Content uploaded by Johannes Kornhuber
Author content
All content in this area was uploaded by Johannes Kornhuber on Sep 17, 2014
Content may be subject to copyright.
Fax +41 61 306 12 34
E-Mail karger@karger.ch
www.karger.com
Original Research Article
Dement Geriatr Cogn Disord 2010;29:448–456
DOI: 10.1159/000312685
Validation of the German Revised Addenbrooke’s
Cognitive Examination for Detecting Mild Cognitive
Impairment, Mild Dementia in Alzheimer’s Disease
and Frontotemporal Lobar Degeneration
P.Alexopoulos
a,b
A.Ebert
c
T.Richter-Schmidinger
a
E.Schöll
b
B.Natale
b
C.A.Aguilar
b,e
P.Gourzis
f
M.Weih
a
R.Perneczky
b
J.Diehl-Schmid
b
T.Kneib
d
H.Förstl
b
A.Kurz
b
A.Danek
c
J.Kornhuber
a
a Department of Psychiatr y and Psychotherapy, Friedrich-Alexander-Universität Erlangen-Nürnberg,
Erlangen , b Department of Psychiatry and Psychotherapy, Technische Universität München, and
c Department of Neurology, Ludwig-Maximilians-Universität München, Munich , and
d Institute of Mathematics,
Carl-von-Ossietzky-Universität Oldenburg, Oldenburg , Germany;
e Department of Child and Adolescent Psychiatry,
Hospital Barros Luco Trudeau, Universidad de Santiago de Chile, Santiago de Chile , Chile;
f Department of
Psychiatry, University of Patras, Patras , Greece
analyses. Results: The optimal cut-off score for the ACE-R for
detecting MCI, AD, and FTLD was 86/87, 82/83 and 83/84, re-
spectively. ACE-R was superior to MMSE only in the detec -
tion of patients with FTLD [area under the curve (AUC): 0.97
vs. 0.92], whilst the accuracy of the two instruments did not
differ in identifying MCI and AD. The ratio of the scores of the
memory ACE-R subtest to verbal fluency subtest contributed
significantly to the discrimination between AD and FTLD
(optimal cut-off score: 2.30/2.31, AUC: 0.77), whereas the
MMSE and ACE-R total scores did not. Conclusion: The Ger-
man ACE-R is superior to the most commonly employed
MMSE in detecting mild dementia in FTLD and in the differ-
ential diagnosis between AD and FTLD. Thus it might serve
as a valuable instrument as part of a comprehensive diag-
nostic workup in specialist centres/clinics contributing to
the diagnosis and differential diagnosis of the cause of de-
mentia. Copyright © 2010 S. Karge r AG, Base l
Key Words
Mild cognitive impairment Alzheimer’s disease
Frontotemporal lobar degeneration Addenbrooke’s
Cognitive Examination, revised edition Mini Mental State
Examination
A b s t r a c t
B a c k g r o u n d / A i m s : The diagnostic accuracy of the German
version of the revised Addenbrooke’s Cognitive Examination
(ACE-R) in identifying mild cognitive impairment (MCI), mild
dementia in Alzheimer’s disease (AD) and mild dementia in
frontotemporal lobar degeneration (FTLD) in comparison
with the conventional Mini Mental State Examination (MMSE)
was assessed. Methods: The study encompasses 76 cogni-
tively healthy elderly individuals, 75 patients with MCI, 56
with AD and 22 with FTLD. ACE-R and MMSE were validated
against an expert diagnosis based on a comprehensive diag-
nostic procedure. Statistical analysis was performed using
the receiver operating characteristic method and regression
Accepted: M arch 30, 2010
Published onlin e: May 26, 2010
Dr. med. P. Alexopoulos, K linik und Polikl inik für Psych iatrie und Psychot herapie
K linikum re chts der Isar der Techni schen Universität Münc hen
Ismaninger Strasse 22 , DE–81675 Munich (Germany)
Tel. +49 89 4140 4279, Fax +49 89 4140 4923
E-Mail panos.alexopoulos @ lrz.tu-muenchen.de
© 2010 S. Karger AG, Basel
1420–8008/10/0295–0448$26.00/0
Accessible online at:
www.karger.com/dem
Validation of the German Revised
Addenbrooke’s Cognitive Examination
Dement Geriatr Cogn Disord 2010;29:448–456
449
Introduction
Due to the rapidly ageing population in the Western
world, the incidence of age-related disorders such as de-
mentia is increasing substantially and dramatically. De-
mentia is predicted to become the largest health care
challenge in modern history because of its medical, social
and philosophical implications
[1] . An early diagnosis of
dementia appears to be of great importance in order to
institute appropriate medical and social interventions.
Alzheimer’s disease (AD) is the most common cause of
dementia
[2] , and frontotemporal lobar degeneration
(FTLD) is the second most common presenile cause of
neurodegenerative dementia
[3] . Depending on its cause,
the clinical phenotypes are quite distinct at the stage of
mild dementia. Patients with mild dementia in AD, for
instance, manifest memory deficits, make repetitive
statements, get lost while driving and have poor judg-
ment, whereas subjects with mild dementia in FTLD pri-
marily present with alterations of speech (language vari-
ant of FTLD) or with behavioural disturbances such as
disinhibition and lack of social awareness (behavioural
variant of FTLD)
[2] .
