The Role of the Behavioral and Social Sciences in Geroscience — Workshop II

Day 1 | February 14, 2024

Welcome and Opening Remarks

Welcome and Opening Remarks

David Braudt, PhD, National Institute on Aging (NIA); Lis Nielsen, PhD, NIA; Stacy Carrington-Lawrence, PhD, NIA; & Richard J. Hodes, MD, NIA

This workshop was the second in a series of three planned workshops on the role of behavioral and social sciences in geroscience research. This series of workshops is intended to stimulate a transdisciplinary dialogue between aging biologists and social and behavioral scientists to advance research examining the ways in which environmental exposures and social and behavioral processes influence the biological processes involved in aging. The workshops are designed to allow researchers from a variety of fields to articulate assumptions, theories, and methods that examine aging over the life course and to review evidence for malleability and reversibility of aging-related biological outcomes. These transdisciplinary forums seek to achieve the longer-term goal of identifying behavioral, social, and environmental intervention targets to delay biological aging, extend the health span, and reduce health disparities at the population level.

The second workshop would provide a forum for sharing frameworks, terminology, and approaches as part of a continued dialogue between aging biology and social science researchers. Two frameworks relevant to geroscience research in the social and behavioral sciences include: a prevention framework and a health disparities framework, both of which require examinations of the interactions between behavioral and social exposures and biology. Leaders at the National Institute on Minority Health and Health Disparities have defined health disparities as preventable health differences that are closely linked with social, economic, and/or environmental disadvantages. Using this definition, the National Institute on Aging (NIA) Health Disparities Research Framework has identified multiple interacting levels of analysis that drive health disparities, which include environmental (e.g., racial segregation), sociocultural (e.g., discrimination, prejudice), and behavioral (e.g., social support systems, emotional regulation) factors and the biological processes they effect (e.g., cellular function, inflammation). The interdependence of the factors that inform health disparities and, ultimately, stress-related aging emphasizes the need for transdisciplinary dialogues between biological and social and behavioral researchers to advance geroscience research.

Changes Across Fields Since Geroscience Workshop Part I

Changes Across Fields Since Geroscience Workshop Part I

Social and Behavioral Predictors of the Pace of Aging: New Findings in 2023‒2024

Terrie Moffitt, PhD, Duke University and King’s College London; & George Kuchel, MD, University of Connecticut

Terrie Moffitt, PhD, Duke University and King’s College London

Since the first Geroscience Workshop was held in October of 2022, the geroscience field has advanced considerably, in part because of the widespread use of third-generation epigenetic (DNAm) clocks that assess whole-body aging as longitudinal decline. For example, the Dunedin Pace of Aging Calculated in the Epigenome (DunedinPACE) has been implemented in 52 large epidemiological cohorts in 15 countries to date, with results reported in 125 publications since 2022. Across multiple cohorts and countries, faster DunedinPACE has been shown to predict Alzheimer’s disease, multimorbidity, and mortality and to correlate with abnormal MRI brain scans, lower educational level, and schizophrenia. DunedinPACE has also demonstrated that early life adversity and other psychological, demographic, and mental health factors correspond with a faster rate of aging across a range of diverse cohorts. Recent studies suggest that DunedinPACE scores may be malleable and responsive to intervention. Finally, several recent research studies have used DunedinPACE to examine racial and ethnic health disparities and the impact of social determinants of health among Black/African American, Hispanic, Han Chinese, Native American, and White populations. Although most studies using DunedinPACE conducted to date have been observational, researchers are now conducting quasi-experiments, natural experiments, and randomized controlled trials. Future geroscience research should examine the mediators of the stress-aging relationship. DunedinPACE is licensed by TruDiagnostic for commercial use and is available for free on Github and BioLearn for research use.

George A Kuchel, MD, University of Connecticut

The Translational Geroscience Network (TGN; R33AG061456), overseen by Dr. Kuchel and his colleagues, has been directly involved in enhancing more than 84 early-stage clinical trials to date. TGN has assisted with several trials and publications in gerotherapeutics, an emerging field in which senolytic therapy has shown particular promise. One NIH-funded randomized clinical trial of gerotherapeutics found that metformin reduced Long-COVID. Other studies have examined the relationship between biomarkers and gerotherapeutics following periods of caloric restriction, which has also been studied extensively as a longevity and healthspan intervention in animal models.

The Journal of Gerontology Series A will publish several papers drawn from studies presented at the Geroscience for the Next Generation Summit held in April 2023, which highlighted interdisciplinary contributions to the recent growth of the field. On November 7, 2023, a Gerontological Society of America pre-conference workshop was held to foster interdisciplinary collaborations in geroscience aimed at building geroscience interest, supplying the geroscience workforce, and alignment of perspectives and educational goals in geroscience research. Cardiology is one discipline that has joined the collaboration, exemplified by the recently published, “Impact of Geroscience on Therapeutic Strategies for Older Adults with Cardiovascular Disease.”

As geroscience continues to move forward, future research directions should include biological aging, lifestyle factors, behavioral considerations, and social determinants of health (SDoH). Ultimately, research should endeavor toward a more holistic view of gerotherapeutics and health promotion that considers the Geriatric 5Ms: Mind, Mobility, Medications, Multicomplexity, and what Matters Most: advanced and personalized care planning.

Discussion

Transdisciplinary Development

In recent years, geroscience has seen a shift from questioning the value of transdisciplinary research to questioning how to best implement it. The transition to a multi-field, whole-person study of aging requires researchers to address some foundational questions about measurement replicability, generalizability, and applicability to different ages or racial and ethnic groups, for example.

Obstacles to Validating Measures

Researchers should continue to attempt to validate SDoH and biological hallmarks of aging (e.g., stress exposure and reactivity, biological hallmarks) in representative and non-White populations; however, funding limitations often require studies to collect data from subgroups, which likely do not represent an entire population. Additionally, many preclinical measures of aging biology have not yet been tested in humans. Translating aging biomarkers developed and tested in preclinical research to clinical trials and population-level studies can be difficult, but such efforts are essential for advancing geroscience research.

A Universal Definition of Aging

Researchers should be aware of how each discipline defines “aging” and recognize that different disciplines may be using the term in different ways. For instance, a distinction could be drawn between aging and the aspects or indicators of aging and, further, between organ- or tissue-specific aging and whole-body aging.

Mechanisms of Disparities

Better measures of the stressors experienced by diverse, racialized, and minoritized groups are needed for a more holistic approach to health-disparities research. Comprehensive measurement of long-term stressors that lead to health disparities, such as racism, is a burgeoning field with more research needed into how different types of long-term stressors contribute to health inequities. Race is sometimes used as a proxy metric for social racialization and the systemic, structural, interindividual, and personal processes associated with racialization. Geroscience researchers examining disparities and SDoH should emphasize the importance of elucidating the causal role of environmental, social, and behavioral mechanisms in shaping health disparities rather than merely documenting the existence of these disparities within specific populations.

Session 1: What do biological scientists need to know about social determinants of health (SDoH) and studying the causal impact of SDoH on biological aging?

Session 1: What do biological scientists need to know about social determinants of health (SDoH) and studying the causal impact of SDoH on biological aging?

Moderator: David Braudt, PhD, NIA
Chair: Kenneth A. Bollen, PhD, University of North Carolina (UNC) at Chapel Hill

Latent Variables and Longitudinal Models with Applications in the Gerosciences

Kenneth A. Bollen, PhD, UNC at Chapel Hill

Many variables central to scientific hypothesis in gerosciences are abstract, such as biological aging, frailty, allostatic load, or stress. While other variables are less abstract, such as heart rate or birth weight, these measures are not error free. Latent variables are able to more accurately model measurement error and thus are more likely to represent the pure form of both types of variables. Latent variables are the subject of scientific hypotheses. When researchers equate the observed variables in their data sets with the latent variables in hypotheses, they risk biased results and flawed tests of hypotheses due to the statistical assumption that these measures are free of any measurement error.

There are, however, options that enable us to model our hypotheses of relations between latent variables that control for measurement error, i.e. structural equation models. These procedures are widely available in the social, behavioral, and health sciences and there are many software options to implement them. An empirical example was reviewed using the fetal origin literature that highlights the lifelong health implications of favorable and unfavorable fetal conditions. Favorable fetal conditions is an abstract latent variable measured with birth weight, birth length, and gestation age. Birth weight is a more concrete variable but is still prone to be measured with error. The model discussed shows that both types of latent variables are accommodated in a single model. And this model supported the hypothesis of treating Favorable Fetal Growth Conditions as a latent variable that influences birth weight, birth length, and gestational age and better represented the data than either a model based on only observed variable and the other multi-dimensional latent variable models tested.

A second problem in geroscience discussed in this talk was the arbitrary selection of longitudinal models when analyzing outcome variables available for the same individuals over time. The longitudinal model that researchers select often differs across studies and investigators. The selection of which model to employ is more often based on arbitrary rather than scientific grounds (e.g., the training that researchers receive). Importantly, the consequences of choosing the wrong longitudinal model are biased assessments of overtime change in outcomes and their causes. The latent variable autoregressive latent trajectory (LV-ALT) model is a flexible approach to longitudinal analysis that allows researchers to systematically test and evaluate which longitudinal model is most appropriate for their hypotheses. For example, a study used the LV-ALT to model measurement error in a single-item measure of self-rated health. Using this approach, the researchers found that an autoregressive model with an individual specific effect fit the data best as compared to competing longitudinal models (e.g., fixed/random effect models, growth curve models).

As this presentation illustrated, researchers can both take account of measurement error in their variables and select the best longitudinal model for a given data set by leveraging structural equation modeling approaches. The two main obstacles to doing so are the lack of awareness of these procedures and the accompanying lack of training.

Leveraging Epigenetic Data to Study the Social Determinants of Health and Aging: Opportunities and Challenges

Lauren Schmitz, PhD, University of Wisconsin, Madison

Economic conditions during the Great Depression varied in different states over time. Dr. Schmitz used this fact to conduct a quasi-natural experiment on the early life impacts of the Great Depression on biological age in later life, using data from the Health and Retirement Study (HRS). Dr. Schmitz used state-level historical data on wages and employment to capture state-based variation in the impact of the Great Depression. She then linked these data to individuals’ state of birth and their biological age later life, as indicated by epigenetic clocks (e.g., GrimAge, DunedinPACE). Exposure to adverse economic conditions in utero was associated with accelerated biological aging. Exposure during childhood was not linked to accelerated aging. These associations likely underestimate the true impact of the Great Depression, as surviving individuals may have had slower biological aging than those who suffered more serious in utero effects.

This study provides evidence for a causal association between in utero exposure to stressful life events and accelerated aging. However, the study has some limitations. Its findings are specific to the Great Depression and its early life effects. Thus, the results of the study may not generalize to other exposures or exogenous shocks. Such naturalistic experiments do not provide information about the mechanisms that lead to accelerated aging. More research is needed on the biological mechanisms by which economic conditions affect fetal environment and later life. These studies would benefit from more extensive data linking social and biological factors, as well as data from more diverse populations to track how the impact of in utero exogenous shocks manifests throughout the life course. Additional studies using data from lower-income countries could also enhance the understanding of how income and exposure to adversity impact epigenetic aging.

Incorporating Biomarkers Into the Study of Racial Inequalities and Health Within Racially Minoritized Populations

Mateo Farina, PhD, University of Texas at Austin

Racial and ethnic inequalities affect multiple domains of health, with effects that can persist across generations. Accounting for these inequalities in health and aging studies is crucial, particularly given the increasing proportion of minoritized older adults in the United States. Moreover, identifying the ways in which aging differs across race and ethnicity can advance the study of aging as a whole. Such research should reflect the fact that race is a social and historical construct that is distinct from genetic ancestry.

Biomarkers of aging, like epigenetic clocks and inflammatory markers, can illuminate the impact of biological and social mechanisms on racial minorities and help to identify aging-related factors that put minoritized populations at increased risk for diseases linked to aging and high morbidity rates. Importantly, the appropriate biomarkers can depend on the social context. An analysis of 2016 HRS found significant accelerated aging among Black and Hispanic participants, based on biological age measured through 22 biomarkers of aging, with Hispanic males exhibiting the fastest biological aging relative to the overall population. The study examined the impact throughout the life course of social and economic factors, such as educational attainment and health insurance coverage. In Dr. Farina’s study, childhood adversity, lower educational attainment, and lack of health insurance correlated with greater levels of biological aging in minoritized populations, which included US- and foreign-born Hispanic older adults and non-Hispanic Black older adults. After accounting for these life course exposures, racial differences in accelerated biological aging were substantially attenuated.

