Early Detection and Differential Diagnosis of

Original Research Article
Dement Geriatr Cogn Disord 2001;12:265280
Accepted: November 8, 2000
Early Detection and Differential Diagnosis of

Alzheimers Disease and Depression with
Neuropsychological Tasks
R. Swainson a J.R. Hodges b C.J. Galton b J. Semple a, d A. Michael e
B.D. Dunn a J.L. Iddon f T.W. Robbins c B.J. Sahakian a
Departments of a Psychiatry, b Neurology and c Experimental Psychology, University of Cambridge,
Beecham Ltd., Addenbrookes Centre for Clinical Investigation, Cambridge, e West Suffolk Hospital,
Bury St. Edmunds, and f CeNeS Pharmaceuticals PLC, Histon, Cambridge, UK
d SmithKline
Key Words
Early detection W Alzheimers disease W Depression W
Neuropsychological assessment W Differential
diagnosis W Neural and pharmacotherapeutic
implications
Abstract
The development of novel treatments for Alzheimers
disease (AD), aimed at ameliorating symptoms and
modifying disease processes, increases the need for early diagnosis. Neuropsychological deficits such as poor
episodic memory are a consistent feature of early-in-thecourse AD, but they overlap with the cognitive impairments in other disorders such as depression, making differential diagnosis difficult. Computerised and traditional tests of memory, attention and executive function
were given to four subject groups: mild AD (n = 26); questionable dementia (QD; n = 43); major depression (n = 37)
and healthy controls (n = 39). A visuo-spatial associative
learning test accurately distinguished AD from depressed/control subjects and revealed an apparent subgroup of QD patients who performed like AD patients.
QD patients performance correlated with the degree of
subsequent global cognitive decline. Elements of contextual and cued recall may account for the tasks sensitivity and specificity for AD.
Introduction
The last 25 years have witnessed a revolution in our

understanding of Alzheimers disease (AD). The recognition that the disease affects primarily, though not exclusively, cholinergic neurones led to the development of the
so-called cholinergic hypothesis and hence to the introduction of cholinesterase inhibitors such as tetrahydroaminoacridine, donepezil and rivastigmine to provide
cognitive enhancement in at least some patients [14]. In
parallel, the application of the techniques of molecular
biology has greatly enhanced our knowledge of the molecular basis of pathology, and there is a real prospect of the
development of interventions aimed at interfering with
the destructive molecular cascade and hence the time
course of disease progression [5, 6]. Irrespective of whether the goals of treatment are directed at ameliorating
symptoms or modifying the disease process, it makes
sense to commence treatment at the earliest stages so that
patients may derive maximal benefit, i.e. can maintain
the highest possible level of function for as long as possible. It should be noted that, for reasons of cost effectiveness and side effects, excluding patients from treatment
for whom it would not be helpful is also necessary. A great
deal of work is in progress to try to find ways of improving
early diagnosis of AD with methods including structural
and functional imaging and ApoE genotyping [79]. A
Copyright 2001 S. Karger AG, Basel
ABC
2001 S. Karger AG, Basel

14208008/01/01240265$17.50/0
Fax + 41 61 306 12 34
E-Mail karger@karger.ch
www.karger.com
Accessible online at:

www.karger.com/journals/dem
Dr. B.J. Sahakian

Department of Psychiatry, University of Cambridge
Box 189, Addenbrookes Hospital, Hills Road
Cambridge CB2 2QQ (UK), Tel. +44 1223 331209
Fax +44 1223 336968, E-Mail jenny.hall@msexc.addenbrookes.anglox.nhs.uk
neuropsychological task focussing on the cognitive difficulties specific to AD may be able to provide a relatively
inexpensive and easily administered method of detection
(and may of course be given in combination with these
other methods to increase accuracy further).
A neuropsychological approach to detection would be
predicated on there being testable deficits in cognition
which are consistent across individuals with the same
neuropathology and distinct from deficits caused by other
neurological or psychiatric conditions. AD pathology, being restricted in the early stages to parts of the medial temporal lobe, may well provide the potential for the development of such neuropsychological tests. The earliest neuropathological changes are seen in the transentorhinal cortex, with the accumulation of neurofibrillary tangles and
neuropil threads [10]. Symptoms are probably not clinically detectable at this stage and indeed may not be
noticed by the patients or their family. When the neuropathology involves the hippocampus and entorhinal cortex,
cognitive impairments become clinically detectable and
at first (probably before the neocortex itself becomes
affected) appear to affect only memory [11, 12]. Patients
at this stage have been referred to as having mild cognitive impairment [13], questionable dementia (QD) [14,
15] or minimal AD [16, 17] and many will qualify for the
probable AD diagnosis (NINCDS-ADRDA) [18] within
a few years, the conversion rate being roughly 1012% per
year compared with 1% per year in the general population
[13]. The capability to diagnose AD at the memoryimpaired stage would potentially be of significant benefit
to these patients. Testing the functioning of the hippocampus and entorhinal cortex may well form the basis of
such a diagnostic strategy, and indeed episodic memory
(memory for unique events which the individual has
experienced, in which the hippocampus is believed to
play a crucial role) is known to be particularly vulnerable
in AD [1924]. It is also important that patients without
AD are not falsely detected by such a test. One of the
most difficult disorders to differentially diagnose from
AD is depression, and contributing to this is the fact that
some episodic memory tests which are very sensitive to
AD particularly effortful tasks such as free recall are
also vulnerable in depression [25, 26]. A range of cognitive functions other than episodic memory, including
semantic memory (knowledge- or fact-based memory),
attention and executive function (high-level control processes), is significantly impaired in the mild stages of AD
[16, 27, 28], and indeed, tests of many different cognitive
capacities have been shown to be predictive of future AD
diagnosis among memory-impaired subjects [13, 29]. It
266
may be that a test of one, or a combination, of these functions would be sufficiently sensitive to the particular neuropathology in AD, as well as being suitably specific (with
regard to excluding conditions such as depression which
are difficult to differentiate clinically from early AD) to be
of diagnostic value. This is the issue which the present
study has sought to address.
In this study, a group of memory-impaired subjects are
being tracked over 2 years with neuropsychological testing
at 8-monthly intervals, the intention ultimately being to
determine whether their clinical outcome could have been
predicted at an earlier time from the results of neuropsychological testing. Three other groups of subjects are also
undergoing repeated testing; these groups will show how
subjects who fall clearly into the categories of mild probable AD, major depression and healthy elderly can be
expected to perform on the same tests over time. This first
report from the study shows how all four groups performed on a wide range of traditional and computerised
tasks when first tested (baseline performance). It also
shows the degree of accuracy with which scores on the various tests are able to classify subjects into diagnostic
groups and how well test performance at baseline predicted the degree of global cognitive decline over the subsequent 8 months. A wide range of tests is being used in
the study, including measures of episodic and semantic
memory, attention and executive function. Many of the
tests included e.g. Mini-Mental State Examination
(MMSE) [30], Alzheimers Disease Assessment Scale
(cognitive subscale) [31], Warringtons Short Recognition
Memory Tests [32] and the Wechsler Memory Scale logical memory recall [33] are used in standard clinical
practice, to assess levels of global, and more specific, cognitive ability. Also included are a range of computerised
tests, mainly taken from the CANTAB (Cambridge Neuropsychological Test Automated Battery; CeNeS Pharmaceuticals PLC; fig. 1) [20, 34]. Many of the CANTAB tests
were developed from the kinds of tests used in lesion and
drug studies with animals, allowing the potential to be
able to relate the deficits of patients to the neuronal and
neurochemical abnormalities produced by the disease
process. Two CANTAB tests paired associates learning
(PAL) and delayed matching to sample (DMTS) have
been used previously in a longitudinal study by Fowler et
al. [14, 15]. That study showed that both tasks were more
sensitive to decline among QD patients than were traditional measures such as the WAIS-R and WMS-R, and
that an error score from PAL, tracked over 6 or 12
months, divided the QD patients into apparently distinct
sub-groups. The performance of approximately half of the
Swainson/Hodges/Galton/Semple/Michael/
Dunn/Iddon/Robbins/Sahakian
Fig. 1. The CANTAB paired associates learning task. 1, 2, 3, 6 or 8 patterns are
displayed one at a time in the available boxes (6 boxes in all but the 8-pattern stage).
Each pattern is then displayed centrally, and the subject has to touch the box in which
the pattern was previously seen. If all responses are correct, the test moves on to the
next stage; otherwise all patterns are redisplayed in their original locations, followed
by another recall phase. Ten presentations and recall phases may be given, following
which, if all patterns have not been placed correctly, the task terminates.
QD subjects on the PAL task worsened, whilst that of the

remaining subjects remained stable. These patterns
matched those of the probable AD and healthy control
subjects, respectively, and Fowler et al. [14, 15] hypothesised that worsening performance on PAL may be an early
indicator of AD. The present study includes both the PAL
and DMTS tasks and thus will serve as an important test
of their utility in the early and differential diagnosis of
AD.
Subjects and Methods