In most cases, the diagnosis of dementia is established
by assessing cognitive deficits and their impact on the
activities of daily liv ing. Neuropsychological instruments
enable clinicians to gather evidence of cognitive impair-
ment, help them to detect dementia and to make a differ-
ential diagnosis. In the light of the long duration (around
1–4 h) of extensive neuropsychological testing, which
limits its utility in routine dementia diagnostic workup,
there is a demand for screening tests
[4] . Ideally, such tests
are sensitive and specific, reliable and valid, quickly ad-
ministered, easily scored, and capture a broad range of
cognitive abilities across various levels of difficulty, thus
being efficient, not only in identifying mild dementia in
AD, but also other forms of dementia
[5, 6] .
Many screening tools for dementia which are available
in German have been criticized for several shortcomings.
For instance, the 7-Minute Screen is difficult to score and
to interpret
[7] . The DemTect [8] does not assess visuo-
spatial ability, known to be closely associated with func-
tional abilities
[9] . The 3MS has been validated in Ger-
many only for the identification of patients with early AD
but not with mild cognitive impairment (MCI) or with
dementias of other aetiologies
[10 ] . The Mini Mental State
Examination (MMSE)
[11] is currently the most widely
used measure of cognitive performance in Germany and
in many other countries. It has been shown to be a useful
tool for dis ting uish ing p eople with d ementia f rom cog ni-
tively healthy individuals and to achieve an acceptable
sensitivity and specificity in detecting dementia in clini-
cal samples
[1 2–1 4] . However, several weaknesses of the
MMSE have been repeatedly demonstrated, including
small number of items, low level of task difficulty and
likelihood of ceiling effects, narrow range of cognitive
domains assessed (including overrepresentation of orien-
tation, underrepresentation of memory tasks and absence
of tasks measuring executive function), limited range of
possible scores for individual items
[6] and low sensitiv-
ity for the detection of patients with MCI
[15 , 16] . Fur-
thermore, the diagnostic accuracy of the MMSE depends
on the patient’s age, education level, and ethnicity
[15 ] .
Drawing on extensive clinical and research expertise
with cognitive tests, Hodges and colleagues developed
the Addenbrooke’s Cognitive Examination test battery
[17 ] and a modified version of it. Changes were intro-
duced into the revised version of Addenbrooke’s Cogni-
tive Examination (ACE-R)
[18 ] in order to avoid ceiling
effects and to provide a more balanced contribution of
the component tests to the final score. Administration of
the ACE-R requires approximately 15 min and is easy to
score and to interpret. It incorporates the MMSE, but
elaborate s on memory, langua ge and visuospatial compo-
nents, and adds component tests of verbal fluency. The
memory component test assesses semantic and episodic
memory. In addition to recall of 3 items taken from the
MMSE, there is a name and address recall and recogni-
tion test for assessing episodic memory. Semantic mem-
ory is tested with 4 questions concerning general knowl-
edge. The language section comprises the naming of 12
l in e d ra wi ngs of i nter me di at e a nd lo w f am i li ar it y, se ma n-
tic comprehension, sentence comprehension, repetition
of words and phrases, reading of irregular words, and
writing a sentence. Executive functions are tested by let-
ter fluency (generating words beginning with the letter P
in 1 min) and category f luency (generating names of an-
imals in 1 min). Visuospatial testing includes the copying
of overlapping pentagons (from the MMSE) and of a wire
cube, the drawing of a clock face, dot counting and iden-
tifying fragmented letters
[18 ] .
The maximum ACE-R score of 100 points is composed
of 5 component test scores: orientation and attention (18
points), memory (26 points), verbal fluency (14 points),
language (26 points) and visuospatial ability (16 points)
[18 ] . The original validation study, which was conducted
at Addenbrooke’s Hospital, UK, included 241 partici-
pants. It suggested an optimal cut-off score of 88 points
for identifying dementia, which was associated with high
sensitivity (94%) and specificity (89%)
[18 ] .
Alexopoulos et al.
Dement Geriatr Cogn Disord 2010;29:448–456
450
The aim of the present study was to investigate the
ability of the German version of the ACE-R to detect pa-
tients with MCI, mild dementia in AD, and mild demen-
tia in FTLD, and to compare its diagnostic accuracy with
that of the MMSE. The ACE-R and MMSE were validated
against the clinical diagnosis based on a comprehensive
diagnostic workup.
Methods, Study Sample and Design
We translated and modified the instrument with advice from
the authors of the original ACE-R. The modifications concerned
the name and address recall and recognition tests, the semantic
memory test and the repetit ion test. Based on the origi nal criteria,
the German addresses were chosen from common street names
and from less well-known towns in order to preclude natural as-
sociations (e.g. Frauenstrasse 24, Spremberg, Brandenburg). In
the semantic memory test, we replaced ‘the current prime minis-
ter’ and ‘the woman who was prime minister’ by ‘the name of the
current chancellor’ and ‘t he name of the previous chancel lor’. The
words in the repetition test (‘Hippopotamus, Exzentrizität, Un-
verhältnismässigkeit, Verantwortungslosigkeit’ and ‘oberhalb,
ausserhalb und unterhalb’) were selected according to the criteria
used in the original English version: length, frequency and diffi-
culty to articulate. A bilingual researcher at the University of
Cambridge performed a back translation into English. Compari-
son of the original English version with the back translation
showed that the new version was simi lar to the origina l one except
for the modified items
[19 ] . The scoring system was not changed.
Like the English original, the German ACE-R can be adminis-
tered in approximately 15 min.