The impacts of SDoH on biological aging may also occur through different pathways across racial groups. Dr. Farina found that loss of a parent was correlated with accelerated epigenetic aging among White participants, but results were inconclusive for Black participants. These distinctions suggest that further research is needed to better understand the causes and consequences of biological aging and how it might differ across population subgroups. Additionally, research advancements should be made on population-specific measures, such as work and life exposures to health risks that differentially affect racialized minorities. Future studies should also account for factors, such as selection bias and social factors, which may affect aging.

Studying Social Determinants of Health and Sociodemographic Factors Across Populations and Their Relationships with Estimating Biological Age

Hiram Beltrán-Sánchez, PhD, University of California, Los Angeles

Modern health disparities are influenced by SDoH, such as social stratification and health systems. By examining a diverse population, researchers can get a fuller understanding of the effects of SDoH on biological aging across different populations. Health disparities manifest throughout the life cycle, which means that longitudinal data are required to examine their impacts and origins. For example, the quality of an individual’s health exposures in early life is directly related to their parents’ social class. In societies with a strong class system, parental and offspring class are directly correlated, so health disparities persist across generations. Longitudinal HRS data can shed light on SDoH impacts, in part because older Americans in the HRS belong to distinct generational cohorts influenced by different social events. The HRS has been able to collect not just follow-up data on the impact of SDoH over time, but also some biomarker data to further analyze the effect of SDoH on aging. In lower-income countries such as in Latin America, many older adults have grown up with limited access to medical technologies and modern health-care infrastructure. Biomarkers in these populations can be particularly important for studying health disparities and biological aging because these countries often lack high-quality health data.

Dr. Beltrán-Sánchez studied biological aging (as indicated by life expectancy at the age of 65) to assess aging among Americans aged 30‒75 who participated in the National Health and Nutrition Examination Survey. His study found that increases in the difference between biological aging and chronological aging lead to a lower life expectancy at the age of 65. This approach found evidence that biological aging clocks can serve as measures of health disparities, but Dr. Beltrán-Sánchez noted the wide variability across different biological aging clock’s predictions of life expectancy based on the kinds of biomarkers used. Future research could utilize biological aging as a predictor of proximate determinants of health or as a mediating factor linking socioeconomic conditions with a particular outcome. In addition, biological aging clocks are often directly dependent on the type and number of biomarkers and endogenous exposures included in the measure. Obesity, for example, can induce dysregulation and act as both an indicator of advanced aging as well as a mechanism that contributes to accelerated aging.

Discussion

Variation in Aging in Nonindustrial Societies

Studies on aging in lower-income countries can inform the understanding of age acceleration and supplement studies in wealthier countries. Important variations in these countries can occur due to nutritional transition, as individuals begin to consume more calorically dense foods. Countries in the late stages of the nutritional transition can have a higher prevalence of diabetes and obesity. Researchers should also consider the socioeconomic variation within lower-income countries, which may lack nationally representative health surveys (such as the HRS), health care records, and other forms of population-level data; their data may also be unavailable for significant periods of time. Thus, in some low- and middle-income countries, biomarkers may be the only reliable method for assessing health and aging due to the lack of national- and population-level health data.

How Biological Scientists and Social Scientists Can Learn From Each Other

Given Dr. Schmitz’s findings related to the in-utero impact of the Great Depression, collaborations between developmental psychologists, social scientists, and researchers studying biological aging may be important to understand how very early life events affect health throughout the life course.

Session 2: What do biological scientists need to know about how psychologists and other social/behavioral scientists examine the impact of acute and chronic stressors on biological processes of aging?

Session 2: What do biological scientists need to know about how psychologists and other social/behavioral scientists examine the impact of acute and chronic stressors on biological processes of aging?

Moderator: Emily Hooker, PhD, NIA
Chair: Elissa Epel, PhD, University of California, San Francisco (UCSF)

Role of Behavioral & Social Factors in Geroscience

Elissa Epel, PhD, UCSF

Researchers can measure several aspects of stress, such as acute and cumulative life stress, resource strain, pregnancy stress, social isolation, neighborhood safety, climate stress, traumatic life events, and daily stressors. Stress biomarkers include systemic inflammation, telomere shortening, mitochondrial impairment, and accelerated epigenetic clock; however, these biomarkers can present differently across different groups. For example, females show more extreme responses to stress across all of these biomarkers, likely because of estrogen-based neuroplasticity and psychosocial factors, but the mechanisms underlying these sex differences are not well understood. The effect of transgenerational stress on future generations is also notable, and some studies have demonstrated that anti-stress interventions during pregnancy make babies healthier and more resilient to stress, in addition to reducing the stress levels of the pregnant person. Certain generational differences in stress were reflected in a recent American Psychological Association survey that asked participants to endorse the statement, “Stress is overwhelming most days.” The results of the survey indicated that young people have more than 3-fold higher degrees of perceived stress than older adults, which could be due to well-documented age-related larger perspectives on life, or resilience to stress. The survey also found that individuals with marginalized social identities (e.g., sexual and gender minority populations) reported higher perceived stress.

To understand how perceived stress and the social hallmarks of stress relate to biological aging, studies that follow individuals throughout their lifespans are necessary. Dr. Epel conducted a longitudinal study that compares aspects of aging biology in premenopausal women who are experiencing extreme chronic stress with low-stress controls. This study found reliable, robust overall differences in stress-based aging biomarkers between the groups. However, short-term changes in stress-based aging biomarker changes (i.e., over two-year increments) were very small, emphasizing that the effects of chronic stress manifest over years, if not decades. Dr. Epel is also conducting a lifespan study which is following a cohort of female individuals from age 10 to age 50. This long-term longitudinal study has found that transgenerational stress and stress during developmental periods affect participants’ rates of biological aging, potentially through weight gain and accelerated onset of puberty. Hypermetabolism may also mediate the effects of stress on aging. A paper recently added to the Stress Measurement Network Literature Library suggests that hypermetabolism may be a vital mechanism involved in whole cell aging, including telomere shortening and reaching the Hayflick limit, and epigenetic aging.

To advance the study of stress and human health, researchers should prioritize interdisciplinary collaboration and the use of common, precise language across disciplines. Specificity when describing stressor exposure and stress responses would allow researchers from multiple disciplines to access stress-related literature. Researchers should define specific stress exposure attributes (e.g., duration, severity, controllability, life domains), timescale (acute, chronic), and lifespan periods (in utero, childhood, adulthood). This also applies to attributes of stress response and response characteristics, global subjective stress, subjective stress within life domains (e.g., pregnancy, retirement), and acute subjective and behavioral stress responses (e.g., emotional reactivity, behavioral coping). The typology and dimensions for describing stress are in the ‘Stress Network paper’s Appendix of Epel et al, 2018.

Do Stress Researchers Love or Hate Good Measurement? A Valentine’s Day Story of Mediocre Measurement & Conceptual Confusion

George M. Slavich, PhD, University of California, Los Angeles

Researchers employ many different measures of stress, but many provide limited information or have weak psychometric properties. For example, a high-profile study of early life adversity found that early life stress interacted with genetic polymorphisms in the promoter region for the serotonin transporter gene to predict risk for depression. The numerous failed attempts to replicate this study varied in their approaches to assessing stress, illustrating the lack of standardized research measures of early life stress. Commonly used measures of stress, such as checklists, provide only imprecise assessments of stress exposures. This can be described by 10 common practices in studies of early life stress: (1) use of brief and imprecise measurement tools, (2) conflation of stress with other study outcomes, (3) timing of stressor exposure generalized or not assessed, (4) different stressors weighted identically, (5) stressors assessed by count or severity but not both, (6) measurement tools assessing acute life events or chronic difficulties but not both, (7) stress reactivity conflated with stress exposure, (8) certain stressors ignored for the sake of the stressor of interest, (9) non-life stress items (e.g. sleep problems, depression) incorrectly defined as such, (10) narrow stress assessment window.

To improve the measurement of early life stress, Dr. Slavich created the Stress and Adversity Inventory (STRAIN). To date, STRAIN has been translated into 28 different languages, and it aims to provide a standardized stress assessment that captures the most accurate, reliable, and replicable results. The computer-based assessment takes up to 25 minutes to complete and uses dynamic interview branching to assess over 115 different lifetime stress exposures. STRAIN uses latent-class trajectory modeling to examine individuals’ stress exposures over time, enables profiling of individual stress levels based on stress exposures, and records demographic, cognitive-emotional, and clinical characteristics. The pattern of associations between acute and chronic stress exposure throughout the lifespan depends on stress signals, some of which can predict stress-related health outcomes in addition to creating marked changes in neuronal structure and other physiological processes.

Early Adversity & Inflammation Links: Through a Life Course Lens

Jessica Chiang, PhD, Georgetown University

Early adversity--i.e., chronic or severe stress experienced during childhood or adolescence--increases risk for a number of chronic illnesses, including cardiovascular disease, respiratory disease, and some cancers. Research points to inflammation as a key pathway linking early adversity to poor health. Although an acute inflammatory response to pathogens or injury is adaptive, when sustained throughout the body for prolonged periods of time, it can elevate risk for chronic illnesses. Many studies have demonstrated a link between early adversity and higher levels of inflammation, but the majority of these studies have not considered the role of time and how this link might vary throughout the life course. A life course perspective is needed to deepen our understanding of the inflammatory and other mechanistic processes through which early adversity comes to shape health and disease risk decades later, and to elucidate optimal times to intervene. However, applying a life course perspective requires data that spans the developmental stages and age range of the life course, which is challenging to collect. Combining meta-analytic techniques with accelerated longitudinal designs in primary data collection can help overcome this challenge.

As an example, Dr. Chiang and her colleagues conducted a meta-analysis of existing studies to test the hypothesis that the link between early adversity and inflammation should strengthen across the lifespan, as effects of early adversity take time to accumulate and take a toll on the health. The meta-analysis included 922 effect sizes from 187 studies across 173,000 individuals ranging from 0‒81 years of age. Statistical parameters and methodological characteristics, such as type of early adversity (socioeconomic status (SES), maltreatment, interpersonal stress, and cumulative stress) and sample age when inflammation was measured, were extracted. Each type of early adversity was linked with higher levels of inflammation, but notably, the effect size of early adversity on inflammation was smallest in childhood and largest in adulthood, supporting the accumulation hypothesis. This meta-analysis was both cost- and time-efficient, and it allowed examination of a large age range and a wide array of stress operationalizations. However, the research was constrained by what was available in the literature and lacked access to individual-level data, limiting the ability to answer more precise questions about stress processes.

The combined use of accelerated longitudinal designs and real-time, diary-based assessments could provide an alternative time-efficient, micro-accessible approach for answering questions centered around how mechanistic processes underlying the effects of early adversity unfold and vary across the life course. For example, to test the hypothesis that early adversity increases sensitivity or responsivity to subsequent stressors that occur later in life, stress (e.g., social conflict with loved one, felt left out, felt embarrassed, etc.) and sleep can be measured every day for two weeks, and using these data, the extent to which stress disrupts sleep can be computed for each individual. This stress responsivity score can then be linked to assessments of early adversity and inflammation to test whether altered stress responsivity functions as a pathway linking early adversity to inflammation. Embedding these assessments within an accelerated longitudinal design, which periodically assesses several different age group cohorts over time, then allows examination of how these links and patterns might change over time through several stages in the life course.

Stress and Aging Biology: Integrating the Role of Time

Rebecca Reed, PhD, University of Pittsburgh

When life events are perceived as stressful, they produce negative emotional responses that can lead to poor health decisions or adverse biological responses, such as activation of the sympathetic-adreno-medullar and hypothalamic-pituitary-adrenal axes. Activation of these axes can lead to physiological changes, including inflammation, immunocompromising, and cardiovascular challenges, which increase the risk for chronic disease. To improve the study of stress and its related effects, stress research needs to better account for time by determining when stressors happen within the life course and examine their severity and duration. The study of intraindividual changes and variability in stress across the lifespan and the effects of stress on biological aging would benefit from the use of multitemporal designs.

