trajectories can be challenging to detect clinically. This poses an impediment to targeting and tailoring prevention and treatment approaches to the individual type of dementia.

Adding to this complexity is the predominance of mixed neuropathologies on autopsy, especially in older people, indicating that multiple pathological pathways are operating simultaneously and may require combinatorial interventions to achieve meaningful improvement. Consequently, different types of neurodegeneration as diagnosed clinically can show large overlaps in pathology at the brain tissue level (i.e., mixed pathology). Given the multiple and potentially intersecting pathways that contribute to AD/ADRD, diverse streams of research focused on individual pathogenetic processes and, most importantly, their integration, are needed to understand risk and resilience factors, as well as the biological processes leading to disease. The findings from such research will support the expansion of the current portfolio of prevention and treatment strategies (discussed in Chapter 4).

This chapter begins by framing the prevention and treatment of AD/ADRD in the context of healthy aging and the optimization of brain health across the life course. It goes on to explore key exposures and mechanistic pathways that may be intervened upon individually or in combination to optimize brain health over the life course and prevent and treat AD/ADRD. It focuses specifically on disease pathways that may be shared across AD/ADRD, as well as disease-agnostic (i.e., disease-nonspecific) resilience mechanisms. With this approach, the committee aims to identify opportunities to maximize the impact of AD/ADRD research investments and advocate for the transition away from the siloed study of individual causes of dementia that has dominated the research landscape and that fails to address the reality of mixed etiology dementia. The chapter ends by identifying research priorities directed at evidence gaps that, if addressed, could guide the development of population- and individual-level AD/ADRD intervention strategies.

HEALTHY AGING, RESILIENCE, AND BRAIN HEALTH ACROSS THE LIFE COURSE

Healthy aging can be defined as “a continuous process of optimizing opportunities to maintain and improve physical and mental health, independence, and quality of life throughout the life course” (PAHO/WHO, 2024). The later years of life present many new opportunities for older individuals, from pursuing new activities and long-neglected passions and even new career paths. Yet the chance to enjoy these opportunities depends to some degree on healthy aging.

As discussed further in this chapter, aging in the biological sense results from the deterioration of myriad molecular and cellular structures and

processes over time. Importantly, however, these changes do not occur on a linear trajectory and do not strictly associate with a person’s age in years (WHO, 2022). Chronological age represents the amount of time a person has been alive. Although there is no single definition for biological age, it is conceptualized in terms of various molecular and cellular changes that accumulate over time (Jazwinski and Kim, 2019). Building on the work of a 2014 trans-institute Geroscience Interest Group Summit sponsored by the National Institutes of Health,¹ López-Otín and colleagues (2023) recently proposed 12 hallmarks of aging, which spanned from the molecular to the meta-organism levels (see Figure 3-1). Manipulation of these hallmarks, which are interdependent, may accelerate or decelerate the biological aging process. Many of these hallmarks overlap with the biochemical and cellular pathologic pathways common to AD/ADRD (discussed later in this chapter), suggesting opportunities to use knowledge of biological aging processes in the development of interventions for AD/ADRD (Fillit et al., 2023).

Biological age is increasingly recognized as a better indicator of chronic health outcomes as compared to chronological age (Ho et al., 2023), with implications for lifespan and brain health span (the period of life during which an individual remains healthy) (Sierra, 2016). Biological age is not necessarily linked to chronological age, given that a person can be biologically younger or older than their chronological age suggests (Sierra et al., 2021; Zhang and Gladyshev, 2020). As a result, there is substantial diversity in the range of physical and cognitive ability seen in older age. While genetics plays a role, a large part of this variability arises from the physical and social environments in which people live and the effect of these environments on their opportunities and health behaviors (i.e., social determinants of health) (WHO, 2022), as supported by increasing evidence from animal and human studies (Durso et al., 2022; Horvath and Levine, 2015; Kalia et al., 2022; Liu et al., 2023a; Quach et al., 2017; Ward-Caviness et al., 2016). Complementary to biological hallmarks of aging, social hallmarks of aging, such as low socioeconomic status, minority status, adverse life events, adverse psychological states, and adverse behaviors, can accelerate aging, while reducing exposure to these social hallmarks can slow the onset of age-related poor health outcomes (Crimmins, 2020; Hooten et al., 2022).

The risk of sporadic forms of dementia increases as individuals age, typically accompanied by age-related biological changes in the brain and throughout the body that may not be observed in individuals with early-onset forms of the diseases (Scheltens et al., 2021). Some characteristics of neurodegenerative disease (e.g., neurodegeneration and cerebral atrophy,

___________________

¹ Geroscience is a transdisciplinary field focused on the relationship between biological aging and age-related diseases (Fillit et al., 2023; Hara et al., 2019).

neuropathological protein accumulation, cognitive decline), gradually manifest over the life course even among individuals who do not develop clinical dementia (Zhang et al., 2023e). Thus, biological pathways underlying both aging and AD/ADRD exist on a continuum and are likely to overlap (Gonzales et al., 2022). Consequently, interventions targeting the underlying biological pathways associated with aging may be useful for delaying or even preventing the onset of AD/ADRD (Cummings et al., 2023a; Fillit et al., 2023; Hara et al., 2019; Sierra, 2016; Sierra et al., 2021).

It should be emphasized that the accumulation of neuropathologic changes with aging does not necessarily lead to cognitive decline. Studies to correlate neuropathologic changes with premortem cognitive function have demonstrated that cognitive health can be maintained despite the accumulation of high levels of neuropathology (Zhang et al., 2023e). This phenomenon has been defined as cognitive resilience, which refers to “the capacity of the brain to maintain cognition and function [despite the presence of] aging and disease” (Stern et al., 2023, p. 3).² The growing understanding of cognitive resilience and the implication for potential intervention strategies is important given that people generally care more about maintaining their cognitive and functional ability than about the neurobiological changes that may occur with aging. Building cognitive reserve or enhancing the brain’s ability to compensate or adapt to neuropathology—which does not necessarily imply a return to a previous state of function but rather the ability to continue to function despite the presence of pathology—are strategies for increasing an individual’s resilience (Stern et al., 2019), and may contribute to preventing, delaying, and/or slowing clinical disease progression.

Cognitive resilience is not dichotomous but rather exists on a continuum that includes better- and worse-than-expected performance for a given level of pathology (Zammit et al., 2023). Thus, mechanisms can either maintain or degrade resilience. For example, psychosocial risk factors such as depression, loneliness, and chronic distress increase risk of clinical dementia independent of effects on typical pathologic features such as amyloid plaques, tau tangles, Lewy bodies, or infarcts (Wilson et al., 2003, 2006, 2007a,b), indicating the existence of yet-to-be discovered mechanisms that degrade resilience. Other factors such as education, cognitive stimulation, socialization, sense of purpose, and physical activity may function in a protective manner to increase or maintain resilience, enabling the brain to

___________________

² Multiple definitions and frameworks have been proposed to clarify terminology and operationalize the concepts of resilience and resistance (Neuner et al., 2022; Stern et al., 2019, 2023; Zammit et al., 2023). For the purposes of this report, resistance to neuropathology is considered to be distinct from resilience and refers to the absence of neuropathology or lower levels than would be expected based on age and other risk factors (e.g., genetic risk) (de Vries et al., 2024; Zammit et al., 2023). This definition for resistance is similar to the definition for brain maintenance provided in one recently published consensus framework (Stern et al., 2023).

compensate for brain pathologies and maintain cognitive function (Zammit et al., 2023). As discussed later in this chapter, the discovery of protective genetic variants has shown that resilience can also be genetically mediated. Thus, there are both modifiable and nonmodifiable factors that contribute to resilience and may in fact interact. The neural mechanisms that underlie resilience are an active area of study (de Vries et al., 2024; Neuner et al., 2022). Understanding how protective variants and other mediators promote resilience may guide new strategies for interventions.

Strategies focused on avoiding gerogens—age-accelerating factors, such as pollution and stress (NASEM, 2020a,b)—and adopting resilience-promoting lifestyle factors, such as healthy diet, exercise, regular sleep patterns, and social activities, are applicable to both antiaging and AD/ADRD prevention efforts and may support the extension of the health span and the optimization of brain health across the life course. Assessment of individuals’ aging clocks using genetic, epigenetic, metabolomic, and other tools could in the future guide more personalized intervention strategies.

THE EXPOSOME AND BRAIN HEALTH ACROSS THE LIFE COURSE

Defining and Measuring the Exposome

The concept of the exposome emerged 2 decades ago and was originally described as encompassing “life-course environmental exposures (including lifestyle factors), from the prenatal period onwards” (Wild, 2005, p. 1848). In recognition of the public health importance of gene–environment interactions, the term was coined in an effort to draw attention to the need to characterize environmental exposures with the same scientific rigor and precision that was at the time being applied to the analysis of the human genome. Understanding the interplay between the genome and the environment depends on the ability to accurately assess environmental exposures (Siroux et al., 2016). Exposomics as a scientific discipline has continued to mature over the last 20 years and was recently defined as “a field that studies the comprehensive and cumulative effects of physical, chemical, biological, and psychosocial mediators that impact biological systems by integrating data from a variety of interdisciplinary methodologies and streams” in order to “enable discovery-based analysis of environmental influences on health” (Banbury Center, 2024). As suggested by this definition, exposures are considered broadly, extending well beyond environmental pollutants (e.g., air pollution, environmental lead) that may commonly be a focus of exposure assessments. Such factors as social determinants of health and health behaviors are also encompassed within the exposome.

For diseases for which genetic variants explain only a limited fraction of the variability in risk and that are presumed to be largely driven by

gene–environment and age-associated interactions (i.e., acquired susceptibility), exposomics may provide one potential approach for better understanding disease pathways. The field relies on advanced data analytics and causal inference methods to identify the relationship between exposome factors and phenotypes of interest (e.g., disease outcomes) (Banbury Center, 2024). Findings from such efforts can guide and complement effective prevention strategies.

A notable challenge, however, relates to how the exposome is defined and comprehensively measured over the entire life course given its vastness of scale and inherent variability over time, as well as how it is linked to health (Siroux et al., 2016). Exposome measurement and evaluation methods use multiple technologies and multilayered data sources, including demographic surveys, administrative data (e.g., Social Security records, health care claims), historical information (e.g., geocoded residential histories), sensors and monitors, satellite data, and analysis of endogenous and exogenous compounds in the body that are biological indicators of exposures (Banbury Center, 2024; Miller, 2024). High-resolution mass spectrometry and other increasingly high-throughput multiomics approaches (e.g., epigenomics, transcriptomics, proteomics, metabolomics) are enhancing the identification of such biological exposure indicators, as well as early molecular changes that precede disease onset (Vineis et al., 2020).

Although a data-layering approach enables the exploration of interactions among different environmental exposures over the life course and consideration of high-risk exposure windows, such large-scale data collection, management, and analysis are costly and integration of these diverse data types requires sophisticated analytic tools (Siroux et al., 2016; Vermeulen et al., 2020). Moreover, the simultaneous analysis of multiple exposures and their effects on health raises concerns regarding risk of misattribution of the effect of one exposure to another (i.e., exposure misclassification) and can make it challenging to identify causal associations because of correlations between exposures in the exposome (Siroux et al., 2016). Data security and privacy considerations must also be considered given that exposome research may involve the linking of some forms of protected health information, such as residential history and genetic information (Safarlou et al., 2023).

Linking the Exposome to Brain Health and AD/ADRD

Findings from twin studies suggest that while genetics is a major contributor to AD/ADRD risk (Gatz et al., 2006), environmental and age-related factors also have a substantial influence (Maloney and Lahiri, 2016; Migliore and Coppedè, 2022; Paulson and Igo, 2011). Yet the role of environmental factors in AD/ADRD etiologies, either directly through effects on

the brain or indirectly via altered brain–body interactions, remain poorly understood (Finch and Kulminski, 2019).

