Association Between Apolipoprotein E4and Cognitive Decline in
Chris J. Packard, DSc,aRudi G.J. Westendorp, MD,dDavid J. Stott, MD,bMuriel J. Caslake, PhD,a
Heather M. Murray, MSc,cJames Shepherd, MD,gGerard J. Blauw, MD,dMichael B. Murphy, MD,g
Edward L.E.M. Bollen, MD,lBrendan M. Buckley, MD,gStuart M. Cobbe, MD,bIan Ford, PhD,cAllan
Gaw, MD,kMichael Hyland, MD,hJ. Wouter Jukema, MD,mAdriaan M. Kamper, MD,dPeter W.
Macfarlane, DSc,bJellemer Jolles, MD,e,fIvan J. Perry, MD,iBrian J. Sweeney,jand Cillian Twomey,
MDhfor the Prospective Study of Pravastatin in the Elderly at Risk Group
OBJECTIVE: To determine the influence of apolipoprotein
Eoncognitivedeclineina cohortofelderlymen and women.
DESIGN: Prospective study.
SETTING: Scotland, Ireland, and the Netherlands.
PARTICIPANTS: Five thousand eight hundred four subjects
aged 70 to 82 from the Prospective Study of Pravastatin in the
Elderly at Risk (PROSPER).
MEASUREMENTS: Subjects were assessed at baseline
and over a mean 3.2-year (range 0.7–4.2) follow-up
for memory (Picture-Word Recall), speed of information
processing (Stroop and Letter-Digit Coding), global cogni-
tive function (Mini-Mental State Examination), and activ-
ities of daily living.
RESULTS: At baseline, subjects with apolipoprotein E4
versus those without E4had poorer memory performance
(mean score difference ?0.20 (95% confidence interval
(CI)5 ?0.31 to ?0.09) for immediate recall and ?0.32
(95% CI5 ?0.48 to ?0.16) for delayed recall and slower
information processing (difference in Stroop, 2.79 seconds,
(95% CI51.20–4.28); Letter-Digit score, ?0.36, (95%
CI5 ?0.77–0.05). Subjects with apolipoprotein E4showed
a greater decline in immediate (?0.22, 95% CI5 ?0.33 to
?0.11) and delayed (?0.30, 95% CI5 ?0.46 to ?0.15)
memory scores but no significant change in speed of
P5.06). Memory scores decreased 2.5% from baseline in
those without E4, 4.3% in E4 heterozygotes (P5.01 for
immediate and P5.03 for delayed, vs no E4) and 8.9% to
13.8% in E4 homozygotes (P5.04 for immediate and
P5.004 for delayed, vs heterozygotes). Apolipoprotein E4
was associated with greater decline in instrumental activities
of daily living (Po.001). Cognitive decline was not associ-
ated with lipoprotein levels.
CONCLUSION: Findings in PROSPER indicate that E4is
associated with more-rapid cognitive decline and may,
therefore, predispose to dementia. J Am Geriatr Soc
Key words: memory; dementia; trial; statin
the general population and the risk factors that lead to the
development of frank dementia. Although age is the main
predictor of cognitive function, investigators have also re-
ported that a history of hypertension, diabetes mellitus,
stroke, depression, and lack of physical activity are fac-
tors.1–6The apolipoprotein E phenotype (apolipoprotein E
is coded by a gene (e) that exhibits allelic variants e2, e3, and
e4) has been linked consistently and strongly to the appear-
ance of Alzheimer’s disease and to some aspects of cognitive
decline in elderly cohorts.1,2,5–9Inheritance of E4(the prod-
uct of e4) especially in the homozygous form, has been as-
sociated in cross-sectional studies with risk of Alzheimer’s
disease and poorer global cognitive function, episodic
memory, and executive function, although the magnitude
of the effect is modest.10,11Longitudinal data also show
good evidence of an association between E4and risk of
Alzheimer’s disease and dementia,12,13although the rela-
major challenge in the quest to promote healthy aging
is to uncover the determinants of cognitive decline in
Address correspondence to Professor Chris J. Packard, Department of Vas-
cular Biochemistry, 4th Floor University Block, Glasgow Royal Infirmary,
Glasgow G31 2ER. E-mail: firstname.lastname@example.org
From theaDepartment of Vascular Biochemistry,bDivision of Cardiovascular
and Medical Sciences, andcRobertson Centre for Biostatistics, University of
Glasgow, Glasgow, Scotland;dSection of Gerontology and Geriatrics, Leiden
University Medical Center, Leiden, The Netherlands; Departments of
ePsychiatry andfNeuropsychology, Maastricht University, Maastricht, The
Netherlands; Departments ofgPharmacology and Therapeutics,hGeriatric
Medicine,iEpidemiology and Public Health, andjNeurology, Cork University
Hospital, Cork, Ireland;kNorth Glasgow Operating Division, National
Health Service Greater Glasgow and Clyde, Glasgow, Scotland; and De-
partmentsoflNeurologyandmCardiology,LeidenUniversity Medical Centre,
Leiden, The Netherlands.
