ArticlePDF Available

Cognitive dedifferentiation as a function of cognitive impairment in the ADNI and MemClin cohorts

Authors:

Abstract and Figures

The cause of cognitive dedifferentiation has been suggested as specific to late-life abnormal cognitive decline rather than a general feature of aging. This hypothesis was tested in two large cohorts with different characteristics. Individuals (n = 2710) were identified in the Alzheimer's Disease Neuroimaging Initiative (ADNI) research database (n = 1282) in North America, and in the naturalistic multi-site MemClin Project database (n = 1223), the latter recruiting from 9 out of 10 memory clinics in the greater Stockholm catchment area in Sweden. Comprehensive neuropsychological testing informed diagnosis of dementia, mild cognitive impairment (MCI), or subjective cognitive impairment (SCI). Diagnosis was further collapsed into cognitive impairment (CI: MCI or dementia) vs no cognitive impairment (NCI). After matching, loadings on the first principal component were higher in the CI vs NCI group in both ADNI (53.1% versus 38.3%) and MemClin (33.3% vs 30.8%). Correlations of all paired combinations of individual tests by diagnostic group were also stronger in the CI group in both ADNI (mean inter-test r = 0.51 vs r = 0.33, p < 0.001) and MemClin (r = 0.31 vs r = 0.27, p = 0.042). Dedifferentiation was explained by cognitive impairment when controlling for age, sex, and education. This finding replicated across two separate, large cohorts of older individuals. Knowledge that the structure of human cognition becomes less diversified and more dependent on general intelligence as a function of cognitive impairment should inform clinical assessment and care for these patients as their neurodegeneration progresses.
Content may be subject to copyright.
www.aging-us.com 1 AGING
www.aging-us.com AGING 2021, Vol. 13, Advance
Research Paper
Cognitive dedifferentiation as a function of cognitive impairment in
the ADNI and MemClin cohorts
John Wallert1,2, Anna Rennie1,3, Daniel Ferreira1, J-Sebastian Muehlboeck1, Lars-Olof Wahlund1,
Eric Westman1,4,*, Urban Ekman1,5,*, and ADNI consortium, and MemClin steering committee#
1Center for Alzheimer Research, Division of Clinical Geriatrics, Department of Neurobiology, Care Sciences, and
Society, Karolinska Institutet, Stockholm, Sweden
2Centre for Psychiatry Research, Department of Clinical Neuroscience, Karolinska Institutet, Stockholm,
Sweden
3Theme Aging, Karolinska University Hospital, Stockholm, Sweden
4Department of Neuroimaging, Centre for Neuroimaging Sciences, Institute of Psychiatry, Psychology and
Neuroscience, King´s College, London, UK
5Medical Unit Medical Psychology, Allied Health Professionals Function, Karolinska
University Hospital, Stockholm,
Sweden
*Shared last authorship
#Data used in preparation of this article were obtained from the Alzheimers’s Disease Neuroimaging Initiative
(ADNI) database (https://adni.loni.usc.edu). As such, the investigators within the ADNI contributed to the design
and implementation of ADNI and/or provided data but did not participate in analysis or writing of this report. A
complete listing of ADNI investigators can be found at: http://adni.loni.usc.edu/wp-
content/uploads/how_to_apply/ADNI_Authorship_List.pdf
Data used in preparation of this article were also obtained from the Stockholm MemClin study. A complete
listing of the MemClin steering committee can be found at: https://ki.se/en/nvs/westman-neuroimaging-group
Correspondence to: Urban Ekman; email: urban.ekman@ki.se
Keywords: aging, cognitive decline, dotage, neurodegeneration, prodromal dementia
Received: December 23, 2020 Accepted: May 13, 2021 Published: May 26, 2021
Copyright: © 2021 Wallert et al. This is an open access article distributed under the terms of the
Creative Commons
Attribution License (CC BY 3.0), which permits unrestricted use, distribution, and reproduction in any medium,
provided the
original author and source are credited.
ABSTRACT
The cause of cognitive dedifferentiation has been suggested as specific to late-life abnormal cognitive
decline
rather than a general feature of aging. This hypothesis was tested in two large cohorts with
different
characteristics. Individuals (n = 2710) were identified in the Alzheimer’s Disease Neuroimaging Initiative
(ADNI)
research database (n = 1282) in North America, and in the naturalistic multi-site MemClin Project database (n
=
1223), the latter recruiting from 9 out of 10 memory clinics in the greater Stockholm catchment area in
Sweden.
Comprehensive neuropsychological testing informed diagnosis of dementia, mild cognitive impairment
(MCI),
or subjective cognitive impairment (SCI). Diagnosis was further collapsed into cognitive impairment (CI: MCI
or
dementia) vs no cognitive impairment (NCI). After matching, loadings on the first principal component
were
higher in the CI vs NCI group in both ADNI (53.1% versus 38.3%) and MemClin (33.3% vs 30.8%). Correlations
of
all paired combinations of individual tests by diagnostic group were also stronger in the CI group in both
ADNI
(mean inter-test r = 0.51 vs r = 0.33, p < 0.001) and MemClin (r = 0.31 vs r = 0.27, p = 0.042).
Dedifferentiation
was explained by cognitive impairment when controlling for age, sex, and education. This finding
replicated
across two separate, large cohorts of older individuals. Knowledge that the structure of human
cognition
becomes less diversified and more dependent on general intelligence as a function of cognitive
impairment
should inform clinical assessment and care for these patients as their neurodegeneration progresses.
www.aging-us.com 2 AGING
INTRODUCTION
The age dedifferentiation hypothesis suggests that as
humans age we become more reliant on general
intelligence (g) for the different cognitive functions [1].
This is a somewhat contested idea in the geriatric field
where it has been suggested that underlying pathology
accounts for dedifferentiation [2]. Based on early
research on dedifferentiation of cognitive abilities in later
life, [3] this hypothesis suggests that the influence of g on
cognitive test performance increases as the biological
constraint that comes with old age renders specific
cognitive abilities more similar [4]. When
dedifferentiation is combined with the (i) known
differentiation of cognitive functioning that occurs due to
skill specialization from young age to adulthood, and (ii)
stability of adult cognitive functioning from 18-65 years
of age, [5] a conceptual U-type lifespan plot can be
drawn (Figure 1) similar to Craik and Bialystok [6].
In the neuroscience literature, age-related neural
dedifferentiation is a fairly robust phenomenon [7].
However, ascribing aging itself causal agency for
dedifferentiation is probably too coarse [6]. Studies
have shown substantial cognitive dedifferentiation in
older samples but only in those with suspected [8] or
diagnosed [9] abnormal neurodegenerative decline. An
age indifferentiation hypothesis has also been proposed
that stresses the stability of cognitive ability over time
[10]. Aside from aging, other potential causes of
dedifferentiation have been described, such as
educational attainment [11]. Control for education and
other potential confounders is needed if one wants to
estimate the unique contribution of neurodegeneration
to dedifferentiation [8].
Dedifferentiation seems in part specific to abnormal
cognitive decline and, consequently, of specific interest
for clinical geriatric populations. Among patients with
subjective cognitive complaints seeking healthcare,
subjective cognitive impairment (SCI) is distinguished
from cognitive impairment (CI) i.e. either Mild
Cognitive Impairment (MCI) or dementia diagnosis
through comprehensive multidisciplinary investigation at
a specialized hospital unit, usually a Memory Clinic [12]
To our knowledge, the dedifferentiation hypothesis has
not been tested in a large, ecologically valid and
representative sample of memory clinic patients.