In the context of brain health, the exposome paradigm can be conceptualized in the form of the neural exposome, which encompasses a diverse array of non-genetic factors that influence neurodegenerative diseases (including AD/ADRD) and other neurological conditions (Tamiz et al., 2022). The neural exposome includes exogenous (e.g., originating from outside the organism such as environmental contaminants), behavioral (e.g., lifestyle), and endogenous factors (e.g., microbiome), as depicted in Figure 3-2. The brain’s dynamic and adaptive nature, characterized by continual change as a function of experience and exposures across the life course, makes the study of the neural exposome complex. Additionally, just as genetic variants function together to influence genetic disease risk,

combined exposures from across the life course similarly are likely to have interactive effects on brain health (Tamiz et al., 2022). This complicates intervention efforts as strategies targeting a single factor may be expected to have limited effectiveness when different factors contribute additively or synergistically to the risk of AD/ADRD. Further complexity stems from the fact that exposures may be happening in early life (in so-called critical windows of development) but not manifesting until later in life. This is the basis for the developmental origins of health and disease hypothesis (Klocke and Lein, 2019; Migliore and Coppedè, 2022).

The human brain continues to develop through adolescence. Consequently, exposures (environmental and social) from gestation through teenage years can disrupt brain development in ways that will not be immediately apparent if the affected regions are not functional until later life or if compensatory mechanisms mask the effects until additional insults accumulate to a level that can no longer be compensated for. Though not related to neurodegeneration, an illustrative example is provided by the effects of monosodium glutamate, which, in high concentrations, can cause neuronal death in the hypothalamus of a developing brain. Deficits from the loss of those neurons does not become apparent until adolescence when the function of those cells is normally activated (Klocke and Lein, 2019). Whether and to what degree AD/ADRD have a developmental basis remains unclear but may be uncovered by further study of the exposome.

There has been growing investment in research on the exposome and, of particular interest to this report, its role in brain aging and disease, with the National Institute of Environmental Health Sciences (NIEHS), the National Institute on Aging (NIA), and the National Institute of Neurological Disorders and Stroke (NINDS) providing support for such research (Cavaliere and Gülöksüz, 2022; NIA, 2020; Tamiz et al., 2022). The growth of research on the exposome and AD/ADRD has created opportunities for greater interagency collaboration to advance critical research across multiple conditions. For example, in 2022, NINDS launched its Office of Neural Exposome and Toxicology, which is focused on interdisciplinary research on the neural exposome and the creation of partnerships within NINDS and across other National Institutes of Health (NIH) institutes, centers, and offices (ICOs), federal agencies (Centers for Disease Control and Prevention, Environmental Protection Agency), advocacy groups, and scientific societies (Tamiz et al., 2022). Additionally, several NIH ICOs (NIA, NINDS, NIEHS, the National Institute of Arthritis and Musculoskeletal and Skin Diseases, and the National Cancer Institute, among others) are cofunding a new coordinating center for exposome research to accelerate precision environmental health (Columbia University, 2024; NIH, 2023).

The contribution of exogenous environmental exposures to AD/ADRD is a topic of active investigation. For example, a growing body of evidence

from epidemiological and experimental studies has linked air pollution (e.g., fine particulate matter, nitrogen dioxide, nitrous oxides) and other environmental pollutants (neurotoxic metals, pesticides) to increased risk of neurologic disorders, including dementia (Bakulski et al., 2012; Liu et al., 2023a; Migliore and Coppedè, 2022; NASEM, 2020a,b; Peters et al., 2019; Shi et al., 2021; Shih et al., 2007; Wilker et al., 2023; Zhang et al., 2023a), although data are lacking for the vast majority of chemicals to which humans are exposed. There is an opportunity to leverage existing databases, emerging technologies for high-throughput screening (Lefevre-Arbogast et al., 2024), and increased understanding of the pathogenesis of AD/ADRD to identify specific environmental chemicals that interact with biological pathways involved in neurodegeneration to promote disease, and conversely, resilience.

Research is also elucidating the influence of social, cultural, and behavioral factors on brain health. Behavioral factors such as physical inactivity, smoking, alcohol consumption, and diet are well-established risk factors for dementia (Livingston et al., 2024; Patten and Lein, 2019). However, recent studies focused on social and economic factors such as educational attainment, wealth, access to health care (Chen and Zissimopoulos, 2018), loneliness and social isolation (Huang et al., 2023), macro-level social systems, and structural influences have identified associations with cognitive impairment and dementia (Lamar et al., 2023; Powell et al., 2020; Reuben et al., 2024; Walsemann et al., 2023). Such research is critical to understanding how social disadvantage and disparities in exposures to environmental hazards (e.g., neighborhood proximity to landfills and major highways that generate traffic-associated air pollution) contribute to differences in AD/ADRD risk across diverse populations. There is also increasing evidence to suggest that gender roles (e.g., household, childrearing, and caregiving roles) and disparities (e.g., access to education, occupational opportunities, autonomy) contribute to gender differences in AD/ADRD incidence (Mielke, 2024). Continuing to uncover these kinds of associations and their interactions will require longitudinal cohort studies with in-depth exposome measurements and analysis.

To understand the exposome and how it interacts with molecular and cellular mechanisms to influence AD/ADRD risk, it is critical to capture the context in which exposures occurred over an individual’s life course (e.g., socioeconomic status, locations in which individuals lived throughout their lives) (Melcher et al., 2024). Such contextual factors can significantly influence the exogenous, endogenous, and behavioral facets of the exposome, and are drivers of disparities in AD/ADRD outcomes (NASEM, 2024).

The linking of social phenotypes (e.g., social vulnerability) to biological phenotypes (e.g., neuropathology markers) can be done prospectively or retrospectively (Kind, 2024; Powell et al., 2020). As an example of the

latter, public data tracing methods can be used to collect life-course residential histories for individuals whose brains were donated to Alzheimer’s Disease Research Center brain banks, thereby linking historic population-level exposure data to existing biobank samples (Melcher et al., 2024). Retrospective approaches, however, are resource intensive and subject to the availability of accurate data. Collection of social phenotype data in parallel to biological data may be less costly and less labor-intensive. The Health and Retirement Study has been collecting social, economic, and demographic data on a nationally representative sample of middle-age and older Americans since 1992, as well as biomarker data on the study respondents (Biomarker Network, 2024). A structural and social determinants of health module has been developed and is being implemented in Alzheimer’s Disease Research Centers to enable the linking of prospectively collected social determinants of health data to neuropathology findings, which will help improve understanding of how social determinants of health modify cognition and AD/ADRD risk (Stites et al., 2022). As noted earlier in this chapter, however, privacy laws and regulations and data security concerns related to protected health information can impede efforts to conduct such linking of studies. Research infrastructure addressing both administrative and legal challenges is needed to facilitate the intake and linkage processes for protected health information (Kind, 2023).

The findings from exposome studies suggest opportunities for individual-level, public health, and policy interventions aimed at primary prevention, such as reducing exposure to environmental contaminants, modifying lifestyle, and eliminating structural factors that contribute to social disadvantage. It is possible to imagine precision prevention approaches that stratify the population based on environmental exposures. However, important knowledge gaps regarding the timing for exposures and resultant changes in brain health remain, impeding efforts to develop life-course approaches to maximize AD/ADRD prevention. Acknowledging the NIH-supported research efforts already underway (NIA, 2023a; NINDS, 2024a), further investment is needed to identify and understand brain-critical exposures across the life course (from the prenatal period to late life); vulnerable age windows (Liu et al., 2023a), such as periods of neurodevelopment and major hormonal changes (e.g., adolescence, pregnancy, menopause); the dose-over-time dependency of effects; and how exposures correlate and interact (Tamiz et al., 2022). Cohort studies (discussed in Chapter 2) and natural experiments (e.g., city-level removal of lead pipes, policies aimed at air pollution reduction) provide opportunities to address these scientific priorities, but this requires multidisciplinary research (breaking down discipline-specific silos as discussed in Chapter 5) and attention to addressing current data gaps (e.g., early-life exposures, reproductive history, social determinants of health).

large effect sizes and can therefore be identified and investigated through family studies. Rare population cohorts (e.g., ARTFL-LEFFTDS Longitudinal Frontotemporal Lobar Degeneration [ALLFTD]³, the Dominantly Inherited Alzheimer’s Network) have been established for the study of these dominantly inherited forms of dementia and have yielded invaluable insights on disease mechanisms, biomarkers, and potential treatment targets (Asken et al., 2023; Bateman et al., 2011; Zhang et al., 2023b).

Genetically determined forms of dementia are not limited to those that are inherited in an autosomal dominant manner. Down syndrome-related AD is another form of genetically determined early onset AD that results from a gene dose effect stemming from an extra copy of the APP gene, which is located on chromosome 21. Neuropathological hallmarks of AD (amyloid plaques and tau neurofibrillary tangles) are nearly universal in people with Down syndrome by the age of 40, and the lifetime risk of dementia in this population is estimated at greater than 90 percent (Fortea et al., 2021). As with autosomal dominant forms of AD/ADRD, the natural history of Down syndrome-related AD follows a predictable course (clinical and biomarker changes). Consortia such as the Alzheimer Biomarkers Consortium—Down Syndrome (NIA, 2024a) and the Alzheimer’s Clinical Trials Consortium—Down Syndrome (ACTC-DS, 2024) are working to advance the development of AD biomarkers and therapies for people with Down syndrome.

A recently published analysis proposes that APOE4 homozygotes may constitute a third form of genetically determined AD (Fortea et al., 2024). The E4 allele of the apolipoprotein E gene (APOE) has long been recognized as a major genetic risk factor for AD with a large effect size (Cacace et al., 2016; Lambert et al., 2023). APOE protein is known to be involved in lipid transport (Raulin et al., 2022), but the mechanisms by which APOE4 increase dementia risk have yet to be fully elucidated and represent an area of intense research.⁴ The proposal of APOE4 homozygotes as a genetic form of AD is based on the near full penetrance of AD neuropathology, as well as the predictability of symptom onset and clinical and biomarker changes over time, characteristics that are also seen with autosomal dominant AD and Down syndrome-related AD. Furthermore, APOE3 and APOE4 heterozygotes exhibit intermediate phenotypes, suggesting the possibility of autosomal semidominant inheritance (Fortea et al., 2024; Genin et al., 2011).

___________________

³ ALLFTD recruits people with autosomal dominant FTD syndromes, as well as people with sporadic FTD.

⁴ NIA organized a workshop on APOE genetics as a major determinant of AD pathobiology in September 2024. More information is available at https://www.nia.nih.gov/research/dab-dgcg-dn/workshops/apoe-genetics-major-determinant-alzheimers-disease-pathobiology (accessed October 11, 2024).

___________________

⁵ The observation of a protective effect for the Christchurch mutation in APOE3 for an individual with an autosomal dominant PSEN1 mutation highlights the importance of APOE as a fundamental driver of AD neuropathology.

___________________

⁶ Sporadic forms make up 70 percent of FTD, a smaller fraction as compared to AD and LBD (Greaves and Rohrer, 2019). A larger number of strong risk loci (i.e., those associated with autosomal dominant forms of disease) have been identified for FTD as compared to AD (see Table 3-1).

diverse ancestral backgrounds, as genetic analysis has shown that continental genetic ancestry plays an important role in AD risk and protection, as evidenced by the variation in risk effect for the APOE4 allele across populations with diverse ancestral backgrounds (Reitz et al., 2023). Observing more population-specific variants increases statistical power in AD/ADRD gene discovery. The Alzheimer’s Disease Sequencing Project has been a major NIA investment in improving understanding of the genetic architecture of AD/ADRD, and recent phases of the initiative have included a focus on expanding the genetic diversity of study populations (see Box 3-1).

