Abstract
Aim
Frailty increases the risk of cognitive decline in older adults, yet the brain structural patterns associated with different frailty subtypes remain unclear. This study examined white matter (WM) alterations across frailty subtypes in community‐dwelling older adults without dementia.
Methods
This cross‐sectional study included participants from the Taiwan Precision Medicine Initiative on Cognitive Impairment and Dementia (TPMIC) cohort. Frailty was assessed using Fried's phenotype and classified into mobility, non‐mobility, and low physical activity subtypes. Cognitive function (attention, memory, and executive function) was evaluated using standardized neuropsychological tests. WM microstructure was measured using diffusion tensor imaging (DTI) metrics, including fractional anisotropy (FA) and mean diffusivity (MD).
Results
Among 297 participants, 91 (30.6%) were pre‐frail/frail. The pre‐frail/frail group showed widespread WM alterations, with the internal capsule (IC) remaining significant after full adjustment and FDR correction (q < 0.05). The mobility subtype was associated with poorer cognitive performance across all domains (all P < 0.01) and showed lower FA and higher MD primarily in motor and cognitive‐related tracts, such as the corpus callosum (all q < 0.05). In contrast, the non‐mobility subtype was associated with poorer attention and executive function, with alterations primarily in emotion‐regulation tracts, such as the cingulum and forceps major (all q < 0.05). No significant WM differences were found for the low physical activity subtype.
Conclusion
Frailty subtypes are associated with distinct WM alteration patterns, reflecting potentially different mechanisms of brain aging. These findings highlight the importance of subtype‐specific approaches to early detection and intervention.
Keywords: cognitive function, diffusion tensor imaging, frailty, older adults, white matter
Frailty is a prevalent condition among older adults, characterized by reduced physiological reserves and increased vulnerability to adverse health outcomes. 1 Fried's frailty phenotype defines frailty using five components: slowness, weakness, exhaustion, weight loss, and low physical activity, underscoring its multidimensional and heterogeneous nature. 1 These components may emerge at different stages of aging, influencing health outcomes through distinct mechanisms. 2 The mobility subtype (slowness and weakness) tends to emerge earlier in aging, whereas the non‐mobility subtype (exhaustion and weight loss) may indicate rapid frailty progression when they appear first. 2 , 3 These distinctions suggest that frailty subtypes may uniquely shape the aging process and contribute to varied outcomes.
Increasing evidence links frailty to cognitive impairment and dementia. 4 , 5 Neuroimaging studies associate frailty with gray matter (GM) atrophy, 6 , 7 , 8 cerebrovascular damage, 9 and higher β‐amyloid burden, 10 all of which contribute to cognitive decline. These findings highlight the strong relationship between frailty, brain aging, and cognitive impairment. While most research has focused on alterations in cerebral cortex and subcortical GM, emerging studies suggest the critical role of white matter (WM) integrity in frailty‐related brain aging. Greater frailty severity has been associated with greater WM hyperintensity volume and a higher burden of cerebral infarcts, 7 , 11 suggesting that WM damage plays a significant role in the neural deterioration linked to frailty. More recently, diffusion tensor imaging (DTI) studies have identified early WM microstructural alterations in frail individuals. 11 , 12 These changes are often detectable before the appearance of overt lesions such as WM hyperintensities on conventional imaging, 13 suggesting that DTI metrics may serve as earlier markers of WM alterations.
Most studies on frailty and brain aging have treated frailty as a unified construct or focused primarily on mobility‐related subtypes, with limited attention to how different subtypes relate to WM alterations. 14 One prior study linked the mobility subtype to cerebellar GM atrophy, whereas the non‐mobility subtype affected non‐cerebellar GM regions. 6 Although frailty has been broadly associated with brain structural changes, the relationships between specific subtypes and WM characteristics remain underexplored. Given the heterogeneity of frailty, investigating whether WM alterations vary across subtypes may provide important insights into its underlying neural mechanisms. In this study, we investigated the associations between frailty subtypes and tract‐specific WM alterations in community‐dwelling older adults. We hypothesized that different subtypes would be associated with distinct patterns of WM alterations.
