University of Southern California Dissertations and Theses

Identifying diverse pathways to cognitive decline in later life using genetic and environmental factors

(USC Thesis Other)

Identifying diverse pathways to cognitive decline in later life using genetic and environmental factors

PDF

Download a page range

Download transcript

Copy asset link

Request this asset

Transcript (if available)

Content Running head: GENES, LIFE STYLES, AND COGNITIVE DECLINE 1

Identifying Diverse Pathways to Cognitive Decline in Later Life Using Genetic and
Environmental Factors
Yusuke Kuroki
University of Southern California

GENES, LIFE STYLES, AND COGNITIVE DECLINE 2

Table of Contents
Page
Abstract 3
Introduction 4
Study 1: Method 15
Study 1: Results 21
Study 2: Method 22
Study 2: Results 25
Discussion 26
References 38
Tables 58
Figures 63
Appendix 68

GENES, LIFE STYLES, AND COGNITIVE DECLINE 3

Abstract
The ε4 allele of the apolipoprotein E is associated with dementia. However, ε4 is neither
necessary nor sufficient to cause dementia and leaves much of variability in dementia
unexplained. One possible reason for unexplained variability is gene-environment (G × E)
interactions. Are there lifestyle factors that can moderate the effect of APOE ε4 on memory? To
address this question, two analytical approaches were proposed. First, the gene-environment
interaction on episodic memory was examined using the structural equation modeling (SEM).
Specifically, the present study asked, “Does the effect of the ε4 on memory vary by education,
job authority, or physical activity?” Second, an innovative exploratory technique called structural
equation model tree (SEM Trees; Brandmaier, von Oertzen, McArdle, & Lindenberger, 2013)
was used. Data from the Wisconsin Longitudinal Study were used. Memory was assessed with
immediate recall, delayed recall and digit ordering. Results of SEM (n = 10,317) showed that
APOE ε4, education, job authority and physical activity affected memory; however, education,
job authority or physical activity did not moderate the effect of APOE ε4 on memory. Results of
SEM Trees (n = 4569) showed that gender, IQ, literacy use at work, personality, smoking, and
four genetic polymorphisms in APOC3, DRD2, CH25H, and SSADH produced complex
interactions and influenced memory decline in late life. However, the influences of genetic
variants on memory were subtle. This study also demonstrates the utility of SEM Trees as a
powerful technique for exploring the data within a theoretical context.

GENES, LIFE STYLES, AND COGNITIVE DECLINE 4

Identifying Diverse Pathways to Cognitive Decline in Later Life Using Genetic and
Environmental Factors
Dementia is a loss of cognitive function that occurs with the death or malfunctions of
neurons in the brain and is common among older adults. According to a recent report, the elderly
population in the U.S. will reach 72 million by 2030, representing nearly 20 percent of the total
U.S. population (Federal Interagency Forum on Aging-Related Statistics, 2012). Payments for
care for people with dementia are estimated to be $200 billion in 2012, and the cost is likely to
grow in the future (Alzheimer’s Association, 2012). Given the rapid growth of the elderly
population, it is increasingly important to find the causes of dementia and safe therapeutic
methods to prevent it.
Dementia is considered strongly influenced by genes. For example, one twin study
estimates heritability of Alzheimer disease (AD) to be 79% (Gatz et al., 2006). Much attention
has been paid to identifying the specific genes involved in pathogenesis of dementia, particularly
AD (reviewed by Tanzi, 2012). To date, the ε4 allele in the APOE gene (APOE ε4) is the only
genetic risk factor for late-onset Alzheimer disease. However, only about 40% of AD patients
have the ε4 allele. It is now commonly believed that dementia is caused by several genes
combined with lifestyle factors. Gene-environment (G × E) interactions may account for
unexplained variance of dementia.
Studying G × E interactions is important. Not only can it deepen the understanding of
fundamental mechanism underlying human diseases, but also it can uncover modifiable
environmental factors that can reduce disease risk for people with particular combinations of
genetic variants (Manuck & McCaffery, 2014). Stated simply, G × E interactions are phenomena
where the effects of some genetic variants on a given trait depend on environmental factors (or
GENES, LIFE STYLES, AND COGNITIVE DECLINE 5

conversely, the effects of some environmental factors on the trait depend on genetic variants). In
1996, expert panels published a statement arguing against screening for carriers of APOE ε4 for
predicting risk of AD in cognitive healthy people because of the absence of preventive treatment
(National Institute on Aging/Alzheimer’s Association Working Group). In the same vein, FDA
recently ordered 23andMe, a personal genomics company, to stop providing its customers their
odds of diseases based on their genetic profiles, because FDA was concerned that the customers
would perceive the results as medical diagnoses. As cheaper genotyping techniques became
available in recent years, there is a great deal of interest in finding environmental factors that can
modify the negative effects of risk genetic variants. With respect to dementia, factors that
influence the onset of dementia can include hypertension, diabetes, diet, physical activity, leisure
activity, occupation, and education (Gatz, Prescott, & Pedersen, 2006; Román, Nash, & Fillit,
2012). Typically, they are considered risk or delaying factors rather than a moderating factor
(Gatz, Prescott, et al., 2006). Are there lifestyle factors that can moderate the effect of APOE ε4
on memory? This is the main research question addressed in the present study.
The present study took two analytical approaches – confirmatory and exploratory. First,
G × E interactions on memory among older adults were examined. Specifically, the present
study investigated whether education, occupation, and physical activity can modify the effect of
APOE ε4 on memory. The relation between these lifestyle factors and cognitive decline and
dementia are reviewed in the following sections. Second, an exploratory technique called SEM
Trees (Brandmaier, et al., 2013) was used to examine potential G × E interactions in a context of
SEM.
Aging and Cognition in Late Life
GENES, LIFE STYLES, AND COGNITIVE DECLINE 6

Even though clinical dementia is rare before age 60, it is clear that cognitive abilities
decline with age. Research has documented various changes in the brain associated with normal
aging, such as inflammation of glia cells, thinning of myelin, reduction in dopamine receptors,
and neural atrophy (Finch & Zelinski, 2009). The basic neural mechanisms remain intact in older
age, but neuro-imaging studies have also found that, compared to young adults, older adults
often show less activity in medial temporal areas during memory encoding or retrieval (e.g.,
Cabeza et al., 2004). Moreover, older adults show greater activity in some brain areas, such as
both frontal and parietal regions during attention tasks (Townsend, Adamo, & Haist, 2006) and
the ventral or dorsal prefrontal cortex during memory tasks (Grady, Yu, & Alain, 2008). The
greater brain activity in older adults during cognitive tasks can be viewed as a compensation for
the brain atrophy brought on by aging, because high-performing older adults tend to show
greater activity than low-performing older adults (Cabeza, Anderson, Locantore, & McIntosh,
2002).
Perhaps because of such compensatory mechanisms, some aspects of cognitive abilities
appear well reserved in old age in the absence of neurological pathology. McArdle, Ferrer-Caja,
Hamagami, and Woodcock (2002) showed long-term retrieval significantly decline at age 36,
whereas comprehension-knowledge does so at age 71. Similarly, Salthouse (2005) argues that
speed, reasoning, and memory decline with age but vocabulary increases until about age 50, after
which remains relatively stable. The cognitive domain first affected by aging may be memory,
though one could argue that some aspects of memory remain intact into older age. In particular,
deficit in episodic memory is considered the hallmark symptom of dementia (but see Storandt,
2008). Episodic memory refers to a type of memory used when remembering particular item or
episode one’s life experience. Importantly, the decline in episodic memory can be observed in
GENES, LIFE STYLES, AND COGNITIVE DECLINE 7

non-demented older adults. Using word recall measures in the Health and Retirement Study (n >
17,000), McArdel, Fisher, and Kadlec (2007) found that word recall performance decreases
linearly with age. According to Finch and Zelinski (2009), it is currently unknown if these early
cognitive declines are precursor of Alzheimer’s disease. More research is needed to understand
the heterogeneity in cognitive aging.
Apolipoprotein E and Cognition
The apolipoprotein E (ApoE) is a protein that binds with lipids and carries cholesterol
and triglycerides through bloodstream. ApoE is abundantly expressed in the brain, mainly by
astrocytes and microglia. Cholesterol released from ApoE-containing lipoproteins is considered
crucial for synaptic formation, stabilization, and plasticity in the central nervous system (Pfrieger,
2003). Importantly, prior studies have delineated how ApoE might clear the brain’s amyloid-β
(Aβ) peptides, a component of senile plaques believed to cause dementia (see a review by Liu,
Kanekiyo, Xu, & Bu, 2013). First, ApoE-containing lipoproteins bind Aβ and facilitate Aβ
uptake (Kim, Basak, & Holtzman, 2009). Second, ApoE drains soluable Aβ into interstitial fluid
or through the blood-brain barrier (Hawkes et al., 2012). Third, ApoE facilitates Aβ proteolytic
degradation by enzymes (Jiang et al., 2008). In essence, ApoE plays a critical role in Aβ
metabolism.
The polymorphisms in the apolipoprotein E gene (APOE) have been shown to influence
the ApoE’s ability to bind to lipids and Aβ (Frieden & Garai, 2012; Zhong & Weisgraber, 2009).
One of three allelic variations of APOE, ε4, is associated with impairment in Aβ uptake,
drainage, and proteolytic degradation (Liu, et al., 2013). Consequently, ε4 allele carriers show
Aβ accumulation in the brain much earlier compared to non-carrier (Fleisher et al., 2013;
Morishima-Kawashima et al., 2000). For example, Fleisher et al. (2013) showed that brain
GENES, LIFE STYLES, AND COGNITIVE DECLINE 8

imaging could detect the presence of Aβ aggregates in ε4 carriers as early as 56 years of age,
compared with about 76 years of age in noncarriers. Moreover, APOE ε4 is the only robust
genetic predictor of dementia and late-onset AD. The risk for Alzheimer disease increases about
three-fold in people with one ε4 allele and about 12-fold in people with two ε4 alleles (Farrer et
al., 1997). APOE ε4 is also associated with poor episodic memory, global cognitive ability,
executive functioning, and perceptual speed among non-demented older adults (Wisdom,
Callahan, & Hawkins, 2011).
However, many ε4 carriers remain cognitively healthy, suggesting there are possible
moderating variables. A recent study reports that ε4 carriers with high education, leisure
activities, or vascular health showed dementia-free survival time similar to that of non-carriers
(Ferrari et al., 2013). The current study focused on three factors: (1) education, (2) occupation,
and (3) physical activity.
Education and Cognition
Education is a robust predictor of dementia among older adults. Literature reviews (Sharp
& Gatz, 2011; Valenzuela & Sachdev, 2006) and a meta-analysis (Meng & D'Arcy, 2012) have
found that higher education is associated with the lower risk of developing dementia. The
relationship between education and dementia is often explained by a theoretical construct,
cognitive reserve. According to Stern and his colleagues (Scarmeas & Stern, 2003; Stern, 2009;
Tucker & Stern, 2011), two neural mechanisms could underlie cognitive reserve: Reserve and
compensation. Neural reserve refers to efficiency, capacity, and flexibility in cognitive
processing in the normal brain. Neural compensation refers to the ability to utilize brain
structures not normally used in the healthy brain. Some people show substantial brain damage
GENES, LIFE STYLES, AND COGNITIVE DECLINE 9

without any clinical symptoms, and these people are said to have superior cognitive reserve.
Education is thought to enhance cognitive reserve.
Can education moderate the effect of APOE ε4? The results seem mixed. In an aging
study in Canada, both APOE ε4 and education were associated with dementia, but their
interaction effect was not significant (Hsiung, Sadovnick, & Feldman, 2004). In the MacArthur
Studies of Successful Aging, ε4 was associated with greater cognitive decline for people with
more education, whereas it did not affect people with less education (Seeman et al., 2005). More
specifically, the authors found that older adults with more education initially performed better
than those with less education on the cognitive tasks; ε4 carriers with more education showed
steeper declines in cognition, eliminating the effect of education. In the Netherlands, more
education was associated with cognitive decline among ε4 carriers, but only on one of six
cognitive measures used in the study, which led the authors to argue that education does not
protect against cognitive decline (Van Gerven, Van Boxtel, Ausems, Bekers, & Jolles, 2012). On
the other hand, low education was associated with cognitive decline only for APOE ε4
homozygotes in older adults in Washington (Shadlen et al., 2005) and increased dementia risk
among ε4 carriers in Sweden and Finland (Wang et al., 2012). The mixed findings in the pattern
of interaction between APOE ε4 and education on cognition may be due to other moderating
variables, such as age. These studies are an observational study in which the participants were
measured at the different initial ages. It has been known that time-based model do not produce
the same results with age-based model of change, even if the time-based model includes age as a
covariate (McArdle, et al., 2002). A study by McArdle and Prescott (2010) is the only one that
tested the APOE ε4-by-education interaction using the age-based models of cognitive decline.
They found that the ε4 allele was associated with greater age-related decline among less
GENES, LIFE STYLES, AND COGNITIVE DECLINE 10

educated people, whereas it was associated with less steeper decline and a lower starting point
among more educated people.
Nonetheless, these studies only examined the effect of education, ignoring life style
factors during adulthood. Studies have shown that education and occupation may independently
contribute to cognitive reserve (Stern et al., 1995; Stern et al., 1994).
Occupation and Cognition
There is evidence that occupation affects the risk of dementia. Like education,
occupations involving complex tasks may enhance cognitive reserve and delay the progress of
dementia. Prior studies have examined what aspects of occupation were associated with the risk
of dementia. A twin study in Sweden created scores for three dimensions of occupational
complexity, i.e., complexity of work with data, people, and things, and examined their effects on
the risk of dementia (Andel et al., 2005). The study found that higher occupational complexity
with people was associated with reduced risk of dementia, after adjusting for the effects of age,
sex, and education. Using the same scoring method for occupational complexity, a Canadian
study similarly found occupational complexity with people and things were associated with
lower risk of dementia (Kröger et al., 2008). According to the scoring method, occupations with
high complexity with people involve “supervising”, “instructing”, “mentoring,” and
“negotiating,” suggesting holding a leadership position may have a protective effect against
dementia. Indeed, a study of nuns in the School Sisters of Notre Dame in Germany found that
both education and holding a leadership position in the institution was associated with reduced
risk of dementia. This is intriguing because the nuns shared an identical environment (e.g.,
similar meals, daily routines, and working conditions). Furthermore, one neuroimaging study
recently found that supervisory experience in midlife was associated greater grey matter volume
GENES, LIFE STYLES, AND COGNITIVE DECLINE 11

in the hippocampus (Suo et al., 2012). In the same study, it was also found that the rate of neural
atrophy was five times slower in those with supervisory experience than those without. The
protective effect of supervisory experience remained significant after adjusting for the effects of
intracranial volume, age, gender, physical activity, depressive symptoms, and APOE ε4. Higher
occupation complexity, particularly being a supervisor, may be as important as education in
reducing the risk of dementia. To our knowledge, there is no study that examined the interaction
effect between job authority and APOE ε4 on memory. The present study was the first one to do
this.
Physical Activity and Cognition
Evidence suggests that physical activity can reduce the risk of dementia through a variety
of mechanisms (e.g., Erickson, Weinstein, & Lopez, 2012; Kempermann et al., 2010; Leckie,
Weinstein, Hodzic, & Erickson, 2012; Radak et al., 2010). First, physical activity increases the
levels of neurotrophins important for brain plasticity. For example, physical activity increases the
levels of brain-derived neurotrophic factor that supports the growth and survival of neurons
(Huang & Reichardt, 2001; Neeper, Gómez-Pinilla, Choi, & Cotman, 1996). Second, physical
activity protects the brain against oxidative stress. It is well-documented in the animal research
that physical activity increases the levels of antioxidant enzymes in the aging brain (e.g., Devi &
Kiran, 2004). Third, physical activity improves vascular health. Critically, physical activity
appears to improve cerebrovascular function by increasing the number of small-caliber vessels
throughout the brain (Bullitt et al., 2009) and cerebral blood volume to dentate gyrus (Pereira et
al., 2007). Fourth, it has been shown in mice that exercise promotes degradation of Aβ plaques,
suggesting that exercise may have a protective effect against AD (Adlard, Perreau, Pop, &
GENES, LIFE STYLES, AND COGNITIVE DECLINE 12