r 2007, Copyright the Authors
Journal compilation r 2007, The American Geriatrics Society
tionship between the E4genotype and age-associated cog-
nitive decline without dementia has been less clear.14One
study15found a link in twins between E4and deterioration
of working memory, and others have reported greater de-
cline in verbal memory and executive function in those with
E4.16–18In contrast, another study13found that possession
of E4did not modify progression of cognitive decline in the
preclinical period of Alzheimer’s disease, and a third
study19reported slower rates of memory decline in E4car-
riers. Likewise, it was found that E4did not contribute to
prediction of cognitive decline in the very old but that
cerebrovascular disease was a major risk factor for this and
for progression to dementia.20It is important, therefore, to
widen the evidence base for the effects of E4on cognitive
decline before the appearance of dementia to assess how
early this genetic variation has an effect. The role of
apolipoprotein E in brain physiology and the precise nature
of the influence of the phenotypic variation on the patho-
genesis of neurodegenerative disease is unknown, although
clues are emerging from molecular studies21and from
investigations of white matter structure.22
The nature of the association between lipid and lipopro-
conflicting results being reported.23–26Recently, low high-
density lipoprotein cholesterol (HDL-C) levels have been
linked to poorer cognitive function, independent of an effect
on cardiovascular disease in the oldest old,23whereas no as-
sociation between total plasma cholesterol or HDL-C and
risk of Alzheimer’s disease was observed in another study.26
The Prospective Study of Pravastatin in the Elderly at
Risk (PROSPER) was a trial of statin use for the prevention
of cardiovascular and cerebrovascular disease in 5,804 men
and women aged 70 to 82.27It demonstrated that pravas-
tatin treatment was associated with a reduction in vascular
events. Part of the trial design was to examine the effect of
statin therapy on cognition. To this end, a battery of cog-
nitive function tests was administered at baseline and an-
nually during follow-up28and linked to initial lipid levels,
apolipoprotein E phenotype, and treatment allocation. It
was reported previously that pravastatin use did not affect
cognitive decline.27Here, the effect of apolipoprotein E
phenotype and lipoprotein levels on cognitive function at
baseline and on cognitive decline are described in this large
cohort. On the basis of previous workin smaller studies and
the role of vascular disease in dementia, the working hy-
pothesis was that apolipoprotein E phenotype and levels of
low-density lipoprotein cholesterol (LDL-C) and HDL-C
would influence cognition.29–31
Study Design and Subjects
cerebrovascular events in older subjects with a history of
vascular disease or at high risk of an event due to a history of
smoking, hypertension, or diabetes mellitus.32A total of
5,804 individuals (2,804 men, 3,000 women; aged 70–82)
were recruited in Scotland, Ireland, and the Netherlands and
randomized to receive 40mg/d pravastatin or placebo daily.
As part of the design of the study, whether statin treatment
had any effect on cognitive decline was assessed by admin-
istering a battery of cognitive function tests.28Determinants
ofcognitive functionatbaselineand ofcognitivedeclinewere
sought from the risk factor profiles, including apolipoprotein
E phenotype, which was assessed during screening. There
were restrictions on entry; plasma cholesterol had to be in the
range of 4.0 to 9.0mmol/L and plasma triglyceride less than
6.0mmol/L. The institutional ethics review boards of the
centers approved the study, and all participants gave written
Plasma cholesterol, triglyceride, LDL-C, and HDL-C were
measured twice at fasting visits during the placebo run-in
phase according to the Lipid Research Clinics protocol33
in a central laboratory that was standardized through
the Centers for Disease Control and Prevention network.
Apolipoprotein E phenotype was determined on plasma
samples using Western blotting following a method
previously established.34Subjects were classified according
to the presence of the E2, E3, or E4bands on gel blots. The
gel phenotyping method shows high concordance (495%)
with genotype testing according to allele-specific oligonu-
cleotide assay (unpublished results).
A detailed description of the cognitive tests used in the
study has been published previously.28The Mini-Mental
State Examination (MMSE) is used widely to screen for
cognitive impairment and dementia.35A cutoff score of 24
pointsor more (out of 30) was used as an inclusion criterion
to eliminate those with poor cognitive function at baseline.
Memory was tested using the Picture-Word Recall test
based on the Groningen-Fifteen Words test.28,36This
measures recall, both immediate and after 20 minutes, of
15 pictures (rather than words, to overcome any language
problem). The outcome variable is the mean number of
correctly recalled pictures over three immediate trials and
number recalled after the delay. Attention and processing
were assessed using the Stroop-Color Word test37and the
Letter-Digit Coding test.28The former, in the key Part III
of the test, presents color names printed in incongruously
colored ink (e.g., the word green printed in blue ink).
Performance, timed in seconds to complete the test,
measures the ability to discard the irrelevant name (green)
in favor of the color of the ink (blue). The latter asks the
subject to fill in digits next to letters according to a key;
the outcome is the number of correct entries in 60 seconds.
Subjects were assessed twice (2 weeks apart) at baseline to
allow for any training effect. The results of the second test
were taken as the starting estimate of cognitive function.
Decline in basic activities of daily living (ADLs) was
assessed using the 20-point variation38of the Barthel
Index39and in extended activities instrumental activities of
daily living (IADLs) using a 14-point score.40All tests were
repeated at 9, 18, and 30 months and at the final trial visit.
Dementia was recorded as an adverse event if diagnosed by
an attending physician; no trial-specific assessment was per-
formed, and it was not an adjudicated endpoint.
Subjects were divided initially into three categories
(E41 (phenotypes E3/4, E2/4, and E4/4), E3/3(the common-
PACKARD ET AL.
NOVEMBER 2007–VOL. 55, NO. 11 JAGS
est phenotype), and E21 (phenotypes E2/3and E2/2)) to test
for an effect of phenotype on cognition. No significant
difference was seen in any result between E3/3and E21
subjects, and these were combined into an E4? category.