Subjects
Four subject groups were included: mild AD (n = 26); QD (n =
43); major depression (n = 37) and normal healthy controls (n = 39).
Consecutive referrals to the Memory Clinic at Addenbrookes Hospital, Cambridge, were approached and screened for suitability for
inclusion into the study; where necessary, other patients were
recruited from the Psychiatry Department of West Suffolk Hospital
and from the Mental Health Services of Addenbrookes NHS Trust.
Healthy controls were recruited from among spouses and friends of
patients, and through an advertisement placed in the Alzheimers
Disease Society newsletter. All subjects were aged between 50 and 80
years at time of recruitment and were screened for the following
exclusion criteria: extrapyramidal signs or hallucinations; vascular
dementias; current cancer treatment (radiotherapy or chemotherapy); uncontrolled diabetes; serious head injury requiring surgical
intervention. Patients having suffered cerebrovascular events (e.g.
TIA or stroke) or epilepsy were excluded at the discretion of a senior
neurologist (J.R.H.). Demographic and clinical data on the four
groups, as collected during the first testing visit (V1), are shown in
table 1.
In the clinic, many patients presenting with deficits in memory
function do not fulfil a neurological or psychiatric diagnosis. These
patients who therefore did not meet the diagnostic criteria for probable AD (NINCDS-ADRDA) [18] or major unipolar depression
Early Detection of AD
Table 1. Clinical and demographic characteristics of the patient and
control groups
AD
n
Age
NART
MMSE
GDS
Ham-D
QD
26
43
68.6 (8.3)
65.0 (9.1)
108.4 (10.1) 118.4 (7.2)
21.0 (3.0)
28.0 (2.4)
7.6 (4.7)
8.9 (5.2)
N/A
N/A
Dep
37
60.8 (8.6)
111.7 (9.5)
28.1 (1.6)
21.9 (5.8)
21.4 (6.2)
Con
39
64.6 (8.5)
119.2 (7.7)
29.4 (0.7)
4.1 (2.7)
N/A
Means (standard deviation). The MMSE provides a measure of

global cognitive function, with a score of 23 or below corresponding
to dementia. The Geriatric Depression Scale (GDS) provides a selfrated measure of depression for elderly subjects, and the Hamilton
Rating Scale for Depression (Ham-D) gives a clinician-rated measure
of symptom severity in depression.
Dep = Depression group; Con = control group.
(DSM-IV) [35] were allocated to the QD group. The range of

impairment in this group was deliberately wide, in order that it
would constitute a clinically representative sample from which
results could be extrapolated to other clinic environments. Included
within this group were, therefore, individuals with subjective complaints of memory loss yet showing normal performance on objective
tests, as well as those with substantial objectively defined deficits
restricted to memory impairment. Patients with memory impairment together with depressive symptoms of insufficient breadth and/
or severity to meet diagnostic criteria for major depression would
also have been allocated to the QD group. In a study of a patient
sample comparable to the QD group in the present study, Tierney et
al. [36] reported that some 24% went on to meet clinical diagnostic
criteria for probable AD within 2 years. The QD group is therefore
likely to contain a substantial proportion of prodromal AD cases,
i.e. patients in whom the pathological process is present yet is currently of insufficient duration or severity to produce symptoms qual-
267
Table 2. Clinical and neuropsychological tests

Reference Task description/psychological functions assessed
Measures analysed
Measures of clinical/global function

NART
MMSE
37
30
Pronunciation of irregular words

Screening test for dementia
Geriatric Depression Scale

ADAS-cog
63
31
Self-reported depression scale

Assessment scale of cognitive symptoms in AD
Predicted verbal IQ
Total score 030 (! 24 indicates
dementia)
Depressive responses/30
Total errors 070
32
2-choice recognition of visually presented words

(e.g. lake, clock) or photographs of male faces
(list of 25 for each)
2-choice recognition of abstract patterns (2 lists of 12)
2-choice recognition of locations of white boxes on
computer screen (4 lists of 5)
4-choice recognition of photographs of doors
(2 lists of 12)
4-choice recognition of abstract patterns sharing
colour or pattern with distractors (10 trials at each
delay: simultaneous, 0, 4, 12 s)
Neuropsychological tests of mnemonic function

Recognition
Warrington Short Recognition Memory Tests
(SRMT) for words and faces
CANTAB pattern recognition
CANTAB spatial recognition
20
20
Doors test from Doors and People tests
65
CANTAB DMTS
20
Cued and free recall

CANTAB PAL
20
Wechsler Logical Memory recall

Semantic naming and fluency
Graded naming test
33
66
New semantic battery naming test
67
Category fluency
68
Neuropsychological tests of attentional function

CANTAB 5-choice reaction time
CANTAB RVIP, 1-sequence and 3-sequence

versions
69
70
Neuropsychological tests of executive function

Stroop colour/word task
71, 72
Letter fluency
73
CANTAB ID/ED attentional set-shifting task
74
One-touch Tower of London planning task
75
Total correct (max. = 25)
Percentage correct
Percentage correct
Total correct (max. = 24)
Total correct at 12 s delay
(max. = 10)
Learning of 1 (2 stages), 2 (2 stages), 3 (2 stages),

6 (1 stage) and 8 (1 stage) visual pattern-location
pairings; 10 presentations available at each stage (fig. 1)
Free recall of 2 story passages immediately and after
30-min delay (25 items per story)
Stage reached (max. = 8);

Number of errors made at 6-pattern
stage (6-pattern errors)1
Total items recalled at 30-min delay
(max. = 50)
Naming of graded-difficulty line drawings

(e.g. kangaroo, centaur)
Naming of line drawings from 8 semantic categories
(e.g. vehicles sledge, train; fruit orange, cherry)
Spontaneous naming of items from 3 semantic
categories: animals, fruit, household items
Total items named (max. = 30)
Total items named in 180 s (60 s

per category)
Speeded response to appearance of visual stimulus,

simple (single location) and 5-choice (5 possible
locations) conditions
Detection of infrequent 3-digit sequences among
rapidly presented digit series (1-sequence, 4-min,
and 3-sequence, 5-min versions)
Accuracy (total hits; max = 25)

and latency of response in 5-choice
condition
A (signal detection measure of
accuracy) and latency of response
in 1-sequence version
Inhibition of automatic word reading during

conflicting ink colour-naming task (e.g. red written
in blue ink, response = blue)
Spontaneous naming of items beginning with a
particular letter: F, A, S
Discrimination learning, involving selective attention
to, and set-shifting between, shape and line stimulus
dimensions
Planning of sequence of moves to achieve a goal
arrangement of coloured balls; difficulty range from
1 to 5 move problems
Interference index (scored as in

Fisher et al., 1990; large negative
score = poor inhibition)
Total items named in 180 s
(60 s per letter)
Number of stages passed (max. = 9;
extra-dimensional set-shift is
8th stage)
Number of attempts to select
correct number of moves needed
at 5-move level
Total items named (max. = 64)
The score allocated to subjects who failed prior to the 6-pattern stage of PAL was arbitrary, and its precise value is therefore not recorded here. The actual score
allocated was chosen simply on the basis that it was higher than any error score actually achieved by any subject attempting the stage and thereby signified poorer
performance on the task. This is not equivalent to the adjustment used by Fowler et al. [14, 15].
268
ifying for a clinical diagnosis [11]. One QD patient was receiving