Cotman, 2005). In summary, physical activity may act on these mechanisms and delay the onset
of dementia in older adults.
Observational studies in older adults tend to find that physical activity is associated with
the risk of dementia (Andel et al., 2008; Churchill et al., 2002; Podewils et al., 2005). Recently,
Brown et al. (2012) utilized actigraphs to measure physical activity of older adults and found that
intensity of physical activity, rather than total activity levels, was associated with better cognitive
performance. Furthermore, randomized-control studies in humans also support the hypothesis
that regular exercise reduces the risk of dementia (Colcombe et al., 2004; Erickson et al., 2011;
Lautenschlager et al., 2008; Rosano et al., 2010). One recent study randomly assigned 120 older
adults to an aerobic exercise group or a stretching control group and examined the effect of
aerobic exercise on the brain and spatial memory (Erickson, et al., 2011). The participants in the
aerobic exercise group engaged in 40-minutes walking three days a week for a year, whereas
those on the control group engaged in stretching and muscle-toning exercises using dumbbells or
resistance bands. It was found that the aerobic exercise increased the size of the hippocampus
and improved spatial memory. Although more studies are needed to elucidate the mechanisms
underlying the relation between physical activity and cognition in humans, prior studies provide
some compelling evidence that physical activity improves memory in older adults.
There is evidence that physical activity may moderate the effect of APOE ε4 on dementia
(Etnier et al., 2007; Schuit, Feskens, Launer, & Kromhout, 2001). One observational study found
physical activity was associated with less amyloid plaques for ε4 carriers, but not for non-carriers
(Head et al., 2012). Other observational studies also found that physically active ε4 carriers
showed increased cortical activation during a memory task relative to sedentary ε4 carriers
(Deeny et al., 2008; J. C. Smith et al., 2011). An experimental study on animal support these
GENES, LIFE STYLES, AND COGNITIVE DECLINE 13

findings suggesting that physical activity is particularly beneficial for ε4 carriers (Nichol, Deeny,
Seif, Camaclang, & Cotman, 2009). Therefore, physical activity may be effective in reducing the
risk of dementia in APOE ε4 carriers.
Physical activity may be more important than education or occupation in protecting the
brain from the effect of APOE ε4. In experimental studies using mice, environmental enrichment
has been shown to induce morphological brain changes (Markham & Greenough, 2006) and
delay the dementia-related brain changes (Berardi, Braschi, Capsoni, Cattaneo, & Maffei, 2007;
Jankowsky et al., 2005); however, environmental enrichment in these studies consisted of many
components, such as toys, social stimulation, larger housing, and running wheels. When the
effect of running wheels was separately examined from other components of environmental
enrichment, the effects of environmental enrichment on the brain were mostly attributable to
running (Bechara & Kelly, 2013; Kobilo et al., 2011; van Praag, Kempermann, & Gage, 1999).
Physical activity may be the only lifestyle factor that can modify the effect of APOE ε4.
The Present Study
The purpose of the present study was to examine whether lifestyle factors, i.e., education,
job authority, and physical activity, could modify the effect of APOE ε4 on memory in late
midlife. Based on past studies, the following hypotheses were generated.
Hypothesis 1. Educational attainment, job authority, and physical activity would have a
protective impact on memory.
Hypothesis 2. Physical activity would moderate the effect of APOE ε4 on memory, but
education attainment and occupation would not. Specifically, ε4 carriers who were not
physically active would show greater impairment in memory than would non-carriers and
physical active ε4 carriers.
GENES, LIFE STYLES, AND COGNITIVE DECLINE 14

The present study took two approaches to study G × E interactions: Confirmatory
structural equation modeling (SEM) and exploratory analysis using tree models. The first
approach is SEM. Measurement error in variables of interest decreases the precision of the
parameters, leading to loss of statistical power. The SEM approach improves measurement by
separating the variance of constructs into one part due to the common factor and one part due to
the unique factors. Of course, statistical power can be increased by increasing sample size;
however, simulated studies have illustrated that increasing measurement accuracy is a more cost-
effective strategy than increasing sample size (Prindle & McArdle, 2013; Wong, Day, Luan,
Chan, & Wareham, 2003). Critically, SEM allows researchers to test their ideas by specifying
relations among observed variables and latent factors. Given that memory is measured with
multiple indicators, SEM may prove a powerful tool for studying G × E interactions.
The second analytical approach is much more exploratory, recursive partitioning.
Recursive partitioning is an exploratory algorithm that partitions the feature space spanned by all
predictors into smaller subsets. In doing so, this algorithm may identify interactions and
nonlinear relationships. Classification and Regression Trees (CART; Breiman, Friedman, Stone,
& Olshen, 1984; J. N. Morgan & Sonquist, 1963) is one method that utilizes recursive
partitioning and is becoming increasingly popular in behavioral sciences (McArdle, 2012).
Classification trees are used for a categorical variable, whereas regression trees are used for a
continuous variable. Using the algorithm, CART chooses one of covariates (i.e., predictors) and
split the data into two parts. The purpose of binary partitioning is to create two subgroups that
are more homogenous with respect to the outcome than the entire sample. Then it creates two
groups using each predictor in the data and chooses a split that produces the most homogenous
groups. It continues to partition the two subgroups in an attempt to further split them according
GENES, LIFE STYLES, AND COGNITIVE DECLINE 15

to all the predictors in the data. Criteria for stopping the algorithm depends on the type of CART
so they are not elaborated here (but see Berk, 2009; Hastie, Tibshirani, & Friedman, 2009; King
& Resick, 2014; Strobl, Malley, & Tutz, 2009). This property of recursive partitioning may be
particularly useful in studying gene-environment interactions, because it is difficult to specify a
priori which genetic variants interact with which aspects of environment (Thomas, 2010). To
deal with such complexity, the present study utilized recursive partitioning. More specifically the
study explored the data using structural equation model trees (SEM Trees), an innovative
method proposed by Brandmaier, von Oertzen, McArdle, and Lindenberger (2013) that apply
recursive partitioning on structural equation models.
Traditional recursive partitioning cannot incorporate prior hypothesis about the data and
deal with more than one dependent variable (e.g., longitudinal data). However, SEM Trees
allows formation of a SEM based on a hypothesis, and the data are partitioned into groups that
explain the largest differences in relation to parameters of the SEM. This algorithm allows
exploration of the data within a theoretical constraint. As Brandmaier et al. claimed, SEM Trees
may provide a useful integration of confirmatory and exploratory approaches. SEM Trees should
be a suitable method to explore complex and heterogeneous longitudinal data.
Study 1: Method
Data and Participants
Data were derived from the Wisconsin Longitudinal Study (WLS). The WLS consisted of
a random sample of 10,317 men and women who graduated from Wisconsin high schools in
1957. The WLS was originally designed to assess graduates’ aspirations for higher education,
but became a longitudinal study of life course. In its earlier waves, the WLS recorded the
graduates’ social background, schooling, work experiences, military service, family, and social
GENES, LIFE STYLES, AND COGNITIVE DECLINE 16

participation. More recent waves recorded various aspects of aging issues, such as family life,
retirement, finance, health, and end-of-life plans. The graduates were surveyed in 1957, 1975,
1992-1993, 2003-2005, and 2010-2011. At different time points, their family members were also
surveyed. Their parents were surveyed in 1957 and 1974. In addition, the WLS randomly
selected one sibling per graduates’ family and surveyed him or her in 1977, 1993-1994, 2004-
2007, and 2010-2011. The graduates’ spouses were surveyed in 2004, and the focal siblings’
spouses were surveyed in 2005. Survey methods consisted of telephone and mailed self-
administered survey.
All 10,317 graduates were considered for the present analyses. The present study
assumed that missing values were missing at random, as defined by Little and Rubin (Little &
Rubin, 1987), and thus full information maximum likelihood (FILM) was used to estimate the
parameters. FIML uses all the available data to optimize the model parameters without
imputation. About 52% of graduates were women. Their average years of age was 64.36 (SD =
0.72) in 2003 – 05 and 71.27 (SD = 0.96) in 2010-11. The basic demographics and
incompleteness are summarized in Table 1.
Measures
Genotypes. Saliva samples were obtained from graduates and siblings in 2010 – 11 and
genotyped for 90 SNPs (see Table 3). The SNPs were chosen based on prior studies showing
associations with an array of conditions, e.g., depression, cancer, diabetes, and cognitive
impairment. All SNP genotypes of the WLS graduates are in Hardy-Weinberg equilibrium their
frequencies were compatible to those reported in the literature for European samples. Oragene
DNA sample collection kits (DNA Genotek Inc, Kanata, Canada) were used to obtain salvia
samples. Genotyping was performed by KBioscience (Hoddesdon, UK). For confirmatory SEM
GENES, LIFE STYLES, AND COGNITIVE DECLINE 17

analysis, a candidate gene was the APOE gene (rs429358 and rs7412). Approximately 27% of
4569 graduates who provided saliva samples had at least one APOE ε4. For SEM Trees, all
SNPs were used.
Education. The years of education (M = 13.39, SD = 2.15) were used. The variable was
centered at 12 for the analyses.
Occupation. Job authority was measured in 1975 with four items: “Do you have authority
to hire and fire others?” “Can you influence pay received by others?” “Do you supervise the
work of others?” “Do you decide what others do and how they do it?” The graduates responded
to these questions in yes (= 1) or no (=0). Their responses were summed. The resulting job
authority score showed mean of 1.74 and standard deviation of 1.71.
Physical activity. Current solitary physical activity was assessed with a question, “During
the past year, how many hours per month did you do vigorous physical activities that you do
alone, such as jogging, swimming, biking, or going to the gym by yourself?” Past solitary
physical activity (10 years ago and 30 years ago) was also assessed on a 3-point scale (never,
rarely, or often). A similar set of questions were asked to assess group physical activity. For the
present study, a dichotomous variable was created where the graduates who never exercised were
scored -0.5 and the graduates who exercised were scored -0.5. In the present sample, 54% of
6262 graduates whose exercise information was available never exercised.
Memory. Memory was assessed with the following tasks: immediate recall, delayed
recall, and digit ordering. In the WLS, these measures are found to form one factor (Yonker,
Hauser, & Freese, 2007). Immediate and delayed recall are both part of the Telephone Interview
of Cognitive Status-Modified (Welsh, Breitner, & Magruder-Habib, 1993) and are used as a
measurement of episodic memory (McArdle, et al., 2007). For immediate recall, the graduates
GENES, LIFE STYLES, AND COGNITIVE DECLINE 18

were told to recall as many words as possible from a list of ten high-frequency words. The
graduates listened to ten words, one every two seconds. For delayed recall, the graduates were
asked to recall the same ten words again approximately 10 minutes after the immediate recall
task. It is important to note that they were not informed that they would have to recall the words
again. Between the immediate and delayed recall tasks, the WLS participants answered a series
of questions about health, volunteering activities etc.
For digit ordering, the graduates listened to a set of digits and were asked to repeat them
from smallest to largest. This was considered a modification of the Wechsler Adult Intelligence
Scale’s digit backward recall (Wechsler, 1997). The graduates started with a set of three digits.
Each correct response was followed by adding one digit. If they incorrectly responded to a three-
digit problem, they were given a score of two. This task ended when the graduates either
incorrectly responded twice in a row or correctly ordered a set of eight digits. Means, standard
deviations and pairwise correlations are shown in Table 2.
Covariates used in Confirmatory SEM. Some covariates were included in the
confirmatory analyses to adjust for their effects on the outcomes. Gender was ascertained in the
1957 survey of the graduates and was included in the present analyses, because previous studies
(e.g., McArdle, et al., 2007) have shown that gender difference is evident in later-life memory.
Furthermore, the graduates’ cognitive ability – measured with the Henmon–Nelson test (Lamke
& Nelson, 1957) during their high school years – was also included in the analyses. The test is a
group-administered multiple-choice assessment of verbal and quantitative ability and produces a
composite score that purportedly represents general intelligence (Lamke, 1958). However, it is
important to note that the test questions consisted of predominantly verbal items, therefore, by
today’s standards, the extent to which the test measures general intelligence is questionable.
GENES, LIFE STYLES, AND COGNITIVE DECLINE 19

