Abstract
Background
Multisystem dysregulation (Dm) shows promise as a metric of aging and predicts mortality. However, Dm needs to be studied with less severe endpoints indicating modifiable aging stages. Physical function, reflecting healthy longevity rather than just longevity, is more relevant to the goals of geroscience but has not been well investigated.
Methods
We tested the association of midlife Dm and its change over ~20 years with physical function in later life in 5 583 the Atherosclerosis Risk in Communities Study cohort participants (baseline mean age 54.7). Dm quantifies the multivariate statistical deviation of 17 physiologically motivated biomarkers relative to their distribution in a young healthy sample at baseline. Physical function was assessed from grip strength and the Short Physical Performance Battery (SPPB). Associations were quantified using linear regression and ordinal logistic regression adjusting for age, sex, race, and education.
Results
Each unit increment in midlife Dm was associated with 1.71 times the odds of having a lower SPPB score. Compared to the first quartile of midlife Dm, the odds ratios of having a lower SPPB score were 1.25, 1.56, and 2.45, respectively, for the second–fourth quartiles. Similar graded association patterns were observed for each SPPB component test and grip strength. An inverse monotonic relationship also was observed between the annual growth rate of Dm and physical function.
Conclusion
Greater Dm and progression in midlife were associated with lower physical function in later life. Future studies on the factors that lead to the progression of Dm may highlight opportunities to preserve physical function.
Keywords: Aging, Multisystem dysregulation, Physical function
While individuals aged 65 and older represented 16% of the population in 2019, the number of older adults will double by 2060 (1). In most societies, advancing age is associated with an exponential increase in burden from many aging-related chronic conditions (2). Similarly, the global burden of disease and disability rises as populations age (3). One of the goals of geroscience is to identify factors that may slow the process of aging and contribute to healthy longevity (4). Doing so requires ways to quantify the aging process with the understanding that the pace and extent of aging-related changes vary among individuals at any given chronological age (5).
Measures that characterize biological age separately from chronological age have been proposed, including molecular measures and multibiomarker measures (6). Among several candidate measures, multisystem dysregulation (Dm) is derived using a multibiomarker algorithm and quantifies how different a person’s physiologic profile is relative to a young healthy reference in the population, with higher values indicating more severe Dm (7). Compared to molecular measures of aging that are typically examined within a single tissue, such as telomere length and epigenetic clocks (6,8,9), Dm is in line with geroscience’s conceptualization of aging as a gradual and progressive deterioration of integrity across multiple systems and could be better suited to reflect the intersecting hallmarks of aging and characterize aging-related physiological and functional changes (2,6,7,10–12).
Although Dm has been validated across populations to predict mortality (7,13–16), a less severe aging-related outcome, such as physical function, implicating healthy longevity rather than just longevity, is more relevant to the goals that geroprotective therapies aim to extend but has not been well investigated (6). Physical function is necessary for independent living and to ensure the quality of life. Decline in physical function performance is an early sign of aging-related physical capacity decline, often preceding disease manifestations and is a potent predictor of morbidity and mortality (17).
Only a few studies have investigated physical function in relation to biological aging measures. Cross-sectional studies suggested that epigenetic age is modestly associated with grip strength in older populations (18,19), but not in young adults (6). No significant association was observed between longitudinal measures of epigenetic aging and physical function measures (6,18). To the best of our knowledge, only one study estimated Dm and reported a modest cross-sectional association with balance, grip strength, and motor coordination in a young adult cohort (6). However, these studies were either small (N < 1 000) (18,19), cross-sectional or had short follow-up time (<10 years) (6,18,19), lacked ethnic diversity (6,18,19), or included only young adults (6). None of the aforementioned studies characterized the progression of biological aging across decades of adult life span in association with a comprehensive battery of physical function.
In this context, we aimed to characterize Dm in midlife (ages 48–67 years) and its change over ~20 years of follow-up with physical function in later life (ages 71–90 years) among Black and White community-dwelling adults. We hypothesized that Dm would capture aging-related physiologic deterioration across multiple systems over the adult life span and be associated with lower physical function at an older age.
Method
Study Population
The Atherosclerosis Risk in Communities (ARIC) study is an ongoing prospective cohort enrolling 15 792 participants aged 45–64 years from 4 communities in the United States in 1987–1989: Forsyth County, NC; Jackson, MS; suburban Minneapolis, MN; and Washington County, MD (20,21). Follow-up visits were conducted in 1990–1992 (visit 2), 1993–1995 (visit 3), 1996–1998 (visit 4), 2011–2013 (visit 5 ARIC Neurocognitive Study [ARIC-NCS]), 2016–2017 (visit 6), and 2018–2019 (visit 7). In-person examinations include a series of detailed interview components, anthropometric and blood pressure measurements, and phlebotomy for laboratory assays. Of 6 538 ARIC participants who returned for the visit 5 examination, 5 583 participants who self-identified as Black or White (Black participants from Minneapolis and Washington County were excluded due to the small numbers) and completed the Short Physical Performance Battery (SPPB) and grip strength examinations were included in this study. Our baseline was the visit 2 examination because it had the most comprehensive set of Dm biomarker measurements. The Institutional Review Boards of all ARIC study sites approved the study protocol, and all cohort participants have given written informed consent at each examination.