Methods
Study Design and Population
This study utilized data from the Taiwan Precision Medicine Initiative on Cognitive Impairment and Dementia (TPMIC), the study design, research focus, and methodology have been previously described. 15 For this cross‐sectional analysis, we included participants who had data on frailty, cognitive function, and DTI measures from July 2021 to May 2022. Of the 334 participants, 19 were excluded due to prevalent dementia and 18 due to missing frailty data, resulting in a final sample of 297 participants. All MRI scans were visually assessed by a board‐certified radiologist and a neurologist as part of the imaging quality control protocol. Participants with significant structural abnormalities were excluded, including those with (1) a history of significant brain injury, (2) brain tumors or large territorial infarcts, (3) extensive leukoencephalopathy indicative of advanced small vessel disease, or (4) severe global or hippocampal atrophy. Based on these criteria, seven participants were excluded, resulting in a final sample of 290 individuals in the MRI analysis. This analysis was based on observational data and did not include a clinical trial or intervention procedures. The study protocol was reviewed and approved by the Institutional Review Board (IRB) of Far Eastern Memorial Hospital (IRB No. 110065‐F) and Cardinal Tien Hospital (IRB No. CTH‐110‐2‐1‐014). All participants provided written informed consent, and the study adhered to the ethical principles of the Declaration of Helsinki.
Assessment of Frailty
Frailty was assessed using Fried's criteria, consisting of slowness, weakness, exhaustion, weight loss, and low physical activity. 1 This operational definition followed a validated Taiwanese adaptation to ensure cultural relevance. 16 Slowness was defined by slow gait speed, adjusted for sex and height, and weakness by grip strength, adjusted for sex and body mass index (BMI). Exhaustion was identified by self‐reported fatigue on more than 3 days in the past week. Weight loss was unintentional loss of over 3 kg or 5% of body weight in the past year. Physical activity was assessed using the Taiwanese version of the International Physical Activity Questionnaire – Short Form (IPAQ‐SF). Although the original IPAQ scoring protocol estimates activity in MET‐minutes per week, in this study, the values were converted into weekly energy expenditure (kcal/week), a method commonly used in Taiwanese community‐based studies to enhance clinical applicability. Participants were classified as having low physical activity if their estimated weekly energy expenditure was <685 kcal for men or <420 kcal for women, thresholds that have been adopted and validated in community‐based frailty studies in Taiwan. 16 Participants meeting three or more criteria were classified as frail, those meeting one or two were pre‐frail, and those meeting none were robust. 1 Frail and pre‐frail individuals were combined due to low frailty prevalence. Participants were categorized into frailty subtypes based on the presence of specific components. 17 The mobility subtype included participants with weakness and/or slowness only. The non‐mobility subtype included those with exhaustion and/or weight loss only. Participants with only low physical activity were classified as low physical activity subtype.
Cognitive Function
Global cognition was evaluated using the Mini‐Mental State Examination (MMSE). Specific domains of cognitive functions, including attention, memory, and executive function, were assessed using neuropsychological tests as previously described. 18 Attention was evaluated using the Logical Memory Test I (LMI) of the Wechsler Memory Scale‐III (WMS‐III), Forward and Backward Digit Span Test (DS), Symbol Substitution Test (SST) of the Wechsler Adult Intelligence Scale‐III (WAIS‐III), and Color Trail Making Test I (CTT1). Memory was assessed using delayed recall of the WMS‐III Logical Memory Test II (LMII). Executive function was assessed using the Color Trail Making Test II (CTT2), Stroop Color and Word Test (SCWT), and Semantic Verbal Fluency (VF). The difference between CTT2 and CTT1 completion times was used to evaluate executive function while controlling for attention. Raw scores were standardized into z‐scores and averaged for each cognitive domain. Trained examiners conducted all tests with standardized instructions on the same day.
