Journal of Clinical and Experimental Neuropsychology

ISSN: 1380-3395 (Print) 1744-411X (Online) Journal homepage: https://www.tandfonline.com/loi/ncen20

Sensitivity of a Preclinical Alzheimer’s Cognitive

Composite (PACC) to amyloid β load in preclinical
Alzheimer’s disease

Lisa Bransby, Yen Ying Lim, David Ames, Christopher Fowler, Joanne
Roberston, Karra Harrington, Peter J. Snyder, Victor L. Villemagne, Olivier
Salvado, Colin L. Masters & Paul Maruff for the AIBL Research Group

To cite this article: Lisa Bransby, Yen Ying Lim, David Ames, Christopher Fowler, Joanne
Roberston, Karra Harrington, Peter J. Snyder, Victor L. Villemagne, Olivier Salvado, Colin L.
Masters & Paul Maruff for the AIBL Research Group (2019): Sensitivity of a Preclinical Alzheimer’s
Cognitive Composite (PACC) to amyloid β load in preclinical Alzheimer’s disease, Journal of
Clinical and Experimental Neuropsychology, DOI: 10.1080/13803395.2019.1593949

To link to this article: https://doi.org/10.1080/13803395.2019.1593949

Published online: 29 Mar 2019.

Submit your article to this journal

View Crossmark data

Full Terms & Conditions of access and use can be found at

https://www.tandfonline.com/action/journalInformation?journalCode=ncen20
JOURNAL OF CLINICAL AND EXPERIMENTAL NEUROPSYCHOLOGY
https://doi.org/10.1080/13803395.2019.1593949

Sensitivity of a Preclinical Alzheimer’s Cognitive Composite (PACC) to amyloid β

load in preclinical Alzheimer’s disease
Lisa Bransbya, Yen Ying Lima, David Amesb,c, Christopher Fowlera, Joanne Roberstona, Karra Harringtona,d,
Peter J. Snydere, Victor L. Villemagnea,f,g, Olivier Salvadoh, Colin L. Mastersa and Paul Maruff for the AIBL
Research Groupa,i
a
The Florey Institute of Neuroscience and Mental Health, The University of Melbourne, Melbourne, VIC, Australia; bNational Ageing Research
Institute, Melbourne, VIC, Australia; cAcademic Unit for Psychiatry of Old Age, St. Vincent’s Health, The University of Melbourne, Melbourne,
VIC, Australia; dCooperative Research Centre for Mental Health, Parkville, Australia; eRyan Institute for Neuroscience, University of Rhode
Island, Kingston, RI, USA; fDepartment of Nuclear Medicine and Centre for PET, Austin Health, Melbourne, VIC, Australia; gDepartment of
Medicine, Austin Health, The University of Melbourne, Melbourne, VIC, Australia; hCommonwealth Scientific Industrial Research
Organization (CSIRO) Preventative Health National Research Flagship, Australian e-Health Research Centre-BiaMedIA, Brisbane, QLD,
Australia; iCogState Ltd., Melbourne, VIC, Australia

ABSTRACT ARTICLE HISTORY

Introduction: Preclinical Alzheimer’s disease (AD) is characterized by amyloid-related cognitive Received 7 April 2018
decline. Reduction in this decline is used to determine the efficacy of drug therapies designed to Accepted 5 March 2019
forestall the disease in preclinical AD clinical trials, measured by a Preclinical Alzheimer’s KEYWORDS
Cognitive Composite (PACC). Most studies estimate rates of cognitive change by comparing Amyloid β; cognitive decline;
cognitively normal (CN) older adults with abnormally high beta-amyloid (Aβ+) to those with Mini-Mental State
low levels (Aβ–). However, participants of preclinical AD clinical trials must be Aβ+ for entry. Examination; Preclinical
Therefore, we estimated the effect of very high amyloid (Aβ++) and Aβ+ on cognitive change Alzheimer’s Cognitive
over three years measured by different versions of the PACC in individuals with preclinical AD. Composite; preclinical
Method: CN older adults underwent Aβ neuroimaging and neuropsychological assessments over Alzheimer’s disease
three years as part of the Australian Imaging, Biomarkers and Lifestyle (AIBL) study. Three
cognitive composite scores were computed: the Alzheimer’s Disease Cooperative Study
(ADCS)–PACC, the ADCS–PACC with no Mini-Mental State Examination (MMSE), and the z-scores
of Attention, Verbal Fluency and Episodic Memory for Nondemented Older Adults (ZAVEN)
composite.
Results: Compared to the Aβ++ group, the Aβ+ group showed a slower rate of cognitive decline
with the largest magnitude of difference reflected by the ADCS–PACC (d = 0.85). The ADCS–PACC
excluding the MMSE and the ZAVEN also reflected a moderate to large magnitude of difference
between groups (d = 0.62, d = 0.72, respectively).
Conclusions: When all individuals have abnormal Aβ, the level of Aβ at baseline is associated
with the rate of subsequent decline. The ADCS–PACC was the most sensitive composite score in
showing that lower Aβ is associated with a slower rate of cognitive decline; however, there are
limitations to the use of the MMSE. These results provide a benchmark of comparison for
preclinical AD clinical trials aiming to slow cognitive deterioration.