Biological Age—Dm
The Dm method, described by Cohen et al. (7,13), measures how deviant an individual’s physiology is relative to a reference norm (7) and is interpreted as a metric of biological age. The calculation of Dm is as follows:
where xis a vector of 17 biomarker values for a given participant at a given visit. µ and S denote the mean and variance-covariance matrix for the same vector of biomarkers from the reference population. Original Dm is approximately log-normally distributed, and its scale depends on the number and scale of the biomarkers included. Following the instructions by Cohen et al. (7,13), the original Dm is standardized using log-transformation and dividing by the standard deviation (SD). We refer to the transformed Dm as Dm.
Reference population
A random sample (N = 1 000) of younger ARIC participants (ages 48–53 years) at visit 2, stratified on sex, race-center distribution, and free of coronary heart disease, stroke, heart failure, hypertension, diabetes, dementia, cancer, and chronic obstructive pulmonary disease within the first 3 years of follow-up served as the internal reference population representative of a physiological healthy state. The health profiles of the reference population were better than that of the ineligible population (Supplementary Table 1).
Vector of biomarkers for Dm
Dm was calculated from the following biomarkers: body mass index (BMI), waist girth, systolic blood pressure (SBP), forced expiratory volume in 1 second (FEV1), forced vital capacity ratio (FEV1/FVC), high density lipoprotein cholesterol (HDL), total cholesterol, triglycerides, high-sensitivity cardiac troponin-T (hs-cTnT), N-terminal probrain natriuretic peptide (NT-proBNP), estimated glomerular filtration rate (eGFR), glucose, lipoprotein-based insulin resistance index (TyG), QRS interval duration, QT interval duration, heart rate, and Cornell voltage. The measurements of the biomarkers are provided in Supplementary Methods, Section 1. Each measurement was standard normal transformed, that is, log-transformation if necessary for normality and then subtracted from the reference mean and divided by the reference SD. Biomarkers were selected to provide a diverse representation of multiple physiologic systems so that Dm could represent global, multisystemic physiologic dysregulation. Disease risk stratification based on Dm is relatively insensitive to the choice of biomarker when multiple, diverse biomarkers are included, although biomarkers with more aberrant values at older age provide a stronger signal (16).
As Dm is a composite measure containing information from 17 biomarkers, it can be affected by the missingness of any biomarker (Supplementary Table 2). Of 5 583 participants included in this study, 351 (6.3%) had missing Dm at visit 2, and 545 (9.8%) had missing Dm at visit 5. We performed multiple imputations by chained equations to address the missing Dm (Supplementary Methods, Section 1) (22). The trace plots depict a good model convergence (Supplementary Figure 1). To assess the robustness of the imputation model, we randomly selected 20% of observed Dm at both visits 2 and 5, converted these values to missing, and reran the imputation model for randomly selected observed values. The correlations between observed values and imputed values were 0.85 at visit 2 and 0.88 at visit 5 (Supplementary Figure 2). We used the imputed data set for our primary analyses.
We calculated an annual Dm growth rate by subtracting the baseline Dm from Dm measured at visit 5, then dividing by the years between baseline and visit 5. We also performed an alternative approach to derive the annual Dm growth rate: we first interpolated Dm biomarkers that were not administered at a specific visit, then estimated the annual Dm growth rate from the random component of mixed effects model regressing Dm of visits 1 to 5 over age (Supplementary Methods, Section 2). Annual Dm growth rate was grouped by quartiles (higher quartiles indicate faster Dm).
Physical Function
Physical function was examined at ARIC visit 5 using grip strength and SPPB, a summary measure of performance within 3 criteria: repeated chair stands, standing balance, and 4-m walk. Based on population-based norms, 0 (poorest) to 4 (best) were assigned to each component, yielding a composite score ranging from 0 to 12. We treated SPPB and each component score as ordinal outcomes.
Participants were timed up to 60 seconds (s) to repeat standing from a seated position, 5 times and as quickly as possible with arms folded across their chest. Participants were scored 0 point if unable to accomplish this task, 1 point if it took 16.7–<60 seconds, 2 points if 13.7–<16.7 seconds, 3 points if 11.2–<13.7 seconds, and 4 points if <11.2 seconds. For the standing balance test (23), the time (up to 10 seconds) that the participant could hold each foot position (side-by-side, semitandem, and tandem feet) was recorded. Beginning with a semitandem stand, if the participant was unable to hold the position for 10 seconds, participants were then tested on the side-by-side stand. Those who completed semitandem were assumed to be able to complete side-by-side and subsequently evaluated in the tandem stand. One point was assigned for completion of the side-by-side stand, another point for completing the semitandem stand, and 1 point for holding tandem position 3–<10 seconds. Two points were given for holding tandem position 10 seconds. Time to walk 4 m at the participant’s usual pace was measured. Participants were scored 0 point if unable to accomplish this task, 1 point if it took ≥8.70 seconds, 2 points if 6.21–<8.70 seconds, 3 points if 4.82–<6.21 seconds, and 4 points if <4.82 seconds. Gait speed (m/s) was calculated by dividing 4 m by the time required.
