Abstract
Background:
Physical functioning (PF) tests are scalable screening tools for neurodegenerative risk that differ by race and ethnicity, reflecting upstream social inequities. Prior work shows that PF relates to Alzheimer’s disease blood biomarkers differentially by race and ethnicity, yet it is unclear if similar associations exist with structural brain outcomes (hyperintensities and brain volumes).
Objective:
This study evaluated associations between PF and white matter hyperintensity volume (WMH), total brain volume (TBV), and hippocampal volume (HV) and assessed heterogeneity by race and ethnicity.
Methods:
Cognitively normal non-Hispanic Black (n = 420), non-Hispanic white (n = 917), and Hispanic (n = 828) older adults (mean [SD] age = 64.7 [8.4]) from the Healthy Aging Brain Study-Health Disparities cohort completed the Timed Up and Go (TUG) and Short Physical Performance Battery (SPPB). Brain imaging was collected via 3 T magnetic resonance imaging (MRI). Multivariable linear models related PF to brain imaging outcomes (normalized by intracranial volume) and assessed heterogeneity by race and ethnicity.
Results:
Bonferroni adjusted significant (p < 0.05) associations were observed between WMH and PF (TUG: β = 0.28, 95% CI = 0.17, 0.40; SPPB:β = −0.19, 95% CI = −0.32, −0.07), TBV and PF (TUG: β = −0.50, 95% CI = −0.61, −0.40; SPPB: β = 0.30, 95% CI = 0.18, 0.42), and HV and PF (TUG: β = −0.28, 95% CI = −0.38, −0.18). Associations were generally similar across race and ethnicity groups for TBV and HV.
Conclusions:
Poorer PF was associated with reduced TBV and HV and greater WMH. Results support PF tests as scalable, non-invasive indicators of brain health in community-based populations.
Keywords: Alzheimer’s disease, hippocampus, magnetic resonance imaging, physical functional performance, white matter
Introduction
Brain and cognitive health in aging populations has emerged as a fundamental challenge for advancing precision medicine and improving long-term health outcomes across representative communities.1 In the United States, the number of older adults is projected to double by 2050 to make up nearly 20% of the population.2 By that time, the global burden of neurodegenerative diseases, such as Alzheimer’s disease (AD), Parkinson’s disease, and related dementias, is expected to reach nearly 153 million cases from an estimated 57.4 million cases in 2019.3 Accordingly, there is a growing need for early, accessible, and cost-effective tools to identify individuals at risk for brain health deterioration, particularly prior to clinical symptom onset and cognitive decline.4–9 Emerging amyloid-reducing therapies have been shown to slow clinical decline when initiated in the early stages of AD, prior to widespread neurodegeneration and cognitive impairment. Identifying at-risk individuals before substantial pathology accumulates is essential to maximize therapeutic benefit and preserve functional independence.10,11
Physical functioning has emerged as a promising, non-invasive indicator of brain health. Performance-based measures such as the Short Physical Performance Battery (SPPB)12 and Timed Up and Go (TUG)13 assess domains including gait speed, balance, and lower extremity strength. Prior research has linked poorer performance on these measures to worse memory, executive function, and processing speed,14 as well as increased risk for cognitive impairment and dementia.15–17 In our prior work, physical functioning was associated with blood-based biomarkers for AD, including Aβ40, Aβ42, total tau, and neurofilament light chain, in a representative community-based sample.18,19 Notably, we found that these associations differed by race and ethnicity, suggesting that functional outcomes related to AD progression may be shaped by environmental, behavioral, and systemic healthcare factors. Such insights can inform more equitable approaches to primary and tertiary prevention of neurodegenerative conditions.
Beyond peripheral biomarkers, physical functioning has also been linked to structural changes in the brain. For instance, higher white matter hyperintensity burden has been associated with impaired mobility and abnormal TUG performance, even in post-stroke and hospital-based samples.20 These associations are thought to reflect disruptions in motor control and executive processing networks. Likewise, structural MRI has consistently identified reduced hippocampal and total brain volume in aging and AD, with strong links to cognitive decline.21–34 Although neuroimaging provides a robust means to assess brain integrity, its complexity, high cost, and infrastructure requirements limit feasibility for large-scale implementation. This has increased interest in more accessible, scalable alternatives, such as physical functioning assessments, that can serve as pragmatic indicators of brain health across diverse populations.
Despite these promising associations, most studies to date relating physical functioning to structural brain scans have relied on small, clinical, or demographically homogenous samples.20,35–39 Few investigations have leveraged large, community-based cohorts that include standardized physical functioning measures, structural neuroimaging, and racial and ethnic diversity sufficient to support subgroup analyses. These design limitations hinder the generalizability of findings and prevent deeper understanding of how structural brain changes relate to physical performance across sociodemographic contexts. Understanding whether associations between physical function and neuroimaging biomarkers differ across racially and ethnically diverse groups is essential for refining early detection strategies and informing tailored interventions.
Accordingly, the purpose of this study was to evaluate the association between structural brain imaging markers, including white matter hyperintensity volume, total brain volume, and hippocampal volume, and physical functioning, as assessed by TUG and SPPB, in a large, community-based sample of cognitively normal older adults. We also examined whether these associations varied across racial and ethnic groups (i.e., heterogeneity of associations). In this context, race and ethnicity are conceptualized as social constructs that reflect differential exposure to structural, environmental, and healthcare factors shaping brain and physical health. This framing allows examination of how contextual influences on aging may manifest in brain–behavior relationships across diverse populations. This work aims to enhance our understanding of how structural brain abnormalities relate to motor function and to determine whether these relationships are consistent across representative populations. Ultimately, these findings may inform risk stratification strategies and support the development of scalable, equitable tools for early identification of brain health deterioration.
Methods
Participants and assessment
Participants in this cross-sectional analysis of clinical examination, interview, and functional exam data were cognitively normal adults enrolled in the Healthy Aging Brain Study-Health Disparities (HABS-HD) study. The HABS-HD study is an ongoing prospective cohort focused on examining health disparities associated with aging, mild cognitive impairment (MCI) and AD among three prominent racial and ethnic groups in the United States (i.e., Hispanic, non-Hispanic Black, non-Hispanic White). A community-based participatory research frame-work was employed to enroll participants through a rolling recruitment process. Eligible participants were those aged at least 30-years at enrollment, who self-identified as Hispanic, non-Hispanic Black, or non-Hispanic White, consented to provide blood samples, were able to undergo neuroimaging procedures, and were able to read and write in English or Spanish. Exclusion criteria at enrollment included individuals with type 1 diabetes, an ongoing infection, a recent or current cancer diagnosis (except non-melanoma skin cancers), a severe mental health condition that might affect cognitive performance (excluding depression), a traumatic brain injury with loss of consciousness within the past 12-months, current or recent alcohol or substance dependency, or a significant medical condition that could influence cognition (e.g., advanced renal failure, severe heart failure, or chronic obstructive pulmonary disease). While HABS-HD collects multimodal biomarker data, including PET and blood-based measures, the present study utilized only functional assessments, neuropsychological evaluations, and MRI data. All study components are administered in either English or Spanish to accommodate participant language preferences.