For example, a study at the University of Pittsburgh is using integrative data analysis to pool raw stressor and biological aging measures across two large epidemiological cohorts of middle- and older-aged adults from the HRS and the English Longitudinal Study of Aging (ELSA). Certain stressful life events are harmonized across HRS and ELSA: death of a child, natural disaster exposure, combat experience, family addiction history, history of physical attack or assault, and exposure to life-threatening illness or accident. Most of these stressful events occurred in young adulthood or midlife, with lower frequencies in childhood and late adulthood. Experiencing a higher total number of stressful life events across the life course correlated with higher levels of inflammation and accelerated epigenetic aging. These findings support the cumulative model of stress, which states that stress accumulates over time to affect biological processes. The findings also supported the sensitive period model of stress, specifically in young adulthood and midlife, potentially because adults at these periods may hold additional roles of responsibility (e.g., caregiving or other support roles) that may further tax their health. Future iterations of this study will incorporate prospective measures, such as checklists to collect objective, childhood-specific stressor data, and decrease potential confounding by differential perceptions of prior stressors.

Dr. Reed also conducted a study in older adults that employed a repeated measurement design to examine intraindividual variability in cognitive reappraisal and stress-related immunological aging Cognitive reappraisal is an emotional regulation technique that reframes an emotional situation to lessen its impact on a person. Dr. Reed assessed older adults’ stress levels, cognitive reappraisal usage, and immunological aging biomarkers twice yearly for up to five years. Greater emotional regulation via cognitive reappraisal was associated with protection from the deleterious effects of stress on immunological aging, both within and across individuals. Higher levels of cognitive reappraisal practice were associated with lower levels of late-differentiated natural killer cells (potentially senescent or nearly senescent natural killer cells), indicating less rapid immunological aging. Additionally, these older adults who used more frequent cognitive reappraisal showed lower levels of inflammation in the context of severe stress. Thus, cognitive reappraisal of stressful life events may promote resilience and recovery from emotional and biological stress.

Discussion

Fostering Collaboration with Other Fields of Study

The field of geroscience could benefit from more interdisciplinary collaboration with social and behavioral science across the lifespan. Notably, many studies that are currently examining early childhood stress do not include geroscience measures. To address this gap, researchers could collect data relevant to geroscience and related to the COVID-19 pandemic to enable study of the pandemic’s impact on biological aging throughout the life course. The COVID-19 pandemic had enormous psychosocial and economic impacts, including employment instability, isolation, and grief. On the other hand, a lifespan study of the effects of this period, despite it being a notably stressful period, may merely reflect what is already known about the relationship between stress and aging. Some early research suggests that, while the COVID-19 pandemic increased stress levels, it did not cause certain correlations identified by pre-pandemic stress research to fluctuate. For example, a recent study found that, although levels of post-traumatic stress disorder (PTSD) in health care workers increased fourfold during the pandemic, the ratio of PTSD scores between women and men remained unchanged.

Inflammation Versus Blood-Based Biomarkers as Indicators of Stress Throughout the Lifespan

Some aging biomarkers are present in cord blood (i.e., infant blood) at birth, indicating the germline impact of stress on offspring and an early predisposition to stress-related aging. By contrast, inflammation is more dynamic and prone to change throughout the life course.

Collecting the Full Picture in Interdisciplinary Research

Rather than making assumptions regarding which life stressors affect aging processes, investigators of all disciplines should assess a wide range of stressors and exposures, including those of unknown relevance. Understanding the bounds of social and environmental stressors is key to confirming and disconfirming stress hypotheses. In this regard, a broad life events perspective could prove to be beneficial.

Cultural variation in stressor types and the impact of these stressors is not well understood. Collaborating with investigators across sociocultural contexts is essential for stress research across contexts.

Finally, researchers need to address selection bias in stress research and in clinical contexts. For example, stress studies collecting data from electronic health records should recognize that health records may be biased by clinicians’ preconceptions of what information is relevant for their patients’ health.

Assessing Resilience

Dynamic probes, such as stress challenge tasks, can reveal psychological and behavioral responses and resilience to stress. Importantly, several open empirical questions remain regarding whether the resilience assessed in lab-based designs operates through the same mechanisms as in daily life or when measured by diary-based methods, such as the ecological momentary assessment.

Future Research

Future geroscience research should study hormetic stress, exercise stress, deep relaxation, and hypermetabolism, while also examining critical, core mechanisms involved in the aging processes.

Summary & Discussion

Day 1 Summary & Discussion

Closing Remarks

Kenneth Bollen, PhD, UNC at Chapel Hill

Key Concepts in Geroscience

Researchers should conduct a systematic exploration of key concepts and domains within each concept to increase specificity when measuring these domains. For example, what are the essential components of SES that should be examined? What constitutes discrimination, and should certain individual aspects of discrimination be considered separately? Similarly, dimensions of aging need to be theoretically and empirically defined.

Definition and Perception

There are many different characteristics of stress responses and stressors, and the line can sometimes be blurred in research. Defining these terms and recognizing the characteristics of stress responses and stressors as independent entities can increase cohesiveness in stress research. Additionally, not all stressors are perceived by participants to have the same emotional or physiological ramification or impact. Researchers should attempt to address these confounds of perception in their studies.

Capturing Electronic Health Records

Certain administrative records, such as birth certificates, may hold information vital to aging studies (e.g., birth weight).

Elissa Epel, PhD, UCSF

Interdisciplinary Practice and Multiple Measures

Researchers should be encouraged to work outside of research silos and to integrate new strategies to investigate the causal mechanisms that drive biological aging. For example, social inequities and stress exposures may affect processes involved cellular restoration (e.g., repair, mitogenesis). In addition, there is a need to incorporate measures of resilience and hormetic stress to understand aspects of positive health and factors like emotional well-being, sleep, relaxation, and rest, which may delay or prevent biological aging. Social disparities are present in sleep data, which shows, for example, that women generally experience lower quality of sleep than men. Variability in sleep quality by gender is especially pronounced in individuals belonging to marginalized groups. More generally, the impact of sex and the reproductive life course on biological aging remains an important and understudied area of geroscience. Finally, geroscience researchers should determine the minimal study periods that reflect reliable changes in aging biomarkers.

Alan Cohen, PhD, Columbia University

Examining Concepts in Practice and Across Cultures

Stress and aging are broad concepts with numerous complexities and subcategories in both biological and social sciences. Social scientists and biology researchers study aging and stress by creating conceptual divisions based on exposure type, timing, and degree or quantity. Both biological and social scientists struggle with similar challenges of clearly describing, defining, and circumscribing key concepts. The field as whole should attempt to find a middle ground between high- and low-level granularity approaches to study these concepts.

Biological and social science researchers should also recognize that it is not yet clear whether language-based concepts in geroscience, such as “aging,” “resilience,” and “stress,” refer to distinct or overlapping biological and behavioral processes. Measures of resilience in a social context may differ markedly from measures of biological resilience and recovery. Examining the similarities and differences between aging and resilience may reveal conceptual revisions that illuminate new dimensions of geroscience research. Furthermore, culture may differently define and shape the experience of psychological, social, and biological stress. Thus, not all measures of stress will translate across cultural contexts.

Day 2 | February 19, 2024

Welcome and Introduction to Day 2

Welcome and Introduction to Day 2

Amanda Boyce, PhD, NIA

Epigenetic Clocks and Biological Aging in Search of Mechanisms

Luigi Ferrucci, MD, NIA

A delicate equilibrium exists between the rate of damage accumulation caused by entropy and resilience mechanisms designed to counteract and repair such damage. Aging, in this context, results from the faltering of these resilience mechanisms, leading to the emergence of entropy and stress-related outcomes such as chronic diseases and disability. Notably, the rate of aging depends not only on biological factors but also on environmental factors such as pollution, poverty, obesity, and personality traits.

Epigenetic clocks use specific changes in the epigenome to estimate biological age. The epigenome consists of chemical modifications to DNA and its associated proteins, which can influence gene expression without altering the underlying sequence. One well-known epigenetic clock is the DNA methylation clock, which measures changes in the methylation patterns of certain CpG sites in the genome. Changes in oxygen levels can affect the activity of enzymes involved in DNA methylation and demethylation processes. Thus, alterations in oxygen levels may contribute to epigenetic changes that are associated with aging. Such alterations can occur due to factors like sedentary lifestyles, arterial stiffness, stress, and anxiety.

The GESTALT study on human immune aging revealed a diverse array of methylation patterns in different cell types in the blood, challenging the assumption that the same changes occur synchronously across different cell types. Analysis of the age-associated hypo- and hypermethylated sites found that ARNT (HIF1β) and REST transcription factor motifs were significantly associated with hypoxia. These findings support the hypothesis that systematic methylation changes with aging may be induced by fluctuations in oxygen availability and energy metabolism. These fluctuations have been linked to fibrosis, impaired microcirculation, autonomic control imbalances, and diminished endothelial reactivity, particularly in nitroxide production. The culmination of these factors leads to local hypoxia and therefore influences the epigenetic clock.

Deciphering the epigenetic clock, not solely in the context of aging but also as a reflection of an organism’s past exposure and responses to different stressors, could offer a holistic understanding of the aging process and its interconnectedness with environmental and physiological factors. This holistic perspective may enable researchers to unravel the molecular imprints left by environmental exposures and stress responses, providing valuable insights into the multifaceted nature of aging and potential targets for interventions.

Session 3: What do social and behavioral scientists need to know about assessing change in the hallmarks of aging over time, including temporal features/variations in hallmarks of aging and normative trajectories?

Session 3: What do social and behavioral scientists need to know about assessing change in the hallmarks of aging over time, including temporal features/variations in hallmarks of aging and normative trajectories?

Moderator: Max Guo, PhD, NIA
Chair: Alan Cohen, PhD, Columbia University

Monitoring Health and Aging Using Deep Data

Michael Snyder, PhD, Stanford Medicine

Health and aging constitute complex interplays of diverse factors, with genetics accounting for merely 16 percent of the lifespan. Recent advancements in technology have revolutionized the ability to measure and understand various facets influencing health, including activity levels, dietary habits, and environmental exposures. The integrative Personal Omics Profile project represents a groundbreaking initiative that includes genome sequencing and continuous monitoring of the epigenome, transcriptome, proteome, and metabolome. To ensure a holistic understanding of an individual’s health, the program employs a multifaceted approach that integrates the analysis of gut microbiomes, regular clinical assessments, participant questionnaires, and wearables for continuous health monitoring. The project samples healthy individuals every three months and collects additional data during adverse events to allow for a deeper understanding of health changes over time. Notably, this approach has enabled researchers to detect major health complications pre-symptomatically, particularly in areas spanning hematology, cardiovascular health, and metabolic diseases. Additionally, longitudinal profiling within the project has revealed that individuals can have distinct aging patterns, or ageotypes. These patterns correlate to four major overlapping but distinct aging pathways that affect immunity, metabolic, liver, and kidney dysregulation.

Building on this innovative approach, remote monitoring, particularly leveraging smartwatches, has proven highly effective in early detection of infections, including the pre-symptomatic identification of COVID-19. Moreover, continuous glucose monitors have revealed personalized responses to different foods. Deep data profiling, enabled by longitudinal tracking and remote monitoring, has emerged as a powerful strategy for gaining comprehensive insights into health and aging. Tracking wellness enhances the potential for personalized health interventions by providing a real-time understanding of an individual’s physical, mental, and emotional states. In addition, the incorporation of environmental exposure monitoring complements the overall deep data profiling approach, establishing a robust framework for the ongoing tracking of health and aging.

Lessons Learned from Studying Extreme Human Longevity

Paola Sebastiani, PhD, Tufts Medical Center

Supercentenarians, defined as individuals aged 110 and older, experience notable delays in the onset of disability and morbidity. These findings are consistent with the compression of morbidity hypothesis, which suggests a reduced period between the onset of disability and the end of life in exceptionally long-lived individuals. Research on this phenomenon has focused on unraveling the factors influencing an individual’s health span to understand how certain genetic, environmental, and lifestyle factors can forestall the onset of age-related diseases and disabilities.