Grip strength in kilograms of force was measured using a Jamar Hydraulic Hand Dynamometer (Chicago, IL) in the participant’s dominant hand. Two trials were taken for each participant, and the better one was used for this analysis. If participants completed only 1 trial, the single trial result was used.
Other Variables
Date of birth, sex, self-reported race, and education were collected at visit 1 by interview. Educational attainment was categorized as (a) less than high school; (b) completed high school or vocational school; and (c) any college, graduate, or professional school. Prevalent and incident coronary heart disease, stroke, heart failure, cancer, chronic obstructive pulmonary disease, diabetes, and dementia were ascertained at in-person examination visits, through annual follow-up interviews, and from hospital records retrieved for all hospitalizations reported by cohort members.
Statistical Analysis
Sociodemographic characteristics, Dm biomarkers, and the medical health conditions at baseline were summarized over baseline Dm quartiles (higher quartiles indicate greater Dm).
The associations of baseline Dm and annual Dm growth rate with physical function were quantified using ordinal logistic regression (for SPPB and its component scores) and linear regression (for the time required for 5 repeated chair stands, gait speed, and grip strength), adjusting for age, sex, race-center, and education. The adjustment for race-center is due to the way the race groups are distributed across the centers. We interpreted the estimated association as a population-average effect independent of basic demographic factors. We initially modeled baseline Dm and annual Dm growth rate continuously. In order to see if there were any nonlinear associations between Dm and physical function, we alternatively modeled baseline Dm and annual Dm growth rate as quartiles, with the first quartile as the reference. Dm above 75th percentile was also deemed to indicate the presence of dysregulation, as suggested by previous studies (15).
Subsequently, we performed subgroup analyses stratified by sex, race, and the presence of aging-related morbidity prior to visit 5 (which would affect physical function, including coronary heart disease, stroke, heart failure, cancer, chronic obstructive pulmonary disease, diabetes, and dementia), and birth cohort (baseline age ≤50, 51–55, 56–60, and >60 years). We also performed sensitivity analyses as follows: (a) restrict the analytic sample to the complete case without any missingness (N = 4 718); (b) use the alternative annual Dm growth rate derived from mixed effects models; and (c) use stabilized inverse probability weighting to account for the cohort attrition due to death and nondeath drop-out. The probabilities of death and nondeath drop-out were estimated separately using logistic regression based on demographic variables, Dm biomarkers, comorbidities, self-reported poor health, number of hospitalizations, and interactions between variables. Dm was calculated using R (version 4.1.0) and RStudio (version 1.4.1717) software. Codes were created and provided by the original author (13). All analyses were performed with Stata version 16.0 (StataCorp LLC, College Station, TX), and a p-value < .05 was considered statistically significant.
Results
The mean age of 5 583 participants at baseline was 54.7 (SD 5.0) years, 3 223 (57.7%) participants were women, and 1 154 (20.7%) were Black. The median length of time between visits 2 and 5 was 20.8 (25th–75th percentile, 20.1–21.4) years. The distributions of Dm are shown in Figure 1, with the greater Dm levels at visit 5 as the population became older. As Dm is an aging-related metric, we grouped participants based on their baseline age in 5-year increments, yielding 4 birth cohorts. Compared to the younger cohort, the older cohort had an average greater baseline Dm as well as a greater increase in Dm from visit 2 to 5 (Supplementary Figure 3). In general, participants with greater Dm were more likely to be older, women, Black, and less educated; they were also likely to have worse Dm biomarker profiles, except that HDL levels were comparable across Dm quartiles (Table 1).
Figure 1.
Distribution of Dm at baseline (visit 2) and visit 5. Dm = multisystem dysregulation; SD = standard deviation.
Table 1.