Continued investment in understanding how different variants affect the functional role of each gene associated with disease risk can help to elucidate the biological mechanisms of AD/ADRD (discussed later in this chapter) and thereby guide the development of more targeted intervention strategies. For example, genetic studies have implicated lysosomal dysfunction and autophagy pathways in AD/ADRD (Bellenguez et al., 2022; Deng et al., 2017). However, caution is required when attempting pathway analysis using GWAS findings as it is not always clear which gene is linked to an identified locus and pathway annotation suffers from a number of limitations (Silberstein et al., 2021). Further development of tools and processes is needed to improve the translation of genetic signals to target pathways. In the meantime, triangulation using a combination of genetic findings and data from experimental studies (in vitro and in vivo) can increase confidence in the identification of affected pathways.

Understanding Gene–Environment Interactions

As noted earlier, there are various molecular and cellular targets through which exposome factors (e.g., lifestyle, social determinants of health, pollutant exposures) may exert pathogenic effects. For example, exposome factors can interact directly with proteins (e.g., bind to and/or alter the function of receptors, enzymes, and structural proteins), but may also alter protein levels or function through interactions with an individual’s epigenome and/or genome. In framing brain health as a trajectory over the life course, an individual’s genetics can be envisioned as defining the major bounds for that trajectory, while the exact course will be determined by the interactions between genes and exposome factors. Perhaps the best studied gene–environment interactions are those involving the APOE gene, but generally these interactions remain poorly understood (Dunn et al., 2019; Migliore and Coppedè, 2022). In fact, environmental and social factors (e.g., socioeconomic status) likely influence observed associations between genetic variants and AD (Reitz et al., 2023), making it challenging to disentangle risks from different sources, which is important for understanding the variation in AD/ADRD risk across diverse populations.

analyses (Pericak-Vance, 2024). Moreover, collaboration among biomedical science experts and experts in bioinformatics and information science will be needed to tackle the data analytic challenges related to managing and integrating these vast and complex datasets and developing tools to extract insights (Hamamoto et al., 2020).

In addition to research focused on understanding how interactions between exposures and genetics lead to disease, other studies are exploring the degree to which genetic risk can be offset by protective aspects of the exposome and the mechanisms behind those protective effects. For example, a recent observational study of associations by Lourida and colleagues (2019) found that a healthy lifestyle (e.g., regular exercise, healthy diet, no current smoking, and alcohol consumption in moderation) could lower but not eliminate dementia risk in individuals with a high genetic risk.

Understanding Molecular and Cellular Mechanisms Associated with Aging and AD/ADRD

Neurodegenerative diseases such as AD/ADRD exhibit a wide range of clinical manifestations, which arise from the loss of specific neurons and synapses in different areas of the brain. While different disorders display some tissue specificity (e.g., degeneration of cholinergic neurons for AD, degeneration of dopaminergic neurons for Parkinson’s disease [PD], degeneration of motor neurons for amyotrophic lateral sclerosis [ALS]), they also feature some anatomical overlaps of affected brain regions and possess shared characteristics and mechanisms, particularly the regional aggregation and spreading of cytosolic or nuclear inclusion proteins. Thus, even when different diseases are associated with distinct genetic variants, common biological themes can be observed (Behl, 2023; Gan et al., 2018).

has been associated with decreased brain volume and impaired cognitive function in AD/ADRD (Franceschi and Campisi, 2014; Hara et al., 2019; Pilling et al., 2015).

As the resident immune cells in the brain, microglia play essential roles in maintaining homeostasis, including defense against infection or damage and phagocytosis of debris or dying cells (Salter and Stevens, 2017). Microglia also secrete tissue rebuilding factors (Vasic et al., 2019), a function that may be targeted by therapies aimed at stimulating brain tissue regeneration (discussed further in Chapter 4). During aging and in the context of some neurodegenerative diseases, microglia can acquire an activated proinflammatory phenotype characterized by morphological and functional changes, which include the release of proinflammatory cytokines and other neurotoxic mediators (Tejera et al., 2019; Wang et al., 2023). While pathways leading to microglial activation are not fully understood, there is some evidence that systemic inflammation can affect the inflammatory response of microglial cells in the brain and interfere with their clearance of aggregating proteins, such as amyloid beta in AD (Tejera et al., 2019). Environmental exposures such as brain chemical injuries or infection that trigger systemic inflammatory responses may thereby contribute to microglial activation (Ahmed et al., 2024; Greve et al., 2023; Jayaraj et al., 2017; Zhang et al., 2023c). Similarly, studies have found a link between proinflammatory microbial metabolites in the gastrointestinal tract and microglial activation and alpha-synuclein aggregation in the brain in PD (Sampson et al., 2016). Perivascular macrophages and astrocytes are also involved in the brain’s innate immune response. Depending on the injury, astrocytes can demonstrate distinct activation states and transcriptional signatures that mediate neurotoxic or neurotrophic effects, which may be modulated by a microglia-astrocyte crosstalk mechanism (Gao et al., 2023; Liddelow et al., 2017).

While not as well studied as the role of the innate immune system in AD/ADRD, the adaptive immune system may also contribute to neurodegenerative disorders and may have both protective and pathogenic roles (Chen and Holtzman, 2022; Mayne et al., 2020). Studies of the adaptive immune system in dementia have primarily focused on the composition and roles of T-cell populations, which appear to infiltrate the brain from the periphery, potentially facilitated by changes to the blood–brain barrier (Chen and Holtzman, 2022). Activated T cells may perpetuate the inflammatory cascade and induce microglia into an activated phenotype by secreting proinflammatory and neurotoxic mediators. Further research is needed to understand the function of specific T-cell populations in different types of dementia. For example, proinflammatory CD4+ T_H17 cells have been implicated in neuronal degeneration in LBD (Gate et al., 2021). CD4+ T cells that exhibit an immune suppressive phenotype, known as Tregs,

are thought to have a neuroprotective role in AD based on animal studies. However, alterations in Treg levels and immunomodulatory activity over the course of neurodegeneration are not well understood and some studies have suggested a detrimental effect of Tregs (Duffy et al., 2018; Gendelman and Appel, 2011; Mayne et al., 2020).

The role of B cells and humoral immune responses in neurodegeneration leading to dementia is less studied (Chen and Holtzman, 2022; Lutshumba et al., 2021), but research in mouse models suggests a role for B cells in AD progression (Kim et al., 2021). Similarly not well understood is the potential role of autoimmunity in neurodegenerative diseases that cause dementia. Some converging lines of evidence suggest that dementia may have an inflammatory autoimmune component (Lindbohm et al., 2022). Links between autoimmune disorders and subsequent dementia based on epidemiological studies, however, have not been as strong as the associations between infections and dementia (Janbek et al., 2023).

Genetic studies provide strong evidence for the link between immune dysfunction, inflammation, and AD/ADRD. Indeed, many genes associated with neurodegenerative diseases regulate responses of the innate immune system (Gan et al., 2018). For example, whole-genome sequencing studies helped identify rare variants in TREM2 (a transmembrane receptor highly expressed in immune system cells, including microglia and myeloid cells) that lead to a two- to threefold increase in AD risk (Gan et al., 2018; Guerreiro et al., 2013; Jonsson et al., 2013; Zhang et al., 2013). In familial FTD-TDP, genetic mutations that lead to insufficient production of the progranulin gene product, which is highly expressed in microglia, are associated with exacerbation of proinflammatory responses and the activation of microglial cells (Baker et al., 2006; Cruts et al., 2006; Yin et al., 2010). The immunomodulatory effects of these and other disease-associated gene mutations implicate aspects of the innate immune system in neurodegenerative diseases (Gan et al., 2018).

To further elucidate the role of inflammation and immune dysregulation in AD/ADRD, key knowledge gaps need to be addressed, including:

• pathways promoting various microglial states in neurodegeneration;
• connections between microglia, astrocytes, neurons, oligodendrocytes, and endothelial cells;
• the role of microglia in the clearance of protein aggregates; and
• the role of the adaptive immune system in neurodegeneration (Lamb, 2024).

Such research may suggest strategies for therapies targeting neuroinflammation (Lin et al., 2023; Tejera et al., 2019). While broad-spectrum anti-inflammatory drugs have not been effective, specifically targeting

some aspects of inflammation may be a more promising approach (Fillit et al., 2023).

Vascular Dysfunction

Vascular pathology is widely acknowledged as a contributing factor to the development of dementia (Gorelick et al., 2011; Hara et al., 2019). Epidemiological evidence supports an association between risk factors for cardiovascular disease, cerebrovascular dysfunction, and cognitive impairment (Corriveau et al., 2016; Pacholko and Iadecola, 2024). Vascular pathologies are thought to be widely prevalent. Findings from two longitudinal cohort studies showed that over 95 percent of individuals diagnosed with probable AD had mixed pathologies observed at autopsy. Approximately 90 percent of those with probable AD had vascular pathologies present (Kapasi et al., 2017).

Vascular contributions to cognitive impairment and dementia arise from age-related alterations in the neurovascular unit. The neurovascular unit consists of various components, including nonfenestrated endothelial cells,⁷ pericytes, smooth-muscle cells, astrocytes, microglia, oligodendroglia, and neurons (see Figure 3-4). These components collectively contribute to the regulation of cerebral blood flow matched to neuronal metabolic demands (Claassen et al., 2021), maintenance of the blood–brain barrier, and communication between the vascular and neural compartments. Age-related changes within the neurovascular unit can disrupt these crucial functions, leading to cognitive impairment and the development of vascular dementia (Cai et al., 2017). During aging, and particularly in the context of neurodegenerative diseases, there is a notable decline in the population of pericytes, resulting in diminished cerebral blood flow delivery and subsequent downstream changes leading to neurodegeneration and dementia (Uemura et al., 2020).

One vascular pathology relevant to AD is vascular breakdown in the brain resulting from blood–brain barrier disruption (Fillit et al., 2023), which may be caused by pathological changes to the neurovascular unit. The integrity of the blood–brain barrier diminishes during normal aging, and this decline becomes even more pronounced in individuals with AD and related dementias, which often results in capillary leakage, brain leukocyte infiltration (Cai et al., 2017), ingress of toxic substances, and reactive response of astrocytes and microglia, which further aggravates neural dysfunction (Watanabe et al., 2020). Additionally, such breakdown also leads to the leakage of fibrinogen into the central nervous system (CNS),

___________________

⁷ “Non-fenestrated endothelium is characterized by low permeability and a high abundance of tight junctions” (Gifre-Renom et al., 2022).

triggering neuroinflammation and causing insoluble fibrin clots to form (Bardehle et al., 2015; Fillit et al., 2023). Fibrinogen is not detectable in a healthy brain; in AD, however, it reaches detectable levels and interacts with amyloid, which then exacerbates clotting, fibrin deposition, and proinflammatory signaling (Fillit et al., 2023; Ryu and McLarnon, 2009; van Oijen et al., 2005; Viggars et al., 2011).

Decreased blood flow to the brain and the disruption of the blood–brain barrier can on their own contribute to neurodegeneration and the development of dementia, but may also be elements of what has recently been described as “multiple hits” models (which include vascular changes) contributing to dementia (Patrick et al., 2019; Steele et al., 2022). Multihit models (see Figure 3-5) describe the convergence of interconnected hits associated with age-related, genetic, vascular, and immune factors (e.g.,

neurovascular dysfunction, neuroinflammation, oxidative stress, endosomal trafficking dysfunction, changes in lipid metabolism, among others) to precipitate AD pathology, including extracellular amyloid plaques and intraneuronal neurofibrillary tangle formation, as well as other alterations that contribute to dementia (Steele et al., 2022).