Baseline characteristics were compared between the E41
and E4? groups using two-sample t-tests for continuous
part of this demographic assessment, overall risk of stroke
was estimated as described previously,41and the computed
risk was compared between E41 and E4? subjects. Cog-
nitive measures at baseline and decline in cognition were
compared between phenotype categories using linear mod-
els adjusted for age, sex, country, educational attainment,
history of vascular disease, history of myocardial infarction,
history of stroke or transient ischemic attack, smoking, use
of antihypertension medication, blood pressure, body mass
index, history of diabetes mellitus, triglyceride and lipopro-
tein cholesterol levels, and where appropriate, version of
test used. Models for change in cognitive function from
baseline (referred to as cognitive decline) were also adjusted
for treatment allocation and baseline test scores. Adjusted
of the three original phenotype categories, and mean differ-
ences with 95% confidence intervals (CIs) and P-values are
given for the E41 and E4? groups. Cognitive decline was
defined as the difference between subjects’ last recorded
follow-up measurement and the second of two baseline
measurements. In further analyses, only subjects with
MMSE scores of 26 to 30 were included (n55,295). Time
to dementia (as a recorded adverse event), or in a separate
analysis to an MMSE score less than 24, in the E41 and
E4? groups was estimated using Cox proportional hazard
models. Kaplan-Meier curves were generated for incident
dementia in the E41 and E4? groups.
The E41 group was split into E4homozygotes (E4/4)
and E4heterozygotes (E2/4and E3/4). Mean results of cog-
nition tests at baseline, and of cognitive decline, were com-
pared between these phenotypes and the E4? group.
Results were also analyzed as percentage change over base-
line. Anaveragepercentagedeclinein memoryfunctionwas
calculated from the results of the immediate and delayed
Picture-Word recall. This variable was then compared in
E4?, E4heterozygotes, and E4homozygotes in subjects
with baseline MMSE scores of 24 to 30 (evaluable subjects,
n55,004), 26 to 30 (n54,611), and 28–30 (n53,456).
Repeated-measures models that included all of the mea-
surements recorded on a subject and incorporated a linear
separation between the groups were investigated, although
these more-complex models did not add statistical power,
possibly because of the additional assumptions made being
invalid. This approach is not reported here; rather the
simpler change from baseline analysis was used.
and HDL-C, linear models were constructed as for apolipo-
protein E analysis. For descriptive purposes, least squares
means and standard errors adjusted for baseline variables, as
noted previously, as well as apolipoprotein E4status, are
given for tertiles of LDL-C and HDL-C. P-values for con-
tinuous measures of HDL-C and LDL-C are also reported.
For all analyses, only complete and reliable test results
of cognitive function (as indicated by the study nurse who
administered the test) were included.
Of the 5,804 subjects recruited to the study, apolipoprotein E
phenotyping was available for 95.5%; 38 (0.7%) were E2/2,
621 (11.2%) were E2/3, 119 (2.2%) were E2/4, 3,496 (63.1%)
were E3/3,1,169 (21.1%) were E3/4,and 101 (1.8%)were E4/4.
Translated into genotypes, the frequencies were in Hardy-
Weinberg equilibrium. Thus, a total of 4,155 subjects were
classified as E4? and 1,389 as E41. In univariate analysis, the
latter group had higher plasma total cholesterol (mean ? stan-
dard deviation 5.84 ? 0.91 vs 5.63 ? 0.90mmol/L, Po.001),
higher LDL-C (3.96 ? 0.81 vs 3.74 ? 0.79mmol/L, Po
.001), lower HDL-C (1.25 ? 0.34 vs 1.29 ? 0.35mmol/L,
Po.001), and higher plasma triglyceride (1.60 ? 0.75 vs
1.52 ? 0.68mmol/L, Po.001) levels than the former. No sig-
nificant difference was observed in blood pressure; sex; history
of coronary disease, stroke, or hypertension; smoking habit;
years of education; or a composite score of stroke risk.41Sub-
jects with diabetes mellitus were approximately 25% less prev-
marginally younger than those who were E4?.
At baseline, age, sex, and educational status were
major determinants of all tests of cognitive function. His-
tory of stroke, diabetes mellitus, and smoking were predic-
not of memory function (i.e., immediate and delayed Pic-
ture-Word Recall). History of vascular disease, history of
stroke, body mass index, diabetes mellitus, and smoking
were predictors of scores in the Barthel and IADL indices
(data not shown).
Influence of Apolipoprotein E Phenotype
Apolipoprotein E phenotype had a significant association
with a number of tests of cognitive function, determined up-
section, none of the E21 versus E3/3comparisons indicated a
significant difference, and these phenotypes were combined
in the E4? category. Subjects in the E41 group performed
significantly less well on the Stroop test and Picture-Word
Recall (immediate and delayed) than subjects in the E4?
group and had marginally, but significantly, poorer scores on
the MMSE (Table 1). No significant influence of apolipo-
protein E4status was seen for the Letter-Digit Coding test or
the Barthel or IADL indices. Restricting the analyses to sub-
jects with baseline MMSE scores of 26 to 30 did not alter
these associations (data not shown).
Follow-up assessments on at least one of the cognitive
tests were available for 98.4% (5,454/5,544) of subjects
with apolipoprotein E phenotyping available, and 4,942
completed at least 3 years of assessment. Of the 90 subjects
without any follow-up assessment, 72 died, 11 refused to
participate or did not attend a follow up cognitive assess-
ment, and seven had a nonfatal adverse event. Over the
average 3.2 years (range 0.7–4.2 years) of follow-up,
apolipoprotein E4 status significantly influenced change
in scores on the cognitive tests associated with memory
(immediate and delayed Picture-Word Recall) but not those
for attention and processing (Stroop, Letter-Digit Coding)
(Table 2). Again, none of the comparisons between E3/3
and E21 subjects was significant. Those in the E41 group
had significantly greater decrements than those in the
E4? group on MMSE, Barthel, and IADL scores. At the
APOLIPOPROTEIN E AND COGNITION IN PROSPER
1779 JAGSNOVEMBER 2007–VOL. 55, NO. 11
end of the study, 85 (6.4%) in the E41 group and 131
(3.3%) in the E4? group had scores less than 24 on the
MMSE (Po.001 for difference in incidence), and there
were 37 cases of dementia reported as an adverse event
(2.7%) in the E41 group and 46 (1.1%) in the E4? group,
a 2.48-fold difference in risk (Po.001). Including only
those with MMSE scores of 26 to 30 at baseline produced
the same qualitative result (data not shown).