Characteristics of Participants Across Dm Quartiles in 1990–1992: The Atherosclerosis Risk in Communities (ARIC) Study (N = 5 583)
| Midlife Dm | ||||
|---|---|---|---|---|
| Quartile 1 | Quartile 2 | Quartile 3 | Quartile 4 | |
| N | 1 354 | 1 442 | 1 399 | 1 388 |
| Age, years, mean (SD) | 53.7 (4.4) | 54.2 (4.7) | 55.1 (5.2) | 55.9 (5.4) |
| Female, N (%) | 762 (56.3) | 825 (57.2) | 803 (57.4) | 833 (60.0) |
| Race-center, N (%) | ||||
| NC/White | 368 (27.2) | 324 (22.5) | 244 (17.4) | 187 (13.5) |
| NC/Black | 8 (0.6) | 19 (1.3) | 28 (2.0) | 26 (1.9) |
| MI/Black | 53 (3.9) | 181 (12.6) | 330 (23.6) | 509 (36.7) |
| MN/White | 512 (37.8) | 509 (35.3) | 404 (28.9) | 307 (22.1) |
| MD/White | 413 (30.5) | 409 (28.4) | 393 (28.1) | 359 (25.9) |
| Education, N (%) | ||||
| <High school | 126 (9.3) | 136 (9.4) | 208 (14.9) | 256 (18.4) |
| High school or vocational school | 626 (46.2) | 605 (42.0) | 572 (40.9) | 559 (40.3) |
| Any college, graduate, or professional school | 602 (44.5) | 701 (48.6) | 619 (44.2) | 573 (41.3) |
| BMI, kg/m2, mean (SD) | 25.9 (3.2) | 26.6 (4.0) | 27.7 (4.8) | 29.5 (6.0) |
| Waist girth, cm, mean (SD) | 91.7 (10.2) | 93.4 (12.1) | 96.4 (13.4) | 101.0 (15.2) |
| SBP, mmHg, mean (SD) | 109.5 (9.6) | 113.1 (12.4) | 118.6 (15.1) | 127.7 (19.8) |
| FEV1, liters, mean (SD) | 3.1 (0.6) | 3.1 (0.7) | 3.0 (0.7) | 2.9 (0.7) |
| FEV1/FVC, %, mean (SD) | 79.4 (2.2) | 78.4 (3.5) | 77.1 (4.3) | 75.7 (4.7) |
| HDL, mg/dL, mean (SD) | 50.3 (13.4) | 51.6 (16.4) | 51.7 (17.8) | 49.8 (18.7) |
| Total cholesterol, mg/dL, mean (SD) | 200.8 (27.6) | 204.9 (34.0) | 208.7 (39.4) | 212.8 (44.5) |
| Triglycerides, mg/dL, mean (SD) | 111.5 (46.7) | 119.7 (62.3) | 130.6 (79.5) | 147.8 (113.6) |
| hs-cTnT, ng/L, mean (SD) | 2.5 (1.8) | 3.0 (2.4) | 3.7 (3.2) | 4.9 (6.6) |
| NT-proBNP, pg/mL, mean (SD) | 54.2 (39.4) | 58.2 (53.2) | 66.0 (67.8) | 74.2 (110.2) |
| eGFR, mL/min/1.73 m2, mean (SD) | 99.5 (8.6) | 98.5 (12.0) | 97.1 (14.9) | 97.0 (17.6) |
| Glucose, mg/dL, mean (SD) | 98.5 (6.4) | 99.7 (8.0) | 101.6 (9.4) | 125.1 (48.9) |
| TyG, mean (SD) | 11032.9 (4802.5) | 12013.7 (6503.4) | 13376.2 (8495.0) | 19264.9 (20656.4) |
| QRS interval duration, ms, mean (SD) | 90.6 (8.3) | 91.3 (9.7) | 91.1 (10.3) | 94.2 (15.3) |
| QT interval duration, ms, mean (SD) | 410.9 (19.2) | 412.6 (24.3) | 414.0 (28.2) | 415.4 (33.0) |
| Heart rate, b/min, mean (SD) | 63.7 (7.1) | 63.9 (8.7) | 64.4 (9.7) | 66.6 (11.3) |
| Cornell voltage, mm, mean (SD) | 1036.1 (363.5) | 1111.1 (442.2) | 1222.3 (514.8) | 1430.8 (620.5) |
Notes: Dm = multisystem dysregulation; SD = standard deviation; BMI = body mass index; SBP = systolic blood pressure; FEV1 = forced expiratory volume in 1 second; FVC = forced vital capacity; HDL = high-density lipoprotein cholesterols; TyG = lipoprotein-based insulin resistance index; hs-cTnT = high-sensitivity cardiac troponin-T; NT-proBNP = N-terminal probrain natriuretic peptide; eGFR = estimated glomerular filtration rate.
We observed a monotonic inverse association of baseline Dm with SPPB score using ordinal logistic regression adjusting for age, sex, race-center, and education (Table 2). Each unit increment in baseline Dm was associated with 1.71 (95% confidence interval [CI]: 1.58, 1.85) times the odds of having a lower SPPB score. We corroborated this consistently higher likelihood of having a lower SPPB score when modeling baseline Dm categorically. Compared to the first quartile of Dm, the odds ratios for a lower SPPB score were 1.25 (1.09, 1.44) for the second quartile, 1.56 (1.35, 1.80) for the third quartile, and 2.45 (2.08, 2.88) for the fourth quartile. Furthermore, a greater Dm growth rate from midlife to older adulthood also was associated with lower SPPB scores in older adulthood, albeit this association was of a smaller magnitude. Regarding the components of the SPPB, the associations were slightly weaker for chair stands and balance tests and stronger for the 4-m walk test. We confirmed this pattern using continuous physical function measures (Table 2). For example, each unit increment in baseline Dm was associated with 0.18 (0.15, 0.22) SD greater time to complete 5 repeated chair stands; had a slower gait speed (−0.27 SD [−0.31, −0.23]); and had a lower grip strength (−0.12 SD [−0.15, −0.09]). This was also true for people with a greater annual Dm growth rate.
Table 2.