One prominent NIH-supported effort in this area is the Molecular Mechanisms of the Vascular Etiology of Alzheimer’s Disease (M^2OVE-AD) initiative, which was established in 2016 and works to examine the link between vascular risk factors, cerebrovascular disease, and AD (NIA, 2016). This consortium brought together over 12 crossdisciplinary research teams from different institutes to use brain tissue and peripheral fluids collected from previous studies that have characterized neurodegeneration and vascular disease to generate various types of multiomics data. Integration of these data with cognitive, neuroimaging, and vascular physiology data is helping to better understand and predict the underlying molecular mechanisms behind vascular pathology and AD risk (AD Knowledge Portal, 2024a).

Aberrant Proteostasis and Deficiency in the Endosomal–Lysosomal (Autophagy) Pathway

The accumulation of misfolded proteins in the CNS is a common feature of many neurodegenerative diseases, implicating a major role for a breakdown in cellular protein quality control and clearance pathways (Gan et al., 2018; Morawe et al., 2012). Protein folding is mediated by molecular chaperones, which are themselves proteins that can become dysregulated. One therapeutic approach under investigation is to use the activity of molecular chaperones to disaggregate misfolded proteins and refold them into their native forms. An alternative approach involves targeting the cellular machinery involved in clearing protein aggregates. Autophagy is the cellular mechanism responsible for breaking down and recycling aggregated misfolded proteins and damaged organelles and plays a major role in proteostasis. In conjunction with the ubiquitin-proteasome and the lysosomal systems, autophagy functions to alleviate cellular stress by breaking down pathogenic forms of aggregate-prone proteins (i.e., TDP-43, amyloid, tau, and alpha-synuclein), lipids, dysfunctional mitochondria, and other organelles in cells (Hu et al., 2015; Scrivo et al., 2018).

This cellular mechanism is particularly critical in postmitotic cells such as neurons, as they are unable to reduce proteotoxic burden and eliminate cellular waste through cell division, rendering them more vulnerable to proteotoxic insults (Nixon, 2013). Thus, functional protein and organelle turnover as provided by a functional autophagy secures neuronal function, stability, and resistance. During the aging process, autophagy may become disrupted, resulting in intracellular buildup of various misfolded

proteins. This age-associated effect can be magnified in the context of neurodegenerative disease (Bartlett et al., 2011; Gonzales et al., 2022).

Aberrant autophagy may contribute to the pathologic accumulation of amyloid plaques, tau tangles, and other protein aggregates found deposited in the brain (e.g., alpha-synuclein, TDP-43) in AD and other forms of neurodegenerative disorders (Fillit et al., 2023; Rubinsztein et al., 2011). Analysis of postmortem human brains affected by AD reveals abnormal autophagy (Lee et al., 2022; Loeffler, 2019). Propagation of misfolded proteins in animal models match observed progression patterns of neurodegeneration in the human brain (Yang et al., 2021). Endosomal–lysosomal dysfunction is one of the early pathophysiological changes that can be observed in neurodegeneration. The lysosome is the executor of autophagic clearance of proteins. Lysosomal dysfunction is acknowledged as a central catalyst of neurodegeneration (Nixon, 2020), and, consequently, lysosomal activity is a new pharmacological target.

Support for this pathway comes from observed mutations or perturbations in autophagy pathway genes in various neurodegenerative diseases. For example, genes implicated in Parkinson’s disease (e.g., PARK2, ATP13A2, PINK1, GBA) encode lysosomal proteins or regulators of endolysosomal trafficking (Abeliovich and Gitler, 2016). In progressive supranuclear palsy, part of the FTD complex of neurodegenerative disorders, GWAS have implicated genes for components of the ubiquitin–proteasome system, including TRIM11 (Jabbari et al., 2018), a gene that also has been shown to be down-regulated in AD (Zhang et al., 2023d). While not fully understood, APOE4 is thought to contribute to dysregulation in the endosome–lysosome and autophagy pathways, which may mediate increased AD risk in APOE4 carriers (Asiamah et al., 2024).

Protein degradation and clearance mechanisms (e.g., stabilization of lysosomal function) are becoming prime pharmacologic therapeutic targets as they have the potential to affect multiple proteinopathies simultaneously (Knopman et al., 2021), which could be especially promising for those with multiple brain pathologies. This is in contrast to approaches such as a focused anti-amyloid or anti-tau immunotherapy that target only one type of protein aggregate.

Mitochondrial and Metabolic Dysfunction

Mitochondrial function is closely intertwined with cell viability, overall cellular function, and the various hallmarks associated with aging. Mitochondria not only play a critical role in cellular respiration—using oxygen to extract, transfer, and generate energy from such molecular substrates as glucose, fat, fatty acids, and amino acids—they are also involved in maintaining calcium and iron balance, regulating cell proliferation and

cell death, facilitating cell signaling, and ensuring proteostasis (Gonzales et al., 2022; Lima et al., 2022). Neurons have a high metabolic demand, and dysfunctional mitochondria disturb the energy supply of neurons. For neurons and glial cells to function optimally, a plentiful and uninterrupted supply of adenosine triphosphate (ATP) is needed, which is best accomplished by oxidative phosphorylation of glucose. As compared to healthy aging, glucose consumption is reduced (i.e., hypometabolism) in AD/ADRD (Kato et al., 2016; Yan et al., 2020). Using fluorodeoxyglucose positron emission topography, disruption to cerebral glucose metabolism in specific brain regions has been detected (Bohnen et al., 2012; Kato et al., 2016; Yan et al., 2020). Reduced glucose utilization can occur prior to the presentation of clinical symptoms as demonstrated in a study involving participants with mild cognitive impairment due to AD (Drzezga et al., 2003; Yan et al., 2020). A longitudinal study assessing data from 128 participants found that glucose uptake in the precuneus, an area of the brain in which there is early amyloid deposition, is reduced in patients 10 years prior to the development of symptoms (Bateman et al., 2012; Yan et al., 2020).

Neurogenic glucose metabolism is perturbed by impaired insulin signaling, resulting in characteristics of AD that mirror the pathophysiology of non-CNS tissues in type 2 diabetes mellitus (Duelli and Kuschinsky, 2001). Moreover, reductions in glucose transporters GLUT1 and GLUT3 were observed in AD patient brains (An et al., 2018) and has been associated with a reduction in brain glucose consumption and cognitive impairment (Landau et al., 2010; Yan et al., 2020). Brain insulin resistance is acknowledged as an early and common feature of AD and can be closely linked to cognitive decline. In fact, there are many pathogenetic connections between brain disorders—including neurodegeneration—diabetes, and insulin resistance (de Galan, 2024).

Oxidation end products are pathological hallmarks of a variety of neurodegenerative pathologies. Mitochondria are a key source of oxidative stress resulting from respiratory chain activities and potential generation of superoxide radicals. Impairment of the activity of mitochondrial respiratory chain complexes are reported in AD, PD, and ALS (Golpich et al., 2017). Sublethal but progressive changes in mitochondrial function have been proposed to drive aging (i.e., the free radical theory of aging) (Harman, 1956) and AD (i.e., the “mitochondrial cascade” hypothesis of AD) (Swerdlow and Khan, 2004). Destabilized mitochondria create more free oxygen radicals than can be buffered sufficiently by the intrinsic antioxidant systems. A previous study reported that impaired mitochondrial function resulting from mitochondrial DNA (mtDNA) damage tends to increase with age, and levels of mtDNA damage were increased in individuals with AD as compared to age-matched controls (de la Monte et al., 2000). Single-cell

analyses have revealed an increase in mtDNA deletions specifically within neurons affected by AD (Krishnan et al., 2012).

In experiments where mtDNA is transferred from donor cells to cells with identical nuclear DNA but lacking mtDNA, it has been demonstrated that mtDNA from individuals with AD is responsible for subtle variations in mitochondrial morphology, biogenesis, membrane potential, oxidative stress, and calcium buffering capacity (Swerdlow et al., 2017). Such distinct mitochondrial characteristics seen in peripheral tissues of individuals with AD, as opposed to control groups, implies that changes in mitochondrial status relevant to the brain can potentially be detected and monitored in peripheral samples. These findings suggest that mitochondrial dysfunction, which may be a cause rather than a result of AD neuropathology, could be an early event in the disease onset and progression (Theurey and Pizzo, 2018; Wilkins and Swerdlow, 2021).

Disturbed Lipostasis and Altered Lipid Metabolism

The brain has an abundance of lipid-rich myelin. Changes in the lipid composition of the brain are observed in both the natural process of aging and various neurodegenerative diseases (Gonzales et al., 2022). Genetic studies—including genetic linkage studies, GWAS, and exome sequencing studies—have repeatedly linked genes/loci related to lipid metabolism (including APOE) with AD (Andrews et al., 2020; Gonzales et al., 2022; Jones et al., 2010; Sims et al., 2020), suggesting a role for aberrations in lipid transport, lipid processing, and metabolism. The field of lipidomics, which involves the investigation of “pathways and networks of cellular lipids in biological systems” (Gonzales et al., 2022), has uncovered AD-associated lipid profiles (Han, 2010; Naudi et al., 2015). For instance, multiple research groups have consistently reported the early accumulation of ceramide levels in brains affected by AD (Cutler et al., 2004; Han et al., 2002). In contrast, sulfatides, a group of sulfoglycolipids highly enriched in myelin, have been shown to be significantly reduced at the earliest clinical stages of AD (Cheng et al., 2013; Gonzales et al., 2022; Han et al., 2003; Irizarry, 2003). Similarly, phospholipid plasmalogen was found to be reduced in the brains of individuals with AD and is associated with disease severity (Goodenowe et al., 2007).

Although the role of lipid metabolism in the pathogenesis of AD/ADRD is still under study, one interesting avenue of lipidomics research is the potential application of lipid mediators to resolve inflammation associated with proteinopathy. Free forms of proresolving lipid mediators made from the enzymatic oxygenation of polyunsaturated fatty acids have been found to be deficient in AD as a result of depletion in the pool of bound or esterified lipid mediators. Improving understanding of the regulation of

lipid esterification and release could suggest effective therapeutic strategies for inflammation resolution (Shen et al., 2022; Taha, 2024).

Cellular Senescence

Cellular senescence can be defined as “a cell state triggered by stressful insults and certain physiological processes, characterized by a prolonged and generally irreversible cell-cycle arrest with secretory features, macromolecular damage, and altered metabolism” (Gorgoulis et al., 2019, p. 814). Senescent cells employ antiapoptotic pathways and halt the cell cycle as a means to evade cell death. Additionally, these cells release various substances such as proinflammatory cytokines, chemokines, growth factors, extracellular remodeling proteins, and other signaling molecules, collectively known as the senescence-associated secretory phenotype (SASP) (Gorgoulis et al., 2019). In the absence of efficient senescent cell clearance, SASP can lead to tissue damage, cell death, or even the conversion of other cells into a senescent state as a propagating phenotype (Yousefzadeh et al., 2020). As people age, senescent cells are found to accumulate throughout the body, including the brain (Yousefzadeh et al., 2020).

Nerve cells are aging over the lifetime and are only stable and functional if the nerve-cell intrinsic defense, repair, and recycling mechanisms are adequately working. Such repair mechanisms include local (axonal, synaptic, soma compartment) protein recycling pathways (see the section above on proteostasis and autophagy) but also DNA surveillance pathways. In postmortem human brain tissues of AD, various senescent cell types have been identified, including astrocytes (Bhat et al., 2012), neurons (Dehkordi et al., 2021; Musi et al., 2018), microglia (Streit and Xue, 2014), oligodendrocyte precursor cells (Zhang et al., 2019), and endothelial cells (Bryant et al., 2020). Unbiased single-cell transcriptomics analysis conducted on the dorsolateral prefrontal cortex of individuals with AD has shown that excitatory neurons are a predominant senescent cell type in the brain and are correlated with the coexistence of intracellular neurofibrillary tangles (Dehkordi et al., 2021). In healthy aging brains, however, analyses of transcriptomics data derived from bulk tissue revealed that the prevailing senescent cell types in the brain were endothelial cells and microglia (Xu et al., 2022). This finding underscores the potential disparities in senescent cell types observed in healthy versus diseased states, but it could also be attributable to the nature of the analyzed material and variations in the predetermined criteria used to define senescence.

