University of Southern California Dissertations and Theses

Associations between longitudinal loneliness, epigenetic age, and dementia risk

(USC Thesis Other)

Associations between longitudinal loneliness, epigenetic age, and dementia risk

PDF

Download a page range

Download transcript

Copy asset link

Request this asset

Transcript (if available)

Content Associations Between Longitudinal Loneliness, Epigenetic Age, and Dementia Risk
by
Morgan Lynch, B.A.
A Thesis Presented to the
FACULTY OF THE USC DORNSLIFE COLLEGE OF LETTERS, ARTS AND SCIENCES
UNIVERSITY OF SOUTHERN CALIFORNIA
In Partial Fulfillment of the
Requirements of the Degree
MASTER OF ARTS
(PSYCHOLOGY)
August 2022
ii
Acknowledgements
It is not often that, as scientists, we are witness and participant simultaneously to our
research. I am sure that each of us reading this experienced loneliness at some point during the
COVID-19 pandemic, and I am no exception. I want to thank everyone who went to work to
fight the coronavirus despite the unknowns. I thank everyone who worked to bring us back to a
more socially-integrated world.
I am extremely grateful for my advisor and mentor, Dr. Christopher Beam, for his
steadfast encouragement and guidance during two challenging years meeting through virtual
boxes. I would also like to thank my committee members Drs. Em Arpawong, Jonas Kaplan, and
Jonathan Stange for their input and feedback.
Finally, I am grateful for my family: my mother Mary Polce-Lynch, my father, John
Lynch, my sister, Annelise Lynch, and my dog Poco, for their love and support. This
accomplishment would not have been possible without any of them.

iii
Table of Contents

Acknowledgements ........................................................................................................................ ii
List of Tables .................................................................................................................................. iv
List of Figures .................................................................................................................................. v
Abstract ........................................................................................................................................... vi
Chapter 1: Introduction .................................................................................................................... 1
1.1 Physiological Dysregulation and Dementia Risk ............................................................ 1
1.2 Epigenetics of Physiological Dysregulation in Dementia ............................................... 3
1.3 Loneliness and Dementia Risk ........................................................................................ 6
1.4 Epigenetics as a Mediator of the Association Between Loneliness and
Dementia Risk ................................................................................................................. 8
1.5 Loneliness as a Moderator of the Association between Epigenetic .................................
Age and Dementia Risk ................................................................................................. 12
1.6 Present Study ................................................................................................................. 13
Chapter 2: Methods ....................................................................................................................... 14
2.1 Sample ........................................................................................................................... 14
2.2 Measures ........................................................................................................................ 14
2.3 Data Analysis ................................................................................................................. 17
2.3.1 Missing Data .................................................................................................................. 18
2.3.2 Baseline Loneliness Mediation ..................................................................................... 18
2.3.3 Mediation and Moderation using LTC .......................................................................... 19
Chatper 3: Results .......................................................................................................................... 23
3.1 Sample Descriptive Statistics ........................................................................................ 23
3.2 Baseline Meditation ....................................................................................................... 24
3.3 Covariance Pattern Mixture Modeling and Growth Mixture Modeling ........................ 25
3.4 Loneliness Latent Class Mediation ................................................................................ 27
3.5 Moderation ..................................................................................................................... 30
Chapter 4: Discussion .................................................................................................................... 33
4.1 Epigenetic Age as a Mediator of Loneliness on Dementia Risk ................................... 33
4.2 Loneliness and Dementia Risk ...................................................................................... 35
4.3 Epigenetic Age and Dementia Risk ............................................................................... 35
4.4 Limitations and Future Directions ................................................................................. 36
References ..................................................................................................................................... 38
Appendix ....................................................................................................................................... 49
iv
List of Tables

Table 1: Sample Descriptive Statistics .......................................................................................... 23
Table 2: Correlation Matrix of Independent and Dependent Variables ........................................ 24
Table 3: Covariance Pattern Mixture Model Comparison Statistics ............................................. 26
Table 4: Growth Mixture Model Comparison Statistics ............................................................... 26
Table 5: Latent Class Descriptions ................................................................................................ 26
Table 6: Descriptive Statistics by Loneliness Latent Class ........................................................... 27
Table 7: Mediation Parameter Estimates, Indirect, Total, and Proportional Effects ..................... 29
Table 8: Multivariate Analysis of Variance Model Results .......................................................... 31

v
List of Figures

Figure 1: Conceptual model of the relationships tested between loneliness trajectories, dementia
risk, and epigenetic age. .................................................................................................................. 5
Figure 2: Growth Mixture Model (GMM) .................................................................................... 21
Figure 3: Mediation Model and Parameter Estimates ................................................................... 29
Figure 4: Forest Plot of Association of DNAm PhenoAge and Dementia Risk in Each Loneliness
Group Between the Unconstrained and Full Model ...................................................................... 32

vi
Abstract

In the absence of a treatment for dementia, identifying modifiable psychosocial risk
factors and biomarkers for dementia are priorities to prevent or delay onset. Loneliness is
regarded as a psychosocial risk factor for dementia. Only three studies to date have explored
whether different trajectories of loneliness across the lifespan influence dementia risk
differentially (Wilson et al., 2007; Kim et al., 2021; Akhter-Khan et al., 2021). Loneliness may
influence biomarkers that are more proximally related to dementia risk. Epigenetic age, for
example, quantifies biological aging that incorporates dysregulated physiological processes that
more accurately characterize the aging process and correlate with psychosocial stress,
Alzheimer’s disease pathology, and cognitive functioning. The present study uses growth
mixture modeling to investigate different latent class trajectories of loneliness in the Health and
Retirement Study using three loneliness measurements collected across eight years. We test
whether different loneliness trajectories differentially predict epigenetic age, dementia risk, and
their association. We find that groups characterizing chronic loneliness and increasing loneliness
across the study window have increased dementia risk, and higher epigenetic age predicted
higher dementia risk. Loneliness trajectory was not found to be associated with epigenetic age,
epigenetic age did not mediate the association between loneliness and dementia risk, and the
association between epigenetic age and dementia risk did not differ between loneliness groups.
Results do not support the physiological dysregulation hypothesis of loneliness on dementia risk,
and suggest epigenetic age is a potential biomarker of dementia risk.

Keywords: dementia, epigenetic age, longitudinal, loneliness, psychosocial risk factors
1

Associations Between Longitudinal Loneliness, Epigenetic Age, and Dementia Risk
Chapter 1: Introduction

Associations between physiological dysregulation and dementia risk are well established
in the literature (Iturria-Medina et al., 2016; Gross et al., 2018; Viswanathan et al., 2009; Levine
& Crimmins, 2012; Marchesi, 2011). Although there is an increasing focus on using epigenetic
markers to quantify the effects of physiological dysregulation on the aging process (Horvath,
2013), few studies have evaluated associations of epigenetic mechanisms on cognitive
impairment, particularly risk of dementia Furthermore, a plethora of psychosocial risk factors
for dementia may exert their risk own dementia through epigenetic changes. The present study
disentangles heterogeneity of a well-replicated psychosocial risk factor for dementia, loneliness,
on the association between epigenetics and dementia risk. The broad aims of this study are to test
how different loneliness trajectories over eight years predict differences in epigenetic age,
dementia risk, and the association between epigenetic aging dementia risk, and to test whether
epigenetic age and dementia risk are negatively associated in older adulthood.

1.1 Physiological Dysregulation and Dementia Risk

Dementia is characterized by the acquired loss of cognitive functioning in at least one
cognitive domain, for which loss is substantial enough to impair everyday functioning (APA,
2013). Cognitive loss is attributed to changes in cell, tissue, and organ structure and function,
which corresponds to different dementias. As such, subtypes of dementia (Alzheimer’s,
Frontotemporal, Vascular, Parkinson’s, Lewy Bodies) are defined by different neuropathology
and symptoms; and all types are characterized by stark physiological changes. For example,
2
cerebrovascular dysfunction is a shared etiology among dementia subtypes. Decreased cerebral
blood flow is associated with from Mild Cognitive Impairment (MCI) to dementia (Chaeo et al.,
2010). Other vascular issues such as hypertension and dyslipidemia are known risk factors for
dementia (Kalaria et al., 2012). Changes in blood flow disrupts blood vessel integrity, leading to
neuropathological white matter hyperintensities and cerebrovascular lesions (Petrovitch et al.,
2005). Further, increases in the prevalence of β-amyloid (Aβ) and tau protein in brain tissue are a
hallmark of Alzheimer’s Disease (AD) (Gallardo & Holtzman, 2019). Calcium signaling is
another implicated physiological pathway in AD. Dysregulation in calcium homeostasis plays a
critical role in the pathogenesis of AD (Sushma & Mondal, 2019).
Disruptions to biological pathways associated with ADRD are caused partly by genetic
predispositions. Twin studies have shown that Alzheimer’s Disease is about 60% heritable (Gatz
et al., 2006), and about 30-60% of the genetic risk for ADRD can be estimated by common
genetic variants in the population (Verheijen & Sleegers, 2018). Genome-wide association
studies have revealed up to 40 variants – common and unique – associated with AD. The APOE
ε4 allele is the largest known genetic risk factor for ADRD, and accounts for 25-50% of the
heritability of ADRD (Andrews et al., 2020), although ε4/ε4 homozygotes are still at only 60%
risk of developing AD; while variants in the APP, PSEN1, and PSEN2 genes virtually guarantee
early-onset AD (Tanzi et al., 2012). Although estimates of ADRD heritability are high, people’s
unique environmental exposures and experiences also account for a large proportion of the
variance in dementia risk (Beam et al., 2020). Thus, the study of the interaction between genetics
and environment is a key area of interest in preventing and delaying disease onset.

3
1.2 Epigenetics of Physiological Dysregulation in Dementia

Epigenetic aging – which indexes the global lifetime toll of environmental and cellular
stress – may correlate with dementia risk. Research at the molecular level demonstrates that gene
expression controls physiological processes implicated in dementia risk (Chouliaras et al., 2010).
Epigenetic mechanisms, which include DNA methylation, histone post-translational
modifications, and non-coding RNAs, result in changes in gene expression that in turn alter
phenotypes (Zarzour et al., 2020). AD is associated with hypermethylation of the APOE gene
that affects amyloid-β processing (Wang et al., 2008), hypomethylation of the amyloid precursor
protein (Tohgi et al., 1999a), and hypomethylation of the promoter region of the tau protein gene
in the parietal cortex, resulting in downregulation of its transcription (Tohgi et al., 1999b) and
greater dementia risk. Epigenetic mechanisms correlate with biomarkers of neuropathology of
dementia, and thus are important intermediary processes to consider in research on
environmental and psychosocial risk factors for dementia.
Reliable effects of epigenetic mechanisms on dementia risk suggest that individual
differences in epigenetic processes may alter the aging process, for better or worse. One way to
quantify epigenetic aging processes is through associating DNA methylation at specific sites on
the genome – Cytosine phosphate Guanines (CpG) markers – with aging processes, and
aggregating their methylation levels. In this way, DNA methylation quantifies the toll that aging
and environmental exposures have on people’s physiological functioning; that is, DNA
methylation is a measure of the age of human cells, tissues, and organs. Both natural events that
occur during aging as well as external and cellular environments influence DNA methylation
(Horvath, 2013; Christensen et al., 2009). By aggregating DNA methylation levels of CpG
markers, measures of biological aging can be estimated, which index how quickly or slowly
4
someone’s DNA has aged irrespective of chronological age. Collectively, these measures are
referred to as “epigenetic clocks.”
Over a dozen epigenetic clocks have been developed in the past decade to quantify
epigenetic aging, also referred to as biological aging and phenotypic aging. A recent review
article compared 15 clocks – chronological (1
st
generation) and biological (2
nd
generation) clocks
– which are based on methylation of 3 to 505 genes (Bergsma & Rogaeva, 2020). Each DNA
methylation clock is unique (Bell et al., 2019), and uses different CpG sites. Epigenetic clocks
are developed for general and specific phenotypes, so not all are relevant for understanding
whether they confound or mediate the association between dementia risk and loneliness. As such,
the current study includes a measure of epigenetic age designated to predict general age-related
outcomes.
Epigenetic age – depending on the clock – may be a biomarker of brain health, and
therefore may have predictive utility for dementia risk. Indeed, epigenetic age predicts
neuropathology that characterizes AD. In one study, epigenetic age of the dorsolateral prefrontal
cortex, an area of the brain vulnerable to neurodegeneration in Alzheimer’s disease (AD), was
found to be associated with neuropathological markers of AD, including amyloid load and
diffuse plaques, and poorer global cognitive functioning (Levine et al., 2015). Additionally,
accelerated age, which is the difference between biological and chronological age, correlates
with poorer cognitive ability (Marioni et al., 2015). However, in a larger study of epigenetic age
based on venous blood samples, epigenetic age acceleration was not associated with increased
dementia risk (Sibbett et al., 2020). Given these mixed findings, the first aim of the present study
is to examine whether epigenetic age predicts dementia risk (Figure 1A, path b).
5
Figure 1: Conceptual model of the relationships tested between loneliness trajectories, dementia
risk, and epigenetic age.