There is now agreement that high levels of beta-amyloid
(Aβ+), detected using cerebrospinal fluid (CSF) sampling
or positron emission tomography (PET) neuroimaging, in
cognitively normal (CN) older adults indicate that the
Alzheimer’s disease (AD) process has begun (Jack et al.,
2018; Sperling et al., 2011). A previous study has shown that
it can take an average of 12 years for an individual with low
levels of Aβ (Aβ–) to become Aβ+, and another 19.2 years
to reach a diagnosis of AD (Villemagne et al., 2013). There
is increasing evidence that, compared to Aβ– older adults,
those who are Aβ+ have showed subtle but progressive
cognitive decline, particularly in episodic memory and
executive function, when followed over a period of 12–-
54 months (Baker et al., 2017). The consistency with which
these relationships have been observed across prospective
observational studies has provided the theoretical founda-
tion for secondary prevention clinical trials of Aβ-reducing
drugs, where it has been proposed that preventing the
accumulation of Aβ in individuals without clinical symp-
toms of dementia will delay or even forestall disease pro-
gression in CN adults (Sperling, Jack, & Aisen, 2011;
Sperling, Mormino, & Johnson, 2014; Sperling et al., 2014).
Clinical trials of preclinical AD now use neuropsy-
chological measures to detect underlying biological

CONTACT Lisa Bransby lisa.bransby@florey.edu.au 30 Royal Parade, Parkville, Victoria 3052, Australia
© 2019 Informa UK Limited, trading as Taylor & Francis Group
2 L. BRANSBY ET AL.
mechanisms as their primary endpoint (Sperling et al.,
2014). As such, the Alzheimer’s Disease Cooperative
Study–Preclinical Alzheimer’s Cognitive Composite
(ADCS–PACC) has been developed to measure the
effects of the anti-Aβ medication solanezumab in the
Anti-Amyloid Treatment in Asymptomatic
Alzheimer’s Disease (A4) trial (Donohue et al., 2014;
Sperling et al., 2014). In fact, data on the extent to
which cognition declined in Aβ+ adults compared to
Aβ– adults provided the basis for estimation of statis-
tical power in the A4 study where it was assumed that
the medication would lead to a 30% slowing of dete-
riorating performance on the ADCS–PACC (Sperling
et al., 2014).
One potential limitation with anticipating the nature
and magnitude of effects of anti-Aβ drugs from models
comparing change in cognition between Aβ+ and Aβ–
groups is that, by definition, all individuals in trials of
preclinical AD must be Aβ+. A more accurate under-
standing of the effects of anti-Aβ medication might be
gained from modeling the effects of variation in Aβ
levels on cognitive decline in individuals whose Aβ
levels are abnormally high to begin with—that is, in
preclinical AD. Comparing cognitive change between
individuals with high (Aβ+) and very high (Aβ++)
levels may therefore provide a more informative
model for estimating the potential effects of slowing
Aβ deposition with anti-Aβ medication.
The cognitive endpoint used in the A4 trial, the
ADCS–PACC, is a composite score of performance
on measures of episodic memory and complex atten-
tion and scores from the Mini-Mental State
Examination (MMSE; Donohue et al., 2014).
However, there remains debate about the extent to
which the inclusion of the MMSE in preclinical AD
composite scores affects the sensitivity of such scores
to changes in Aβ levels. This is because in CN older
adults, the MMSE has shown poor test–retest relia-
bility, ceiling effects, and low sensitivity to impair-
ment (Spencer et al., 2013). Thus, including the
MMSE in a composite score might reduce its sensi-
tivity to changes in Aβ levels in CN adults (Lim
et al., 2016; Spencer et al., 2013). Furthermore, as
studies have shown that verbal fluency measures, as
measures of executive function, are sensitive to Aβ-
related change (Insel et al., 2015; Lim et al., 2014;
Lim et al., 2016; Papp et al., 2016; Papp, Rentz,
Orlovsky, Sperling, & Mormino, 2017), it was pro-
posed that the MMSE measure be replaced by
a measure of verbal fluency. This yielded the z-scores
of Attention, Verbal Fluency and Episodic Memory
for Nondemented Older Adults (ZAVEN) composite
(Lim et al., 2016). This ZAVEN composite showed
greater sensitivity to Aβ+ related cognitive decline
than composite scores that include the MMSE (Lim
et al., 2016).
The aim of this study was to understand whether
lower levels of Aβ burden are likely to translate to
slower cognitive decline in Aβ+ CN older adults over
36 months, a similar time period to that of a clinical
trial. This was achieved by estimating differences in the
rate of cognitive change over 36 months between cog-
nitively normal older adults with abnormally high
levels of Aβ (Aβ+) and those with very high levels of
abnormal Aβ (Aβ++). The first hypothesis was that
compared to Aβ++, Aβ+ would be associated with
slower rates of decline in performance on composite
measures of episodic memory and executive function.
The second aim of this study was to determine whether
differences in Aβ+ related cognitive change between
groups manifest differently using composite measures
of episodic memory and executive function that did or
did not include the MMSE. The second hypothesis was
that the exclusion of the MMSE would improve the
sensitivity of cognitive composite scores to Aβ+ related
cognitive decline. Finally, we also aimed to explore how
Aβ level, when treated as a continuous variable, is
related to cognitive decline.
Method
Participants
Participants were recruited from a group of cognitively
normal (CN) older adults over the age of 60 years enrolled
in the Australian Imaging, Biomarkers and Lifestyle (AIBL)
study. The AIBL study is a prospective longitudinal study
seeking to understand the natural history of AD. Details of
recruitment have been described previously (Ellis et al.,
2009). Briefly, exclusion criteria included schizophrenia,
depression (15-item Geriatric Depression Score ≥6),
Parkinson’s disease, symptomatic stroke, uncontrolled dia-
betes, sleep apnea, and alcohol consumption exceeding two
standard drinks per day for women or four per day for
men. Participants undergo medical, psychiatric, and neu-
ropsychological assessments approximately every
18 months. At each assessment, a clinical review panel
considers all available medical, psychiatric and neuropsy-
chological information to classify clinical status (Ellis et al.,
2009). Clinical classification is blinded to neuroimaging
results. Aβ positivity was a key selection criterion for the
current study. This is in line with the aim of this study to
model rates of cognitive decline for different levels of
abnormal Aβ in preclinical AD. As such, including Aβ–
participants who are not considered to be in the preclinical
stage of AD in the analysis would detract from the aim of