Dedifferentiation in these patients, whom have all
undergone the memory clinic investigation and been
diagnosed as either CI (MCI or dementia) or No CI
(SCI), can be studied and the association of CI with
dedifferentiation estimated.
The present study thus tested the dedifferentiation
hypothesis as a function of CI in two cohorts, each of
considerable size and each including both individuals
with CI and NCI, but with important differences
regarding their geographical location (ADNI: North
America [13], MemClin: Sweden) [12], diagnostic
setting (research setting; clinical setting), sample
selection, and data characteristics. After controlling for
age, education, and sex, we hypothesized that as CI
comes into play, dedifferentiation is greater and the
influence of g on cognitive test performance is higher.
We also expected dedifferentiation to be more clear in
the ADNI database compared to MemClin since ADNI
applied a set of additional exclusion criteria to a
priori differentiate diagnostic groups whereas
MemClin did not have such exclusion but instead by
design includes patients that are the most difficult to
diagnostically differentiate requiring full memory
clinic investigation.
RESULTS
Summary statistics for patients subgrouped by dataset
and by group are available in Table 1. Within each
dataset, patients with CI were older, had completed
fewer years of education, more likely male, and
performed worse on psychometric tests, compared to
NCI patients. Across datasets, patients were similar in
age but ADNI patients had more years of education.
ADNI patient counts were more evenly dispersed across
groups relative to MemClin. In addition, differences in
psychometric performance across CI and NCI groups
were slightly more pronounced in ADNI compared to
MemClin.
Table 2 shows a higher proportion of variance in
psychometric test scores explained by PC1 for the
CI group vs NCI, across both ADNI and MemClin.
Applying the Kaiser rule retained one less PC
for the CI group vs NCI in ADNI. However,
equal amount of PCs were retained across groups
in MemClin. Regarding individual PC1 test loadings
for key cognitive domains, the averages were
slightly higher and less variable across domains
for the CI group vs NCI in ADNI but not in
MemClin.
All informative test pair correlations (105 for ADNI;
190 for MemClin) sorted from weakest to strongest
magnitude stratified by database and group are
depicted in Figure 2 showing dedifferentiation by
impairment in both datasets, i.e. generally stronger
linear associations between test pairs among CI vs NCI
patients (ADNI: mean inter-test r = 0.51 vs r = 0.33,
p < 0.001) and MemClin: r = 0.31 vs r = 0.27,
p = 0.042). We also see that the dedifferentiation
pattern is more pronounced in ADNI compared to
MemClin but present in both.
www.aging-us.com 3 AGING
DISCUSSION
After controlling for age, education and sex through
propensity score matching, dedifferentiation was
specifically associated with cognitive impairment in the
present study. This was shown in two large, high-
quality cohorts that differed in several aspects; the
research oriented ADNI and the naturalistic cohort
MemClin. Our findings are in line with previous studies
showing age-related dedifferentiation of cognitive
abilities, yet does not focus on aging with respect to
dedifferentiation. Instead, our findings corroborate
previous research highlighting the role of impairment of
cognitive functions as explanatory for cognitive
dedifferentiation in the latter portion of life, rather than
explained by aging per se [8, 9].
Research on dedifferentiation, as the phenomenon is
defined and investigated with data, predominantly
focuses on cognitive change processes that are global in
their nature. There are however more specific and also
subtle changes to human cognition in late life.
Cognitive reorganization has been suggested, [14] as
well as findings suggesting that differentiation and
dedifferentiation processes can be simultaneously
ongoing as part of a cognitive restructuring process that
is compensatory to newly developed cognitive
deficiency [15]. There is also the concept of cognitive
reserve suggesting interindividual differences with
regards to the amount of complex cognitive activity
experienced during life. This experience in turn
determines interindividual differences in accumulated
cognitive reserve which functions as a flexible and
active buffer to cognitive decline, for some but not for
others, as neurodegeneration progresses [16]. There are
likely both fixed and modifiable factors producing
variability in cognitive reserve across individuals that
may in turn influence dedifferentiation caused by
neurodegeneration. In our study we did control for
education and thus also controlled for cognitive reserve
by proxy so the present findings can hardly be
explained by cognitive reserve. One can conclude that
more research on these largely interlinked cognitive
processes and their relationship with dedifferentiation is
needed.
Clinical perspective
To our knowledge, this is the first time cognitive
impairment as explanatory variable for dedifferentiation
is found in two large and separate cohorts in which each
patient’s cognition has been thoroughly examined and
diagnosed. The present study therefore puts particular
emphasis on the dedifferentiation phenomenon in the
context of geriatric care and its patient population in
abnormal cognitive decline, i.e. patients with MCI or
dementia.
One important issue that the present study highlights
pertains to the cognitive profile of a patient, and whether
this profile of performance on psychometric tests is
relatively similar across cognitive domains gauged by the
tests. Clinicians often reason that an “even” profile is a
sign of healthy cognitive functioning, and vice versa. The
present study problematizes such reasoning as it suggests
that an even cognitive profile can de facto be due to
Figure 1. Conceptual plot for the degree of dependence of cognitive test scores on general intelligence (g) as a function of
cognitive development and specialization in young age, stability in adult age, and decline in old age. The present study focus is
highlighted.
www.aging-us.com 4 AGING
Table 1. Descriptive statistics in ADNI and MemClin within each dataset stratified as CI
and NCI.
ADNI
MemClin
CI (n = 884)
NCI (n = 398)
NCI (n = 152)
Age (yrs)
74.33 (7.40)
74.24 (5.68)
75.32 (5.70)
Education (yrs)
15.76 (2.86)
16.40 (2.68)
13.54 (3.74)
Male sex
517 (58.8)
188 (47.2)
61 (40.1)
Diagnosis, three class
Dementia*/AD
503 (56.9)
0 (0.0)
0 (0.0)
MCI
381 (43.1)
0 (0.0)
0 (0.0)
CN/SCI
0 (0.0)
398 (100.0)
152 (100.0)
Key psychometric tests
AVLT 1
4.05 (1.57)
5.48 (1.73)
5.64 (1.73)
AVLT delayed recall
2.64 (3.42)
7.85 (3.76)
10.11 (2.96)
TMT B / TMT 3
142.97 (81.77)
82.86 (41.07)
46.62 (16.62)
MMSE
26.30 (2.73)
29.07 (1.16)
28.91 (1.36)
Data are mean (SD) and count (%). Psychometric tests are raw scores. *MemClin includes other
subtypes, most frequently AD, vascular dementia, and mixed dementia. ADNI, Alzheimer’s
Disease Neuroimaging Initiative; MemClin, Memory Clinic project; AD, Alzheimer’s disease;
AVLT, Rey Auditory Verbal Learning Test; CI, Cognitive impairment; NCI, No cognitive
impairment; CN, Cognitively normal; MCI, Mild cognitive impairment; MMSE, Mini Mental State
Examination; SCI, Subjective cognitive impairment; TMT, Trail-Making Test.
Table 2. Principal component analysis (PCA) in the separate ADNI and MemClin datasets by CI
vs NCI groups after matching.