The validation study was carried out at 3 university memory
clinics, located in Erlangen and in Munich, Germany. The exam-
ination of the pa rticipants included a history from t he patient and
from an informant, medical, neurological and psychiatric exami-
nation, laboratory screening, brain imaging (MRI or CT) and a
neuropsychological examination based on the German version of
the Consortium to Establish a Registry for Alzheimer’s Disease
Neuropsychological Assessment Battery (CERAD-NAB)
[20] ,
which incorporates the MMSE. The additional neuropsychologi-
cal examination comprised a f lexible battery (Trail Making Test
[21] , Bayer Activities of Daily Living Scale [22] , and Neuropsychi-
atric Inventory
[23] ), the components of which varied according
to the aims of the neuropsychological assessment of each patient
as defined by the clinician.
The participants were German-speaking and had adequate v i-
sion and hearing, although many wore glasses and some required
a hearing aid. The ACE-R was administered after the clinical and
neuropsychological examination. The components of the ACE-R
that are identical to those of the MMSE were not administered
twice in order to preclude learning effects. The clinical diagnoses
were established independent of the performance of participants
on the ACE-R. The diagnosis of dementia and the assessment of
its overall severity were based on the criteria of the ICD-10 clas-
sification system
[24] . AD diagnosis was based on the criteria of
the National Institute of Neurological and Communicative Dis-
orders and Stroke-Alzheimer’s Disease and Related Disorders As-
sociation (NINCDS-ADRDA) for probable AD
[25] . The diagno-
sis of FTLD was established according to the revised Lund-Man-
chester criteria
[26] . To e ns ur e t ha t p at i en ts wi t h d em ent i a h ad no t
crossed the threshold to moderate dementia, patients with a score
below 15 poi nts on t he MMSE we re excluded fr om the s tudy. Thi s
score has been found to discriminate mild from moderate demen-
tia
[27] . MMSE staging has been proven to be an effective clinical
ins tr um ent for tr ac ki ng t he st age s o f de men ti a
[27] . T he diagnosis
of MCI followed the revised consensus criteria of the Internation-
al Working Group on Mild Cognitive Impairment
[28] . The cog-
nitively healthy controls were recruited among spouses and
friends of patients of the 3 memory clinics. They had normal cog-
nitive performance according to the MMSE, were independent in
their activities of daily living and did not have any memory com-
pl ain ts. Subje cts with s eri ous me dic al, p sychia tri c or ne uro log ica l
disorders which could affect cognitive functioning (e.g. major
depression, schizophrenia, seizure disorder, head injury) or with
a score on the MMSE of ! 28 were excluded.
The study protocol was approved by the Ethics Committee of
the Medical Faculty of the Friedrich-Alexander-Universität Er-
langen-Nürnberg, Germany.
S t a t i s t i c a l A n a l y s e s
Statistical analyses were implemented in PASW Statistics 17.0
for Windows (SPSS, Chicago, Ill., USA); p values of less than 0.05
were considered to indicate statistical significance.
Differences with regard to demographic variables, MMSE
scores and component and total scores of the ACE-R among cog-
nitively healthy controls and patients with MCI, AD and FTLD
were tested using one-way analysis of variance. Pairwise compar-
isons were performed using Bonferroni’s test. 2
tests were em-
ployed for nominal (categorical) data. If differences attained sta-
tistical significance, a linear regression analysis was carried out
to investigate possible associations between the demographic
variable which varied significantly across the groups and the par-
ticipants’ performance in the ACE-R.
The first step of the ana lysis of the utilit y of the ACE-R and the
MMSE encompassed a logistic regression analysis, assessing the
extent to which ACE-R and MMSE scores predict the clinical di-
agnosis. ACE-R and MMSE scores were fed as explanatory vari-
ables and the clinical diagnosis as outcome variable. If the utility
analysis revealed significantly higher accuracy of the ACE-R in
discriminating between cognitively healthy controls and patients
in comparison with the MMSE, a logistic regression analysis was
carried out to assess which ACE-R component tests contributed
to the higher accuracy of the instrument. The second step com-
prised the performance of a receiver operating characteristic
(ROC) curve analysis. The area under the curve (AUC) was used
as a measure of the accuracy of the ACE-R and MMSE to distin-
guish between the groups of participants. AUC values of less than
1.0 (perfect test) refer to excellent ( 1 0.9), good ( 1 0.8), fair ( 1 0.7)
and poor ( 1 0.6) accuracy
[29] . Differences between AUCs were
assessed with the StAR software for the statistical comparison
of ROC curves
[30] . The ROC was also used to select an optimal
cut-off value below which an individual has a very high chance
of suffering from the aforementioned clinical syndromes and
disorders.
Validation of the German Revised
Addenbrooke’s Cognitive Examination
Dement Geriatr Cogn Disord 2010;29:448–456
451
R e s u l t s
The present study refers to 56 patients with mild de-
mentia in AD, 75 patients with MCI, 22 patients with
mild dementia in FTLD (14 with the frontal variant of
FTLD and 8 with the language variant) and 76 cognitive-
ly healthy controls. Demographic data and MMSE and
ACE-R total and component scores of the four groups of
individuals are presented in table1 . There were no statis-
tical differences regarding age and years of education be-
tween the control group and the patient groups. The MCI
group included significantly fewer female participants
than the healthy controls (Fisher’s exact test, p = 0.015).
The linear regression analysis, using the ACE-R scores as
dependent variable and diagnosis (MCI vs. no cognitive
deficits) and gender distribution as independent factors
(F = 28,523, p ! 0.0001), did not reveal a significant im-
pact of gender distribution on ACE-R performance (stan-
dardized partial regression coefficient of clinical diagno-
sis = –0.03, p = 0.72), whereas the impact of the clinical
diagnosis was significant (standardized partial regres-
sion coefficient of clinical diagnosis = –0.523, p ! 0.0001).
No significant differences were detected between the
FTLD and the AD groups with regard to age, education
and gender.