this study. Additionally, as current preclinical AD trials
require participants to be Aβ+ for enrolment, the inclusion
of Aβ– participants in the analyses may skew estimations of
cognitive deterioration, thus making it less informative for
power calculations.
Materials and measures
Neuroimaging
Aβ imaging with PET was conducted using one of
three radioligands: 11C-Pittsburgh Compound B (PiB),
18
F-Florbetapir (FBP) or 18F-Flutemetamol (FLUTE).
A PET standardized uptake value ratio (SUVR) was
calculated for each participant. A linear regression
transformation was applied to the SUVR for each
F-18 Aβ tracer to transform it into a “PiB-like” SUVR
unit.8. These were termed BeCKeT (Before the
Centiloid Kernel Transformation; Villemagne et al.,
2014). Participants were excluded if their SUVR/
BeCKeT score was below 1.4. Participants with
a SUVR between 1.4 and 1.9 (inclusive of 1.4) were
classified as high Aβ (Aβ+). Participants with a SUVR
score of 1.9 or more were classified as very high Aβ
(Aβ++). The cutoff value of 1.4 was modified from 1.5,
which has been used previously and was obtained from
a hierarchical cluster analysis (Lim et al., 2016; Rowe
et al., 2013). A cutoff of 1.4 was used in the current
study to be more inclusive when analyzing Aβ+ parti-
cipants only. The cutoff value of 1.9 obtained from
a ROC curve analysis was used as the optimal cut
point to discriminate age-matched CN older adults
from AD patients (Lim et al., 2016; Rowe et al., 2013).
Demographic and clinical characteristics
Age and gender were self-reported. Premorbid IQ was
assessed using the Wechsler Test of Adult Reading
(WTAR; Wechsler, 2001). The Hospital Anxiety and
Depression Scale (HADS; Zigmond & Snaith, 1983) and
the Geriatric Depression Scale (GDS; Yesavage et al., 1982)
were used to measure levels of depression and anxiety. The
Clinical Dementia Rating (CDR) scale was also included to
contribute to the clinical information of the current sample.
Table 1. Description of composite scores and neuropsychological tests included.
Composite score
Alzheimer’s Disease Cooperative Study–
Preclinical Alzheimer’s Cognitive Composite score
Alzheimer’s Disease Cooperative Study–
Preclinical Alzheimer’s Cognitive Composite score excluding the Mini-
Mental State Examination
Z-scores of Attention, Verbal Fluency and Episodic Memory for
Nondemented Older Adults
Note. MMSE = Mini-Mental State Examination; CVLT–II = California Verbal Learning Test–Second Edition; D-KEFS = Delis–Kaplan Executive Function System.
JOURNAL OF CLINICAL AND EXPERIMENTAL NEUROPSYCHOLOGY 3
Neuropsychological assessment
The AIBL test battery comprises neuropsychological tests
that were selected to cover main cognitive domains affected
by AD, and has been described previously (Ellis et al.,
2009). A selection of these have been used to compute the
composite scores. These include: the Mini-Mental State
Examination (MMSE; Folstein, Robins, & Helzer, 1983)
to measure global cognition; California Verbal Learning
Test–Second Edition (CVLT–II) delayed free recall (Delis,
Kramer, Kaplan, & Ober, 2000) to measure episodic mem-
ory; Logical Memory II (Wechsler Memory Scale, WMS;
Story A only) delayed recall (Wechsler, 1945) to measure
episodic memory; Delis–Kaplan Executive Function
System (D-KEFS) verbal (letter) fluency (Delis, Kaplan, &
Kramer, 2001) to measure executive function; and Digit
Symbol–Coding subtests of the Wechsler Adult
Intelligence Scale–Third Edition (WAIS–III; Wechsler,
1997) to measure complex attention (see Table 1).
Calculation of PACC scores
Composite scores were calculated by averaging standar-
dized scores for each outcome measure from the neurop-
sychological tests. The raw scores for the outcome
measures were standardized to z-scores using the baseline
mean and standard deviation derived from the current
sample. The subtests that form the ADCS–PACC were
selected based on a review of the literature that suggested
that the main cognitive domains to show deterioration in
preclinical AD were episodic memory (both list learning
and paragraph recall), executive function, and orientation
(Donohue et al., 2014). As such, a similar approach has
been adopted to the construction of the current multido-
main cognitive composites, and the tests selected are reli-
able and well validated and are widely used across
multiple AD cohorts (Lim et al., 2016).
Procedure
The current study was conducted based on data col-
lected in the AIBL study. The process of consent,
recruitment, and classification has been described in
detail previously (Ellis et al., 2009). The AIBL study has
Abbreviation Neuropsychological tests included
ADCS–PACC MMSE, CVLT–II, Logical Memory II, and Digit
Symbol–Coding
ADCS–PACC no CVLT–II, Logical Memory II, and Digit Symbol–Coding
MMSE
ZAVEN CVLT–II, Logical Memory II, Digit Symbol–Coding, and
D-KEFS verbal fluency
4 L. BRANSBY ET AL.
been approved by the regulations of three institutional
research and ethics committees, with all participants
providing written informed consent prior to participa-
tion. Participants attended the research facility for clin-
ical and neuropsychological assessments every
18 months. Neuropsychological assessments were con-
ducted in 90-min sessions by trained research assistants
according to standardized protocols. For participants
included in the current study, neuroimaging aligned
with one of these time points.
Data analysis
Data selection
Data from participants who had not undergone neu-
roimaging or did not have two subsequent cognitive
assessments following neuroimaging were excluded
from analysis. For the data included, the time point at
which neuroimaging had taken place was now classi-
fied as baseline, and the two subsequent time points
were now classified as 18- and 36-month visits. All data
from other time points were excluded.
Demographic and clinical characteristics
Analysis began by investigating differences in demo-
graphic and clinical characteristics between Aβ+ and
Aβ++ groups to determine whether they needed to be
added as covariates in subsequent analyses. An inde-
pendent samples t test was used to analyze age. A chi-
square test of independence was used to analyze sex.
Due to distributional characteristics a Mann–Whitney
test was used to analyze premorbid IQ, depression, and
anxiety. For all analyses, Aβ status was entered as
a dichotomous and categorical independent variable.
Difference in rates of cognitive change between Aβ+
and Aβ++ groups
Three planned comparisons were conducted using
repeated measures linear mixed-effects models
(LMMs) with maximum likelihood estimation and
an unstructured covariance matrix, for each cogni-
tive composite score. Linear mixed modeling was
used because it is able to model both fixed and
random effects, accounting for variability typical of
longitudinal studies (Lim et al., 2016). In the ana-
lyses, group (Aβ+, Aβ++), time (baseline, 18 months,
36 months) and Group × Time interaction were
entered as fixed factors, participant as a random
factor, and cognitive composite score as the depen-
dent variable. Cohen’s d was calculated to determine
the magnitude of difference between rates of cogni-
tive change between groups over 36 months. These
analyses were repeated with the MMSE and letter
fluency as individual measures entered as the depen-
dent variable. This was done to elicit a comparison
between these two measures, the inclusion or exclu-
sion of which is what differentiates each composite
score.
Relationship between Aβ level as a continuous
variable and cognitive decline
Three planned comparisons were conducted using
repeated measures LMMs for each cognitive composite
score to determine the relationship between Aβ level as
a continuous variable and cognitive change. In the
analyses, SUVR, time (baseline, 18 months, 36 months)
and SUVR × Time interaction were entered as fixed
factors, participant as a random factor, and cognitive
composite score as the dependent variable. Statistical
significance was set at p < .05.
Results
Figure 1 summarizes the data selection process for the
current study. Participants who had not undergone
neuroimaging (n = 343), who did not have two subse-
quent neuropsychological visits following neuroima-
ging (n = 328), and who were classified as being Aβ–
(n = 215) were excluded from the current study. This
resulted in a total of 66 CN participants.
Demographic and clinical characteristics
At baseline, there were no statistically significant dif-
ferences between the Aβ+ and Aβ++ groups on any
demographic or clinical characteristics (Table 2).
However, age was added as a covariate to adjust for
any covariance between age and cognitive decline.
Effect of Aβ group on cognitive change using
cognitive composite scores
The LMM analyses indicated statistically significant
group by time interactions for each composite score.
Table 3 summarizes these interactions and the group
mean slopes for both groups for each cognitive com-
posite score. The group mean slopes for both groups
are displayed in Figure 2 for each composite score.
Relative to the Aβ++ group, the Aβ+ showed a slower
rate of decline on each of the composite scores.
The magnitude of difference between Aβ groups
measured by the cognitive composite scores
The magnitude of the difference in rate of change
between groups for each composite was moderate to
 JOURNAL OF CLINICAL AND EXPERIMENTAL NEUROPSYCHOLOGY 5