The present analysis excluded participants classified with MCI or dementia based on a comprehensive neuropsychological evaluation and expert consensus review. The neuropsychological evaluation incorporated measures of cognitive ability, executive function, memory, language, and premorbid intelligence. Informant-reported Clinical Dementia Rating (CDR) assessments were used to corroborate cognitive status and functional abilities. An independent expert panel reviewed the neuropsychological outcomes, CDR ratings, and participants’ medical histories to reach a consensus in categorizing individuals as cognitively normal, or having MCI or dementia (see Petersen et al. (2025) for greater detail on the methods for determining cognitive status).40
Physical functioning
Physical functioning was the primary dependent variable of interest in the present study. Physical functioning was measured during the functional examination with the TUG and SPPB tests. The TUG evaluates functional mobility by timing participants as they rise from a standard armchair, walk 3 meters to a designated mark, turn, return to the chair, and sit down. Participants are directed to complete the task “at a normal comfortable pace” per standard protocol. The timing commences when the participant lifts off the seat and concludes upon contact with the chair upon sitting. This measure has demonstrated reliability in distinguishing cognitive and functional limitations and serves as a valid indicator of gait speed, balance, and activities of daily living.13 Recent research also indicates its association with AD plasma biomarkers among HABS-HD samples.18,19
The SPPB was developed by the National Institutes of Health’s National Institute on Aging (NIA) to assess balance, lower extremity strength, and overall functional capacity in older adults. This test comprises three distinct components: a walking task, a sit-to-stand task, and a balance task, each scored individually and combined for a total functional score (ranging from 0 to 12). The walking task involves a 3- or 4-meter walk at a “comfortable pace” to gauge typical gait speed. The sit-to-stand task requires participants to transition from sitting to standing without assistance; if successful, participants are asked to repeat this movement five times without using their arms. The balance task involves standing unassisted in three positions—feet together, semi-tandem, and full tandem—for up to 10 s each, with performance times recorded. Points are allocated based on completion times, with higher scores reflecting superior performance across the walking, sit-to-stand, and balance tasks (e.g., faster times or longer balance durations yield more points). The SPPB is recognized as a dependable and accurate tool for evaluating functional health and has proven effective as a screening method for detecting associated health conditions.12
Magnetic resonance imaging
Imaging was performed using a 3 T Siemens Magnetom Skyra (T1-weighted MPRAGE: TR = 2300 ms, TE = 2.93 ms, voxel size = 1.1 × 1.1 × 1.2 mm; FLAIR: TR = 4800 ms, TE = 441 ms, voxel size = 1.0 × 1.0 × 1.2 mm; FOV = 256 × 240 mm) or 3 T Siemens Magnetom Vida (T1-weighted MPRAGE: TR = 2300 ms, TE = 2.98 ms, flip angle = 9°, voxel size = 1 × 1 × 1 mm; FLAIR: TR = 4800 ms, TE = 441 ms, voxel size = 1.0 × 1.0 × 1.2 mm; FOV = 256 × 240 mm) whole-body scanner using the ADNI3 MRI protocol. The scanner model was included as a covariate in all multivariable regression models to address potential scanner-related differences in image resolution and quality. Additionally, we performed a sensitivity analysis restricting analyses to participants scanned on the Magnetom Vida.
Total brain volume and white matter hyperintensity volume were derived from T1-weighted structural MRI using standardized HABS-HD imaging pipelines.41 T1 images were converted to NIfTI format, bias-corrected using ANTs N4, and processed with FreeSurfer v5.3.0 to obtain whole brain volume; all segmentations underwent visual quality control. FreeSurfer v5.3.0 was used to maintain consistency with the established HABS-HD processing pipeline. Internal benchmarking against a later FreeSurfer release indicated fewer segmentation failures for HABS-HD data using v5.3.0. Intracranial volume and hippocampal volume were estimated using HippoDeep (single available version). HippoDeep was used for these measures because internal HABS-HD comparisons against FreeSurfer v5.3.0 demonstrated fewer segmentation failures and improved performance for hippocampal segmentation and intracranial volume estimation. All outputs underwent pass/fail visual quality control and no manual editing was performed. White matter hyperintensity volume was quantified using combined T1 and FLAIR images processed in SPM12 (SPM 12.r7219, MATLAB 2019B) with the Lesion Segmentation Toolbox (v3.0.0) employing the lesion growth algorithm, which generates lesion probability maps thresholded at 0.3 to produce white matter hyperintensity masks, and outputs were visually inspected for quality. Total brain volume and hippocampal volume were each normalized by intracranial volume using the proportion method (biomarker volume divided by ICV) as described by Nordenskjöld et al. (2015).42 White matter hyperintensity volume was also normalized by ICV and then logarithmically transformed to reduce skewness.
Covariates and comorbidities
Variables included in the analysis as covariates included age (recorded in years), sex at birth, educational attainment (measured in years), and MRI scanner model (Skyra, Vida). Race and ethnicity were self-reported and categorized as non-Hispanic White, non-Hispanic Black, and Hispanic. These categories are used as social constructs representing exposure to structural and social determinants of health; models did not attempt to isolate causal effects of race/ethnicity, and residual confounding by unmeasured social/environmental factors is possible. Comorbidities included the presence of hypertension, dyslipidemia, and diabetes mellitus (each recorded as yes or no). Height and weight were not available in the analytic dataset; therefore, additional adjustment for body size (e.g., BMI) was not performed. These data were collected as part of a structured interview administered questionnaire.
Statistical analysis
All variables of interest were evaluated for missingness; no variable had greater than 6% missing data and therefore a complete case analysis was employed in all remaining analyses. All imaging biomarkers were standardized using z-scores to facilitate comparisons across measures by expressing values in terms of standard deviations from the mean. The primary independent (white matter hyperintensity volume, total brain volume, hippocampal volume) and dependent variables (TUG time, SPPB score) of interest were evaluated to identify any outlying observations and deviations from a normal distribution, where appropriate. Specifically, for TUG time, an interquartile range (IQR)-based method (2 × IQR) with an adjusted threshold to account for the skewness of the data was used to identify and address outliers, in addition to a clinical threshold (TUG time > 20 s) to identify functionally significant outliers. Any data points exceeding either the adjusted IQR threshold or the clinical threshold were flagged as outliers and set to missing. No other outliers were found for the other variables of interest. TUG time and SPPB score were additionally categorized for secondary/exploratory analyses using commonly applied clinical cutpoints (Table 4). Specifically, participants with TUG time less than or equal to 13.5 s were classified as higher physical functioning, and those with TUG time greater than 13.5 s were classified as lower physical functioning.43 For SPPB, scores greater than or equal to 10 were classified as higher physical functioning.44,45
Table 4.