Various longitudinal studies, such as the New England Centenarian Study, Long Life Family Study, and Integrative Longevity Omics, demonstrate the pivotal role of genetics in promoting extreme old age in optimal physical condition. Researchers aim to translate genetic factors into molecular signatures that can serve as modifiable elements for a comprehensive understanding of health span promotion. Metabolomics findings within these studies have unveiled intriguing associations between specific metabolites and chronological age, longevity, and mortality. There is growing interest in the aggregation of these findings into metabolomic clocks that assess an individual’s biological age based on distinct metabolic markers and hold promise in providing a detailed and personalized understanding of aging processes.

Improving the collection of phenotypic data for longitudinal studies involves a comprehensive approach that integrates historical data with ongoing dynamic data collection. Dynamic data collection encompasses the continuous monitoring of individuals’ aging processes, including genetics (e.g., somatic mutations and DNA methylation), disability, morbidity, and molecular data. The central challenge of longitudinal studies lies in determining the optimal timing and frequency of these measurements, given the inherently dynamic nature of molecular data. The primary objective of aging studies revolves around characterizing overarching aging processes rather than merely identifying acute responses. Investigations, such as those into the apolipoprotein E (APOE) gene’s impact on extreme longevity, have yielded significant results by establishing associations between genetic variations and molecular signatures linked to lipid regulation and inflammation. However, exercising caution when interpreting findings is crucial because certain biomarkers exhibit varying behaviors across different age groups. Additionally, statistical biases may affect the accuracy of age prediction models, especially concerning older individuals. Hence, a rigorous process of investigation and validation is imperative before deploying omics clocks as reliable measures of biological age. Such meticulous scrutiny ensures the credibility and applicability of findings derived from phenotypic data in longitudinal studies. Despite the strides made in understanding aging, funding constraints pose a significant barrier to fully leveraging the extensive data generated from longitudinal studies. Addressing these challenges collectively will pave the way for robust, detailed, and personalized approaches to promote health span.

An Epigenetic Clocks Perspective

Raghav Sehgal, PhD Student, Yale University (Dr. Sehgal’s research was funded by the Gruber Foundation, Impetus Grants, and NIH)

Researchers have developed various biological clocks using DNA methylation signals, proteomes, phenomes, and clinical biomarkers to measure and predict an individual’s aging phenotype, including their pace of aging and time remaining until death. Researchers have developed three generations of epigenetic clocks based on CpG methylation. First generation epigenetic clocks (e.g., Hannum, Horvath1) measure an individual’s biological age; second generation epigenetic clocks (e.g., PhenoAge, GrimAge) predict time remaining until death; and third generation epigenetic clocks (e.g., DunedinPACE) predict the pace of aging. First- and second-generation epigenetic clocks yield nonspecific, error-prone predictions about an individual’s aging phenotype. To overcome these shortcomings, researchers conducted principal component analysis (PCA) to extract the covariance between multiple CpG sites and to create principal component (PC) epigenetic clocks (e.g., PCPhenoAge). Unlike first- and second-generation epigenetic clocks, PC epigenetic clocks use many CpG sites to predict aging phenotypes, which minimizes the probability of a single CpG site skewing data results. Additionally, age-related signals are highly unlikely to covary across many CpG sites, allowing PC epigenetic clocks to detect signals more accurately than first- and second-generation epigenetic clocks. PC epigenetic clocks are more reliable than first- and second-generation epigenetic clocks because they yield more specific aging phenotype predictions, and they are less error-prone.

The CENTRAL trial assessed how a low-fat diet affected epigenetic age. Researchers randomized 60 participants to a low-fat diet for 18 months. Epigenetic clocks were used to measure whole body aging as well as system-specific aging. Age deviation is associated both with the causes of aging (e.g., low education, low SES, body mass index [BMI]) and the outcomes of aging (e.g., mortality, frailty, diabetes, cardiovascular disease). No significant effects were found for whole-body aging. However, low-fat diets reduced the age deviation in the musculoskeletal system by 2.4 years, in the metabolic system by 1.85 years, in the kidneys by 1.2 years, in the heart by 0.95 years, in the blood by 0.55 years, and in inflammation by 1.7 years.

The PRISMO longitudinal study assessed the effects of military deployment-related PTSD on epigenetic age in 108 individuals who were deployed to Afghanistan. Blood samples were collected pre-deployment, one-month post-deployment, and six months post-deployment. As with the CENTRAL trial, PRISMO found no significant changes in whole-body aging, as measured by the second generation GrimAge epigenetic clock. However, examination of organ-specific epigenetic clocks found that six months after deployment, lung aging increased by 2.6 years, musculoskeletal aging increased by 2.1 years, liver aging increased by 1.9 years, heart aging increased by 1.3 years, and inflammation aging increased by 1.35 years. Thus, organ-specific epigenetic clocks appear to provide more detailed information than whole-body epigenetic clocks on the impact of social, psychological, and behavioral interventions.

Conceptual Challenges in the Measurement

Alan Cohen, PhD, Columbia University

Dr. Cohen defines aging as the inexorable loss of information over time in a complex adaptive system, which leads to declining capacity of the system to respond to changing conditions and perpetuate itself in its environment. Aging mechanisms include all processes that involve the loss of information over time. The hallmarks of aging are those mechanisms that are relatively universal. They can perhaps be used to measure the aging process. However, hallmarks of aging are not the only factors to consider when assessing aging because the hallmarks do not include specific aging mechanisms, such as wing wear in insects and diabetes in humans. Measuring aging is challenging because aging mechanisms are context-dependent and their role in the aging process of a biological system depends on the environment at a given time.

Resilience is an individual’s ability to return to a healthy biological state. Ecological theory on resilience states that low resilience and early warning signs, such as high variance, temporal autocorrelation, and cross-correlation, are indicators of critical transitions in complex systems. Critical transitions are abrupt shifts between two states and often result in a collapse of a complex system. Drawing on the ecological theory on resilience, researchers compared 11 biomarkers in blood panels of hemodialysis patients from Canada and Japan. Three months before death, biomarker variability dramatically increased. This shows that human biological resilience can be estimated from dynamic system properties, as in other complex systems.

Further research is needed for investigators to better define and measure aging. However, current biomarkers of aging are incomplete and can be difficult to interpret at the population level. Similarly, resilience is difficult to measure longitudinally on a large scale. Wearable technologies that continuously track participants’ medical, biological, and exercise data may help address these challenges.

Discussion

Aging Biomarkers in Social and Behavioral Sciences

Biomarkers of aging predict a diverse array of outcomes and exposures that accelerate the onset of age-related diseases, disabilities, and end of life. Second and third generation epigenetic clocks are better at measuring the biological process of aging than first generation epigenetic clocks. Therefore, social and behavioral scientists should use these later generation epigenetic clocks in future studies. It is essential to broadly assess molecular signatures of the biology of aging rather than assess only methylation markers thought to be causally involved in aging.

DNA Methylation

DNA hypermethylation and hypomethylation at CpG sites are strong indicators of aging. Organ-specific epigenetic clocks use DNA methylation signals to measure and predict an individual’s aging phenotype and are preferred over whole-body epigenetic clocks because of their specificity. However, it is still difficult to determine whether DNA methylation is causal or reactive to aging. Additionally, assessing DNA methylation requires blood draws, which are difficult to routinely collect in multi-year trials.

Timescales

Investigators must become more actively involved in defining the timescale of social and behavioral impacts on aging. Many environmental impacts on biological systems cannot be observed until years later. Typically, funding determines a study’s timescale, an approach which does not always yield the most useful data. For example, without repeated measurements over time, immunosenescence (the alteration of immune functions due to aging) appears to be a linear process occurring gradually in everyone, whereas, in fact, immunosenescence is nonlinear and the risk of experiencing it increases with age. To accurately detect immunosenescence in longitudinal studies, investigators must frequently obtain blood samples from their participants. Analyzing data from longitudinal studies can be challenging because of tracking difficulties when participants withdraw from a study or pass away.

Adaptive Capacity

Adaptive capacity refers to a biological system’s ability to adapt to its surroundings. To determine a biological system's adaptive capacity, investigators must develop measures that capture the system’s intrinsic health—the capacity of the biological system to maintain its internal, biological dynamic equilibrium. Existing measures that may capture intrinsic health include heart rate and homeostatic dysregulation, but neither is likely to be very precise. Ideally, adaptive capacity should be assessed at more than one time point, which can be accomplished through wearable technologies that continuously track medical, biological, and exercise data.

Biological Impacts of Aging

Aging involves a progressive loss of a biological system’s self-regulation in the face of environmental changes. As a biological system ages, it loses its ability to self-regulate at the cell, organelle, and organ levels, resulting in an overall breakdown. Therefore, the biological system is more vulnerable to environmental changes and has lower adaptability as it ages. To better understand the aging process, researchers should compare changes in participants’ epigenetic clocks to age-related changes in transcriptomic, proteomic, and metabolomic processes.

Session 4: What do social and behavioral scientists need to know about malleability/reversibility in the hallmarks of aging?

Session 4: What do social and behavioral scientists need to know about malleability/reversibility in the hallmarks of aging?

Moderator: Amanda Boyce, PhD, NIA
Chair: Stephen Kritchevsky, PhD, Wake Forest University School of Medicine

Generative AI for Drug Discovery, Biomarker Development, and Social & Behavioral Aspects of Aging Research

Alex Zhavoronkov, PhD, Insilico Medicine (Dr. Zhavoronkov is the chairman of the board, executive director, chief executive officer, and a shareholder of Insilico Medicine.)

“Artificial Intelligence in Longevity Medicine,” published in the Nature Aging journal in 2021, describes how deep generative reinforcement learning in artificial intelligence (AI) could be used to inform aging research, drug discovery, and clinical practices. Deep generative reinforcement learning is a learning mode in which computers interact with an environment and receive feedback to adjust their decision-making strategy. The paper described how computers solved complex problems by using deep neural networks (DNNs), which are machine learning algorithms that mimic the information processing of the brain. Researchers can add multiple data types to DNNs and train them to predict a subject’s age and health status, generate synthetic data, learn the fundamental human biology of aging, and understand the relationship between aging and disease.

Discovering a new drug takes approximately 12 years and costs billions of dollars. The most important part of drug discovery is identifying a therapeutic target and understanding the role it plays in the mechanism of disease. Generative AI can be used for therapeutic target identification, small molecule generation and optimization, preclinical pharmacology, clinical trial design, and biomarker development. AI can help investigators get their drugs to phase II clinical trials in under five years with a small budget, and generative AI-based tools can further accelerate the drug discovery process. For example, as described in “AlphaFold Accelerates Artificial Intelligence Powered Drug Discovery: Efficient Discovery of a Novel CDK20 Small Molecule Inhibitor,” three generative AI tools were able to identify a therapeutic target and synthesize small molecule inhibitors that treat hepatocellular carcinoma. Similarly, investigators at Insilico Medicine use the Pharma.AI platform to identify therapeutic targets and determine whether a target is implicated in age-related diseases, and they have leveraged the platform to create the world’s first AI-developed anti-fibrotic drug for idiopathic pulmonary fibrosis.

Dr. Zhavoronkov’s team has recently published several scientific papers showing that AI can predict a subject’s biological and psychological age. After processing biological, psychological, social, and behavioral data, AI found that biological age is strongly correlated with psychological age (which is defined as an AI-generated construct derived from self-reported measures of lifestyle, health, and wellbeing) and that optimism is an important factor for predicting psychological age. Additionally, investigators used AI to integrate multiple data types across all species, acquire knowledge about age and disease, and query for therapeutic targets associated with various diseases. Investigators at Insilico Medicine developed Precious3GPT, a multimodal platform that will help investigators integrate omics data, identify therapeutic targets, generate synthetic data, and better understand the aging process.

The Heat Shock Response: Proteostasis, Resilience & Neurohormonal Stress

Rick Morimoto, PhD, Northwestern University

The heat shock response (HSR) is among the most ancient cellular stress responses in biology and geroscience and a central component in stress resilience, which is the ability of an organism to adapt and survive exposure to diverse forms of environmental and physiological stress. The HSR is a universal genetic response to protect cells against environmental stress including elevated temperatures, oxidants, and other perturbations of protein stability and function that challenge healthy aging. The HSR together with the proteostasis network (PN) ensures that all proteins are properly expressed and functional during development and in health and protected against misfolding and aggregation when exposed to environmental stress and aging.