donepezil.
A group of healthy control subjects was recruited primarily to act
as a healthy control group for comparison with the QD patients; these
groups were well-matched in terms of age and premorbid IQ, as estimated by the NART [37] (table 1). Two further groups were recruited
to act as positive control groups to the QD patients. Patients diagnosed with AD or with major unipolar depression were tested to provide the profiles of cognitive performance produced by these particular conditions, against which the profiles of undiagnosed (QD)
patients could be compared. Patients with comorbid diagnoses of
both AD and depression, therefore, were excluded.
All AD patients met the NINCDS-ADRDA [18] criteria for probable AD. AD patients scoring below 17 on the MMSE at clinic were
not included because of the expected difficulty in administering computerised tests to such patients after 2 years. Patients in this group
tended, on average, to be somewhat older and with lower estimated
verbal IQ than those in the QD and matched control groups (table 1),
reflecting the increasing prevalence of AD in older age groups and the
deterioration of global cognitive processes, including language, in
dementia. Four AD patients were receiving donepezil; 2 were receiving rivastigmine. All AD patients lived at home, some independently, but most with a carer. All depressed patients met DSM-IV criteria
[35] for major unipolar depression. Of these, 4 patients were tested in
the West Suffolk Hospital where they were inpatients, and the rest
were tested in their own homes. No depressed patient scored below
24 on the MMSE.
All subjects were paid 4 per hour, up to a maximum of 20 for
each visit (with a visit comprising 13 sessions as necessary). A complimentary box of chocolates was also given at the end of each visit to
thank subjects for their participation. All subjects gave written
informed consent. The study was approved by the relevant local
research ethics committees and was carried out according to the principles of the Declaration of Helsinki [38].
(3 in PAL and DMTS, 1 in logical memory, GNT, Stroop, ID/ED

and 5-choice reaction time). The reasons for missing data included:
colour-blindness, the subject being too tired to continue and refusal
to co-operate with a particular test. In the case of the one-touch Tower of London test, only 10 subjects in the AD group were able to
attempt the task. (During the trials of the original CANTAB Stockings of Cambridge test which were given as training, the rest of this
group showed that they were unable to understand the rules of the
task, making further testing clearly inappropriate).
Statistical Analyses
Missing Data Points and Eligibility for Testing

It was not always possible to collect all the required data from
every subject, although missed scores were in practice extremely rare
Data were analysed using the SPSS for Windows 9.0.0 package
for the PC [39]. Three forms of analysis were carried out on the baseline data: group comparisons, individual patient classification analyses and assessments of predictive validity.
Differences between group mean (or median) performance were
tested using one-way ANOVA or non-parametric Kruskal-Wallis
ANOVA. Planned orthogonal contrasts (which require no reduction
in the significance level applied) were carried out to address the specific hypotheses of the study: firstly, whether the tests were sensitive
to AD in comparison with all other groups (AD vs. QD/depression/
controls); secondly, whether the tests were sensitive to QD when
assessed against a combined control group of depressed and healthy
control subjects (QD vs. depression/controls), and thirdly, whether
the tests were sensitive to depression when compared with healthy
controls (depression vs. controls). Because the AD group were on
average slightly older and of lower estimated IQ than the QD, depression and control groups, the scores of an older, less intelligent group
of controls are provided in the results tables. These data are taken
from groups of healthy control subjects between the ages of 70 and 79
and of estimated IQ 100109, and are available for most of the
CANTAB tests (CeNeS Pharmaceuticals PLC).
In order more rigorously to examine the potential diagnostic value of a test, it was necessary to assess how well it could categorise
individual patients into different groups. This was done firstly by
calculating the degree of overlap between the ranges in the AD and
combined depression/control groups (i.e. the number of individual
scores from either group within the overlapping region as a percentage of the total number of scores). In addition, logistic regression
analysis was carried out using a forward stepwise (likelihood ratio)
method for entry of variables [40]. The variable contributing most
significantly to classification was entered first; any variables which,
when also entered, improved this classification would also be incorporated into the final model. (QD patients, belonging at this point in
the study in neither classification group, were excluded from the classification analyses).
At the time of writing, V2 (8-month follow-up) data were available for the entire QD group, but not for all subjects in the other
groups. (This was due to the fact that recruitment of the QD patients
was complete before recruitment of the other groups; follow-up data
on all tests for the QD group and the other groups will be analysed in
future publications from this study.) MMSE scores at follow-up were
therefore available for the QD patients and provide a useful measure
of global ability 8 months subsequent to initially completing the neuropsychological tests. The change in performance on the MMSE over
8 months was calculated for the QD group and used as an interim
measure of cognitive decline. This decline score was used to corre-
Neuropsychological Testing
Baseline Visit 1 (V1)
Where possible, subjects were tested at their own home over 2
sessions, although in some cases a third session was needed in order
to complete all of the tests. These sessions were completed within 3
weeks of each other. Duration of these sessions was usually about 2 h,
but could be longer if the subject wished to continue; a substantial
break was usually given about half way through each session to minimise fatigue. To minimise differential practice effects, tests were
given with counterbalanced forward:reverse ordering in each group.
Tests given are described in table 2.
Follow-Up Visits 24 (V2, V3, V4)
At 8, 16 and 24 months (B 4 weeks) after the baseline visit, all
subjects are being tested again on the complete battery of tests (with
the exception of the NART which is only given at V1 and V4;
table 2), given in the same order as for the same subject at V1. These
visits are still in progress.
269
Table 3. Tests of cognitive function: group scores
Task
AD
QD
Global cognitive function

ADAS-cog
22.38 (1.11)
11.14 (0.91)
9.32 (0.64)
6.67 (0.38)
Recognition
Warrington SRMT words
Warrington SRMT faces
Pattern recognition
Spatial recognition
Doors recognition
DMTS (correct 12 s delay)
17.31 (0.55)
19.31 (0.54)
58.17 (1.99)
60.96 (2.73)
9.46 (0.48)
4.56 (0.30)
22.33 (0.54)
22.00 (0.51)
80.04 (2.11)
75.47 (1.86)
14.98 (0.57)
6.14 (0.29)
23.08 (0.48)
22.57 (0.51)
83.11 (1.74)
77.30 (1.56)
16.03 (0.62)
6.69 (0.31)
24.21 (0.21)
23.64 (0.22)
86.86 (1.53)
81.28 (1.63)
17.03 (0.47)
7.28 (0.29)
n/a
n/a
75.0 (1.63)
72.2 (1.91)
n/a
5.98 (0.30)
Cued/free recall
Logical memory (30 min recall)
PAL (stages passed)
PAL (6-pattern errors)1
1.96 (0.67)
4.84 (0.29)
42.81 (2.84)
12.42 (1.44)
7.30 (0.17)
12.80 (2.38)
16.14 (1.15)
7.74 (0.09)
9.37 (1.39)
19.87 (1.02)
7.85 (0.08)
7.79 (1.09)
n/a
7.00 (0.08)
n/a
Semantic naming/fluency
Graded naming test
Semantic naming
Category fluency
14.2 (1.3)
57.3 (1.0)
28.5 (1.8)
23.3 (0.7)
62.0 (0.3)
48.6 (2.0)
22.4 (0.7)
63.0 (0.2)
50.4 (2.0)
24.1 (0.7)
62.7 (0.3)
61.1 (2.1)
n/a
n/a
n/a
24.65 (0.10)
24.70 (0.13)
24.72 (0.10)
n/a
Divided attention
5-choice reaction time (accuracy) 23.48 (0.52)
5-choice reaction time
(latency), ms
492 (30)
Sustained attention
RVIP 1-sequence (A)
RVIP 1-sequence (latency), ms
Executive function
Stroop interference
Letter fluency
ID/ED (stages completed)
One-touch Tower (selections to
solution, 5 moves)
415 (12)
Dep
412 (17)
Con
390 (14)
Elderly Con
n/a
n/a
0.918 (0.017)
0.978 (0.004)
0.963 (0.005)
0.988 (0.003)
570 (34 )
482 (23)
461 (12)
417 (11)
n/a
n/a
10.6 (1.7)
26.3 (2.4)
5.52 (0.65 )
4.1 (1.4)
44.5 (2.4)
8.49 (0.13)
7.4 (1.5)
36.2 (2.6)
7.54 (0.34)
4.6 (1.4)
49.8 (2.2)
8.51 (0.17)
n/a
n/a
7.00 (0.36)
3.35 (0.31)
1.98 (0.13)
2.07 (0.13)
1.72 (0.09)
n/a
Means (SEM). n/a = Not assessed; Dep = depression group; Con = control group.
In order to indicate the actual number of errors made, only subjects attempting the stage contributed to the mean
(SEM) score presented here.
1
late with neuropsychological test scores collected at V1 in order to

indicate how well baseline test performance predicted global cognitive decline among the QD patients. Non-parametric Spearmans
rank coefficients were used to produce these correlations; these analyses were one-tailed because the hypothesis being tested was that a
poor baseline test score among patients at an early stage of the disease
would be predictive of future global decline.
Transformed Data
Untransformed scores are displayed in table 3 and in figure 2.
However, in order to conform to the assumptions of normality and
homogeneity of variance, some data were transformed prior to parametric analysis: logarithmic transformation (x = log10(y)) was used
270
for latencies and arcsine transformation (x = 2 ! arcsiney) for proportions.