ADNI
MemClin
CI (n = 392)
NCI (n = 392)
CI (n = 143)
NCI (n = 143)
% variance explained by PC1
53.1
38.3
33.3
30.8
N factors with eigenvalues > 1
2
3
5
5
Individual test loadings on PC1
Working memory
(AVLT 1)
0.24 0.27 0.28 0.27
Episodic memory
(AVLT delayed recall; AVLT 5)
0.26 0.32 0.28 0.29
Executive function
(TMT B; TMT 3)
0.24 0.13 0.18 0.19
General
(MMSE)
0.25 0.10 0.20 0.18
Patients have been propensity score matched 1:1 without replacement on age, education and sex. PCA
was thereafter performed through single value decomposition. Values are calculated with imputed and
propensity score matched data controlling for age, education, and sex. ADNI, Alzheimer’s Disease
Neuroimaging Initiative; MemClin, Memory Clinic Project; AVLT, Rey Auditory Verbal Learning Test; CI,
Cognitive impairment; NCI, No cognitive impairment; MMSE, Mini Mental State Examination.
dedifferentiation of cognitive abilities. Most clinicians
are well-aware that a cognitive test profile that is similar
across domains but significantly worse across these
domains relative to normative test scores usually
signifies some form of cognitive decline. If however
dedifferentiation presents itself early on for a patient,
it might obscure a serious condition if the clinician
applies the “even profile” heuristic when presented
www.aging-us.com 5 AGING
with a cognitive profile that is a bit lower in
performance relative to applicable test norms but
similar across cognitive domains. Simply because AD
is very common, and because the classic AD profile is
uneven with fairly distinct underperformance on
episodic memory tests, does not allow for diagnostic
heuristics to be applied insofar that possible
dedifferentiation is ignored. Further, if a cognitive
profile of a patient is even because of dedifferentiation
and then also coupled with inadequately estimated
premorbid cognition, proper diagnosis of MCI could
be substantially delayed for patients initially classified
as cognitively healthy or with only subjective cognitive
impairment.
Cognitive status conditions are difficult to diagnose and
require thorough examination by a multidisciplinary
team at a memory clinic to achieve sufficient diagnostic
accuracy for the most difficult to separate patients.
In clinical practice, stronger dedifferentiation with
progressing neurodegeneration means that impaired
patients depend to a greater extent on their general
intelligence because their task specific skills developed
earlier in life are deteriorating. This suggests that
important lessons are to be learned regarding how these
patients become, for instance, cognitively overburdened
in concrete daily situations, during which they
previously could offload their cognitive ability through
skill heuristics but which are no longer accessible, or
not as easily accessible to them.
Limitations and strengths
All observational research is limited by possibly
remaining residual confounding. This includes our
study, since there may be other factors than cognitive
impairment that produces dedifferentiation of cognitive
abilities in late life. Factors such as biological age [15],
Figure 2. Sorted correlation strengths across all informative test pairs in the ADNI (n = 105) and MemClin (n = 190) datasets.
Notice that inverse test scores (e.g. TMT) had been rescaled prior and that the assumption of positive manifold is slightly violated (a few
negative correlations), possibly due to stochasticity. ADNI, Alzheimer’s Disease Neuroimaging Initiative; MemClin, Memory Clinic Project; CI,
Cognitive Impairment; NCI, No Cognitive Impairment; TMT, Trail-Making Test.
www.aging-us.com 6 AGING
cognitive reserve [17], or brain compensation [18] might
also produce dedifferentiation, and we cannot exclude the
possibility that compensatory mechanisms co-exist with
cognitive impairment, especially in the early phases. We
were not able to directly control for all these factors in
the present study, but through propensity score matching
on age, education, and sex, it is reasonable to assume that
we control for the bulk of such possible confounding
indirectly. Control for premorbid ability was not deemed
necessary as years of education was controlled for and is
considered a good proxy for premorbid ability.
Moreover, the different prevailing retrospective methods
for estimating premorbid ability; using education and
demographics, or semantic knowledge test performance
(e.g. WAIS-IV Information), or a specific pronunciation
type test (e.g. NART [19], ISW [20]), have their
respective limitations such as overestimation and
conceptual biases. More research is likely needed in
which other potential confounders are controlled for,
simultaneously being wary of the modelling pitfall of
overadjustment bias, i.e. controlling for intermediary
variables situated on the suggested causal path from
cognitive impairment to dedifferentiation. Our study was
also limited by data being cross-sectional. There are
longitudinal measurements on some variables in both
ADNI and MemClin but a more complete recording of
psychometric performance across time is needed to study
dedifferentiation over time within individuals. Another
important feature in our study was the multiple
dimensions for which the ADNI and MemClin cohorts
differ. For instance, ADNI is a North American research
database, employing in part different psychometric tests,
brain imaging techniques, diagnostic methods, and
patient inclusion procedures compared to the Swedish
MemClin database. Because of these differences, the two
datasets could not be combined and analysed as a whole.
We could however use the strengths of this study feature
by investigating, and also finding, dedifferentiation
in the two datasets separately, leveraging their
individually fairly large size and generally high quality of
measurements. Another important difference relates
to differing sampling procedures for which ADNI
deliberately sampled healthy controls whereas MemClin
controls where SCI patients. The pattern of
dedifferentiation was also more pronounced in ADNI
compared to MemClin. Importantly, MemClin is a new
large-scale naturalistic database with high ecological
validity and generalizability to the clinical geriatric
population in its specialized care setting provides both a
rare strength, and a complement to the excellent ADNI
database for conducting such studies as the present one.
CONCLUSIONS
Dedifferentiation of cognitive abilities in late life was
investigated and identified in two large and independent
cohorts. Through adjustment for age, education, and
sex, an independent association of cognitive impairment
on dedifferentiation was found. The meta-cognitive
aspect of dedifferentiation is important and should be
accounted for by clinicians as they diagnose, treat, and
care for their patients.
MATERIALS AND METHODS
Study population
We included patients from the Alzheimer’s Disease
Neuroimaging Initiative (ADNI, n = 1282) [13] and the
Memory Clinic Project (MemClin, n = 1223) [12]. The
diagnostic procedures for ADNI [21] and MemClin
have previously been described [12]. For ADNI, the
diagnosis procedure relied on subjective and objective
cognitive measures but was independent of biomarker
information. For MemClin, diagnosis was determined at
multidisciplinary consensus meetings based on
cognitive testing, neurological examination, brain
imaging, biomarkers and other routine clinical measures
as available [12, 22, 23]. The diagnosed cognitive
statuses of dementia, MCI, and SCI were collapsed into
two classes: CI and NCI. CI represents impairment of
clinically diagnosed severity whereas NCI represents no
objective impairment.
Cognitive variables
The administered neuropsychological batteries of ADNI
and MemClin are described elsewhere [12, 14] and
covered similar cognitive domains and subdomains with
some differences by original design. Importantly however,
across main analysis comparisons of dedifferentiation in
on CI vs NCI variables were identical in each database.
Unless further defined, psychometric tests are total raw
scores. Some psychometric variables were reverse coded
as needed. Psychometric tests are separated with semi-
colon (;) and subtests belonging to the same test are
separated with dash (/) as follows.
From ADNI we selected patient performance on 15
variables: the Mini Mental State Examination (MMSE);
Clock test; Copy test; Rey Auditory Verbal Learning
Test (AVLT) trial 1/2/3/4/5/total score/delayed recall/
recognition; semantic fluency (animals); Trail-Making
Test (TMT) A/B; and Boston Naming Test spontaneous
recall.