DTI Acquisition and White Matter Tracts
Two MRI sequences were used to perform DTI: a 20‐direction diffusion‐weighted echo‐planar imaging (EPI) sequence with b‐values of 0 and 1000 s/mm2 and a 71‐direction multi‐shell diffusion‐weighted EPI sequence with b‐values of 0, 1000, and 2000 s/mm2 (6 b0 images, 20 b1000 images, and 45 b2000 images). Imaging parameters for Sequence 1 were repetition time (TR) = 5200 ms, echo time (TE) = 92 ms, 35 axial interleaved slices (slice thickness = 4 mm), field‐of‐view (FOV) = 220 × 220 mm, matrix = 128 × 128, and voxel size = 1.7 × 1.7 × 4 mm. Sequence 2 had TR = 9600 ms, TE = 97 ms, 68 axial interleaved slices (slice thickness = 2.2 mm), FOV = 220 × 220 mm, matrix = 100 × 100, and voxel size = 2.2 × 2.2 × 2.2 mm. DTI analysis was conducted using PANDA 1.3.1, a MATLAB toolbox utilizing FSL 6.0. 19 A brain mask was used to remove the skull with FSL's brain extraction tool. Following that, FSL's eddy correction algorithm was implemented to correct head motion artifacts and eddy current‐induced distortions and motion artifacts. The diffusion tensors were then reconstructed using the DTIFIT command, yielding DTI metrics. Notably, when analyzing 71‐direction multi‐shell diffusion MRI datasets, only inner‐shell diffusion‐weighted images with b‐values of 0 and 1000 s/mm2 were used to perform diffusion tensor modeling. To compare the DTI metrics derived from raw DTI data in different spatial resolutions, we performed PANDA to write different subjects' DTI metrics into the same standard space by registering their images into the FMRIB58_FA template in the MNI space at 1 1 × 1 × 1 mm resolution via FMRIB's Nonlinear Image Registration Tool (FNIRT), a non‐linear registration approach. Each individual's fractional anisotropy (FA) image was first registered onto this template, and the resultant warping transformation was then used to write a mean diffusivity (MD) image into the MNI space. Ultimately, the PANDA pipeline incorporates the ICBM‐DTI‐81 WM labels atlas and the JHU WM tractography atlas to parcellate the entire WM into multiple regions of interest (ROIs), enabling subsequent ROI‐based analyses. 19 , 20 , 21 Combined association test (ComBat), a harmonization technique, was conducted to eliminate protocol variations in DTI data. 22 FA and MD, as complementary diffusion metrics, were used to characterize WM microstructure in subsequent analyses. FA reflects the directional coherence of water diffusion and is influenced by axonal density, fiber alignment, and degree of myelination. Lower FA values may indicate microstructural disruption, such as reduced axonal density or loss of myelination. MD, in contrast, tends to increase in the presence of tissue rarefaction or extracellular fluid accumulation resulting from edema or inflammation. Together, these metrics provide distinct yet complementary insights into WM microstructure. FA values range from 0 to 1, with lower values indicating less organized fiber structure, while higher MD values reflect increased water diffusion, often associated with axonal loss or extracellular matrix breakdown. 23 , 24 , 25 , 26
The following WM tracts were selected as ROIs: the superior longitudinal fasciculus (SLF), inferior longitudinal fasciculus (ILF), inferior fronto‐occipital fasciculus (IFOF), uncinate fasciculus (UF), cingulum (including the cingulate gyrus and hippocampal segments), anterior thalamic radiation (ATR), corticospinal tract, internal capsule (IC), corpus callosum (CC; subdivided into the genu, body, and splenium), and forceps major and minor. In addition, three cerebellar peduncles, the inferior (ICP), middle (MCP), and superior (SCP), were included. These tracts were chosen for their extensive coverage of functionally relevant brain regions and their established roles in motor control, visuospatial processing, memory, and executive function. 27 The rationale for tract selection is provided in Table S1.
Statistical Analyses
Differences in baseline characteristics between non‐frail and pre‐frail/frail participants were examined using t‐tests and chi‐square tests, as appropriate. Multiple linear regression models were conducted to assess associations between frailty status, frailty subtypes, cognitive functions, and DTI metrics. All models were first adjusted for age, sex, and education (Model 1), and further adjusted for hypertension, hyperlipidemia, stroke, and depressive symptoms in Model 2. To address the issue of multiple comparisons across 17 tracts and two DTI metrics (FA and MD), we applied false discovery rate (FDR) correction using the Benjamini–Hochberg procedure across all 34 tract–metric combinations. Results with a q‐value <0.05 were considered statistically significant. All analyses were performed using SAS 9.4 and R (Version 4.0.4).
Results
Of 297 participants with a mean age of 70, 35.7% were male. Based on frailty status, 206 were classified as robust, while 91 were pre‐frail and six were frail, resulting in a combined pre‐frail/frail prevalence of 30.6%. Among the pre‐frail participants, 75 met one criterion, and 10 met two. Compared to the robust group, individuals in the pre‐frail/frail group were significantly older and had a higher prevalence of hypertension and depressive symptoms (Table 1). No significant group differences were observed in sex, education years, smoking status, alcohol consumption, history of diabetes, hyperlipidemia, coronary artery disease, or stroke. Key frailty components were exclusively present in the pre‐frail/frail group (all P‐values <0.001).