Selection of Task Measures for Analysis
It was not necessary within the present study to analyse all of the
available measures for every task. Examination of graphically displayed data guided selection of a smaller number of meaningful measures for further analysis. Usually, the most stringent score that
assumed to load most heavily on the psychological function which
the task was being used to measure was selected. In other cases,
where a significant proportion of the severest patients were unable
even to attempt such stages of a task, less stringent scores were used
to avoid conducting analyses with a good deal of missing data.
Fig. 2. Scores of individual subjects on PAL, ADAS-cog, pattern recognition and logical memory recall (i.e. the four
tests best able to classify AD and depression/control subjects). PAL and ADAS-cog are scored in terms of errors, so a
high score indicates poor performance, whereas pattern recognition and logical memory recall are scored in terms of
items correct, so a low score indicates poor performance. F indicates that the subjects failed to reach the 6-pattern
stage.
Tests of Global Cognitive and Mnemonic Function

Means and standard deviations of all mnemonic task
measures are displayed in table 3. All differences between
group means were significant to at least the p ! 0.001 level
of significance. Therefore, planned orthogonal contrasts
were carried out for all tasks as detailed in Materials and
Methods. The results of these comparisons are detailed in
table 4.
Contrasting AD with QD/Depression/Controls. On all

measures, the AD group performed significantly more
poorly than the combined QD/depression/control group
to at least the p ! 0.001 level of significance. As shown in
table 3, in comparison to a more elderly healthy control
group also (where these data were available for the
CANTAB tests), the AD group performed very poorly.
Contrasting QD with Depression/Controls. The QD
subjects performed poorly compared to the depression/
control subjects on the following tasks: the ADAS-cog test
of global cognitive function; three tests of recognition
memory (pattern recognition, doors recognition and
DMTS); the two recall tests (logical memory and PAL),
Results
Sensitivity of the Task Measures to Differences in

Mean Group Performance
271
Table 4. Tests of cognitive function: group contrasts
Task
Contrast
Overlap
AD vs. Dep/Con, %
AD vs. QD/Dep/Con
QD vs. Dep/Con
Dep vs. Con

ADAS-cog
***
U = 131
p ! 0.001
**
U = 1,090
p = 0.002
***
U = 404.0
p = 0.001
21
Recognition
***
n.s.
Pattern recognition
***
Spatial recognition
***
Doors recognition
***
DMTS
(correct 12 s delay)
***
U = 557
p = 0.06
U = 0.62
p = 0.27
t = 1.4
p = 0.15
t = 1.53
p = 0.13
t = 1.29
p = 0.20
t = 1.43
p = 0.16
99
***
U = 1,315
p = 0.06
U = 1,324
p = 0.08
t = 2.28
p = 0.02
t = 1.76
p = 0.08
t = 2.41
p = 0.02
t = 2.45
p = 0.015
n.s.
U = 295
p ! 0.001
U = 495
p ! 0.001
t = 10.24
p ! 0.001
t = 6.92
p ! 0.001
t = 8.99
p ! 0.001
t = 5.41
p ! 0.001
U = 132
p ! 0.001
t = 9.75
p ! 0.001
U = 228
p ! 0.001
U = 1,282
p = 0.016
t = 2.01
p = 0.056
U = 1,053
p = 0.002
n.s.
U = 603
p = 0.18
t = 0.64
p = 0.53
U = 474
p = 0.015
23
t = 8.82
p ! 0.001
U = 462
p ! 0.001
t = 9.36
p ! 0.001
n.s.
t = 0.10
p = 0.92
U = 1,201
p = 0.012
t = 3.03
p = 0.003
n.s.
t = 1.62
p = 0.11
U = 677
p = 0.62
t = 3.79
p ! 0.001
89
t = 4.26
p ! 0.001
t = 4.06
p ! 0.001
n.s.
t = 0.75
p = 0.46
t = 1.18
p = 0.24
n.s.
t = 0.24
p = 0.81
t = 0.97
p = 0.34
97
U = 644
p ! 0.001
t = 4.53
p ! 0.001
n.s.
U = 1,597
p = 0.84
t = 1.75
p = 0.08
***
U = 391
p = 0.001
t = 2.04
p = 0.04
95
t = 2.76
p = 0.007
t = 5.38
p ! 0.001
U = 730
p ! 0.001
t = 5.83
p ! 0.001
n.s.
t = 0.91
p = 0.36
t = 0.52
p = 0.61
U = 1,466
p = 0.25
t = 0.57
p = 0.57
n.s.
t = 1.63
p = 0.11
t = 4.03
p ! 0.001
U = 511
p = 0.008
t = 1.99
p = 0.049
95
Cued/free recall
PAL
(stages passed)
PAL
(6-pattern errors)
Logical memory
(30 min recall)
***
***
***
Semantic naming/fluency
Graded naming test
***
Semantic naming
***
Category fluency
***
Divided attention
(accuracy)
(latency), ms
***
***
Sustained attention
RVIP 1-sequence (A))
***
RVIP 1-sequence
(latency), ms
***
Executive function
Stroop interference
**
Letter fluency
***
ID/ED
(stages completed)
One-touch Tower (selections
to solution, 5 moves)1
***
***
n.s.
*
n.s.
*
*
*
**
*
***
n.s.
n.s.
n.s.
n.s.
n.s.
n.s.
n.s.
n.s.
n.s.
n.s.
n.s.
*
n.s.
***
n.s.
***
**
*
78
26
58
53
70
7
27
94
33
77
92
78
99
45
Dep = Depression group; Con = control group; U = Mann-Whitney U test; t = Students t test.
* p ! 0.05; ** p ! 0.01; *** p ! 0.001; n.s. p 1 0.05.
Only 10 AD patients were able to attempt the one-touch Tower task.
272
and two tests of semantic memory function (the semantic

naming test and category fluency). Two recognition tests
(the Warrington Short Recognition Memory Tests for
words and faces) were insensitive to this contrast, as was
the graded naming test of semantic memory.
Contrasting Depression with Controls. Those measures
which failed to show differences between QD and depression/control subjects also failed to show differences between the depressed and control subjects. Of those which
had been sensitive to impairments in the QD group,
ADAS-cog, logical memory recall and category fluency
showed poorer performance in depressed than control
subjects, whereas pattern recognition, doors recognition,
DMTS, PAL and semantic naming showed no significant
differences between depressed patients and controls.
Tests of Attentional/Executive Function
All of the attentional and executive test measures
showed overall group differences to at least the p ! 0.01
level of significance. The test scores and results of the
between-group contrasts, described below, are shown in
tables 3 and 4.
Contrasting AD with QD/Depression/Controls. All of
the tests used were sensitive to the deficits of AD patients
when compared to the other three groups combined.
Again, where an elderly control score was available as
was the case for ID/ED stages completed the AD group
can be seen to have performed very poorly in comparison.
Contrasting QD with Depression/Controls. In contrast
to the results from the mnemonic tasks, none of the attentional/executive tasks showed a significant difference in
mean performance between the QD and depression/control groups. This does not mean that the QD patients did
not perform poorly as a group on these tasks, the lack of
significance possibly being due to poor performance of the
depressed patients masking impaired QD performance. It
does mean, however, that these tasks would probably be
of little use in differentiating individuals in the very early
stages of AD from those who were depressed.
Contrasting Depression with Controls. Most of the
attentional/executive tasks showed significant differences
between depressed patients and controls. The RVIP test
of sustained attention, letter fluency, the ID/ED attentional set-shifting task and the one-touch Tower test of
planning all were sensitive to impairments in the depressed group, whilst the 5-choice reaction time test of
divided attention and the Stroop test of inhibition revealed no significant differences between these two
groups.
Efficacy of Task Measures in Patient Classification