From MemClin we selected 20 variables: MMSE;
Wechsler Adult intelligence Scale 4th edition (WAIS-IV)
Information; Block design; Digit Span total/forward/
backward; AVLT 1/2/3/4/5/total score; Rey Complex
Figure copy; Delis-Kaplan Executive Function System
(D-KEFS) Verbal fluency FAS/semantic fluency/shifting
www.aging-us.com 7 AGING
(correct answers)/shifting (correct shiftings); and TMT
1/2/3. TheHiveDB was used for MemClin data
management [24].
Statistical analyses
All analyses were mirrored and run separately in both
ADNI and MemClin. Due to missing data some variables
were excluded (cut-off > 25% missing). Remaining
missing data sorted by % missing of total (in ADNI: TMT
B [1.9], NART [0.7], sex [0.3], age [0.3], edu [0.3] and in
MemClin: AVLT total [23.8], D-KEFS TMT 3 [22.8],
RCFT copy [22.6], WAIS-IV Block design [20.9], D-
KEFS TMT 1 [20.7] was deemed acceptable, assumed to
be missing at random (MAR) and imputed with k Nearest
Neighbour (kNN) [25, 26] applying the unweighted
Gower distance metric [27] with k set to three.
To control for confounding of the Dedifferentiation ~
CI association, propensity score matching [28, 29] was
performed prior to main analysis. A logit model was
constructed estimating the probability for CI with age
at examination, years of education, and sex as
predictors. Exposed (CI) where thereafter propensity
score matched 1:1 with unexposed (NCI) without
replacement on this probability applying SD = 0.05
caliper width. Figure 3 plots the matching diagnostics
and resulting across-group balance in the confounders.
Figure 3. Propensity score matching diagnostics in the ADNI and MemClin datasets. (A) Patient counts after matching are shown
as a function of their individual propensity score and overlayed with density plots, stratified by level of cognitive impairment. (B) Mean
distance followed by single-covariate balance by group calculated before (Unadjusted) and after (Adjusted) matching. ADNI matched: n CI=
392, n NCI = 392. MemClin matched: n CI = 143, n NCI = 143. ADNI, Alzheimer’s Disease Neuroimaging Initiative; MemClin, Memory Clinic
Project; CI, cognitive impairment; NCI, no cognitive impairment.
www.aging-us.com 8 AGING
Main analysis
After matching, principal component analysis (PCA)
was performed on the mean centered and unit variance
scaled psychometric variables through single value
decomposition. Percentage explained variance in
psychometric variables by the first unrotated component
(PC1), a proxy for general cognitive ability, was
compared across the matched CI and NCI groups
expecting a smaller % explained variance by PC1 among
scores in the CI vs the NCI group. We also report n
components with eigenvalues > 1 (Kaiser criterion)
expecting fewer selected components by this criterion for
CI vs NCI. Loadings on PC1 by key psychometric tests
that tap central cognitive domains were expected to be
higher in CI vs NCI groups. Supplementary Tables 1, 2
show PCA analysis by subgroups after propensity score
matching for both the ADNI and the MemClin cohort.
Correlation coefficients calculated across all informative
pairings of psychometric variables were thereafter sorted
from weakest to strongest and assessed by t-test and also
by visual inspection comparing the strengths of
correlations across matched CI and NCI groups expecting
average test-pair correlations to be higher in the CI vs the
NCI group.
Additional statistics
Unless further explained, bivariate summary statistics
are presented as mean (SD) for numeric variables and
count (%) for factors. Statistical significance was set
to 5%.
Analyses were performed in R version 4.0.1 using
packages caret, cobalt, corrplot, data.table, dummies,
factoextra, FactoMineR, foreign, ggplot2, gridExtra,
haven, MatchIt, matrixStats, mice, readxl, tableone, and
VIM.
AUTHOR CONTRIBUTIONS
JW and UE designed the study. JW drafted the
manuscript and analyzed data. All authors interpreted
the results, revised the manuscript, and approved the
final submission.
ACKNOWLEDGMENTS
We are deeply thankful to the ADNI and MemClin
participants and clinicians as they made this study
possible.
Data collection and sharing for this project was funded
by the Alzheimer's Disease Neuroimaging Initiative
(ADNI) (National Institutes of Health Grant U01
AG024904) and DOD ADNI (Department of Defense
award number W81XWH-12-2-0012). ADNI is funded
by the National Institute on Aging, the National Institute
of Biomedical Imaging and Bioengineering, and through
generous contributions from the following: AbbVie,
Alzheimer’s Association; Alzheimer’s Drug Discovery
Foundation; Araclon Biotech; BioClinica, Inc.; Biogen;
Bristol-Myers Squibb Company; CereSpir, Inc.;
Cogstate; Eisai Inc.; Elan Pharmaceuticals, Inc.; Eli Lilly
and Company; EuroImmun; F. Hoffmann-La Roche Ltd
and its affiliated company Genentech, Inc.; Fujirebio;
GE Healthcare; IXICO Ltd.; Janssen Alzheimer
Immunotherapy Research and Development, LLC.;
Johnson and Johnson Pharmaceutical Research and
Development LLC.; Lumosity; Lundbeck; Merck and
Co., Inc.; Meso Scale Diagnostics, LLC.; NeuroRx
Research; Neurotrack Technologies; Novartis
Pharmaceuticals Corporation; Pfizer Inc.; Piramal
Imaging; Servier; Takeda Pharmaceutical Company; and
Transition Therapeutics. The Canadian Institutes of
Health Research is providing funds to support ADNI
clinical sites in Canada. Private sector contributions are
facilitated by the Foundation for the National Institutes of
Health (https://www.fnih.org/). The grantee organization
is the Northern California Institute for Research and
Education, and the study is coordinated by the
Alzheimer’s Therapeutic Research Institute at the
University of Southern California. ADNI data are
disseminated by the Laboratory for Neuro Imaging at the
University of Southern California.
CONFLICTS OF INTEREST
The authors declare that they have no conflicts of interest.
FUNDING
This study was financially supported by the Swedish
Research Council (VR), the Swedish Research
Council for Health, Working Life and Welfare
(FORTE), the Swedish Foundation for Strategic
Research (SSF), the Strategic Research Programme in
Neuroscience at Karolinska Institutet (StratNeuro), the
Åke Wiberg foundation, Hjärnfonden, Alzheimerfonden,
Demensfonden, The Söderström König Foundation,
Stiftelsen Olle Engkvist Byggmästare, Loo och Hans
Ostermans stiftelse för medicinsk forskning, and
Birgitta och Sten Westerberg. The funding sources had
no role in the study design, data collection, analysis,
interpretation, or the writing of the manuscript.
REFERENCES
1. Reinert G. Comparative Factor Analytic Studies of
Intelligence Throughout the Human Life-span. The
West Virginia University Conference on Life-Span
www.aging-us.com 9 AGING
Developmental Psychology; Morgantown, West
Virginia University 1970.
2. Dodge HH, Marson DC. Illuminating cognitive
dedifferentiation at the end of life. Neurology. 2012;
78:111011.
https://doi.org/10.1212/WNL.0b013e31824f80b9
PMID:22491863
3. Balinsky B. An analysis of the mental factors of various
age groups from nine to sixty. Genetic Psychology
Monographs. 1941; 23:191234.