Perhaps this is why the test is not widely used these days. Nonetheless, it was included as a
covariate as it is correlated with educational attainment, psychological well-being, and physical
health in the WLS data (Hauser & Palloni, 2011; Seltzer et al., 2005). Because of the present
study’s focus on job authority, it was important to adjust for the effects other job characteristics
on memory. Literacy use, assessed in 1975, was a self-reported number of hours in a week they
spent reading, writing, and dealing with written materials at work. Manual labor, also assessed in
1975, was a self-reported number of hours in a week they spent working with hands, tools, or
equipment at work. The variables’ summary statistics are presented in Table 1. Gender was half
effect coded, with men = -0.5, women = 0.5, whereas the continuous covariates were centered at
their respective means.
Statistical Analysis
McArdle and Prescott (2010) illustrated several approaches to studying the G × E effect,
including longitudinal SEM. Following their work, the present study conducted longitudinal
SEM analyses. A common factor model with a latent change score was used in the present study
(see Figure 1). The squares in the path diagram represent observed variables, which were
immediate recall, delayed recall, and digit ordering in the present study. Using these observed
variables, a latent factor was constructed within each occasion (F[1] in 2003-05 and F[2] in
2010-11). The factor loadings (λ) and unique variances (ψ) were held invariant across time. A
third latent factor, ΔF, represents a change between the two common factors, F[1] and F[2]. A
latent change score can be constructed by defining F[2] as the sum of the initial score and the
latent change score (i.e., F[2] = F[1] + ΔF). The effect of the initial score (F[1]) on the change
score (ΔF) was also estimated (β1). A triangle represents a constant of 1; the mean of the memory
factor score (F[1]) was fixed at 0 and β0 is the mean of the latent changes (ΔF). Though not
GENES, LIFE STYLES, AND COGNITIVE DECLINE 20

shown in the figure, variable intercepts were forced to be equal over time and specific
covariances were added for each memory measure.
To test the G × E hypotheses in the present study, the two memory factors (F[1] and ΔF)
were regressed on education, job authority, physical activity, APOE ε4 genotype, three
interaction terms (i.e., APOE ε4 by education, APOE ε4 by job authority, and APOE ε4 by
physical activity) and four covariates (i.e., gender, Henmon-Nelson scores, literacy use at work,
and manual labor). Specifically the following models were fitted:
Model 1: The effects of on education, job authority, physical activity, APOE ε4 genotype,
and three interaction terms on F[1] and ΔF were fixed to 0. But F[1] and ΔF were
regressed on the covariates (i.e., gender, Henmon-Nelson scores, literacy use at work, and
manual labor).
Model 2: The effects of APOE ε4 on F[1] and ΔF were added to Model 1 but the effect of
education, job authority, physical activity on F[1] and ΔF remained fixed at 0.
Model 3: The effects of education, job authority, physical activity, and APOE ε4 on F[1]
and ΔF were added to Model 3.
Model 4: The effects of the interaction terms, APOE ε4 × education and APOE ε4 × job
authority, on F[1] and ΔF were added to Model 3.
Model 5: The effects of the interaction term, APOE ε4 × physical activity, were added to
Model 4.
These nested models were fitted to the data and their fit indices were compared.
Hypothesis 1 is supported if Model 3 fits better than Model 2. Hypothesis 2 is supported if
Model 4 does not lead to a better fit but Model 5 does. The data were analyzed using OpenMx
GENES, LIFE STYLES, AND COGNITIVE DECLINE 21

(Boker et al., 2011) in R 2.15.2 (R Development Core Team, 2012) and Mplus 5.2 (Muthén &
Muthén, 1998-2008).
Study 1: Results
As a first step toward fitting the common factor model with a latent change score
(Figure1), the assumption of factorial invariance across time was examined. To do so, all the
parameters (i.e., factor means, factor variances, factor loadings, variable intercepts, and
residuals) were forced to be equal across 2003-05 and 2010-11 (χ
2
= 1520, df = 17, RMSEA =
0.124), and then some were relaxed with each subsequent model. Full information maximum
likelihood was used to estimate the parameters; consequently 5,792 graduates who had at least
one score of the three memory measurements across two occasions were included in the
parameter estimation. When the factor means were allowed to be different across time, it resulted
in significant improvement in fit (Δχ
2
/Δdf = 539/1, RMSEA = 0.100). Furthermore, allowing the
factor means and variances to vary across time led to improvement in fit (Δχ
2
/Δdf = 258/1,
RMSEA = 0.087). However, allowing the factor loadings to vary across time did not led to
improvement in fit (Δχ
2
/Δdf = 7/2, RMSEA = 0.093), and thus there was no evidence of different
factor structure across time. The results are summarized in Table 4.
Next, the common factor model with a latent change score (Figure1) was fitted to the data
(n = 5,792). The model included three memory measurements from two occasions (i.e., 2003-05
and 2010-11) without any covariates. The model’s fit was χ
2
= 185 (df = 12) and εa = 0.050
(90%CI = 0.044, 0.056). However, the model showed factor loadings of 1.000 (fixed), 1.020, and
0.318 for immediate recall, delayed recall, and digit ordering, respectively. The model showed a
latent change mean of -0.654 and a latent change variance of 1.301, suggesting that the memory
factor declined across time as expected. The effect of the initial score (F[1]) on the latent change
GENES, LIFE STYLES, AND COGNITIVE DECLINE 22

score (ΔF) was β1 = -0.654, meaning higher initial scores were associated with greater decline. In
addition to OpenMx in R, Mplus was used for this part and the results were replicated (see
Appendix A for Mplus code). However, it was clear that the factor loading of digit ordering was
lower than those of immediate and delayed recall, suggesting that digit ordering did not measure
the same construct as the word recall items. Therefore, it was decided that the present study
would focus only on the word recall measurements. The new model is shown in Figure 2.
Memory[1] is an average number of words recalled in immediate and delayed recall tasks in
2003-05 and ΔMemory is the change in the average number of words recalled across 2003-05
and 2010-11. The present study’s hypotheses remained the same, except Memory[1] replaced
F[1] and ΔMemory replaced ΔF. The results of this simple model indicated that the average
words recalled in 2003-05 was 5.125, and one recalled word in 2003-05 was associated with -
0.705 words of memory decline. Thus, on average, the present sample declined -0.627 words
across two measurement occasions.
Next, covariates were introduced to the model to test the present study’s hypotheses. The
results of goodness-of-fit comparison are summarized in Table 5. In Model 1, the memory
variables (i.e., Memory[1] and Δmemory) were regressed on gender, Henmon-Nelson score,
manual labors, and literacy use work. In Model 2, the effects of APOE ε4 on the memory
variables were estimated, and this improved the fit from Model 1 (ΔL = 30, Δdf = 2, p < 0.01).
Furthermore, Model 3, in which the effects of education, job authority, and physical activity on
the memory factors were estimated, resulted in improvement in fit from Model 2 (ΔL = 70, Δdf =
6, p < 0.01). However, adding interaction terms in Model 4 (ΔL = 3, Δdf = 4, p = 0.57) and
Model 5 (ΔL = 1, Δdf = 2, p = 0.55) did not improve the fit. Thus, among the five models, Model
3 (i.e., estimating the effects of APOE ε4, education, job authority, and physical activity on the
GENES, LIFE STYLES, AND COGNITIVE DECLINE 23

memory factors and excluding three interactions terms) appeared adequate and reasonable for the
data. The results of Model 3 showed that the increases in educational level had a positive impact
on the latent memory factor in 2003-05 and latent memory change factor. APOE ε4 did not
influence memory in 2003-05 but had a negative impact on the latent memory change factor.
Both job authority and physical activity showed a positive impact on memory in 2003-05, but not
on the latent memory change factor. Model 3’s summarized numeric results are shown in Figure
3. The R programming codes for Model 3 is shown in Appendix B.
Study 2: Method
Structural Equation Model Trees
SEM Trees uses model-based recursive partitioning (Brandmaier, von Oertzen, McArdle,
& Lindenberger, 2012). Traditional decision trees, such as CART (Breiman, et al., 1984),
recursively partition the data to maximize differences in the observed outcome, whereas SEM
Trees partitions the data to maximize difference in parameters of a SEM.
1. Build a SEM based on hypotheses about the data. This is a template model.
2. Fit the template model to the current data using a maximum-likelihood estimation
procedure and obtain likelihood statistic of misfit.
3. Choose one covariate and split the data into two with respect to the covariate. Fit the
template model to the two groups and obtain likelihood statistics of misfit; Repeat this for
all the covariates.
4. Choose the split that produced the best likelihood statistics. If this split produces
significantly better likelihood statistics than the template model, repeat the procedure
with Step 1 for all submodels. Otherwise terminate.
GENES, LIFE STYLES, AND COGNITIVE DECLINE 24

Due to multiple testing, we cannot use a traditional significance level, 0.05, in Step 4. In
SEM Trees, the Bonferroni correction or cross-validation (CV) method are available to correct
this. As per recommendations by Brandmaier et al. (2012), 5-fold CV were utilized.
Covariates used in SEM Trees. A large number of covariates collected during the
graduates’ adolescence, young adulthood, and old adulthood were included in the present
exploratory analysis. From the 1957 data, the following covariates were included in the SEM
Trees analysis: Population of town, teacher’s evaluation of the graduates (dichotomous),
graduates’ marriage status in 1957 (three categories; single, married or other), and parental
income (continuous; in hundreds).
From the 1975 data, the following covariates were included: Income (continuous; in
hundreds), marriage status, marriage status (three nominal categories; single, married or other),
number of months in military (continuous), job type (12 nominal categories; e.g., agriculture,
construction, professional, entertainment), frequency of working under time pressure (ordinal; 1
= “always”, 5 = “never”), whether they lived with their parents during childhood (dichotomous),
father’s education (continuous; in years), and mother’s education (continuous; in years).
From the 1992-93 data, the following covariates were included: income, total value of
saving (continuous), total value of investment (continuous), marriage status (three nominal
categories; single, married or other), employment status (dichotomous), number of job spells
from ’75 to ’92, numbers of hours spent dealing with written materials at work (continuous),
numbers of hours spent working with hands at work (continuous), frequency of working under
time pressure (ordinal; 1 = “always”, 5 = “never), frequency of exercise (1 = “three or more
times per week”, 4 = less than once per month), and six variables regarding chronic diseases
(dichotomous; i.e., hypertension, diabetes, cancer, heart problems, stroke, and arthritis).
GENES, LIFE STYLES, AND COGNITIVE DECLINE 25

From the 2003-05 data, the following covariates were included: income, marriage status
(three nominal categories; single, married or other), employment status (dichotomous), numbers
of hours spent dealing with written materials at work (continuous), numbers of hours spent
working with hands at work (continuous), retirement status (dichotomous), child death (three
categories; alive, death, or childless), sibling death (three categories; alive, death, or no sibling),
parental death (dichotomous; alive vs. dead), six variables regarding chronic diseases
(dichotomous; i.e., hypertension, diabetes, cancer, heart problems, stroke, and arthritis), body
mass index based on reported height and weight. Also maternal physical abuse (e.g., “Up-until
you were 18, to what extent did your mother treat you in a way that you would now consider
physical abuse”), paternal sexual abuse (e.g., “Up-until you were 18, to what extent did your
father treat you in a way that you would now consider sexual abuse?”), and sexual abuse
perpetrated by person other than fathers (e.g., “Up-until you were 18, to what extent did any
other person treat you in a way that you would now consider sexual abuse?”) were used. The R
programming codes for SEM tree is shown in Appendix C.
Study 2: Results
SEM Trees was run, using the simple regression model (Figure 2) as a base model.
However, the influence of Memory[1] on ΔMemory was fixed across groups in the tree.
Furthermore, it proved computationally too demanding to include so many continuous variables
in the analysis, therefore they were converted to quartiles (1 = Q1, 2 = Q2, 3 = Q3, and 4 = Q4); in
addition, people without genetic data were excluded, resulting in a sample size of 4569 (2220
men). The resulting tree is Figure 4. The first selected split was gender. For men, the next split
was having graduated from college. Among college-graduated men, the final split was the 3
rd

quartile of the hours in a week they spent working written materials at work in 1975. Among
GENES, LIFE STYLES, AND COGNITIVE DECLINE 26

men were not college educated, the next was a SNP of the CH25C gene (rs3802657). The next
split for those with G:G in the CH25C gene was openness to experience. The final split for men
with less than the median score of openness was rs2760118, a SNP in the SSADH gene.
The median of the Henmon-Nelson was the first split for women (Figure 5). Next, the
median of conscientiousness scores was the next split for women with lower-than-median
Henmon-Nelson scores. Women who scored higher than median in conscientiousness were split
by history of regularly smoking. For women with higher-than-median Henmon-Nelson scores,
the next split was produced by a SNP of the APOC3 gene (rs45537037). Women with C:C or
T:T in the APOC SNP was then split by a SNP of the DRD2 gene (rs2242592).
Discussion
The present study examined whether lifestyle factors, i.e., education, occupation, and
physical activity, could modify the effect of APOE ε4 on memory decline in late midlife. It was
hypothesized that these lifestyle factors have a positive impact on memory and memory decline.
Consistent with the first hypothesis, education, job authority, and physical activity influenced the
latent memory factors. Education showed a positive impact on both the latent memory factor at
2003-05 and the latent memory change factor. However, job authority and physical activity
showed a positive impact only on the latent memory factor at 2003-05. It was also hypothesized
that physical activity would moderate the effect of APOE ε4 on memory and memory decline,
whereas education attainment or job authority would not. Contrary to the hypothesis, none of
three lifestyle factors moderated the effects of APOE ε4 on memory. The finding is similar to
those from prior observational studies where exercise was found to be beneficial but did not
moderate the effects of APOE ε4 on the risk of dementia (Lindsay et al., 2002; Sabia, Kivimaki,
Kumari, Shipley, & Singh-Manoux, 2010; Taaffe et al., 2008). Many researchers delineated the
GENES, LIFE STYLES, AND COGNITIVE DECLINE 27