Concerning the performance on ACE-R component
tests, patients with MCI had significantly lower scores on
all ACE-R component tests than cognitively healthy con-
trols except for attention/orientation and visuospatial
ability (memory p ! 0.0001, verbal fluency p ! 0.0001,
language p = 0.016). They also had lower total scores on
MMSE and on ACE-R in comparison with healthy con-
trols (p ! 0.0001 for both tests). Patients with AD and
FTLD performed worse than healthy controls on all
ACE-R component tests, as well as on the MMSE and the
ACE-R total score (all p ! 0.0001). Regarding differences
in the performance between AD and FTLD patients, the
former scored higher on the verbal fluency ACE-R com-
ponent test (p = 0.044) and lower on ACE-R memory
tasks (p = 0.008) compared to the latter. However, there
were no differences between the two groups concerning
the ACE-R total score and the rest of the cognitive do-
mains assessed with the ACE-R.
Distinction between Patients with MCI and
Cognitively Healthy Controls
The logistic regression analysis (likelihood ratio 2
=
73.74, p ! 0.0001) with clinical diagnosis (MCI vs. healthy
controls) as dependent variable and ACE-R (regression
coefficient = –0.16, p ! 0. 00 01) and MMSE scor es (regre s-
sion coefficient = –0.87, p ! 0.0001) as independent fac-
tors revealed that both instruments can significantly con-
tribute to the discrimination between MCI and cogni-
tively healthy controls. The results of the ROC analyses,
which showed that the AUC of the ACE-R was slightly
and not significantly (p 1 0.05) larger than that of the
MMSE, are displayed in figure 1 and table2 .
Distinction between Patients with Mild Dementia in
AD and Cognitively Healthy Controls
The ACE-R contributed and the MMSE strongly tend-
ed to contribute significantly to the discrimination be-
Table 1. Description of study sample, and component and composite mean scores 8 SD on the ACE-R and on the MMSE
Group variable Cognitively
healthy controls
MCI Mild dementia
in AD
Mild dementia
in FTLD
Number of subjects 76 75 56 22
Age, years 69.64 8 7.53 67.83 8 8.01 72.00 8 8.18 69.64 8 6.18
Gender:female ratio, % 61.5 40.0
a 64.2 40.9
Education, years 11.78 8 2.51 12.00 8 3.27 11.02 8 2.63 11.70 8 3.52
MMSE score 29.09 8 0.73 27.29 8 1.82
a 23.21 8 3.25
a
23.45 8 5.54
a
ACE-R, total score 90.37 8 4.99 81.34 8 9.09
a
64.80 8 11.32
a
64.50 8 17.82
a
Orientation and attention ACE-R component test 17.95 8 0.23 17.31 8 0.97 14.30 8 3.07
a 14.91 8 3.74
a
Memory ACE-R component test 22.45 8 2.70 18.04 8 4.75
a
10.46 8 5.26
a 14.14 8 6.24
a, b
Verbal fluency ACE-R component test 9.82 8 2.06 7.99 8 2.56
a
6.21 8 3.60
a 4.36 8 2.68
a, b
Language ACE-R component test 25.13 8 1.25 23.53 8 2.30
a
21.23 8 4.24
a 19.18 8 6.43
a
Visuospatial ability ACE-R component test 15.05 8 1.53 14.65 8 1.75 12.59 8 2.98
a 11.41 8 3.50
a
a Significant difference from cognitively healthy controls, p < 0.05.
b
Significant difference from AD patients, p < 0.05.
Alexopoulos et al.
Dement Geriatr Cogn Disord 2010;29:448–456
452
tween AD and cognitively healthy controls according to
the logistic regression analysis (likelihood ratio 2
=
167.68, p ! 0.0001) with clinical diagnosis (AD vs. healthy
controls) as dependent variable and ACE-R (regression
coefficient = –0.44, p = 0.02) and MMSE scores (regres-
sion coefficient = –6.48, p = 0.05) as independent factors.
The AUCs did not differ according to the ROC analyses
( table2 ; fig.2 ).
Distinction between Patients with Mild Dementia in
FTLD and Cognitively Healthy Controls
The logistic regression analysis (likelihood ratio 2
=
72.28, p ! 0.0001) with clinical diagnosis (FTLD vs.
healthy controls) as dependent variable and ACE-R (re-
gression coefficient = –0.28, p = 0.01) and MMSE scores
(regression coefficient = –0.85, p = 0.19) as independent
factors revealed that only the ACE-R significantly con-
tributes to the discrimination between FTLD and cogni-
tively healthy controls. Due to this significant difference
between the accuracy of the two instruments, a further
logistic regression analysis (likelihood ratio 2
= 80.8,
p ! 0.0001) was carried out to assess which ACE-R com-
ponent tests contribute significantly to the discrimina-
tion between FTLD and cognitively healthy controls. The
ACE-R component tests and the clinica l diagnosis (FTLD
0
0.2
0.4
0.6
0.8
1.0
Sensitivity
MMSE
ACE-R
0 0.2 0.4 0.6 0.8 1.0
1 – specificity
0
0.2
0.4
0.6
0.8
1.0
Sensitivity
0 0.2 0.4 0.6 0.8 1.0
1 – specificity
MMSE
ACE-R
F i g . 1 . ACE-R and MMSE ROC for the detection of patients with
MCI.
F i g . 2 . ACE-R and MMSE ROC for the detection of patients with
mild dementia in AD.