4. Balsamo M, Romanelli R, Saggino A. The de-
differentiation hypothesis in normal elderly persons.
Percept Mot Skills. 2010; 110:8588.
https://doi.org/10.2466/PMS.110.1.85-88
PMID:20391873
5. Rönnlund M, Sundström A, Nilsson LG. Interindividual
differences in general cognitive ability from age 18 to
age 65 years are extremely stable and strongly
associated with working memory capacity. Intelligence.
2015; 53:5964.
6. Craik FI, Bialystok E. Cognition through the lifespan:
mechanisms of change. Trends Cogn Sci. 2006;
10:13138.
https://doi.org/10.1016/j.tics.2006.01.007
PMID:16460992
7. Koen JD, Rugg MD. Neural Dedifferentiation in the
Aging Brain. Trends Cogn Sci. 2019; 23:54759.
https://doi.org/10.1016/j.tics.2019.04.012
PMID:31174975
8. Batterham PJ, Christensen H, Mackinnon AJ.
Comparison of age and time-to-death in the
dedifferentiation of late-life cognitive abilities. Psychol
Aging. 2011; 26:84451.
https://doi.org/10.1037/a0023300
PMID:21534687
9. Wilson RS, Segawa E, Hizel LP, Boyle PA, Bennett DA.
Terminal dedifferentiation of cognitive abilities.
Neurology. 2012; 78:111622.
https://doi.org/10.1212/WNL.0b013e31824f7ff2
PMID:22491858
10. Juan-Espinosa M, García LF, Escorial S, Rebollo I, Colom
R, Abad FJ. Age dedifferentiation hypothetis: Evidence
from the WAIS III. Intelligence. 2002; 30:395408.
11. Christensen H. What cognitive changes can be
expected with normal ageing? Aust N Z J Psychiatry.
2001; 35:76875.
https://doi.org/10.1046/j.1440-1614.2001.00966.x
PMID:11990887
12. Ekman U, Ferreira D, Muehlboeck JS, Wallert J, Rennie
A, Eriksdotter M, Wahlund LO, Westman E. The
MemClin project: a prospective multi memory clinics
study targeting early stages of cognitive impairment.
BMC Geriatr. 2020; 20:93.
https://doi.org/10.1186/s12877-020-1478-3
PMID:32138686
13. Mueller SG, Weiner MW, Thal LJ, Petersen RC, Jack C,
Jagust W, Trojanowski JQ, Toga AW, Beckett L. The
Alzheimer’s disease neuroimaging initiative.
Neuroimaging Clin N Am. 2005; 15:86977.
https://doi.org/10.1016/j.nic.2005.09.008
PMID:16443497
14. Bock O, Haeger M, Voelcker-Rehage C. Structure of
executive functions in young and in older persons.
PLoS One. 2019; 14:e0216149.
https://doi.org/10.1371/journal.pone.0216149
PMID:31071104
15. Gonzalez-Burgos L, Hernández-Cabrera JA, Westman E,
Barroso J, Ferreira D. Cognitive compensatory
mechanisms in normal aging: a study on verbal fluency
and the contribution of other cognitive functions.
Aging (Albany NY). 2019; 11:4090106.
https://doi.org/10.18632/aging.102040
PMID:31232698
16. Stern Y. Cognitive reserve. Neuropsychologia. 2009;
47:201528.
https://doi.org/10.1016/j.neuropsychologia.2009.03.0
04 PMID:19467352
17. Gonzalez-Burgos L, Barroso J, Ferreira D. Cognitive
reserve and network efficiency as compensatory
mechanisms of the effect of aging on phonemic
fluency. Aging (Albany NY). 2020; 12:2335178.
https://doi.org/10.18632/aging.202177
PMID:33203801
18. Gonzalez-Burgos L, Pereira JB, Mohanty R, Barroso J,
Westman E, Ferreira D. Cortical Networks
Underpinning Compensation of Verbal Fluency in
Normal Aging. Cereb Cortex. 2021. [Epub ahead of
print.]
https://doi.org/10.1093/cercor/bhab052
PMID:33866353
19. Nelson HE. National Adult Reading Test (NART): For the
Assessment of Premorbid Intelligence in Patients with
Dementia: Test Manual, NFER-Nelson, Windsor, UK.
1982.
20. Almkvist O, Tallberg IM. Cognitive decline from
estimated premorbid status predicts neurodegeneration
in Alzheimer’s disease. Neuropsychology. 2009;
23:11724.
https://doi.org/10.1037/a0014074
PMID:19210039
21. Petersen RC, Aisen PS, Beckett LA, Donohue MC,
Gamst AC, Harvey DJ, Jack CR Jr, Jagust WJ, Shaw LM,
Toga AW, Trojanowski JQ, Weiner MW. Alzheimer’s
www.aging-us.com 10 AGING
Disease Neuroimaging Initiative (ADNI): clinical
characterization. Neurology. 2010; 74:20109.
https://doi.org/10.1212/WNL.0b013e3181cb3e25
PMID:20042704
22. Wallert J, Westman E, Ulinder J, Annerstedt M, Terzis
B, Ekman U. Differentiating Patients at the Memory
Clinic With Simple Reaction Time Variables: A
Predictive Modeling Approach Using Support Vector
Machines and Bayesian Optimization. Front Aging
Neurosci. 2018; 10:144.
https://doi.org/10.3389/fnagi.2018.00144
PMID:29881341
23. Wallert J, Ekman U, Westman E, Madison G. The worst
performance rule with elderly in abnormal cognitive
decline. Intelligence. 2017; 64:917.
24. Muehlboeck JS, Westman E, Simmons A. TheHiveDB
image data management and analysis framework.
Front Neuroinform. 2014; 7:49.
https://doi.org/10.3389/fninf.2013.00049
PMID:24432000
25. Beretta L, Santaniello A. Nearest neighbor imputation
algorithms: a critical evaluation. BMC Med Inform
Decis Mak. 2016 (Suppl 3); 16:74.
https://doi.org/10.1186/s12911-016-0318-z
PMID:27454392
26. Jönsson P, Wohlin C, editors. An Evaluation of k-
Nearest Neighbour imputation Using Likert Data. 10th
International Symposium on Software Metrics; 2004:
IEEEXplore.
27. Kowarik A, Templ M. Imputation with R package VIM. J
Stat Softw. 2016; 74:116.
28. Rosenbaum PR, Rubin DB. The Central Role of the
Propensity Score in Observational Studies for Causal
Effects. Biometrika. 1983; 70:4155.
29. Ho D, Kosuke I, King G, Stuart E. Matching as
Nonparametric Preprocessing for Reducing Model
Dependence in Parametric Causal Inference. Political
Analysis. 2007; 15:199236.
www.aging-us.com 11 AGING
SUPPLEMENTARY MATERIALS
Supplementary Tables
Supplementary Table 1. Principal component analysis by subgroups after propensity score
matching, ADNI.