benefits of physical activity on the brain (Erickson, et al., 2011; Kempermann, et al., 2010;
Leckie, et al., 2012; Radak, et al., 2010), but it might not directly counteract the damages caused
by APOE ε4. Kemperman et al. (2010) argues that physical activity prepares the brain for
learning by signaling proliferation of neural precursor cells. Therefore physical activity may not,
by itself, help neuronal survival.
The present study’s methods might also explain the null finding. The null finding might
have been caused by the insufficient number of repeated measurements. McArdle and Prescott
(2010) utilized seven waves of immediate and delayed recall in the Health and Retirement Study
and found a significant G × E interaction; The present study unfortunately consisted of only two
waves. Finally, the effects of lifestyles on psychological or health outcomes are difficult to study
(Taubes, 1995); lifestyle factors, such as occupational characteristics, leisure activities, and diet,
are correlated among each other and may change over time, even though the covariates were
treated as time-invariant in the present study. It might have been more appropriate to examine
the present study’s research question with a randomized-control study. At any rate, the present
study did not provide evidence that the effect of APOE on memory could be modified by
education, job authority, or physical activity.
In the exploratory stage of the present study, an innovative method called SEM trees was
applied. Gender produced the first split, women outperforming men. This is consistent with
gender being the robust predictor of memory measure. Recent studies suggest that, in terms of
cognitive function among older adults, women tend to outperform men in developed country
(e.g., Langa et al., 2008) whereas men tend to outperform women in developing countries (e.g.,
Lee, Shih, Feeney, & Langa, 2014). The second split was produced by education for men. As
stated previously, low educational attainment is one of the better predictors of poor cognitive
GENES, LIFE STYLES, AND COGNITIVE DECLINE 28

functioning in later life (Valenzuela & Sachdev, 2006). Education is hypothesized to contribute
to cognitive reserve because educated people are likely to engage in stimulating activities, which
moderate age-related cognitive declines by building more efficient brain networks as well as
alternative brain networks. On the other hand, the Henmon-Nelson scores produced a second
split for women. Perhaps this is because the Henmon-Nelson scores – predominantly reflecting
reading ability – might be a better indicator of cognitive reserve than education for these women
educated in Wisconsin during 1950’s. Consistent with this idea, the present study found that
literacy use at work produced a split for college educated men, and a previous study also found
that literacy is a better indicator of cognitive reserve than education attainment for older adults of
ethnic minorities (Manly, Byrd, Touradji, Sanchez, & Stern, 2004). Manly et al. speculate this
might be because societal forces may have prevented some people from attending higher
education institution. In the WLS data, the Henmon-Nelson score are correlated with gender,
educational attainment, parental income, and other socioeconomic variables, and thus it is
difficult to tease apart cognition from genetic or environmental factors. However, it is clear that
memory in late life is affected by socioeconomic status as well as ability.
A SNP of the CH25H gene produced a split in men without college degree. The gene
produces an enzyme called cholesterol 25-hydroxylase involved in cholesterol and lipid
metabolism (Lund, Kerr, Sakai, Li, & Russell, 1998). Therefore some speculated that the gene
may play a role in Alzheimer’s disease (e.g., Papassotiropoulos et al., 2005), even though the
evidence is mixed (e.g., A. R. Morgan et al., 2007). Nonetheless, the present study’s outcome
variable was not Alzheimer’s disease but episodic memory. It is possible that, because CH25H
plays a role in cholesterol/lipid metabolism, it affects the brain health in general, not just
Alzheimer’s disease. Similarly, the present study found that a SNP of the APOC3 produced a
GENES, LIFE STYLES, AND COGNITIVE DECLINE 29

split in women. APOC3 is a component of triglyceride-rich lipoproteins and variants of APOC3
have been associated with lipid metabolism and heart disease (Muendlein et al., 2008; Pollex et
al., 2007). Moreover, one study found that variants of APOC3 were associated with cognitive
status (C. E. Smith et al., 2009). Even though it is not clear why CH25H and APOC3 showed up
in the tree while APOE4 did not, evidence suggests that genes that play a role in lipid
metabolism influence cognition in late life.
One of more interesting findings in the present study was a role of personality in SEM
Trees. The WLS measured the five dimensions of personality known as the “Big Five” (i.e.,
conscientiousness, extraversion, agreeableness, neuroticism, and openness to experience).
Conscientiousness – characterized as orderly, self-disciplined, and goal-orientated – produced a
split in women and openness to experience – characterized as intellectually curious – produced a
split in men. Researchers of personality and health have argued that personality influences health
outcomes because personality influences health-related behaviors (e.g., Adler & Matthews,
1994). In particular, conscientiousness is thought to influence health across the life-span
(Shanahan, Hill, Roberts, Eccles, & Friedman, 2012). A meta-analysis found that
conscientiousness-related characteristics were negatively correlated with risk behaviors (e.g.,
drug use, risky sexual behaviors, unhealthy diet) and positively correlated with healthy behaviors
(e.g., exercise; Boggs & Roberts, 2004); therefore it is no surprise that conscientiousness people
tend to live longer (Kern & Friedman, 2008). However, prior studies examining relations
between conscientiousness and cognitive functioning did not show consistent results (e.g.,
Furnham & Chamorro-Premuzic, 2006; Moutafi, Furnham, & Crump, 2002; Soubelet &
Salthouse, 2011; Wolf & Ackerman, 2005). One reason for such inconsistency might be due to
the diversity of cognitive measurements used in these studies. But research on openness and
GENES, LIFE STYLES, AND COGNITIVE DECLINE 30

cognition shows more consist findings. For example, one review found that openness to
experience was correlated with general intelligence (r = .33) and crystallized intelligence (r
= .30) (Ackerman & Heggestad, 1997). Neuroscience further supports a notion that both
personality influences the brain. A recent neuroimaging study showed that lower
conscientiousness was associated with greater age-related decreases in cerebral white matter,
amygdala, and parahippocampal volumes (Jackson, Balota, & Head, 2011). Similarly,
openness was associated with better integrity of white matter in the corona radiata and superior
longitudinal fasciculus tracts that interconnect prefrontal cortex in healthy adults aged 18 to 54
years (Xu & Potenza, 2012). The processes by which personality affect the health is not well
understood, but some researchers, such as Israel et al. (2014), believe that personality is a robust
correlate of lifestyles and health behaviors and thus personality measurements may be useful in
personalizing preventive medicine. More studies are needed to delineate how and why
personality influences cognitive function in late life.
Two more SNPs produced splits in the tree: rs2760118 (SSDAH) in men and rs2242592
(DRD2) in women. SSDAH, or succinic semialdehyde dehydrogenase, is an enzyme that is
involved in the gamma-amino butyric acid (GABA) catabolism. GABA’s primary role is to
regulate the brain’s excitability. The SSDAH deficiency leads to the accumulation of GABA,
which is toxic to the brain (Gibson, 2005). Importantly, the SSDAH encoded by the less frequent
T allele of this SNP was found to be 17.5% less active than one encoded by the more frequent C
allele (Blasi et al., 2002) and was associated with lower cognitive performance (Plomin et al.,
2004). Perhaps more relevant to the present study, another study also found that the T allele of
the SNP was associated with lower cognitive functioning (assessed by the Mini Mental State
Examination) and survival in the elderly population (Rango et al., 2008). Moreover, Chabris et al.
GENES, LIFE STYLES, AND COGNITIVE DECLINE 31

(2012) examined associations between g and 12 specific genetic variants, including SSADH and
APOE, in three independent samples and found that only SSADH was significantly associated
with g, though the effect disappeared after Bonferroni correction. They concluded the effect of
the SSADH SNP may be observed only under certain environmental conditions. Consistent with
such idea, the present study found that the SNP was important for men who were 1) not college
educated, 3) G:G genotype in rs3802657, and 3) lower than median in openness to experience.
However, in the present sample, the more frequent allele was associated with lower word recall.
This makes it unclear as to the role of the SSADH in cognitive aging. Looking at the tree, the
gene seems to be affecting those from lower SES. Future research first needs to examine if the
finding will be replicated in the different sample.
A SNP in the DRD2 gene, which codes for dopamine D2 receptor (DRD2), produced a
split in women. Dopamine is a monoamine neurotransmitter essential for memory, learning, and
motivation (Wise, 2004), as it promotes protein synthesis in hippocampal neurons (W. B. Smith,
Starck, Roberts, & Schuman, 2005). Prior studies have shown that the polymorphisms in the
DRD2 gene influence the density of DRD2 (Pohjalainen et al., 1998) and are associated with
cognitive functioning (Markett, Montag, Diekmann, & Reuter, 2014; Markett, Montag, & Reuter,
2010; Markett, Montag, Walter, & Reuter, 2011; Rodriguez-Jimenez et al., 2006). Importantly,
recent evidence indicates that aging increases the effects of the DRD2 polymorphism on episodic
memory (Chowdhury, Guitart-Masip, Bunzeck, Dolan, & Duzel, 2012; Li et al., 2013;
Papenberg et al., 2013). Papenberg et al., for example, found that older adults (59 – 71 years of
age) with two or three genotypes associated with higher availability of dopamine receptors
remembered more pictorial information than older adults with one or no beneficial genotypes.
However, the researchers did not find such effects in younger adults (20 – 31 years). Papenberg
GENES, LIFE STYLES, AND COGNITIVE DECLINE 32

et al. argued that the genetic effects on cognition would be larger in older people because of their
depleting neuroanatomical or neurochemical resources. Prior studies have demonstrated that the
availability of dopamine in the brain decreases with age. Therefore, it could be argued that the
negative genetic effects become salient in older adults. For the SNP in the DRD2 gene that
produced the split in the present study, a C allele is a minor allele. The SEM Tree’s split was
based on whether or not women had the C allele (i.e., T:T vs. C:T or C:C). Comparing the
parameter estimates of two groups produced by this split, those with a C allele might be
considered at greater risk for memory impairment, though the difference is subtle. Relatively
large differences can be observed in the variance, but not in the mean or change.
Finally, women who have ever smoked cigarette regularly showed a lower mean and
greater decline of word recall compared to women who have never smoked regularly. Prior
studies have shown that cigarette smoking accelerates aging (see review by Bernhard, Moser,
Backovic, & Wick, 2007). The effect of smoking on dementia and cognitive decline is also well
documented (Anstey, von Sanden, Salim, & O'Kearney, 2007; Peters et al., 2008). Cigarette
smoke consists of more than 7000 chemicals such as hydrogen cyanide, lead, ammonia (Centers
for Disease and Control and Prevention, 2010). These chemicals can induce oxidative stress in
multiple organs in the body (Rueff-Barroso et al., 2010). A recent animal study experimentally
showed that cigarette smoking increased oxidative stress and reduced of the expression of pre-
synaptic proteins in the hippocampus (Ho et al., 2012). Moreover, the study showed that
smoking increased sAPPb and the accumulation of amyloid-beta peptide in the CA3 and dentate
gyrus region. The present study confirmed that, among many lifestyle factors, cigarette smoking
is harmful for memory in late life.
Limitations
GENES, LIFE STYLES, AND COGNITIVE DECLINE 33

The study had severe limitations. The WLS participants are not representative of either
the Wisconsin population or the North American population. First, everyone in the WLS
completed high school, even though approximately 25% of young people in Wisconsin did not
graduate from high schools in the late 1950s (Sewell & Hauser, 1975). This is unfortunate given
education was a key element in this study’s hypothesis. Second, the sample did not include
people of racial/ethnic minorities, reflecting the Wisconsin demography during 1950s. This
limits the generalizability of the study’s results. Therefore, the study needs to be replicated using
data representative of the North American population, such as Health and Retirement Study
(Juster & Suzman, 1995). As a clarifying note, I attempted to gain access to the HRS genetic data
but was rejected because I did not have a Ph.D. at the time, even though I indicated in my
application that my advisor, Jack McArdle (co-investigator of the HRS), would oversee my study.
In my view, the qualifications for obtaining access to the genetic are stricter than they need to be.
Researchers who work in a credible research institution and have means to protect sensitive data
should be eligible for accessing the data regardless of their educational attainment. Also, in the
present study, SEM Trees did not split when the number of people in resulting groups was less
than 200. Therefore, the present study might have missed an effect of infrequent genotypes. One
good strategy might have been to explore the data by letting the tree grow freely.
Another limitation is the missing data problem. Given that the WLS started more than 50
years ago, it was expected that attrition would be a problem in the present study. Hauser (2005)
found that lower Henmon-Nelson score and academic performance in high school were
associated with attrition in the WLS. Therefore, in the present study, the Henmon-Nelson score
was included in the confirmatory SEM, and FILM estimator was used to adjust, though not
perfectly, for the effects of incompleteness. McArdle (2013) discussed various means of dealing
GENES, LIFE STYLES, AND COGNITIVE DECLINE 34

with longitudinal attrition and proposed creating sampling weights based on decision tree model.
The advantage of using decision tree model to examine attrition is its exploratory approach in
detecting nonlinearities and interactions. Unfortunately, sampling weights could not be applied
to the present analyses because the R packages used – OpenMx and semtree – did not directly
support them.
Moreover, the current study only tested one cognitive domain due to a limited scope of
cognitive measurements available in the WLS data. In addition to the measurements used in the
present study, the WLS includes longitudinal data of the Wechsler Adult Intelligence Scale’s
similarities task. For this task, participants were asked how two things are similar (e.g., “how are
oranges and bananas alike?”). The task is thought to measure abstract verbal reasoning ability
(but see Horn & McArdle, 1992). Also, the WLS contain longitudinal data participants’
vocabulary by asking participant to name words starting with either the letter “L” or the letter “F”
in one minute, as many of them as possible. Though these cognitive domains are interesting, the
present study focused on memory because it was expected to decline at greater rate than verbal
ability assessed by these measurements. As stated earlier, research on cognitive aging
demonstrated that different cognitive abilities exhibit different developmental trajectories.
McArdle et al. (2002) used the Woodcock-Johnson Psycho-Educational Battery – a
comprehensive test of cognitive abilities – to study to age-related change. Consequently, unlike
the present study, they were able to analyze developmental properties of several cognitive
domains. Similarly, McArdle, Fisher and Kadlec (2007) used data from the Health and
Retirement Study, which included immediate word recall, delayed word recall, Serial 7s,
backward counting, object naming, dates, vocabulary, and similarities – thus more
comprehensive than the WLS. The WLS and the Health and Retirement Study have some
GENES, LIFE STYLES, AND COGNITIVE DECLINE 35