Treatment assignment (i.e., to placebo or pravastatin
therapy) was included in the multivariate models. It was
reported previously that pravastatin use during the trial had
no significant effect on change in cognitive function.42This
was further tested by examining the interaction between
apolipoprotein E phenotype, treatment, and cognition. No
significant associations were found with any of the tests,
confirming the lack of effect of statin therapy on cognitive
decline (data not shown).
Gene Dose Effect
Further analyses were performed in which E4homozygotes
At baseline, neither E4homozygotes nor E4heterozygotes
differed from E4? on the Letter-Digit Coding test or on the
Table 1. Comparison of Baseline Measures of Cognition
According to Apolipoprotein E Phenotype
E4? Mean Difference
Stroop Part III (seconds to complete) (n55,163)
Letter-Digit Coding (number correct) (n55,185)
Picture-Word Recall (number recalled) (n55,222)
Barthel Index (score) (n55,544)
Instrumental activity of daily living index (score) (n55,544)
Mini-Mental State Examination (score) (n55,479)
2.79 (1.30, –4.28)
?0.36 (0.77, –0.05).08
?0.20 (?0.31, ?0.09)
?0.32 (?0.48, ?0.16)
?0.04 (?0.08, –0.00) .08
?0.05 (?0.11, –0.01) .12
?0.21 (?0.30, ?0.12)
?E41 includes E4/4, E4/3, and E4/2phenotypes, E21 includes E3/2and E2/2.
wAdjusted for age, sex, country, education, history of vascular disease, history
of myocardial infarction, history of diabetes mellitus, history of stroke or
transient ischemicattack, smoking,use ofantihypertensivemedication, blood
pressure, body mass index, low-density lipoprotein cholesterol, high-density
lipoprotein cholesterol, triglyceride, and where applicable test version.
zComparison of mean test results in E41 (E4/4, E4/2, E4/3) subjects versus E4?
(E3/3, E2/3, E2/2) subjects.
§Apolipoprotein E41 group result was significantly different from E3/3and
Table 2. Comparison of Change in Cognitive Function
According to Apolipoprotein E Phenotype
Cognitive Test and
E41 versus E4?
Stroop Part III (seconds to complete) (n54,897)§
Letter-Digit Coding (number correct) (n54,953)
Picture Word Recall (number recalled) (n55,004)
Barthel Index (score) (n55,453)
Instrumental activity of daily living index (score) (n55,454)
Mini-Mental State Examination (score) (n55,260)
?E41 includes E4/4, E4/3, E4/2; E21 includes E2/3, E2/2.
wChange score (last recorded result minus baseline result) adjusted for age,
sex, country, education, history of vascular disease, history of myocardial
infarction, history of diabetes mellitus, history of stroke or transient ischemic
attack, smoking, use of antihypertensive medication, blood pressure, body
mass index, low-density lipoprotein cholesterol, high-density lipoprotein
cholesterol, triglyceride, and where applicable test version, plus baseline test
score and treatment allocation.
zComparison of mean change in E41 and E4? groups.
§A positive change in the Stroop test indicates deteriorating performance.
kApolipoprotein E41 group result was significantly different from E3/3result.
#Apolipoprotein E41 group result was significantly different from E21 result.
10.95 (0.39–2.29) .17
?0.26 (0.53–0.01) .06
?0.22 (0.33, ?0.11)
?0.30 (0.46, ?0.15)
?0.13 (0.24, ?0.02) .02
?0.32 (0.45, ?0.18)
?0.27 (0.39, ?0.16)
PACKARD ET AL.
NOVEMBER 2007–VOL. 55, NO. 11 JAGS
Barthel or IADL indices. MMSE scores of E4homozygotes
were significantly lower than those with E4? (P5.004), but
the comparison between the former and E4heterozygotes
was not significant (P5.11), although the means differed.
Memory tests exhibited a trend toward stepwise reduction
with E4gene dosage. E4homozygotes and heterozygotes had
significantly lower scores than E4?, although possibly be-
cause of small subgroup size, the mean scores of homozy-
gotes while tending to be lower than those of heterozygotes,
were not significantly different (P5.17 for immediate mem-
ory; P5.07 for delayed memory).
The decline in performance on the Stroop and Letter-
Digit Coding tests was not significantlydifferent in E4?,E4
heterozygotes, and E4homozygotes (Table 3). For the im-
mediate and delayed memory tests, the decrement in mean
score for the E4homozygotes was significantly greater than
that for the heterozygotes (P5.006 for immediate; Po.001
for delayed memory). In turn, the decline for E4heterozy-
gotes was greater than that for E4? (P5.001 for imme-
diate memory; P5.004 for delayed memory). A similar
pattern was present for change in MMSE (Po.001 for E4
homozygotes vs heterozygotes; Po.001 for heterozygotes
vs E4?). To test the effect of baseline MMSE score on the
association between apolipoprotein E phenotype and de-
cline in memory function, an average percentage decline in
the scores for the immediate and delayed Picture-Word re-
call was computed. This combined variable gave significant
differences for E4homozygotes, E4heterozygotes, and E4?
subjects in the whole cohort (Figure 1) and the apolipo-
protein E effect persisted when subjects were restricted to
those with MMSE scores of 26 to 30. A similar pattern was
seen in subjects with baseline MMSE scores of 28 to 30,
with an overall significant effect of phenotype; intergroup
comparisons were at the border of significance, probably
because of smaller numbers of patients.