CI group (n = 392)
PC1
PC2
PC3
PC4
PC5
PC6
Standard deviation
2.8219
1.3263
0.9675
0.9160
0.8132
0.7417
Proportion of variance
0.5309
0.1173
0.0624
0.0559
0.0441
0.0367
Cumulative proportion of variance
0.5309
0.6481
0.7106
0.7665
0.8109
0.8473
Individual test loadings
MMSE
-0.2520
0.1396
-0.1920
0.1648
0.0883
-0.3796
Clock test
-0.2030
0.3575
0.1946
0.2835
0.1469
-0.5835
Copy test
-0.1161
0.3926
0.5224
0.3829
-0.5205
0.2608
AVLT 1
-0.2438
-0.0350
0.2880
-0.5600
-0.1158
-0.1179
AVLT 2
-0.3024
-0.1348
0.1988
-0.2179
0.0112
0.0164
AVLT 3
-0.3122
-0.1978
0.1170
-0.0538
-0.0688
-0.0297
AVLT 4
-0.3178
-0.1944
0.0405
0.0427
-0.0366
-0.0579
AVLT 5
-0.3133
-0.2110
-0.007
0.0967
0.0227
-0.0274
AVLT total
-0.3376
-0.1873
0.1156
-0.0969
-0.0328
-0.0432
AVLT delayed
-0.2597
-0.3144
-0.0800
0.2842
-0.0341
0.1792
AVLT recognition
-0.2380
-0.1628
-0.2310
0.4567
0.1842
0.2374
Semantic fluency Animals
-0.2447
0.2077
-0.3321
-0.1575
-0.2641
0.3054
TMT A
-0.2110
0.3900
0.0924
-0.1281
0.4338
0.4752
TMT B
-0.2371
0.3590
-0.1082
-0.1179
0.4078
0.0247
BNT spontaneous recall
-0.1887
0.2703
-0.5576
-0.1300
-0.4681
-0.1346
NCI group (n = 392)
PC1
PC2
PC3
PC4
PC5
PC6
Standard deviation
2.3962
1.4179
1.1249
0.9771
0.9502
0.8920
Proportion of variance
0.3828
0.1340
0.0844
0.0637
0.0602
0.0530
Cumulative proportion of variance
0.3828
0.5168
0.6012
0.6648
0.7250
0.7781
Individual test loadings
MMSE
-0.1033
0.2569
-0.0807
0.2215
0.8312
-0.3164
Clock test
-0.0857
0.4187
-0.4669
-0.2221
-0.0037
0.2391
Copy test
-0.0427
0.3620
-0.6225
0.0317
-0.1470
0.0698
AVLT 1
-0.2725
-0.1185
-0.0793
0.3971
-0.1515
0.1061
AVLT 2
-0.3411
-0.1028
-0.0246
0.2235
-0.0387
0.1217
AVLT 3
-0.3689
-0.1083
0.0042
-0.0018
-0.0122
0.0360
AVLT 4
-0.3658
-0.1307
-0.0038
-0.0182
0.0403
0.0145
AVLT 5
-0.3515
-0.0721
-0.0563
-0.1104
0.0426
0.0731
AVLT total
-0.4055
-0.1252
-0.0336
0.0942
-0.0194
0.0791
AVLT delayed
-0.3231
-0.1202
-0.0599
-0.2049
0.0512
-0.0404
AVLT recognition
-0.1926
-0.0750
0.0149
-0.7437
0.0230
-0.3323
Semantic fluency Animals
-0.1910
0.2635
0.1019
0.2485
-0.1818
-0.5921
TMT A
-0.1501
0.3627
0.4126
-0.1139
0.1677
0.3251
TMT B
-0.1276
0.4136
0.3973
-0.0374
-0.0088
0.3469
BNT spontaneous recall
-0.1220
0.4015
0.1854
0.0170
-0.4408
-0.3316
The six first PCs are presented from each PCA with values rounded to four decimals.
www.aging-us.com 12 AGING
Supplementary Table 2. Principal component analysis by subgroups after propensity score
matching, MemClin.
CI group (n = 143)
PC1
PC2
PC3
PC4
PC5
PC6
Standard deviation
2.5793
1.6616
1.3282
1.1957
1.0906
1.0205
Proportion of variance
0.3326
0.1380
0.0882
0.0715
0.0595
0.0521
Cumulative proportion of variance
0.3326
0.4707
0.5589
0.6304
0.6898
0.7419
Individual test loadings
MMSE
-0.1988
-0.1080
0.0587
-0.1530
0.2559
-0.0030
WAIS-IV Information
-0.1298
-0.0395
0.1048
-0.1516
0.1234
0.7551
WAIS-IV Block design
-0.1742
-0.2480
-0.2584
-0.1019
0.3309
0.0431
WAIS-IV Digit Span total
-0.1450
-0.3873
0.3153
0.2209
-0.0588
-0.1371
WAIS-IV Digit Span forward
-0.0792
-0.3317
0.3611
0.2444
-0.1733
0.0587
WAIS-IV Digit Span backward
-0.1210
-0.2652
0.3618
0.2657
0.1127
-0.1494
AVLT 1
-0.2829
0.1337
0.0979
0.0761
0.1170
-0.1240
AVLT 2
-0.3070
0.1999
0.1075
0.1975
0.1104
-0.1092
AVLT 3
-0.2974
0.2651
0.0445
0.1520
-0.0157
0.0118
AVLT 4
-0.3033
0.2720
0.0372
0.0623
-0.0658
-0.0244
AVLT 5
-0.2765
0.2981
-0.0128
0.0203
0.0028
0.1165
AVLT total
-0.3386
0.2674
0.0583
0.0998
0.0174
0.0088
RCFT copy
-0.0668
-0.1565
-0.0032
-0.1243
0.7314
-0.1314
D-KEFS verbal fluency FAS
-0.2083
-0.2186
0.0862
-0.1674
-0.1854
0.3260
D-KEFS semantic fluency
-0.2559
-0.1693
0.0231
-0.2147
-0.1335
0.1894
D-KEFS semantic fluency shifting
(n correct answers)
-0.2587
-0.1101
-0.0233
-0.4408
-0.1806
-0.3041
D-KEFS sematic fluency shifting
(n correct shiftings)
-0.2459
-0.1066
-0.0341
-0.4630
-0.2262
-0.2907
D-KEFS TMT 1
-0.1558
-0.1980
-0.3632
0.2493
-0.2421
-0.0078
D-KEFS TMT 2
-0.1673
-0.1599
-0.4736
0.2316
0.0396
-0.0196
D-KEFS TMT 3
-0.1806
-0.2081
-0.4052
0.2607
-0.0422
0.0703
NCI group (n = 143)
PC1
PC2
PC3
PC4
PC5
PC6
Standard deviation
2.4816
1.7304
1.3896
1.2732
1.1431
0.9989
Proportion of variance
0.3079
0.1497
0.0966
0.0811
0.0653
0.0499
Cumulative proportion of variance
0.3079
0.4576
0.5542
0.6352
0.7006
0.7505
Individual test loadings
MMSE
-0.1808
0.0205
0.0990
0.0844
-0.2895
0.3132
WAIS-IV Information total
-0.1336
-0.2211
-0.2660
0.3220
-0.3035
0.0016
WAIS-IV Block Design total
-0.1521
-0.3192
-0.0710
0.1119
-0.3895
0.0099
WAIS-IV Digit Span total
-0.2008
-0.2993
0.3490
0.2186
0.1891
-0.0636
WAIS-IV Digit Span forward
-0.1590
-0.2314
0.4195
0.1737
0.1976
0.0732
WAIS-IV Digit Span backward
-0.1354
-0.1883
0.4554
0.1949
0.0663
-0.2893
AVLT 1
-0.2679
0.1209
0.0926
0.0427
0.1153
0.0244
AVLT 2
-0.2987
0.2470
0.0378
0.0316
-0.0670
0.0269
AVLT 3
-0.2727
0.3191
0.0107
0.0232
-0.1281
-0.0905
AVLT 4
-0.3020
0.2807
0.0031
0.0001
0.0411
-0.0802
AVLT 5
-0.2927
0.2702
-0.0095
0.0198
-0.0523
-0.0932
AVLT total
-0.3404
0.2976
0.0251
0.0267
-0.0233
-0.0582
RCFT copy
-0.1355
-0.1915
-0.2582
0.3432
-0.3464
-0.1807
www.aging-us.com 13 AGING
D-KEFS verbal fluency FAS
-0.1924
-0.0926
-0.0806
0.1030
0.2350
0.6857
D-KEFS semantic fluency
-0.2502
-0.1383
-0.2887
0.0221
0.2414
0.2829
D-KEFS semantic fluency shifting
(n correct answers)
-0.2437
-0.1726
-0.2865
-0.1547
0.3440
-0.2959
D-KEFS semantic fluency shifting
(n correct shiftings)
-0.2310
-0.2233
-0.3211
-0.1214
0.2617
-0.2950
TMT 1
-0.1371
-0.1993
-0.0566
-0.3436
-0.0115
0.1187
TMT 2
-0.1711
-0.2340
0.1461
-0.4784
-0.2258
-0.0418
TMT 3
-0.1883
-0.1177
0.1792
-0.4917
-0.2875
0.0866
The six first PCs are presented from each PCA.