Estimating the relative odds of poor physical functioning based on clinically relevant cutoffs for TUG time and SPPB scores using brain imaging biomarkers, HABS-HD.
| TUG |
SPPB OR (95% CI) |
|||||
|---|---|---|---|---|---|---|
| OR (95% CI) | Raw p | Adj.2 p | Raw p | Adj.2 p | ||
|
| ||||||
| Brain Imaging | ||||||
| WMH | 1.48 (1.10, 2.00) | 0.008 | 0.03 | 1.40 (1.18, 1.65) | <0.01 | <0.01 |
| TBV | 0.46 (0.34, 0.61) | <0.01 | <0.01 | 0.53 (0.45, 0.63) | <0.01 | <0.01 |
| HV | 0.57 (0.43, 0.75) | <0.01 | <0.01 | 0.69 (0.59, 0.81) | <0.01 | <0.01 |
CI: confidence interval; HABS-HD: Health and Aging Brain Study—Health Disparities; NHB: non-Hispanic Black; NHW: non-Hispanic White; TBV: total brain volume normalized by intracranial volume; HV: hippocampal volume normalized by intracranial volume; TUG: Timed Up and Go; WMH: log transformed white matter hyperintensity. Models are adjusted for age, sex, educational level, MRI scanner model, and the presence of any comorbidity (hypertension, dyslipidemia, and/or diabetes mellitus).
Descriptive statistics, comprising means and standard deviations and frequencies and percentages, were computed overall and by race and ethnicity group for age, sex, education, comorbid conditions, MRI scanner model, TUG time (seconds; high/low functioning), SPPB score (score; high/low functioning), and brain imaging markers (unstandardized). Multivariable least squares linear regression models were used to investigate the relation between the standardized brain imaging markers and physical functioning outcomes while adjusting for covariates with product terms included to test for brain imaging and race and ethnicity interactions. Because SPPB is an ordinal, bounded scale and distributions suggested ceiling effects (see Supplemental Figure 1A–D), a sensitivity analysis was conducted using ordinal logistic regression models for SPPB. These models included the same covariate set and race/ethnicity interaction terms as the primary linear models. Further exploratory stratified analyses by race and ethnicity were performed to consider potential subgroup variations. Model assumptions, including linearity, homoscedasticity, and normality of residuals, were verified using variance inflation factors (VIF < 10), residual plots, Q-Q plots, and residual histograms. Finally, multivariable logistic regressions were used to estimate the odds of clinically meaningful diminished physical function per a one-unit increase in each of the brain imaging biomarker (i.e., per 1 SD increase in white matter hyperintensity volume, total brain volume, hippocampal volume). All models included age, sex, education, hypertension, dyslipidemia, type 2 diabetes, and scanner model as covariates. Statistical significance was set at the 0.05 level, with Bonferroni adjustments for multiple comparisons given the exploratory nature of the analyses.
This study protocol was reviewed and approved by the UNTHSC IRB protocols UNTHSC 2016–128 and 2020–125. Each participant (or his/her legal representative) signed written informed consent to participate in the study.
Results
The HABS-HD cohort consisted of 3229 participants with a baseline assessment between 2018 and 2023. Of those, a total of 1051 (32.5%) were excluded based on the a priori exclusion criteria (missing data, n = 303; MCI, n = 580; dementia, n = 168). Among the remaining 2178 participants 42.2% (n = 920) were non-Hispanic White, 19.5% (n = 424) were non-Hispanic Black, and 38.3% (n = 834) were Hispanic (see Figure 1; Table 1).
Figure 1.
Flowchart to derive the final analytic sample of participants, HABS-HD.
Table 1.
Descriptive statistics of key characteristics of analytic sample by race and ethnicity, HABS-HD.
| Characteristic | Total sample (n = 2178) | Non-Hispanic White (n = 920) | Non-Hispanic Black (n = 424) | Hispanic (n = 834) |
|---|---|---|---|---|
|
| ||||
| Age, mean (SD) | 64.7 (8.4) | 68.2 (8.3) | 62.0 (7.3) | 62.3 (7.6) |
| Sex, n (%) | ||||
| Male | 737 (33.8) | 353 (38.4) | 127 (30.0) | 257 (30.8) |
| Female | 1441 (66.2) | 567 (61.6) | 297 (70.1) | 577 (69.2) |
| Education (years), mean (SD) | 13.6 (4.3) | 15.7 (2.5) | 15.3 (2.6) | 10.3 (4.6) |
| Hypertension, n (%) | ||||
| No | 806 (37.0) | 391 (42.5) | 93 (21.9) | 322 (38.6) |
| Yes | 1372 (63.0) | 529 (57.5) | 331 (78.0) | 512 (61.4) |
| Dyslipidemia, n (%) | ||||
| No | 668 (30.7) | 286 (31.1) | 150 (35.4) | 232 (27.8) |
| Yes | 1510 (69.3) | 634 (68.9) | 274 (64.6) | 602 (72.2) |
| Diabetes mellitus, n (%) | ||||
| No | 1684 (77.3) | 808 (87.8) | 323 (76.2) | 553 (66.3) |
| Yes | 494 (22.7) | 112 (12.2) | 101 (23.8) | 281 (33.7) |
| TUG, n (%) | ||||
| Time, mean (SD) | 9.2 (1.5) | 9.0 (1.5) | 9.0 (1.6) | 9.4 (1.5) |
| High PF | 2101 (99.0) | 897 (99.5) | 410 (98.6) | 794 (98.6) |
| Low PF | 22 (1.0) | 5 (0.6) | 14 (1.4) | 11 (1.4) |
| SPPB, n (%) | ||||
| Total score, mean (SD) | 11.0 (1.7) | 11.0 (1.7) | 11.1 (1.7) | 10.9 (1.8) |
| High PF | 1889 (86.7) | 802 (87.2) | 373 (88.0) | 714 (85.6) |
| Low PF | 289 (13.3) | 118 (12.8) | 51 (12.0) | 120 (14.4) |
| Brain imaging, mean (SD) | ||||
| WMH | 4.3 (30.5) | 3.4 (6.3) | 9.1 (59.5) | 2.8 (24.0) |
| HV | 3237.2 (370.2) | 3234.3 (395.0) | 3195.8 (348.5) | 3251.4 (351.0) |
| TBV | 99.9e04 (10.3e04) | 102.4e04 (10.4e04) | 98.5e04 (9.9e04) | 97.6e04 (9.7e04) |
| Scanner | ||||
| Skyra | 1323 (60.7) | 669 (72.7) | 1 (0.2) | 653 (78.3) |
| Vida | 855 (39.3) | 251 (99.8) | 423 (99.8) | 181 (21.7) |
HABS-HD: Health and Aging Brain Study—Health Disparities; HV: hippocampal volume; PF: physical functioning; SD: standard deviation; TUG: Timed Up and Go; SPPB: Short Physical Performance Battery; WMH: white matter hyperintensities; TBV: total brain volume.