The HSR occurs when all cells and organisms are exposed to a transient acute heat shock of 5 to 10 degrees above ambient temperature resulting in the rapid induction of heat shock proteins also known as molecular chaperones. Preconditioning organisms to mild heat shocks can help them acquire thermotolerance and survive severe heat shocks later through faster activation of molecular chaperones. Molecular chaperones are essential for the synthesis, folding, assembly, and degradation of nearly all proteins in the cell and plays a crucial role in organismal survival under diverse stress conditions. Molecular chaperones together with the ubiquitin proteasome system and the autophagy-lysosome pathway are essential to prevent protein misfolding, aggregation and the accumulation of amyloid species in protein conformational diseases (e.g., Alzheimer’s disease, multiple myeloma, cystic fibrosis). However, as the organism ages, molecular chaperones get depleted, resulting in an accumulation of misfolded proteins and a less robust HSR. Failure of the HSR in aging is thought to contribute to all age-associated degenerative diseases.

The HSR is regulated by the transcription factor, heat shock factor 1 (HSF1), that binds to promoters of genes such as Hsp70. HSF1 is positively regulated by SIRT--a deacetylase that regulates metabolism and negatively regulated by chaperones such as Hsp70. After the HSR genes are transcribed, expression of chaperones, in turn, protects newly synthesized and metastable proteins from instability and becoming damaged. However, when organisms are chronically exposed to heat shock and other forms of stress, the HSR self-attenuates. During self-attenuation, the binding of Hsp70 to HSF1 represses HSF1 and the HSR, which could result in unharmful, misfolded proteins known as mild folding variants. It is normal for organisms to have some polymorphic variation that leads to mild folding variants. Typically, the proteostasis network then refold the proteins correctly; however, if the organism has a severe mutation, aggregation-prone proteins can accumulate, overwhelm the proteostasis network, and prevent the refolding of mild folding variants. Severe mutations can cause various diseases and result in the dysfunction of diverse cellular pathways as the organism ages. The ability of the HSR to respond robustly declines with aging; this is associated with signals from germline stem cells and reduced expression of a jumonji (JmjC) demethylase protein, which results in molecular chaperone chromatin being compressed by the polycomb repressive complex 2 thus preventing HSF1 from binding to the Hsp70 gene. Thus, the decline of the HSR in aging is programmed.

Previous studies have identified an intriguing relationship between neurohormonal stress and the HSR. Exposure of rats to neurohormonal stress is known to activate the hypothalamic-pituitary-adrenal axis and induce Hsp70 expression selectively in the adrenal cortex and vascular endothelium. However, stress does not induce Hsp70 gene expression in rats without pituitary glands; instead, investigators must administer exogenous adrenocorticotropic hormone (ACTH) to evoke a similar response. After ACTH is administered, the adrenal glands activate HSF1 and Hsp70. Stress induction of the Hsp70 gene declines with age and HSF1 levels remain constant, but HSF1 exhibits a decreased ability to bind to the Hsp70 gene over time. The age-related decline in Hsp70 gene expression likely increases the vulnerability of the organism to stress and further contributes to the aging process. It is worth noting that the expression of stress hormones such as the glucocorticoid receptor is strongly dependent upon molecular chaperones for function and response. Because each organism responds to differently to environmental stress, this may result in unique combinations of stress hormone regulation mechanisms. As humans age, the number of abnormal protein assemblies and lysosomes increases, resulting in neurodegeneration and cognitive impairment. Aging and ethnicity are a major risk factors for all neurodegenerative diseases. For example, Hispanics and African Americans have a higher risk of developing neurodegenerative diseases than Caucasians.

It has been unclear whether geroscience, as a field, is prepared to assess stress resilience at multiple levels, ranging from the cell and molecular level to the population level. However, Dr. Morimoto used this multi-level approach to describe the role of heat shock proteins and cellular stress responses in organismal physiology and susceptibility to disease. Through his multi-level geroscience research, Dr. Morimoto demonstrated that molecular stress caused by the accumulation of misfolded proteins is directly linked to organismal neurohormonal stress. He also proposed that rejuvenating the proteostasis network, molecular chaperones, and HSF1 activity could have beneficial effects for neurohormonal stress. Although aging is known to be associated with a decline in cellular stress responses and proteostasis networks, additional studies are required to determine how this molecular variation affects other aspects of health. In general, multi-level geroscience research will require identification of the right model organisms and research paradigms.

Malleability/Reversibility in Hallmarks of Aging: Inflammatory Biomarkers

Stephen Kritchevsky, PhD, Wake Forest University School of Medicine

The FDA recognizes 7 kinds of biomarkers. In geroscience most of the attention is on: (1) response biomarkers, (2) prognostic biomarkers, and (3) surrogate biomarkers. Investigators use response biomarkers to demonstrate whether therapeutic intervention evoked a biological response and use prognostic biomarkers to predict improved health outcomes in clinical trials. Surrogate biomarkers serve as substitutes for clinically meaningful outcomes. Surrogate markers are both response and prognostic markers, but even it a biomarker is a response and a prognostic marker it may not be a surrogate. A validated surrogate biomarkers help investigators identify potential therapeutic interventions; however, it is difficult to prove that a biomarker is an acceptable surrogate to the U.S. Food and Drug Administration (FDA) because they require trials showing efficacy with respect to a clinical outcome of interest to validate surrogate biomarkers. Examples of successful surrogate biomarkers are bone mineral density to predict fracture risk and blood pressure to predict stroke risk. The Health, Aging, and Body Composition (Health ABC) study used various biomarkers to identify risk factors for functional decline in older adults. Comparisons of findings from the Health ABC study with other aging studies identified four biomarkers as strong predictors of mortality in older individuals: (1) interleukin 6 (IL6), (2) grip strength, (3) walking speed, and (4) the forced expiratory volume in 1 second (FEV1).

Much of our current understanding of age-related biologic pathways stems from caloric restriction studies (CR) in animals. In humans, trials employing CR, typically to achieve weight loss, can help investigators anticipate which biomarkers will be responsive in trials of CR mimetic therapies. A meta-analysis of all intentional weight loss trials in older adults found a 15 percent reduction in all-cause mortality across all treatment arms, indicating that CR may help to prolong participants’ lifespans. Three additional weight loss studies, ADAPT, IDEA, and CLIP, found that CR can reduce IL6 plasma concentrations. In the SECRET weight loss study, the following biomarkers were used to construct a biomarker index: IL6, tumor necrosis factor receptor 1 (TNFR1), growth/differentiation factor 15 (GDF15), cystatin C, and N-terminal pro–B-type natriuretic peptide (NT-proBNP). In the SECRET study, neither CR nor exercise significantly changed specific biomarker concentrations, but CR did have a significant impact on the biomarker index. The planned TAME trial will aim to determine whether participants taking metformin experience delayed development or progression of age-related chronic diseases. Weight loss studies in older adults have shown that CR can reduce inflammatory biomarkers (e.g., IL6) and it is associated with lower all-cause mortality. However, it is unclear whether the reduction in all-cause mortality is caused by CR or lower plasma concentrations of inflammatory biomarkers. To understand whether inflammatory markers are ‘surrogates’ one would need to show that the extent to which CR lowers inflammatory markers reflects the extent to which mortality risk is also affected. Since health-related outcomes are strongly age-related, it may be that interventions late in the life-course are too late, and interventions to prolong participants’ lifespans may produce larger health benefits when performed earlier in the life course.

Discussion

Detection of Meaningful Change in Aging Studies

In aging studies, meaningful change is a relative term that changes depending on the desired or undesired outcomes of the studies. When assessing meaningful change, investigators must allow sufficient time for the biological and phenotypic changes under study to occur. Typically, it is unclear how biological changes are related to phenotypic changes and how the intensity of the intervention affects both changes. Some investigators have proposed that conducting animal studies could help them better understand the relationships between biological changes, phenotypic changes, and interventions; however, the results of animal studies do not always translate well to human studies. Animal study results are not always reproducible in human studies because the endpoint in both studies is usually all-cause mortality, and typically, animals have different causes of death than humans. Therefore, it is essential for investigators to identify the biological systems and pathways in animals that are similar to those in humans and design interventions that target those pathways.

Investigators must remember that it may take a long time to detect the meaningful changes induced by their interventions and to understand the difference between indicators of a biological system’s failure and indicators of meaningful change. For example, if an investigator discovers a misfolded protein in a subject, this is an indicator of a biological system’s failure that could have occurred several decades earlier. Additionally, when conducting aging studies, investigators should account for interindividual variation and the possibility of imprecise assays when documenting the assays results. Investigators still have not reached a consensus on the appropriate sample size and methods that are required to detect meaningful change in aging studies.

As participants age, their rate of aging does not occur linearly; instead, aging accelerates and decelerates at certain time points throughout the subject’s lifetime. Investigators should dedicate more resources to understanding how to identify those periods of acceleration and deceleration so that they can improve the quality of life of older adults through more effective interventions. To identify periods of aging acceleration and deceleration, investigators need longitudinal data so that they can understand each subject’s baseline characteristics and accurately identify when meaningful changes occur. In longitudinal studies involving participants in early to middle adulthood, there is still uncertainty on whether biomarkers such GDF15 and IL6 should be monitored, as these biomarkers may only be informative in older populations.

Inclusive Data in Clinical Trials and Datasets

During clinical trial recruitment, investigators should optimize diversity and recruit individuals from a variety of different backgrounds. Currently, clinical trials are not representative of the general population, which makes it difficult to assess the applicability of the trial’s interventions in the real world. Similarly, it is imperative that investigators ensure that the human datasets they add to DNNs are representative of the general population so that they will not introduce biases into their trials.

FDA Priorities

Many pharmaceutical companies do not allocate their resources to developing drugs that would slow aging because FDA does not define aging as a disease. Without an incentive to market their drugs and make money to offset clinical trial costs, it is unlikely that pharmaceutical companies will research new drugs to better manage aging. FDA is more concerned with treating chronic diseases in older adults than preserving physical strength in older adults. Likewise, many drugs that are approved by FDA treat chronic degenerative diseases, which occur when an individual loses their stress responsiveness and becomes maladaptive to their environment.

Session 5: Optimal study designs for bringing social and behavioral sciences and biological sciences together to better understand aging

Session 5: Optimal study designs for bringing social and behavioral sciences and biological sciences together to better understand aging

Moderator: Lis Nielsen, PhD, NIA
Chair: Eileen Crimmins, PhD, University of Southern California (USC)

The Role of the Behavioral and Social Sciences in Geroscience

Eileen M. Crimmins, PhD, USC

Analyses of nationally representative datasets can provide insight into the factors that contribute to variability in aging, including socioeconomic, behavioral, and psychological causes of differential aging. Since 1992, HRS has collected longitudinal data for cohorts of people over age 50. Similar ongoing studies across the globe have harmonized their data with HRS and one another, enabling large-scale parallel analyses. Starting in 2006, HRS incorporated biological measures in its data collection, initially physiological functioning, HbA1c, CRP, mainly cholesterol and blood pressure measures. In 2016, HRS began collecting extensive biological data (e.g., genome-wide association study [GWAS] data, markers of neurodegeneration, DNA methylation, transcriptomics, telomere length, and many additional biomarkers).

Using HRS biological data, researchers can study the factors that contribute to variability in aging outcomes, such as mortality, multimorbidity, cognitive dysfunction, and difficulties with activities of daily living (ADLs). Biological aging is a latent concept, and in some cases is modeled as such. Biological aging can be influenced by a variety of measures including genetics, Biological Age Score (e.g. composed of 22 biomarkers), Epigenetic Age Score (e.g. measured via GrimAge), transcriptomics, telomere length, and mitochondrial DNA copy number. We found this combination of biological variables explains less than 15 percent of the variability in mortality, multimorbidity, cognitive dysfunction, and difficulties with ADLs. A combination of these biological factors and social factors (e.g., demographics, SES, childhood health and hardship, adult trauma, psychological factors, and health behaviors), explains approximately 25 percent of the same variability in mortality, multimorbidity, and cognitive dysfunction, as well as approximately 16 percent of variability in difficulties with ADLs. Therefore, both social factors and biological processes related to resilience and adversity should be incorporated into the concept of how biologists and social scientists think about aging.