Article
Cognitive processing relies on the functional coupling between the cerebrum and cerebellum. However, it remains unclear how the 2 collaborate in amnestic mild cognitive impairment (aMCI) patients. With functional magnetic resonance imaging techniques, we compared cerebrocerebellar functional connectivity during the resting state (rsFC) between the aMCI and healthy control (HC) groups. Additionally, we distinguished coupling between functionally corresponding and noncorresponding areas across the cerebrum and cerebellum. The results demonstrated decreased rsFC between both functionally corresponding and noncorresponding areas, suggesting distributed deficits of cerebrocerebellar connections in aMCI patients. Increased rsFC was also observed, which were between functionally noncorresponding areas. Moreover, the increased rsFC was positively correlated with attentional scores in the aMCI group, and this effect was absent in the HC group, supporting that there exists a compensatory mechanism in patients. The current study contributes to illustrating how the cerebellum adjusts its coupling with the cerebrum in individuals with cognitive impairment.
Article
Full-text available
Elucidating compensatory mechanisms underpinning phonemic fluency (PF) may help to minimize its decline due to normal aging or neurodegenerative diseases. We investigated cortical brain networks potentially underpinning compensation of age-related differences in PF. Using graph theory, we constructed networks from measures of thickness for PF, semantic, and executive–visuospatial cortical networks. A total of 267 cognitively healthy individuals were divided into younger age (YA, 38–58 years) and older age (OA, 59–79 years) groups with low performance (LP) and high performance (HP) in PF: YA-LP, YA-HP, OA-LP, OA-HP. We found that the same pattern of reduced efficiency and increased transitivity was associated with both HP (compensation) and OA (aberrant network organization) in the PF and semantic cortical networks. When compared with the OA-LP group, the higher PF performance in the OA-HP group was associated with more segregated PF and semantic cortical networks, greater participation of frontal nodes, and stronger correlations within the PF cortical network. We conclude that more segregated cortical networks with strong involvement of frontal nodes seemed to allow older adults to maintain their high PF performance. Nodal analyses and measures of strength were helpful to disentangle compensation from the aberrant network organization associated with OA.
Article
Full-text available
Compensation in cognitive aging is a topic of recent interest. However, factors contributing to cognitive compensation in functions such as phonemic fluency (PF) are not completely understood. Using cross-sectional data, we investigated cognitive reserve (CR) and network efficiency in young (32-58 years) versus old (59-84 years) individuals with high versus low performance in PF. ANCOVA was used to investigate the interaction between CR, age, and performance in PF. Random forest and graph theory analyses were conducted to study the contribution of cognition to PF and efficiency measures, respectively. Higher CR increased performance in PF and reduced age-related differences in PF. A slightly higher number of cognitive functions contributed to performance in high CR groups. The networks were more integrated in high CR individuals, both in the older age and high-performance groups. The strength and segregation of the networks were decreased in high-performance groups with high CR. We conclude that PF decreases less with age in individuals with higher CR, possibly due to a greater capacity to recruit non-linguistic cognitive networks, and efficient use of language networks, thereby integrating information in a rapid way across less fragmented networks. High CR and network efficiency seem to be important factors for cognitive compensation.
Article
Full-text available
Background: There remains a lack of large-scale clinical studies of cognitive impairment that aim to increase diagnostic and prognostic accuracy as well as validate previous research findings. The MemClin project will amass large quantities of cross-disciplinary data allowing for the construction of robust models to improve diagnostic accuracy, expand our knowledge on differential diagnostics, strengthen longitudinal prognosis, and harmonise examination protocols across centres. The current article describes the Memory Clinic (MemClin) project's study-design, materials and methods, and patient characteristics. In addition, we present preliminary descriptive data from the ongoing data collection. Methods: Nine out of ten memory clinics in the greater Stockholm area, which largely use the same examination methods, are included. The data collection of patients with different stages of cognitive impairment and dementia is coordinated centrally allowing for efficient and secure large-scale database construction. The MemClin project rest directly on the memory clinics examinations with cognitive measures, health parameters, and biomarkers. Results: Currently, the MemClin project has informed consent from 1543 patients. Herein, we present preliminary data from 835 patients with confirmed cognitive diagnosis and neuropsychological test data available. Of those, 239 had dementia, 487 mild cognitive impairment (MCI), and 104 subjective cognitive impairment (SCI). In addition, we present descriptive data on visual ratings of brain atrophy and cerebrospinal fluid markers. Conclusions: Based on our current progress and preliminary data, the MemClin project has a high potential to provide a large-scale database of 1200-1500 new patients annually. This coordinated data collection will allow for the construction of improved diagnostic and prognostic models for neurodegenerative disorders and other cognitive conditions in their naturalistic setting.
Article
Full-text available
Verbal fluency has been widely studied in cognitive aging. However, compensatory mechanisms that maintain its optimal performance with increasing age are not completely understood. Using cross-sectional data, we investigated differentiation and dedifferentiation processes in verbal fluency across the lifespan by analyzing the association between verbal fluency and numerous cognitive measures within four age groups (N=446): early middle-age (32-45 years), late middle-age (46-58 years), early elderly (59-71 years), and late elderly (72-84 years). ANCOVA was used to investigate the interaction between age and fluency modality. Random forest models were conducted to study the contribution of cognition to semantic, phonemic, and action fluency. All modalities declined with increasing age, but semantic fluency was the most vulnerable to aging. The most prominent reduction in performance was observed during the transition from middle-age to early elderly, when cognitive variables stopped contributing (differentiation), and new cognitive variables started contributing (dedifferentiation). Lexical access, processing speed, and executive functions were among the most contributing functions. We conclude that the association between age and verbal fluency is masked by age-specific influences of other cognitive functions. Differentiation and dedifferentiation processes can coexist. This study provides important data for better understanding of cognitive aging and compensatory processes.