group, the magnitude of this difference was slightly less

343 participants who had not
undergone neuroimaging were
excluded
than that observed for the ADCS–PACC (d = 0.85, p
< .01) and the ZAVEN (d = 0.72, p < .01). Reanalyses
with age as a covariate did not alter the nature or
magnitude of group differences on the composite
scores.

Comparison of the effect of Aβ on MMSE and

328 participants who did not have
two subsequent
neuropsychological visits
following neuroimaging were
excluded
verbal fluency individual scores over 36 months
The LMM analyses did not indicate statistically signifi-
cant group by time interactions on the MMSE or verbal
fluency scores. Similarly, both the MMSE (d = 0.15, p
= .54) and verbal fluency (d = 0.24, p = .34) reflected
small and statistically insignificant magnitudes of dif-
ference between Aβ groups.

(N=281)
215 Aβ- participants were
excluded
Figure 1. Sample selection process. CN = cognitively normal;
Aβ+ = beta-amyloid positive; Aβ– = beta-amyloid negative;
AIBL = Australian Imaging, Biomarkers and Lifestyle study.
Relationship between Aβ level at baseline and
cognitive performance
The LMMs indicated statistically significant relation-
ships between SUVR at baseline and each composite
score over 36 months. Similar to the previous analyses,
the ADCS–PACC reflected a z-score change of −0.42
(0.78) per 18 months associated with each unit increase
in SUVR, which was larger than the other composite
scores. However, the rate of decline on each composite
score was not statistically significant from each other
(95% confidence interval, CI, overlap). Table 3 and
Figure 3 summarize these results.

Discussion
large. While cognitive decline measured using the The first hypothesis that, compared to CN older adults
ADCS–PACC with no MMSE (d = 0.62, p = .01) was with very high levels of abnormal Aβ (Aβ++), CN older
significantly less in the Aβ+ group than in the Aβ++ adults with high levels (Aβ+) would show a slower rate

Table 2. Demographic and clinical characteristics of the sample by Aβ status.

CN Aβ+ (n = 34) CN Aβ++ (n = 32)
M (SD) n (%) M (SD) n (%) p
Female sex 18 (52.9) 13 (40.6) .316
Age (years) 72.47 (6.57) 75.19 (6.89) .106
Premorbid IQ 109.73 (5.96) 110.47 (6.47) .345
GDS 0.61 (0.97) 0.75 (0.88) .322
HADS Depression 2.41 (2.02) 2.97 (2.49) .372
HADS Anxiety 3.62 (2.10) 3.81 (3.14) .887
CDR 0.03 (0.12) 0.03 (0.12) .951
CDR sum of boxes 0.03 (0.12) 0.03 (0.12) .951
MMSE 28.74 (1.02) 28.91 (1.17) .530
Digit Symbol 57.71 (13.76) 57.06 (13.69) .964
Logical Memory II 11.97 (5.03) 10.48 (3.93) .272
CVLT–II 11.88 (3.46) 11.56 (3.06) .442
Letter Fluency (FAS) 42.82 (12.69) 45.22 (9.08) .019
Note. Aβ = beta-amyloid; CN = cognitively normal; GDS = Geriatric Depression Scale; HADS = Hospital Anxiety and Depression Scale;
CDR = Clinical Dementia Rating Scale; MMSE = Mini-Mental State Examination; CVLT–II = California Verbal Learning Test delayed recall.
6 L. BRANSBY ET AL.