Note. Figure 1A: Mediation of baselines loneliness (UCLA 2008 score) on dementia risk through
epigenetic age. Figure 1B: Mediation using latent classes from Growth Mixture Modeling as IV. Figure
1C: Moderation of latent classes on association between DNAm PhenoAge and Dementia Risk.

In the present study, the association between dementia risk and an epigenetic clock that
measures general physiological age will be tested. Known as DNAm PhenoAge (Levine et al.,
2018), this second-generation epigenetic clock was trained on clinical measures, in addition to
chronological age, to account for differences in physiological functioning among individuals of
the same chronological age. DNAm PhenoAge includes phenotypic measures of physiological
dysregulation (e.g., C-reactive protein, creatine, white blood cell count), DNA methylation (CpG
sites), clinical measures, and chronological age in its prediction of epigenetic age (Levine et al.,
6
2018). Physiological status is implicated in both loneliness and dementia, thus, DNAm
PhenoAge is a supported biomarker for the present study.

1.3 Loneliness and Dementia Risk

Loneliness is conceptualized as the negative feelings that result from a lack of sufficient
social connection, either in quantity of quality (Perlman & Peplau, 1981; Goossens et al., 2015).
It is distinguished conceptually from, and weakly correlated with, objective social isolation (r =
0.16 – 0.25; Coyle et al., 2012; Matthews et al., 2016; Tomaka et al., 2006; Cornwell et al.,
2009).
Loneliness has been found to be associated with poorer cognitive performance and
declines in global cognition (Holwerda et al., 2014; Tilvis et al., 2004) and risk for Alzheimer’s
disease (Wilson et al., 2007). However, recent systematic reviews have revealed diverging
conclusions, with studies reporting both significant (Lara et al., 2019; Solmi et al., 2020) and
non-significant associations (Penninkilampi et al., 2018) between loneliness and dementia.
Among ten studies showing significant results, eight studies yield small effect sizes (0.10 - 0.30),
and two report medium effect sizes (Holwerda et al., 2014; Wilson et al., 2007). In another
recent meta-analysis, Victor (2021) tested the proposition that loneliness causes dementia and
found that of eleven studies, five reported a significant relationship between loneliness and
dementia, with relative and hazard risk of dementia ranging from 15% to 64% amongst those
who were lonely. Together, this research suggests that loneliness likely increases risk for
dementia marginally; however, previous studies have not categorized loneliness to account for
heterogeneity. That is, those with short durations of loneliness in late adulthood might not be at
7
risk for dementia in the same way that people with chronic loneliness across mid- to late
adulthood might be.
The experience of loneliness is not uniform across individuals – some people experience
chronic or increasing loneliness whereas others report never feeling lonely. For example, 15-30%
of the population report that loneliness is a chronic state (Hawkley & Cacioppo, 2013). For
others, loneliness follows a U-shaped trend across the lifespan that affects young adults and older
adults (> 70 years of age) more than all other age groups (Pinquart & Sorensen, 2001; Beam &
Kim, 2020). One theory that could explain increasing loneliness in older adulthood is
socioemotional selectivity theory (SST). SST describes how motivations for the quality and
quantity of social connections change with age. For example, in young adulthood, individuals are
more future-oriented, and tend to seek relationships that will help them gain new knowledge.
Goals shift in mid-adulthood, and people tend to become more person-focused rather than group-
focused. As such, they more often invest in a small number of relationships that bring meaning
and positive emotion than novel experience (Ziaei & Fischer, 2016). Loneliness may be a
byproduct of investing in stronger, but fewer, social relationships because with age it is more
likely that loved ones will pass away, and there may be a paucity of high-quality connections to
replace the lost relationships (Carstensen, 1992) thus increasing risk for loneliness in older
adulthood. As people age, factors like proximity to a larger community, personality traits,
emotional resiliency may increase or decrease chances of creating new meaningful relationships.
Indeed, research has found that there are factors that affect lifespan loneliness risk; including
peer acceptance (Renshaw & Brown, 1993), childhood trauma (Kochenderfer-Ladd & Wardrop,
2001), bereavement (Grimby, 1993), and social network size and diversity (Hawkley &
8
Cacioppo, 2007). While SST may explain general increasing trends of loneliness with age,
individual differences affect risk for loneliness across the lifespan.
Although baseline loneliness scores predict dementia risk, only a few studies have tested
whether change in loneliness predicts dementia risk, with differing conclusions. In addition to
the inconclusive evidence for loneliness on dementia risk, the temporal associations between
them are not well understood. It may be the case that memory functioning has a lagged effect on
loneliness in which dementia onset precedes loneliness (Ayalon et al., 2016). Given the
heterogeneity of loneliness across individuals and the lifespan, and the inconsistent relationship
between loneliness and dementia, it is important to identify different trajectories of loneliness
across the lifespan. Although recent studies show that worsening loneliness with age is not
associated with greater dementia risk (Kim et al., 2021), it may be the case that people with
chronic loneliness with age may have greater dementia risk, as found by Akhter-Khan et al.
(2021). As shown in the conceptual figure (Figure 1B), an additional aim of the current study is
test whether different loneliness trajectories, as described by intercepts and slopes, predict
dementia risk directly (path c’) and through epigenetic age (paths a*b, c).

1.4 Epigenetics as a Mediator of the Association Between Loneliness and Dementia Risk

The mechanisms by which loneliness translates from psychological experience to
epigenetic age and dementia risk remain unclear. Research suggests at least three different
models by which loneliness can influence morbidity and mortality: health risk behaviors,
psychological distress, and physiological dysregulation (Hawkley & Cacioppo, 2010). Support
for the health behaviors model comes from evidence that lonely individuals engage in health
damaging behaviors including reduced physical activity (Hawkley et al., 2009), poor sleep
9
quality (Cacioppo et al., 2002), poor dietary choices (Steptoe et al., 2004), and smoking (Shankar
et al., 2011), which in turn have adverse effects on cardiovascular functioning (Mullington et al.,
2009). The psychological distress model posits that loneliness is closely associated with
psychopathology such as depression (Cacioppo et al., 2010), psychosis (DeNiro, 1995), and
suicidal ideation (Stravynski & Boyer, 2001), which are risk factors for morbidity and mortality.
Models that apply to health outcomes are likely to also apply to cognitive outcomes. Finally, the
physiological dysregulation model states that loneliness is associated with dementia through
physiological changes. The present study focuses on physiological dysregulation as the potential
mechanism by which loneliness affects dementia risk, and uses epigenetic age as an estimate and
marker of physiological dysregulation.
Research has shown that the effects of loneliness exacerbate physiological aging
(Hawkley & Cacioppo, 2007). In other words, homeostatic processes that can become more
dysregulated with age (e.g., cardiovascular physiology – blood pressure, neuroendocrine
functioning – cortisol, epinephrine levels), are more dysregulated in lonely individuals (Hawkley
& Cacioppo, 2007). Additionally, loneliness acts as an independent risk factor for morbidity,
including dementia, after controlling for other confounds (O’Luanaigh et al. 2012; Lara et al.,
2019). The physiological dysregulation model and the psychological distress models are not
mutually exclusive. Kim et al. (2020), for example, found that functional ability, social
participation, self-rated health, and depressive symptoms significantly mediated effects of
loneliness on dementia risk. Taken together, this evidence suggests that the effects of loneliness
on dementia risk might include physiological mechanisms of which epigenetic age could
indicate.
10
The stress sensitization model (SSM) may explain the effects of chronic or repeated
loneliness as a psychosocial stress across the lifespan. SSM was originally applied to recurrence
of major depressive episodes, yet it may also explain recurrent or sustained loneliness in older
adulthood (Monroe & Harkness, 2005). The SSM, based on Post’s kindling hypothesis (Post,
1992), states that repeated stressors may generate increased affective and behavioral responses
(Stroud, 2018; Post, 1992). The effects of loneliness may be cyclical between generation and
sensitization of loneliness occurrences, whereby individuals who experience loneliness may not
only be more likely to experience loneliness later, but they may also be more sensitive to
experiences that increase likelihood of loneliness, like interpersonal rejection. Indeed, there
appears to be a cumulative effect of chronic loneliness on health outcomes, including higher
mortality risk relative to individuals who are “situationally lonely” (Shiovitz-Ezra & Ayalon,
2010).
People who experience chronic loneliness may be at risk for accelerated aging, putting them
at greater risk for dementia. Transcriptional processes and genetic expression appear to be part of
the process by which loneliness could lead to older epigenetic age. In line with SSM, loneliness
might trigger stress reactivity (Boss et al., 2015), resulting in increased circulating levels of the
stress hormone cortisol (Adam et al., 2006; Steptoe et al., 2004). Further, prolonged
hypercortisolism may cause neuronal damage associated with altered cognition (Epel, 2009).
Paradoxically, glucocorticoid receptors, the activation sites of cortisol on cell membranes, have
anti-inflammatory effects, but lonely individuals are at higher risk for inflammation-mediated
diseases. One possible explanation that has been tested is that glucocorticoid receptors can
become inactive as a result of reduced expression of glucocorticoid receptor genes, itself an
epigenetic process (Pace et al., 2007). Cole et al. (2007), for example, identified 209 genes that
11
were expressed differently in leukocytes of high lonely individuals, compared to low-lonely
individuals. Genes involved in immune activation, transcription control, and cell proliferation
were up-regulated whereas genes supporting mature B lymphocyte function and type I interferon
response were down-regulated in lonely individuals. Together, these epigenetic changes result in
reduced anti-inflammatory responses, which explains why loneliness is associated with
inflammation-mediated morbidity despite increases in cortisol. Further, these associations held
after controlling for demographic, psychological, or medical characteristics. Thus, transcriptional
processes and genetic expression appear to be part of the process by which loneliness implicates
immunological functioning.
Social isolation in animals is also associated with epigenetic changes in the brain. In animal
models, researchers can manipulate social contact and study downstream molecular effects.
Researchers have found that social isolation affects gene expression in host immune responses
and pathogens (Levine & Mody, 2003; Cacioppo et al., 2002). In adult mice, social isolation for
3 months was found to lead to increased global DNA methylation in the midbrain, and increased
histone methylation (H3K4) in the midbrain (Siuda et al, 2014). Additionally, in Drosophila
(fruit flies), genes that encode for activity-regulated transcription factors in dopaminergic
neurons respond to social isolation (Agarwal et al., 2018).
The effect of cumulative loneliness across the lifespan on epigenome-wide processes,
epigenetic aging, has not been explored. Effects of psychosocial stress on epigenetic aging has
focused mostly on early life stressors, with discordant evidence for lifetime exposures to
psychosocial stress. Epigenetic age correlates with low early-life SES (Austin et al., 2018),
childhood trauma (Han et al., 2018), threatening early life adversities (Sumner et al., 2019), and
racial discrimination during adolescence without family support (Brody et al., 2016). However,
12
adult levels of SES did not correlate with epigenetic age (Austin et al., 2018). In another study,
cumulative lifetime stress, but not current or childhood stress, was associated with epigenetic age
(Zannas et al., 2015). Loneliness, however, has not been investigated as a predictor of advanced
epigenetic age despite prior research that supports altered gene expression in chronically lonely
people (Cole et al., 2007).
Taken together, a preponderance of research suggests that psychosocial stressors alter the
epigenome, depending on type and recency. Following this reasoning, loneliness may
differentially alter the epigenome – depending on duration and consistency of exposure – which
may in turn differentially affect dementia risk. The first aim of this study is to test whether
epigenetic age mediates the relationship between loneliness and dementia risk, and whether
effects differ between longitudinal loneliness patterns.