Associations of Dm in 1990–1992 and Annual Dm Growth Rate With Physical Function Performance in 2011–2013: The Atherosclerosis Risk in Communities (ARIC) Study (N = 5 583)
| Midlife Dm* | |||||
|---|---|---|---|---|---|
| Physical Function Measures | Quartile 1 | Quartile 2 | Quartile 3 | Quartile 4 | Per 1 Unit Increment |
| SPPB†,‡ | 1 (Ref) | 1.25 (1.09, 1.44) | 1.56 (1.35, 1.80) | 2.45 (2.08, 2.88) | 1.71 (1.58, 1.85) |
| SPPB—Chair stands†,‡ | 1 (Ref) | 1.24 (1.08, 1.42) | 1.40 (1.21, 1.62) | 1.97 (1.67, 2.32) | 1.51 (1.39, 1.63) |
| SPPB—Balance†,‡ | 1 (Ref) | 1.10 (0.91, 1.33) | 1.48 (1.22, 1.78) | 2.10 (1.72, 2.56) | 1.55 (1.41, 1.70) |
| SPPB—4 meters walk†,‡ | 1 (Ref) | 1.35 (1.12, 1.63) | 1.81 (1.49, 2.19) | 2.75 (2.24, 3.36) | 1.69 (1.54, 1.85) |
| 1 SD—Time required for 5 chair stands (s)§ | 0 (Ref) | 0.08 (0.02, 0.15) | 0.16 (0.09, 0.23) | 0.32 (0.24, 0.40) | 0.18 (0.15, 0.22) |
| 1 SD—Gait speed (m/s)§ | 0 (Ref) | −0.12 (−0.19, −0.06) | −0.27 (−0.34, −0.20) | −0.48 (−0.56, −0.41) | −0.27 (−0.31, −0.23) |
| 1 SD—Grip strength (kg)§ | 0 (Ref) | −0.05 (−0.10, 0.01) | −0.13 (−0.18, −0.07) | −0.21 (−0.27, −0.15) | −0.12 (−0.15, −0.09) |
| Annual Dm* Growth Rate | |||||
| SPPB†,‡ | 1 (Ref) | 1.16 (1.01, 1.35) | 1.29 (1.11, 1.50) | 1.71 (1.46, 1.99) | 1.29 (1.22, 1.37) |
| SPPB—Chair stands†,‡ | 1 (Ref) | 1.06 (0.91, 1.22) | 1.24 (1.07, 1.45) | 1.56 (1.34, 1.82) | 1.26 (1.19, 1.34) |
| SPPB—Balance†,‡ | 1 (Ref) | 1.09 (0.91, 1.32) | 1.09 (0.90, 1.32) | 1.28 (1.05, 1.56) | 1.12 (1.04, 1.21) |
| SPPB—4 meters walk†,‡ | 1 (Ref) | 1.13 (0.94, 1.36) | 1.28 (1.05, 1.55) | 1.72 (1.42, 2.09) | 1.29 (1.20, 1.38) |
| 1 SD—Time required for 5 chair stands (s)§ | 0 (Ref) | 0.00 (−0.07, 0.07) | 0.04 (−0.03, 0.12) | 0.13 (0.06, 0.21) | 0.08 (0.05, 0.11) |
| 1 SD—Gait speed (m/s)§ | 0 (Ref) | −0.06 (−0.12, 0.01) | −0.13 (−0.21, −0.06) | −0.27 (−0.34, −0.19) | −0.13 (−0.16, −0.10) |
| 1 SD—Grip strength (kg)§ | 0 (Ref) | −0.02 (−0.08, 0.04) | −0.09 (−0.15, −0.03) | −0.10 (−0.16, −0.05) | −0.05 (−0.07, −0.03) |
Notes: SPPB = short physical performance battery; Dm = multisystem dysregulation; SD = standard deviation.
*Dm was calculated from body mass index, waist girth, forced expiratory volume in 1 second, forced vital capacity ratio, estimated glomerular filtration rate, systolic blood pressure, N-Terminal ProBrain Natriuretic Peptide, high-sensitivity cardiac troponin-T, high density lipoprotein cholesterol, total cholesterol, triglycerides, glucose, lipoprotein-based insulin resistance index-TyG, heart rate, QRS interval duration, QT interval duration, and Cornell voltage.
†SPPB is an ordinal outcome ranging from 0 to 12, with each component (chair stands, balance, and 4-m walk) ranging from 0 to 4. A lower level indicates a poorer physical functional performance.
‡Odds ratios (95% confidence intervals) of lower physical function score from ordinal logistic regression including visit 2 Dm, annual Dm growth rate, age, sex, race-center, and education.
§Linear regression model including visit 2 Dm, annual Dm growth rate, age, sex, race-center, and education.
Although the magnitude of the odds ratios varied, the monotonic inverse associations between midlife Dm and Dm growth rate with physical function were largely consistent across subgroups of sex, race, the presence of aging-related morbidity, and birth cohort (Supplementary Tables 3–6). Finally, our results remained robust when performing a complete case analysis (Supplementary Table 7), or using an alternative approach to derive annual Dm growth rate (Supplementary Table 8), or after accounting for cohort attrition and selection bias (Supplementary Table 9).
Discussion
In this community-based cohort, a multibiomarker, algorithmic estimate of biological age (Dm) at midlife characterizing aging-related Dm showed monotonic inverse associations with physical function in later life. Similar graded association patterns were observed for each SPPB component test and grip strength, with the 4-m walk test showing a greater magnitude of association than the others. The progression of Dm over 20 years from midlife to older adulthood also was inversely associated with physical function in older adulthood. These associations were largely consistent across sex, race, the presence of aging-related morbidity, and birth cohort.