Table 3. Effect of abnormal Aβ level, as a group and continuous variable, on cognitive decline over 36 months in CN older adults.
CN Aβ+ CN Aβ++
Group Time Group × Time (n = 34) n = 32) SUVR
Slope Slope
F (df) p F (df) p F (df) p M (SD) M (SD) slope M (SD) p
ADCS–PACC 1.19 (1, 139.25) .277 20.44 (1, 129.24) <.001 11.92 (1, 129.24) <.001 −0.04 (0.29) −0.29 (0.29) −0.42 (0.78) <.001
ADCS–PACC (with no 0.22 (1, 120.22) .638 25.10 (1, 128.17) <.001 6.29 (1, 128.17) .013 −0.08 (0.25) −0.23 (0.25) −0.35 (0.67) <.001
MMSE)
ZAVEN 0.80 (1, 113.14) .372 19.91 (1, 129.16) <.001 8.49 (1, 129.16) .004 −0.04 (0.22) −0.20 (0.22) −0.32 (0.58) <.001
Note. Aβ = beta-amyloid; CN = cognitively normal; SUVR = standardized uptake value ratio; ADCS–PACC = Alzheimer’s Disease Cooperative Study–Preclinical
Alzheimer’s Cognitive Composite score; MMSE = Mini-Mental State Examination; ZAVEN = Z-scores of Attention, Verbal Fluency and Episodic Memory for
Nondemented Older Adults. Mean slope is calculated as z-score decline per year.