While tracing the cause of brain senescent cell emergence in humans is not currently feasible, studies conducted on rodents revealed that such emergence could be triggered by the endogenous accumulation of tau (Bussian et al., 2018; Musi et al., 2018) or amyloid beta protein (Zhang et

al., 2019), insulin resistance (Chow et al., 2019), and immune dysfunction (Yousefzadeh et al., 2021). Contributing exposomal factors may include a high-fat diet (Ogrodnik et al., 2019), chronic ethanol exposure from alcohol consumption (Sun et al., 2023), chronic unpredictable stress (Lin et al., 2021), and brain injury (Arun et al., 2020). Further elucidation of the role of cellular senescence may be achieved through comparative studies, including widescale multiomics studies, of brain regions that are normally spared in AD/ADRD (e.g., resistant) and vulnerable regions. Knowing that age is the prime risk factor for the most prominent neurodegenerative disorders in humans, the effect of age-related changes and the senescence status of different brain cell types need to be understood in more detail.

The Cellular Senescence Network (SenNet) is a collaborative research effort launched by the NIH Common Fund to comprehensively identify and characterize the heterogeneity of senescent cells in multiple human (and other model organism) tissues, in various states of health, and across the lifespan. Ultimately, the network aims to develop public atlases of these cells and create tools and technologies (e.g., biomarker panels, computer and experimental model systems, imaging tools) that build from other Common Fund programs to advance the development of therapeutics (NIH, 2024; SenNet Consortium, 2022).

Epigenetic Dysregulation

DNA methylation in the brain trends toward global decreases with aging, which has been associated with enhanced gene expression (De Jager et al., 2014; Pellegrini et al., 2021). The pattern and frequency of epigenetic changes—in particular, DNA methylation at cytosine-guanine pair sites—have been proposed as a mechanism for comparing chronological and biological age (i.e., an epigenetic clock) (Gonzales et al., 2022). Differential methylation patterns have been observed in genes associated with AD, such as APP and MAPT, among others, when comparing individuals with AD to control subjects (De Jager et al., 2014; Iwata et al., 2014; Yu et al., 2015). As age advances, increases in chromatin remodeling and chromatin heterogeneity are also observed. In addition to DNA methylation, histone acetylation typically decreases with aging, resulting in a chromatin structure that is more condensed and consequent changes in transcription (Gonzales et al., 2022; Peleg et al., 2016). By analyzing postmortem human brain tissue, a recent study uncovered an increase in the expression of two histone acetyltransferases, which were associated with both amyloid beta pathology and neurodegeneration (Nativio et al., 2018, 2020).

Alterations in levels of multiple mitochondrial RNAs have been found in AD cases, which may play a role in multiple aspects of pathology, such as modulating amyloid beta and tau production and function, synaptic

plasticity, neuronal growth, apoptosis, and inflammatory response. Mitochondrial RNAs may therefore be useful as potential biomarkers and therapeutic targets (Zhang and Bian, 2021).

Understanding the relationships between the exposome, age, epigenetic and posttranslational changes, and such age-related diseases as AD/ADRD holds promise for disease prediction (Grodstein et al., 2020; Shireby et al., 2020) and intervention efforts. As potentially reversible contributors to AD/ADRD, epigenetic and posttranslational alterations are appealing targets for therapeutic interventions (Hara et al., 2019; Migliore and Coppedè, 2022). Realizing such potential will require the systematic analysis and comparison of epigenetic and posttranslational changes in aged and diseased (AD/ADRD) brain tissue and identification of areas of overlap.

Toward an Integrative Biological Framework for AD/ADRD

As illustrated in Figure 3-3 above and described throughout this section, the process of neurodegeneration is influenced by a number of cell types and pathogenic processes. Importantly, the factors and mechanisms affecting the onset and progression of disease act along the life course, meaning they occur to different extents as the brain ages, and may also interact with each other. These processes (e.g., inflammation, vascular dysfunction, altered proteostasis and autophagy, mitochondrial and metabolic changes) and cellular players (e.g., neurons, endothelial and glial cells) comprise a complicated biological framework for neurodegeneration. It is increasingly acknowledged that age-associated neurodegenerative diseases, such as AD/ADRD, are the result of complex and multifactorial events. However, all the pathways described are disease relevant and need to be taken into account when creating a more integrative understanding of events taking part in age-related neurodegeneration (Behl, 2023; Lehmann et al., 2023).

By enabling better discernment of the initial triggers, cascading pathophysiologic sequences, and biological systems working to counteract those perturbations to restore homeostasis, such a model could potentially guide combination approaches to AD/ADRD interventions that sequentially target key steps in the etiologic cascade and that integrate age-dependent changes. For example, nonpharmacological interventions implemented early in life for individuals at risk for dementia (e.g., based on genetic and exposome risks) could be followed later in life with an anti-inflammatory therapy, and potentially, a pathology-specific (e.g., amyloid) immunotherapy to avoid microglia activation. Such a combination approach may be necessary to achieve effect sizes greater than those observed to date for strategies targeting a single pathology and pathway.

Some progress has been made in the development of integrated models for molecular and cellular pathways contributing to neurodegeneration

(Balusu et al., 2023; De Strooper and Karran, 2016; Gan et al., 2018). In such models, a biochemical phase features perturbations in normal cellular processes that contribute to the protein aggregation commonly observed in neurodegenerative diseases; protein aggregation, on the other hand, is not part of the pathological process in dementia types based on vascular changes (vascular dementia). The biochemical phase is followed by a cellular phase during which cells in the brain react to the protein aggregates. A key element of the cascade proposed here is the generation of an innate inflammatory reaction mediated by microglia, which triggers reactions from other cells and ultimately leads to neuronal cell death. The resulting damage to the brain, some potentially irreversible, leads to the clinical phase (cognitive impairment and clinical dementia) (De Strooper and Karran, 2016; NASEM, 2024).

While described linearly for simplicity, within this cascade there are several feedback mechanisms, adding to the model complexity, as depicted in Figure 3-3 earlier in this chapter. On the other hand, this proposed sequence leaves behind the neuron-only modeling of neurodegeneration and considers also the effect of other nonneuronal cell types. As neuroinflammatory endophenotypes and biomarkers for different biochemical pathways are developed, it may be possible to further unravel the temporal relationship between inflammatory responses and other pathophysiological aspects of AD/ADRD (Arafah et al., 2023; Hampel et al., 2020).

While the aforementioned cascade provides a useful starting place for understanding the interdependent molecular and cellular processes contributing to AD/ADRD, a comprehensive model needs to account not only for pathologic factors occurring within the brain, but also the role of the exposome, brain–peripheral body interactions, and mechanisms of resilience.

The Role of the Exposome and Brain–Peripheral Body Interactions

Upstream of the intracellular biochemical events leading to the different pathogenic processes are the contributions of the exposome to pathogenesis, which need to be incorporated into an integrative model. While it is clear that genetic variants associated with AD/ADRD can increase susceptibility to and/or accelerate many of the hallmarks of aging, the contributions of exposome factors, mediated either through gene–environment interactions or through direct biochemical and/or cellular effects, have not been well characterized and are important for understanding differences in AD/ADRD across diverse populations. As discussed earlier in this chapter, efforts are needed to identify exposure-related pathways that modulate or converge with other AD/ADRD risk factors or pathways to mediate pathogenesis. For example, air pollution may alter the blood–brain barrier (Oppenheim et al., 2013) and has also been shown to reduce levels of

esterified lipid mediators, which, as noted earlier, are involved in resolving inflammation and have been found to be deficient in AD (Shen et al., 2022; Taha, 2024), indicating the potential for synergistic deleterious effects.

A growing body of evidence suggests that infections, both CNS specific and systemic, can increase the risk of dementia and that these effects may be mediated in part by inflammation, although pathogen-specific pathways may also be involved (Itzhaki, 2021; Sipilä et al., 2021; Vestin et al., 2024), particularly for pathogens that can infect the brain, such as herpes simplex virus (Itzhaki, 2021) and SARS-CoV-2 (Liu et al., 2023b). Furthermore, behavioral factors such as diet and physical activity contribute to cardiovascular risk factors, such as hypertension, diabetes, and obesity (Gonzalez et al., 2024; Marquez et al., 2024) and are increasingly understood to contribute to neurodegeneration through vascular pathways (NASEM, 2018), but behavioral factors may also affect AD/ADRD risk through epigenetic effects (Maloney and Lahiri, 2016; Migliore and Coppedè, 2022).

The potential for extra-CNS infections to increase dementia risk and the well-established cardiovascular risk factors for dementia underscore the need for brain–peripheral body interactions to be represented in a more holistic integrated model (Hampel et al., 2019). AD/ADRD are not diseases exclusive to neurons but involve the whole body since the brain and the (rest of the) body are tightly connected through metabolic, immune, and vascular systems (Hampel et al., 2019). As discussed earlier in this chapter, vascular dysfunction may be linked to such neurodegenerative mechanisms as neuroinflammation and impeded clearance of protein aggregates. Thus, targeting cardiovascular risk factors in the periphery (e.g., atherosclerosis, hypertension) has the potential to affect multiple potential AD/ADRD pathways in the brain in a disease-agnostic manner, in addition to effects on cardiovascular diseases such as heart disease and stroke (preventive interventions targeting cardiovascular risk factors are discussed in Chapter 4).

Another notable example of such brain–peripheral body interactions is the gut microbiome and its potential contribution to AD/ADRD through the gut–brain axis. Microbial metabolites may exert effects on the brain through multiple pathways (e.g., immune, endocrine, enteric nervous system), and are thought to have a wide range of effects including altering neuronal transcription, compromising the blood–brain barrier integrity, and triggering neuroinflammation (Wang et al., 2021). Ongoing research is seeking to elucidate how gut microbiota interact with certain risk gene settings and exposome factors, such as air pollution (Dutta et al., 2022), to influence the onset and/or progression of AD/ADRD. In addition to the gut–brain axis, connections via the lung–brain and renal–CVD–brain axes have also been studied (Finch and Kulminski, 2019; Owaki et al., 2024).

and has been identified as a therapeutic target for AD (Hurst et al., 2023; Zammit et al., 2023). Cellular studies aiming to define resistance and adaptive factors that provide protection against neurodegeneration-associated challenges and stress (e.g., oxidative and endoplasmic reticulum stress) may complement in vivo investigations (Chakraborty et al., 2019; Pham et al., 2023). To advance precision approaches to AD/ADRD interventions, resilience-focused research and an integrative biological framework for AD/ADRD need to consider the interplay of biological and sociocultural factors and the consequent potential for heterogeneity in resilience determinants and trajectories across different population groups. An important example is the observation of sex and gender differences in resilience, which may stem from myriad interrelated factors requiring further study, including neuroanatomy, genetic architecture, and influences of gender roles on modifiable risk factors (Arenaza-Urquijo et al., 2024).

Targeting resilience factors represents an appealing alternative to targeting individual AD/ADRD pathologies, especially given the prevalence of mixed pathologies. In the future, a cognitive resilience index

based on the ability to measure known resilience factors could facilitate a precision medicine approach to interventions, which may include combinations of resilience behaviors, such as physical activity, and novel resilience therapies aimed at maintaining brain health and late-life cognitive function.