The association between the biological age indicator Dm and physical function could be explained by the accumulation of aging-related deficits and loss of resiliency and function in multiple physiologic systems across organs, inclusive of skeletal muscle (19). As a metric of age, Dm identifies individuals who age more rapidly than others and are therefore more susceptible to the adverse effects of aging, such as loss of muscle mass, function, and strength due to fat accumulation and excessive inflammation, which further leads to walking limitations, loss of function, and an increased risk for disability and morbidity (24).
To the best of our knowledge, only one previous study has related Dm to physical function. This study was conducted in a 1-year birth cohort and reported a modest cross-sectional association between Dm and physical function measured at age 38 years (6). The study included primarily White individuals. Because its physical function measures were taken at age 38 years, it may also lack the variability of changes in physical function associated with aging. Our study extends the limited evidence on the association of Dm with physical function by providing an extensive follow-up and multidimensional assessment of physical function among older adults.
The observed robust monotonic associations of Dm with the SPPB composite score, its individual components, and grip strength suggest a general lower physical function in relation to biological aging. Physical function was measured in the ARIC cohort using a comprehensive physical performance battery that consists of chair stands, standing balance, and 4-m walk tests, and each of them reflects a specific aspect of physical function. While the chair stands test mainly reflects leg strength and balance, the walk test involves the coordination of multiple organ systems, including muscles, nerves, vision, joints, heart, and lungs (25). The observed magnitude of the association was the strongest for the 4-m walk test compared to the other components, suggesting the latter as a potential marker of the Dm.
We also observed an inverse association between the Dm growth rate over the 20-year course from midlife to older adulthood with physical function, independent of midlife Dm levels. Our results are in line with previous studies describing an acceleration in multisystemic dysregulation with chronological age (26). Baseline Dm primarily can be viewed as the accumulation of Dm until the time of measurement. Part of the measured change in Dm correlates with baseline Dm, while the other estimates deterioration in physical function among those who experience a more rapid aging process than the group norm. This has been interpreted to mean that Dm captures the progressive breakdown of homeostasis and loss of the capacity to respond to stressors (26).
The term biological age has been applied to the observations that individuals differ in manifest age as they age chronologically. To date, translation of the insights into the aging processes gained from model organisms to humans has been made difficult by the long duration of the human life span and the diversity of the exposures proposed to influence human aging. For a better understanding of aging trajectories and of eventual intervention and prevention efforts, quantification of biological age and its progression focused on earlier life epochs is needed. Age acceleration quantified using epigenetic age was not found to be associated with knee extension strength or walking speed in a previous study (19). In that study, as in most studies investigating epigenetic age, peripheral blood leukocytes were the tissue being assayed with uncertain correlation with methylation patterns in skeletal muscle tissue (8,19). In contrast, Dm reflects the systemic nature of homeostatic dysregulation of aging and measures the degree of deviation from the group norm in the overall physiologic condition of an individual at a given point.
Instead of a stochastic process towards dysfunction and ultimately death, geroscience characterizes aging as an adaptive process that can be amenable to modification (27,28), and seeks to identify deterministic mechanisms that modulate aging and interindividual variability in the pace of aging (29). A focus on middle or younger adulthood would be more effective than reversing aging and its effects in later life if the goal is to extend healthy longevity and preserve physical function. Dm is estimated using an algorithm that reduces a high-dimensional biomarker space into a single measure and provides a way to quantify and track aging-related homeostasis disturbances before disease onsets. Dm levels increased with the number of biomarkers included but were less sensitive to biomarker choice in this study (data not shown) and in previous studies (16), as long as these biomarkers were sufficiently diverse to represent different physiologic systems and were less correlated with each other. Also, unlike frailty indices used to quantify deficit accumulation in older adults, Dm has a great potential to quantify the gradual and often unnoticed Dm that accumulates over decades and might not become more clinically evident until later in life. The strong association between midlife Dm and its progression with physical function corroborates the potential use of this metric of biological age during the adult life span as a tool in aging research and as a target for geroprotective interventions.
Limitations of this study are acknowledged. The effect of medication use on Dm biomarker values could not be considered, introducing a likely small but unmeasured degree of misclassification. Dm in our study was measured at 2-time points, 20 years apart. A sensitivity analysis based on Dm values imputed at all 5 cohort visits nonetheless yielded similar results. We also note that at baseline, the ARIC study consisted of Black and White men and women aged 48–67 years of age, sampled from the residents of 4 U.S. communities; generalization of the results of this study to other populations and settings warrants caution.
In conclusion, greater Dm at midlife, and a faster growth in Dm over the adult life span, were associated with lower physical function in later life. Future studies on the factors that lead to the onset and progression of Dm during midlife may identify opportunities to preserve functional abilities in later life.
Supplementary Material
Acknowledgments
The authors thank the staff and participants of the Atherosclerosis Risk in Communities study for their important contributions. The authors acknowledge the contributions of G.H., Kenan Distinguished Professor of Epidemiology at the University of North Carolina at Chapel Hill Gillings School of Global Public Health. G.H. was central to the successful development and execution of this manuscript. He passed away on June 11, 2022, during the completion of this manuscript.