Measures of Overlap between Groups
Table 4 shows the degree of overlap between AD and
combined depression/control groups for each task measure. Measures producing the least overlap are those most
able to assign individual subjects appropriately to these
two categories. The most effective classifier by this method was PAL 6-pattern errors (7% overlap). This task, as
well as pattern recognition (26% overlap), was sensitive to
the QD versus depression/control, but not the depression
versus control contrast in the between-group comparisons. ADAS-cog (21%) and logical memory recall (27%
overlap) also gave reasonably high separation despite
being sensitive to impairments in depression; thus, although depressed patients produced fewer correct responses on these tasks than did controls, a large proportion of AD subjects scored below the level of even the
most impaired depressed subject. Figure 2 shows the performance of individuals from each group on the four most
effective tasks for classification. In the case of PAL 6-pattern errors, not only can the ranges of scores of the AD
patients be seen to be almost separate from that of the
depression and control subjects (depression and control
ranges being essentially equivalent), but examination of
the individual data from the QD group reveals another
interesting feature: that is, there is a visually obvious split
in these scores. Whilst the majority of the QD patients
scored well within the depression/control range, 12 QD
patients performed very much like the AD patients, making a higher number of errors than any control or
depressed patient. Such a split was not noticeable for any
other of the scores analysed, not even for the other effective classifiers shown in figure 2.
In general, the tests of attention and executive function
distinguished less effectively between AD and depression/
control subjects than did the mnemonic tasks. The data in
table 4 show that in the cases of the RVIP A), letter fluency and ID/ED stages completed scores, the depression
group performed substantially more poorly than the QD
group. The degree of overlap between the depression and
combined QD/control groups was calculated to show
whether or not these measures could be useful in identifying depressed patients in the clinic: it was found that
97% of the data (in the cases of ID/ED stages passed and
letter fluency) and 99% (in the case of RVIP A)) fell in the
overlapping region. Therefore, on an individual basis, it is
unlikely that these scores could be used as markers of
depression.
273
Table 5. Correlations of principal task
measures with global cognitive decline
Task
Correlation with global (MMSE) decline

ADAS-cog
rs = 0.19
p = 0.12
n.s.
Mnemonic tasks
Pattern recognition
Spatial recognition
Doors recognition
DMTS (correct 12 s delay)
PAL (stages passed)
PAL (6-pattern errors)
Logical memory (30 min recall)
Graded naming test
Semantic naming
Category fluency
rs = 0.13
rs = 0.23
rs = 0.15
rs = 0.03
rs = 0.19
rs = 0.29
rs = 0.39
rs = 0.33
rs = 0.24
rs = 0.03
rs = 0.09
rs = 0.20
p = 0.21
p = 0.07
p = 0.17
p = 0.42
p = 0.11
p = 0.03
p ! 0.01
p = 0.02
p = 0.06
p = 0.43
p = 0.28
p = 0.11
n.s.
n.s.
n.s.
n.s.
n.s.
*
**
*
n.s.
n.s.
n.s.
n.s.
rs = 0.21
rs = 0.32
rs = 0.01
rs = 0.24
rs = 0.01
rs = 0.25
rs = 0.01
p = 0.09
p = 0.02
p = 0.47
p = 0.06
p = 0.47
p = 0.06
p = 0.47
n.s.
*
n.s.
n.s.
n.s.
n.s.
n.s.
rs = 0.26
p = 0.05
Attentional and executive tasks

5-choice reaction time (accuracy)
5-choice reaction time (latency), ms
RVIP 1-sequence (A))
RVIP 1-sequence (latency), ms
Stroop interference
Letter fluency
ID/ED (stages completed)
One-touch Tower (selections to
solution, 5 moves)
For significances see table 4.

Spearmans correlation coefficient of V1 neuropsychological test score with V2V1
MMSE score.
Logistic Regression Analysis

All task measures were entered into a logistic regression analysis [40] with outcome being AD or combined
depression/control group membership. Only the PAL 6pattern errors measure needed to be entered into the analysis, i.e. a model using just this PAL score was able to
classify individuals with 98% accuracy (only 1 subject
from each group being misclassified), and no other measure being included in the model would have been able to
achieve a better classification.
Predictive Capabilities of Neuropsychological
Measures
The degree of decline in performance on the MMSE (a
test of global cognitive function) was correlated with neuropsychological test performance at V1 to assess whether
baseline test performance was significantly poorer in
those QD individuals who would experience greater glob-
274
al cognitive decline over 8 months. A correlation matrix

of V1 test scores with V2V1 global decline score is shown
in table 5. Tests which predicted global decline came from
the mnemonic, attentional and executive categories, but
not the global cognitive function category. Thus, the
ADAS-cog global cognitive function scores were unable to
predict global decline as assessed by the MMSE, but measures from four neuropsychological tasks produced significant correlations with global cognitive decline. These
measures were: the number of correct responses at the 12second delay of DMTS; the number of stages passed, as
well as the number of errors made at the 6-pattern stage,
of PAL; latency of response in the 5-choice reaction time
task; and the number of selections to solution on the 5move problems of the one-touch Tower test of planning.
In all cases, poor test performance correlated with decline
in global cognitive function.
Discussion
This study has compared groups of probable AD, QD,

major depression and healthy control subjects on a number of neuropsychological tests in order to identify those
tests which are most likely to be of use in helping to diagnose very early AD. Whereas several mnemonic tests were
sensitive to QD, other (attentional and executive) tests
were more sensitive to depression. Scores on a test of
visuo-spatial PAL very accurately classified individuals as
belonging to either the AD or the combined depression/
control group. This score also allowed detection of two
apparently distinct sub-groups of QD patients. Those QD
patients who performed within the same range as the
diagnosed AD patients may already be in the early stages
of the disease. Scores from four tests, including the
CANTAB PAL and DMTS tests, correlated with the
degree of global cognitive decline over the 8 months following testing, indicating a degree of predictive validity
for these scores.
The potential to detect AD in its very early stages by
means of neuropsychological investigation has generated
huge interest in recent years [1315, 29, 36, 4146]. Episodic memory is generally regarded to be especially compromised in AD [1924], yet in order to qualify for a diagnosis of probable AD, deficits must be present in at least
one cognitive domain other than memory, and it can be
seen that when compared with healthy elderly individuals, even patients in the early stages of AD tend to be
impaired on a wide range of cognitive functions [16, 28].
It may not be surprising therefore that when prospective
studies have tracked the progress of initially non-demented subjects over a number of years, many neuropsychological measures have been found to be related to later
AD diagnosis [13, 29]. Because conditions such as depression, which are difficult to differentiate from early AD
clinically, also compromise many of the same cognitive
processes, tests must have considerable specificity to aid
differential diagnosis. The present study has differentiated the particular profile of cognitive disturbance in
early AD from that seen in clinical depression and identified which neuropsychological tests are likely to identify
individuals in the very early stages of AD from among
those with isolated memory loss. Specifically, we have
asked:
Which tests are sensitive to the cognitive deficits of
questionable dementia over and above those seen in
depression and healthy ageing?
Can these tests be used to classify demented patients
on an individual patient basis?
Which, if any, tests are more sensitive to depression

than to early AD, and could these contribute to individual
patient classification?
Which tests are already proving to be of value in predicting global cognitive decline among the patients with
isolated memory loss?
These questions will now be addressed in turn.
The Sensitivity of Neuropsychological Tests to QD

We assume that tests sensitive to impairments in the
QD group are probably detecting patients with very early,
or prodromal, AD, such patients being expected to constitute a reasonable proportion of the QD group. Whereas
all of the cognitive tasks analysed showed significant
impairments among the diagnosed AD group compared
to all other groups, only a restricted set of measures was
sensitive to impairments in the undiagnosed QD group
compared to the combined depressed and control group.
Those tests were: ADAS-cog, pattern recognition, doors
recognition, DMTS, PAL, logical memory recall, semantic naming and category fluency. Tasks insensitive to this
comparison were: the Warrington Short Recognition
Memory Tests (words and faces), spatial recognition, the
graded naming test, and all of the tasks of attention and
executive function. These results suggest that, on the
whole, QD patients differ from depressed and healthy
elderly subjects more significantly in terms of memory
than in terms of attentional or executive function. The
ADAS-cog, although nominally included as a test of global
cognitive function, was probably detecting differences
between the QD and depression/control groups in terms
of mnemonic function, as most of the errors made on this
task by QD, depressed and control subjects were made in
the word recall and word recognition sections of the test.
This result is consistent with the staging of cognitive
impairment in early AD, episodic memory being impaired initially, followed by impairments in either semantic memory or attention [16]. Because the contrast was
with depressed patients as well as controls, these data also
show that tests loading highly on mnemonic processes
may be useful in differentiating very early AD from
depression. Although it is probably safe to assume that
some QD patients were already in the early stages of AD
at the time of testing, we have as yet no information on the
composition of the rest of the group. Conceivably, very
few patients in this group who were actually suffering
from, for example, undiagnosed Lewy body disease or
frontotemporal lobar degeneration, could affect analyses
275
based on mean group performance. Thus, some caution

must be exercised at this point when interpreting the
impairments of the QD group.
Classification of Individual Patients as Demented or