Participants were a mean (SD) age of 64.7 (8.4) years with 13.6 (4.3) years of education, and the majority were female (66.2%), self-reported hypertension (63.0%) and dyslipidemia (69.3%), while nearly one-quarter (22.7%) reported having diabetes mellitus. The mean (SD) TUG time for the sample was 9.2 (1.5) seconds which accounted for nearly the entire sample (99.0%) qualifying as having high physical functioning. Accordingly, the TUG-based logistic regression models should be interpreted as exploratory and may be statistically unstable due to sparse outcome counts. The total mean (SD) score for the SPPB among all participants was 11.0 (1.7), allowing for 86.7% of participants to be classified as high functioning based on this test. For the brain imaging measures, the mean (SD) for white matter hyperintensity volume, total brain volume, and hippocampal volume were 4.3 (30.5), 99.9e04 (10.3e04), and 3237.2 (370.2), respectively. MRI scans were acquired on two scanner models, with 60.7% conducted on the Skyra and 39.3% on the Vida.
Table 2 displays the linear associations between measures of physical function and brain imaging outcomes for all participants and with interactions by race and ethnicity group after adjusting for age, sex, and educational level (years), and the presence of hypertension, dyslipidemia, and/or diabetes mellitus; Figure 2 provides a visual summary of the primary main-effect estimates.
Table 2.
Linear associations between measures of physical function and brain imaging outcomes,a HABS-HD.
| TUG time |
SPPB score |
||||||||
|---|---|---|---|---|---|---|---|---|---|
| β (SE) | 95% CI | Raw p | Adj.b p | β (SE) | 95% CI | p | Adj.b p | ||
|
| |||||||||
| Brain Imaging | |||||||||
| WMH | 0.28 (0.06) | 0.17, 0.40 | <0.001 | <0.001 | −0.19 (0.06) | −0.32, −0.07 | <0.001 | 0.02 | |
| TBV | −0.50 (0.05) | −0.61, −0.40 | <0.001 | <0.001 | 0.30 (0.06) | 0.18, 0.42 | <0.001 | <0.01 | |
| HV | −0.28 (0.05) | −0.38, −0.18 | <0.001 | <0.001 | 0.13 (0.06) | 0.02, 0.25 | 0.02 | 0.14 | |
| Interactions | |||||||||
| WMH | |||||||||
| NHW | Ref. | Ref. | |||||||
| NHB | −0.19 (0.09) | −0.38, −0.02 | 0.03 | 0.31 | 0.07 (0.10) | −0.13, 0.27 | 0.48 | 1.00 | |
| Hispanic | −0.11 (0.07) | −0.26, 0.03 | 0.12 | 1.00 | −0.04 (0.08) | −0.20, 0.13 | 0.65 | 1.00 | |
| TBV | |||||||||
| NHW | Ref. | Ref. | |||||||
| NHB | 0.04 (0.08) | −0.13, 0.20 | 0.43 | 1.00 | 0.06 (0.10) | −0.12, 0.25 | 0.52 | 1.00 | |
| Hispanic | 0.15 (0.07) | 0.02, 0.28 | 0.03 | 0.34 | 0.15 (0.08) | −0.00, 0.30 | 0.05 | 0.60 | |
| HV | |||||||||
| NHW | Ref. | Ref. | |||||||
| −0.02 (0.09) | −0.21, 0.16 | 0.80 | 1.00 | 0.22 (0.11) | 0.01, 0.42 | 0.04 | 0.47 | ||
| Hispanic | 0.09 (0.07) | −0.04, 0.23 | 0.18 | 1.00 | 0.13 (0.08) | −0.02, 0.29 | 0.09 | 1.00 | |
CI: confidence interval; HABS-HD: Health and Aging Brain Study—Health Disparities; NHB: non-Hispanic Black; NHW: non-Hispanic White; TBV: total brain volume normalized by intracranial volume; HV: hippocampal volume normalized by intracranial volume; TUG: Timed Up and Go; WMH: log transformed white matter hyperintensity.
Notes:
Models are adjusted for age, sex, educational level, MRI scanner model, and the presence of any comorbidity (hypertension, dyslipidemia, and/or diabetes mellitus);
Bonferroni-adjusted p-values were reported.
Figure 2.
Associations of physical functioning with brain imaging outcomes. Forest plot displaying adjusted β coefficients and 95% confidence intervals from multivariable linear regression models examining associations of white matter hyperintensity volume (WMH), total brain volume (TBV), and hippocampal volume (HV) with Timed Up and Go (TUG) time and Short Physical Performance Battery (SPPB) score. Models include all scanner types and adjust for age, sex, education, hypertension, dyslipidemia, diabetes mellitus, and scanner model.
There were significant linear associations between white matter hyperintensity volume and TUG (β = 0.28, SE = 0.06, adjusted p < 0.01) and SPPB (β = −0.19, SE = 0.06, adjusted p = 0.02), total brain volume and TUG (β = −0.50, SE = 0.05, adjusted p < 0.01) and SPPB (β = 0.30, SE = 0.06, adjusted p < 0.01), and hippocampal volume and TUG (β = −0.28, SE = 0.05, adjusted p < 0.01). Tests for interactions by race and ethnicity revealed the association between greater white matter hyperintensity volume and slower TUG time was attenuated among non-Hispanic Black participants compared to non-Hispanic Whites (β = −0.19, SE = 0.09, p = 0.03), however after Bonferroni adjustment this association failed to reach significance (adjusted p = 0.31). Similarly, the association between lower total brain volume and slower TUG time was attenuated among Hispanic participants relative to non-Hispanic Whites (β = 0.15, SE = 0.07, p = 0.03, adjusted p = 0.34). In contrast, the positive association between hippocampal volume and SPPB score was amplified among non-Hispanic Black participants compared to non-Hispanic Whites (β = 0.22, SE = 0.11, p = 0.04, adjusted p = 0.47). In sensitivity analyses restricted to Magnetom Vida scans, effect estimates were directionally consistent with the primary models and of similar magnitude; reduced precision in the restricted sample attenuated statistical significance for some total brain volume/hippocampal volume associations (Supplemental Table 1). In the sensitivity analysis using ordinal logistic regression for SPPB, associations were directionally consistent with the primary linear models and remained statistically significant for all three MRI markers (Supplemental Table 2).