Influence of Lipoprotein Levels
Examination of the association between lipoprotein plasma
levels and cognitive function at baseline revealed generally
weaker relationships than seen with apolipoprotein E4sta-
tus (Table 4). Results of the Picture-Word Recall tests were
similar in all tertiles of LDL-C and HDL-C. The Letter-
Digit test was significantly related to LDL-C, with poorer
Table 3. Apolipoprotein E4 Status and Cognitive Function
Least Square Mean (Standard Error)
Stroop Part III (seconds to complete)z
Letter-Digit Coding (number correct)
Picture-Word Recall (number recalled)
Barthel Index (score)
Instrumental activities of daily living index (score)
Mini-Mental State Examination (score)
?E4? includes E2/2, E2/3, E3/3; E4heterozygotes include E4/3, E4/2; E4homozygotes are E4/4. The number of subjects given is the maximum in each group. Data are
adjusted for age, sex, country, education, history of vascular disease, history of myocardial infarction, history of diabetes mellitus, history of stroke or transient
ischemic attack, smoking, use of antihypertensive medication, blood pressure, body mass index, low-density lipoprotein cholesterol, high-density lipoprotein
cholesterol, triglyceride, and where applicable test version. Change score (last recorded result minus baseline result) was also adjusted for baseline test score and
baseline test score and treatment allocation.
wP-value for test of heterogeneity across groups.
zNote a positive change in the Stroop test indicates a deteriorating performance.
§E4heterozygotes or E4homozygotes differ significantly from E4?.
??E4homozygotes differ significantly from E4heterozygotes.
APOLIPOPROTEIN E AND COGNITION IN PROSPER
1781 JAGSNOVEMBER 2007–VOL. 55, NO. 11
scores being seen in the lowest LDL-C tertile, and the Stroop
test showed a trend in the same direction. The Barthel and
IADL indices were significantly related to LDL-C level, with
lower scores again being seen in the lowest tertiles. No sig-
nificant association was observed between these indices and
Lipoprotein concentrations at baseline had no signifi-
cant effect on change in cognitive function or decrease in
score on the MMSE or the Barthel or IADL index (Table 4).
This study of the determinants of cognition in elderly adults
found a marked effect of apolipoprotein E phenotype on
memory performance that was apparent in the cross-sec-
tional comparison at baseline and in the change over an
average of 3.2 years of follow-up. Greater decline in mem-
ory in those with apolipoprotein E4occurred irrespective of
baseline levels of global cognitive function (as assessed by
MMSE). This suggested that the phenotype has a general
population effect,not justaninfluence inthose withsigns of
impaired cognition. Furthermore, the association between
apolipoprotein E4and greater decline on the IADL index
indicates that, at a population level, this phenotype has
clinically important effects even in those who do not exhibit
frank dementia. In contrast to the changes in memory per-
formance, over this time scale, this genetic variation affect-
ed tests of the attention and processing cognitive domain to
a much lesser extent. Plasma lipoprotein levels had limited
influence on cognition at baseline and on its decline during
After the initial reports of an association between Alz-
heimer’s disease and apolipoprotein E phenotype,7it has
been demonstrated repeatedly that subjects possessing apo-
lipoprotein E4, particularly homozygotes, have a high risk
of developing dementia.10–15The underlying mechanism is
thought in part to be more beta-amyloid deposits in the
presence of E4than with other isoforms of the apolipopro-
tein,8,43although it is increasingly clear that apolipoprotein
E phenotype has multiple effects on neurobiology.21Alz-
heimer’s disease, characterized by episodic memory loss, is
likely to be the end result of a long subclinical degenerative
process, and the question arises as to when and how apo-
has been noted,11,13early studies of the influence of apo-
lipoprotein E4on the rate of cognitive decline in the appar-
entlynormalpopulation involvedsmallnumbersof subjects
and included those with frank dementia or cognitive im-
pairment at baseline or during subsequent evaluations. For
example, one study44reported a greater decrease in MMSE
score and in information processing speed in elderly sub-
jects with no initial cognitive impairment who were
apolipoprotein E4 carriers; results for memory function
were less clearcut, showing no significant drop. In that
study, there was a fall of 0.8 to 1.4 points in the MMSE
score, indicating that a number of subjects developed sig-
nificant cognitive impairment during the assessment period.
In the Cognitive Function andAgeing Study, apolipoprotein
E phenotype was linked to risk of dementia (as determined
according to MMSE score) in the cross-sectional survey, but
no association was seen with incident dementia over 6
years.45Results supporting a link between apolipoprotein
E4and memory impairment came from a study of 611
Decline in Picture-Word Recall
P = .001
P = 0.008
P = .004
Overall P < .001
P < .001
P = 0.019
P = .002
Overall P < .001
P = 0.012
P = 0.058
P = 0.059
Overall P = .010
MMSE 24 – 30
MMSE 26 – 30
MMSE 28 – 30
Figure 1. Decline in memory performance in apolipoprotein E4homozygotes, E4heterozygotes, and those without E4. An average
percentage change was calculated from the results of the immediate and delayed Picture-Word Recall scores. The influence of
apolipoprotein E4status on this variable was examined in subjects with differing Mini-Mental State Examination (MMSE) score
ranges at baseline. Least square mean changes (with standard errors) are given for E4? subjects, E4heterozygotes (E4het), and E4
homozygotes (E4hom). These were adjusted for the factors listed in the footnote to Table 1 and for treatment allocation. Significance
levels are given for an overall effect and for intergroup comparisons. The number of evaluable subjects within a specific baseline
MMSE range is provided (i.e., those with apolipoprotein E analysis and follow-up cognitive function scores available).
PACKARD ET AL.