1.5 Loneliness as a Moderator of the Association between Epigenetic Age and Dementia
Risk
So far, we have presented how epigenetic age might mediate effects of loneliness on
dementia risk. However, dementia risk and epigenetic age may be bidirectionally associated. The
directional relationship between blood-based epigenetic clocks and dementia risk has not been
thoroughly tested, with some evidence showing that epigenetic age negatively predicts follow-up
cognition in the blood-based clocks GrimAge (Hillary et al., 2021) and BioAge (Wu et al.,
2021). Moreover, because aging is the common underlying risk factor for both higher epigenetic
age, and higher dementia risk, it may be the case that they are bidirectionally associated. The
final aim of this study is to examine whether longitudinal loneliness moderates the effect of
epigenetic age on dementia risk (Figure 1C, path d
i
).

13
1.6 Present Study

Because prior research shows that baseline loneliness reliably predicts ADRD, we first test
whether epigenetic age mediates effects of baseline loneliness on dementia risk (i.e., TICSm
score). The first test is represented in Figure 1A. Next, because loneliness changes in late
adulthood within individuals, particularly after age 75 (Beam & Kim, 2020), we will then model
different loneliness trajectories. We predict that longitudinal modeling will show that loneliness
follows at least 4 different trajectories in older adults – including a stable low lonely class, a
chronically lonely class, a moderately increasingly lonely class, and moderately decreasingly
lonely class. These latent classes will be referred to as loneliness trajectory class (LTC). We will
use these LTCs as multicategorical IVs in a mediation of LTC on dementia risk by epigenetic
age (represented in Figure 1B). We predict that epigenetic age will partially mediate the
association between LTC and dementia risk in high and increasing LTCs. We predict that LTCs
of high and increasing loneliness will have statistically significant negative relative direct effects
of LTC on dementia risk (Figure 1B, c’
i
), statistically significant negative associations with
epigenetic age, and statistically significant relative indirect and total effects of DNAm PhenoAge
on dementia risk in comparison to the reference LTC (Figure 1B, a*b, c+a*b). Lastly, in our
third analysis we will test whether the association between epigenetic age and dementia risk is
moderated by LTC. We predict that the association between epigenetic age and dementia risk
will be negatively correlated, and will be moderated by LTC such that higher loneliness classes
will have a stronger negative correlation between DNAm PhenoAge and TICSm (Figure 1C, d
i
).

14
Chapter 2: Methods

2.1 Sample

Data come from three waves of the Health and Retirement Study (HRS). The HRS is an
ongoing longitudinal study that began in 1992 that surveys a representative sample of Americans
over 50 years of age. Baseline Core interviews are conducted face-to-face and follow-up
interviews are conducted either by phone or in-person every two years, with psychosocial and
lifestyle questionnaires completed every 4 years. Core interviews assess demographics, physical
health, cognition, family structure, functional limitations, housing, employment history,
disability, widowhood and divorce, and self-report psychosocial functioning. The HRS conducts
additional studies including the Venous Blood Study, and the Aging, Demographics, and
Memory (ADAMS) study, and Healthy Cognitive Aging Project (HCAP) that include validation
and estimation of biological aging and dementia risk measures.
The sample in the present study includes all individuals who completed the UCLA
Loneliness Scale in 2008, 2012 or 2016 through the Core HRS, provided blood samples for
epigenetic analysis in 2016 through the Venous Blood Study and HCAP, and completed
cognitive assessment for dementia in 2016 through the Core HRS (N = 1814; 59% Female; Mean
Age = 70).

2.2 Measures

Loneliness: A modified version of the UCLA Loneliness Scale, 11-item, was
administered to all participants in 2008, 2012, and 2016. The 11 items were selected from the
13-item scale based on published factor loadings with older adults (Russell, 1996; Hawkley et
al., 2005). This measure asks participants to rate how much of the time they feel alone, part of a
15
group of friends, etc. A linear index for each time point (2008, 2012, 2016) was created from the
sum of scores from all 11 questions. Higher scores on the UCLA Loneliness Scale indicate
higher levels of loneliness (R Core Team, 2020). A score of 11 would mean an individual hardly
ever, or never, felt lonely; a score of 23 or higher would mean an individual endorsed often
feeling lonely at least once. Composite reliability of the UCLA Loneliness Scale in the HRS was
assessed for each time point. Reliability of the measure was found to be substantial in 2008 (ω =
0.88, SE = 0.01), 2012 (ω = 0.88, SE = 0.01), and 2016 (ω = 0.87, SE = 0.01).
Dementia risk: Dementia risk was determined using a composite measure of cognitive
functioning developed from the Telephone Interview for Cognitive Status (TICS-M) (Weir et al.,
2011; Brandt et al., 1988). The 27-point index includes scores from 1) immediate and delayed
10-noun free recall test to measure memory (0 to 20 points); 2) a serial sevens subtraction test to
measure working memory (0 to 5 points); and 3) a counting backwards test to measure speed of
mental processing (0 to 2 points). Crimmins (2011) classified the 27-point continuous variable
into cutoffs for Dementia (0-6), Cognitively Impaired by not Demented (CIND) (7-11), and
Normal (12-27). The continuous variable is used in the present study; cutoffs are provided for
interpretation.
Epigenetic age: Levine et al. (2018)’s epigenetic clock, DNAm PhenoAge was generated
by first developing a phenotypic age using a nationally representative sample of 9,926 adults,
and validated in 6,209 adults, where it was found to be associated with a 9% increase in the risk
of mortality from aging-related diseases, and secondly correlating phenotypic age with DNA
methylation sites; 513 CpG sites were selected that were predictive of phenotypic age. The
phenotypic aging measures used in the calculation of DNAm PhenoAge are: albumin, creatine,
16
glucose (serum), C-reactive protein (log), Lymphocyte percent, mean red cell volume, red cell
distribution width, alkaline phosphatase, white blood cell count, and age (Levine et al., 2018).
Blood samples in this study were collected through the Venous Blood Study, within about
four weeks of the HRS Core interview. Participants were recommended to fast before blood
collection. Phlebotomists collected 50.5mL of blood in 6 tubes, including 1 10mL EDTA whole
blood tube which was used for DNA extraction. DNA methylation was measured using the
Infium Methylation EPIC BeadChip at the University of Minnesota (Crimmins et al., 2020).
Demographics: Sex (2008) was coded based on participant’s sex at birth, male or female.
Education (2008) was coded based on the highest grad of school or year of college completed.
Cohort was dummy coded according to cohort delineations summarized by the RAND Center for
the Population of Aging (2010). Race (2008) was coded into the categories: White/Caucasian,
Black/African American, and Other.
Covariates: Body Mass Index (2016) was calculated using English measurements
according to the formula BMI =
!"# % &'(
)*+,-#
!
. Self-Rated Health (2016) was assessed using a single
item “Would you say your health is excellent, very good, good, fair, or poor?” Higher scores
indicate poorer self-rated health. Current Smoking Status (2016) was assessed using a single item
“Do you smoke cigarettes now? (not including pipes, cigars, or e-cigarettes)” and coded as 1 =
Yes, 2 = No, 8 = Don’t Know. Depression (2016) was measured using the CES-D 11-item,
which was administered to participants in 2016 as part of the Harmonized Cognitive Assessment
Protocol (Radloff, 1977). Due to convergent validity concerns, Item 5, which asks participants
whether they felt lonely over the past two weeks, was excluded from this study. A composite
measure was conducted from the remaining 10 items. Higher scores indicate higher depressive
symptoms. Social Isolation was coded as the amount of contact with four types of relationships:
17
spouse, children, friends, other family on a scale of 0 (social integration) to 4 (social isolation).
Coding followed the same procedures used in Sutin et al. (2020). In each relationship domain,
participants answered whether they had contact with the other person, including in-person
contact, phone calls, e-mail/writing less than once per month (0), or once or more per month (1).
APOE ε4 status was coded as the number of ε4 alleles, e.g., a participant with an APOE
genotype of 34 received a code of 1.

2.3 Data Analysis

First, we present results from basic descriptive analyses. These include means and
standard deviations of key variables (i.e., UCLA Loneliness scores, DNAm PhenoAge, TICSm).
Here, we also present correlations among these variables. Second, we present findings from a
mediation analysis using participants’ baselines loneliness score, DNAm PhenoAge, and TICSm,
to test whether epigenetic age mediates effects of baselines loneliness on dementia risk. Third,
we present the analysis using all three measurements of loneliness (i.e., 2008, 2012, and 2016).
This analysis consists of three parts. In the first, we identify latent loneliness trajectories using
Growth Mixture Modeling (GMM). In the second, we first present findings of a mediation
analysis in which LTC is used instead of baseline loneliness. Finally, we present findings from a
Multivariate Analysis of Variance (MANOV A) test to determine whether different latent
longitudinal loneliness classes differentially predict epigenetic age, dementia risk, and the
correlation between them. All analyses were conducted in Mplus 8.6 (Muthén & Muthén, 2018).
All analyses adjust for sex, education, race, cohort, baseline age, current depression, current
body mass index, current smoking level, current self-rated health, current objective social
isolation, and APOE ε4 allele count. Bonferroni corrections were used for multiple comparisons.
18
2.3.1 Missing Data

Missing data on the UCLA Loneliness Scale, at any timepoint, was analyzed to determine
whether data are missing completely at random (MCAR). Missingness was associated with
DNAm PhenoAge (t = -12.81, p < 0.05), objective social isolation (t = -2.05, p < 0.05), BMI (t =
2.31, p < 0.05), current smoking (t = 3.60, p < 0.05), CESD scores in 2016 (t = 17.91, p < 0.05),
and race (t = 6.54, p < 0.05). Missingness was not associated with TICSm (t = -1.46, p = 0.14),
nor sex (t = -1.04, p = 0.30). Since variables of interest are related to missingness on the UCLA
Loneliness Scale, it is determined that data are missing at random (MAR). Missing data were
estimated using full information maximum likelihood.

2.3.2 Baseline Loneliness Mediation

To test the prediction that epigenetic aging is a mechanism by which loneliness has an
effect on dementia risk, we test a mediation model using the baseline (2008) loneliness variable
as the IV , DNAm PhenoAge as the mediator, and dementia risk as the DV (Figure 1B).

19
Figure 1: Conceptual model of the relationships tested between loneliness trajectories, dementia
risk, and epigenetic age.

Note. Figure 1A: Mediation of baselines loneliness (UCLA 2008 score) on dementia risk through
epigenetic age. Figure 1B: Mediation using latent classes from Growth Mixture Modeling as IV. Figure
1C: Moderation of latent classes on association between DNAm PhenoAge and Dementia Risk.