cognitive measurements in common because the WLS attempted to increase comparability with
the Health and Retirement Study. Though the WLS was not originally designed as a study of
cognition or aging, the WLS researcher added a new cognitive assessment task call number
series in the most recent wave. The scope of cognitive domains is far from complete but the
WLS continues to evolve and may offer many future opportunities for research on cognitive
aging.
More measurement issues need to be considered. At the time of proposing this study, the
WLS had only genotyped 90 SNPs of the saliva samples, and therefore many genetic variations
were unmeasured. Recently, Pamela Herd, the principal investigator of the WLS, received NIH
funding to genotype more SNPs. This will further increase the compatibility with the Health and
Retirement Study and provide more opportunities for research on gene-environment interaction
and cognitive aging.
Moreover, physical activity was measured with self-report questionnaires. Recall bias and
social desirability bias could have affected how they responded to the questions. Physical activity
in the present analyses was a binary variable where two groups were 1) never exercised and 2)
exercised in the last 30 years. Researchers recommend, however, more rigorous and frequent
exercise for older adults. For example, the American Heart Association and the American
College of Sports Medicine recommend older adults to engage in moderate aerobic physical
activity for at least 30 min on five days per week or intense aerobic physical activity for at least
20 min on three days per week (Nelson et al., 2007). In the WLS, recent physical activity was
assessed by asking the participants how many hours they engaged in physical activity in the last
month. Asking older adults to quantify the extent of physical activity in this manner may be too
demanding, particularly for those with memory problems. Therefore, one recommendation for
GENES, LIFE STYLES, AND COGNITIVE DECLINE 36

the WLS is to include physical activity measurement that is easier for them to self-evaluate and
respond to. For example, a study by Ferencz et al. (2014) asked how frequently participants
exercised in 6-point Likert scale where choices were every day, several times a week, two-to-
three times per month, less than two-to-three times per month, and never. This allowed Ferencz
et al. to create a 3-level physical activity variable, informed by the American Heart Association’s
recommendations: Inadequate (two-to-three times per month in light and/or moderate/intense
exercise; health enhancing (light exercise several times per week or every day); and fitness-
enhancing (moderate/intense exercise several times per week or every day). Fifty-four percent of
the WLS participants claimed to have never exercise in the last 30 years, but what makes
difference in late-life cognition may be recent physical activity as demonstrated by randomized-
control studies of physical activity and cognition (Heyn, Abreu, & Ottenbacher, 2004; Hindin &
Zelinski, 2012). The present study’s results urge a better assessment of physical activity for
future research using the WLS data.
Despite these limitations, the present study had some strengths as well. The present study
took two approaches: confirmatory and exploratory. McArdle (2011) argued that, instead of
searching for the “best” model, researchers need to come into an analysis with ideas about their
data and confirm their ideas. Only after the confirmatory analyses, should researchers explore the
data. This sequence is important for what Raymond Cattell called inductive-hypothetico-
deductive spiral, an on-going process of theory development, testing, exploration, and theory
revision (McArdle, 2011, as cited in p. 332). By using both confirmatory and exploratory
analyses, the present study’s results set the stage for many potential analyses that might be
conducted using other longitudinal data of cognitive aging. Following McArdle’s argument, the
present study then proceeded to the SEM Trees phase where the data set was explored. The
GENES, LIFE STYLES, AND COGNITIVE DECLINE 37

present study was the first study to apply SEM Trees and demonstrated its utility using the well-
known data set. As all the analyses were conducted in R, researchers with the appropriate
knowledge of SEM and R programming language can easily attempt to replicate these results as
well as explore their own data. There are many longitudinal cognitive data available, leading to
future research opportunities to deepen the understanding of mechanisms underlying cognitive
aging.
In summary, the present study suggests an important role of lifestyle factors in memory
in older adults. Even though the effects of APOE ε4 on memory were not modified by
educational attainment, job authority or physical activity, these lifestyle factors showed positive
impacts on memory. In particular, the effect of physical activity is as strong as the negative effect
of APOE ε4 on memory. Therefore, the results suggest that physical activity may be an
important element of an inexpensive lifestyle intervention method to improve cognitive
functioning in older adults. However, changing one’s lifestyle is difficult. In 2011, a group of
prominent social scientists discussed what the most important questions that social scientists
should tackle were. They concluded that the most important question was “How can we induce
people to look after their health?” (Giles, 2011). Educating the public about healthy lifestyle
through research is one thing, but inducing people to embrace healthier lifestyles is another. It is
a big challenge for researchers.

GENES, LIFE STYLES, AND COGNITIVE DECLINE 38

References
Ackerman, P. L., & Heggestad, E. D. (1997). Intelligence, personality, and interests: Evidence
for overlapping traits. . Psychological Bulletin, 121, 219-245.
Adlard, P. A., Perreau, V. M., Pop, V., & Cotman, C. W. (2005). Voluntary exercise decreases
amyloid load in a transgenic model of Alzheimer's disease. The Journal of Neuroscience,
25, 4217-4221.
Adler, N., & Matthews, K. (1994). Health psychology: Why do some people get sick and some
stay well? Annual Review of Psychology,, 45, 229–259.
Alzheimer’s Association. (2012). 2012 Alzheimer’s Disease Facts and Figures. Alzheimer’s &
Dementia, 8(2). Retrieved from http://www.alz.org/downloads/facts_figures_2012.pdf
Andel, R., Crowe, M., Pedersen, N. L., Fratiglioni, L., Johansson, B., & Gatz, M. (2008).
Physical exercise at midlife and risk of dementia three decades later: A population-based
study of Swedish twins. Journal of Gerontology: Medican Sciences, 63A, 62-66.
Andel, R., Crowe, M., Pedersen, N. L., Mortimer, J. A., Crimmins, E., Johansson, B. (2005).
Complexity of work and risk of Alzheimer's disease: A population-based study of
Swedish twins. Journal of Gerontology: Psychological Sciences, 60B(251-258).
Anstey, K. J., von Sanden, C., Salim, A., & O'Kearney, R. (2007). Smoking as a risk factor for
dementia and cognitive decline: a meta-analysis of prospective studies. American Journal
of Epidemiology, 166, 367-378.
Bechara, R. G., & Kelly, A. M. (2013). Exercise improves object recognition memory and
induces BDNF expression and cell proliferation in cognitively enriched rats. Behavioural
Brain Research, 245, 96-100.
GENES, LIFE STYLES, AND COGNITIVE DECLINE 39

Berardi, N., Braschi, C., Capsoni, S., Cattaneo, A., & Maffei, L. (2007). Environmental
enrichment delays the onset of memory deficits and and reduces neuropathological
hallmarks in a mouse model of Alzheimer-like neurodegeneration. Journal of Alzheimer's
disease, 11, 359-370.
Berk, R. A. (2009). Statistical learning from a regression perspective. New York, USA: Springer.
Bernhard, D., Moser, C., Backovic, A., & Wick, G. (2007). Cigarette smoke -- an aging
accelerator? Experimental Gerontology, 42, 160-165.
Blasi, P., Boyl, P. P., Ledda, M., Novelletto, A., Gibson, K. M., Jakobs, C. (2002). Structure of
human succinic semialdehyde dehydrogenase gene: identification of promoter region and
alternatively processed isoforms. Molecular Genetics and Metabolism, 76, 348–362.
Boggs, T., & Roberts, B. W. (2004). Conscientiousness and health-related behaviors: A meta-
analysis of the leading behavioral contributors to mortality. Psychological Bulletin,
130(6), 887-919.
Boker, S. M., Neale, M. C., Maes, H. H., Wilde, M. J., Spiegel, M., Brick, T. R. (2011).
OpenMx: An open source extended structural equation modeling framework.
Psychometrika, 76, 306-317.
Brandmaier, A. M., von Oertzen, T., McArdle, J. J., & Lindenberger, U. (2012). Structural
equation model trees. Psychological Methods, Advance online publication. doi:
10.1037/a0030001
Brandmaier, A. M., von Oertzen, T., McArdle, J. J., & Lindenberger, U. (2013). Structural
equation model trees. Psychological Methods, 71-86. doi: 10.1037/a0030001
Breiman, L., Friedman, J., Stone, C. J., & Olshen, R. A. (1984). Classification and regression
tree. Monterey, California: Wadsworth & Brooks.
GENES, LIFE STYLES, AND COGNITIVE DECLINE 40

Brown, B. M., Peiffer, J. J., Sohrabi, H. R., Mondal, A., Gupta, V. B., Rainey-Smith, S. R.
(2012). Intense physical activity is associated with cognitive performance in the elderly.
Translational Psychiatry, 2, e191. doi: 10.1038/tp.2012.118
Bullitt, E., Rahman, F. N., Smith, J. K., Kim, E., Zeng, D., Katz, L. M. (2009). The effect of
exercise on the cerebral vasculature of healthy aged subjects as visualized by MR
angiography. American Journal of Neuroradiology, 30, 1857–1863.
Cabeza, R., Anderson, N. D., Locantore, J. K., & McIntosh, A. R. (2002). Aging gracefully:
compensatory brain activity in high-performing older adults. NeuroImage, 17, 1394–1402.
Cabeza, R., Daselaar, S. M., Dolcos, F., Prince, S. E., Budde, M., & Nyberg, L. (2004). Task-
independent and task-specific age effects on brain activity during working memory,
visual attention and episodic retrieval. Cerebral Cortex, 14, 364–375.
Centers for Disease and Control and Prevention. (2010). Chemicals in Tobacco Smoke, from
http://www.cdc.gov/tobacco/data_statistics/sgr/2010/consumer_booklet/chemicals_smok
e/
Chabris, C. F., Hebert, B. M., Benjamin, D. J., Beauchamp, J., Cesarini, D., Loos, M. v. d.
(2012). Most reported genetic associations with General intelligence are probably false
positives. Psychological Science, 23, 1314-1323.
Chowdhury, R., Guitart-Masip, M., Bunzeck, N., Dolan, R. J., & Duzel, E. (2012). Dopamine
modulates episodic memory persistence in old age. The Journal of Neuroscience, 32,
14193-14204.
Churchill, J. D., Galvez, R., Colcombe, S., Swain, R. A., Kramer, A. F., & Greenough, W. T.
(2002). Exercise, experience and the aging brain. Neurobiology of Aging, 23, 941-955.
GENES, LIFE STYLES, AND COGNITIVE DECLINE 41

Colcombe, S. J., Kramer, A. F., Erickson, K. I., Scalf, P., McAuley, E., Cohen, N. J. (2004).
Cardiovascular fitness, cortical plasticity, and aging. Proceedings of the National
Academy of Sciences, 101, 3316-3321.
Deeny, S. P., Poeppel, D., Zimmerman, J. B., Roth, S. M., Brandauer, J., Witkowski, S. (2008).
Exercise, APOE, and working memory: MEG and behavioral evidence for benefit of
exercise in epsilon4 carriers. Biological Psychology, 78, 179-187.
Devi, S. A., & Kiran, T. R. (2004). Regional responses in antioxidant system to exercise training
and dietary vitamin E in aging rat brain. Neurobiology of Aging, 25, 501-508.
Erickson, K. I., Voss, M. W., Prakash, R. S., Basak, C., Szabo, A., Chaddock, L. (2011).
Exercise training increases size of hippocampus and improves memory. Proceedings of
the National Academy of Sciences, 108, 3017–3022.
Erickson, K. I., Weinstein, A. M., & Lopez, O. L. (2012). Physical activity, brain plasticity, and
Alzheimer's disease. Archives of Medical Research, 43, 615-621.
Etnier, J. L., Caselli, R. J., Reiman, E. M., Alexander, G. E., Sibley, B. A., Tessier, D. (2007).
Cognitive performance in older women relative to ApoE-ε4 genotype and Aerobit fitness.
Medicine and Science in Sports and Exercise, 39(1), 199-207.
Farrer, L. A., Cupples, L. A., Haines, J. L., Hyman, B., Kukull, W. A., Mayeux, R. (1997).
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. Journal of the American Medical Association, 278, 1349-1356.
Federal Interagency Forum on Aging-Related Statistics. (2012). Older Americans 2012: Key
Indicators of Well-Being. Washington, DC: Government Printing Office Retrieved from
GENES, LIFE STYLES, AND COGNITIVE DECLINE 42

http://www.agingstats.gov/agingstatsdotnet/Main_Site/Data/2012_Documents/Docs/Entir
eChartbook.pdf.
Ferencz, B., Laukka, E. J., Welmer, A.-K., Kalpouzos, G., Angleman, S., Keller, L. (2014). The
benefits of staying active in old age: Physical activity counteracts the negative influence
of PICALM, BIN1, and CLU risk alleles on episodic memory functioning. Psychology
and Aging, Online First Publication. doi: 10.1037/a0035465
Ferrari, C., Xu, W.-L., Wang, H.-X., Winblad, B., Sorbi, S., Qiu, C. (2013). How can elderly
apolipoprotein E ε4 carriers remain free from dementia? Neurobiology of Aging, 34, 13-
21.
Finch, C. E., & Zelinski, E. M. (2009). Normal aging of brain structure and cognition:
Evolutionary perspectives. Research in Human Development, 2, 69-82.
Fleisher, A. S., Chen, K., Liu, X., Ayutyanont, N., Roontiva, A., Thiyyagura, P. (2013).
Apolipoprotein E ε4 and age effects on florbetapir positron emission tomography in
healthy aging and Alzheimer disease. Neurobiology of Aging, 34, 1-12.
Frieden, C., & Garai, K. (2012). Structural differences between apoE3 and apoE4 may be useful
in developing therapeutic agents for Alzheimer’s disease. Proceedings of the National
Academy of Sciences, 109, 8913–8918.
Furnham, A., & Chamorro-Premuzic, T. (2006). Personality, intelligence, and general knowledge.
Learning and Individual Differences, 16, 79-90.
Gatz, M., Prescott, C. A., & Pedersen, N. L. (2006). Lifestyle risk and delaying factors.
Alzheimer Disease and Associated Disorders, 20, S84 - S88.
GENES, LIFE STYLES, AND COGNITIVE DECLINE 43