Table 3 displays stratified analyses by race and ethnicity to further explore these interactions. Consistent with the overall findings, significant linear associations were observed between both physical functioning tests and total brain volume and across all race and ethnicity groups (Bonferroni adjusted p < 0.05). Regarding white matter hyperintensity volume, the association with TUG time remained significant among non-Hispanic White (β = 0.29, 95% CI: [0.16, 0.41], adjusted p < 0.01) and Hispanic participants (β = 0.21, 95% CI: [0.10, 0.33], adjusted p < 0.01), but was attenuated and non-significant among non-Hispanic Black participants (β = 0.05, 95% CI: [−0.10, 0.21], p = 1.00). Similar results were observed for TUG time and hippocampal volume where there were significant associations among non-Hispanic White (β = −0.24, 95% CI: [−0.37, −0.13], adjusted p < 0.01) and Hispanic participants (β = −0.23, 95% CI: [−0.34, −0.12], adjusted p < 0.01), but not among non-Hispanic Black participants after Bonferroni adjustment (β = −0.27, 95% CI: [−0.46, −0.09], unadjusted p < 0.01, adjusted p = 0.07). For SPPB, a similar attenuation in the white matter hyperintensity volume association was observed among non-Hispanic Black participants (β = −0.10, 95% CI: [−0.25, 0.06], adjusted p = 1.00), while significant associations were found in non-Hispanic White and Hispanic participants.
Table 3.
Race and ethnicity heterogeneity of association between TUG time and brain imaging outcomesa, HABS-HD.
| Non-Hispanic White |
Non-Hispanic Black |
Hispanic |
||||||||||
|---|---|---|---|---|---|---|---|---|---|---|---|---|
| β (SE) | 95% CI | Raw p | Adj.2 p | β (SE) | 95% CI | Raw p | Adj.2 p | β (SE) | 95% CI | Raw p | Adj.2 p | |
|
| ||||||||||||
| TUG time | ||||||||||||
| WMH | 0.29 (0.06) | 0.16, 0.41 | <0.01 | <0.01 | 0.05 (0.08) | −0.10, 0.21 | 0.49 | 1.00 | 0.21 (0.06) | 0.10, 0.33 | <0.01 | <0.01 |
| TBV | −0.46 (0.06) | −0.58, −0.34 | <0.01 | <0.01 | −0.46 (0.08) | −0.62, −0.29 | <0.01 | <0.01 | −0.40 (0.06) | −0.51, −0.28 | <0.01 | <0.01 |
| HV | −0.24 (0.06) | −0.37, −0.13 | <0.01 | <0.01 | −0.27 (0.09) | −0.46, −0.09 | <0.01 | 0.07 | −0.23 (0.06) | −0.34, −0.12 | <0.01 | <0.01 |
| SPPB score | ||||||||||||
| WMH | −0.23 (0.07) | −0.37, −0.08 | <0.01 | 0.04 | −0.10 (0.08) | −0.25, 0.06 | 0.22 | 1.00 | −0.26 (0.07) | −0.40, −0.13 | <0.01 | <0.01 |
| TBV | 0.35 (0.07) | 0.21, 0.49 | <0.01 | <0.01 | 0.36 (0.09) | 0.18, 0.53 | <0.01 | <0.01 | 0.48 (0.07) | 0.35, 0.61 | <0.01 | <0.01 |
| HV | 0.15 (0.07) | 0.02, 0.29 | 0.03 | 0.46 | 0.36 (0.09) | 0.18, 0.55 | <0.01 | <0.01 | 0.28 (0.07) | 0.15, 0.42 | <0.01 | <0.01 |
Models are adjusted for age, sex, educational level, MRI scanner model, and the presence of any comorbidity (hypertension, dyslipidemia, and/ordiabetes mellitus).
CI: confidence interval; HABS-HD: Health and Aging Brain Study—Health Disparities; NHB: non-Hispanic Black; NHW: non-Hispanic White; TBV: total brain volume normalized by intracranial volume; HV: hippocampal volume normalized by intracranial volume; TUG: Timed Up and Go; WMH: log transformed white matter hyperintensity. Models are adjusted for age, sex, educational level, MRI scanner model, and the presence of any comorbidity (hypertension, dyslipidemia, and/or diabetes mellitus).
Finally, Table 4 presents the odds of poor physical functioning, based on clinically relevant cutoffs for TUG time and SPPB score, in relation to brain imaging biomarkers. After adjusting for age, sex, educational level, and the presence of hypertension, dyslipidemia, diabetes mellitus and MRI scanner type, higher total brain volume and hippocampal volume were both associated with a Bonferroni adjusted significantly lower odds of poor physical function across both performance tests. Specifically, for each additional standardized unit of total brain volume, there was a 54% lower odds (OR = 0.46; 95% CI: [0.34, 0.61]; p < 0.01) of poor physical functioning using the TUG test and a 47% lower odds (OR = 0.53; 95% CI: [0.45, 0.63]; p < 0.01) of poor SPPB performance. Similarly, higher hippocampal volume was associated with lower odds of poor functioning for both TUG (OR = 0.57; 95% CI: [0.43, 0.75]; p < 0.01) and SPPB (OR = 0.69; 95% CI: [0.59, 0.81]; p < 0.01). White matter hyperintensity volume was significantly associated with greater odds of poor physical functioning on both tests. Each standardized unit increase in white matter hyperintensity volume was associated with a 48% increase in the odds of poor TUG performance (OR = 1.48, 95% CI: [1.10, 2.00], p = 0.03) and a 40% increase in the odds of poor SPPB performance (OR = 1.40, 95% CI: [1.18, 1.65], p < 0.01).
Discussion
This study examined the association between physical functioning and structural brain imaging markers among a large, community-based cohort of cognitively normal adults. The principal findings indicate that physical functioning, measured using both the TUG and SPPB tests, was significantly associated with total brain volume, hippocampal volume, and white matter hyperintensity volume in the overall sample. Associations for total brain and hippocampal volumes were generally consistent across race and ethnicity groups. In contrast, white matter hyperintensity associations appeared attenuated among non-Hispanic Black participants, with non-significant adjusted associations for both TUG and SPPB in stratified analyses.