Model Systems and Other Tools for Elucidating Molecular and Cellular Pathways Leading to AD/ADRD

Considering the various and interrelated molecular and cellular pathways discussed in the sections above, one central question remaining is how to model complex brain diseases that develop over decades and are affected by various sets of risk genes, as well as their interplay with peripheral organs (i.e., system health) and the environment (i.e., exposome). NIH has made substantial investments to support the creation of model systems to serve as an experimental basis for studying disease pathogenesis and facilitating drug discovery. As no single model is able to fully recapitulate the human condition, and each has its strengths and weaknesses, a variety of in vivo and in vitro models have been developed to answer different kinds of scientific questions. Thus, it is important to adequately define the key questions at the start of a scientific endeavor so they may drive the development and selection of an appropriate model. In the understandable absence of an ideal model, the use of multiple models in combination can enhance the robustness of the experimental findings.

___________________

⁸ Further information is available at https://www.model-ad.org/ (accessed June 5, 2024).

AD. Ultimately, the aim is to integrate these models into preclinical testing pipelines to identify and evaluate novel therapies (MODEL-AD, 2024; NIA, 2023b). MODEL-AD, like most animal model development to date, has primarily focused on AD, with a heavy bias toward amyloid pathways, although some models for LBD (Lim et al., 2011; Magen and Chesselet, 2011; Sander et al., 2023; Stylianou et al., 2020) and FTD (Ahmed et al., 2017; Kashyap et al., 2023; Roberson, 2012) have been created. To continue to expand the therapeutic pipeline to other nonamyloid targets and related dementias, there is a need for further investment in the creation and characterization of models that can inform those efforts, including mixed-pathology models. Animal models are also critical to understanding synaptic and cellular contributors to resilience (de Vries et al., 2024; Neuner et al., 2022).

In addition to mice, other animals with the potential to model AD/ADRD include fruit flies (Trotter et al., 2017); rats; dogs, which naturally develop amyloid plaques with age (Noche et al., 2024); and nonhuman primates (NHPs) (Padmanabhan and Gotz, 2023). Rats are commonly used in pharmacokinetic studies but also may be superior to mice in modeling inflammation (Du et al., 2017; Seok et al., 2013) and complex cognitive behaviors. NIA’s intramural research program is conducting a longitudinal study of neurocognitive aging in rats—Successful Trajectories of Aging: Reserve and Resilience in RatS (STARRRS)—which will generate shareable resources of phenotypic and neuroimaging data, as well as biological samples (NIA, 2024c)

As compared to rodents, NHPs are more similar to humans with regards to neuroanatomy and neurophysiology (NASEM, 2023), which may make them useful models for neurodegenerative diseases such as AD/ADRD. Some NHPs (great apes, macaques) develop amyloid plaques with age but they do not appear to regularly spontaneously develop other features of human AD pathology that lead to dementia (e.g., neurofibrillary tangles) (Beckman et al., 2021). Consequently, there is interest in genetically engineering more human-relevant NHP models for AD/ADRD. NIH recently funded the Marmosets As Research Models of AD (MARMO-AD) program, which is exploring transgenic marmosets as models to study primate‐specific cellular and molecular mechanisms that contribute to the pathogenesis and progression of AD (Sukoff Rizzo et al., 2023). Genetic engineering of new NHP models is still in its infancy, however, and may be slowed by the high cost and logistic hurdles of working with NHPs, as well as ethical considerations (Padmanabhan and Gotz, 2023).

Consistent with NIH policy requirements (NIH, 2015), NIH-funded AD/ADRD studies using animal models need to consider sex as a biological variable, beginning with the development of research questions and extending to analysis of results and reporting of findings (e.g., sex disaggregated).

In doing so, preclinical animal studies can help to improve understanding of sex-specific differences in AD/ADRD (Arenaza-Urquijo et al., 2024).

In Vitro Models

To complement AD/ADRD research using animal models, there has been increasing focus on the development of in vitro models derived from human cells in the hopes of generating more human-relevant findings. Cell-based systems for studying human diseases were revolutionized in the early 2000s by the discovery of human induced pluripotent stem cells (iPSCs) (Yamanaka, 2012), which can be used to generate most human tissue cell types. Patient-derived iPSCs have been used to develop human brain organoids and brain microphysiological systems (i.e., tissue chips), which are being explored as potential model systems to study AD/ADRD. The generation of iPSCs from participants involved in cohort studies provides opportunities to link data from the cell-based models to longitudinal clinical data (NIA, 2022a).

Additional and further advanced in vitro systems that can recapitulate patient-like pathophysiology are emerging and represent potential alternatives to conventional animal- and cell-based models (Amartumur et al., 2024). For example, brain or cerebral organoids are three-dimensional (3D) cellular aggregates that when generated from human iPSCs exhibit some similarities to the structural and functional features of specific regions of a human brain, including neural circuit formation and neuroimmune and neurovascular interactions. Such models have been used to compare amyloid clearance by microglia carrying the APOE4 or the APOE3 allele (Paşca, 2019) and to elucidate early pathogenic events resulting from mutations in the MAPT gene that gives rise to an inherited form (autosomal dominant) of FTD (Bowles et al., 2021). As their architecture and functionality more closely resemble an embryonic brain, brain organoids may generally be better suited to the study of neurodevelopment than age-related neurodegeneration (Paşca, 2019).

Microphysiological systems are similarly 3D cellular systems intended to recapitulate specific aspects of human physiology but differ from organoids in that they are formed on microfluidic chips, giving researchers more control over the development of the composition and architecture of the tissue and the ability to model such features as vascular perfusion (Hargrove-Grimes et al., 2021). Coculture systems that can enable the investigation of complex cellular interactions (e.g., among neurons, astrocytes, microglia, endothelial cells) (Cetin et al., 2022) have the potential to help answer pressing scientific questions regarding the perturbations in cellular activity that lead to the loss of homeostasis in the brain (De Strooper and Karran, 2016). For some scientific questions regarding AD/ADRD biology, such coculture systems

may better model in vivo conditions and have greater translational relevance than cellular models involving only a single cell type (Cetin et al., 2022). Furthermore, the ability to connect chips with different tissues in a modular fashion can enable the study of interactions of peripheral body organs with the brain. A workshop hosted by NIH in 2022 focused on opportunities for microphysiologic systems to serve as alternatives to animal models for AD/ADRD and their potential role in drug discovery and development (NIA, 2022a). This workshop highlighted examples of efforts underway to develop 3D cell-based models for AD/ADRD including an organoid model for AD and tissue chip models for FTD and the blood–brain barrier.

An important limitation of iPSC-derived organoid and microphysiologic models is that the process of reprogramming patient-derived cells into iPSCs results in the loss of their somatic epigenetic patterns, including aging, exposure, and disease-related markers (Balusu et al., 2023). A promising complementary approach that may circumvent this limitation is induced neuron (iN) models, which are generated by converting already differentiated cells, commonly fibroblasts, into neurons (Balusu et al., 2023; Traxler et al., 2022). In contrast to iPSC models, iN models retain to some degree (albeit not well characterized) the epigenetic profile and metabolic state of the patients they are derived from and demonstrate enhanced expression of mature neuronal markers and improved functionality (Balusu et al., 2023). Employing iN models may lead to a better understanding of neuronal state and fate changes, which in turn could guide intervention approaches to manipulation of cell fate with the goal of promoting neuronal resilience, repair, and, potentially, rejuvenation (Traxler et al., 2023). Further comparative evaluations of iPSC and iN models are needed to better characterize the advantages and limitations of each and their use as complementary systems.

Of course, in vitro models are not suited to some aspects of AD/ADRD research, such as elucidating the underlying effects of such factors as exercise, hypertension, psychological stress, sociobehavioral factors, and sleep, which generally require the study model to be an integrated organism. Other considerations regarding findings from human-derived cell models include potential variability resulting from such factors as differences in cell sources, genetic backgrounds, and culture methods (including their ability to mimic physiologic conditions) (Slanzi et al., 2020), as well as how representative existing models are of the population and different dementia phenotypes.

Single-cell sequencing of samples from brains collected postmortem from individuals with AD/ADRD have the potential to generate data that are complementary to cell-based models, such as uncovering affected neuronal and nonneuronal cell types, vulnerable brain regions, and changes (e.g., somatic DNA mutations, altered gene expression) that precede neuronal dysfunction and death (Balusu et al., 2023; Miller et al., 2022). Advances in

multiomics approaches are enabling the analysis of genomic, transcriptomic, metabolomic, lipidomic, and epigenomic information from single cells (Vandereyken et al., 2023), providing new opportunities to elucidate molecular pathways leading to dementia, especially when coupled with emerging spatial multiomics methods. In addition, multiomics approaches in combination with GWAS and genetic single-nucleotide polymorphism studies have the potential to define “a cluster of core alterations in AD” and can advance a more holistic approach in AD/ADRD research (Argyriou et al., 2024). As discussed later in this chapter, findings from studies using unbiased GWAS and multiomics approaches require significant investment in follow-on basic, mechanistic research (including studies using in vivo models) to understand their biology and how they might be targeted therapeutically.

One challenge with the high-dimensional analyses of single cells from postmortem brain samples is that the disease has generally already progressed to late or end stage and data can only be captured from the more resistant, surviving neurons. To address these limitations, data from single-cell analyses of human brain samples could be supplemented with single-cell multiomics data from animal models, which may provide insights into earlier-stage dysfunction (Balusu et al., 2023). Of note, the Brain Research Through Advancing Innovative Neurotechnologies® (BRAIN) initiative recently released a comprehensive single-cell spatial map of the mouse brain and identified thousands of cell types, underscoring the importance of leveraging large-scale resources developed by other institutions.

ACCELERATING THE TRANSLATION OF BASIC, EPIDEMIOLOGICAL, AND SOCIAL SCIENCES RESEARCH FOR THE DEVELOPMENT OF INTERVENTION STRATEGIES

As described in this chapter, diverse streams of epidemiological, social science, and mechanistic research have been pursued, funded by NIH and others (Stephan et al., 2024). An expansion beyond this breadth of inquiry is appropriate given the breadth of the exposome and the longitudinal and molecular complexity of pathways that lead to AD/ADRD. All of these factors contribute to the individuality of this group of neurodegenerative diseases, as increasingly supported by recent advances in biomarkers and subtyping studies (Tijms et al., 2024), as discussed further in Chapter 4.

Advances in the understanding of the risks associated with different types of dementia and the molecular and cellular pathways that mediate those risks need to be accompanied by efforts to translate this new knowledge into intervention strategies. These efforts can be guided by advancements in blood-based or other biomarkers to stratify risks into all-cause dementia, AD, or related dementia risks.

Translational Epidemiology Approaches: Moving from Risk Factors to Public Health Interventions and Back to Stratified Epidemiology

Research to guide interventions must evaluate causation, and yet for decades the statistical training typically offered to clinical researchers has emphasized “risk factor” epidemiology (i.e., the identification of correlates of disease without rigorous evaluation of causality). Revolutionary progress in conceptualization and methodology for causal inference started in the 1980s and created substantial debates in clinical research (Pearl, 2009). Technical tools to support causal inference in the absence of formal randomization continue to improve, alongside rigorous frameworks for recognizing the limitations of evidence from randomized and nonrandomized studies (Hernán, 2018). This progress is highly relevant to AD/ADRD research and yet very few researchers have adopted these tools. The slow adoption is attributable to a combination of statistical difficulty, the limited formal training available to many researchers (the field is deeply interdisciplinary), or training that occurred prior to the causality revolution and thus did not incorporate the tools. As a result, the quality of evidence available even for routine questions that have been tackled in countless studies, such as the effects of alcohol use on AD/ADRD, or common medications, remains thin.

Many observational studies are vulnerable to identical biases, so they offer little new evidence to complement one another. More formal frameworks to integrate evidence across study designs, populations, and data types are needed. This includes work on measurement harmonization and cross-walking, meta-analyses and meta-regression, and evidence triangulation (Lawlor et al., 2016; Thurmond, 2001). For example, randomized controlled trials are not always or even often feasible, especially for preventive interventions for a disease that develops over decades. As a result, conclusions about preventive strategies rest heavily on studies that were not intentionally designed randomized controlled trials.