Non-Demented
To contribute usefully to diagnostic techniques, neuropsychological measures must be able not only to detect
differences between mean group performances, but also
accurately classify individual patients into the appropriate diagnostic categories. The PAL 6-pattern errors
measure was the most useful measure in this sense, producing only 7% overlap between AD patients and the
combined non-demented (depressed and healthy) controls. The next most successful tasks for classification
were ADAS-cog, pattern recognition and logical memory
recall, with 21, 26 and 27% overlap, respectively. Logistic
regression analysis produced a model whereby the PAL
single 6-pattern errors measure classified patients with
98% accuracy. This single task measure alone was required to achieve such a high level of sensitivity and specificity and the addition of no other measure improved on
the accuracy of this model. Applying the PAL 6-pattern
errors measure to QD patients had the effect of distinctly
sub-grouping individuals into either the AD-like or the
depressed/control-like range of performance (fig. 2), an
effect not achieved with such resolution by any other test.
A similar sub-grouping of QD patients into AD-like and
control-like performance on the basis of PAL score was
apparent in the study by Fowler et al. [15], an important
difference being that the sub-groups in the Fowler study
were defined on the basis of a longitudinal decline score
over 6 and 12 months, whereas those in the present study
were defined according to a single baseline score. Whether
the apparent categorisation reported here is meaningful in
terms of how accurately it indicates the presence or
absence of early dementia in these QD patients remains to
be addressed by the ongoing follow-up assessments.
Sensitivity of Neuropsychological Tests to Depression

None of the task measures analysed showed greater
impairments in depression than in mild probable AD.
Thus, although different psychological functions may potentially be relatively more vulnerable to depression than
to mild AD, the global severity of the relatively mild AD
group in this study appears to have progressed at least as
276
far as to include those functions which are compromised

in depression. An alternative approach, therefore, compares depression with very early AD, which we may
assume cautiously to describe a subgroup of the QD
group. The degree of overlap between depressed patients
and a combined non-depressed control group (QD and
healthy control subjects) was very high (195%), so even
the measures which were sensitive to depression in the
between-group comparisons would not be useful for classifying individual patients in the clinic as suffering from
depression rather than early AD. Nevertheless, although
the QD and depression groups were not directly compared statistically, it was clear that on the RVIP, letter
fluency and ID/ED tasks, depressed patients performed
more poorly than QD patients, and that on the ID/ED
task at least, depressed patients were impaired, whereas
the performance of the QD group was indistinguishable
from that of controls. Thus the impairments of the QD
group on memory tests were not simply an exaggerated
form of the type of cognitive impairment seen in depression, demonstrating that very early AD may be possible to
differentiate from depression on the basis of neuropsychological test performance.
Predictive Capabilities of Neuropsychological Testing

Are the present neuropsychological measures capable
of predicting future global cognitive decline, as some
authors have suggested is the case with other measures
[47]? Correlations were computed for the QD group as a
whole between baseline performance on individual neuropsychological tasks and degree of subsequent deterioration of general cognitive function, as indicated by a poorer
score on the MMSE at 8-month follow-up compared with
baseline. A significant correlation would indicate that
poor initial test scores could predict a high rate of future
cognitive decline; this is important because a steeper rate
of decline would in most cases reflect a poorer prognosis
than the relatively gradual cognitive decline associated
with ageing. Whilst the ADAS-cog measure of global function did not correlate with decline as measured by the
MMSE, some neuropsychological scores (poor baseline
performance on the PAL, DMTS, 5-choice reaction time
and one-touch Tower tests) predicted global (MMSE)
decline. Thus, these tests may be sensitive enough to
detect the cognitive effects of neuropathology in early AD
in individuals who will go on to receive a diagnosis of
probable AD. Although several measures showed significant predictive capabilities, these results must be assessed
in the light of their specificity and sensitivity for AD. On

these grounds, the predictive capacity of the PAL test
which is very insensitive to depression may prove to be
the most useful. Correlations between the neuropsychological tests and MMSE decline were significant, but relatively small in magnitude; nevertheless, the finding of significant correlations between specific aspects of cognition
and decline as measured by a task involving a number of
cognitive domains is remarkable, especially considering
the failure of ADAS-cog, itself a global measure, to correlate with global decline as measured by the MMSE.
The current study adds to the literature on early detection of AD in a number of important ways. Firstly, most
studies examining early deficits in AD [1316, 27, 36, 46]
have included patients from no other diagnostic category
for comparison, thus relying on comparisons of patients
who have (or who have later gone on to receive) a diagnosis of AD with healthy elderly subjects. (Exceptions are
studies comparing AD with frontotemporal lobar degeneration, FTLD [48, 49], and AD with Parkinsons disease
[20]). The inclusion in the present study of a group of
patients with major depression a disorder which is
important to consider in the differential diagnosis of AD
makes the high degree of specificity achieved with certain
neuropsychological tests, especially PAL, particularly relevant to clinical practice. That AD patients did not have
depression allowed interpretation of the results of the AD
group unconfounded by depression-induced changes in
cognition [50]. It would be of interest in future studies to
determine whether the cognitive effects of AD and depression are additive or synergistic. Many of the tests used
in the present study have also been used with other disorders that affect memory in the elderly. One such disorder,
which may also be difficult to differentiate from AD in a
clinical situation, is FTLD [48, 51]. The frontal variant of
this disorder shows a particular pattern of impairment on
the ID/ED task in that patients make a particularly high
number of errors on the reversal stages of the task, a pattern interpreted as indicating poor learning of altered
stimulus-reward contingencies subsequent to damage to
the orbitofrontal cortex [52]. This contrasts with the unimpaired performance at all stages of the test of the QD
patients in the present study and could potentially contribute to differential diagnosis of FTLD from AD.
Secondly, we have confirmed the finding from the
Fowler study that measures of relatively precise cognitive
function seem to be more effective at predicting global
decline than the more global measures used clinically, and
we have extended the finding to a wider range of neuropsychological tests (compare PAL, DMTS, 5-choice reac-
tion time and one-touch Tower with ADAS-cog in table 5).

A third important finding is that a single neuropsychological score PAL 6-pattern errors can split patients
with memory impairments into two apparently separate
sub-groups, and that this score correlates significantly
with subsequent global cognitive decline. Fowler et al.
[15] showed that classifying QD subjects on the basis of
decline versus stable performance over 6 or 12 months in
performance on PAL produced apparently separate subgroups which may have corresponded to dementing versus non-dementing populations. By excluding QD patients with abnormal neuropsychological performance
from their study, Fowler et al. [15] studied a restricted
range of patients who would on average have been at a
rather earlier stage of the disease process than those in the
present study, and a decline in performance over time was
necessary to define an AD-like sub-group of QD patients.
The present study included a number of patients with significant (although isolated) memory impairments, and
therefore was able to show that PAL performance could
be just as poor (in terms of 6-pattern errors, at least) in
patients who did not qualify for a diagnosis of probable
AD as in those who did, even at their initial assessment.
The continuing follow-up of the QD patients in the
present study may identify a proportion of patients like
those from the Fowler study (presently in the normal-performance sub-group) whose global decline is predicted by
a PAL performance which declines abnormally over time.
The remaining follow-up phases of this study will be of
crucial importance in addressing whether sub-groups
based on either poor (at baseline) or declining performance on a single neuropsychological task are clinically
meaningful. Specifically, it will become clear whether
such a single task can predict, with a sufficiently high
degree of accuracy, which memory-impaired patients will
go on to receive a diagnosis of probable AD.
Neural Substrates Underlying Cognitive Dysfunction

in AD and Depression
We have seen that QD mainly involves impairments in
a range of memory functions (including recognition, recall, naming and fluency tasks), whereas depression impairs attentional and executive function more than memory. An analysis of the neural bases underlying these different although overlapping patterns of impairment
may explain, for example, why particular tasks such as
PAL are so sensitive and specific for early AD.
277
Evidence from lesion studies in rats [53], monkeys [54]

and humans [55], as well as functional imaging studies in
normal human volunteers [56, 57], suggest that those temporal lobe structures particularly susceptible to early AD,
such as the hippocampus, parahippocampal gyrus and
entorhinal cortex are implicated in the main cognitive
component of the PAL test, such as stimulus-context and
object-location association processes and episodic memory. By contrast, the cognitive deficits in depression have
been attributed to underlying impairments in effortful
rather than automatic processing, being seen on tasks
requiring strategy and the re-allocation of cognitive resources [58, 59], and associated with prefrontal cortex
dysfunction [60]. A profile of impairments on tasks sensitive to frontal lobe dysfunction was found in elderly
depressed patients by Beats et al. [50], and in the present
study, impairments were most pronounced on attentional
and executive tasks, and on those mnemonic tasks which
would be expected to be benefit from a strategic approach,
i.e. logical memory recall and category fluency. Thus,
tasks requiring a high level of context memory, but a relatively low level of strategy would be expected not only to
be sensitive, but also specific, for early AD. Indeed, supportive tasks, such as those with cues given at recall, are
particularly specific for early dementia [46, 61, 62]. This
may explain the finding of extremely high specificity of
the PAL task in the present study. The PAL task may be
seen as a cued recall task, with the centrally displayed pat-
tern providing a cue for recall of the associated spatial