2.3.3 Mediation and Moderation using LTC

To model heterogeneity in the loneliness data, we employ growth mixture modeling
(GMM) to parse out discrete latent growth trajectories of loneliness, and to test associations
between loneliness trajectories and distal outcomes. We begin by identifying longitudinal
loneliness trajectory groups in the data. First, we determined the covariance pattern to use in
Growth Mixture Modeling using Covariance Pattern Mixture Modeling (CPMM). Using CPMM
has been found to lead to less biased class-specific growth parameters (McNeish & Harring,
20
2019). Unstructured, toeplitz, and compound symmetric structures were tested to model the
loneliness data over time. To determine the appropriate covariance structure, we examined the
Bayesian and Aikaike information criterion indices, entropy values, and considered substantive
meaning in model selection following guidance from McNeish & Harring (2020).
Growth Mixture Modeling (GMM) was used to identify classes of longitudinal loneliness
trends. This person-centered approach classifies individuals into distinct groups and is chosen
over alternative techniques, such as latent class analysis, to allow for random effects of repeated
measures to be modeled using the underlying covariance structure. GMM is a flexible modeling
approach that allows for different sub-models to be tested. As an example, the variance and
covariance estimates for the growth factors within each class can be fixed to zero (known as
latent class growth analysis), such that individual growth trajectories within a class are
homogeneous (Jung & Wickrama, 2008). These variances and covariances can be freed so that
intraindividual heterogeneity within each class can be modeled.
A generic model for each class specified in GMM is illustrated in Figure 2. Latent
variables estimated are the intercept (I) and slope (S), which will vary by class membership.
Together, I and S predict Loneliness Trajectory Class (LTC). Covariance between the intercept
and slope is fixed to 0. Manifest variables are UCLA Loneliness Scale score in 2008 (2008 L),
UCLA Loneliness Scale score in 2012 (2012 L), and UCLA Loneliness Scale score in 2016
(2016 L). The effect of the intercept on each manifest variable is fixed to 1, and the effect of the
slope on each manifest variable varies by person to take into account heterogeneity in age at
measurement at each wave. GMM with three, four, and five, and six latent classes will be tested.
Model selection will be determined by comparison of fit statistics (e.g., Akaike information
21
criterion (AIC), Bayesian information criterion (BIC), Log-Likelihood, and entropy) and
consideration of substantive meaning.

Figure 2: Growth Mixture Model (GMM)

To test the prediction that epigenetic aging is a mechanism by which loneliness has an
effect on dementia risk, we test a mediation model using the multicategorical variable –
loneliness latent classes (k) – as the IV , DNAm PhenoAge as the mediator, and dementia risk as
the DV (Figure 1B). Mediation modeling using the loneliness latent classes (k) was structured
following Hayes and Preacher (2014). First, k-1 indicator codes were created. Group four, the
low, small increasing group, was used as the reference group because it is the largest group and
22
likely represents the baseline in the population. Next, DNAm PhenoAge and TICSm scores were
regressed onto the indicator codes (L1-L4 in Figure 3), and TICSm score was regressed onto
DNAm PhenoAge. Relative indirect and total effects, and proportions were calculated as:

𝑖𝑛𝑑𝑖𝑟𝑒𝑐𝑡
.
= 𝑎
.
∗𝑏 𝑡𝑜𝑡𝑎𝑙
.
= 𝑐
.
+𝑎
.
∗𝑏 𝑝𝑟𝑜𝑝
.
=
./0.1234
"
45467
"

Next, we test the hypothesis that the relationship between DNAm PhenoAge and dementia risk is
bi-directional using Multivariate Analysis of Variance (MANOV A) to investigate the covariation.

23
Chatper 3: Results

3.1 Sample Descriptive Statistics

The final sample included 1814 people with an average age of 70, 1072 of whom are
female, and 742 of whom are male. The sample included 1402 white participants (77%), 282
Black participants (16%) and 130 participants who identified their race as “Other” (7%). Sample-
level and group-level descriptive statistics for loneliness scores, covariates, and dependent
variables are provided in Table 1.
Table 1: Sample Descriptive Statistics

Full Sample
M SD
Age (2016) 69.76 9.77
DNAm PhenoAge 57.71 10.06
TICSm 14.8 4.22
BMI 34.5 8.26
CES-D 1.69 2.07
PGS for Cognition 0.02 1
UCLA 8 16.5 4.7
UCLA 12 16.59 4.77
UCLA 16 16.92 4.81
N %
Sex
Female 1072 59.1
Male 742 40.9
Race

White 1402 77.29
Black 282 15.55
Other 130 7.17
Cohort 1 (1890-1923) 109 6.01
Cohort 2 (1924-1930) 518 28.56
Cohort 3 (1931-1941) 277 15.27
Cohort 4 (1942-1947) 384 21.17
Cohort 5 (1948-1953) 445 24.53
Cohort 6 (1954-1959) 64 3.53

24
Preliminary correlations provided initial support for investigating whether different
trajectories of loneliness moderate the associations between epigenetic age and dementia risk.
Table 2 presents the intercorrelations among key variables. The correlations between loneliness
across time points range from .60 to .65, such that higher loneliness is associated with higher
loneliness at a later time point, and a previous one. The consistent correlation across all
measurements suggests that scores are relatively stable over time. There are small correlations
observed between loneliness and epigenetic age, on average. There is a small negative
correlation observed between dementia risk and epigenetic age. Loneliness at each time point is
associated with dementia risk, although this effect is small. Before adjusting for covariates,
estimates of the correlation between DNAm PhenoAge and dementia is small in the sample (r =
-.28, 95% CI [-0.32, -0.24]) and small–medium across groups (r = -.16 – -.31, 95% CIs reported
in Table 8).

Table 2: Correlation Matrix of Independent and Dependent Variables

3.2 Baseline Meditation

We find that baseline loneliness (UCLA Loneliness Scale in 2008) is significantly
associated with dementia risk 𝛽 = −0.05, 95% CI[−0.10,−0.01],𝑝 < .05, and DNAm
PhenoAge 𝛽 = −0.03,95% CI [−0.05,−0.01],𝑝 < .05. DNAm PhenoAge was also associated
with dementia risk 𝛽 = −0.12, 95% CI[−0.21,−.02], 𝑝 < .05, such that higher epigenetic age
25
predicted lower TICSm scores, i.e., higher likelihood of dementia. The indirect effect of
loneliness on dementia risk as a result of DNAm PhenoAge (a*b) was not statistically significant
at the alpha cutoff value but trended toward statistical significance, 𝐵 =
0.003,95% CI [0,0.01],𝑝 = .09. The total effect of direct and indirect effects was estimated to
be 𝐵 = −0.05,95% CI [−0.10,−0.01],𝑝 < .05. Although DNAm PhenoAge does not mediate
the association between baseline loneliness and dementia risk at a statistically significant level,
there is some evidence that epigenetic age may be a plausible partial mediating mechanism.
Given these estimates are small, we next tested whether loneliness change over time provides
more meaning to these relationships.

3.3 Covariance Pattern Mixture Modeling and Growth Mixture Modeling

Model fit statistics for CPMM for each structure is provided in Table 3. No one class-
solution emerged as the best solution according to all metrics. The unstructured covariance
pattern had the largest -LL and lowest entropy, the toeplitz had the second lowest AIC, and
second lowest BIC, and the compound symmetric had the lowest AIC and lowest BIC. Priority
was given to AIC and BIC indices in model selection, and consideration of measurement
structure. Thus, a compound symmetric covariance structure (e.g., a, a, a) was used for Growth
Mixture Modeling. We tested three-to-six class unconditional (i.e., no covariates) growth mixture
models to determine the optimal number of latent classes. A five-class solution was selected as
the best model according to: lowest AIC and BIC values, smallest negative log-likelihood,
second lowest entropy, and interpretability (Table 4). While the three-class solution had lower
entropy, -LL, AIC, and BIC values were higher. Sattora-Bentler Scaled Chi-Square Difference
Tests were used to test differences between models. Each comparison model was significantly
26
different from the nested model. Latent class parameters for the five-class model are shown in
Table 5.

Table 3: Covariance Pattern Mixture Model Comparison Statistics

AIC BIC Entropy -LL
unstructured 22479.67 22655.78 0.62 -11207.84
toeplitz 22477.81 22626.4 0.61 -11211.91
compound symmetric 22474.92 22595.99 0.62 -11215.46

Table 4: Growth Mixture Model Comparison Statistics

Classes Entropy AIC BIC -Log-Likelihood Parameters Compared DLL Ddf p
3 0.57 22637 22714 -11304.77 14 -
4 0.63 22528 22627 -11246.39 18 3 106.83 4 0.000
5 0.62 22474 22595 -11215.46 22 4 77.46 4 0.000
6 0.65 22475 22602 -11431.56 26 5 324.66 4 0.000

Table 5: Latent Class Descriptions

Latent Class Description N Intercept Slope
1 Moderate 286 17.91 -0.16
2 Low 489 14.44 0.11
3 High Declining 155 25.42 -0.50
4 Low Increasing 539 12.83 0.21
5 Moderate Declining 345 21.39 -0.45
Note. Bolded values are statistically significant at the 0.05 level.

27
Table 6: Descriptive Statistics by Loneliness Latent Class

Moderate Low High Declining
Low
Increasing
Moderate
Declining
M SD M SD M SD M SD M SD
Age (2016) 69.88 10.32 70.51 9.7 67.96 9.67 69.89 9.64 69.18 9.57
DNAm PhenoAge 57.61 9.88 58.48 10.08 56.96 10.32 57.15 10.33 57.89 9.6
TICSm 14.43 4.43 14.9 4.28 13.91 4.03 15.55 3.92 14.21 4.27
BMI 34.1 8.21 34.41 8.06 35.81 9.98 33.88 7.63 35.34 8.59
CES-D 1.78 2.03 1.48 1.97 3.28 2.6 0.99 1.45 2.45 2.29
PGS for Cognition 0.22 1.09 -0.01 0.93 -0.2 1.04 -0.01 0.99 0.01 1.01
UCLA 8 17.93 2.91 14.46 2.63 25.02 2.8 12.65 1.74 21.06 3.34
UCLA 12 17.95 2.77 14.38 2.3 25.72 2.76 12.59 1.68 21.23 2.92
UCLA 16 17.96 2.95 15.2 2.68 25.78 2.91 12.66 1.77 21.25 2.95
N % N % N % N % N %
Sex
Female 168 58.74 268 54.81 88 56.77 341 63.27 207 60
Male 118 41.26 221 45.19 67 43.23 198 36.73 138 40
Race

White 222 77.62 381 77.91 113 72.9 433 80.33 253 73.33
Black 36 12.59 73 14.93 25 16.13 77 14.29 71 20.58
Other 28 9.79 35 7.16 17 10.97 29 5.38 21 6.09
Cohort 1 (1890-1923) 23 8.04 31 6.34 14 9.03 24 4.45 17 4.93
Cohort 2 (1924-1930) 82 28.67 145 29.65 30 19.35 164 30.43 97 28.12
Cohort 3 (1931-1941) 41 14.34 85 17.38 19 12.26 85 15.77 47 13.62
Cohort 4 (1942-1947) 53 18.53 103 21.06 37 23.87 109 20.22 82 23.77
Cohort 5 (1948-1953) 66 23.08 107 21.88 51 32.9 131 24.3 90 26.09
Cohort 6 (1954-1959) 18 6.29 13 2.66 4 2.58 19 3.53 10 2.9

3.4 Loneliness Latent Class Mediation

Next, to test the extent to which epigenetic aging mediates the relationship between
loneliness and dementia risk, we tested the relative indirect and total effects of high loneliness
and moderate loneliness classes, that is, the indirect and total effect of belonging to a high LTC
on the dependent variable, relative to the reference group, the largest group. As shown in Figure
3, DNAm PhenoAge was significantly associated with TICSm, B = -0.03, 𝛽 =
−0.12,95% CI [−0.22,−0.02] ,𝑝 < .05. DNAm PhenoAge did not significantly mediate the
association between any loneliness group and dementia risk (𝑝
.
𝑠 > .05). All total effects were
nonsignificant, however total effects approached significance in the moderate group (L1), 𝐵 =
28
−0.50, 95% CI [−1.02,0.02], and high declining group (L3), 𝐵 = −0.60,
95% CI [−1.26,0.07]. These total and direct effects in L1 and L3, in reference to group 4,
suggest that higher loneliness groups are associated with lower TICSm scores (i.e., higher
dementia risk). In summary, given the insignificant indirect effects, DNAm PhenoAge did not
significantly mediate the association between indicator codes (i.e., dummy coded loneliness
groups) and TICSm. Indirect effects accounts for 1% of the total effect for L1, 8% for L2, 0% for
L3, and 2% for L5. Parameter estimates of main effects, indirect effects, total effects, and
proportions are reported in Table 7.