Importantly, not all health and aging outcomes will be explainable to the same extent and by the same factors. Improving overall understanding of factors contributing to variability in health and aging outcomes requires more specific definitions of: (1) the relative importance of the various hallmarks of aging, (2) linkages among these hallmarks, (3) linkages with downstream health and aging outcomes, and (4) linkages to SDoH.

Sampling, Biospecimen, and Molecular Marker Considerations for Aging Health Research

Allison Kupsco, PhD, Columbia University

When designing research projects centered around human biosamples, researchers should consider: (1) the research question, (2) feasibility of collecting biosamples of interest, (3) cell types present in different types of biosamples, and (4) whether a cohort already exists with available biosamples. Laboratory quality control practices can help ensure accurate and replicable results of biosample assays. Biosamples should be stored at appropriate temperatures for long-term storage and aliquoted into small volumes to eliminate the need for multiple freeze-thaw cycles, which can damage sample quality. Prior to beginning biosample assays, technicians should evaluate sample quality.
Detailed procedure information should be recorded for each assay run, including the name of the technician running the assay, assay run date, and positions of biosamples and plates. All these factors can introduce variability in results. Proper randomization of sample and plate positions can mitigate this variability. To ensure replicability, 5 to 10 percent of samples should be run in duplicate, even for assays with well-established pipelines. Lastly, assay result quality should be monitored and evaluated.

Researchers currently use various well-established biomarkers as well as some emerging and alternative aging biomarkers. Epigenetic clocks are the most well-studied biomarkers of aging across the full life course. Other biomarkers, such as glucocorticoid exposure scores as a readout of stress and inflammaging scores for inflammation, may also provide relevant information about the aging process and factors that influence aging.

Multiomics approaches as well as extracellular vesicle (EV) biomarkers are also used in aging studies. Researchers are currently developing proteomic, transcriptomic, and chromatin immunoprecipitation (CHIP)-based clocks as biomarkers. Integration of multiple biomarkers of aging can provide a more comprehensive and detailed view of the life course. EV-based biomarkers may provide detailed information about intercellular communication, cellular function, and mechanistic information. Neuron-derived EVs are released from the nervous system and can cross the blood-brain barrier, enabling EV collection via blood sampling. Researchers can isolate neuron-derived EVs using neuronal and microglial markers for downstream assays and analyses that provide information on neuronal function. Newer aging biomarkers still in development may be ready for use at smaller scales. Prior to scaling these biomarkers for use in hundreds of thousands of samples, researchers need to consider biomarker stability over time, overall generalizability, and causality, as well as further streamline technical assay and analysis pipelines.

Optimal Study Designs for Bringing the Social and Behavioral Sciences and the Biological Sciences Together to Better Understand Aging

Daniel Belsky, PhD, Columbia University

Molecular changes that accumulate over the lifespan lead to loss of integrity and resilience capacity of cells, tissues, and organs, causing a decline in functional capacity and onset of chronic disease, disability, and ultimately death. In laboratory animals, slowing or reversing the accumulation of these molecular changes can slow decline in system integrity and extend healthy lifespan. Measurement of these molecular changes in humans is challenging. However, new data science methods that integrate omics data and large-scale clinical data make it possible quantify these biological processes of aging through data-driven proxy measures. These proxies allow for measurement of biological aging from early in the lifespan.

Integrating multiple components into aging scores allows for assessment of risks for aging phenotypes and responsiveness to interventions to prevent disease onset. Social and behavioral scientists can use these molecular and cellular changes to track aging trajectories across the lifespan before aging symptoms are readily apparent.

While molecular data can be used to derive composite biomarkers of aging, the use of molecular data to make measurements does not directly translate into measurement of molecular mechanisms of aging. In general, data-driven methods for the quantification of aging are agnostic to biological mechanisms. Epigenetic, proteomic, and metabolomic clocks are not measures of the hallmarks of aging; instead, they are measures of molecular signatures downstream from core mechanistic and biological processes that cause the aging process. For example, epigenetic clocks estimate biological aging from DNA methylation. However, based on molecular experiments, the relationship between DNA methylation and gene expression is bidirectional. The interpretation of molecular clocks is further complicated by cell type heterogeneity in biosamples. Currently, observed changes in DNA methylation-based composite measurements are difficult to distinguish from changes in the abundances of cell types with distinct DNA methylation patterns. In addition, molecular changes may be upstream or downstream from tissue and organ changes that occur during the aging process.

In addition to the identification, validation, and refinement of measures of biological aging, future research should focus on establishing causal relationships between molecular changes and aging phenotypes. Dr. Belsky recommended the following to support future aging research:

1. Support for population-representative longitudinal datasets to enable studies examining which aging biomarkers predict aging and health outcomes.
2. Support studies using sibling difference designs to correlate differences in the exposome with differences in aging biomarkers.
3. Support studies using quasi-experimental approaches to estimate causal relationships between exposures and aging and health outcomes.
4. Support randomized controlled trials assessing the effects of interventions on aging biomarkers, putative biological mechanisms, and aging and health outcomes.
5. Provide dedicated awards and/or supplements to develop improved data sharing infrastructure to increase data sharing speed
6. Support the formation of interdisciplinary consortia to define best practices for aging research and platforms for disseminating these best practices.

Working with Environmentally Diverse Populations

Amanda Lea, PhD, Vanderbilt University

Many diseases of aging are rare in subsistence-level communities, likely because of lifestyle differences. Accelerometry data show that subsistence-level groups have much higher levels of physical activity compared to urban groups. The ultra-processed foods that make up more than half of the American diet are unavailable in subsistence-level settings. Inflammation patterns are also different between subsistence-level communities and high-income settings—inflammation in subsistence-level communities is dynamic and pathogen-driven, while inflammation in high-income countries is chronic, sterile, and age-associated. In addition, while there is wide variability across societies, wealth inequality is lower and social support is higher in subsistence-level societies compared to societies in high-income countries.

To determine the types, timing, and biological mechanisms of environmental factors that matter for aging processes, researchers study within-population variation using a combination of anthropological (e.g., ethnographies) and biological strategies (e.g., biosample collection). By partnering with populations undergoing various types and degrees of acculturation, market integration, and exposure to urban environments within a single generation, researchers gain access to populations with a wide range of exposure to urban lifestyles and can assess the impacts of urban lifestyles on health and aging.

The Turkana Health and Genomics Project is an ongoing study of the Turkana people who have historically practiced nomadic pastoralism in remote and rural areas of northwest Kenya. The Turkana people are rapidly transitioning to more urban and westernized lifestyles. Various data types are collected under this project:

• survey data for diet, subsistence, housing, built environment, health, reproductive history, family and social environment, and early life experiences;
• cardiometabolic health measures for waist and hip circumferences, BMI, body fat percentage, lipid panel, blood glucose level, hemoglobin A1C (HbA1c), blood pressure, arterial stiffness, and arterial wall thickness;
• physical function and health measures for hand and knee x-rays, bone mineral density, leg and hand grip strengths, sit stand test, walking speed, and accelerometry;
• bioassay data for genetic variation, complete blood count, cytokines, CHIP, DNA methylation, and gene expression.

Initial data from this project show that on average, urban adults have worse cardiometabolic health. Notably, individuals in rural areas with higher SES have better health compared to those living in urban areas with higher SES. In addition, early life and current lifestyle factors can have additive effects on adult health. However, exposure to early life adversity does not necessarily predict adult health.

The Turkana Health and Genomics Project, along with other similar projects, utilize integrative anthropological and biological approaches, enabling the study of systemic and structural factors that drive lifestyle change and health at the individual level. This integrative approach also enables the study of determinants of health and the potential biological mechanisms that link these determinants to aging and health outcomes. Subsistence-level communities undergoing rapid acculturation often exhibit extreme between-individual and within-lifetime environmental variation—an important consideration when calculating study statistical power. Lastly, by partnering with communities, researchers can help address community-driven questions.

Challenges with the Turkana Health and Genomics Project design include: (1) difficulty collecting cross-sectional and longitudinal data, (2) presence of multiple confounds due to the nonrandomization of environments, and (3) lack of validation of common measurement tools in subsistence-level populations. However, studying aging beyond high-income contexts enables the identification of potentially modifiable factors and strategies to improve health and equity in biomedicine. These studies can help elucidate which drivers of aging are generalizable versus culturally specific and how different social structures (e.g., unequal versus egalitarian structures) affect biological outcomes of aging.

Discussion

Dimensionality of Biological Aging and the Relationship of Biological Aging and Its Measures

Multidimensional models of biological aging use multiple measures, and the inclusion of multiple measures may introduce a higher degree of error into models due to inherent errors in each measure respectively. If multiple measures assess the same dimension of aging, combining them without consideration of measurement error can reduce overall accuracy. In addition, the relationships between the concept of biological aging and the measures of biological aging often used remains unclear—more work needs to address if or when a measure could be a cause of biological aging or the result of biological aging. These relationships can have significant implications for analysis of aging biomarkers especially given that most data reduction techniques assume perfect measurement of the included variables.

Use of Biomarkers to Understand Biological Mechanisms of Aging

Biomarkers of aging can serve multiple functions, depending on perspective. From the clinical perspective, biomarkers are predictive tools; the actual measure itself matters less than its ability to predict aging and detect responses to aging interventions. Biomarkers can also act as surrogate endpoints of aging, enabling assessment of aging interventions on a much shorter timeframe. From the biological and mechanistic perspective, biomarkers are essential for mechanistic studies to understand the underlying mechanisms of aging. These mechanistic studies can help inform clinical study designs. Ultimately, aging biomarkers may help bridge research involving aging interventions and research on biological mechanisms.

Generalizability of Aging Biomarker Clocks

Methylation and other molecular clocks are now trained on health measures rather than on chronological age, and they are inherently different empirically and philosophically from older generations of aging clocks. Some newer epigenetic clocks with robust predictive capacities are derived from immune cell measures; however, these are subject to confounds, such as variation in different immune cell abundancies over time and short-term changes irrelevant to aging. However, future clocks derived from immune cell measures will likely control for cell type abundancies. In addition, despite validation across hundreds of studies over the globe that show robust predictive ability for morbidity and mortality, these clocks may lack generalizability to outlier groups (i.e., very long-lived individuals). Newer aging clocks should be tested in long-lived populations to better understand their generalizability.

The construction of aging clocks is informed by the geroscience hypothesis that links aging with decline in health. However, assuming older individuals are sick and younger individuals are healthy can have implications for downstream interpretations of results. Therefore, researchers are also developing other aging measures that represent the ways aging damages the body, including hazard of mortality and rate of decline in the integrity of multiple organ systems.

Biomarkers for Past Environmental Exposures

Using omics to measure past environmental exposures requires consideration of sensitivities to short-term changes and to the stability of markers of past exposures. For example, epigenetic markers of smoking remain stable for decades, while the timescales for transcriptomic and metabolomic changes in response to smoking are much shorter. Compared to newer omics, epigenomics has been extensively studied and standardized in the context of aging. Metabolomics and lipidomics lack standardization, but ultimately they may provide critical insights regarding environmental exposures and biological mechanisms.

Development of Stress Biomarkers

Stress biomarkers have the potential to link SDoH to aging trajectories. However, hormonal stress biomarkers have largely failed for measuring stress because they are also involved in many basic metabolic processes as well as stress responses. Newer stress biomarkers within individual cell types may be relevant to aging. However, these stress biomarkers likely affect the hallmarks of aging in opposite directions due to complex compensatory mechanisms, and some of them may be incidental consequences of aging that are not linked to health outcomes, similar to gray hair. Therefore, more complex networks of measures will likely be required to measure stress in a more comprehensive way. Using neural networks and machine learning enables an unbiased approach to identifying networks of physiological and molecular changes that are most informative for stress in the context of aging.

Bridging Individual- and Population-Level Studies

Truly understanding disparities in aging trajectories requires both individual- and population-level studies. By using a sequential process, population-level studies can identify drivers of accelerated aging, which, in turn, informs the design of subsequent individual-level studies aimed at understanding the mechanisms of how these factors impact aging. However, mechanistic findings of causality at the individual level with a significant magnitude of effect do not always translate to explaining the macro-level population differences in aging outcomes. This lack of translatability may be because some social factors impact the individual while certain ones, such as racism and discrimination, occur at the population level.