NOVEMBER 2007–VOL. 55, NO. 11JAGS
elderly clergymen, in which it was seen that E4had a more-
pronounced influence on the decline in episodic memory
than on other cognitive domains.46Again, the 16.7% inci-
dence of clinically evident Alzheimer’s disease may have in-
fluenced the result. More recently, in a report on the long-
who were free of obvious dementia at baseline, apolipopro-
tein E4was associated at 7 years but not at 3 years of follow-
up with reduced memory function and decreased ability in
other cognitive areas.47Furthermore, in 840 cognitively nor-
mal elderly subjects followed for 3.5 years, apolipoprotein E
phenotype and depression contributed to the appearance of
mild cognitive impairment, apparently in an additive fash-
ion.6In contrast, in a population-based study13in which
subjects with dementia at baseline and at various stages of
follow-up were excluded from the analysis, a greater preclin-
ical decline in MMSE was present in participants who went
on to develop dementia, but apolipoprotein E status had no
influence on this change or on the rate of cognitive decline in
those who were free of dementia throughout. Thus, debate
continues about the influence of apolipoprotein E on cogni-
tion in the healthy aging population without dementia.
The present study has a number of features that should
be borne in mind in assessing its conclusions. It followed a
large cohort, and the majority of subjects had good cogni-
tive function at baseline. Because of short follow-up, few
developed dementia or poor general cognitive function.
(Mean decrease in MMSE score was 0.13, and 89.3% of
E41 and 92.8% of E4? subjects had MMSE scores 424 at
their final assessment.) Dementia was not a formal study-
specific diagnosis but was recorded as an adverse event. On
entry, subjects were selected to have a history of, or to be at
risk of, a vascular event, and the results should be extrap-
olated to the general public in light of this design feature. In
addition, half of the subjects were allocated to receive
pravastatin, but the drug had no discernible effect on cog-
nitive decline, and it is likely that treatment did not con-
found the result.27
Overall, a consistent influence of apolipoprotein E4on
observed at baseline and during follow-up. This contrasted
with the relative lack of influence of E4on attention and pro-
was evidence of a gene dosage effect, with E4homozygotes
faring worse than heterozygotes. These results echo the find-
ings of a meta-analysis of cross-sectional studies by,11which
concluded that, in the general aging population, the influence
to certain domains of cognitive function. The findings of the
current study, together with other,11,15but not all,13,20com-
parable observations, raise the possibility that apolipoprotein
E4is a determinant of the trajectory of memory loss (and con-
sequent cognitive decline) in the elderly general population.
Table 4. Comparison of Baseline Measures of Cognition and Cognitive Decline According to Tertile of Low-Density
Lipoprotein Cholesterol (LDL-C) and High-Density Lipoprotein Cholesterol (HDL-C)
Stroop Part III
(seconds to complete)
1.00 .75.62 .16
Barthel Index (score) .001 .59.08 .34
Instrumental activity of
daily living index (score)
.01.96 .06 .23
?The numbers for each test are shown in Table 1 for baseline models and Table 2 for change models.
wTertiles of LDL-C were I o3.40, II 3.40–4.10, III 44.10mmol/L, tertiles of HDL-C were I o1.10, II 1.10–1.37, III 41.37 mmol/L.
§P-value for continuous measure of HDL-C or LDL-C (as appropriate).
zLeast squares mean (standard error) for baseline result or for change (last recorded results minus baseline result) score adjusted for age, sex, country, education,
history of vascular disease, history of myocardial infarction, history of stroke or transient ischemic attack, smoking, use of antihypertensive medication, blood
pressure, body mass index, HDL-C or LDL-C (i.e., HDL-C was included in the model when LDL-C levels were tested and vice versa), plasma triglyceride,
apolipoprotein E4, and where applicable test version. Change was also adjusted for baseline test score and treatment allocation.
APOLIPOPROTEIN E AND COGNITION IN PROSPER
1783 JAGSNOVEMBER 2007–VOL. 55, NO. 11
Vascular dementia and Alzheimer’s disease have a con-
siderable degree of overlap in terms of neuropsychological
impairments, although there is a tendency for vascular de-
mentia to be associated with greater deficits in attention,
speed of information processing, and executive function,48
whereas subjects with Alzheimer’s disease have more prob-
lems with semantic memory.49,50Thus, it was of interest to
evaluate whether plasma lipoproteins potentially linked to
vascular disease influenced specific domains of cognitive im-
pairment. It was found that associations between lipoprotein
levels and cognitive function tests were of borderline signifi-
canceor not significant at baseline and during follow-up. The
most notable observation was that low LDL-C levels inde-
pendent of vascular disease history were associated with
similar trend. Analogous findings were reported in the Leiden
85 Plus study,23with low HDL-C being linked to poorer
MMSE scores in subjects with and without a history of car-
diovascular disease, although in that study, LDL-C showed
no association with cognition. One possible explanation for
are nutritionally compromised, and this leads to both
lipoproteins being reduced, although a direct influence of
lipoproteins on some cognitive aspects of disability cannot be
excluded. Similar results were reported recently in a study
that found no association between cholesterol or HDL-C and
risk of dementia in a community-based study of cognitively
intact elderly subjects.26
In conclusion, this large-scale cognition study embed-
ded in a clinical trial revealed that, over an average 3.2-year
period, the presence of the apolipoprotein E4isoform was
associated with greater decline in memory performance
without apparently affecting attention or processing ability.
The apolipoprotein E4 isoform was linked also to more-
pronounced deterioration in indices of ADLs and IADLs.
This manuscript was prepared with the excellent secretarial
assistance of Ms. Shelley Wilkie. We are indebted to Mrs.
Dorothy Bedford and her technical staff for their help with
the analysis of apolipoprotein E phenotype. The authors
gratefully acknowledge the contribution of Dr. Peter Houx,
who sadly died during the course of this study.
The PROSPER study group comprised
Executive Committee: (Glasgow) J. Shepherd (Chairman
Macfarlane, C. J. Packard, D. J. Stott; (Leiden) G. J. Blauw
(Principal Investigator), E. L. E. M. Bollen, A. M. Kamper, R.