29
Figure 3: Mediation Model and Parameter Estimates

Note. L1 = Moderate relative to Low Increasing, L2 = Low relative to Low Increasing, L3 = High
Declining relative to Low Increasing, L5 = Moderate Declining relative to Low Increasing. ** p < .05, *
p < .1

Table 7: Mediation Parameter Estimates, Indirect, Total, and Proportional Effects
Model Estimates Relative Indirect Effect Relative Total Effect Proportion I/D

a (SE) b (SE) c (SE) B 95%CI B 95%CI B 95%CI
L1 -0.26 (0.51) -0.03 (0.01)** -0.51 (0.26)* 0.007 [-0.03, 0.05] -0.5 [-1.27, -0.18] -0.02 [-0.07, 0.05]
L2 0.47 (0.43) -0.03 (0.01)** -0.07 (0.22) -0.01 [-0.04, 0.01] -0.08 [-0.78, 0.15] 0.16 [-0.73, 1.05]
L3 -0.16 (0.65) -0.03 (0.01)** -0.60 (0.34)* 0.004 [-0.03, 0.04] -0.6 [-1.26, 0.07] -0.01 [-0.07, 0.05]
L5 0.31 (0.52) -0.03 (0.01)** -0.31 (0.25) -0.01 [-0.04, 0.02] -0.32 [-0.82, 0.18] 0.03 [-0.07, 0.12]
Note. Estimates are unstandardized. ** p < .05, * p < .1

30
3.5 Moderation
Multivariate Analysis of Variance (MANOV A) and post-hoc t-tests were conducted to test
between-group differences in dementia risk scores and DNAm PhenoAge. TICSm scores were
statistically significantly worse (i.e., lower) for the high-declining compared to the low-
increasing (𝐵 = −1.68,95% CI [−1.03,−2.33]), and the moderate group was lower than the
low-increasing group (𝐵 = −1.04,95% CI [−1.59,−0.50]), and the moderate declining group
was lower than the low group (𝐵 = −1.27,95% CI [−1.77,−0.76]), (𝑝
.
𝑠 > .05, Bonferroni
corrected for multiple comparisons). No significant differences in PhenoAge or the correlation
between PhenoAge and TICSm were found between any pair of LTC groups. After adjusting for
covariates (Table 8), LTC group did not significantly predict differences in DNAm PhenoAge or
dementia risk. Chi-square difference tests indicated that the full model significantly modeled the
data better than the static model, and the static significantly modeled the data better than the
unconstrained model (𝑝
.
𝑠 > .05). Indeed, the correlation between DNAm PhenoAge was
reduced in magnitude and no longer statistically significant from zero in all groups except for the
low loneliness group. As depicted in Figure 4, correlations between DNAm PhenoAge and
dementia risk for each LTC group in the unconstrained model were much larger (in blue) than in
the full model which adjusted for covariates (in red). Results suggest that loneliness latent
classes do not differentially affect the correlation between epigenetic aging and dementia risk,
after controlling for relevant health variables, demographics, and other psychosocial risk factors
(i.e., social isolation, depression).
Due to the known high correlation between loneliness and depressive symptomatology,
additional models were run with and without depression to investigate whether depression is
driving effects. Appendix B shows the parameter estimates of these models, which indicate that
controlling for depression does not significantly change parameter estimates.
31

Table 8: Multivariate Analysis of Variance Model Results

32
Figure 4: Forest Plot of Association of DNAm PhenoAge and Dementia Risk in Each Loneliness
Group Between the Unconstrained and Full Model

33
Chapter 4: Discussion

Understanding how loneliness correlates with dementia risk has been a driving question in
the literature. A major effort in tackling that question is testing whether physiological
dysregulation is the primary mechanism. The present study adds to the literature on the
physiological dysregulation hypothesis of loneliness by testing associations between a marker of
physiological functioning, epigenetic age, with loneliness and dementia risk.

4.1 Epigenetic Age as a Mediator of Loneliness on Dementia Risk

We found that epigenetic age did not mediate the relationship between loneliness and
dementia risk in the baseline loneliness model, or the latent class model. This finding implies
that loneliness in older adulthood may not affect the epigenome, and thus may not be the
mechanism by which loneliness affects dementia risk. Secondly, the present study found little
evidence to support the hypothesis that there are different associations between epigenetic age
and different patterns of longitudinal loneliness, and thus questions the physiological
mechanisms of loneliness on dementia risk.
As found in previous studies on loneliness and cognition (Kim et al., 2020) and dementia
(Kim et al., 2021), physiological mechanisms do not appear to confound or mediate the
association between loneliness and dementia. Although the initial unconstrained MANOVA
indicated that there appeared to be different associations between epigenetic age and dementia
risk between loneliness latent classes, the effect disappeared after accounting for related lifestyle
factors that also affect epigenetic age and dementia risk. Further, results of the mediation
analyses suggest that epigenome-wide aging is unlikely to mediate the relationship between
longitudinal or baseline loneliness and dementia risk once covariates are included. Thus, other
34
areas of functioning may be critical to examine in the relationship between loneliness and
dementia. For example, in line with the health risk behavior hypothesis, lifestyle decisions such
as diet and exercise may be more critical to the effect of loneliness on dementia risk.
Notably, in the low group (Table 8, Model 3), dementia risk and epigenetic age were
significantly correlated in the full model. The low group is larger than the high and moderate
groups – thus, it, along with the low increasing group, likely represent the population. This
finding indicates that the general population of older adults is low in loneliness and among this
group, epigenetic age is associated with dementia risk a small degree, based on the small Pearson
r correlation estimate observed in the fully adjusted MANOVA model. The correlation is
negative – which means that higher epigenetic age is related to higher dementia risk.
The current paper cannot fully provide evidence to disprove the physiological dysregulation
hypothesis of loneliness, however. Limitations of the measure of physiological dysregulation,
DNAm PhenoAge, and the timing of epigenetic age measurement, in addressing this hypothesis
cannot be discounted. For example, stochastic epigenetic changes in aging, or epigenetic changes
unrelated to loneliness, may have interfered with detecting an effect in DNAm PhenoAge in the
sample. CpG sites may have become methylated during aging in the sample for reasons unrelated
to loneliness that were not measured in the current study (e.g., diet changes, environmental
exposures) and this may have interacted with an effect of loneliness on DNAm PhenoAge.
Additionally, DNAm PhenoAge was measured at the same time as the third timepoint of
loneliness measurement, and dementia risk measurement, thus DNAm PhenoAge does not fully
represent a mediating mechanism. Lastly, epigenetic age was lower than chronological age in the
sample (Table 6), indicating that this sample is relatively healthy. This sample may not be
35
representative of the population’s epigenetic age, dementia risk, and loneliness, and thus may not
have had the variability to capture an effect of epigenetic age as a mediating mechanism.

4.2 Loneliness and Dementia Risk

We replicated previous studies, including those that used HRS data, in finding that baseline
levels of loneliness predicted differences in dementia risk in older adults (Wilson et al., 2007;
Victor et al., 2021). The present study extended scientific knowledge by showing that
longitudinal loneliness change, represented through latent classes, is also associated with
dementia risk in the hypothesized direction. This finding is consistent with findings from Akhter-
Khan et al (2021) which found that persistent loneliness conferred higher risk of dementia onset,
and differs from Kim et al. (2021) which found that increases in loneliness was not associated
with higher dementia risk. The present study was unique in using growth mixture modeling to
characterize longitudinal loneliness. High and moderate loneliness classes were significantly
differentially associated with dementia risk, compared to the low increasing lonely class, in the
negative direction. This suggests that high or moderate loneliness – regardless of slope – during
an 8-year span in older adulthood is a risk factor for dementia. Further, given the differential
associations, the present study supports the use of growth mixture modeling in studying dynamic
psychosocial risk factors on dementia risk in older adulthood.

4.3 Epigenetic Age and Dementia Risk

DNAm PhenoAge was found to be significantly associated with dementia risk. This finding
extends previous research showing that other second-generation clocks are predictive of
cognitive decline (Hillary et al., 2021; Wu et al., 2020). Specifically, this study found that the
36
epigenetic clock PhenoAge clock is a potential marker of dementia risk, adding to research
indicating other blood-based epigenetic clocks may be predictive of neuropathology.
Importantly, this study provides evidence for only DNAm PhenoAge, and the finding cannot be
extended to other second-generation clocks, because each epigenetic clock is unique in its
predictive capacities given the unique CpG sites selected for the clock. This study provides
evidence that epigenetic aging, as measured by DNAm PhenoAge, may underlie later-onset
dementia. Given there is a paucity of reliable blood-based biomarkers for dementia, this finding
is critical, and future research should replicate this finding in other samples.

4.4 Limitations and Future Directions

The present study has several limitations. First, we used one measure of physiological
functioning DNAm PhenoAge, which was developed for general age-related conditions. There
are other clocks that might be more useful for understanding the association between loneliness
and dementia risk. Second, the timeline of measurements poses a threat to internal validity.
Epigenetic age is not measured at each wave, therefore directionality between psychological
experiences and biological mechanisms cannot be concluded. Interpretations from the present
study are limited because it did not include measures of epigenetic age at each time point to fully
test a mediation mechanism. Third, missing data on the UCLA Loneliness Scale is a concern for
internal validity as well because variables of interest were related to missingness. Our analyses,
however, included covariates of missingness that decreased potential bias in parameter estimates
associated with systematic missingness. Fourth, the diagnoses of dementia are not made in the
HRS Base samples, so the dementia risk variable is a proxy measure of dementia based on
37
cognitive performance only. However, Crimmins et al. (2011) showed that TICSm accurately
classified participants as cognitively normal, dementia, or CIND 69.2% of the time.
Physiological dysregulation, as indexed by one epigenetic age measure, was not found to
mediate the effects of loneliness on dementia risk. Additionally, loneliness was unrelated to the
association between epigenetic age and dementia once sex, education, race, cohort, baseline age,
current depression, body mass index, current smoking level, self-rated health, objective social
isolation, and APOE ε4 allele count were included in the model. As chronic loneliness and social
disconnection is known to lead to altered transcription processes (Cole et al. 2007), future studies
should replicate current study results before drawing firm conclusions about the utility of
epigenetic age as a mechanism for understanding the association between loneliness and
dementia. Indeed, there are numerous epigenetic clocks that could be considered but were not in
the current study.
Future studies should determine the directionality between physiological markers of
aging using repeated measures of both epigenetic age and psychosocial factors. The amount of
time by which psychosocial factors take to impact the epigenome is unknown, and may vary
across the lifespan, thus future studies should measure epigenetic changes in small intervals to
determine the optimal measurement window to detect changes.

38
References

Adam, E. K., Hawkley, L. C., Kudielka, B. M., & Cacioppo, J. T. (2006). Day-to-day dynamics
of experience--cortisol associations in a population-based sample of older
adults. Proceedings of the National Academy of Sciences of the United States of
America, 103(45), 17058–17063. https://doi.org/10.1073/pnas.0605053103.

Akhter-Khan, S. C., Tao, Q., Ang, T., Itchapurapu, I. S., Alosco, M. L., Mez, J., Piers, R. J.,
Steffens, D. C., Au, R., & Qiu, W. Q. (2021). Associations of loneliness with risk of
Alzheimer's disease dementia in the Framingham Heart Study. Alzheimer's & dementia :
the journal of the Alzheimer's Association, 17(10), 1619–1627.
https://doi.org/10.1002/alz.12327.

American Psychiatric Association. (2013). Diagnostic and statistical manual of mental
disorders (5th ed.). https://doi.org/10.1176/appi.books.9780890425596.

Andrews, S. J., Fulton-Howard, B., & Goate, A. (2020). Interpretation of risk loci from genome-
wide association studies of Alzheimer's disease. The Lancet. Neurology, 19(4), 326–335.
https://doi.org/10.1016/S1474-4422(19)30435-1.

