF ? Recognition Memory Test for Faces10; PA ? paired associate learning.11
Figure 2. Metabolite ratios. Presymptomatic mutation car-
riers are identified by subject number as given in table 2,
with Subject 1 having the least and Subject 7 the greatest
time remaining before expected symptom onset. (A)
N-Acetyl aspartate (NAA)/creatine (Cr); (B) myo-inositol
(MI)/Cr; (C) choline-containing compounds (Cho)/Cr; (D)
March (1 of 2) 2006
were over 20 years from expected onset both had
NAA/MI ratios within the control range, but all oth-
ers, including those 7 and 10 years before expected
onset, had NAA/MI outside the control range.
Increased MI/Cr, but not decreased NAA/Cr, has
previously been reported in patients with MCI,6a
group at increased risk of developing AD. A further
study by the same investiators21found that elevated
MI/Cr occurred only in patients with the amnestic
subtype of MCI, who are most likely to develop AD.
Our finding of lower NAA/Cr in mutation carriers, if
confirmed with larger numbers, might suggest that a
presymptomatic decline in NAA is restricted to FAD.
However, both decreased NAA and increased MI con-
centrations have been reported before the onset of
AD in patients with Down syndrome. In that study,
older subjects appeared to have a greater elevation
in MI, although no further decrease in NAA.7
Interestingly, we found that NAA/MI declined
with increasing proximity to expected time of onset
of AD symptoms, reflecting a strong relationship
with increasing MI/Cr and more modest association
with decreasing NAA/Cr. This suggests that metabo-
lite changes in the posterior cingulate region, partic-
ularly increasing MI, may occur progressively in the
presymptomatic phase of FAD. Although we ac-
knowledge that we can only estimate the age at on-
set of these subjects, families with AD mutations
develop symptoms at fairly consistent ages (table 1),
and the strength of the association is nevertheless
NAA is found exclusively in neurons, and decrease
in this metabolite is widely considered to reflect neu-
ronal loss or dysfunction. The pathophysiologic im-
portance of elevated MI in AD is unresolved. A
leading hypothesis is that it reflects increased glial
cell numbers; MI is present at high concentrations in
activated glia, where it is thought to play an impor-
tant osmoregulatory role. It has also been suggested
that alterations in cellular detoxification pathways,
and in the inositol triphosphate intracellular second
messenger cycle, may account for increases in this
metabolite.2,3More detailed knowledge of how these
metabolite derangements reflect cellular abnormali-
ties and their progression with time would aid our
understanding of the sequence of pathologic changes
Further studies are warranted to define more ac-
curately how early these metabolic changes may be
determined, to assess whether metabolite measures
may be used to predict which patients with MCI will
progress to fulfill a diagnosis of sporadic AD, and
ultimately whether this noninvasive technique can
identify asymptomatic elderly individuals at risk of
The authors thank the subjects and their families. Assistant psy-
chologists of the National Hospital for Neurology and Neurosur-
gery performed the neuropsychological assessments.
1. Braak H, Braak E. Neuropathological staging of Alzheimer-related
changes. Acta Neuropathol 1991;82:239–259.
2. Jones RS, Waldman AD. 1H-MRS evaluation of metabolism in Alzhei-
mer’s disease and vascular dementia. Neurol Res 2004;26:488–495.
3. Valenzuela MJ, Sachdev P. Magnetic resonance spectroscopy in AD.
4. Mosconi L. Brain glucose metabolism in the early and specific diagnosis
of Alzheimer’s disease FDG-PET studies in MCI and AD. Eur J Nucl
Med Mol Imag 2005;32:486–510.
5. Scahill RI, Schott JM, Stevens JM, Rossor MN, Fox NC. Mapping the
evolution of regional atrophy in Alzheimer’s disease: unbiased analysis
of fluid-registered serial MRI. Proc Natl Acad Sci USA 2002;99:4703–
6. Kantarci K, Jack CR, Xu YC, et al. Regional metabolic patterns in mild
cognitive impairment and Alzheimer’s disease: a 1H MRS study. Neu-
7. Huang W, Alexander GE, Daly EM, et al. High brain myo-inositol levels
Figure 3. Relationship between N-acetyl aspartate/myo-
inositol and number of years until expected onset in
presymptomatic mutation carriers (diamonds). Control
values (triangles) are given for comparison.