Gatz, M., Reynolds, C. A., Fratiglioni, L., Johansson, B., Mortimer, J. A., Berg, S. (2006). Role
of genes and environments for explaining Alzheimer Disease. Archives of General
Psychiatry, 63, 168-174.
Gibson, K. M. (2005). Gamma-hydroxybutyric aciduria: A biochemist’s education from a
heritable disorder of GABA metabolism. Journal of Inherited Metabolic Disease, 28,
247-265.
Giles, J. (2011). Social science lines up its biggest challenges. Nature, 470(18-19). doi:
doi:10.1038/470018a
Grady, C. L., Yu, H., & Alain, C. (2008). Age-related differences in brain activity underlying
working memory for spatial and nonspatial auditory information. Cerebral Cortex, 18,
189–199.
Hastie, T., Tibshirani, R., & Friedman, J. (2009). The Elements of Statistical Learning: Data
mining, inference, and prediction Retrieved from
http://statweb.stanford.edu/~tibs/ElemStatLearn/
Hauser, R. M. (2005). Survey response in the long run: the Wisconsin Longitudinal Study. Field
Methods, 17, 3–29.
Hauser, R. M., & Palloni, A. (2011). Adolescent IQ and survival in the Wisconsin Longitudinal
Study. The Journals of Gerontology, Series B: Psychological Sciences and Social
Sciences, 66B(S1), i91–i101.
Hawkes, C. A., Sullivan, P. M., Hands, S., Weller, R. O., Nicoll, J. A. R., & Carare, R. O. (2012).
Disruption of arterial perivascular drainage of amyloid-β from the brains of mice
expressing the human APOE ε4 allele. PLoS One, 7(7), e41636.
GENES, LIFE STYLES, AND COGNITIVE DECLINE 44

Head, D., Bugg, J. M., Goate, A. M., Fagan, A. M., Mintun, M. A., Benzinger, T. (2012).
Exercise engagement as a moderator of APOE effects on amyloid deposition. Archives of
Neurology, 69, 636-643.
Heyn, P., Abreu, B. C., & Ottenbacher, K. J. (2004). The effects of exercise training on elderly
persons with cognitive impairment and dementia: A meta-analysis. Archives of Physical
Medicine and Rehabilitation, 85, 1694-1704.
Hindin, S. B., & Zelinski, E. M. (2012). Extended practice and aerobic exercise interventions
benefit untrained cognitive outcomes in older adults: A meta-analysis. Journal of
American Geriatrics Society, 60, 136-141.
Ho, Y.-S., Yang, X., Yeung, S.-C., Chiu, K., Lau, C.-F., Tsang, A. W.-T. (2012). Cigarette
smoking accelerated brain aging and induced pre-Alzheimer-like neuropathology in rats.
PLoS One, 7(5), e36752.
Horn, J. L., & McArdle, J. J. (1992). A practical and theoretical guide to measurement invariance
in aging research. Experimental Aging Research, 18(3), 117-144.
Hsiung, G.-Y. R., Sadovnick, A. D., & Feldman, H. (2004). Apolipoprotein E ε4 genotype as a
risk factor for cognitive decline and dementia: Data from the Canadian Study of Health
and Aging. Canadian Medical Association Journal, 171, 863-867.
Huang, E. J., & Reichardt, L. F. (2001). Neurotrophins: Roles in neuronal development. Annual
Review of Neuroscience, 24, 677-736.
Israel, S., Moffitt, T. E., Belsky, D. W., Hancox, R. J., Poulton, R., Roberts, B. W. (2014).
Translating personality psychology to help personalize preventive medicine for young
adult patients. Journal of Personality and Social Psychology, 106(3), 484-498.
GENES, LIFE STYLES, AND COGNITIVE DECLINE 45

Jackson, J., Balota, D. A., & Head, D. (2011). Exploring the relationship between personality
and regional brain volume in healthy aging. Neurobiology of Aging, 32, 2162-2171.
Jankowsky, J. L., Melnikova, T., Fadale, D. J., Xu, G. M., Slunt, H. H., Gonzales, V. (2005).
Environmental enrichment mitigates cognitive deficits in a mouse model of Alzheimer's
disease. The Journal of Neuroscience, 25, 5217-5224.
Jiang, Q., Lee, C. Y. D., Mandrekar, S., Wilkinson, B., Cramer, P., Zelcer, N. (2008). ApoE
promotes the proteolytic degration of Aβ. Neuron, 58(5), 681–693.
Juster, T., & Suzman, R. (1995). Overview of the Health and Retirement Study. Journal of
Human Resources, 30, S7-56.
Kempermann, G., Fabel, K., Ehninger, D., Babu, H., Leal-Galicia, P., Garthe, A. (2010). Why
and how physical activity promotes experience-induced brain plasticity. Frontiers in
Neuroscience, 4(189), 1-9.
Kern, M. L., & Friedman, H. S. (2008). Do conscientious individuals live longer? A quantitative
review. Health Psychology, 27, 505-512.
Kim, J., Basak, J. M., & Holtzman, D. M. (2009). The role of apolipoprotein E in Alzheimer's
disease. Neuron, 63, 287–303.
King, M. W., & Resick, P. A. (2014). Data Mining in Psychological Treatment Research: A
Primer on Classification and Regression Trees. Journal of Consulting and Clinical
Psychology, Advance online publication. doi: 10.1037/a0035886
Kobilo, T., Liu, Q.-R., Gandhi, K., Mughal, M., Shaham, Y., & Van Praag, H. (2011). Running
is the neurogenic and neurotrophic stimulus in environmental enrichment. Learning &
Memory, 18, 605-609.
GENES, LIFE STYLES, AND COGNITIVE DECLINE 46

Kröger, E., Andel, R., Lindsay, J., Benounissa, Z., Verreault, R., & Laurin, D. (2008). Is
complexity of work associated with risk of dementia. American Journal of Epidemiology,
167, 820-830.
Lamke, T. A. (1958). Henmon-Nelson Tests of Mental Ability. Journal of Consulting
Psychology, 22, 241.
Lamke, T. A., & Nelson, M. J. (1957). Henmon-Nelson Tests of Mental Ability (Revised ed.).
Boston: MA: Houghton Mifflin.
Langa, K. M., Larson, E. B., Karlawish, J. H., Cutler, D. M., Kabeto, M. U., Kim, S. Y. (2008).
Trends in the prevalence and mortality of cognitive impairment in the United States: Is
there evidence of a compression of cognitive morbidity? Alzheimer’s & Dementia, 4,
134-144.
Lautenschlager, N. T., Cox, K. L., Flicker, L., Foster, J. K., van Bockxmeer, F. M., Xiao, J.
(2008). Effect of physical activity on cognitive function in older adults at riskk for
Alzheimer disease: A randomized trial. Journal of American Medical Association, 300,
1027-1037.
Leckie, R. L., Weinstein, A. M., Hodzic, J. C., & Erickson, K. I. (2012). Potential moderators of
physical activity on brain health. Journal of Aging Research. doi: 10.1155/2012/948981
Lee, J., Shih, R., Feeney, K., & Langa, K. M. (2014). Gender disparity in late-life cognitive
functioning in India: Findings from the Longitudinal Aging Study in India. Journals of
Gerontology, Series B: Psychological Sciences and Social Sciences. doi:
10.1093/geronb/gbu017
GENES, LIFE STYLES, AND COGNITIVE DECLINE 47

Li, S.-C., Papenberg, G., Nagel, I. E., Preuschhof, C., Schröder, J., Nietfeld, W. (2013). Aging
magnifies the effects of dopamine transporter and D2 receptor genes on backward serial
memory. Neurobiology of Aging, 358e351-358e310.
Lindsay, J., Laurin, D., Verreault, R., Hébert, R., Helliwell, B., Hill, G. B. (2002). Risk Factors
for Alzheimer’s Disease: A Prospective Analysis from the Canadian Study of Health and
Aging. American Journal of Epidemiology, 156, 445-453.
Little, R. J. A., & Rubin, D. B. (1987). Statistical analysis with missing data. New York: Wiley.
Liu, C.-C., Kanekiyo, T., Xu, H., & Bu, G. (2013). Apolipoprotein E and Alzheimer disease:
Risk, mechanisms and therapy. Nature Review Neurology, 9, 106-118. doi:
10.1038/nrneurol.2012.263
Lund, E. G., Kerr, T. A., Sakai, J., Li, W. P., & Russell, D. W. (1998). cDNA cloning of mouse
and human cholesterol 25-hydroxylases, polytopic membrane proteins that synthesize a
potent oxysterol regulator of lipid metabolism. The Journal of Biological Chemistry, 272,
34316–34327.
Manly, J. J., Byrd, D., Touradji, P., Sanchez, D., & Stern, Y. (2004). Literacy and cognitive
change among ethnically diverse elders. International Journal of Psychology, 39, 47-60.
Manuck, S. B., & McCaffery, J. M. (2014). Gene-environment interaction. Annual Review of
Psychology, 65, 41-70.
Markett, S., Montag, C., Diekmann, C., & Reuter, M. (2014). Dazed and confused: A molecular
genetic approach to everyday cognitive failure. Neuroscience Letters. doi:
10.1016/j.neulet.2014.02.052
Markett, S., Montag, C., & Reuter, M. (2010). The association between dopamine
DRD2polymorphisms and working memory capacity is modulated by a
GENES, LIFE STYLES, AND COGNITIVE DECLINE 48

functionalpolymorphism on the nicotinic receptor gene CHRNA44JT. Journal of
CognitiveNeuroscience, 9, 1944–1954.
Markett, S., Montag, C., Walter, N. T., & Reuter, M. (2011). Evidence for the modality
independence of the genetic epistasis between the dopaminergic and cholinergic system
on working memory capacity. European Neuropsychopharmacology, 21(2), 216-220.
Markham, J. A., & Greenough, W. T. (2006). Experience-driven brain plasticity: beyond the
synapse. Neuron Glia Biology, 1, 351-363.
McArdle, J. J. (2011). Some ethical issues in factor analysis. In A. T. Panter & S. K. Sterba
(Eds.), Handbook of ethics in quantitative methodology. London: Routledge.
McArdle, J. J. (2012). Exploratory data mining using CART in the behavioral science. In H.
Cooper, P. M. Camic, D. L. Long, A. T. Panter, D. Rindskopf & K. J. Sher (Eds.), APA
handbook of research methods in psychology, vol 3: Data analysis and research
publication. Washington DC: American Psychological Association.
McArdle, J. J. (2013). Dealing with longitudinal attrition using logistic regression and decision
tree analyses. In J. J. McArdle & G. Ritschard (Eds.), Contemporary Issues in
Exploratory Data Mining in the Behavioral Sciences (pp. 282-311). New York:
Routledge.
McArdle, J. J., Ferrer-Caja, E., Hamagami, F., & Woodcock, R. W. (2002). Comparative
longitudinal structural analyses of the growth and decline of multiple intellectual abilities
over the life span. Developmental Psychology, 38, 115-142.
McArdle, J. J., Fisher, G. G., & Kadlec, K. M. (2007). Latent variable analyses of age trends of
cognition in the Health and Retirement Study, 1992–2004. Psychology and Aging, 22(3),
525-545.
GENES, LIFE STYLES, AND COGNITIVE DECLINE 49

McArdle, J. J., & Prescott, C. A. (2010). Contemporary modeling of gene × environmental
effects in randomized multivariate longitudinal studies Perspectives on Psychological
Science, 5(5), 606-621.
Meng, X., & D'Arcy, C. (2012). Education and dementia in the context of the cognitive reserve
hypothesis: A systematic review with meta-analyses and qualitative analyses. PLoS One,
7, e38268. doi: doi:10.1371/journal.pone.0038268
Morgan, A. R., Turic, D., Jehu, L., Hamilton, G., Hollingworth, P., Moskvina, V. (2007).
Association studies of 23 positional/functional candidate genes on chromosome 10 in
late-onset Alzheimer's disease. American Journal of Medical Genetics Part B:
Neuropsychiatric Genetics., 144B, 762-770.
Morgan, J. N., & Sonquist, J. A. (1963). Problems in the analysis of survey data: And a proposal.
Journal of the American Statistical Association, 58, 415-434. doi: 10.2307/2283276
Morishima-Kawashima, M., Oshima, N., Ogata, H., Yamaguchi, H., Yoshimura, M., Sugihara, S.
(2000). Effect of Apolipoprotein E Allele ε4 on the Initial Phase of Amyloid β-Protein
Accumulation in the Human Brain. The American Journal of Pathology, 15, 2093–2099.
Moutafi, J., Furnham, A., & Crump, J. (2002). Demographic and personality predictors of
intelligence: a study using the Neo Personality Inventory and the Myers–Briggs Type
Indicato. European Journal of Personality, 17, 79-94.
Muendlein, A., Saely, C. H., Marte, T., Schmid, F., Koch, L., Rein, P. (2008). Synergistic Effects
of the Apolipoprotein E ε3/ε2/ε4, the Cholesteryl Ester Transfer Protein TaqIB, and the
Apolipoprotein C3 –482 C>T Polymorphisms on their Association with Coronary Artery
Disease. Atherosclerosis, 199, 179–186.
GENES, LIFE STYLES, AND COGNITIVE DECLINE 50

Muthén, L. K., & Muthén, B. O. (1998-2008). Mplus User's Guide. Fifth Edition. Los Angeles,
CA: Muthén & Muthén.
National Institute on Aging/Alzheimer’s Association Working Group. (1996). Apolipoprotein E
genotyping in Alzheimer’s disease. The Lancet, 347, 1091-1095.
Neeper, S. A., Gómez-Pinilla, F., Choi, J., & Cotman, C. W. (1996). Physical activity increases
mRNA for brain-derived neurotrophic factor and nerve growth factor in rat brain. Brain
Research, 726, 49–56.
Nelson, M. E., Rejeski, W. J., Blair, S. N., Duncan, P. W., Judge, J. O., King, A. C. (2007).
Physical activity and public health in older adults: Recommendation from the American
College of Sports Medicine and the American Heart Association. Circulation, 116, 1094-
1105.
Nichol, K., Deeny, S. P., Seif, J., Camaclang, K., & Cotman, C. W. (2009). Exercise improves
cognition and hippocampal plasticity in APOE ε4 mice. Alzheimer's & Dementia, 5, 287-
294.
Papassotiropoulos, A., Lambert, J. C., Wavrant-De Vrieze, F., Wollmer, M. A., von der Kammer,
H., Streffer, J. R. (2005). Cholesterol 25-hydroxylase on chromosome 10q is a
susceptibility gene for sporadic Alzheimer's disease. Neurodegenerative Diseases, 2,
233-241.
Papenberg, G., Bäckman, L., Nagel, I. E., Nietfeld, W., Schröder, J., Bertram, L. (2013).
Dopaminergic gene polymorphisms affect long-term forgetting in old age: Further
support for the magnification hypothesis. Journal of Cognitive Neuroscience, 25, 571–
579.
GENES, LIFE STYLES, AND COGNITIVE DECLINE 51