a. of cognitive decline was supported. Decline over three

b.
c.
Figure 2. Differences in rates of cognitive decline measured by (a)
the ADCS–PACC, (b) the ADCS–PACC excluding the MMSE, and (c)
the ZAVEN between Aβ groups over 36 months. Slope is calculated
as z-score decline per year. Shaded regions represent 95% CIs.
ADCS–PACC = Alzheimer’s Disease Cooperative Study–Preclinical
Alzheimer’s Cognitive Composite score; MMSE = Mini-Mental State
Examination; ZAVEN = Z-scores of Attention, Verbal Fluency and
Episodic Memory for Nondemented Older Adults; Aβ = beta-
amyloid; CI = confidence interval; BL = baseline.
years in performance on each of the composite scores
studied was less in the Aβ+ group than in the Aβ++
groups. The effect size observed for differences in this
rate of cognitive decline varied between 0.62 and 0.85
(Table 3, Figure 2). The error associated with the
estimate of difference in rates of cognitive decline for
each composite overlapped, indicating that these did
not differ from one another. These data show that
within preclinical AD, lower Aβ levels at baseline are
associated with slower rates of cognitive decline.
The second hypothesis that the exclusion of the
MMSE from composite measures of cognitive func-
tion would improve their sensitivity to Aβ+ related
cognitive decline was not supported. The removal of
the MMSE from the ADCS–PACC reduced the mag-
nitude of difference in rate of cognitive change
detected between groups, with Cohen’s d decreasing
from 0.85 to 0.62. The finding that the MMSE did
contribute to the sensitivity of cognitive composite
scores to Aβ-related cognitive decline in
preclinical AD is consistent with the data showing
the validity of the ADCS–PACC initially (Donohue
et al., 2014). However, in these analyses, claims about
sensitivity of the ADCS–PACC were based on com-
parisons of performance between Aβ– and Aβ+ indi-
viduals. Given that Aβ+ is a necessary criterion for
preclinical AD, this study was comparing cognitive
decline between individuals for whom AD had already
begun and individuals without AD. The current data
extend this by showing that when all individuals are
Aβ+ (i.e., meet criteria for preclinical AD), the MMSE
does play a role in the sensitivity of the ADCS–PACC
to Aβ.
Our exploration of Aβ level at baseline, when treated
as a continuous measure, showed that a higher SUVR
was associated moderately with worsening cognitive
performance over time on each of the cognitive com-
posite scores. These results provide further evidence
 JOURNAL OF CLINICAL AND EXPERIMENTAL NEUROPSYCHOLOGY
a.
b.
c.
Figure 3. The relationship between Aβ level at baseline, as
a continuous measure, and cognitive performance on (a) the
ADCS–PACC, (b) the ADCS–PACC excluding the MMSE, and (c)
the ZAVEN over 36 months. Slope is calculated as z-score
decline per year. Shaded regions represent 95% CIs. ADCS–
PACC = Alzheimer’s Disease Cooperative Study–Preclinical
Alzheimer’s Cognitive Composite score; MMSE = Mini-Mental
State Examination; ZAVEN = Z-scores of Attention, Verbal
Fluency and Episodic Memory for Nondemented Older Adults;
Aβ = beta-amyloid; CI = confidence interval; SUVR = standar-
dized uptake value ratio.
that when all individuals have abnormal Aβ, the level
of Aβ is associated with cognitive decline.
The current study is one of the first to show that
variation in the extent of Aβ positivity, even when all
individuals are Aβ+, is associated with the nature and
rate of subsequent cognitive decline (Lim et al., 2016).
Specifically, for individuals whose SUVR was above 1.4
but below 1.9, the rate of decline detected by all
7
composite scores was lower than that observed in indi-
viduals with a SUVR of 1.9 or higher at baseline.
However, there was substantial variability surrounding
the differences in the rate of decline between groups
(Figure 2), which was likely due to the small sample
size of the current study and reinforces the need for the
results to be replicated in a larger group.
In the context of AD, Aβ levels are not associated
with clinical characteristics and, in particular, with
level of cognitive impairment. This conclusion is
based mainly on postmortem studies that have
reported the strength of associations between Aβ pla-
que counts and clinical disease severity prior to death,
as well as on data from studies of living patients on
associations between SUVR and performance on neu-
ropsychological tests (Aizenstein et al., 2008; Bennett
et al., 2006). Importantly, many clinical trials targeting
Aβ in participants with mild to moderate AD have
failed to show any improvement on cognition (Egan
et al., 2018). Such data have been used to argue that
clinical trials of Aβ lowering medications are mis-
guided (Herrup, 2015). However, the results from the
present study indicate that in individuals with
preclinical AD, higher Aβ levels are associated with
more rapid cognitive decline. Thus, keeping Aβ levels
low at this preclinical stage is likely to be associated
with slower disease progression (Sperling et al., 2011;
Sperling et al., 2014).
The current results also show that the magnitude of
association between Aβ level and subsequent cognitive
decline is related to the combination of tests used to
measure cognition. Both the ADCS–PACC (d = 0.85)
and ZAVEN (d = 0.72) composite scores showed
a large effect of Aβ load in the preclinical AD sample.
However, previous comparisons have indicated that the
inclusion of the MMSE in a composite score will
reduce its sensitivity to cognitive change (Lim et al.,
2016). In CN older adults, the MMSE is known to have
poor test–retest reliability, ceiling effects, and restricted
sensitivity to subtle cognitive impairment (Spencer
et al., 2013). However, the ADCS–PACC that includes
the MMSE (d = 0.85) reflected the largest effect size out
of all composite scores in the current study, and
removing the MMSE from the ADCS–PACC reduced
its sensitivity to differences in rates of cognitive decline
associated with Aβ level (d = 0.62). This controversial
result needs to be considered taking into account the
integrity of the MMSE. Previous analysis has demon-
strated that the MMSE is not a robust or reliable
assessment tool, as patients with moderate to severe
memory impairment were still able to achieve near to
perfect scores (Lacy, Kaemmerer, & Czipri, 2014).
Therefore, the MMSE might accurately screen the
8 L. BRANSBY ET AL.
mental state of the subject at the time, but it might fail
to reliably detect any underlying pathology.
The MMSE was excluded from the ZAVEN on the
basis that no studies have shown performance on this
test to be abnormal, or to decline over time in
preclinical AD (Lim et al., 2016). Further, a previous
study showed that adding a measure of letter fluency to
a cognitive composite score increased its sensitivity to