Table 1.
Characteristics of participants by frailty status
| Total | Robust | Pre‐frail/frail | P‐value | |
|---|---|---|---|---|
| n | 297 | 206 | 91 | |
| Demographics | ||||
| Age, years | 69.9 (6.4) | 69.4 (6.1) | 71.1 (6.8) | 0.029 |
| Male, n (%) | 106 (35.7) | 70 (34.0) | 36 (39.6) | 0.427 |
| Education, years | 10.1 (4.6) | 10.4 (4.5) | 9.5 (4.7) | 0.146 |
| Smoking, n (%) | 64 (21.8) | 40 (19.4) | 24 (27.3) | 0.180 |
| Alcohol drinking, n (%) | 94 (32.0) | 62 (30.1) | 32 (36.4) | 0.358 |
| Diabetes, n (%) | 62 (21.2) | 40 (19.5) | 22 (25.0) | 0.369 |
| Hypertension, n (%) | 129 (44.0) | 80 (39.0) | 49 (55.7) | 0.012 |
| Hyperlipidemia, n (%) | 102 (34.9) | 65 (31.7) | 37 (42.5) | 0.101 |
| Coronary artery disease, n (%) | 34 (11.6) | 20 (9.8) | 14 (15.9) | 0.191 |
| Stroke, n (%) | 10 (3.4) | 4 (2.0) | 6 (6.8) | 0.081 |
| Depressive symptoms*, n (%) | 32 (11.1) | 12 (5.9) | 20 (23.0) | <0.001 |
| Frailty components | ||||
| Walking speed, m/s | 0.85 (0.23) | 0.90 (0.21) | 0.74 (0.23) | <0.001 |
| Slowness, n (%) | 15 (5.1) | 0 (0.0) | 15 (16.5) | |
| Grip strength, kg | 24.5 (8.1) | 25.1 (7.6) | 23.3 (8.9) | 0.071 |
| Weakness, n (%) | 18 (6.1) | 0 (0.0) | 18 (19.8) | |
| Exhaustion, n (%) | 11 (3.7) | 0 (0.0) | 11 (12.1) | |
| Weight loss, n (%) | 14 (4.7) | 0 (0.0) | 14 (15.4) | |
| IPAQ, kcal/week | 2538.1 (3402.5) | 3276.3 (3780.7) | 867.0 (1203.3) | <0.001 |
| Low physical activity, n (%) | 56 (18.9) | 0 (0.0) | 56 (61.5) | |
| Cognitive function | ||||
| MMSE | 27.0 (3.2) | 27.5 (2.3) | 25.9 (4.4) | <0.001 |
| Logic memory I | 29.6 (12.8) | 30.8 (12.7) | 26.8 (12.7) | 0.014 |
| Logic memory II | 15.9 (9.4) | 17.1 (9.3) | 12.9 (8.9) | <0.001 |
| Digit span | 20.7 (5.9) | 21.0 (5.4) | 20.1 (6.8) | 0.212 |
| Symbol substitution | 55.2 (20.8) | 57.9 (19.9) | 49.0 (21.5) | 0.001 |
| Colored trails 1, sec | 64.2 (47.6) | 57.2 (32.0) | 80.3 (69.3) | <0.001 |
| Colored trails 2–1, sec | 75.9 (52.7) | 70.8 (42.9) | 87.9 (69.4) | 0.011 |
| Stroop test, interference | −2.1 (10.1) | −1.6 (9.8) | −3.2 (10.6) | 0.226 |
| Semantic verbal fluency | 35.3 (9.0) | 35.9 (8.1) | 33.9 (10.7) | 0.085 |
A geriatric depression scale (GDS) score greater than five was classified as indicative of depressive symptoms.
Data presented as mean (SD) unless otherwise indicated. IPAQ, International Physical Activity Questionnaire Short Form; MMSE, Mini‐Mental Status Examination.
Associations Between Frailty and Cognitive Function
Pre‐frail/frail participants were associated with poorer memory performance after accounting for covariates (Table 2). When examining frailty subtypes, both the mobility and non‐mobility subtypes were associated with worse cognitive functions compared to robust individuals. Specifically, the mobility subtype was significantly associated with poorer performance across all cognitive domains, including attention, memory, and executive function (Table S2). In contrast, the non‐mobility subtype was significantly associated with lower attention scores, and the low physical activity subtype did not show significant associations with any cognitive domains.