Pereira, A. C., Huddleston, D. E., Brickman, A. M., Sosunov, A. A., Hen, R., McKhann, G. M.
(2007). An in vivo correlate of exercise-induced neurogenesis in the adult dentate gyrus.
Proceedings of the National Academy of Sciences, 104, 5638–5643.
Peters, R., Poulter, R., Warner, J., Beckett, N., Burch, L., & Bulpitt, C. (2008). Smoking,
dementia and cognitive decline in the elderly, a systematic review. BMC Geriatrics,
8(36). doi: 10.1186/1471-2318-8-36
Pfrieger, F. W. (2003). Cholesterol homeostasis and function in neurons of the central nervous
system. Cellular and Molecular Life Sciences, 60, 1158–1171.
Plomin, R., Turic, D. M., Hill, L., Turic, D. E., Stephens, M., Williams, J. (2004). A functional
polymorphism in the succinate-semialdehyde dehydrogenase (aldehyde dehydrogenase 5
family, member A1) gene is associated with cognitive ability. Molecular Psychiatry, 9,
582-586.
Podewils, L. J., Guallar, E., Kuller, L. H., Fried, L. P., Lopez, O. L., Carlson, M. (2005).
Physical activity, APOE genotype, and dementia risk: Findings from the Cardiovascular
Health Cognition Study. American Journal of Epidemiology, 161, 639-651.
Pohjalainen, T., Rinne, J. O., Någren, K., Lehikoinen, P., Anttila, K., Syvälahti, E. K. G. (1998).
The A1 allele of the human D2 dopamine receptor gene predicts low D2 receptor
availability in healthy volunteers. Molecular Psychiatry, 3(3), 256-260.
Pollex, R. L., Ban, M. R., Young, T. K., Bjerregaard, P., Anand, S. S., Yusuf, S. (2007).
Association between the -455T>C promoter polymorphism of the APOC3 gene and the
metabolic syndrome in a multi-ethnic sample. BMC Medical Genetics, 8: 80. Retrieved
from http://www.ncbi.nlm.nih.gov/pmc/articles/PMC2241585/
GENES, LIFE STYLES, AND COGNITIVE DECLINE 52

Prindle, J. J., & McArdle, J. J. (2013). An examination of statistical power in multigroup
dynamic structural equation models. Structural Equation Modeling: A Multidisciplinary
Journal, 19, 351-371.
R Development Core Team. (2012). R: A language and environment for statistical computing.
Vienna, Austria: R Foundation for Statistical Computing Retrieved from http://www.R-
project.org/
Radak, Z., Hart, N., Sarga, L., Koltai, E., Atalay, M., Ohno, H. (2010). Exercise plays a
preventive role against Alzheimer's disease. Journal of Alzheimer's Disease, 20, 777-783.
Rango, F. D., Leone, O., Dato, S., Novelletto, A., Bruni, A. C., Berardelli, M. (2008). Cognitive
functioning and survival in the elderly: The SSADH C538T polymorphism. Annals of
Human Genetics, 72, 630-635.
Rodriguez-Jimenez, R., Hoenicka, J., Jimenez-Arriero, M. A., Ponce, G., Bagney, A., Araguez,
M. (2006). Performance in the Wisconsin Card Sorting Test and the C957T
polymorphism of the DRD2 gene. Neuropsychobiology, 54(3), 166-170.
Román, G. C., Nash, D. T., & Fillit, H. (2012). Translating current knowledge into dementia
prevention. Alzheimer Disease and Associated Disorders, 26, 295-299.
Rosano, C., Venkatraman, V. K., Guralnik, J. M., Newman, A. B., Glynn, N. W., Launer, L.
(2010). Psychomotor speed and functional brain MRI 2 years after completing a physical
activity treatment. Journal of Gerontology Series A: Biological Sciences and Medical
Sciences.
Rueff-Barroso, C. R., Trajano, E. T., Alves, J. N., Paiva, R. O., Lanzetti, M., Pires, K. M. (2010).
Organ-related cigarette smoke-induced oxidative stress is strain-dependent. Medical
Science Monitor, 16, BR218–BR226.
GENES, LIFE STYLES, AND COGNITIVE DECLINE 53

Sabia, S., Kivimaki, M., Kumari, M., Shipley, M. J., & Singh-Manoux, A. (2010). Effect of
Apolipoprotein E ε4 on the association between health behaviors and cognitive function
in late midlife. Molecular Nerodegeneration, 5. Retrieved from
http://www.molecularneurodegeneration.com/content/5/1/23
Scarmeas, N., & Stern, Y. (2003). Cognitive reserve and lifestyle. Journal of Clinical and
Experimental Neuropsychology, 25, 625.
Schuit, A. J., Feskens, E. J. M., Launer, L. J., & Kromhout, D. (2001). Physical activity and
cognitive decline, the role of the apolipoprotein e4 allele. Medicine and Science in Sports
and Exercise, 33, 772-777.
Seeman, T. E., Huang, M.-H., Bretsky, P., Crimmins, E., Launer, L., & Guralnik, J. M. (2005).
Education and APOE-e4 in longitudinal cognitive decline: MacArthur Studies of
Sucessful Aging. The Journal of Gerontology: Psychological Sciences, 60B(2), 74-83.
Seltzer, M. M., Floyd, F., Greenberg, J., Lounds, J., Lindstrom, M., & Hong, J. (2005). Life
Course Impacts of Mild Intellectual Deficits. American Journal on Mental Retardation
110(6), 451-468.
Sewell, W. H., & Hauser, R. M. (1975). Education, Occupation and Earnings: Achievement in
the Early Career. New York: Academic Press.
Shadlen, M.-F., Larson, E. B., Wang, L., Phelan, E. A., McCormick, W. C., Jolley, L. (2005).
Education modifies the effect of apolipoprotein epsilon 4 on cognitive decline.
Neurobiology of Aging, 26, 17-24.
Shanahan, M. J., Hill, P. L., Roberts, B. W., Eccles, J., & Friedman, H. S. (2012).
Conscientiousness, health, and aging: The life course of personality model.
Developmental Psychology. doi: 10.1037/a0031130
GENES, LIFE STYLES, AND COGNITIVE DECLINE 54

Sharp, E. S., & Gatz, M. (2011). The relationship between education and dementia: An updated
systematic review. Alzheimer Disease and Associated Disorders, 25, 289-304.
Smith, C. E., Tucker, K. L., Scott, T. M., Rompay, M. V., Mattei, J., Lai, C.-Q. (2009).
Apolipoprotein C3 polymorphisms, cognitive function and diabetes in Caribbean origin
Hispanics. PLoS One, 4(5), e5465.
Smith, J. C., Nielson, K. A., Woodard, J. L., Seidenberg, M., Durgerian, S., Antuono, P. (2011).
Interactive effects of physical activity and APOE-ε4 on BOLD semantic memory
activation in healthy elders. Neuroimage, 54, 635-644.
Smith, W. B., Starck, S. R., Roberts, R. W., & Schuman, E. M. (2005). Dopaminergic
stimulation of local protein synthesis enhances surface expression
of GluR1 and synaptic transmission in hippocampal neurons. Neuron, 45, 765–779.
Soubelet, A., & Salthouse, T. A. (2011). Personality-cognition relations across adulthood.
Developmental Psychology, 47, 303-210.
Stern, Y. (2009). Cognitive reserve. Neuropsychologia, 47, 2015-2028.
Stern, Y., Alexander, G. E., Prohovnik, I., Stricks, L., Link, B., Lennon, M. C. (1995).
Relationship between lifetime occupation and parietal flow: Implications for a reserve
against Alzheimer’s disease pathology. Neurology, 45, 55-60.
Stern, Y., Gurland, B., Tatemichi, T. K., Tang, M. X., Wilder, D., & Mayeux, R. (1994).
Influence of education and occupation on the incidence of Alzheimer’s disease. Journal
of American Medical Association, 271, 1004-1010.
Storandt, M. (2008). Cognitive deficits in the early stages of Alzheimer's disease. Current
Directions in Psychological Science, 17, 198-202.
GENES, LIFE STYLES, AND COGNITIVE DECLINE 55

Strobl, C., Malley, J., & Tutz, G. (2009). An introduction to recursive partitioning: Rationale,
application, and characteristics of classification and regression trees, bagging, and
random forests. Psychological Methods, 14(4), 323–348.
Suo, C., Leon, I., Brodaty, H., Trollor, J., Wen, W., Sachdev, P. (2012). Supervisory experience
at work is linked to low rate of hippocampal atrophy in late life. NeuroImage, 63, 1542-
1551.
Taaffe, D. r., Irie, F., Masaki, K. H., Abbott, R. D., Petrovitch, H., Ross, G. W. (2008). Physical
activity, physical function, and incident dementia in elderly men: The Honolulu-Asia
Aging Study. The Journals of Gerontology, 63A, 529-535.
Tanzi, R. E. (2012). The genetics of Alzheimer disease. Cold Spring Harbor Perspectives in
Medicine, 2, a006296. doi: 10.1101/cshperspect.a006296
Taubes, G. (1995). Epidemiology faces its limits. Science, 269(5221), 164-165+167-169.
Thomas, D. C. (2010). Gene-environment-wide association studies: Emerging approach. Nature
Review Genetics, 11, 259-272.
Townsend, J., Adamo, M., & Haist, F. (2006). Changing channels: an fMRI study of aging and
cross-modal attention shifts. NeuroImage, 31, 1682–1692.
Tucker, A. M., & Stern, Y. (2011). Cognitive reserve in aging. Current Alzheimer Research, 8,
354-360.
Valenzuela, M. J., & Sachdev, P. (2006). Brain reserve and dementia: A systematic review.
Psychological Medicine, 36, 441-454.
Van Gerven, P. W. M., Van Boxtel, M. P. J., Ausems, E. E. B., Bekers, O., & Jolles, J. (2012).
Do apolipoprotein E genotype and educational attainment predict the rate of cognitive
GENES, LIFE STYLES, AND COGNITIVE DECLINE 56

decline in normal aging? A 12-Year follow-up of the Maastricht Aging Study.
Neuropsychology, 26(459-472).
van Praag, H., Kempermann, G., & Gage, F. H. (1999). Running increases cell proliferation and
neurogenesis in the adult mouse dentate gyrus. Nature Neuroscience, 2, 266-270.
Wang, H.-X., Gustafson, D. R., Kivipelto, M., Petersen, N. L., Skoog, I., Winblad, B. (2012).
Education halves the risk of dementia due to apolipoprotein ε 4 allele: A collaborative
study from the Swedish Brain Power initiative. Neurobiology of Aging, 33, 1007.e1001–
1007.e1007.
Wechsler, D. (1997). Weschler Adult Intelligence Scale (3 ed.). New York: Psychological
Corporation.
Welsh, K., Breitner, J., & Magruder-Habib, K. (1993). Detection of dementia in the elderly using
telephone screening of cognitive status. Neuropsychiatry, Neuropsychology, and
Behavioral Neurology, 6, 111-117.
Wisdom, N. M., Callahan, J. L., & Hawkins, K. A. (2011). The effects of apolipoprotein E on
non-impaired cognitive functioning: A meta-analysis. Neurobiology of Aging, 32, 63-74.
Wise, R. A. (2004). Dopamine, learning and motivation. Nature Reviews Neuroscience, 5, 483-
494.
Wolf, M. B., & Ackerman, P. L. (2005). Extraversion and intelligence: A meta-analytic
investigation. Personality and Individual Differences, 39, 531-542.
Wong, M. Y., Day, N. E., Luan, J. A., Chan, K. P., & Wareham, N. J. (2003). The detection of
gene-environment interaction for continuous traits: Should we deal with measurement
error by bigger studies or better measurement? International Journal of Epidemiology, 32,
51-57.
GENES, LIFE STYLES, AND COGNITIVE DECLINE 57

Xu, J., & Potenza, M. N. (2012). White matter integrity and five-factor personality measures in
healthy adults. NeuroImage, 59.1, 800-807.
Yonker, J. A., Hauser, R. M., & Freese, J. (2007). The dimensionality and measurement of
cognitive functioning at age 65 in the Wisconsin Longitudinal Study. Center for
Demography and Ecology Working Paper 2007-06. Unpublished manuscript. University
of Wisconsin-Madison. Madison, WI. Retrieved from
http://www.ssc.wisc.edu/cde/cdewp/2007-06.pdf
Zhong, N., & Weisgraber, K. H. (2009). Understanding the Association of Apolipoprotein E4
with Alzheimer Disease: Clues from Its Structure. Journal of Biological Chemistry, 284,
6027-6031.

GENES, LIFE STYLES, AND COGNITIVE DECLINE 58

Table 1
Descriptive Statistics (N = 10,317)
Variable M (SD) Minimum Maximum Missing
Gender* 0.02 (0.50) -0.5 0.5 0
Age at 2003-05 64.36 (0.72) 62 73 3052
Age at 2010-11 71.27 (0.96) 68 78 4349
Years of education 13.39 (2.15) 12 20 705
Job authority 1.74 (1.71) 0 4 2533
Physical activity* -0.04 (0.50) -0.5 0.5 4055
APOE4* -0.23 (0.44) -0.5 0.5 5895
Henmon-Nelson 58.28 (26.31) 0 100 0
Literacy use at work 17.63 (14.89) 0 69 2579
Manual labor at work 22.13 (17.63) 0 69 2566
Note. * = half effect coded. For gender, -0.5 = men and 0.5 = women. For physical activity, -0.5
= never exercised, .5 = have exercised. For APOE4 -0.5 = have no ε4, 0.5 = have at least one ε4.