Article
Full-text available
Introduction Using factor analysis, several studies reported that higher-order cognitive control involves separable executive functions. However, the number and definition of the purported functions differed between studies. One possible explanation for this discrepancy is that executive functions don’t exhibit a clear factorial structure, i.e., there is no clear dichotomy between executive function tests which are well-correlated (representing a common factor) and those which are poorly correlated (representing distinct factors). We scrutinize this explanation separately in data from young and from older persons. Methods & results Young and older volunteers completed cognitive tests of the purported executive functions shifting, updating, inhibition and dual-tasking (two tests per function). Confirmatory and exploratory factor analyses yielded, for either age group, factorial structures that were within the range reported in literature. More importantly, when correlations between tests were sorted in ascending order, and were then fitted them by piecewise linear regression with a breakpoint, there was no evidence for a distinct breakpoint between low and high correlations in either age group. Correlations between tests were significantly higher in older compared to young participants, and the pattern of test pairs with high and with low correlations differed between age groups. Discussion The absence of a breakpoint indicates that executive function tests don’t segregate into well-correlated and poorly correlated pairs, and therefore are not well suited for factor analyses. We suggest that executive functions are better described as a partly overlapping rather than a factorial structure. The increase of correlations in older participants supports the existence of age-related dedifferentiation, and the dissimilarity of correlations in the two age groups supports the existence of age-related reorganization.
Article
Full-text available
BACKGROUND: Mild Cognitive Impairment (MCI) and dementia differ in important ways yet share a future of increased prevalence. Separating these conditions from each other, and from Subjective Cognitive Impairment (SCI), is important for clinical prognoses and treatment, socio-legal interventions, and family adjustments. With costly clinical investigations and an aging population comes a need for more cost-efficient differential diagnostics. METHODS: Using supervised machine learning, we investigated nine variables extracted from simple reaction time data with respect to their single and conjoined ability to discriminate both MCI/dementia, and SCI/MCI/dementia, compared to – and together with – established psychometric tests. One-hundred-twenty elderly patients (age range = 65-95 years) were recruited when referred to full neuropsychological assessment at a specialized memory clinic in urban Sweden. A freely available simple reaction time task served as index test and was administered and scored objectively by a computer before diagnosis of SCI (n = 17), MCI (n = 53), or dementia (n = 50). As reference standard, diagnosis was decided through the multidisciplinary memory clinic investigation. Bonferroni-Holm corrected p-values for constructed models against the null model are provided. RESULTS: Algorithmic feature selection for the two final multivariable models was performed through recursive feature elimination with 3x10-fold cross-validation resampling. For both models, this procedure selected seven predictors of which five were simple reaction time variables. When used as input for a soft-margin, radial-basis support vector machine model tuned via Bayesian optimization, the leave-one-out cross-validated accuracy of the final model for MCI/dementia classification was good (Accuracy = .806 [.716, .877], P < .001) and the final model for SCI/MCI/dementia classification held some merit (Accuracy = .650 [.558, .735], P < .001). These two models are implemented in a freely available application for research and educatory use. CONCLUSIONS: Simple reaction time variables hold some potential in conjunction with established psychometric tests for differentiating MCI/dementia, and SCI/MCI/dementia in these difficult-to-differentiate memory clinic patients. While external validation is needed, their implementation within diagnostic support systems is promising.
Article
Full-text available
Compared to best performances, worst performances on multi-trial psychometric tests often show stronger correlations with other g-loaded cognitive tests, which is known as the Worst Performance Rule (WPR). While worst performances may be more sensitive or specific to cognitive decline, clinical psychometric research and neuropsychological practice tends to neglect the WPR. Here, we examined the WPR-paradigm relative to abnormal cognitive decline. Specifically, we studied the WPR with binned simple reaction time task responses when rank-correlated with five different estimates of psychometric g within a memory clinic sample (n = 103) of elderly diagnosed with either Mild Cognitive Impairment (MCI) (n = 53) or dementia (n = 50). Three of the g-estimates were composite scores constructed from 2, 6, and 28 established test scores. Results showed a consistent WPR-pattern in the whole sample for each of the five estimates (block design rs = − 0.201 to − 0.120; digit span rs = − 0.284 to − 0.112; g2 rs = − 0.311 to − 0.162; g6 rs = − 0.314 to − 0.107; g28 rs = − 0.269 to − 0.121). Our findings contradict classical test theory, and highlight the underused potential of the WPR when assessing cognitive dysfunction in elderly patients.
Article
Full-text available
The package VIM (Templ, Alfons, Kowarik, and Prantner 2016) is developed to explore and analyze the structure of missing values in data using visualization methods, to impute these missing values with the built-in imputation methods and to verify the imputation process using visualization tools, as well as to produce high-quality graphics for publications. This article focuses on the different imputation techniques available in the package. Four different imputation methods are currently implemented in VIM, namely hot-deck imputation, k-nearest neighbor imputation, regression imputation and iterative robust model-based imputation (Templ, Kowarik, and Filzmoser 2011). All of these methods are implemented in a flexible manner with many options for customization. Furthermore in this article practical examples are provided to highlight the use of the implemented methods on real-world applications. In addition, the graphical user interface of VIM has been re-implemented from scratch resulting in the package VIMGUI (Schopfhauser, Templ, Alfons, Kowarik, and Prantner 2016) to enable users without extensive R skills to access these imputation and visualization methods.
Article
Full-text available
Background Nearest neighbor (NN) imputation algorithms are efficient methods to fill in missing data where each missing value on some records is replaced by a value obtained from related cases in the whole set of records. Besides the capability to substitute the missing data with plausible values that are as close as possible to the true value, imputation algorithms should preserve the original data structure and avoid to distort the distribution of the imputed variable. Despite the efficiency of NN algorithms little is known about the effect of these methods on data structure. Methods Simulation on synthetic datasets with different patterns and degrees of missingness were conducted to evaluate the performance of NN with one single neighbor (1NN) and with k neighbors without (kNN) or with weighting (wkNN) in the context of different learning frameworks: plain set, reduced set after ReliefF filtering, bagging, random choice of attributes, bagging combined with random choice of attributes (Random-Forest-like method). Results Whatever the framework, kNN usually outperformed 1NN in terms of precision of imputation and reduced errors in inferential statistics, 1NN was however the only method capable of preserving the data structure and data were distorted even when small values of k neighbors were considered; distortion was more severe for resampling schemas. Conclusions The use of three neighbors in conjunction with ReliefF seems to provide the best trade-off between imputation error and preservation of the data structure. The very same conclusions can be drawn when imputation experiments were conducted on the single proton emission computed tomography (SPECTF) heart dataset after introduction of missing data completely at random.
Article
Many cognitive abilities decline with age even in the absence of detectable pathology. Recent evidence indicates that age-related neural dedifferentiation, operationalized in terms of neural selectivity, may contribute to this decline. We review here work exploring the relationship between neural dedifferentiation, cognition, and age. Compelling evidence for age effects on neural selectivity comes from both non-human animal and human research. However, current data suggest that age does not moderate the observed relationships between neural dedifferentiation and cognitive performance. We propose that functionally significant variance in measures of neural dedifferentiation reflects both age-dependent and age-independent factors. We further propose that the effects of age on neural dedifferentiation do not exclusively reflect detrimental consequences of aging.