Table 2.
Associations between frailty status and cognitive function (pre‐frail/frail vs. robust)
| Model 1 (n = 297) | Model 2 (n = 285) | |
|---|---|---|
| Cognitive domain | β (95% C.I.) | β (95% C.I.) |
| Attention | −0.19 (−0.32, −0.05)** | −0.13 (−0.28, 0.01) |
| Memory | −0.31 (−0.52, −0.11)** | −0.29 (−0.50, −0.07)* |
| Executive Function | −0.13 (−0.29, 0.03) | −0.14 (−0.32, 0.03) |
P < 0.05;
P < 0.01.
Model 1 adjusted for age, sex, and education.
Model 2 adjusted for age, sex, education, hypertension, hyperlipidemia, stroke, and depressive symptoms.
Further analysis of the individual frailty components revealed patterns consistent with the subtype findings. Weakness was significantly associated with poorer performance across all cognitive domains, mirroring the broad impact observed for the mobility subtype. Weight loss was significantly associated with poorer attention and executive function performance, while exhaustion was linked specifically to executive function (Table S3).
Associations Between Frailty Status and White Matter Alterations
Group differences in FA and MD were examined across 17 WM ROIs between pre‐frail/frail and robust participants (Table 3). In unadjusted analyses, the pre‐frail/frail group showed significantly lower FA and higher MD in multiple tracts, including SLF, IFOF, UF, ATR, cingulum (cingulate gyrus), IC, all segments of the CC, and forceps minor. Only IC remained significantly different between groups after controlling for potential confounders and applying FDR correction for multiple comparisons.
Table 3.
Differences in fractional anisotropy (FA) and mean diffusivity (MD) across frailty groups in selected white matter tracts
| White matter tract | Robust (n = 201) | Pre‐frail/frail (n = 89) | |
|---|---|---|---|
| Superior longitudinal fasciculus | FA | 0.38 (0.02) | 0.37 (0.03)* |
| MD | 6.74 (0.33) | 6.90 (0.48)* | |
| Inferior longitudinal fasciculus | FA | 0.42 (0.02) | 0.42 (0.03) |
| MD | 7.10 (0.31) | 7.22 (0.33)* | |
| Inferior fronto‐occipital fasciculus | FA | 0.41 (0.02) | 0.40 (0.03)* |
| MD | 7.22 (0.36) | 7.43 (0.48)* | |
| Uncinate fasciculus | FA | 0.38 (0.02) | 0.37 (0.03)* |
| MD | 7.58 (0.36) | 7.73 (0.44)* | |
| Cingulum‐cingulate gyrus | FA | 0.44 (0.03) | 0.43 (0.04)* |
| MD | 6.57 (0.38) | 6.72 (0.43)* | |
| Cingulum‐hippocampus | FA | 0.33 (0.03) | 0.32 (0.03) |
| MD | 7.55 (0.45) | 7.71 (0.47)* | |
| Anterior thalamic radiation | FA | 0.36 (0.03) | 0.35 (0.03)* |
| MD | 7.91 (0.90) | 8.32 (1.19)* | |
| Corticospinal tract | FA | 0.56 (0.02) | 0.56 (0.03) |
| MD | 6.57 (0.22) | 6.68 (0.36)* | |
| Internal capsule | FA | 0.57 (0.02) | 0.56 (0.03)*, † |
| MD | 6.45 (0.29) | 6.66 (0.46)*, † | |
| Genu of Corpus Callosum | FA | 0.54 (0.03) | 0.53 (0.05)* |
| MD | 7.51 (0.51) | 7.77 (0.70)* | |
| Body of Corpus Callosum | FA | 0.50 (0.04) | 0.48 (0.05)* |
| MD | 8.24 (0.54) | 8.49 (0.71)* | |
| Splenium of Corpus Callosum | FA | 0.61 (0.03) | 0.60 (0.04)* |
| MD | 7.87 (0.46) | 8.08 (0.62)* | |
| Forceps major | FA | 0.55 (0.03) | 0.54 (0.04)* |
| MD | 7.67 (0.63) | 7.85 (0.75) | |
| Forceps minor | FA | 0.41 (0.03) | 0.40 (0.03)* |
| MD | 7.51 (0.41) | 7.70 (0.52)* | |
| Middle Cerebellar Peduncle | FA | 0.45 (0.02) | 0.45 (0.03)* |
| MD | 7.18 (0.35) | 7.25 (0.47) | |
| Inferior Cerebellar Peduncle | FA | 0.43 (0.04) | 0.42 (0.05) |
| MD | 6.96 (0.46) | 7.10 (0.62) | |
| Superior Cerebellar Peduncle | FA | 0.52 (0.03) | 0.52 (0.03)* |
| MD | 9.36 (0.63) | 9.47 (0.74) |
Significant group difference based on FDR‐adjusted P‐values (q < 0.05).