Table 2. Optimal cut-off scores and diagnostic utility of the ACE-
R and the MMSE for identifying MCI, mild dementia in AD and
mild dementia in FTLD
ACE-R MMSE
Distinction between patients with MCI and cognitively healthy
controls
Optimal cut-off score 86/87 28/29
Sensitivity 0.82 0.78
Specificity 0.68 0.73
Area under the curve 0.83 0.81
Distinction between patients with mild dementia in AD and
cognitively healthy controls
Optimal cut-off score 82/83 27/28
Sensitivity 0.92 1.00
Specificity 0.96 0.93
Area under the curve 0.99 0.99
Distinction between patients with mild dementia in FTLD and
cognitively healthy controls
Optimal cut-off score 83/84 27/28
Sensitivity 0.88 0.73
Specificity 0.96 0.97
Area under the curve 0.97 0.92
Validation of the German Revised
Addenbrooke’s Cognitive Examination
Dement Geriatr Cogn Disord 2010;29:448–456
453
vs. healthy controls) were entered into the model as inde-
pendent and dependent variables, respectively. The ACE-
R verbal fluency component test (regression coefficient =
–0.73, p = 0.01) and the language component test (regres-
sion coefficient = –0.92, p = 0.02) were significantly as-
sociated with the clinical diagnosis, whereas the ACE-R
memory component test (regression coefficient = –0.24,
p = 0.17), the visuospatial component test (regression co-
efficient = 0.38, p = 0.37) and the attention-orientation
component test (regression coefficient = –3.31, p = 0.07)
were not. The AUC of the ACE-R was larger than that of
the MMSE (0.97 vs. 0.92), though the difference did not
attain statistical significance (p 1 0.05) ( table2 ; fig.3 ).
Regarding the utility of the ACE-R in discriminating
between healthy controls and patients suffering from the
frontal variant of FTLD, the results of the regression and
ROC analyses did not deviate from the aforementioned
findings concerning all FTLD patients. In short, the pa-
tients with the frontal variant of FTLD did not differ from
healthy controls in demographic characteristics. Only
the ACE-R (regression coefficient = –0.27, p = 0.02) pre-
dicted the clinical diagnosis with highly significant ac-
curacy and not the MMSE (regression coefficient = –0.77,
p = 0.28) according to the regression analysis (likelihood
ratio 2
= 54.27, p ! 0.0001). The ROC also revealed a su-
periority of the ACE-R to MMSE in discriminating be-
tween healthy controls and patients with the frontal vari-
ant of FTLD (AUC: 0.96 vs. 0.92). Such analyses were
not performed for patients with the language variant of
FTLD due to the limited subsample size.
Distinction between Patients with Mild Dementia in
FTLD and Patients with Mild Dementia in AD
The memory to verbal fluency (M/VF) ratio was cal-
culated since the FTLD and AD groups differed signifi-
cantly with regard to memory and verbal f luency ACE-R
subtest scores. Its value was significantly higher in the
FTLD group than in the AD group [mean (SD): 4.13
(2.77) vs. 2.18 (2.07), t = –2.87, p = 0.01]. The logistic re-
gression analysis (likelihood ratio 2
= 9.91, p = 0.02) with
clinical diagnosis (FTLD vs. AD) as dependent variable
and ACE-R M/VF ratio (regression coefficient = 0.35, p =
0.01), ACE-R total score (regression coefficient = 0.03,
p = 0.35) and MMSE score (regression coefficient = –0.06,
p = 0.61) as independent factors revealed that only the
ACE-R M/VF ratio is a significant predictor of the clini-
cal diagnosis. According to the ROC, the ACE-R M/VF
ratio (AUC = 0.77, optimal cut-off score = 2.30/2.31, sen-
sitivity = 0.70, specificity = 0.76) was shown to be more
useful in the distinction between FTLD and AD with fair
0
0.2
0.4
0.6
0.8
1.0
Sensitivity
0 0.2 0.4 0.6 0.8 1.0
1 – specificity
MMSE
ACE-R
0 0.2 0.4 0.6 0.8 1.0
1 – specificity
0
0.2
0.4
0.6
0.8
1.0
Sensitivity
MMSE
ACR-R
ACR-R M/VF ratio
F i g . 3 . ACE-R and MMSE ROC for the detection of patients with
mild dementia in FTLD.
F i g . 4 . ACE-R M/VF ratio, ACE-R and MMSE ROC for the dis-
tinction between patients with mild dementia in FTLD and pa-
tients with mild dementia in AD.
Alexopoulos et al.
Dement Geriatr Cogn Disord 2010;29:448–456
454
accuracy than the ACE-R total score (AUC = 0.60) or the
MMSE (AUC = 0.64) ( fig.4 ), the accuracy of which was
poor. The AUCs of the MMSE and the ACE-R total score
were not significantly different from 0.5 (p 1 0.05), a val-
ue indicating that the results are not better than chance
at predicting the presence of FTLD or AD.
Discussion
The role of neurocognitive examination is central in
the diagnostic workup of cognitive impairment in ageing.
The present study was performed to evaluate the accu-
racy of the German version of the ACE-R in the distinc-
tion between MCI, mild dementia in AD, FTLD, and cog-
nition in normal ageing.
The ACE-R was not found to be more accurate than
the MMSE in distinguishing patients with MCI from
cognitively healthy controls. The high values of the AUCs
for both tests (0.83 for the ACE-R and 0.81 for the MMSE)
and the significant association between ACE-R and
MMSE scores (both p ! 0.0001) with the clinical diagno-
sis according to the regression analysis, implying a good
accuracy, should be interpreted with caution, since the
MMSE has not been proven to be effective in identifying
mild cognitive deficits
[15 , 16] . These findings possibly
reflect a common problem in the recruitment of elderly
control subjects. Such volunteers typically represent a se-
lected group of very healthy and highly functioning el-
derly, differing from the general elderly population, and
especially from the elderly individuals referred for de-
mentia evaluation
[10 , 31–33] . Studies of distinct cogni-
tively impaired and cognitively normal samples maxi-
mize test performance characteristics
[34] . Thus, from
the clinician’s perspective, emphasis should be placed on
the absence of ACE-R superiority to the MMSE in distin-
guishing healthy controls from patients with MCI.