Table 3 Metabolite ratios
mean (95% CI)
mean (95% CI)
% change in PMC compared
with controls (95% CI)
NAA/Cr 1.57 (1.44–1.70) 1.75 (1.66–1.84)10%2 (22–18%2) 0.02
MI/Cr 0.71 (0.61–0.81)0.59 (0.50–0.70)19%1 (1%2–44%1) 0.07
Cho/Cr 0.61 (0.55–0.67)0.60 (0.52–0.70) 1%1 (13%2–17%1) 0.90
NAA/MI2.22 (1.81–2.72)2.95 (2.42–3.59) 25%2 (32–41%2) 0.03
*p Value from t test on log values with unequal variance.
PMC ? presymptomatic mutation carrier; NAA ? N-acetyl aspartate; Cr ? creatine; MI ? myo-inositol; Cho ? choline-containing com-
March (1 of 2) 2006
in the predementia phase of Alzheimer’s disease in adults with Down’s Download full-text
syndrome: a 1H MRS study. Am J Psychiatry 1999;156:1879–1886.
8. Folstein M, Folstein S, McHughs P. The “Mini Mental State”: a practi-
cal method for grading the cognitive state of patients for the clinician. J
Psychiatr Res 1975;12:189–198.
9. Wechsler D. Manual for the Wechsler Adult Intelligence Scale. Revised.
New York: Psychological Corp., 1981.
10. Warrington EK. Manual for the Recognition Memory Test for Words
and Faces. Windsor: NFER-Nelson, 1984.
11. Warrington EK. The Camden Memory Tests manual. Hove: Psychology
12. Warington EK, James M. The Visual Object and Space Perception
Battery. Bury St. Edmunds, UK: Thanes Valley Test Co., 1991.
13. Nelson HE. A modified card sorting test sensitive to frontal lobe defects.
14. Jackson M, Warrington EK. Arithmetic skills in patients with unilat-
eral cerebral lesions. Cortex 1986;22:611–620.
15. McKenna P, Warrington EK. The Graded Naming Test. Windsor, UK:
16. Nelson H. The National Adult Reading Test (NART): test manual.
Windsor: NFER-Nelson, 1982.
17. Jack CR, Petersen RC, Xu YC, et al. Prediction of AD with MRI-based
hippocampal volume in mild cognitive impairment. Neurology 1999;52:
18. McKhann G, Drachman D, Folstein M, Katzman R, Price D, Stadlan
EM. Clinical diagnosis of Alzheimer’s disease: report of the NINCDS-
ADRDA Work Group under the auspices of Department of Health and
Human Services Task Force on Alzheimer’s Disease. Neurology 1984;
19. Lleo A, Berezovska O, Growdon JH, Hyman BT. Clinical, pathological,
and biochemical spectrum of Alzheimer disease associated with PS-1
mutations. Am J Geriatr Psychiatry 2004;12:146–156.
20. Schott JM, Fox NC, Frost C, et al. Assessing the onset of structural
change in familial Alzheimer’s disease. Ann Neurol 2003;53:181–188.
21. Kantarci K, Petersen RC, Boeve BF, et al. 1H MRS findings differ
among mild cognitive impairment syndromes. Proc Int Soc Magn Res
DISAGREE? AGREE? HAVE A QUESTION? HAVE AN ANSWER?
Respond to an article in Neurology through our online Correspondence system:
• Visit www.neurology.org
• Access specific article on which you would like to comment
• Click on “Correspondence: Submit a response” in the content box
• Enter contact information
• Upload your Correspondence
• Press Send Response
Correspondence will then be transmitted to the Neurology Editorial Office for review. Accepted material will be posted
within 10–14 days of acceptance. Selected correspondence will subsequently appear in the print Journal. See our
Information for Authors at www.neurology.org for format requirements.
March (1 of 2) 2006