Agarwal, P., Chung, P., Heberlein, U., & Kent, C. (2018). Social isolation-induced epigenetic
and transcriptional changes in Drosophila dopaminergic neurons. BioRxiv, 1–52.
https://doi.org/10.1101/443226.

Ayalon, L., Shiovitz-Ezra, S., & Roziner, I. (2016). A cross-lagged model of the reciprocal
associations of loneliness and memory functioning. Psychology and Aging, 31(3), 255–
261. https://doi.org/10.1037/pag0000075.

Austin, M. K., Chen, E., Ross, K. M., McEwen, L. M., Maclsaac, J. L., Kobor, M. S., et al.
(2018). Early-life socioeconomic disadvantage, not current, predicts accelerated epige-
netic aging of monocytes. Psychoneuroendocrinology, 97(January), 131–134. https://doi.
org/10.1016/j.psyneuen.2018.07.007.

Beam, C. R., Kaneshiro, C., Jang, J. Y., Reynolds, C. A., Pedersen, N. L., & Gatz, M. (2020). A
Twin Study of Sex Differences in Genetic Risk for All Dementia, Alzheimer's Disease
(AD), and Non-AD Dementia. Journal of Alzheimer's disease : JAD, 76(2), 539–551.
https://doi.org/10.3233/JAD-191192

Beam, C. R. and Kim, A. J. (2020). Psychological sequelae of social isolation and loneliness
might be a larger problem in young adults than older adults. Psychological Trauma:
Theory, Research, Practice, and Policy, 12, S58.

Bell, C. G., Lowe, R., Adams, P. D., Baccarelli, A. A., Beck, S., Bell, J. T., Christensen, B. C.,
Gladyshev, V. N., Heijmans, B. T., Horvath, S., Ideker, T., Issa, J. P. J., Kelsey, K. T.,
Marioni, R. E., Reik, W., Relton, C. L., Schalkwyk, L. C., Teschendorff, A. E., Wagner,
W., … Rakyan, V. K. (2019). DNA methylation aging clocks: Challenges and
39
recommendations. Genome Biology, 20(1), 1–24. https://doi.org/10.1186/s13059-019
1824-y.

Bergsma, T., & Rogaeva, E. (2020). DNA Methylation Clocks and Their Predictive Capacity for
Aging Phenotypes and Healthspan. Neuroscience
Insights. https://doi.org/10.1177/2633105520942221.

Boomsma, D. I. (2013). Twin, association and current “omics” studies. Journal of Maternal-
Fetal & Neonatal Medicine, 26(S2), 9–12. doi:10.3109/14767058.2013.830405.

Boss, L., Kang, D. H., & Branson, S. (2015). Loneliness and cognitive function in the older
adult: a systematic review. International psychogeriatrics, 27(4), 541–553.
https://doi.org/10.1017/S1041610214002749.

Brandt, J., Spencer, M, Folstein, M. (1988) The Telephone Interview for Cognitive
Status. Neuropsychiatry Neuropsychol Behav Neurol, 1, 111–117.

Brody, G. H., Miller, G. E., Yu, T., Beach, S. R. H., & Chen, E. (2016). Supportive family
environments ameliorate the link between racial discrimination and epigenetic aging: A
replication across two longitudinal cohorts. Psychological Science, 27(4), 530–541.
https://doi.org/10.1177/0956797615626703.

Cacioppo, J. T., Hawkley, L. C., Crawford, L. E., Ernst, J. M., Burleson, M. H., Kowalewski, R.
B., Malarkey, W. B., Van Cauter, E., & Berntson, G. G. (2002). Loneliness and health:
potential mechanisms. Psychosomatic medicine, 64(3), 407–417.
https://doi.org/10.1097/00006842-200205000-00005.

Cacioppo, J. T., Hawkley, L. C., Ernst, J. M., Burleson, M., Berntson, G. G., Nouriani, B., &
Spiegel, D. (2006). Loneliness within a nomological net: An evolutionary perspective.
Journal of Research in Personality, 40, 1054–1085. doi:10.1016/j.jrp.2005.11.007.

Cacioppo, J. T., Hawkley, L. C., & Thisted, R. A. (2010). Perceived social isolation makes me
sad: 5-year cross-lagged analyses of loneliness and depressive symptomatology in the
Chicago Health, Aging, and Social Relations Study. Psychology and aging, 25(2), 453–
463. https://doi.org/10.1037/a0017216.

Carstensen, L.L. (1992) Social and emotional patterns in adulthood: support for socioemotional
selectivity theory, Psychological Aging, 7(3), 331-338.

Chao, L. L., Buckley, S. T., Kornak, J., Schuff, N., Madison, C., Yaffe, K., Miller, B. L.,
Kramer, J. H., & Weiner, M. W. (2010). ASL perfusion MRI predicts cognitive decline
and conversion from MCI to dementia. Alzheimer Disease and Associated Disorders,
24(1), 19–27. https://doi.org/10.1097/WAD.0b013e3181b4f736.

Chen, B. H., Marioni, R. E., Colicino, E., Peters, M. J., Ward-Caviness, C. K., Tsai, P. C., …
40
Horvath, S. (2016). DNA methylation-based measures of biological age: Meta-analysis
predicting time to death. Aging, 8(9), 1844–1865.

Chouliaras, L., Rutten, B. P., Kenis, G., Peerbooms, O., Visser, P. J., Verhey, F., van Os, J.,
Steinbusch, H. W., & van den Hove, D. L. (2010). Epigenetic regulation in the
pathophysiology of Alzheimer's disease. Progress in neurobiology, 90(4), 498–510.
https://doi.org/10.1016/j.pneurobio.2010.01.002.

Christensen, B. C., Houseman, E. A., Marsit, C. J., Zheng, S., Wrensch, M. R., Wiemels, J. L.,
Nelson, H. H., Karagas, M. R., Padbury, J. F., Bueno, R., Sugarbaker, D. J., Yeh, R. F.,
Wiencke, J. K., & Kelsey, K. T. (2009). Aging and environmental exposures alter tissue-
specific DNA methylation dependent upon CpG island context. PLoS genetics, 5(8),
e1000602. https://doi.org/10.1371/journal.pgen.1000602.

Cole, S. W., Hawkley, L. C., Arevalo, J. M., Sung, C. Y., Rose, R. M., & Cacioppo, J. T. (2007).
Social regulation of gene expression in human leukocytes. Genome Biology, 8(9).
https://doi.org/10.1186/gb-2007-8-9-r189.

Cook, S. E., Marsiske, M., & McCoy, K. J. (2009). The use of the Modified Telephone Interview
for Cognitive Status (TICS-M) in the detection of amnestic mild cognitive
impairment. Journal of geriatric psychiatry and neurology, 22(2), 103–109.
https://doi.org/10.1177/0891988708328214.

Cornwell, E. Y., & Waite, L. J. (2009). Social disconnectedness, perceived isolation, and health
among older adults. Journal of health and social behavior, 50(1), 31–48.
https://doi.org/10.1177/002214650905000103.

Coyle, C. E., & Dugan, E. (2012). Social isolation, loneliness and health among older
adults. Journal of aging and health, 24(8), 1346–1363.
https://doi.org/10.1177/0898264312460275.

Crimmins, E.M., Kim, J.K., Langa, K.M., & Weir, D.R. (2011). Assessment of cognition using
surveys and neuropsychological assessment: the health and retirement study and the
aging, demographics, and memory study. The Journals of Gerontology, Series B:
Psychological Sciences and Social Sciences, 66B(S1), i162–i171,
doi:10.1093/geronb/gbr048.

DeNiro D. A. (1995). Perceived alienation in individuals with residual-type schizophrenia. Issues
in mental health nursing, 16(3), 185–200. https://doi.org/10.3109/01612849509006934.

Epel, E., Daubenmier, J., Moskowitz, J. T., Folkman, S., & Blackburn, E. (2009). Can meditation
slow rate of cellular aging? Cognitive stress, mindfulness, and telomeres. Annals of the
New York Academy of Sciences, 1172, 34–53. https://doi.org/10.1111/j.1749-
6632.2009.04414.x.

Gallardo, G., & Holtzman, D. M. (2019). Amyloid-β and Tau at the Crossroads of Alzheimer's
41
Disease. Advances in experimental medicine and biology, 1184, 187–203.
https://doi.org/10.1007/978-981-32-9358-8_16.

Gassen, N. C., Chrousos, G. P., Binder, E. B., & Zannas, A. S. (2016). Life stress, glucocor-
ticoid signaling, and the aging epigenome: Implications for aging-related diseases.
Neuroscience and Biobehavioral Reviews, 74(Pt. B), 356–365. https://doi.org/10.1016/j.
neubiorev.2016.06.003.

Gatz, M., Reynolds, C. A., Fratiglioni, L., Johansson, B., Mortimer, J. A., Berg, S., Fiske, A., &
Pedersen, N. L. (2006). Role of genes and environments for explaining Alzheimer
disease. Archives of general psychiatry, 63(2), 168–174.
https://doi.org/10.1001/archpsyc.63.2.168

Goossens, L., van Roekel, E., Verhagen, M., Cacioppo, J. T., Cacioppo, S., Maes, M., &
Boomsma, D. I. (2015). The Genetics of Loneliness: Linking Evolutionary Theory to
Genome-Wide Genetics, Epigenetics, and Social Science. Perspectives on Psychological
Science, 10(2), 213–226. https://doi.org/10.1177/1745691614564878.

Grimby A. (1993). Bereavement among elderly people: grief reactions, post-bereavement
hallucinations and quality of life. Acta psychiatrica Scandinavica, 87(1), 72–80.
https://doi.org/10.1111/j.1600-0447.1993.tb03332.x.

Gross, A., Chu, N., Carlson, M. C., & Xue, Q. (2018). Role of Multisystem Physiological
Dysregulation in Cognitive Impairment and Frailty. Innovation in Aging, 2(Suppl 1), 23.
https://doi.org/10.1093/geroni/igy023.084.

Hackett, R. A., Hamer, M., Endrighi, R., Brydon, L., & Steptoe, A. (2012). Loneliness and
stress-related inflammatory and neuroendocrine responses in older men and
women. Psychoneuroendocrinology, 37(11), 1801–1809.
https://doi.org/10.1016/j.psyneuen.2012.03.016.

Hawkley, L. C., & Cacioppo, J. T. (2007). Aging and Loneliness: Downhill Quickly? Current
Directions in Psychological Science, 16(4), 187–191. https://doi.org/10.1111/j.1467-
8721.2007.00501.x.

Hawkley, L. C., & Cacioppo, J. T. (2010). Loneliness matters: a theoretical and empirical review
of consequences and mechanisms. Annals of behavioral medicine : a publication of the
Society of Behavioral Medicine, 40(2), 218–227. https://doi.org/10.1007/s12160-010-
9210-8.

Hawkley, L. C., Thisted, R. A., & Cacioppo, J. T. (2009). Loneliness predicts reduced physical
activity: cross-sectional & longitudinal analyses. Health psychology: official journal of
the Division of Health Psychology, American Psychological Association, 28(3), 354–
363. https://doi.org/10.1037/a0014400.

Hawkley, L. C., & Cacioppo, J. T. (2013). Social connectedness and health. In C. Hazan & M. I.
42
Campa (Eds.), Human bonding: The science of affectional ties (p. 343–364). The
Guilford Press.

Hayes, A. F., & Preacher, K. J. (2014). Statistical mediation analysis with a multicategorical
independent variable. The British journal of mathematical and statistical
psychology, 67(3), 451–470. https://doi.org/10.1111/bmsp.12028

Health and Retirement Study, (2021). HRS CORE public use dataset. Produced and distributed
by the University of Michigan with funding from the National Institute on Aging (grant
number NIA U01AG009740). Ann Arbor, MI.

Health and Retirement Study, (2021). Epigenetic Clock restricted use dataset. Produced and
distributed by the University of Michigan with funding from the National Institute on
Aging (grant number NIA U01AG009740). Ann Arbor, MI.