Contributor Information
Yifei Lu, Department of Epidemiology, University of North Carolina at Chapel Hill, Chapel Hill, North Carolina, USA.
James R Pike, Department of Biostatistics, University of North Carolina at Chapel Hill, Chapel Hill, North Carolina, USA.
Anna M Kucharska-Newton, Department of Epidemiology, College of Public Health, University of Kentucky, Lexington, Kentucky, USA.
Priya Palta, Departments of Medicine and Epidemiology, Columbia University Irving Medical Center, New York, New York, USA; Department of Neurology, University of North Carolina at Chapel Hill, Chapel Hill, North Carolina, USA.
Eric A Whitsel, Department of Epidemiology, University of North Carolina at Chapel Hill, Chapel Hill, North Carolina, USA; Department of Medicine, University of North Carolina at Chapel Hill, Chapel Hill, North Carolina, USA.
Ganga S Bey, Department of Epidemiology, University of North Carolina at Chapel Hill, Chapel Hill, North Carolina, USA.
Anthony S Zannas, Department of Psychiatry, University of North Carolina at Chapel Hill, Chapel Hill, North Carolina, USA; Department of Genetics, University of North Carolina at Chapel Hill, Chapel Hill, North Carolina, USA.
B Gwen Windham, School of Medicine, University of Mississippi Medical Center, Jackson, Mississippi, USA.
Keenan A Walker, Laboratory of Behavioral Neuroscience, National Institute on Aging, Baltimore, Maryland, USA.
Michael Griswold, School of Medicine, University of Mississippi Medical Center, Jackson, Mississippi, USA.
Gerardo Heiss, Department of Epidemiology, University of North Carolina at Chapel Hill, Chapel Hill, North Carolina, USA.
Funding
The Atherosclerosis Risk in Communities study has been funded in whole or in part with Federal funds from the National Heart, Lung, and Blood Institute, National Institutes of Health, Department of Health and Human Services, under Contract nos. (75N92022D00001, 75N92022D00002, 75N92022D00003, 75N92022D00004, and 75N92022D00005).
Conflict of Interest
None declared.
Author Contributions
Conception and study design: Y.L. and G.H.; data access and analysis: Y.L. and J.R.P.; preparation of manuscript: Y.L. and G.H.; interpretation of data: all authors; critical revision of work for important intellectual content: all authors.
References
- 1. World Health Organization. Accessed March 22, 2022.https://www.cdc.gov/chronicdisease/resources/publications/factsheets/promoting-health-for-older-adults.htm#:~:text=In%202019%2C%2054.1%20million%20US,expected%20to%20reach%2080.8%20million
- 2. Belsky DW, Caspi A, Houts R, et al. Quantification of biological aging in young adults. Proc Natl Acad Sci USA. 2015;112(30):E4104. doi: 10.1073/pnas.1506264112 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 3. Vos T, Flaxman AD, Naghavi M, et al. Years lived with disability (YLDs) for 1160 sequelae of 289 diseases and injuries 1990–2010: a systematic analysis for the Global Burden of Disease Study 2010. Lancet. 2012;380(9859):2163–2196. doi: 10.1016/S0140-6736(12)61729-2 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 4. Moskalev A, Chernyagina E, Tsvetkov V, et al. Developing criteria for evaluation of geroprotectors as a key stage toward translation to the clinic. Aging Cell. 2016;15(3):407–415. doi: 10.1111/acel.12463 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 5. Mitnitski AB, Mogilner AJ, Rockwood K. Accumulation of deficits as a proxy measure of aging. ScientificWorldJournal. 2001;1:323–336. doi: 10.1100/tsw.2001.58 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 6. Belsky DW, Moffitt TE, Cohen AA, et al. Eleven telomere, epigenetic clock, and Biomarker-Composite quantifications of biological aging: do they measure the same thing? Am J Epidemiol. 2018;187(6):1220–1230. doi: 10.1093/aje/kwx346 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 7. Cohen AA, Li Q, Milot E, et al. Statistical distance as a measure of physiological dysregulation is largely robust to variation in its biomarker composition. PLoS One. 2015;10(4):e0122541. doi: 10.1371/journal.pone.0122541 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 8. Horvath S. DNA methylation age of human tissues and cell types. Genome Biol. 2013;14(10):R115. doi: 10.1186/gb-2013-14-10-r115 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 9. Hannum G, Guinney J, Zhao L, et al. Genome-wide methylation profiles reveal quantitative views of human aging rates. Molecular Cell. 2013;49(2):359–367. doi: 10.1016/j.molcel.2012.10.016 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 10. Kirkwood TB. Understanding the odd science of aging. Cell. 2005;120(4):437–447. doi: 10.1016/j.cell.2005.01.027 [DOI] [PubMed] [Google Scholar]
- 11. Klemera P, Doubal S. A new approach to the concept and computation of biological age. Mech Ageing Dev. 2006;127(3):240–248. doi: 10.1016/j.mad.2005.10.004 [DOI] [PubMed] [Google Scholar]