Aβ-related cognitive decline, and this was replicated in
the current results showing an increase in Cohen’s
d from 0.62 to 0.72 (Lim et al., 2016). In the current
study there was a statistically significant difference on
letter fluency scores between Aβ groups at baseline,
with the Aβ++ group performing on average three
words better than the Aβ+ group (Table 2). After
36 months the Aβ++ group performed slightly worse
on the measure than the Aβ+ group. However, this
difference was not statistically significant over time (d
= 0.24, p = .34). There was also no significant differ-
ence between Aβ groups on the MMSE (d = 0.15, p
= .54) as an individual measure. These results were
expected given that the ADCS–PACC and ZAVEN
did not significantly differ from one another.
However, the ZAVEN should be considered as
a better composite score as it still demonstrated sensi-
tivity to the difference between Aβ groups (d = 0.72, p
< .01), is composed of measures that are not prone to
ceiling effects (unlike the MMSE), and is more reliable,
demonstrated by tighter standard deviations than the
ADCS–PACC, suggesting that the inclusion of the
MMSE in the ADCS–PACC is associated with more
variance.
There are limitations in the current study that war-
rant consideration. First, the sample size of the current
study was relatively small and is likely to have played
a role in the large variability around the results.
Therefore, it is important for results to be replicated
in a larger sample of CN older adults with abnormal
Aβ. Additionally, the AIBL study is not an epidemio-
logical sample, but rather a convenience sample
(Brodaty et al., 2014). As a result, cognitively normal
participants recruited were well educated and had few
untreated medical or psychiatric illnesses. On average,
premorbid IQ was relatively high. Therefore, it would
also be important for these findings to be replicated in
population-based samples. Another limitation of cur-
rent older cohort studies of preclinical Alzheimer’s
disease is that a substantial portion (30%) already
meet criteria for Aβ positivity. As it is not possible to
determine how long these individuals have been Aβ+
for, we have used baseline Aβ levels as an estimation
(i.e., someone with higher Aβ levels may have been Aβ
+ for longer). Future studies that conduct serial Aβ
assessments on Aβ– adults will clarify whether number
of years of Aβ+ provides predictive information over
and above that obtained from continuous Aβ levels, as
has been presented in this study. A final limitation of
the current study is that Aβ is included as the sole
biomarker, and was only measured at a single time
point. Future studies should explore the role of bio-
markers and their relationship with cognition, and the
relationship between change in biomarker levels and
subsequent change in cognitive function.
The current results demonstrate that there is impor-
tant information that needs to be understood beyond
the traditional dichotomy of Aβ levels as positive or
negative (Lim et al., 2016). Our results show that even
within the Aβ+ group, there remains much variance in
the rate of cognitive decline. Other studies have also
shown that the rate of Aβ accumulation is related to
the rate of cognitive decline (Villemagne et al., 2013).
As these factors have implications for clinical trial out-
comes, future studies should investigate the role of
both Aβ levels at baseline and rate of Aβ accumulation
in predicting cognitive decline in CN older adults.
These limitations notwithstanding, the current study
demonstrated that when CN individuals have abnor-
mal Aβ, the level of Aβ at baseline is associated with
the rate of subsequent cognitive decline. Our results
suggest that in Aβ+ CN older adults, lower levels of Aβ
burden are likely to translate to slower cognitive
decline over a time period similar to that of a clinical
trial. Differences in the rate of cognitive change over
36 months between CN older adults with abnormally
high levels of Aβ (Aβ+) and those with very high levels
of abnormal Aβ (Aβ++) were estimated for three cog-
nitive composite scores. It was also demonstrated that
Aβ level, as a continuous variable, is related to cogni-
tive decline over 36 months. Although the ADCS–
PACC reflected the largest difference between Aβ
groups, previous research suggests that the MMSE is
not a reliable measure of cognitive decline in CN older
adults. Thus, the ZAVEN might provide better utility
for characterizing Aβ-associated rates of cognitive
change in preclinical AD and will likely be sensitive
to detecting the benefits in cognition associated with
lowering abnormal levels of Aβ in a clinical trial. Taken
together, these results provide a benchmark of compar-
ison for preclinical AD clinical trials that aim to slow
cognitive deterioration.
Acknowledgments
Alzheimer’s Australia (Victoria and Western Australia)
assisted with promotion of the study and the screening of
telephone calls from volunteers. The AIBL team wishes to
 JOURNAL OF CLINICAL AND EXPERIMENTAL NEUROPSYCHOLOGY
thank the clinicians who referred patients with AD to the
study: Brian Chambers, Edmond Chiu, Roger Clarnette,
David Darby, Mary Davison, John Drago, Peter Drysdale,
Jacqueline Gilbert, Kwang Lim, Nicola Lautenschlager, Dina
LoGiudice, Peter McCardle, Steve McFarlane, Alastair
Mander, John Merory, Daniel O’Connor, Ron Scholes,
Mathew Samuel, Darshan Trivedi, and Michael Woodward.
We thank all those who participated in the study for their
commitment and dedication to helping advance research into
the early detection and causation of AD.
Disclosure statement
C.L.M. is an advisor to Prana Biotechnology Ltd and
a consultant to Eli Lilly. P.J.S. is a scientific consultant to
Cogstate Ltd. P.M. is a full-time employee of Cogstate Ltd. D.
A. has served on scientific advisory boards for Novartis, Eli
Lilly, Janssen, and Pfizer Inc. V.L.V. served as a consultant
for Bayer Pharma.
Funding
Funding for the study was provided in part by the study
partners: Commonwealth Scientific Industrial and research
Organization (CSIRO), Edith Cowan University (ECU),
Mental Health Research institute (MHRI), National Ageing
Research Institute (NARI), Austin Health, CogState Ltd. The
study also received support from the National Health and
Medical Research Council (NHMRC); the Dementia
Collaborative Research Centres program (DCRC2); the
Science and Industry Endowment Fund (SIEF); and the
Cooperative Research Centre for Mental Health (CRCMH);
and the New Energy and Industrial Technology
Development Organization (NEDO) Japan [grant to V.L.V.].
References
Aizenstein, H. J., Nebes, R. D., Saxton, J. A., Price, J. C.,
Mathis, C. A., Tsopelas, N. D., … Klunk, W. E. (2008).
Frequent amyloid deposition without significant cognitive