Although race/ethnicity interaction terms suggested possible heterogeneity in the white matter hyperintensity–TUG association, these did not remain statistically significant after Bonferroni adjustment and should be interpreted cautiously. Overall, these findings suggest that the relationship between structural brain integrity and physical functioning was largely similar across groups, with the main exception being greater heterogeneity for white matter hyperintensity volume. The finding that greater white matter hyperintensity volume and lower total brain and hippocampal volumes are associated with poorer physical functioning is consistent with prior studies in both community-based and clinical populations, using various measures of physical functioning and different aspects of brain structure. Specifically, Makizako et al. (2013) found that greater white matter hyperintensity volume and lower gray matter volumes were linked to poor balance and fall risk in older adults with MCI.38 These findings were later supported with similar findings in other samples of older adults.20,35,36 Shen et al. (2016) prospectively observed that white matter hyperintensity volume was associated with fall risk using the TUG in a hospital-based sample of older adults,20 while Beauchet et al. (2016) found that lower gray matter volume in frontal regions was linked to slow gait cross-sectionally.35 Other subsequent studies from Demnitz et al. (2017) and Tian et al. (2019) observed that lower brain volumes overall, and in certain regions, were associated with motor deficits in aging populations, reinforcing the brain volume-physical function link.36,39 Most recently, Laurienti et al. (2023) further linked reduced gray matter volumes to functional declines, highlighting network-level brain changes in older adults.37
Our results add further support to this literature by demonstrating consistent associations between both SPPB and TUG scores and structural brain volume measures. Importantly, these associations remained robust across models and race and ethnicity groups, suggesting a more universal link between brain volume and functional mobility in aging adults. In contrast, white matter hyperintensity volume associations showed greater heterogeneity: in stratified analyses, white matter hyperintensity volume was not significantly associated with TUG or SPPB among non-Hispanic Black participants after multiple-comparison adjustment. This pattern may reflect limited precision, restricted range of physical functioning, or unmeasured social/vascular determinants of white matter hyperintensity volume; therefore, subgroup-specific white matter hyperintensity volume findings should be interpreted cautiously and evaluated in longitudinal analyses with richer covariate measurement. Furthermore, our findings extend the literature to a community-representative, cognitively normal cohort, suggesting that TUG and SPPB may have potential clinical utility as accessible correlates of structural brain integrity. However, longitudinal studies are needed to determine whether physical functioning predicts subsequent brain changes or cognitive decline.
Although the primary aim of this study was to examine the relation between physical function and brain structure, it is important to consider these findings in the context of prior literature on racial and ethnic disparities in physical functioning. Differences in functional performance across race and ethnicity have been consistently documented in both cross-sectional and longitudinal analyses, with prior work highlighting a range of contributing factors including occupational exposures, socioeconomic status, comorbid disease burden, and unequal access to healthcare.46–49 In our sample, non-Hispanic Black and Hispanic participants had higher prevalence of hypertension, dyslipidemia, and diabetes, and lower educational attainment relative to non-Hispanic White participants. These factors were accounted for in our models, though residual confounding remains possible. Despite these differences, our findings did not reveal substantial heterogeneity of association by race or ethnicity between brain structure and physical functioning. This may reflect true underlying consistency in these associations across groups, or it may reflect the limitations of power in detecting interaction effects, particularly in smaller subgroups like non-Hispanic Black participants (n = 424). Other previous studies on this topic have not fully explored race and ethnicity differences,20,35–39 likely due to homogeneous samples, thereby limiting our ability to compare findings. However, Makizako et al. (2013) suggested that social determinants of health may amplify white matter hyperintensity effects in MCI populations.38 Continued work is needed to understand how structural and social determinants shape brain aging and to clarify the conditions under which race and ethnicity may meaningfully alter these pathways. Any observed differences across racial and ethnic groups should be interpreted as reflecting lived context and accumulated exposure to social and structural determinants of health, rather than innate biological differences.
The current study had several important limitations that should be considered when interpreting the findings. First, the multivariable models adjusted for medical comorbidities, including hypertension, dyslipidemia, and diabetes mellitus, given their known potential to confound the relation between physical functioning and brain structure. However, these variables were assessed via self-report and coded as binary indicators, which may be subject to misclassification. Future work should consider incorporating clinically verified diagnoses, a broader range of chronic conditions (e.g., cerebrovascular disease, chronic kidney disease), and measures of disease duration and/or severity. As discussed previously, race and ethnicity in this study represent social constructs that likely capture a complex interplay of environmental, behavioral, and structural factors. However, we were limited to individual-level covariates available in the analytic dataset (e.g., education and cardiometabolic comorbidities). Unmeasured social and environmental factors (e.g., income, neighborhood conditions, health care access) may confound or mediate associations between MRI markers and physical functioning and may contribute to observed differences in association patterns across groups. Accordingly, group-specific estimates should be interpreted as descriptive and hypothesis-generating rather than as evidence of biologic differences. Another key limitation is the absence of physical activity data in the current models. Physical activity is a known modifier of brain structure and physical function and may explain some of the observed associations between performance-based physical functioning and imaging biomarkers. Tian et al. (2019) highlighted the role of physical fitness in preserving brain volumes, suggesting that including physical activity data could clarify the directionality of our observed total brain and hippocampal volume associations.39 Including device-based and self-reported measures of physical activity in future models would strengthen causal inference and clarify the direction of these associations. Additionally, while the sample was racially and ethnically diverse, subgroup sizes varied, which may have limited the power to detect smaller differences in race-ethnicity interaction terms. Height and weight were not available for additional adjustment; however, all MRI volumetric outcomes were normalized by intracranial volume, which mitigates confounding by head size. Finally, the cross-sectional nature of this analysis limits conclusions regarding temporal sequencing or causality. Although the HABS-HD is a longitudinal cohort, this analysis was restricted to baseline data. Despite these limitations, this study included a large and racially and ethnically diverse sample of cognitively normal adults, which enhances generalizability and helps address critical gaps in prior research. Future work should continue to examine how physical functioning relates to brain health across diverse populations and explore how this information can be used to inform preventive strategies for neurodegenerative diseases.