- 12. Gruenewald TL, Seeman TE, Ryff CD, Karlamangla AS, Singer BH. Combinations of biomarkers predictive of later life mortality. Proc Natl Acad Sci USA. 2006;103(38):14158–14163. doi: 10.1073/pnas.0606215103 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 13. Cohen AA, Milot E, Yong J, Seplaki CL, Fülöp T, Bandeen-Roche K, Fried LP. A novel statistical approach shows evidence for multi-system physiological dysregulation during aging. Mech Ageing Dev. 2013;134(3–4):110–117. doi: 10.1016/j.mad.2013.01.004 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 14. Li Q, Wang S, Milot E, et al. Homeostatic dysregulation proceeds in parallel in multiple physiological systems. Aging Cell. 2015;14(6):1103–1112. doi: 10.1111/acel.12402 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 15. Ghachem A, Fried LP, Legault V, et al. Evidence from two cohorts for the frailty syndrome as an emergent state of parallel dysregulation in multiple physiological systems. Biogerontology. 2021;22(1):63–79. doi: 10.1007/s10522-020-09903-w [DOI] [PMC free article] [PubMed] [Google Scholar]
- 16. Cohen AA, Milot E, Li Q, Legault V, Fried LP, Ferrucci L. Cross-population validation of statistical distance as a measure of physiological dysregulation during aging. Exp Gerontol. 2014;57:203–210. doi: 10.1016/j.exger.2014.04.016 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 17. Guralnik JM, Ferrucci L, Pieper CF, et al. Lower extremity function and subsequent disability: consistency across studies, predictive models, and value of gait speed alone compared with the short physical performance battery. J Gerontol A Biol Sci Med Sci. 2000;55(4):M221–M231. doi: 10.1093/gerona/55.4.M221 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 18. Marioni RE, Shah S, McRae AF, et al. The epigenetic clock is correlated with physical and cognitive fitness in the Lothian Birth Cohort 1936. Int J Epidemiol. 2015;44(4):1388–1396. doi: 10.1093/ije/dyu277 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 19. Sillanpää E, Laakkonen EK, Vaara E, et al. Biological clocks and physical functioning in monozygotic female twins. BMC Geriatr. 2018;18(1):83. doi: 10.1186/s12877-018-0775-6 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 20. The Atherosclerosis Risk in Communities (ARIC) Study: Design and Objectives. The ARIC investigators. Am J Epidemiol. 1989;129(4):687–702. doi: 10.1093/oxfordjournals.aje.a115184 [DOI] [PubMed] [Google Scholar]
- 21. Wright JD, Folsom AR, Coresh J, et al. The ARIC (Atherosclerosis Risk in Communities) Study: JACC Focus Seminar 3/8. J Am Coll Cardiol. 2021;77(23):2939–2959. doi: 10.1016/j.jacc.2021.04.035 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 22. van Buuren S. Multiple imputation of discrete and continuous data by fully conditional specification. Stat Methods Med Res. 2007;16(3):219–242. doi: 10.1177/0962280206074463 [DOI] [PubMed] [Google Scholar]
- 23. Guralnik JM, Simonsick EM, Ferrucci L, et al. A short physical performance battery assessing lower extremity function: association with self-reported disability and prediction of mortality and nursing home admission. J Gerontol. 1994;49(2):M85–M94. doi: 10.1093/geronj/49.2.m85 [DOI] [PubMed] [Google Scholar]
- 24. Bellettiere J, Nguyen S, Eaton CB, et al. The short physical performance battery and incident heart failure among older women: the OPACH study. Am J Prev Cardiol. 2021;8:100247. doi: 10.1016/j.ajpc.2021.100247 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 25. Matsushita K, Ballew SH, Sang Y, et al. Ankle-brachial index and physical function in older individuals: the Atherosclerosis Risk in Communities (ARIC) study. Atherosclerosis. 2017;257:208–215. doi: 10.1016/j.atherosclerosis.2016.11.023 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 26. Milot E, Morissette-Thomas V, Li Q, Fried LP, Ferrucci L, Cohen AA. Trajectories of physiological dysregulation predicts mortality and health outcomes in a consistent manner across three populations. Mech Ageing Dev. 2014;141–142:56–63. doi: 10.1016/j.mad.2014.10.001 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 27. Goldman DP, Cutler D, Rowe JW, et al. Substantial health and economic returns from delayed aging may warrant a new focus for medical research. Health Aff (Millwood). 2013;32(10):1698–1705. doi: 10.1377/hlthaff.2013.0052 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 28. Burch JB, Augustine AD, Frieden LA, et al. Advances in geroscience: impact on healthspan and chronic disease. J Gerontol A Biol Sci Med Sci. 2014;69(suppl 1):S1–S3. doi: 10.1093/gerona/glu041 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 29. Gonzalez-Freire M, Diaz-Ruiz A, Hauser D, et al. The road ahead for health and lifespan interventions. Ageing Res Rev. 2020;59:101037. doi: 10.1016/j.arr.2020.101037 [DOI] [PMC free article] [PubMed] [Google Scholar]
Associated Data
This section collects any data citations, data availability statements, or supplementary materials included in this article.