G. J. Westendorp; (Cork) M. B. Murphy (Principal Investiga-
tor), B. M. Buckley, M. Hyl, I. J. Perry.
Endpoint Committee: S. M. Cobbe (Chairman), J. W.
Jukema, P. W. Macfarlane, A. E. Meinders, D. J. Stott, B. J.
Sweeney, C. Twomey.
Data and Safety Monitoring Committee: W. V. Brown
(Chairman), H. C. Diener, J. Feely, I. Ford (Nonvoting),
T. Pearson, S. Pocock, P. A. van Zwieten.
Conflictof Interest:Thestudywassupported byagrant
from Bristol-Myers Squibb. This work was supported
by a grant from the sponsor of the PROSPER trial. The
editor in chief has determined that the authors have no
conflict of interest related to this manuscript. CP has
received research support or honoraria from AstraZeneca,
Sanofi-Aventis, MSD, Schering Plough, and Glaxo Smith-
Kline (GSK). MC has received research support or hono-
raria from AstraZeneca, GSK, and Sanofi-Aventis. RW has
received research support from Bristol-Myers Squibb. JS
has received research support or honoraria from Astra-
Zeneca, GSK, MSD, Merck, and Schering-Plough. BB has
received research support or honoraria from Pfizer, Bristol
Myers Squibb, AstraZeneca, and Servier. AG has received
research support from AstraZeneca. SC has received re-
search support from AstraZeneca.
Author Contributions: All authors contributed to the
interpretation of the results and drafting of the manuscript
(CJP wrote the first draft). The executive committee (CJP,
RGW, JS, GJB, MBM, ELEMB, BMB, SMC, IF, AG, MM,
design and conduct of PROSPER. MC undertook the apo-
lipoprotein E analysis. HMM was responsible for statistical
analysis of the cognitive decline findings. JJ oversaw inter-
pretation of cognitive function tests.
Sponsor’s Role: PROSPER was sponsored by Bristol-
Myers Squibb Ltd and company personnel on the Executive
Committee helped design and conduct the trial. The spon-
sor had no role in the present analysis.
1. Dik MG, Deeg DJ, Bouter LM et al. Stroke and apolipoprotein E epsilon4 are
independent risk factors for cognitive decline: A population-based study.
2. Haan MN, Shemanski L, Jagust WJ et al. The role of APOE epsilon4 in mod-
ulating effects of other risk factors for cognitive decline in elderly persons.
3. Launer LJ, Ross GW, Petrovitch H et al. Midlife blood pressure and dementia:
The Honolulu-Asia aging study. Neurobiol Aging 2000;21:49–55.
4. Ott A, Stolk RP, van Harskamp F et al. Diabetes mellitus and the risk of
dementia: The Rotterdam Study. Neurology 1999;53:1937–1942.
5. Podewils LJ, Guallar E, Kuller LH et al. Physical activity, APOE genotype, and
dementia risk: Findings from the Cardiovascular Health Cognition Study. Am
J Epidemiol 2005;161:639–651.
6. Geda YE, Knopman DS, Mrazek DA et al. Depression, apolipoprotein E ge-
notype, and the incidence of mild cognitive impairment: A prospective cohort
study. Arch Neurol 2006;63:435–440.
7. Saunders AM, Strittmatter WJ, Schmechel D et al. Association of apolipopro-
teinE allele epsilon 4with late-onset familial and sporadic Alzheimer’s disease.
8. Smith JD. Apolipoproteins and aging: Emerging mechanisms. Ageing Res Rev
9. Mahley RW, Huang Y. Apolipoprotein E: From atherosclerosis to Alzheimer’s
disease and beyond. Curr Opin Lipidol 1999;10:207–217.
10. Farrer LA, Cupples LA, Haines JL et al.Effects of age, sex, and ethnicity on the
association between apolipoprotein E genotype and Alzheimer disease. A
meta-analysis. APOE and Alzheimer Disease Meta Analysis Consortium.
11. Small BJ, Rosnick CB, Fratiglioni L et al. Apolipoprotein E and cognitive
performance: A meta-analysis. Psychol Aging 2004;19:592–600.
12. Wilson RS, Schneider JA, Barnes LL et al. The apolipoprotein Eepsilon 4 allele
and decline in different cognitive systems during a 6-year period. Arch Neurol
13. Bunce D, Fratiglioni L, Small BJ et al. APOE and cognitive decline in preclin-
icalAlzheimer diseaseand non-demented aging.Neurology2004;63:816–821.
14. Small BJ, Graves AB, McEvoy CL et al. Is APOE-epsilon4 a risk factor for
cognitive impairment in normal aging? Neurology 2000;54:2082–2088.
15. Reynolds CA, Prince JA, Feuk L et al. Longitudinal memory performance
during normal aging: Twin association models of APOE and other Alzheimer
candidate genes. Behav Genet 2006;36:185–194.
16. Swan GE, Lessov-Schlaggar CN, Carmelli D et al. Apolipoprotein E epsilon4
and change in cognitive functioning in community-dwelling older adults. J
Geriatr Psychiatry Neurol 2005;18:196–201.
PACKARD ET AL.
NOVEMBER 2007–VOL. 55, NO. 11JAGS
17. Deary IJ, Whiteman MC, Pattie A et al. Apolipoprotein E gene variability and Download full-text
cognitive functions at age 79: A follow-up of the Scottish mental survey of
1932. Psychol Aging 2004;19:367–371.
18. Hofer SM, Christensen H, MacKinnon AJ et al. Change in cognitive func-
tioning associated with apoE genotype in a community sample of older adults.
Psychol Aging 2002;17:194–208.