Establishing causal relationships at the population level requires specific study designs. Some social scientists make causal inferences based on emulated clinical trials and family-based methods and samples. Findings from quasi-experimental studies can also build evidence toward causality at the population level. However, these approaches require more nationally representative data, such as that available in the HRS international family of studies. In addition, data on long-lived populations are becoming increasingly available through the Exceptional Longevity Translational Resources (ELITE) portal. Challenges to identifying causality using observational data include the lack of randomization and differential exposure to perturbations. For example, in subsistence-level societies undergoing rapid acculturation, certain inherent individual traits and factors may dictate the level of a person’s exposure to acculturation. Implementation of quasi-experimental studies likely requires partnership with smaller population aging studies that may have more flexibility to alter their protocols than large, nationally representative studies such as HRS. Proper design of quasi-experimental and family-based studies of aging requires expertise from social, behavioral, and biological scientists.

Closing Remarks

Closing Remarks

Max Guo, PhD, NIA

Based on workshop discussions, three identified main takeaways were identified:

1. There is a need for better biomarkers, such as comprehensive panels of measures, as well as a better understanding of the causal processes linking these biomarkers to various aging processes.
2. There is a need for more accurate, objective, and quantitative measures and proxies for measuring behavioral and social phenomena to better understand connections between aging and behavioral and social factors.
3. Bridging social and behavioral factors with biological aging may require better biological indicators of eustress (including hormetic stress), resiliency to stress, distress, and levels of stress exposure.

Day 1 (Wednesday, Feb. 14)

11-11:05 a.m.: Welcome

• David Braudt , PhD, Program Official, Population and Social Processes Branch, DBSR

11:05-11:20 a.m.: Opening Remarks

• Lis Nielsen , Ph.D., director, Division of Behavioral and Social Research, NIA
• Stacy Carrington-Lawrence , Ph.D., acting director, Division of Aging Biology, NIA
• Richard J. Hodes , M.D., director, National Institute on Aging

11:20-11:35 a.m.: Speakers introduce themselves (name, university, field[s])

11:35 a.m.-12:15 p.m.: Changes across fields since the first workshop

• 11:35-11:45 a.m.: Terrie Moffitt, Ph.D., Duke University and King’s College London
• 11:45-11:55 a.m.: George Kuchel , M.D., University of Connecticut
• 11:55 a.m.-12:15 p.m.: Discussion/Q&A

12:15-12:30 p.m.: Break

12:30-2:20 p.m.: Session 1 | What do biological scientists need to know about social determinants of health (SDoH) and studying the causal impact of SDoH on biological aging?

The goal of this session is to provide an overview of:

• Definitions and operationalizations of SDoH
• The importance of longitudinal assessments of social exposures and the use of longitudinal methods — including the incorporation of time-varying and confounding variables — used by social scientists to study the impact of SDoH on aging
• Data sources leveraged by social and behavioral scientists to study the effects of social and behavioral determinants of health and aging across the life course
• The successes and limitations of recent work in the field

NIA Moderator: David Braudt , PhD, Program Official, Population and Social Processes Branch, DBSR

• 12:30-12:50 p.m.: Kenneth A. Bollen , Ph.D., Session 1 chair — University of North Carolina, Chapel Hill
• 12:50-1:10 p.m.: Lauren Schmitz , Ph.D., University of Wisconsin, Madison
• 1:10-1:30 p.m.: Mateo Farina , Ph.D., University of Texas at Austin
• 1:30-1:50 p.m.: Hiram Beltrán-Sánchez , Ph.D., University of California, Los Angeles

Each speaker is asked to: (1) describe the theoretical underpinnings of one or two prototypical projects from their program of research and (2) identify the strengths and weaknesses of their approach to studying the link between SDoH and biological aging.

1:50-2:20 p.m.: Session 1 Discussion/Q&A

Guiding Questions for the Discussion:

• How might the weaknesses of the measures, data, and/or methodology used to study the impact of SDoH on aging be addressed?
• What are the gaps in extant data resources that limit the research and questions pursued by social and behavioral scientists conducting research integrating social and behavioral science and geroscience?
• What are the characteristics of “ideal” data sources for studying links between SDoH and biological aging and to what extent are relevant biospecimens accessible for future/additional assays?

2:20-2:35 p.m.: Break

2:35-4:25 p.m.: Session 2 | What do biological scientists need to know about how psychologists and other social/behavioral scientists examine the impact of acute and chronic stressors on biological processes of aging?

The goal of this session is to provide an overview of:

• The typology that informs how behavioral and social scientists conceptualize and measure stress, including differentiation between stressor exposures and responses to stressors
• The experimental, observational, and longitudinal methods used to examine the link between stressor exposure across the lifespan, psychological and physiological responses to stressors, aging biology, and mortality
• The theoretical and operational definitions used to characterize individual-level and population-level “resilience” to stressors — or the ability to “bounce back from, or successfully adapt to stressors” — in biobehavioral and biosocial research.

NIA Moderator: Emily Hooker , PhD, Program Official, Individual Behavioral Processes Branch, DBSR

• 2:35-2:55 p.m.: Elissa Epel , Ph.D., Session 2 chair — University of California, San Francisco
• 2:55-3:15 p.m.: George Slavich , Ph.D., University of California, Los Angeles
• 3:15-3:35 p.m.: Jessica Chiang , Ph.D., Georgetown University
• 3:35-3:55 p.m.: Rebecca Reed , Ph.D., University of Pittsburgh

Each speaker is asked to: (1) describe the theoretical underpinnings of one or two prototypical projects from their program of research, (2) highlight experimental and/or observational methods used in their field to study stress, and (3) identify the strengths and weaknesses of their approach to studying the link between stressor exposure and aging biology.

3:55-4:25 p.m.: Session 2 Discussion/Q&A

Guiding Questions for the Discussion:

• What questions remain about the links between stressor exposure, aging biology, and healthspan?
• Are the impacts of stressors and stress responses on health reversable and how best might researchers measure and/or model this?
• How and when should assessments of stress reactivity and stress resilience be included in investigations of the association between stress exposures and biological aging?

4:25-4:30 p.m.: Break

4:30-4:55 p.m.: Day 1 Summary & Discussion Closing Remarks

NIA Moderator: Emily Hooker , PhD, Program Official, Individual Behavioral Processes Branch, DBSR

• Kenneth A. Bollen , PhD, University of North Carolina, Chapel Hill
• Alan Cohen, PhD, Columbia University
• Elissa Epel , PhD,University of California, San Francisco

4:55-5:00 p.m.: Closing Remarks

• Emily Hooker , PhD, Program Official, Individual Behavioral Processes Branch, DBSR

Day 2 (Thursday, Feb. 29)

11-11:05 a.m.: Welcome

• Amanda Boyce , PhD, Chief of the Aging Physiology Branch, Division of Aging Biology, NIA

11:05-11:25 a.m.: Introduction to Day 2

• Luigi Ferrucci , M.D., scientific director, National Institute on Aging

11:25 a.m.-1:10 p.m.: Session 3 | What do social and behavioral scientists need to know about assessing change in the hallmarks of aging over time, including temporal features/variations in hallmarks of aging and normative trajectories?

The goals of this session are to share:

• The current understanding of our ability to capture meaningful change in the hallmarks of aging over time; and
• Examples of experimental manipulations and/or environmental exposures used in aging biology to examine change in markers of biological aging; and
• The extent to which measures of biological resilience are used to capture change in biological aging and how these measures are operationalized.

NIA Moderator: Max Guo , PhD, Chief of the Cell Biology Branch, Division of Aging Biology, NIA

• 11:25-11:45 a.m.: Michael Snyder , Ph.D., Stanford Medicine
• 11:45-12:05 p.m.: Paola Sebastiani , Ph.D., Tufts Medical Center
• 12:05-12:25 p.m.: Raghav Sehgal , Ph.D. student, Yale University
• 12:25-12:45 p.m.: Alan Cohen , Ph.D., Session 3 chair, Columbia University

Each speaker is asked to: (1) describe the theoretical underpinnings of one or two prototypical projects from their program of research and (2) identify the strengths and weaknesses of their approach to studying the biology of aging over time.

12:45-1:10 p.m.: Session 3 Discussion/Q&A

Guiding Questions for the Discussion:

• How are biological scientists addressing divergences in the time scales between environmental or experimental exposures and biological change? How are they approaching and modeling time-varying and confounding variables?
• What resources are lacking to conduct research on trajectories of change in biological measures of aging?
• How and when should measures of biological, physiological, or functional resilience be included in investigations of change in biological aging?

1:10-1:30 p.m.: Break

1:30-2:55 p.m.: Session 4 | What do social and behavioral scientists need to know about malleability/reversibility in the hallmarks of aging?

The goals of this session are to provide an overview of:

• Measuring, operationalizing, and interpreting meaningful changes in aging hallmarks post-intervention, and consideration of complex interactions and confounds; and
• Theoretical and empirical definitions related to models of biological resilience in aging intervention studies; and
• Consideration of the timeframe of resilience and modeling time-varying and dynamic changes; and
• Expectations of efficacy in interventions and other practical considerations.

NIA Moderator: Amanda Boyce , PhD, Chief of Aging Physiology Branch, Division of Aging Biology, NIA

• 1:30-1:50 p.m.: Alex Zhavoronkov , Ph.D., Inscilico Medicine
• 1:50-2:10 p.m.: Richard Morimoto , Ph.D., Northwestern University
• 2:10-2:30 p.m.: Steve Kritchevsky , Ph.D., Session 4 chair, Wake Forest University

Each speaker is asked to: (1) describe the theoretical underpinnings of one or two prototypical projects from their program of research and (2) identify the strengths and weaknesses of their approach to studying interventions to slow or reverse aging.

2:30-2:55 p.m.: Session 4 Discussion/Q&A

Guiding Questions for the Discussion:

• For which hallmarks of aging is there good evidence of malleability and/or reversibility, and what constitutes meaningful change?
• What is known about the dose, timing, and intensity of interventional exposure needed to see meaningful change? How are these changes best measured?
• What data are needed to test the most pressing questions?

2:55-3:05 p.m.: Break

3:05-3:45 p.m.: Session 5 | Optimal study designs for bringing the social and behavioral sciences and the biological sciences together to better understand aging?

With the goal of catalyzing geroscience-relevant research at the intersection of social and behavioral research and aging biology to elucidate change and potential reversibility of biological aging, this session will:

• Consider data sets/study designs that are ideal for research at the intersection of social/behavioral and aging biology in human populations; and
• Identify opportunities to capitalize on existing data collection efforts to evaluate the link between social determinants of health, stress exposures, and aging biology; and
• Identify the characteristics of ideal longitudinal research for examining geroscience in human populations; and
• Identify the characteristics of an ideal social/behavioral interventional study for geroscience research in humans; and
• Identify remaining gaps and opportunities for data collection and/or study design to better understand factors driving individual or population/group differences in trajectories of biological aging.

NIA Moderator: Lis Nielsen , PhD, Director, Division of Behavioral and Social Research, NIA

• 3:05-3:15 p.m.: Eileen Crimmins , Ph.D., Session 5 chair, University of Southern California
• 3:15-3:25 p.m.: Daniel Belsky , Ph.D., Columbia University
• 3:25-3:35 p.m.: Allison Kupsco , Ph.D., Columbia University
• 3:35-3:45 p.m.: Amanda Lea , Ph.D., Vanderbilt University

Each speaker is asked to: (1) briefly describe their field(s) and focus of their research and (2) present study-design considerations for future, geroscience-relevant, research at the intersection of social and behavioral research and aging biology.

3:45-4:50 p.m.: Open Discussion | All panelists and speakers from Day 1 and Day 2

The goals for the open discussion are to articulate principles for appropriate study designs to elucidate the extent to which psychological, social, or environmental exposures impact biological aging, identify priority research gaps and current opportunities for integrating approaches across fields.

• Speakers and panelists will also be able to make a comment on one of the following based on the discussions from the meeting:
  • What will I do differently in my research program? What concrete steps would be helpful at bridging these fields? What is my main takeaway from the meeting?