information. Displaying the 68 potential locations as
well as the pattern at recall probably provides enough support to enable a very high proportion of healthy elderly
and depressed subjects to perform at a higher level than
AD patients.
In summary, the PAL test is sensitive and specific for
AD and preliminary analysis suggests that it may also be
sensitive to the specific memory deficits of prodromal
AD. Because of its requirement to combine object-based
and contextual information, PAL may be particularly sensitive to the entorhinal/hippocampal damage sustained in
AD, and the provision of supportive recall cues may
explain its relative insensitivity to the strategic, or effortful, processes which may underlie poor memory function
in the elderly and in depression.
Acknowledgments
We would like thank Drs. Tom Dening, Deborah Girling and
Carol Gregory for referring patients; Claire Fisher, Melissa Jefferies,
Claire ONeill and Isabel Stow for testing patients; Carolyn Crane for
recruiting patients, and all of the patients and control subjects for
allowing us to test them. This project was funded by an MRC-LINK
grant to B.J.S., J.H.R., T.W.R. and J.S., and completed within the
MRC Co-operative Group in Brain, Behaviour and Neuropsychiatry. The CANTAB tests are available from CeNeS Pharmaceuticals
PLC [www.cenes.com].
References
1 Sands LP, Katz I, Schneider L: Assessing individual patients for cognitive benefits from acetylcholinesterase inhibitors. Alzheimer Dis Assoc Disord 1999;13:2633.
2 Eagger SA, Levy R, Sahakian BJ: Tacrine in
Alzheimers disease. Lancet 1991;337:989
992.
3 Rogers SL, Farlow MR, Doody RS, Mohs R,
Friedhoff LT, Donepezil Study Group: A 24week, double-blind, placebo-controlled trial of
donepezil in patients with Alzheimers disease.
Neurology 1998;50:136145.
4 Spencer CM, Noble S: Rivastigmine: A review
of its use in Alzheimers disease. Drugs Aging
1998;13:391411.
5 Small GW: Treatment of Alzheimers disease:
Current approaches and promising developments. Am J Med 1998;104:32S38S.
278
6 Schenk D, Barbour R, Dunn W, Gordon G,

Grajeda H, Guido T, Hu K, Huang J, JohnsonWood K, Dhan K, Kholodenko D, Lee M, Liao
Z, Lieberburg I, Motter R, Mutter L, Soriano F,
Shopp G, Vasquez N, Vandevert C, Walker S,
Wogulis M, Yednock T, Games D, Seubert P:
Immunization with amyloid-beta attenuates
Alzheimer-disease-like pathology in the
PDAPP mouse. Nature 1999;400:173177.
7 Fox NC, Warrington EK, Rossor MN: Serial
magnetic resonance imaging of cerebral atrophy in preclinical Alzheimers disease. Lancet
1999;353:2125.
8 Jobst KA, Barnetson LPD, Shepstone BJ, OPTIMA: Accurate prediction of histologically
confirmed Alzheimers disease and the differential diagnosis of dementia: The use of
NINCDS-ADRDA and DSM-III-R criterial,
SPECT, X-ray CT, and Apo E4 in medial temporal lobe dementias. Int Psychogeriatr 1998;
10:271302.
9 Small GW, Leiter F: Neuroimaging for diagnosis of dementia. J Clin Psychiatry 1998;59:
47.
10 Braak H, Braak E: Staging of Alzheimers disease-related neurofibrillary changes. Neurobiol

Aging 1995;16:271284.
11 Hodges JR: The amnestic prodrome of Alzheimers disease. Brain 1998;121:16011602.
12 Grober E, Dickson D, Sliwinski M, Buschke H,
Katz M, Crystal H, Lipton R: Memory and
mental status correlates of modified Braak
staging. Neurobiol Aging 1999;20:573579.
13 Petersen RC, Smith GE, Waring SC, Ivnik RJ,
Tangalos EG, Kokmen E: Mild Cognitive Impairment. Arch Neurol 1999;56:303308.
14 Fowler KS, Saling MM, Conway EL, Semple
JM, Louis WJ: Computerized delayed matching to sample and paired associate performance
in the early detection of dementia. Applied
Neuropsychology 1995;2:7278.
15 Fowler KS, Saling MM, Conway EL, Semple
JM, Louis WJ: Computerized neuropsychological tests in the early detection of dementia:
Prospective findings. J Int Neuropsychol Soc
1997;3:139146.
16 Perry RJ, Watson P, Hodges JR: The nature

and staging of attention dysfunction in early
(minimal and mild) Alzheimers disease: Relationship to episodic and semantic memory impairment. Neuropsychologia 2000;38:252
271.
17 Perry R, Hodges J: Fate of patients with questionable (very mild) Alzheimers disease: Longitudinal profiles of individual subjects decline. Dement Geriatr Cogn Disord 2000;11:
342349.
18 McKhann G, Drachman D, Folstein M, Katzman R, Price D, et al: Clinical diagnosis of Alzheimers disease: Report of the NINCDSADRDA Work Group under the auspices of
Department of Health and Human Services
Task Force on Alzheimers Disease. Neurology
1984;34:939944.
19 Corkin S, Growdon J, Nissen M, Huff F, Freed
D, Sagar J: Recent advances in the neuropsychological study of Alzheimers disease; in
Wurtman R, Corkin S, Growdon H (eds): Alzheimers Disease: Advances in Basic Research
and Therapies. Cambridge, Center for Brain
Sciences and Metabolism Charitable Trust,
1984, pp 7593.
20 Sahakian BJ, Morris RG, Evenden JL, Heald
A, Levy R, Philpot M, Robbins TW: A comparative study of visuospatial memory and learning in Alzheimer-type dementia and Parkinsons disease. Brain 1988;111:695718.
21 Welsh K, Butters N, Hughes J, Mohs R, Heyman A: Detection of abnormal memory decline
in mild cases of Alzheimers disease using
CERAD neuropsychological measures. Arch
Neurol 1991;48:278281.
22 Welsh K, Butters N, Hughes J, Mohs R, Heyman A: Detection and staging of dementia in
Alzheimers disease. Use of the neuropsychological measures developed for the Consortium
to Establish a Registry for Alzheimers Disease.
Arch Neurol 1992;49:448452.
23 Petersen RC, Smith GE, Ivnik RJ, Kokmen E,
Tangalos EG: Memory function in very early
Alzheimers disease. Neurology 1994;44:867
872.
24 Zakzanis KK: Quantitative evidence for neuroanatomic and neuropsychological markers in
dementia of the Alzheimers type. J Clin Exp
Neuropsychol 1998;20:259269.
25 Lichtenberg PA, Ross T, Millis SR, Manning
CA: The relationship between depression and
cognition in older adults: A cross-validation
study. J Gerontol B Psychol Sci Soc Sci 1995;
50B:2532.
26 Zakzanis KK, Leach L, Kaplan E: On the
nature and pattern of neurocognitive function
in major depressive disorder. Neuropsychiatry
Neuropsychol Behav Neurol 1998;11:111
119.
27 Hodges JR, Patterson K: Is semantic memory
consistently impaired early in the course of Alzheimers disease? Neuroanatomical and diagnostic implications. Neuropsychologia 1995;
33:441459.
28 Orgogozo JM, Fabrigoule C, Dartigues JF: Early diagnosis of Alzheimers disease with simple
neuropsychological tests; in Iqbal K, Swaab D,
Winblad B, Wisniewski H (eds): Alzheimers
Disease and Related Disorders. Wiley & Sons,
1999, pp 127135.
29 Linn RT, Wolf PA, Bachman DL, Knoefel JE,
Cobb JL, Belanger AJ, Kaplan EF, DAgostino
RB: The preclinical phase of probable Alzheimers disease. Arch Neurol 1995;52:485
490.
30 Folstein MF, Folstein SE, McHugh PR: MiniMental State: A practical method for grading
the cognitive state of patients for the clinician.
J Psychiatr Res 1975;12:189198.
31 Rosen WG, Mohs RC, Davis KL: A new rating
scale for Alzheimers disease. Am J Psychiatry
1984;141:13561364.
32 Warrington EK: Camden Memory Tests.
Hove, Psychology Press, 1996.
33 Wechsler D: Wechsler Memory Scale Revised. San Antonio, Psychological Corporation, 1987.
34 Robbins T, James M, Owen A, Sahakian B,
McInnes L, Rabbitt P: Cambridge Neuropsychological Test Automated Battery (CANTAB): A factor analytic study of a large sample
of normal elderly volunteers. Dementia 1994;
5:266281.
35 American Psychiatric Association: Diagnostic
and Statistical Manual of Mental Disorders, ed
4 (DSM-IV). Washington, APA, 1994.
36 Tierney MC, Szalai JP, Snow WG, Fisher RH,
Nores A, Nadon G, Dunn E, St George-Hyslopp PH: Prediction of probable Alzheimers
disease in memory-impaired patients: A prospective longitudinal study. Neurology 1996;
46:661665.
37 Nelson HE: National Adult Reading Test
(NART) Test Manual. Windsor, NFER-Nelson, 1982.
38 World Medical Association: Declaration of
Helsinki: Recommendations guiding physicians in biomedical research involving human
subjects. World Medical Association, 1996.
39 SPSS Inc: SPSS for Windows 9.0.0. Chicago,
SPSS, 1998.
40 Hosmer DW, Lemeshow S: Applied Logistic
Regression. New York, Wiley, 1989.
41 Amieva H, Rouch-Leroyer I, Fabrigoule C,
Dartigues J-F: Deterioration of controlled processes in the preclinical phase of dementia: A
confirmatory analysis. Dement Geriatr Cogn
Disord 2000;11:4652.
42 Bowen J, Teri L, Kukull W, McCormick W,
McCurry SM, Larson EB: Progression to dementia in patients with isolated memory loss.
Lancet 1997;349:763765.
43 Fabrigoule C, Rouch I, Taberly A, Letenneur L,
Commenges D, Mazaux J-M, Orgogozo J-M,
Dartigues J-F: Cognitive process in preclinical
phase of dementia. Brain 1998;121:135141.
44 Fox NC, Warrington EK, Seiffer AL, Agnew
SK, Rossor MN: Presymptomatic cognitive deficits in individuals at risk of familial Alzheimers disease. Brain 1998;121:16311639.
45 Jacobs DM, Sano M, Dooneief G, Marder K,