According to our findings, both the German ACE-R
and the MMSE have very high accuracy in detecting mild
dementia in AD according to the ROC analysis (both
AUCs = 0.99). These extremely high levels of accuracy
should, however, be treated with caution, owing to the
findings of a recent review study which did not point out
such a high utility of the MMSE in detecting patients with
dementia in AD
[35] . The ACE-R scores were significant-
ly associated with the clinical diagnosis according to the
logistic regression analysis (p = 0.02). Since the ACE-R
encompasses features of learning materials over a series
of trials and a delayed recall component, it was expected
to be superior to t he MMSE in identi fying mild dementia
in AD. Nevertheless, the shorter MMSE with a p value of
0.05 was shown to tend to predict the clinical diagnosis
with high accuracy and to be equally effective in the de-
tection of patients with mild dementia in AD as the lon-
ger ACE-R according to the ROC analysis. Our findings
are in line with the observation that no brief cognitive test
has yet been found to be clinically superior over ot hers in
identifying AD
[4] . In the light of the current short con-
sultation times, the equal effectiveness of the two instru-
ments to detect mild dementia in AD may act as a disin-
centive to clinical usage of the ACE-R.
ACE-R was found superior to MMSE in detecting mild
dementia in FTLD according to the regression and ROC
analyses. Though the AUCs of the two instruments dif-
fered (AUC = 0.97 vs. 0.92), the difference did not reach
statistical significance. The absence of statistical signifi-
cance can be explained by the relatively limited sample
size of the FTLD group. The superiority of the ACE-R can
be attributed to its component tests that assess cognitive
domains affected by FTLD (i.e. executive function, verbal
fluency)
[36] but are not covered by the MMSE. The re-
gression analysis revealed a highly significant association
between the clinical diagnosis and the scores of the verbal
fluency ACE-R component test (p = 0.01) and the lan-
guage component test (p = 0.02), which are not or only to
a limited extent assessed by the MMSE.
The ACE-R assesses a broad range of cognitive abili-
ties and provides a wide profile of cognitive functions/
dysfunctions. It contributes to drawing a differentiated
picture of cognitive deficits with the objective of support-
ing diagnosis and differential diagnosis. In the present
study, the comparison of the AD and the FTLD groups
on the ACE-R component tests revealed differences in
memory and verbal fluency performance. AD patients
performed better in verbal f luency tasks, whereas the
FTLD patients’ performance in memory tasks was sig-
nificantly higher. These observations are concordant
with the discrepancies of the profiles of cognitive impair-
ment in these two clinical entities that are highlighted in
the literature. For similar levels of overall cognitive de-
cl ine , FTL D patients h ave a rela tive prese rvat ion of mem-
ory and greater deficits in frontal functioning than AD
patients
[36 , 37] . Calculating the ACE-R M/VL ratio fur-
ther supports with fair accuracy (AUC = 0.77) the differ-
entiation between AD and FTLD.
The German ACE-R was found to detect mild demen-
tia in FTLD more effectively and to differentiate, with the
ACE-R M/VF ratio, between FTLD and AD patients sig-
nificantly better than the MMSE, whilst the accuracy of
the two instruments in detecting MCI and mild dementia
Validation of the German Revised
Addenbrooke’s Cognitive Examination
Dement Geriatr Cogn Disord 2010;29:448–456
455
References
1 Minati L, Edginton T, Bruzzone MG, Giac -
cone G: Current concepts in Alzheimer’s dis-
ease: a multidisciplinary review. Am J Alz -
heimers Dis Other Demen 2009;
24: 95–121.
2 Chapman DP, Williams SM, Strine TW,
Anda R F, Moore MJ: Dementi a and its impli-
cations for public health. Prev Chronic Dis
2006;
3:A34.
3 Harvey RJ, Skelton-Robinson M, Rossor
MN: The prevalence and causes of dementia
in people under the age of 65 years. J Neurol
Neurosurg Psychiatry 2003;
74: 1206–1209.
4 Feldman HH, Jacova C, Robillard A, Garcia
A, Chow T, B orr ie M, Schi pper HM, B lai r M,
Kertesz A , Chertkow H: Diagno sis and treat-
men t of de mentia . 2. Dia gnos is. C MAJ 200 8;
178: 825–836.
5 Shulman KI: Clock-drawing: is it t he ideal
cognitive screening test? Int J Geriatr Psy-
chiatry 2000;
15: 548–561.
6 Jones TG, Schinka JA, Vanderploeg RD,
Small BJ, Gr aves AB, Mortime r JA: 3MS nor-
ma tiv e da ta f or t he e lde rly. A rc h Cl in N eur o-
psychol 2002;
17: 171–177.
7 Ihl R, Grass-Kapanke B, Lahrem P, Brink-
meyer J, Fischer S, Gaab N, Kaupmannsen-
necke C: Development and validation of
a test for early diagnosis of dementia with
differentiation from depression (TFDD).
Fortsch r Neurol Psychiatr 20 00;
68: 413–422.