GENES, LIFE STYLES, AND COGNITIVE DECLINE 59

Table 2
Means, Standard Deviations, and Pairwise Correlations of Memory Measurements
Imm ‘04 Delay ‘04 Digit ‘04 Imm ‘11 Del ‘11 Dig ‘11
M 6.141 4.089 5.882 5.464 3.443 5.633
SD 1.759 2.151 1.626 1.454 1.785 1.417
Missing 4933 5095 4988 5745 5751 5775
Correlations
Imm ‘04 1.000
Del ‘04 0.720 1.000
Dig ‘04 0.299 0.250 1.000
Imm ‘11 0.304 0.273 0.214 1.000
Del ‘11 0.279 0.357 0.172 0.626 1.000
Dig ‘11 0.142 0.115 0.327 0.257 0.193 1.000

GENES, LIFE STYLES, AND COGNITIVE DECLINE 60

Table 3
A list of SNPs Measured in WLS
Gene SNP Gene SNP Gene SNP
INHBA rs2237436 COMT rs4680 ACVR2A rs1424954
INHBB rs11902591 AR rs6152 ACVR2B rs3749386
INHA rs35118453 DTNBP1 rs760761 ALK1 rs2071219
INHA rs2059693 BDNF rs6265 GABBR2 rs1435252
CYP2A6 rs5031016 5-HTR1A rs1800046 GABBR2 rs2779562
A2M rs669 5-HTR2A rs6314 CYP11B2 rs1799998
GnRH novel 5-HTR2A rs7997012 FMR1 rs1805420
IL1A rs17561 5-HTR2C rs6318 GnRHR rs17635850
LHB rs6521 5HTT rs25533 IL6 rs1800795
LHR rs4073366 MAOA rs3788862 FST rs3797297
PGR rs1042838 IGF1 rs12313279 FST rs12152850
FTO rs1421085 StAR rs3990403 TFAM rs1937
ER rs7761133 CH25H rs3802657 DTNBP1 rs1018381
FSH rs6169 FADS2 rs174575 BCKDHB rs4502885
FSHR rs6166 FADS2 rs1535 BDNF rs908867
APOE rs429358 HFE rs1799945 DRD2 rs17529477
APOE rs7412 DRD2 rs2242592 DRD2 rs6277
BRCA-1 rs1799966 DRD2 rs1800497 DRD2 rs4245147
BRCA-1 rs4986852 DRD4 rs1800955 HERC rs12913832
BRCA-2 rs144848 DRD5 rs2227845 TFAM rs2306604
CHRM2 rs8191992 DAT1 rs11564774 ESR1 rs3853248
CHRM2 rs2061174 DAT1 rs2963238 ESR1 rs12664989
KIBRA rs17070145 TPH1 rs1799913 HTT rs8076005
SNAP25 rs363050 TPH2 rs11178997 HSD17B1 rs592389
SSADH rs2760118 APOC-3 CC rs45537037 HSD17B1 rs12602084
CTSD rs17571 MT1A rs11640851 HTR2A rs6312
OXTR rs2254298 ADIPOQ rs2241766 ALDH5A1 rs2267539
PEPCK-C rs707555 ADIPOQ rs1501299 ALDH5A1 rs3761793
CYP11A1 rs8039957 LEPR rs1137100 ALDH5A1 rs807517
DISC1 rs821616 PPARG rs1805192 HTR1A rs878567

GENES, LIFE STYLES, AND COGNITIVE DECLINE 61

Table 4.
Results of Testing Invariance across Time
All equal Different means Different means
and variance
Different means,
variance, and
loadings
2003-05 2010-11 2003-05 2010-11 2003-05 2010-11 2003-05 2010-11
F mean = 0 = 0 = 0 -0.637 = 0 -0.637 = 0
F variance 2.015 2.015 1.997 1.997 2.419 1.415 2.398 1.468
immediate =1.000 =1.000 =1.000 =1.000 =1.000 =1.000 =1.000 =1.000
delayed 1.161 1.161 1.116 1.116 1.137 1.137 1.151 1.087
digit 0.357 0.357 0.353 0.353 0.357 0.357 0.360 0.346
covariance 0.783 0.904 0.867 0.874
χ
2
1520 934 676 669
df 17 16 15 13
RMSEA 0.124 0.100 0.087 0.093
Δχ
2
/Δdf -- 539/1 258/1 7/2

GENES, LIFE STYLES, AND COGNITIVE DECLINE 62

Table 5
Results of Model Comparison
-2LL ΔDF Δ2LL AIC P-value
Model 1 377648 --- --- 207076 ---
Model 2 377618 2 30 207050 p < 0.01
Model 3 377548 6 70 206992 p < 0.01
Model 4 377545 4 3 206992 p = 0.57
Model 5 377544 2 1 207000 p = 0.55

GENES, LIFE STYLES, AND COGNITIVE DECLINE 63

Figure 1. A common factor model with a latent change score. imm = immediate recall; del =
delayed recall; dig = digit ordering; [1] = 2003-05 interview; [2] = 2010-11 interview.

GENES, LIFE STYLES, AND COGNITIVE DECLINE 64

Figure 2. Numeric results of the baseline model

GENES, LIFE STYLES, AND COGNITIVE DECLINE 65

Figure 3. Numeric results of Model 3

GENES, LIFE STYLES, AND COGNITIVE DECLINE 66

Education
College
Literacy use at
work
M = 5.484
Decline = -0.836
σ
2
1
= 2.888
σ
2
e
= 1.870
N = 261
No college
CH25H
rs3802657
Openness
G:G
G:T
T:T
Memory[1] mean = M
Δ Memory mean = Decline
Memory[1] Variance = σ
2
1
Residual Variance = σ
2
e
> Q3 < Q3
M = 5.012
Decline = -0.460
σ
2
1
= 2.341
σ
2
e
= 1.553
N = 473
M = 4.632
Decline = -0.381
σ
2
1
= 2.438
σ
2
e
= 0.932
N = 298
M = 4.639
Decline = -0.445
σ
2
1
= 3.037
σ
2
e
= 1.418
N = 370
SSADH
rs2760118
M = 4.340
Decline = -0.423
σ
2
1
= 2.278
σ
2
e
= 1.437
N = 351
M = 3.917
Decline = -0.318
σ
2
1
= 2.673
σ
2
e
= 1.380
N = 265
> median < median
G:G
A:A
A:G

Figure 4. SEM Tree for men

GENES, LIFE STYLES, AND COGNITIVE DECLINE 67

Henmon-
Nelson
APOC3
rs45537037
T:T
Memory[1] mean = M
Δ Memory mean = Decline
Memory[1] Variance = σ
2
1
Residual Variance = σ
2
e
M = 5.618
Decline = -0.534
σ
2
1
= 2.668
σ
2
e
= 1.322
N = 477
DRD2
rs2242592
M = 5.907
Decline = -0.647
σ
2
1
= 3.047
σ
2
e
= 1.837
N = 385
M = 5.660
Decline =-0.455
σ
2
1
= 2.257
σ
2
e
= 2.106
N = 372
Ever
smoked
M = 5.619
Decline = -1.012
σ
2
1
= 4.018
σ
2
e
= 1.933
N = 266
M = 5.759
Decline = -0.936
σ
2
1
= 3.200
σ
2
e
= 1.617
N = 266
Conscientious
M = 5.001
Decline = -0.728
σ
2
1
= 1.881
σ
2
e
= 2.764
N = 471
C:T
C:C
T:T
C:T
C:C
< median > median
No Yes
> median
< median

Figure 5. SEM tree for women

GENES, LIFE STYLES, AND COGNITIVE DECLINE 68

Appendix A
Mplus code used for the model discussed on the confirmatory SEM.

DATA:
FILE= memory.dat;
VARIABLE:
NAMES = imm04, del04, dig04, imm11, del11, dig11 ;

USEVARIABLES ARE imm04 del04 dig04 imm11 del11 dig11 ;

MISSING = all(-999) ;
analysis: estimator = ml ;

model:
!SPECIFYING COG FACTORS.
f1 by imm04@1 ;
f1 by del04 (lam2) ;
f1 by dig04 (lam3) ;

f2 by imm11@1 ;
f2 by del11 (lam2);
f2 by dig11 (lam3);

!SETTING UP LATENT CHANGE SCORE

change by imm04@0 ;

f2 on change@1 ;
f2 on f1@1 ;
change on f1 (beta1);
[f1@0 ] (mean0);
[f2@0 ] ;
[change] (mean1) ;

f1 (varf1) ;
change (varchange);
f2@0;

!FIXING UNIQUENESS invariant
imm04 (r1);
del04 (r2);
dig04 (r3);

imm11 (r1);
del11 (r2);
GENES, LIFE STYLES, AND COGNITIVE DECLINE 69

dig11 (r3);

!FIXING VARIABLE INTERCEPTS INVARIANT
[imm04 ] (inter1);
[del04 ] (inter2);
[dig04 ] (inter3);
[imm11 ] (inter1);
[del11 ] (inter2);
[dig11 ] (inter3);

!ALLOWING SPECIFICS CORRELATED
imm04 with imm11 ;
del11 with del04 ;
dig04 with dig11 ;

output: sampstat STANDARDIZED tech1 patterns;

GENES, LIFE STYLES, AND COGNITIVE DECLINE 70

Appendix B
The following code of the OpenMx package in R was used for Model 3. “com1” = average of
immediate recall ’04 and delayed recall ’04, “dif” = difference between the average word recall
in ’11 and ’04.
datamodel = data.frame(com1,dif)
library(OpenMx)

model1 = mxModel(
name = "regression model",
type = "RAM",
mxData(observed = datamodel, type = "raw"),
manifestVars = c("com1", "dif"),
mxPath(
from = "com1",
to = "dif",
arrows = 1,
free = TRUE,
values = -.5,
labels = "beta"
),
mxPath(
from = "one",
to = c("com1", "dif"),
arrows = 1,
free = c(TRUE, TRUE),
values = c(1, 1),
labels = c("mean1", "mean2")
),
#variances and residual
mxPath(
from = c("com1", "dif"),
arrows = 2,
free = c(TRUE, TRUE),
values = c(1, 1),
labels = c("FVAR1","FVAR2")
)

)

factorfit1 = mxRun(model1)
summary(factorfit1)

GENES, LIFE STYLES, AND COGNITIVE DECLINE 71

Appendix C
The following code was used for SEM tree analysis. “com1” = average of immediate recall ’04
and delayed recall ’04, “dif” = difference between the average word recall in ’11 and ’04. The
object “datsem” consists of com1, dif, and all the covariates, but the covariates are not shown in
the following code.
datsem = data.frame(com1, dif,…)

library(semtree.base)
library(semtree.OpenMx)

changemodel = mxModel(
name = "regression model",
type = "RAM",
mxData(observed = datsem, type = "raw"),
manifestVars = c("com1", "dif"),
mxPath(
from = "com1",
to = "dif",
arrows = 1,
free = TRUE,
values = -.5,
labels = "beta"
),
mxPath(
from = "one",
to = c("com1", "dif"),
arrows = 1,
free = c(TRUE, TRUE),
values = c(1, 1),
labels = c("mean1", "mean2")
),

#variances and residual
mxPath(
from = c("com1", "dif"),
arrows = 2,
free = c(TRUE, TRUE),
values = c(1, 1),
labels = c("FVAR1","FVAR2")
)

GENES, LIFE STYLES, AND COGNITIVE DECLINE 72

)

my.control = semtree.control(alpha = 0.05)
my.control$min.N = 200
print(my.control)
mytree = semtree(changemodel, control = my.control, global.constraints = c("beta"))
plot(mytree)

Abstract (if available)

Abstract The ε4 allele of the apolipoprotein E is associated with dementia. However, ε4 is neither necessary nor sufficient to cause dementia and leaves much of variability in dementia unexplained. One possible reason for unexplained variability is gene‐environment (G × E) interactions. Are there lifestyle factors that can moderate the effect of APOE ε4 on memory? To address this question, two analytical approaches were proposed. First, the gene‐environment interaction on episodic memory was examined using the structural equation modeling (SEM). Specifically, the present study asked, “Does the effect of the ε4 on memory vary by education, job authority, or physical activity?” Second, an innovative exploratory technique called structural equation model tree (SEM Trees

Linked assets

University of Southern California Dissertations and Theses

Conceptually similar

PDF

Emergent literacy skills and home literacy environment of Latino children who live in poverty

PDF

Regularized structural equation modeling

PDF

Estimation of nonlinear mixed effects mixture models with individually varying measurement occasions

PDF

A functional use of response time data in cognitive assessment

PDF

Attrition in longitudinal twin studies: a comparative study of SEM estimation methods

PDF

A comparison of classical methods and second order latent growth models for longitudinal data analysis

PDF

Wave induced hydrodynamic complexity and transport in the nearshore

PDF

Rasch modeling of abstract reasoning in Project TALENT

PDF

An exploratory study of physical activity across different adult groups

PDF

Improving reliability in noncognitive measures with response latencies

PDF

On the latent change score model in small samples

PDF

Combination of quantile integral linear model with two-step method to improve the power of genome-wide interaction scans

PDF

Sources of stability and change in the trajectory of openness to experience across the lifespan

PDF

Effects of the perceived and objectively assessed environment on physical activity in adults and children

PDF

Pushmi-pullyu and little c: a search for the structure of personal creativity in a general population

PDF

Evaluating social-cognitive measures of motivation in a longitudinal study of people completing New Year's resolutions to exercise

PDF

The association of cerebrovascular disease risk factors with brain structure and its modification by genetic variation

PDF

Genetic and environmental risk factors for childhood cancer

PDF

Latent change score analysis of the impacts of memory training in the elderly from a randomized clinical trial

PDF

Family conflict, negative mood, and adolescents' daily problems in school

Asset Metadata

Creator Kuroki, Yusuke (author)

Core Title Identifying diverse pathways to cognitive decline in later life using genetic and environmental factors

School College of Letters, Arts and Sciences

Degree Doctor of Philosophy

Degree Program Psychology

Publication Date 07/14/2014

Defense Date 05/20/2014

Publisher University of Southern California (original), University of Southern California. Libraries (digital)

Tag aging,data‐mining,episodic memory,gene‐environment interaction,OAI-PMH Harvest,structural equation models

Format application/pdf (imt)

Language English

Contributor Electronically uploaded by the author (provenance)

Advisor McArdle, John J. (committee chair), Gauderman, William James (committee member), John, Richard S. (committee member), Zelinski, Elizabeth M. (committee member)

Creator Email yusukekuroki@gmail.com

Permanent Link (DOI) https://doi.org/10.25549/usctheses-c3-438231

Unique identifier UC11287051

Identifier etd-KurokiYusu-2671.pdf (filename),usctheses-c3-438231 (legacy record id)

Legacy Identifier etd-KurokiYusu-2671.pdf

Dmrecord 438231

Document Type Dissertation

Format application/pdf (imt)

Rights Kuroki, Yusuke

Type texts

Source University of Southern California (contributing entity), University of Southern California Dissertations and Theses (collection)

Access Conditions The author retains rights to his/her dissertation, thesis or other graduate work according to U.S. copyright law. Electronic access is being provided by the USC Libraries in agreement with the a...

Repository Name University of Southern California Digital Library

Repository Location USC Digital Library, University of Southern California, University Park Campus MC 2810, 3434 South Grand Avenue, 2nd Floor, Los Angeles, California 90089-2810, USA