There are two general strategies for evaluating causality in these settings: (1) control for confounders that might influence the exposure and also AD/ADRD or (2) identify a quasi-random source of variation in exposure and evaluate how that influences AD/ADRD risk (Matthay et al., 2020). Combining evidence from both approaches offers a far stronger basis for causal conclusions. Additional opportunities for evidence triangulation include

identifying settings where the confounders are likely to be entirely different or finding negative control outcomes. These approaches align well with the policy emphasis in Chapter 4, highlighting the strategies of public and institutional policies to translate epidemiological findings into public health impact for prevention of AD/ADRD. Policies are often directly amenable to evaluation via quasi-experimental approaches, for example, leveraging approximately random variation in the timing of policy adoption and implementation. Major progress already achieved in declining age-specific incidence rates in many settings suggest rigorous evaluation of the changes that account for those declines may point the way to future breakthroughs.

In addition, as blood tests for AD/ADRD are implemented in real-world settings, the availability of biomarker data will enable epidemiological research to stratify the population and differentiate risk factors for AD from those for related dementias. This will enable the development of tailored interventions in both public health and therapeutics.

Opportunities to Accelerate Discoveries into Interventions for AD/ADRD

At present there are more than 100 drugs being evaluated in clinical trials for effectiveness against AD (Cummings et al., 2023b), many targeting pathways not specific to any one AD/ADRD pathology, such as those described earlier in this chapter. More than 150 trials are evaluating nonpharmacological interventions (Kelley, 2023). Despite hope for each approach, the reality is that most phase 2 trials will fail. Thus, improving our ability to learn from failures, as well as improving the probability of success for the pharmacological interventions entering the clinic, needs to be a research priority as it has the greatest potential to improve the delivery of treatments for AD/ADRD. Negative clinical trials need to be analyzed to identify points of failure in translation and the potential causes, and this information needs to be fed back into earlier phases of the research and development pipeline, as discussed further in Chapter 4. However, this can only occur if trials are well designed and have specific and validated target engagement or pharmacodynamic biomarkers informing upon the hypothesized mechanism of action. Currently only 44 percent of clinical trials have fluid biomarkers fulfilling this role and the numbers drop in industry-sponsored trials (Marlies Oosthoek, 2024), and even fewer have imaging measures because of the associated costs.

This is a significant barrier to the translation of clinical research back to basic mechanisms.

Success of clinical drug discovery research can be seen two ways. First, success is achieving a positive primary outcome providing evidence of efficacy on an accepted endpoint. This type of success leads to phase 3 studies to demonstrate application across a larger and more diverse patient population. However, this does not occur often; instead, many drug discovery trials fail to meet their primary outcome. Outside of safety, the most important secondary outcome is that the intervention engages the mechanism and provides sufficient activity such that an efficacy response would have been observed if that mechanism was related to the disease. The majority of drug discovery trials do not invest sufficiently in this area and thus, despite the investment in clinical research, the field advances slowly and leaves many questions pertaining to why efficacy was not achieved.

The probability of success for pharmacological interventions entering the clinic is low and requires improvements in target diversity, validation, and enablement. NIH has intensified infrastructure investments in these areas and has developed several interconnected translational research programs that support the identification, validation, and enablement of targets as part of a precision medicine approach to drug development for AD (see Figure 3-6) (NIA, 2023b). These strategic investments in the earliest phases of drug discovery and preclinical research have resulted in the generation of large and diverse (e.g., genetic, exposome, functional multiomics) datasets that can be used for the identification of novel targets.

The Accelerating Medicines Partnership® (see Box 3-4), a cross-sector partnership launched in 2014 and managed by the Foundation for the NIH has used several avenues to identify pathways and nodes specific to certain cell types that are affected in AD (NIA, 2024d). The Target Enablement to Accelerate Therapy Development for Alzheimer’s Disease (TREAT-AD) centers function to bridge the efforts of AMP-AD and those to enable novel targets through early drug discovery using animal models developed

through MODEL-AD (see Box 3-5). The Centrally Linked Longitudinal Peripheral Biomarkers of Alzheimer’s Disease in Multiethnic Populations (CLEAR-AD) program is exploring through a diverse set of samples centrally linked and longitudinal peripheral biomarkers to enable personalized medicine approaches (see Box 2-6). Coupled with the investments in the Alzheimer’s Disease Research Centers, these programs function to derisk novel targets and facilitate their translation toward clinical development (NIH, 2019) or the Small Business Innovative Research program. Considering the comparatively small number of clinical trials for dementia relative to other high-burden diseases such as cardiovascular disease and cancer (see Chapter 1), these programs are critical to enabling the expansion of early-phase trials of preventive and therapeutic interventions for AD/ADRD, as discussed further in Chapter 5.

However, the importance of considering both AD and related dementias cannot be overstated. As the field has expanded beyond amyloid-targeting mechanisms, investigators have found that the mechanistic hypotheses for

these conditions often overlap (as discussed earlier in this chapter). This overlap suggests a shared pathophysiology, which could potentially be targeted by similar therapeutic strategies. NIH-funded programs, such as AMP-AD, TREAT-AD, MODEL-AD, and CLEAR-AD, can be instrumental in this regard with the expansion of their mandate beyond AD and into related dementias. Where parallel programs for related dementias already exist (e.g., AMP-PDRD), opportunities for integration should be pursued to break down silos (discussed further in Chapter 5). This expansion and integration would not only enhance the understanding of shared disease mechanisms but also facilitate the development of novel therapeutic targets and personalized medicine approaches. By considering and prioritizing mechanisms that might have relevance across AD/ADRD and supporting

research to develop mechanism-specific translational biomarkers, these programs can help to diversify and validate potential targets, thereby increasing the likelihood of successful clinical trials. This comprehensive approach is crucial for advancing the fight against these debilitating conditions.

A well-recognized barrier to advancing interventions for AD/ADRD and many other diseases is the so-called valley of death—the large gap that spans early-stage scientific research (generally in academic settings) that provides mechanistic insights and identifies promising therapeutic targets and the later-stage, lower-risk programs that are more attractive to pharmaceutical companies and investors (Parrish et al., 2019). As a result of the valley of death, many potentially effective therapeutic strategies are not pursued. Strategies to bridge the valley of death can include

• establishing challenge programs to spur innovation and the acceleration of translational research;
• strategically targeting public or philanthropic funding to de-risk novel intervention approaches to incentivize private-sector investment and advance research on strategies that the pharmaceutical industry may not have an interest in, such as repurposing generic drugs (Smith, 2024); and
• creating public–private partnerships, which can draw on the talents of academia in discovery and those of industry in bringing therapies to market (Atkinson et al., 2023; HHS, 2012).

Partnerships also have a role in accelerating translational research outside of the drug development space, as discussed further in Chapter 5. Partnerships between different public entities (e.g., NIH partnership with housing or education departments) can facilitate more cross-disciplinary

approaches and better leverage existing investments that may affect brain health and AD/ADRD (Kremer, 2024).

RESEARCH PRIORITIES

The advancement of knowledge of risk and protective exposures for diverse populations, elucidation of the complex biological basis of AD/ADRD, and the rapid translation of basic and epidemiological research discoveries to clinical trials are essential to the development of effective strategies for the prevention and treatment of AD/ADRD. In recent years, NIH has made significant investments in basic research focused on elucidating diverse molecular and cellular processes of AD/ADRD vulnerability and resilience (NIA, 2023d, 2024f). Infrastructure investments to support these efforts have included the creation of large-scale, collaborative research programs to support the identification of disease mechanisms; application of emerging analytic approaches; development of novel therapeutic targets and open-science tools; and mechanisms to support early-stage clinical research. The committee identified four research priorities that build upon these existing efforts and investments. These research priorities are listed in Table 3-2 and include key scientific questions and near-term research opportunities that would advance progress on the research priorities.

Morelli, F. Moreno, K. Morgan, T. Mosley, M. M. Nothen, C. Muchnik, S. Mukherjee, B. Nacmias, T. Ngandu, G. Nicolas, B. G. Nordestgaard, R. Olaso, A. Orellana, M. Orsini, G. Ortega, A. Padovani, C. Paolo, G. Papenberg, L. Parnetti, F. Pasquier, P. Pastor, G. Peloso, A. Perez-Cordon, J. Perez-Tur, P. Pericard, O. Peters, Y. A. L. Pijnenburg, J. A. Pineda, G. Pinol-Ripoll, C. Pisanu, T. Polak, J. Popp, D. Posthuma, J. Priller, R. Puerta, O. Quenez, I. Quintela, J. Q. Thomassen, A. Rabano, I. Rainero, F. Rajabli, I. Ramakers, L. M. Real, M. J. T. Reinders, C. Reitz, D. Reyes-Dumeyer, P. Ridge, S. Riedel-Heller, P. Riederer, N. Roberto, E. Rodriguez-Rodriguez, A. Rongve, I. R. Allende, M. Rosende-Roca, J. L. Royo, E. Rubino, D. Rujescu, M. E. Saez, P. Sakka, I. Saltvedt, A. Sanabria, M. B. Sanchez-Arjona, F. Sanchez-Garcia, P. S. Juan, R. Sanchez-Valle, S. B. Sando, C. Sarnowski, C. L. Satizabal, M. Scamosci, N. Scarmeas, E. Scarpini, P. Scheltens, N. Scherbaum, M. Scherer, M. Schmid, A. Schneider, J. M. Schott, G. Selbaek, D. Seripa, M. Serrano, J. Sha, A. A. Shadrin, O. Skrobot, S. Slifer, G. J. L. Snijders, H. Soininen, V. Solfrizzi, A. Solomon, Y. Song, S. Sorbi, O. Sotolongo-Grau, G. Spalletta, A. Spottke, A. Squassina, E. Stordal, J. P. Tartan, L. Tarraga, N. Tesi, A. Thalamuthu, T. Thomas, G. Tosto, L. Traykov, L. Tremolizzo, A. Tybjaerg-Hansen, A. Uitterlinden, A. Ullgren, I. Ulstein, S. Valero, O. Valladares, C. Van Broeckhoven, J. Vance, B. N. Vardarajan, A. van der Lugt, J. Van Dongen, J. van Rooij, J. van Swieten, R. Vandenberghe, F. Verhey, J. S. Vidal, J. Vogelgsang, M. Vyhnalek, M. Wagner, D. Wallon, L. S. Wang, R. Wang, L. Weinhold, J. Wiltfang, G. Windle, B. Woods, M. Yannakoulia, H. Zare, Y. Zhao, X. Zhang, C. Zhu, M. Zulaica, EADB, GR@ACE, DEGESCO, EADI, GERAD, Demgene, FinnGen, ADGC, CHARGE, L. A. Farrer, B. M. Psaty, M. Ghanbari, T. Raj, P. Sachdev, K. Mather, F. Jessen, M. A. Ikram, A. de Mendonca, J. Hort, M. Tsolaki, M. A. Pericak-Vance, P. Amouyel, J. Williams, R. Frikke-Schmidt, J. Clarimon, J. F. Deleuze, G. Rossi, S. Seshadri, O. A. Andreassen, M. Ingelsson, M. Hiltunen, K. Sleegers, G. D. Schellenberg, C. M. van Duijn, R. Sims, W. M. van der Flier, A. Ruiz, A. Ramirez, and J. C. Lambert. 2022. New insights into the genetic etiology of Alzheimer’s disease and related dementias. Nature Genetics 54(4):412-436.

Bhat, R., E. P. Crowe, A. Bitto, M. Moh, C. D. Katsetos, F. U. Garcia, F. B. Johnson, J. Q. Trojanowski, C. Sell, and C. Torres. 2012. Astrocyte senescence as a component of Alzheimer’s disease. PLoS ONE 7(9):e45069.