Bell KL, Stern Y: Neuropsychological detection and characterization of preclinical Alzheimers disease. Neurology 1995;46:957962.
46 Small BJ, Fratiglioni L, Almkvist O, Herlitz A,
Backman L: Cognitive predictors of incident
Alzheimers disease: A prospective longitudinal study. Neuropsychology 1997;11:413420.
47 Flicker C, Ferris SH, Reisberg B: Mild cognitive impairment in the elderly: Predictors of
dementia. Neurology 1991;41:10061009.
48 Hodges J, Patterson K, Ward R, Garrard P,
Bak T, Perry R, Gregory C: The differentiation
of semantic dementia and frontal lobe dementia (temporal and frontal variants of frontotemporal dementia) from early Alzheimers disease: A comparative neuropsychological study.
Neuropsychology 1999;13:3140.
49 Perry R, Hodges J: Differentiating frontal and
temporal variant frontotemporal dementia
from Alzheimers disease. Neurology 2000;54:
22772284.
50 Beats BC, Sahakian BJ, Levy R: Cognitive performance in tests sensitive to frontal lobe dysfunction in the elderly depressed. Psychol Med
1996;26:591603.
51 Gregory C, Serra-Mestres J, Hodges J: Early
diagnosis of the frontal variant of frontotemporal dementia: How sensitive are standard neuroimaging and neuropsychologic tests? Neuropsychiatry Neuropsychol Behav Neurol
1999;12:128135.
52 Rahman S, Sahakian BJ, Hodges JR, Rogers
RD, Robbins TW: Specific cognitive deficits in
mild frontal variant fronto-temporal dementia.
Brain 1999;122:14691493.
53 McDonald RJ, White NM: A triple dissociation of memory systems: Hippocampus, amygdala, and dorsal striatum. Behav Neurosci
1993;107:322.
54 Miyashita Y, Kameyama M, Hasegawa I, Fukushima T: Consolidation of visual associative
long-term memory in the temporal cortex of
primates. Neurobiol Learn Mem 1998;70:197
211.
55 Smith M, Milner B: The role of the right hippocampus in the recall of spatial location. Neuropsychologia 1981;19:781793.
56 Owen AM, Milner B, Petrides M, Evans AD: A
specific role for the right parahippocampal gyrus in the retrieval of object-location: A positron emission tomography study. J Cogn Neurosci 1996;8:588602.
57 Maguire EA, Frith CE, Burgess N, Donnett JG,
OKeefe J: Knowing where things are: Parahippocampal involvement in encoding object locations in virtual large-scale space. J Cogn Neurosci 1998;10:6176.
58 Channon S, Baker J, Robertson MM: Working
memory in clinical depression: An experimental study. Psychol Med 1993;23:8791.
59 Baker JE, Channon S: Reasoning in depression:
Impairment on a concept discrimination learning task. Cogn Emot 1995;9:579597.
279
60 Norman D, Shallice T: Attention to action.

Willed and automatic control of behaviour; in
Davidson R, Schwartz G, Shapiro D (eds):
Consciousness and Self-Regulation, New York,
Plenum Press, 1986, vol 4, 118.
61 Buschke H, Sliwinski MJ, Kuslansky G, Lipton
RB: Diagnosis of early dementia by the Double
Memory Test: Encoding specificity impoves
diagnostic sensitivity and specificity. Neurology 1997;48:989997.
62 Herlitz A, Hill RD, Fratiglioni L, Backman L:
Episodic memory and visuospatial ability in
detecting and staging dementia in a community-based sample of very old adults. J Gerontol
A Biol Sci Med Sci 1995;50:M107M113.
63 Yesavage JA, Brink TL, Rose TL, Lum O,
Huang V, Adey M, Leirer VO: Development
and validation of a geriatric depression screening scale: A preliminary report. J Psychiatr Res
1982;17:3749.
64 Hamilton M: A rating scale for depression. J
Neurol Neurosurg Psychiatry 1960;23:5662.
280
65 Baddeley A, Emslie H, Nimmo-Smith I: Doors

and People. Bury St Edmunds, Thames Valley
Test Company, 1994.
66 McKenna P, Warrington EK: Graded Naming
Test. Windsor, NFER-NELSON, 1983.
67 Bozeat S, Lambon Ralph MA, Patterson K,
Garrard P, Hodges JR: Non-verbal semantic
impairment in semantic dementia. Neuropsychologia 2000;38:12071215.
68 Newcombe F: Missile Wounds of the Brain.
London, Oxford University Press, 1969.
69 Sahakian BJ, Owen AM, Morant NJ, Eagger
SA, Boddington S, Crayton L, Crockford HA,
Crooks M, Hill K, Levy R: Further analysis of
the cognitive effects of tetrahydroaminoacridine (THA) in Alzheimers disease: Assessment
of attentional and mnemonic function using
CANTAB. Psychopharmacology 1993;110:
395401.
70 Sahakian BJ, Jones GMM, Levy R, Gray JA,
Warburton DM: The effects of nicotine on
attention, information processing, and shortterm memory in patients with dementia of the
Alzheimer type. Br J Psychiatry 1989;154:797
800.
71 Trenerry MR, Crosson B, DeBoe J, Lever WR:

Stroop Neuropsychological Screening Test.
Florida, Psychological Assessment Resources,
1989.
72 Fisher LM, Freed DM, Corkin S: Stroop colour-word test performance in patients with
Alzheimers disease. J Clin Exp Neuropsychol
1990;12:745758.
73 Benton AL: Differential effects of frontal lobe
disease. Neuropsychologia 1968;6:5360.
74 Downes JJ, Roberts AC, Sahakian BJ, Evenden
JL, Morris RG, Robbins TW: Impaired extradimensional shift performance in medicated
and unmedicated Parkinsons disease: Evidence for a specific attentional dysfunction.
Neuropsychologia 1989;27:13291343.
75 Owen AM, Sahakian BJ, Hodges JR, Summers
BA, Polkey CE, Robbins TW: Dopamine-dependent frontostriatal planning deficits in early
Parkinsons disease. Neuropsychology 1995;9:
126140.

Early Detection and Differential Diagnosis of

Uploaded by

Document Information

Original Title

Copyright

Available Formats

Share this document

Share or Embed Document

Sharing Options

Did you find this document useful?

Is this content inappropriate?

Copyright:

Available Formats

Early Detection and Differential Diagnosis of

Uploaded by

Copyright:

Available Formats

Original Research Article

Dement Geriatr Cogn Disord 2001;12:265280

Accepted: November 8, 2000