8 Kalbe E, Kessler J, Calabrese P, Smith R,
Passmore AP, Brand M, Bul lock R: Demtect:
a new, sensitive cognitive screening test to
support the diagnosis of mild cognitive im-
pairment and early dementia. Int J Geriatr
Psychiatry 2004;
19: 136–143.
9 Perry RJ, Hodges JR: Relationship bet ween
functional and neuropsychological perfor-
mance in early Alzheimer disease. Alzhei -
mer Dis Assoc Disord 2000;
14: 1–10.
10 Alexopoulos P, Nadler K, Cramer B, Her-
pertz SC, Kurz A: Validation of a short test
(3MS-R) for detecting Alzheimer’s disease.
Fortschr Neurol Psychiatr 2007;
75: 728–736.
11 Folstein M F, Folstein SE , McHugh PR: Mini-
mental state. A practical met hod for grading
the cognitive state of patients for the clini-
cian. J Psychiatr Res 1975;
12: 189–198.
12 Mohs RC, Rosen WG, Davis KL: The Alz -
heimer’s disease assessment scale: an instru-
ment for assessing treat ment eff icacy. Psy-
chopharmacol Bull 1983;
19: 448–450.
13 Rosen WG, Mohs RC, Davis KL: A new rat-
ing scale for Alzheimer’s disease. Am J Psy-
chiatry 1984;
141: 1356–1364.
14 Burkart M, Heun R, Maier W, Benkert O:
Dementia screening in routine clinical prac-
tice. A comparative analysis of MMSE,
SIDAM and ADAS. Ner venarzt 1998;
69:
983–990.
in AD did not differ. Since ACE-R takes around 15 min
to perform, whereas administering the MMSE lasts only
approximately 8 min
[38] , the thorny question is posed
about the clinical and diagnostic concept in terms of
which the ACE-R could be most appropriately deployed.
Measuring a broad range of cognitive abilities and com-
prising tasks and questions of variable difficulty, the
ACE-R provides an objective and differentiated global
profile of cognitive performance. This could serve in spe-
cialist centres/clinics as an invaluable instrument in a
comprehensive diagnostic workup which also encom-
passes clinical, behavioural and functional features sup-
plemented by an informant report and neuroimaging
[39] .
Our study has several limitations. The participants
were recruited at university centres. Applicability and re-
liability of the ACE-R in community samples require fur-
ther investigation. The evaluation was confined to pa-
tients with MCI, AD and FTLD. Hence, we were not in
the position to assess the effectiveness of the ACE-R in
detecting other forms of dementia, such as dementia with
Lewy bodies or vascular dementia. The reproducibility
and repeatability of the ACE-R were not evaluated in this
study, but as the ACE-R assesses cognitive functions in a
well-standardized and objective manner, the rater-relat-
ed bias is likely to be low. The statistics of positive and
negative predictive values were not calculated, because
the groups of patients and healthy controls were recruited
independently and not from a population, as well as due
to the fact that these va lues cannot be compared between
different diagnostic tests such as the ACE-R and the
MMSE
[40] . Furthermore, the clinical diagnosis, which
was based on a comprehensive diagnostic procedure and
on international diagnostic criteria, was used as the ulti-
mate gold standard. In spite of the high validity of the
diagnostic criteria, the clinical diagnoses are not always
confirmed at autopsy
[41] . Thus, the possibly erroneous
clinical assessments should also be taken into account.
The validity of the ACE-R could therefore be lower than
our results suggest.
The diagnosis and differential diagnosis of dementia
remain a clinically integrated process, with neurocogni-
tive testing providing only supportive evidence. The
ACE-R, a brief and easily administered bedside test, as-
sesses a broad range of cognitive abilities, provides a dif-
ferentiated objective picture of cognitive impairment and
detec ts patients wit h mi ld dementia i n FT LD w ith higher
accuracy than the common MMSE. It could emerge as a
valid screening instrument that supports clinical and dif-
ferential diagnosis of dementia, in terms of the clinical
and diagnostic concept of specialist centres/clinics.
Acknowledgment
We thank Dr. Eneida Mioshi (Department of Clinical Neuro-
science, University of Cambridge, Cambridge, UK, and Prince of
Wales Medical Research Institute, Sydney, N.S.W., Australia) for
her assistance in the ACE-R translation into German.
Alexopoulos et al.
Dement Geriatr Cogn Disord 2010;29:448–456
456
15 B ous ta ni M, P ete rso n B, Ha nso n L , Ha rr is R ,
Lohr K N: Screening for dementia i n primary
care: a summar y of the ev idence for the US.
Preventive Services Task Force. Ann Intern
Med 2003;
138: 927–937.
16 Kurz A, Diehl J, Riemenschneider M, Per-
neczky R, Lautenschlager N: Mild cognitive
disorder. Questions of def inition, diagnosis,
prognosis and therapy. Nervenarzt 2004;
75:
6–15.
17 Mathuranath PS, Nestor PJ, Berrios GE, Ra-
kowicz W, Hodges JR: A brief cognitive test
battery to differentiate Alzheimer’s disease
and frontotemporal dementia. Neurology
2000;
55: 1613–1620 .
18 Mioshi E, Dawson K, Mitchell J, Arnold R,
Hodges JR : The Addenbrooke’s cognitive ex-
amin ation revised (ACE-R): a brief cog nitive
test battery for dementia screening. Int J
Geriatr Psychiatry 2006;
21: 1078–1085.
19 Alexopoulos P, Greim B, Kurz A: Brief
asses sment of cogniti ve mental s tatus in ger-
man: the Addenbrooke’s Cognitive Exami-
nation (ACE) and the Addenbrooke’s
Cognitive Examination Revised (ACE-R).