Bhattarai, P., T. Iniyan Gunasekaran, M. E. Belloy, D. Reyes-Dumeyer, D. Jülich, H. Tayran, E. Yilmaz, D. Flaherty, B. Turgutalp, G. Sukumar, C. Alba, E. Martinez McGrath, D. N. Hupalo, D. Bacikova, Y. Le Guen, R. Lantigua, M. Medrano, D. Rivera, P. Recio, T. Nuriel, N. Ertekin-Taner, A. F. Teich, D. W. Dickson, S. Holley, M. Greicius, C. L. Dalgard, M. Zody, R. Mayeux, C. Kizil, and B. N. Vardarajanet. 2024. Rare genetic variation in fibronectin 1 (FN1) protects against APOEε4 in Alzheimer’s disease. Acta Neuropathologica 147(1):70.

Biomarker Network. 2024. Health and Retirement Study (HRS). https://gero.usc.edu/cbph/network/studies-with-biomarkers/the-health-and-retirement-study-hrs/ (accessed August 16, 2024).

Bohnen, N. I., D. S. Djang, K. Herholz, Y. Anzai, and S. Minoshima. 2012. Effectiveness and safety of 18F-FDG PET in the evaluation of dementia: A review of the recent literature. Journal of Nuclear Medicine 53(1):59-71.

Bowles, K. R., M. C. Silva, K. Whitney, T. Bertucci, J. E. Berlind, J. D. Lai, J. C. Garza, N. C. Boles, S. Mahali, K. H. Strang, J. A. Marsh, C. Chen, D. A. Pugh, Y. Liu, R. E. Gordon, S. K. Goderie, R. Chowdhury, S. Lotz, K. Lane, J. F. Crary, S. J. Haggarty, C. M. Karch, J. K. Ichida, A. M. Goate, and S. Temple. 2021. ELAVL4, splicing, and glutamatergic dysfunction precede neuron loss in MAPT mutation cerebral organoids. Cell 184(17):4547-4563, e4517.

Bryant, A. G., M. Hu, B. C. Carlyle, S. E. Arnold, M. P. Frosch, S. Das, B. T. Hyman, and R. E. Bennett. 2020. Cerebrovascular senescence is associated with tau pathology in Alzheimer’s disease. Frontiers in Neurology 11:575953.

S. Gentleman, G. Logroscino, G. E. Serrano, T. G. Beach, I. G. McKeith, A. J. Thomas, J. Attems, C. M. Morris, L. Palmer, S. Love, C. Troakes, S. Al-Sarraj, A. K. Hodges, D. Aarsland, G. Klein, S. M. Kaiser, R. Woltjer, P. Pastor, L. M. Bekris, J. B. Leverenz, L. M. Besser, A. Kuzma, A. E. Renton, A. Goate, D. A. Bennett, C. R. Scherzer, H. R. Morris, R. Ferrari, D. Albani, S. Pickering-Brown, K. Faber, W. A. Kukull, E. Morenas-Rodriguez, A. Lleó, J. Fortea, D. Alcolea, J. Clarimon, M. A. Nalls, L. Ferrucci, S. M. Resnick, T. Tanaka, T. M. Foroud, N. R. Graff-Radford, Z. K. Wszolek, T. Ferman, B. F. Boeve, J. A. Hardy, E. J. Topol, A. Torkamani, A. B. Singleton, M. Ryten, D. W. Dickson, A. Chio, O. A. Ross, J. R. Gibbs, C. L. Dalgard, B. J. Traynor, and S. W. Scholz. 2021. Genome sequencing analysis identifies new loci associated with Lewy body dementia and provides insights into its genetic architecture. Nature Genetics 53(3):294-303.

Chow, H.-M., M. Shi, A. Cheng, Y. Gao, G. Chen, X. Song, R. W. L. So, J. Zhang, and K. Herrup. 2019. Age-related hyperinsulinemia leads to insulin resistance in neurons and cell-cycle-induced senescence. Nature Neuroscience 22(11):1806-1819.

Ciani, M., C. Bonvicini, C. Scassellati, M. Carrara, C. Maj, S. Fostinelli, G. Binetti, R. Ghidoni, and L. Benussi. 2019. The missing heritability of sporadic frontotemporal dementia: New insights from rare variants in neurodegenerative candidate genes. International Journal of Molecular Science 20(16):E3903.

Claassen, J., D. H. J. Thijssen, R. B. Panerai, and F. M. Faraci. 2021. Regulation of cerebral blood flow in humans: Physiology and clinical implications of autoregulation. Physiological Reviews 101(4):1487-1559.

Clark, K., Y. Y. Leung, W. P. Lee, B. Voight, and L. S. Wang. 2022. Polygenic risk scores in Alzheimer’s disease genetics: Methodology, applications, inclusion, and diversity. Journal of Alzheimer’s Disease 89(1):1-12.

Cocean, A. M., and D. C. Vodnar. 2024. Exploring the gut-brain axis: Potential therapeutic impact of psychobiotics on mental health. Progress in Neuro-Psychopharmacology & Biological Psychiatry 134:111073.

Collins, H. M., and S. Greenfield. 2024. Rodent models of Alzheimer’s disease: Past misconceptions and future prospects. International Journal of Molecular Science 25(11):6222.

Columbia University. 2024. NIH Award Creates Columbia-Led Exposomics Coordinating Center. https://www.publichealth.columbia.edu/news/nih-award-creates-columbia-led-exposomics-coordinating-center (accessed November 11, 2024).

Corriveau, R. A., F. Bosetti, M. Emr, J. T. Gladman, J. I. Koenig, C. S. Moy, K. Pahigiannis, S. P. Waddy, and W. Koroshetz. 2016. The science of vascular contributions to cognitive impairment and dementia (VCID): A framework for advancing research priorities in the cerebrovascular biology of cognitive decline. Cellular and Molecular Neurobiology 36(2):281-288.

Crimmins, E. M. 2020. Social hallmarks of aging: Suggestions for geroscience research. Ageing Research Review 63:101136.

Cruts, M., I. Gijselinck, J. van der Zee, S. Engelborghs, H. Wils, D. Pirici, R. Rademakers, R. Vandenberghe, B. Dermaut, J.-J. Martin, C. van Duijn, K. Peeters, R. Sciot, P. Santens, T. De Pooter, M. Mattheijssens, M. Van den Broeck, I. Cuijt, K. l. Vennekens, P. P. De Deyn, S. Kumar-Singh, and C. Van Broeckhoven. 2006. Null mutations in progranulin cause ubiquitin-positive frontotemporal dementia linked to chromosome 17q21. Nature 442(7105):920-924.

Cummings, J., A. M. Leisgang Osse, and J. Kinney. 2023a. Geroscience and Alzheimer’s disease drug development. Journal of Prevention of Alzheimer’s Disease 10(4):620-632.

Cummings, J., Y. Zhou, G. Lee, K. Zhong, J. Fonseca, and F. Cheng. 2023b. Alzheimer’s disease drug development pipeline: 2023. Alzheimer’s & Dementia (N Y) 9(2):e12385.

Cutler, R. G., J. Kelly, K. Storie, W. A. Pedersen, A. Tammara, K. Hatanpaa, J. C. Troncoso, and M. P. Mattson. 2004. Involvement of oxidative stress-induced abnormalities in ceramide and cholesterol metabolism in brain aging and Alzheimer’s disease. Proceedings of the National Academy of Sciences USA 101(7):2070-2075.

Waldo, J. S. Yokoyama, M. A. Nalls, P. Momeni, A. B. Singleton, J. Hardy, and V. Escott-Price. 2024. Genome-wide analyses reveal a potential role for the MAPT, MOBP, and APOE loci in sporadic frontotemporal dementia. American Journal of Human Genetics 111(7):1316-1329.

Marlies Oosthoek, L. V., A. de Wilde, B. Bongers, D. Antwi-Berko, P. Scheltens, P. van Bokhoven, E. G. B. Vijverberg, and C. E. Teunissen. 2024. Utilization of fluid-based biomarkers as endpoints in disease-modifying clinical trials for Alzheimer’s disease: A systematic review. Alzheimer’s Research & Therapy 16:93.

Marquez, F., W. Tarraf, A. M. Stickel, K. A. Gonzalez, F. D. Testai, J. Cai, L. C. Gallo, G. A. Talavera, M. L. Daviglus, S. Wassertheil-Smoller, C. DeCarli, N. Schneiderman, and H. M. Gonzalez. 2024. Hypertension, cognitive decline, and mild cognitive impairment among diverse Hispanics/Latinos: Study of Latinos—Investigation of Neurocognitive Aging Results (SOL-INCA). Journal of Alzheimer’s Disease 97(3):1449-1461.

Marsit, C. J. 2015. Influence of environmental exposure on human epigenetic regulation. Journal of Experimental Biology 218(Pt 1):71-79.

Martinez-Feduchi, P., P. Jin, and B. Yao. 2024. Epigenetic modifications of DNA and RNA in Alzheimer’s disease. Frontiers in Molecular Neuroscience 17:1398026.

Matthay, E. C., E. Hagan, L. M. Gottlieb, M. L. Tan, D. Vlahov, N. E. Adler, and M. M. Glymour. 2020. Alternative causal inference methods in population health research: Evaluating tradeoffs and triangulating evidence. SSM Population Health 10:100526.

Mayne, K., J. A. White, C. E. McMurran, F. J. Rivera, and A. G. de la Fuente. 2020. Aging and neurodegenerative disease: Is the adaptive immune system a friend or foe? Frontiers in Aging Neuroscience 12:572090.

Mehta, R. I., and R. I. Mehta. 2023. The vascular-immune hypothesis of Alzheimer’s disease. Biomedicines 11(2):408.

Mei, Z., G. Liu, B. Zhao, Z. He, and S. Gu. 2023. Emerging roles of epigenetics in lead-induced neurotoxicity. Environment International 181:108253.

Melcher, E. M., L. Vilen, A. Pfaff, S. Lim, A. DeWitt, W. R. Powell, B. B. Bendlin, and A. J. H. Kind. 2024. Deriving life-course residential histories in brain bank cohorts: A feasibility study. Alzheimer’s & Dementia 20(5):3219-3227.

Mensikova, K., R. Matej, C. Colosimo, R. Rosales, L. Tuckova, J. Ehrmann, D. Hrabos, K. Kolarikova, R. Vodicka, R. Vrtel, M. Prochazka, M. Nevrly, M. Kaiserova, S. Kurcova, P. Otruba, and P. Kanovsky. 2022. Lewy body disease or diseases with Lewy bodies? NPJ Parkinsons Disease 8(1):3.

Mez, J., J. Chung, G. Jun, J. Kriegel, A. P. Bourlas, R. Sherva, M. W. Logue, L. L. Barnes, D. A. Bennett, J. D. Buxbaum, G. S. Byrd, P. K. Crane, N. Ertekin-Taner, D. Evans, M. D. Fallin, T. Foroud, A. Goate, N. R. Graff-Radford, K. S. Hall, M. I. Kamboh, W. A. Kukull, E. B. Larson, J. J. Manly, J. L. Haines, R. Mayeux, M. A. Pericak-Vance, G. D. Schellenberg, K. L. Lunetta, and L. A. Farrer. 2017. Two novel loci, COBL and SLC10A2, for Alzheimer’s disease in African Americans. Alzheimer’s & Dementia 13(2):119-129.

Mielke, M. 2024. Sex and gender differences in Alzheimer’s disease and Alzheimer’s disease related dementias. Commissioned Paper for the Committee on Research Priorities for Preventing and Treating Alzheimer’s Disease and Related Dementias.

Migliore, L., and F. Coppedè. 2022. Gene–environment interactions in Alzheimer disease: The emerging role of epigenetics. Nature Reviews Neurology 18:643-660.

Miller, G. 2024. Presentation at Committee on Research Priorities for Preventing and Treating Alzheimer’s Disease and Related Dementias Public Workshop, Hybrid. January 16–17, 2024.