19. Reitz C, Luchsinger JA, Tang MX et al. Stroke and memory performance in
elderly persons without dementia. Arch Neurol 2006;63:571–576.
20. Qiu C, Winblad B, Fratiglioni L. Cerebrovascular disease, APOE epsilon4
allele and cognitive decline in a cognitively normal population. Neurol Res
21. Huang Y. Apolipoprotein E and Alzheimer disease. Neurology 2006;66:
22. Persson J, Lind J, Larsson A et al. Altered brain white matter integrity in
healthy carriers of the APOE epsilon4 allele: A risk for AD? Neurology 2006;
23. van Exel E, de Craen AJ, Gussekloo J et al. Association between high-density
lipoprotein and cognitive impairment in the oldest old. Ann Neurol 2002;
24. Kivipelto M, Helkala EL, Laakso MP et al. Midlife vascular risk factors and
Alzheimer’s disease in later life: Longitudinal, population based study. BMJ
25. Wieringa GE, Burlinson S, Rafferty JA et al. Apolipoprotein E genotypes and
serum lipid levels in Alzheimer’s disease and multi-infarct dementia. Int J
Geriatr Psychiatry 1997;12:359–362.
26. Li G, Shofer JB, Kukull WA et al. Serum cholesterol and risk of
Alzheimer disease: A community-based cohort study. Neurology 2005;65:
27. Shepherd J, Blauw GJ, Murphy MB et al. Pravastatin in elderly individuals at
risk of vascular disease (PROSPER): A randomised controlled trial. Lancet
28. Houx PJ, Shepherd J, Blauw GJ et al. Testing cognitive function in elderly
populations: The prosper study. Prospective Study of Pravastatin in the Elderly
at Risk. J Neurol Neurosurg Psychiatry 2002;73:385–389.
29. Kalmijn S, Feskens EJ, Launer LJ et al. Cerebrovascular disease, the apolipo-
protein e4 allele, and cognitive decline in a community-based study of elderly
men. Stroke 1996;27:2230–2235.
30. Kalmijn S, Foley D, White Let al. Metabolic cardiovascular syndrome and risk
of dementia in Japanese-American elderly men. The Honolulu-Asia Aging
Study. Arterioscler Thromb Vasc Biol 2000;20:2255–2260.
31. O’Brien JT, Erkinjuntti T, Reisberg B et al. Vascular cognitive impairment.
Lancet Neurol 2003;2:89–98.
32. Shepherd J, Blauw GJ, Murphy MB et al. The design of a prospective study of
Pravastatin in the Elderly at Risk (PROSPER). PROSPER study group. PRO-
spective Study of Pravastatin in the Elderly at Risk. Am J Cardiol 1999;
33. The lipid research clinics program epidemiology committee. Plasma lipid dis-
tributions in selected North American populations: The lipid research clinics
program prevalence study. Circulation 1979;60:427–439.
34. Havekes LM, de Knijff P, Beisiegel U et al. A rapid micromethod for apolipo-
protein E phenotyping directly in serum. J Lipid Res 1987;28:455–463.
35. Folstein MF, Folstein SE, McHugh PR. ‘‘Mini-mental state’’. A practical
method for grading the cognitive state of patients for the clinician. J Psychiatr
36. Brand N,JollesJ.Learning andretrievalrateofwordspresentedauditorilyand
visually. J Gen Psychol 1985;112:201–210.
37. Houx PJ, Jolles J, Vreeling FW. Stroop interference: Aging effects assessed with
the Stroop Color-Word Test. Exp Aging Res 1993;19:209–224.
38. Hobart JC, Lamping DL, Freeman JA et al. Evidence-based measurement:
Which disability scale for neurologic rehabilitation? Neurology 2001;57:
39. Mahoney FI, Barthel DW. Functional evaluation: The Barthel Index. Md State
Med J 1965;14:61–65.
40. Lawton MP, Brody EM. Assessment of older people: Self-maintaining and
instrumental activities of daily living. Gerontologist 1969;9:179–186.
41. Wolf PA, D’Agostino RB, Belanger AJ et al. Probability of stroke: A risk profile
from the Framingham Study. Stroke 1991;22:312–318.
42. Shepherd J, Blauw GJ, Murphy MB et al. Pravastatin in elderly individuals at
risk of vascular disease (PROSPER): A randomised controlled trial. Lancet
43. Weisgraber KH, Mahley RW. Human apolipoprotein E: The Alzheimer’s
disease connection. FASEB J 1996;10:1485–1494.
44. Dik MG, Jonker C, Comijs HC et al. Memory complaints and APOE-epsilon4
accelerate cognitive decline in cognitively normal elderly. Neurology 2001;
45. Yip AG, Brayne C, Easton D et al. Apolipoprotein E4 is only a weak predictor
of dementia and cognitive decline in the general population. J Med Genet
46. Wilson RS, Schneider JA, Barnes LL et al. The apolipoprotein Eepsilon 4 allele
and decline in different cognitive systems during a 6-year period. Arch Neurol
47. Bretsky P, Guralnik JM, Launer L et al. The role of APOE-epsilon4 in lon-
gitudinalcognitive decline: MacArthur Studies ofSuccessfulAging. Neurology
48. O’Brien JT. Vascular cognitive impairment. Am J Geriatr Psychiatry 2006;
49. Oosterman JM, Scherder EJ. Distinguishing between vascular dementia and
Alzheimer’s disease by means of the WAIS: A meta-analysis. J Clin Exp
50. Poore QE, Rapport LJ, Fuerst DR et al. Word list generation performance in
Alzheimer’s disease and vascular dementia. Neuropsychol Dev Cogn B Aging
Neuropsychol Cogn 2006;13:86–94.
APOLIPOPROTEIN E AND COGNITION IN PROSPER
1785JAGS NOVEMBER 2007–VOL. 55, NO. 11