impairment among the elderly. Archives of Neurology, 65
(11), 1509–1517.
Baker, J. E., Lim, Y. Y., Pietrzak, R. H., Hassenstab, J.,
Snyder, P. J., Masters, C. L., & Maruff, P. (2017).
Cognitive impairment and decline in cognitively normal
older adults with high amyloid-β: A meta-analysis.
Alzheimer’s & Dementia (Amsterdam, Netherlands), 6,
108–121.
Bennett, D. A., Schneider, J. A., Arvanitakis, Z., Kelly, J. F.,
Aggarwal, N. T., Shah, R. C., & Wilson, R. S. (2006).
Neuropathology of older persons without cognitive
impairment from two community-based studies.
Neurology, 66(12), 1837–1844.
Brodaty, H., Mothakunnel, A., de Vel-Palumbo, M., Ames,
D., Ellis, K. A., Reppermund, S., … Sachdev, P. S. (2014).
Influence of population versus convenience sampling on
sample characteristics in studies of cognitive aging. Annals
of Epidemiology, 24(1), 63–71.
9
Delis, D., Kaplan, E., & Kramer, J. (2001). Delis-Kaplan
executive function system (D-KEFS). Canadian Journal
of School Psychology, 20(1–2), 117–128.
Delis, D. C., Kramer, J. H., Kaplan, E., & Ober, B. A. (2000).
California verbal learning test – Second edition. Adult
version. manual. Test.
Donohue, M. C., Sperling, R. A., Salmon, D. P., Rentz, D.
M., Raman, R., Thomas, R. G., … Aisen, P. S. (2014).
The preclinical Alzheimer cognitive composite. Journal
of the American Medical Association Neurology, 71(8),
961.
Egan, M. F., Kost, J., Tariot, P. N., Aisen, P. S., Cummings, J. L.,
Vellas, B., … Michelson, D. (2018). Randomized trial of ver-
ubecestat for mild-to-moderate Alzheimer’s disease. The New
England Journal of Medicine, 378(18), 1691–1703.
Ellis, K. A., Bush, A. I., Darby, D., De Fazio, D., Foster, J.,
Hudson, P., … Ames, D. (2009). The Australian imaging,
biomarkers and lifestyle (AIBL) study of aging:
Methodology and baseline characteristics of 1112 indivi-
duals recruited for a longitudinal study of Alzheimer’s
disease. International Psychogeriatrics, 21(4), 672.
Folstein, M. F., Robins, L. N., & Helzer, J. E. (1983). The
mini-mental state examination. Archives of General
Psychiatry, 40(7), 812.
Herrup, K. (2015). The case for rejecting the amyloid cascade
hypothesis. Nature Neuroscience, 18(6), 794–799.
Insel, P. S., Mattsson, N., Mackin, R. S., Kornak, J., Nosheny,
R., Tosun-Turgut, D., … Weiner, M. W. (2015).
Biomarkers and cognitive endpoints to optimize trials in
Alzheimer’s disease. Annals of Clinical and Translational
Neurology, 2(5), 534–547.
Jack, C. R., Bennett, D. A., Blennow, K., Carrillo, M. C.,
Dunn, B., Haeberlein, S. B., & Silverberg, N. (2018). NIA-
AA research framework: Toward a biological definition of
Alzheimer’s disease. Alzheimer’s and Dementia, 14, 535–
562.
Lacy, M., Kaemmerer, T., & Czipri, S. (2014). Standardized
mini-mental state examination scores and verbal memory
performance at a memory center: Implications for cogni-
tive screening. American Journal of Alzheimer’s Disease
and Other Dementias, 30(2), 145–152.
Lim, Y. Y., Maruff, P., Pietrzak, R. H., Ames, D., Ellis, K. A.,
Harrington, K., … Rowe, C. C. (2014). Effect of amyloid
on memory and non-memory decline from preclinical to
clinical Alzheimer’s disease. Brain, 137(1), 221–231.
Lim, Y. Y., Snyder, P. J., Pietrzak, R. H., Ukiqi, A., Villemagne, V.
L., Ames, D., … Maruff, P. (2016). Sensitivity of composite
scores to amyloid burden in preclinical Alzheimer’s disease:
Introducing the Z-scores of attention, verbal fluency, and
episodic memory for nondemented older adults composite
score. Alzheimer’s and Dementia: Diagnosis, Assessment and
Disease Monitoring, 2, 19–26.
Papp, K. V., Mormino, E. C., Amariglio, R. E., Munro, C., Dagley,
A., Schultz, A. P., … Rentz, D. M. (2016). Biomarker validation
of a decline in semantic processing in preclinical alzheimer’s
disease. Neuropsychology, 30(5), 624–630.
Papp, K. V., Rentz, D. M., Orlovsky, I., Sperling, R. A., &
Mormino, E. C. (2017). Optimizing the preclinical
Alzheimer’s cognitive composite with semantic proces-
sing: The PACC5. Alzheimer’s and Dementia:
Translational Research and Clinical Interventions, 3(4),
668–677.
10 L. BRANSBY ET AL.
Rowe, C. C., Bourgeat, P., Ellis, K. A., Brown, B., Lim, Y. Y.,
Mulligan, R., … Villemagne, V. L. (2013). Predicting
Alzheimer disease with beta-amyloid imaging: Results
from the Australian imaging, biomarkers, and lifestyle
study of ageing. Annals of Neurology, 74(6), 905–913.
Spencer, R. J., Wendell, C. R., Giggey, P. P., Katzel, L. I.,
Lefkowitz, D. M., Siegel, E. L., & Waldstein, S. R. (2013).
Psychometric limitations of the mini-mental state exam-
ination among nondemented older adults: An evaluation
of neurocognitive and magnetic resonance imaging corre-
lates. Experimental Aging Research, 39(4), 382–397.
Sperling, R., Mormino, E., & Johnson, K. (2014). The evolu-
tion of preclinical Alzheimer’s disease: Implications for
prevention trials. Neuron, 84, 608–622.
Sperling, R. A., Aisen, P. S., Beckett, L. A., Bennett, D. A.,
Craft, S., Fagan, A. M., … Phelps, C. H. (2011). Toward
defining the preclinical stages of Alzheimer’s disease:
Recommendations from the National Institute on Aging-
Alzheimer’s Association workgroups on diagnostic guide-
lines for Alzheimer’s disease. Alzheimer’s and Dementia.
doi:10.1016/j.jalz.2011.03.003
Sperling, R. A., Jack, C. R., & Aisen, P. S. (2011). Testing the
right target and right drug at the right stage. Science
Translational Medicine, 3(111), 111cm33–111cm33.
Sperling, R. A., Rentz, D. M., Johnson, K. A., Karlawish, J.,
Donohue, M., Salmon, D. P., & Aisen, P. (2014). The A4
study: Stopping AD before symptoms begin? Science
Translational Medicine, 6(228), 228fs13–228fs13.
Villemagne, V. L., Burnham, S., Bourgeat, P., Brown, B., Ellis,
K. A., Salvado, O., … Masters, C. L. (2013). Amyloid β
deposition, neurodegeneration, and cognitive decline in
sporadic Alzheimer’s disease: A prospective cohort study.
The Lancet Neurology, 12(4), 357–367.
Villemagne, V. L., Dore, V., Yates, P., Brown, B., Mulligan,
R., Bourgeat, P., et al. (2014). En attendant centiloid.
Journal of Advanced Research, 2, 723–729.
Wechsler, D. (1945). A standardized memory scale for clin-
ical use. The Journal of Psychology, 19(1), 87–95.
Wechsler, D. (1997). WAIS-III administration and scoring
manual. San Antonio, TX: The Psychological
Corporation. doi:10.1177/1073191102009001003
Wechsler, D. (2001). Wechsler test of adult reading:
Examiner’s manual. San Antonio, TX: The Psychological
Corporation.
Yesavage, J. A., Brink, T. L., Rose, T. L., Lum, O., Huang, V.,
Adey, M., & Leirer, V. O. (1982). Development and vali-
dation of a geriatric depression screening scale: A preli-
minary report. Journal of Psychiatric Research, 17(1), 37–
49.
Zigmond, A. S., & Snaith, R. P. (1983). The hospital anxiety
and depression scale. Acta Psychiatrica Scandinavica, 67
(6), 361–370.