4:50-5 p.m.: Closing Remarks

• Stacy Carrington-Lawrence , Ph.D., acting director, Division of Aging Biology, NIA
• Lis Nielsen , Ph.D., director, Division of Behavioral and Social Research, NIA

Appendix 2: Participants

listed alphabetically under header

NIA Workshop Planning Committee

Amanda Boyce, Program Director and Chief of Aging Physiology Branch, Division of Aging Biology, NIA
David Braudt, Program Officer, Population and Social Processes Branch, Division of Behavioral and Social Research, NIA
Stacy Carrington-Lawrence, Deputy Director, Division of Aging Biology, NIA
Melissa Gerald, Program Director, Division of Behavioral and Social Research, NIA
Max Guo, Chief, Cell Biology Branch, Division of Aging Biology, NIA
Emily Hooker, Program Officer, Individual Behavioral Processes Branch, Division of Behavioral and Social Research, NIA
Jonathan King, Senior Scientific Advisor to the Division Director, Division of Behavioral and Social Research, NIA
Marianna Molina, Social Science Analyst, Division of Behavioral and Social Research, Individual Behavioral Processes Branch, NIA
Lis Nielsen, Director, Division of Behavioral and Social Research, NIA
Janine Simmons, Deputy Director, Office of Behavioral and Social Sciences Research, NIH
Fei Wang, Chief, Translational Research Branch, Division of Aging Biology, NIA

External Workshop Steering Committee

Eileen Crimmins, AARP Chair in Gerontology, Director, Center on Biodemography and Population Health, University of Southern California Leonard Davis School of Gerontology
Elissa Epel, Professor, Psychiatry, Weill Institute for Neurosciences, University of California, San Francisco
George Kuchel, Travelers Chair in Geriatric and Gerontology, Professor of Medicine, Director, Center on Aging, University of Connecticut
Terrie Moffitt, Duke University and King’s College London

Invited Speakers

Daniel Belsky, Associate Professor, Epidemiology, Columbia University
Hiram Beltrán-Sánchez, Associate Professor, Community Health Sciences, University of California-Los Angeles
Kenneth Bollen, Henry Rudolph Immerwahr Distinguished Professor, Psychology and Neuroscience, Sociology, Head of Methods Unit at the Carolina Population Center, University of North Carolina-Chapel Hill
Jessica Chiang, Assistant Professor, Psychology, Georgetown University
Alan Cohen, Associate Professor and Chair in Biological Complexity and Health Longevity, Department of Family Medicine and Emergency Medicine, Butler Columbia Aging Center, Mailman School of Public Health, Columbia University
Mateo Farina, Assistant Professor, Department of Human Development and Family Sciences, Population Research Center, University of Texas at Austin
Steve Kritchevsky, Professor, Gerontology and Geriatric Medicine, Wake Forest University School of Medicine
Allison Kupsco, Assistant Professor, Columbia University
Amanda Lea, Assistant Professor, Department of Biological Sciences, Vanderbilt University
Richard Morimoto, Bill and Gayle Cook Professor of Biology, Director of the Rice Institute for Biomedical Research, Northwestern University
Rebecca Reed, Assistant Professor, Psychology, University of Pittsburgh
Lauren Schmitz, Assistant Professor, La Follette School of Public Affairs, University of Wisconsin-Madison
Paola Sebastiani, Professor, Tufts Medical Center
Raghav Sehgal, PhD Student, Yale University
George Slavich, Professor, Department of Psychiatry and Biobehavioral Sciences, University of California, Los Angeles
Michael Snyder, Chair, Department of Genetics, Director, Center for Genomics and Personalized Medicine, Stanford University
Alex Zhavoronkov, Founder and CEO, Insilico Medicine

NIA Staff in Attendance

Siobhan Addie, Health Science Policy Analyst, Division of Aging Biology, NIA
Christy Carter, Program Officer for Training and Workforce Development, Division of Aging Biology, NIA
Michele Evans, Senior Investigator (Clinical), Office of the Scientific Director, NIA
Luigi Ferrucci, Scientific Director, NIA
Roberto Flores, Director, Aging Microbiome & Virome Program, NIA
Jennifer Fox, Program Officer, Translational Research Branch, Division of Aging Biology, NIAYih-Woei Fridell, Program Officer, Cell Biology Branch, Division of Aging Biology, NIA
Yi-Ping Fu, Health Scientist Administrator, Cell Biology Branch, Division of Aging Biology, NIA
Hongwei Gao, Program Officer, Aging Physiology Branch, Division of Aging Biology, NIA
Evan Hadley, Director, Division of Geriatrics and Clinical Gerontology, NIA
Erin Harrell, Program Official, Individual Behavioral Processes Branch, Division of Behavioral and Social Research, NIA
Richard Hodes, Director, NIA
Petra Jacobs, Director, Office of Behavioral and Social Clinical Trials, Division of Behavioral and Social Research, NIA
Dinesh John, Program Director, Digital Health and Technology Solutions, Individual Behavior Processes Branch, NIA
Lyndon Joseph, Program Officer, Division of Geriatrics and Clinical Gerontology, NIA
Pragati Katiyar, Program Officer, Division of Aging Biology, NIA
Theresa Kim, Program Officer, Population and Social Processes Branch, Division of Behavioral and Social Research, NIA
Hans-Peter Kohler, Frederick J. Warren Professor of Demography Professor of Sociology, Co-Director, Population Aging Research Center, NIA
Tong Li, Program Director, Neurobiology of Aging and Neurodegeneration Branch, Division of Neuroscience, NIA
Janetta Lun, Scientific Review Branch Section Chief, Social and Behavioral Sciences, NIA
Nicholas McNeill, Division of Behavioral and Social Research, NIA
Jennifer Merickel, Clinical Trials Coordinator and Health Science Administrator, Clinical Trials Office, NIA
Manuel Moro, Health Scientist Administrator, Translational Research Branch, Division of Aging Biology, NIA
Carmen Moten, Health Scientist Administrator, Scientific Review Branch, NIA
Ann Namkung, Program Director of HIV and Aging Research, Division of Aging Biology, NIA
Liz Necka, Program Director, Individual and Behavioral Processes Branch, Division of Behavioral and Social Research, NIA
Maggie Nellissery, Program Officer, Clinical Trials Branch, Division of Geriatrics and Clinical Gerontology, NIA
Carol Nguyen, Scientific Program Specialist, Division of Geriatrics and Clinical Gerontology, NIA
Lisa Onken, Director, Behavior Change and Intervention Program, Division of Behavioral and Social Research, NIA
Andras Orosz, Program Director, Cell Biology Branch, Division of Aging Biology, NIA
Gianina Ramona Dumitrescu, Chief, Clinical Sciences Section, Scientific Review Branch, NIA
Melissa Riddle, Health Scientist Administrator, Division of Behavioral and Social Research, NIA
Rajasri Roy, Scientific Review Officer, Clinical Sciences Section, Scientific Review Branch, NIA
Irina Sazonova, Program Officer, Clinical Trials Branch, Division of Geriatrics and Clinical Gerontology, NIA
Luke Stoeckel, Program Director, Mechanistic and Translational Decision Science, Division of Behavioral and Social Research, NIA
Dory Sullivan, Program Analyst, Office of the Director, Division of Behavioral and Social Research, NIA
Mulualem Tilahun, Program Officer, Aging Physiology Branch, Division of Aging Biology, NIA
Delany Torres Salazar, Training Officer, Division of Behavioral and Social Research, NIA
John Williams, Program Officer, Aging Physiology Branch, Division of Aging Biology, NIA
Beth Wilmot, AAAS Science & Technology Policy Fellow, NIA
Sue Zieman, Medical Officer, Division of Geriatrics and Clinical Gerontology, NIA

Other Attendees

Lisa Barry, Associate Professor of Psychiatry, University of Connecticut Center on Aging
Alessandro Bartolomucci, Professor of Integrative Biology and Physiology, University of Minnesota
Ilaria Bellantuono, Professor, Musculoskeletal Aging, University of Sheffield
Natalia Bobba Alves, Postdoctoral Researcher, Columbia University
Sam Caton, Senior Lecturer, Population Health, University of Sheffield
Michelle Chang
Alan Cohen, Associate Professor of Environmental Health Sciences, Robert N. Butler Columbia Aging Center, University of Columbia
Charlie Collinge, Research Assistant, University of Minnesota
Juan Del Toro, Assistant Professor, Department of Psychology, University of Minnesota
Dwayne Diggs, Support Technician, NIH
Ellis Dillon, Assistant Professor of Public Health Sciences, University of Connecticut on Aging
Breno Diniz, Associate Professor of Psychiatry, University of Connecticut Center on Aging
Nahed El Kassar, Physician, Division of Geriatrics and Clinical Gerontology, NIH
Jessica Faul, Research Associate Professor, Survey Research Center, Institute for Social Research, University of Michigan
Jason Fletcher, Vilas Distinguished Achievement Professor of Public Affairs, University of Wisconsin-Madison
Richard Fortinsky, Professor, Department of Medicine, University of Connecticut Center on Aging
David Furman, Associate Professor, Buck Institute
Lauren Gaydosh, Assistant Professor of Sociology, Center on Aging and Population Sciences, University of Texas at Austin
Daniel Holman, Lecturer in Sociology and Public Health, University of Sheffield
Iliana Kohler, Practice Associate Professor in Sociology, Associate Director of the Population Studies Center, University of Pennsylvania
Anastasia Leshchyk, PhD Candidate, Boston University
Gordon Lithgow, Vice President, Academic Affairs and Professor, Buck Institute
Yaro Markov, PhD Candidate, Yale University
Anna Marsland, Biological & Health Program Chair, Professor of Psychology, University of Pittsburgh
Renu Pillai, Senior Research Fellow, University of Washington
Jesse Poganik, Instructor, Brigham and Women’s Hospital, Harvard Medical School
Aric Prather, Professor, Psychiatry and Behavioral Sciences, University of California, San Francisco
Nalini Raghavachari, Program Officer, Clinical Gerontology Branch, Division of Geriatrics and Clinical Gerontology
Allison Ramiller, Product Manager, XPRIZE
Julie Robison, Professor, University of Connecticut Center on Aging
Roshan Sharafieh, Assistant Professor, Surgery, University of Connecticut Health
Noah Snyder-Mackler, Associate Professor, Life Sciences, Arizona State University
Lu Tian
Frances Tran, Prize Designer, XPRIZE
Indira Turney, Postdoctoral Fellow, Columbia University
Eva Wang
Yu Zhou, Assistant Professor, Buck Institute

Appendix 3: Links Shared in Meeting Chat

Peer-Reviewed Publications

• How deep single-cell and temporal phenotyping detected major biological differences between older men and women in response to the Prevnar vaccine: https://doi.org/10.1038/s41590-023-01717-5
• The long-term effects of prenatal cytokine exposure: https://doi.org/10.1073/pnas.2014464118
• How multigenerational factors influence later life outcomes: https://doi.org/10.1038/s41562-023-01796-2
• Prospective versus retrospective measures of early-life stress: https://doi.org/10.1001/jamapsychiatry.2019.0097
• How cellular restoration can combat the effects of stress: https://doi.org/10.1037/rev0000453
• How changes in aging clocks relate to changes in health outcomes: https://doi.org/10.1093/gerona/glad251
• Observing changes in DNA methylation clocks over time:
  • https://doi.org/10.1093/aje/kwx346
  • https://doi.org/10.1093/gerona/gly060
  • https://doi.org/10.7554%2FeLife.51507
  • https://doi.org/10.1038/s43587-022-00357-y
• The impacts of stress on hallmarks of aging: https://doi.org/10.1016/j.neubiorev.2023.105359
• Using exposomics to study the effects of environmental toxicants on aging: https://doi.org/10.1093/exposome/osac002
• Causal relationships of markers used in aging clocks: https://doi.org/10.1177/096228021875913

Ongoing Studies

• The Social, Behavioral, and Economic COVID Coordinating Center will examine some long-term impacts of COVID-19. List of current studies.
• Midlife in the United States ( MIDUS ): A National Longitudinal Study of Health and Well-Being as an example of a longitudinal study that includes acute stress reactivity laboratory paradigms.
• The Irish Longitudinal Study on Ageing ( TILDA ) as an example of a large study that measures orthostasis.
• The Legacy Effects of CALERIE study is examining the effects of caloric restriction 10 years post-intervention.

Other Resources

• Guidance on measuring early life adversity from the Stress Measurement Network
• Exceptional Longevity Translational Resources ( ELITE ) Portal