In summary, this study found that physical functioning, as measured by TUG and SPPB, was consistently associated with total brain and hippocampal volume among cognitively normal adults from a representative community-based sample. These associations were robust across analytic models and largely consistent across race and ethnicity groups, suggesting a more universal link between structural brain integrity and physical performance in aging populations. Although some evidence of race and ethnicity heterogeneity of association was observed, these findings were modest and did not consistently reach statistical significance. The absence of strong differences may reflect the cognitively normal status of the sample, the relatively younger age distribution, or unmeasured social and behavioral factors that influence both brain health and physical function. Importantly, this study adds to a growing body of literature supporting the utility of physical functioning tests as scalable, non-invasive indicators of brain health, particularly in settings where advanced neuroimaging is not feasible. While limitations related to measurement precision, residual confounding, and cross-sectional design should be acknowledged, the study’s strengths—including a large, community-based, diverse cohort and rigorous analytic approach—support the relevance of these findings. Continued investigation into the behavioral, environmental, and structural contributors to brain health disparities is warranted, as is longitudinal research to clarify the temporal dynamics between physical functioning and neurodegeneration. Specifically, modeling physical functioning–to–brain imaging to clarify temporality and quantify the extent to which earlier physical functioning predicts subsequent changes in white matter hyperintensity and brain volumes. Such models may help define clinically meaningful risk stratification approaches and identify individuals who may benefit from earlier evaluation or targeted intervention. These efforts will be critical for refining risk profiles and guiding equitable strategies for early detection and prevention of AD and related conditions.
Supplementary Material
Supplemental material for this article is available online.
Acknowledgements
The authors have no acknowledgments to report.
Funding
The authors disclosed receipt of the following financial support for the research, authorship, and/or publication of this article: Research reported on this publication was supported by the National Institute on Aging of the National Institutes of Health under Award Numbers R01AG054073, R01AG058533, R01AG070862, P41EB015922 and U19AG078109. The content herein is solely the responsibility of the authors and does not necessarily represent the official views of the National Institutes of Health.
Footnotes
Declaration of conflicting interests
The authors declared the following potential conflicts of interest with respect to the research, authorship, and/or publication of this article: S.E.O. has multiple patents on precision medicine for neurodegenerative diseases and is the founding scientist of Cx Precision Medicine. The other authors declare no conflict of interest.
Ethical considerations
This study protocol was reviewed and approved by the UNTHSC IRB protocols UNTHSC 2016–128 and 2020–125.
Consent to participate
Each participant (or his/her legal representative) signed written informed consent to participate in the study.
Consent for publication
Not applicable
Data availability statement
The datasets generated during and/or analyzed during the current study are available in the Image & Data Archive (IDA) run by the Laboratory of Neuro Imaging (LONI), https://ida.loni.usc.edu/login.jsp.
References
- 1.Anderson LA and McConnell SR. Cognitive health: an emerging public health issue. Alzheimers Dement 2007; 3: S70–S73. [DOI] [PubMed] [Google Scholar]
- 2.Vincent GK. The next four decades: The older population in the United States: 2010 to 2050. Washington, DC: US Department of Commerce, Economics and Statistics Administration, 2010. [Google Scholar]
- 3.Nandi A, Counts N, Chen S, et al. Global and regional projections of the economic burden of Alzheimer’s disease and related dementias from 2019 to 2050: a value of statistical life approach. eClinicalMed 2022; 51: 101580. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 4.Morley JE, Morris JC, Berg-Weger M, et al. Brain health: the importance of recognizing cognitive impairment: an IAGG consensus conference. J Am Med Dir Assoc 2015; 16: 731–739. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 5.Plassman BL, Williams JW Jr, Burke JR, et al. Systematic review: factors associated with risk for and possible prevention of cognitive decline in later life. Ann Intern Med 2010; 153: 182–193. [DOI] [PubMed] [Google Scholar]
- 6.Alberdi A, Aztiria A and Basarab A. On the early diagnosis of Alzheimer’s disease from multimodal signals: a survey. Artif Intell Med 2016; 71: 1–29. [DOI] [PubMed] [Google Scholar]
- 7.Graham SA, Lee EE, Jeste DV, et al. Artificial intelligence approaches to predicting and detecting cognitive decline in older adults: a conceptual review. Psychiatry Res 2020; 284: 112732. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 8.Risacher SL and Saykin AJ. Neuroimaging and other biomarkers for Alzheimer’s disease: the changing landscape of early detection. Annu Rev Clin Psychol 2013; 9: 621–648. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 9.DeKosky ST and Marek K. Looking backward to move forward: early detection of neurodegenerative disorders. Science 2003; 302: 830–834. [DOI] [PubMed] [Google Scholar]
- 10.Cummings J Anti-amyloid monoclonal antibodies are transformative treatments that redefine Alzheimer’s disease therapeutics. Drugs 2023; 83: 569–576. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 11.Karran E and De Strooper B. The amyloid hypothesis in Alzheimer disease: new insights from new therapeutics. Nat Rev Drug Discov 2022; 21: 306–318. [DOI] [PubMed] [Google Scholar]
- 12.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: M85–M94. [DOI] [PubMed] [Google Scholar]
- 13.Christopher A, Emily K, Hannah O, et al. The reliability and validity of the Timed Up and Go as a clinical tool in individuals with and without disabilities across a life-span: a systematic review. Disabil Rehabil 2021; 43: 1799–1813. [DOI] [PubMed] [Google Scholar]
- 14.Donoghue OA, Horgan NF, Savva GM, et al. Association between timed Up-and-Go and memory, executive function, and processing speed. J Am Geriatr Soc 2012; 60: 1681–1686. [DOI] [PubMed] [Google Scholar]
- 15.de Oliveira Silva F, Ferreira JV, Placido J, et al. Stages of mild cognitive impairment and Alzheimer’s disease can be differentiated by declines in timed up and go test: a systematic review and meta-analysis. Arch Gerontol Geriatr 2019; 85: 103941. [DOI] [PubMed] [Google Scholar]