Eur J Geriatr 2007;
9175–180.
20 Thalmann B, Mönsch AU: The Consortium
to Establish a Registry for A lzheimer’s Dis-
ease. Neuropsychologische Testbatterie. Ba-
sel, Memory Clinic Basel, 1997.
21 Reitan RM, Wolfson D: Validit y of the Trail
Making Test as an ind icator of organic brain
damage. Percept Mot Ski lls 1985;
8: 271–276.
22 Hindmarch I, Lehfeld H, de Jongh P, Erzig-
keit H: The Bayer Activities of Daily Living
Scale (B -ADL). Dement Geri atr Cogn Disord
1998;
9(suppl 2):20–26.
23 Cumm ings JL, Mega M, Gray K, Rosenberg-
Thompson S, Carusi DA, Gornbein J: The
neuropsychiatric inventory: comprehensive
assessment of psychopatholog y in dementia.
Neurology 1994;
44: 2308–2314.
24 Dilling H, Mombour W, Schmidt MH,
Schulte-Markwort E , Weltgesundheitsorga -
nisation (eds): Internationale Klassifikation
psychischer Störungen. ICD-10 Kapitel V
(F). Klinisch-diagnostische Leitlinien. Bern,
Huber, 1994.
25 McKhann G, Drachman D, Folstein M,
Katzma n R, Price D, Stad lan EM: Cli nical di-
agnosis of Alzheimer’s disease: report of the
NINCDS-ADRDA work group under the
au spices of d epa rtm ent of hea lth a nd hu man
services task force on Alzheimer’s disease.
Neurology 1984;
34: 939–344.
26 Neary D, Snowden JS, Gustafson L, Passant
U, Stuss D, Black S, Freedman M, Kertesz A,
Robert PH, Albert M, Boone K, Miller BL,
Cummings J, Benson DF: Frontotemporal
lobar degeneration: a consensus on clinical
diagnostic criteria. Neurology 1998;
51:
1546–1554.
27 Kraemer HC, Taylor JL, Tink lenberg JR, Ye-
savage JA: The stages of Alzheimer’s disease:
a reappraisal. Dement Geriatr Cogn Disord
1998;
9: 299–308.
28 Winblad B, Palmer K, Kivipelto M, Jelic V,
Fratiglioni L, Wahlund LO, Nordberg A,
Backman L, Albert M, Almkvist O, Arai H,
Basun H, Blennow K, de Leon M, DeCarli C,
Erkinjuntti T, Giacobini E, Gra ff C, Hardy J,
Jack C, Jorm A , Ritchie K, van Du ijn C, Viss-
er P, Petersen RC: Mild cognitive impair-
ment – Beyond controversie s, towards a con-
sensus: report of the internationa l working
group on mild cognitive impairment. J In-
tern Med 2004;
256: 240–246.
29 Perneczky R, Pohl C, Sorg C, Hartmann J,
Komossa K, Alexopoulos P, Wagenpfeil S,
Kurz A: Complex activities of daily living in
mild cognitive impairment: conceptual and
diagnostic issues. Age Ageing 2006;
35: 240–
245.
30 Vergara IA, Nora mbuena T, Ferrada E, Slater
AW, Melo F: StAR: a simple tool for the sta-
tistical comparison of ROC curves. BMC
Bioinformatics 2008;
9: 265.
31 Stokholm J, Vogel A, Johannsen P, Walde-
mar G: Validation of the Da nish Adden-
brooke’s cognitive examination as a screen-
ing test in a memory clinic. Dement Geriatr
Cogn Disord 2009;
27: 361–365.
32 Alexopoulos P, Greim B, Nadler K, Martens
U, Krecklow B, Domes G, Herpertz S, Kurz
A: Validation of the Addenbrooke’s cogni-
tive examination for detecting early Alz -
heimer’s disease and mild vascular dementia
in a German population. Dement Geriatr
Cogn Disord 2006;
22: 385–391.
33 Alexopoulos P, Perneczky R, Cramer B,
Grimmer T, Kurz A: Validation of a short
telephone test (T3MS) for the diagnosis of
cognitive impairment. Fortschr Neurol Psy-
chiatr 2006;
74: 329–336.
34 Brodat y H, Low LF, Gibson L, Bur ns K: What
is the best dementia screening instrument
for general pra ctitioners to use? A m J Geriatr
Psychiatry 2006;
14: 391–400.
35 Ho ls in ger T, D ev eau J, Bou st a ni M , W il li a ms
JW Jr: Does this patient have dementia?
JAMA 2007;
297: 2391–2404.
36 Diehl J, Kurz A: Frontotemporal dementia:
patient characteristics, cognition, and be-
haviour. Int J Geriatr Psychiatry 2002;
17:
914–918.
37 Bozoki AC, Farooq MU: Frontotemporal lo-
bar degeneration insights from neuropsy-
chology and neuroimaging. Int Rev Neuro-
biol 2009;
84: 185–213.
38 Tangalos EG, Smith GE, Ivnik RJ, Petersen
RC, Kokmen E, Kurland LT, Offord KP, Pa-
risi JE: The Mini-Mental State Examination
in general medical practice: clinical utilit y
and acceptance. Mayo Clin Proc 1996;
71:
829–837.
39 Woodford HJ, Geor ge J: Addenbrooke’s cogni-
tive examination-revised in day-to-day clini-
cal practice. Age Ageing 2008;
37: 350 –351.
40 Peat J, Barton B: Medical Statistics, ed 2.
Malden, Blackwell, 2006.
41 Först l H: Primär kort ikale Demenzen. Stutt-
gart, Thieme, 2000.