Hillary, R. F., Stevenson, A. J., Cox, S. R., McCartney, D. L., Harris, S. E., Seeboth, A.,
Higham, J., Sproul, D., Taylor, A. M., Redmond, P., Corley, J., Pattie, A., Hernández,
M., Muñoz-Maniega, S., Bastin, M. E., Wardlaw, J. M., Horvath, S., Ritchie, C. W.,
Spires-Jones, T. L., McIntosh, A. M., … Marioni, R. E. (2021). An epigenetic predictor
of death captures multi-modal measures of brain health. Molecular psychiatry, 26(8),
3806–3816. https://doi.org/10.1038/s41380-019-0616-9

Holwerda, T. J., Deeg, D. J., Beekman, A. T., van Tilburg, T. G., Stek, M. L., Jonker, C., &
Schoevers, R. A. (2014). Feelings of loneliness, but not social isolation, predict dementia
onset: results from the Amsterdam Study of the Elderly (AMSTEL). Journal of
neurology, neurosurgery, and psychiatry, 85(2), 135–142. https://doi.org/10.1136/jnnp-
2012-302755.

Hopwood, C. J., Donnellan, M. B., Blonigen, D. M., Krueger, R. F., McGue, M., Iacono, W. G.,
& Burt, S. A. (2011). Genetic and environmental influences on personality trait stability
and growth during the transition to adulthood: a three-wave longitudinal study. Journal of
personality and social psychology, 100(3), 545–556. https://doi.org/10.1037/a0022409.

Horvath S. (2013). DNA methylation age of human tissues and cell types. Genome
biology, 14(10), R115. https://doi.org/10.1186/gb-2013-14-10-r115.

Iturria-Medina, Y., Sotero, R. C., Toussaint, P. J., Mateos-Pérez, J. M., Evans, A. C., Weiner, M.
W., et al. (2016). Early role of vascular dysregulation on late-onset Alzheimer’s disease
based on multifactorial data-driven analysis. Nat. Commun, 7, 11934. doi:
10.1038/ncomms11934.

Jung, T., & Wickrama, K. A. S. (2008). An Introduction to Latent Class Growth Analysis and
Growth Mixture Modeling. Social and Personality Psychology Compass, 2(1), 302–317.
https://doi.org/10.1111/j.1751-9004.2007.00054.x.

Kalaria, R. N., Akinyemi, R., & Ihara, M. (2012). Does vascular pathology contribute to
43
Alzheimer changes?. Journal of the neurological sciences, 322(1-2), 141–147.
https://doi.org/10.1016/j.jns.2012.07.032.

Kim, A. J., Beam, C. R., Greenberg, N. E., and Burke, S. L. (2020). Health factors as potential
mediators of the longitudinal effect of loneliness on general cognitive ability. The
American Journal of Geriatric Psychiatry 28, 1272–1283.

Kim, A. J., Gold, A. I., Fenton, L., Pilgrim, M., Lynch, M., Climer, C. R., Penichet, E. N., Kam,
A., & Beam, C. R. (2021). A Genetically Informed Longitudinal Study of Loneliness and
Dementia Risk in Older Adults. Frontiers in genetics, 12, 661474.
https://doi.org/10.3389/fgene.2021.661474.

Kochenderfer-Ladd, B., & Wardrop, J. L. (2001). Chronicity and instability of children's peer
victimization experiences as predictors of loneliness and social satisfaction
trajectories. Child development, 72(1), 134–151. https://doi.org/10.1111/1467-
8624.00270.
Lara, E., Martín-María, N., De la Torre-Luque, A., Koyanagi, A., Vancampfort, D., Izquierdo,
A., & Miret, M. (2019). Does loneliness contribute to mild cognitive impairment and
dementia? A systematic review and meta-analysis of longitudinal studies. Ageing
Research Reviews, 52(March), 7–16. https://doi.org/10.1016/j.arr.2019.03.002.

Levine, S., & Mody, T. (2003). The long-term psychobiological consequences of intermittent
postnatal separation in the squirrel monkey. Neuroscience and Biobehavioral Reviews,
27(1-2), 83–89. https://doi.org/10.1016/S0149-7634(03)00011-3.

Levine, M. E., & Crimmins, E. M. (2012). Sarcopenic obesity and cognitive functioning: The
mediating roles of insulin resistance and inflammation? Current Gerontology and
Geriatrics Research, 2012. https://doi.org/10.1155/2012/826398.

Levine, M. E., Lu, A. T., Bennett, D. A., & Horvath, S. (2015). Epigenetic age of the pre-frontal
cortex is associated with neuritic plaques, amyloid load, and Alzheimer’s disease related
cognitive functioning. Aging, 7(12), 1198–1211.

Levine, M. E., Lu, A. T., Quach, A., Chen, B. H., Assimes, T. L., Bandinelli, S., Hou, L.,
Baccarelli, A. A., Stewart, J. D., Li, Y., Whitsel, E. A., Wilson, J. G., Reiner, A. P., Aviv,
A., Lohman, K., Liu, Y., Ferrucci, L., & Horvath, S. (2018). An epigenetic biomarker of
aging for lifespan and healthspan. BioRxiv, 10(4), 573–591.
https://doi.org/10.1101/276162.

Marchesi V. T. (2011). Alzheimer's dementia begins as a disease of small blood vessels,
damaged by oxidative-induced inflammation and dysregulated amyloid metabolism:
implications for early detection and therapy. FASEB journal: official publication of the
Federation of American Societies for Experimental Biology, 25(1), 5–13.
https://doi.org/10.1096/fj.11-0102ufm.

Marioni, R. E., Shah, S., McRae, A. F., Chen, B. H., Colicino, E., Harris, S. E., Gibson, J.,
44
Henders, A. K., Redmond, P., Cox, S. R., Pattie, A., Corley, J., Murphy, L., Martin, N.
G., Montgomery, G. W., Feinberg, A. P., Fallin, M. D., Multhaup, M. L., Jaffe, A. E.,
Joehanes, R., … Deary, I. J. (2015). DNA methylation age of blood predicts all-cause
mortality in later life. Genome biology, 16(1), 25. https://doi.org/10.1186/s13059-015-
0584-6.

Maslow, A. H. (1943). A theory of human motivation. Psychological Review, 50(4), 370–
396. https://doi.org/10.1037/h0054346.

Matthews, T., Danese, A., Wertz, J., Odgers, C. L., Ambler, A., Moffitt, T. E., & Arseneault, L.
(2016). Social isolation, loneliness and depression in young adulthood: a behavioural
genetic analysis. Social psychiatry and psychiatric epidemiology, 51(3), 339–348.
https://doi.org/10.1007/s00127-016-1178-7.

McNeish, D., Harring, J. (2020). Covariance pattern mixture models: Eliminating random effects
to improve convergence and performance. Behav Res, 52, 947–979.
https://doi.org/10.3758/s13428-019-01292-4.

Monroe, S. M., & Harkness, K. L. (2005). Life Stress, the "Kindling" Hypothesis, and the
Recurrence of Depression: Considerations From a Life Stress Perspective. Psychological
Review, 112(2), 417–445. https://doi.org/10.1037/0033-295X.112.2.417.

Mullington, J.M, Haack, M., Toth, M., Serrador, J.M, Meier-Ewert, H.K. (2009). Cardiovascular,
inflammatory, and metabolic consequences of sleep deprivation. Prog Cardiovasc Dis,
51(4), 294-302.

Muthén, L. K., & Muthén, B. O. (2017). Mplus: Statistical Analysis with Latent Variables:
User's Guide (Version 8). Los Angeles, CA.

O'Luanaigh, C., O'Connell, H., Chin, A. V., Hamilton, F., Coen, R., Walsh, C., Walsh, J. B.,
Coakley, D., Molloy, A., Scott, J., Cunningham, C. J., & Lawlor, B. A. (2012).
Loneliness and vascular biomarkers: the Dublin Healthy Ageing Study. International
journal of geriatric psychiatry, 27(1), 83–88. https://doi.org/10.1002/gps.2695

Pace, T. W., Hu, F., & Miller, A. H. (2007). Cytokine-effects on glucocorticoid receptor
function: relevance to glucocorticoid resistance and the pathophysiology and treatment of
major depression. Brain, behavior, and immunity, 21(1), 9–19.
https://doi.org/10.1016/j.bbi.2006.08.009

Penninkilampi, R., Casey, A. N., Singh, M. F., & Brodaty, H. (2018). The Association between
Social Engagement, Loneliness, and Risk of Dementia: A Systematic Review and Meta-
Analysis. Journal of Alzheimer's disease : JAD, 66(4), 1619–1633.
https://doi.org/10.3233/JAD-180439.

45
Perlman, D., & Peplau, L. A. (1981). Toward a Social Psychology of Loneliness. In R. Gilmour,
& S. Duck (Eds.), Personal Relationships: 3. Relationships in Disorder (pp. 31-56).
London: Academic Press.

Petrovitch, H., Ross, G. W., Steinhorn, S. C., Abbott, R. D., Markesbery, W., Davis, D., Nelson,
J., Hardman, J., Masaki, K., Vogt, M. R., Launer, L., & White, L. R. (2005). AD lesions
and infarcts in demented and non-demented Japanese-American men. Annals of
neurology, 57(1), 98–103. https://doi.org/10.1002/ana.20318.

Pinquart, M., & Sörensen, S. (2001). Gender differences in self-concept and psychological well-
being in old age: a meta-analysis. The journals of gerontology. Series B, Psychological
sciences and social sciences, 56(4), P195–P213.
https://doi.org/10.1093/geronb/56.4.p195.

Post, R. M. (1992). Transduction of psychosocial stress into the neurobiology of recurrent
affective disorder. American Journal of Psychiatry, 149, 999–1010.

Ram, N., & Grimm, K. J. (2009). Growth Mixture Modeling: A Method for Identifying
Differences in Longitudinal Change Among Unobserved Groups. International journal of
behavioral development, 33(6), 565–576. https://doi.org/10.1177/0165025409343765.

R Core Team (2020). R: A language and environment for statistical computing. R Foundation for
Statistical Computing, Vienna, Austria. URL https://www.R-project.org/.

Renshaw, P. D., & Brown, P. J. (1993). Loneliness in middle childhood: Concurrent and
longitudinal predictors. Child Development, 64(4), 1271–
1284. https://doi.org/10.2307/1131339.

Russell, D. (1996). UCLA Loneliness Scale (Version 3): Reliability, validity, and factor
structure. Journal of Personality Assessment, 66, 20-40.

Shankar, A., McMunn, A., Banks, J., Steptoe, A. (2011). Loneliness, social isolation, and
behavioral and biological health indicators in older adults. Health Psychol, 30(4), 377-
385. doi: 10.1037/a0022826. PMID: 21534675.

Shiovitz-Ezra, S., & Ayalon, L. (2010). Situational versus chronic loneliness as risk factors for
all-cause mortality. International psychogeriatrics, 22(3), 455–462.
https://doi.org/10.1017/S1041610209991426.

Sibbett, R. A., Altschul, D. M., Marioni, R. E., Deary, I. J., Starr, J. M., & Russ, T. C. (2020).
DNA methylation-based measures of accelerated biological ageing and the risk of
dementia in the oldest-old: A study of the Lothian Birth Cohort 1921. BMC Psychiatry,
20(1), 1–15. https://doi.org/10.1186/s12888-020-2469-9

Siuda, D., Wu, Z., Chen, Y., Guo, L., Linke, M., Zechner, U., Xia, N., Reifenberg, G., Kleinert,
H., Forstermann, U., & Li, H. (2014). Social isolation-induced epigenetic changes in
46
midbrain of adult mice. Journal of physiology and pharmacology : an official journal of
the Polish Physiological Society, 65(2), 247–255.

Solmi, M., Veronese, N., Galvano, D., Favaro, A., Ostinelli, E. G., Noventa, V., Favaretto, E.,
Tudor, F., Finessi, M., Shin, J. I., Smith, L., Koyanagi, A., Cester, A., Bolzetta, F.,
Cotroneo, A., Maggi, S., Demurtas, J., De Leo, D., & Trabucchi, M. (2020). Factors
Associated With Loneliness: An Umbrella Review Of Observational Studies. Journal of
affective disorders, 271, 131–138. https://doi.org/10.1016/j.jad.2020.03.075.