- 16.Ibrahim A, Singh DKA and Shahar S. ‘Timed Up and Go’ test: age, gender and cognitive impairment stratified normative values of older adults. PLoS One 2017; 12: e0185641. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 17.Mirelman A, Weiss A, Buchman AS, et al. Association between performance on Timed Up and Go subtasks and mild cognitive impairment: further insights into the links between cognitive and motor function. J Am Geriatr Soc 2014; 62: 673–678. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 18.Knell G, Hall JR, Large S, et al. Alzheimer’s disease plasma biomarkers and physical functioning in a diverse sample of adults. Alzheimers Dement 2025; 21: e14322. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 19.O’Bryant SE, Petersen M, Hall JR, et al. Plasma biomarkers of Alzheimer’s disease are associated with physical functioning outcomes among cognitively normal adults in the multiethnic HABS-HD cohort. J Gerontol A Biol Sci Med Sci 2023; 78: 9–15. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 20.Shen DC, Wu SL, Shi YZ, et al. The correlation between white matter hyperintensity and balance disorder and fall risk: an observational, prospective cohort study. Chronic Dis Transl Med 2016; 2: 173–180. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 21.Catana C, Drzezga A, Heiss W-D, et al. PET/MRI for neurologic applications. J Nucl Med 2012; 53: 1916–1925. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 22.Villemagne VL, Fodero-Tavoletti MT, Masters CL, et al. Tau imaging: early progress and future directions. Lancet Neurol 2015; 14: 114–124. [DOI] [PubMed] [Google Scholar]
- 23.Barkhof F and Scheltens P. Imaging of white matter lesions. Cerebrovasc Dis 2002; 13: 21–30. [DOI] [PubMed] [Google Scholar]
- 24.Pietroboni AM, Colombi A, Carandini T, et al. Amyloid PET imaging and dementias: potential applications in detecting and quantifying early white matter damage. Alzheimers Res Ther 2022; 14: 33. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 25.Jack CR Jr, Petersen RC, Xu YC, et al. Prediction of AD with MRI-based hippocampal volume in mild cognitive impairment. Neurology 1999; 52: 1397–1397. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 26.Aschenbrenner AJ, Gordon BA, Benzinger TL, et al. Influence of tau PET, amyloid PET, and hippocampal volume on cognition in Alzheimer disease. Neurology 2018; 91: e859–e866. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 27.Fox NC and Freeborough PA. Brain atrophy progression measured from registered serial MRI: validation and application to Alzheimer’s disease. J Magn Reson Imaging 1997; 7: 1069–1075. [DOI] [PubMed] [Google Scholar]
- 28.Jack CR Jr, Shiung M, Gunter J, et al. Comparison of different MRI brain atrophy rate measures with clinical disease progression in AD. Neurology 2004; 62: 591–600. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 29.Xia C, Makaretz SJ, Caso C, et al. Association of in vivo [18F]AV-1451 tau PET imaging results with cortical atrophy and symptoms in typical and atypical Alzheimer disease. JAMA Neurol 2017; 74: 427–436. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 30.Prins ND and Scheltens P. White matter hyperintensities, cognitive impairment and dementia: an update. Nat Rev Neurol 2015; 11: 157–165. [DOI] [PubMed] [Google Scholar]
- 31.Tosto G, Zimmerman ME, Carmichael OT, et al. Predicting aggressive decline in mild cognitive impairment: the importance of white matter hyperintensities. JAMA Neurol 2014; 71: 872–877. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 32.Sluimer JD, van der Flier WM, Karas GB, et al. Whole-brain atrophy rate and cognitive decline: longitudinal MR study of memory clinic patients. Radiology 2008; 248: 590–598. [DOI] [PubMed] [Google Scholar]
- 33.Lee S, Viqar F, Zimmerman ME, et al. White matter hyperintensities are a core feature of Alzheimer’s disease: evidence from the dominantly inherited Alzheimer network. Ann Neurol 2016; 79: 929–939. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 34.Pini L, Pievani M, Bocchetta M, et al. Brain atrophy in Alzheimer’s disease and aging. Ageing Res Rev 2016; 30: 25–48. [DOI] [PubMed] [Google Scholar]
- 35.Beauchet O, Allali G, Annweiler C, et al. Association of motoric cognitive risk syndrome with brain volumes: results from the GAIT study. J Gerontol A Biol Sci Med Sci 2016; 71: 1081–1088. [DOI] [PubMed] [Google Scholar]
- 36.Demnitz N, Zsoldos E, Mahmood A, et al. Associations between mobility, cognition, and brain structure in healthy older adults. Front Aging Neurosci 2017; 9: 155. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 37.Laurienti PJ, Miller ME, Lyday RG, et al. Associations of physical function and body mass index with functional brain networks in community-dwelling older adults. Neurobiol Aging 2023; 127: 43–53. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 38.Makizako H, Shimada H, Doi T, et al. Poor balance and lower gray matter volume predict falls in older adults with mild cognitive impairment. BMC Neurol 2013; 13: 102. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 39.Tian Q, Resnick SM, Davatzikos C, et al. A prospective study of focal brain atrophy, mobility and fitness. J Intern Med 2019; 286: 88–100. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 40.Petersen ME, Zhou Z, Hall JR, et al. Health and Aging Brain Study-Health Disparities (HABS-HD) methods and partner characteristics. Alzheimers Dement (N Y) 2025; 11: e70140. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 41.Tsiknia AA, Tennant VR, Lee N, et al. Diabetes and cortical thickness in ethnically diverse cognitively normal older adults. Alzheimers Dement (Amst) 2025; 17: e70088. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 42.Nordenskjöld R, Malmberg F, Larsson E-M, et al. Intracranial volume normalization methods: considerations when investigating gender differences in regional brain volume. Psychiatry Res Neuroimaging 2015; 231: 227–235. [DOI] [PubMed] [Google Scholar]
- 43.Shumway-Cook A, Brauer S and Woollacott M. Predicting the probability for falls in community-dwelling older adults using the Timed Up & Go test. Phys Ther 2000; 80: 896–903. [PubMed] [Google Scholar]
- 44.Bernabeu-Mora R, Medina-Mirapeix F, Llamazares-Herrán E, et al. The Short Physical Performance Battery is a discriminative tool for identifying patients with COPD at risk of disability. Int J Chron Obstruct Pulmon Dis 2015; 10: 2619–2626. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 45.Pavasini R, Guralnik J, Brown JC, et al. Short Physical Performance Battery and all-cause mortality: systematic review and meta-analysis. BMC Med 2016; 14: 215. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 46.Dunlop DD, Song J, Manheim LM, et al. Racial/ethnic differences in the development of disability among older adults. Am J Public Health 2007; 97: 2209–2215. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 47.Kelley-Moore JA and Ferraro KF. The black/white disability gap: persistent inequality in later life? J Gerontol B Psychol Sci Soc Sci 2004; 59: S34–S43. [DOI] [PubMed] [Google Scholar]
- 48.Louie GH and Ward MM. Socioeconomic and ethnic differences in disease burden and disparities in physical function in older adults. Am J Public Health 2011; 101: 1322–1329. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 49.Manton KG and Gu X. Changes in the prevalence of chronic disability in the United States black and nonblack population above age 65 from 1982 to 1999. Proc Natl Acad Sci U S A 2001; 98: 6354–6359. [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.
Supplementary Materials
Data Availability Statement
The datasets generated during and/or analyzed during the current study are available in the Image & Data Archive (IDA) run by the Laboratory of Neuro Imaging (LONI), https://ida.loni.usc.edu/login.jsp.