Steptoe, A., Owen, N., Kunz-Ebrecht, S. R., & Brydon, L. (2004). Loneliness and
neuroendocrine, cardiovascular, and inflammatory stress responses in middle-aged men
and women. Psychoneuroendocrinology, 29(5), 593–611. https://doi.org/10.1016/S0306-
4530(03)00086-6.

Steptoe, A., Shankar, A., Demakakos, P., & Wardle, J. (2013). Social isolation, loneliness, and
all-cause mortality in older men and women. Proceedings of the National Academy of
Sciences of the United States of America, 110(15), 5797–5801.
https://doi.org/10.1073/pnas.1219686110.

Stravynski, A., & Boyer, R. (2001). Loneliness in relation to suicide ideation and parasuicide: a
population-wide study. Suicide & life-threatening behavior, 31(1), 32–40.
https://doi.org/10.1521/suli.31.1.32.21312.

Stroud, C. (2018). The Stress Sensitization Model. The Oxford Handbook of Stress and Mental
Health. DOI: 10.1093/oxfordhb/9780190681777.013.16.

Sushma, & Mondal, A. C. (2019). Role of GPCR signaling and calcium dysregulation in
Alzheimer's disease. Molecular and cellular neurosciences, 101, 103414.
https://doi.org/10.1016/j.mcn.2019.103414.

Sumner, J. A., Colich, N. L., Uddin, M., Armstrong, D., & McLaughlin, K. A. (2019). Early
experiences of threat, but not deprivation, are associated with accelerated biological
aging in children and adolescents. Biological Psychiatry, 85(3), 268–278. https://doi.
org/10.1016/j.biopsych.2018.09.008.

Tanzi R. E. (2012). The genetics of Alzheimer disease. Cold Spring Harbor perspectives in
medicine, 2(10), a006296. https://doi.org/10.1101/cshperspect.a006296.

Tilvis, R. S., Kähönen-Väre, M. H., Jolkkonen, J., Valvanne, J., Pitkala, K. H., & Strandberg, T.
E. (2004). Predictors of cognitive decline and mortality of aged people over a 10-year
period. The journals of gerontology. Series A, Biological sciences and medical
sciences, 59(3), 268–274. https://doi.org/10.1093/gerona/59.3.m268.

Tohgi, H., Utsugisawa, K., Nagane, Y., Yoshimura, M., Genda, Y., & Ukitsu, M. (1999a).
Reduction with age in methylcytosine in the promoter region -224 approximately -101 of
47
the amyloid precursor protein gene in autopsy human cortex. Brain research. Molecular
brain research, 70(2), 288–292. https://doi.org/10.1016/s0169-328x(99)00163-1.

Tohgi, H., Utsugisawa, K., Nagane, Y., Yoshimura, M., Ukitsu, M., & Genda, Y. (1999b). The
methylation status of cytosines in a tau gene promoter region alters with age to
downregulate transcriptional activity in human cerebral cortex. Neuroscience
letters, 275(2), 89–92. https://doi.org/10.1016/s0304-3940(99)00731-4.

Tomaka, J., Thompson, S., & Palacios, R. (2006). The Relation of Social Isolation, Loneliness,
and Social Support to Disease Outcomes Among the Elderly. Journal of Aging and
Health, 18(3), 359–384. https://doi.org/10.1177/0898264305280993.

Verheijen, J., & Sleegers, K. (2018). Understanding Alzheimer Disease at the Interface Between
Genetics and Transcriptomics, Trends in Genetics, 34(6), 434-447.
https://doi.org/10.1016/j.tig.2018.02.007.

Victor, C. R., Rippon, I., Nelis, S. M., Martyr, A., Litherland, R., Pickett, J., Hart, N., Henley, J.,
Matthews, F., & Clare, L. (2020). Prevalence and determinants of loneliness in people
living with dementia: Findings from the IDEAL programme. International Journal of
Geriatric Psychiatry, 35(8), 851–858. https://doi.org/10.1002/gps.5305.

Victor C. R. (2021). Is Loneliness a Cause or Consequence of Dementia? A Public Health
Analysis of the Literature. Frontiers in psychology, 11, 612771.
https://doi.org/10.3389/fpsyg.2020.612771.

Viswanathan, A., Rocca, W. A., & Tzourio, C. (2009). Vascular risk factors and dementia: how
to move forward?. Neurology, 72(4), 368–374.
https://doi.org/10.1212/01.wnl.0000341271.90478.8e.

Wang, S. C., Oelze, B., & Schumacher, A. (2008). Age-specific epigenetic drift in late-onset
Alzheimer's disease. PloS one, 3(7), e2698.
https://doi.org/10.1371/journal.pone.0002698.

Weir, D.R., Faul, J.D. , Langa, K.M. (2011). Proxy interviews and bias in cognition measures
due to non-response in longitudinal studies: a comparison of HRS and ELSA.
Longitudinal and Life Course Studies, 2(2), 170-184. doi:10.14301/llcs.v2i2.116.

Wilson, R. S., Krueger, K. R., Arnold, S. E., Schneider, J. A., Kelly, J. F., Barnes, L. L., Tang,
Y., & Bennett, D. A. (2007). Loneliness and risk of Alzheimer disease. Archives of
General Psychiatry, 64(2), 234–240. https://doi.org/10.1001/archpsyc.64.2.234.

Wu, J. W., Yaqub, A., Ma, Y., Koudstaal, W., Hofman, A., Ikram, M. A., Ghanbari, M., &
Goudsmit, J. (2021). Biological age in healthy elderly predicts aging-related diseases
including dementia. Scientific reports, 11(1), 15929. https://doi.org/10.1038/s41598-021-
95425-5.
48
Zannas, A. S. (2019). Epigenetics as a key link between psychosocial stress and aging: Concepts,
evidence, mechanisms. Dialogues in Clinical Neuroscience, 21(4), 389–396.
https://doi.org/10.31887/DCNS.2019.21.4/azannas.
Zarzour, A., Kim, H. W., & Weintraub, N. L. (2019). Epigenetic Regulation of Vascular
Diseases. Arteriosclerosis, thrombosis, and vascular biology, 39(6), 984–990.
https://doi.org/10.1161/ATVBAHA.119.312193.

Ziaei, M., Fischer, H. (2016). Emotion and Aging. Neuroimaging Personality, Social Cognition,
and Character. ISBN: 978-0-12-800935-2.

49
Appendix

Appendix A: Items in the UCLA Loneliness Scale (11 item)
Instructions: the next questions are about how you feel about different aspects of your
life: How much of the time do you feel….
1. You lack companionship?
2. Left out?
3. Isolated from others?
4. That you are “in tune” with the people around you?
5. Alone?
6. That there are people you can talk to?
7. That there are people you can turn to?
8. That there are people who really understand you?
9. That there are people you feel close to?
10. Part of a group of friends?
11. That you have a lot in common with the people around you?
Coding: 1 = Often, 2 = Some of the time, 3 = Hardly ever or never

50
Appendix B: Unconstrained Model With Depression and Full Model Without Depression

Note. Parameter estimates of covariates are unstandardzied. Bolded values are significant at the 0.05
level.

Abstract (if available)

Abstract In the absence of a treatment for dementia, identifying modifiable psychosocial risk factors and biomarkers for dementia are priorities to prevent or delay onset. Loneliness is regarded as a psychosocial risk factor for dementia. Only three studies to date have explored whether different trajectories of loneliness across the lifespan influence dementia risk differentially (Wilson et al., 2007; Kim et al., 2021; Akhter-Khan et al., 2021). Loneliness may influence biomarkers that are more proximally related to dementia risk. Epigenetic age, for example, quantifies biological aging that incorporates dysregulated physiological processes that more accurately characterize the aging process and correlate with psychosocial stress, Alzheimer’s disease pathology, and cognitive functioning. The present study uses growth mixture modeling to investigate different latent class trajectories of loneliness in the Health and Retirement Study using three loneliness measurements collected across eight years. We test whether different loneliness trajectories differentially predict epigenetic age, dementia risk, and their association. We find that groups characterizing chronic loneliness and increasing loneliness across the study window have increased dementia risk, and higher epigenetic age predicted higher dementia risk. Loneliness trajectory was not found to be associated with epigenetic age, epigenetic age did not mediate the association between loneliness and dementia risk, and the association between epigenetic age and dementia risk did not differ between loneliness groups. Results do not support the physiological dysregulation hypothesis of loneliness on dementia risk, and suggest epigenetic age is a potential biomarker of dementia risk.

Linked assets

University of Southern California Dissertations and Theses

Conceptually similar

PDF

Are life events differentially associated with dementia risk by gender? A twin study

PDF

Neuroimaging markers of risk & resilience to brain aging and dementia

PDF

Untreated hyperglycemia associated with tau pathology and worse cognitive performance in older adults

PDF

Dementia, co-morbid depressive symptoms, and mortality in older adults: a twin study

PDF

Alzheimer’s disease: dysregulated genes, ethno-racial disparities, and environmental pollution

PDF

Brainstem structural integrity in the progression of Alzheimer's disease

PDF

Underlying neural mechanisms of depression and dementia

PDF

Arterial stiffness and resting-state functional connectivity in older adults

PDF

Long-term blood pressure variability across the clinical and biomarker spectrum of Alzheimer’s disease

PDF

The neuroscience of ambivalent and ambiguous feelings

PDF

Cumulative risk as a moderator of multisystemic therapy effects for juvenile offenders

PDF

The design, implementation, and evaluation of accelerated longitudinal designs

PDF

Contributions of dynamic cerebrovascular function to cognitive decline and dementia: development and validation of a novel neuroimaging approach

PDF

Age-related differences in amygdala subregion functional connectivity with Ventromedial Prefrontal Cortex and Dorsolateral Prefrontal Cortex

PDF

Characterizing brain aging with neuroimaging, health, and genetic data

PDF

Self-perceptions of Aging in the Context of Neighborhood and Their Interplay in Late-life Cognitive Health

PDF

Family aggression exposure and community violence exposure associated with brain volume in late adolescence: a comparison of automated versus manual segmentation

PDF

Neurobiological correlates of fathers’ transition to parenthood

PDF

Dedifferentiation of emotion regulation strategies in the aging brain: an MVPA investigation

PDF

Epigenetic regulation of endocrine aging transitions of the perimenopausal and menopausal brain

Asset Metadata

Creator Lynch, Morgan (author)

Core Title Associations between longitudinal loneliness, epigenetic age, and dementia risk

School College of Letters, Arts and Sciences

Degree Master of Arts

Degree Program Psychology

Degree Conferral Date 2022-08

Publication Date 07/22/2022

Defense Date 07/22/2022

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

Tag dementia,epigenetic age,Loneliness,longitudinal,OAI-PMH Harvest,psychosocial risk factors

Format application/pdf (imt)

Language English

Contributor Electronically uploaded by the author (provenance)

Advisor Beam, Christopher (committee chair), Arpawong, Thalida (committee member), Kaplan, Jonas (committee member), Stange, Jonathan (committee member)

Creator Email lynmorg@usc.edu,morganelynch@gmail.com

Permanent Link (DOI) https://doi.org/10.25549/usctheses-oUC111373987

Unique identifier UC111373987

Legacy Identifier etd-LynchMorga-10915

Document Type Thesis

Format application/pdf (imt)

Rights Lynch, Morgan

Type texts

Source 20220722-usctheses-batch-961 (batch), 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 author, as the original true and official version of the work, but does not grant the reader permission to use the work if the desired use is covered by copyright. It is the author, as rights holder, who must provide use permission if such use is covered by copyright. The original signature page accompanying the original submission of the work to the USC Libraries is retained by the USC Libraries and a copy of it may be obtained by authorized requesters contacting the repository e-mail address given.

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

Repository Email cisadmin@lib.usc.edu