Indicates a significant association in linear regression models, adjusted for age, sex, education, hypertension, hyperlipidemia, stroke, and depressive symptoms, based on FDR‐adjusted P‐values (q < 0.05).
Data expressed as mean (standard deviation).
FA, fractional anisotropy; MD, mean diffusivity (10−4 mm2/s).
Associations Between Frailty Subtypes and White Matter Alterations
Group differences in WM microstructure were further examined across frailty subtypes using tract‐specific FA and MD metrics (Table 4). Participants classified under the mobility subtype showed widespread alterations, with significantly lower FA in several tracts, including the ATR, cingulum (cingulate), IC, CC (genu, and splenium), forceps major and minor, and MCP. Corresponding MD increases were only observed in IC. The non‐mobility subtype showed a more localized pattern of differences, with significantly lower FA in the IC, genu, and splenium of CC, and forceps major. No significant MD differences were observed for this group. In contrast, the low physical activity subtype was not associated with significant FA or MD differences across any WM tracts.
Table 4.
Associations between frailty subtypes and diffusion metrics of white matter tracts
| Subtypes of frailty (Robust as reference n = 201) | Mobility frailty (n = 16) | Non‐mobility frailty (n = 13) | Low physical activity (n = 46) | |
|---|---|---|---|---|
| White Matter Tract | β (95% C.I.) | β (95% C.I.) | β (95% C.I.) | |
| Superior longitudinal fasciculus | FA | −1.26 [−2.36, −0.17] | −1.51 [−2.74, −0.29] | 0.43 [−0.23, 1.09] |
| MD | 0.27 [0.10, 0.44]* | 0.10 [−0.08, 0.28] | −0.05 [−0.15, 0.05] | |
| Inferior longitudinal fasciculus | FA | −1.32 [−2.56, −0.07] | −1.02 [−2.42, 0.38] | 0.58 [−0.21, 1.36] |
| MD | 0.16 [0.01, 0.31] | 0.10 [−0.07, 0.27] | −0.04 [−0.13, 0.06] | |
| Inferior fronto‐occipital fasciculus | FA | −1.41 [−2.62, −0.20] | −1.14 [−2.50, 0.22] | 0.51 [−0.22, 1.25] |
| MD | 0.32 [0.14, 0.49]* | 0.11 [−0.08, 0.30] | −0.002 [−0.11, 0.10] | |
| Uncinate fasciculus | FA | −1.31 [−2.50, −0.11] | −0.55 [−1.87, 0.76] | −0.42 [−1.14, 0.29] |
| MD | 0.19 [−0.002, 0.38] | 0.03 [−0.18, 0.24] | −0.002 [−0.12, 0.11] | |
| Cingulum‐cingulate gyrus | FA | −2.81 [−4.37, −1.25]* | −1.94 [−3.77, −0.10] | 0.71 [−0.31, 1.72] |
| MD | 0.22 [0.04, 0.40] | 0.21 [−0.01, 0.42] | −0.06 [−0.17, 0.05] | |
| Cingulum‐hippocampus | FA | −0.42 [−2.03, 1.19] | −0.98 [−2.78, 0.82] | 0.14 [−0.89, 1.18] |
| MD | 0.14 [−0.08, 0.35] | 0.21 [−0.03, 0.45] | −0.01 [−0.14, 0.13] | |
| Anterior thalamic radiation | FA | −1.95 [−3.18, −0.73]* | −1.37 [−2.81, 0.06] | 0.50 [−0.26, 1.26] |
| MD | 0.57 [0.14, 0.99] | 0.22 [−0.28, 0.72] | −0.07 [−0.33, 0.19] | |
| Corticospinal tract | FA | −1.37 [−2.55, −0.19] | −1.17 [−2.50, 0.16] | 0.73 [−0.02, 1.47] |
| MD | 0.20 [0.08, 0.32]* | 0.09 [−0.03, 0.22] | −0.06 [−0.13, 0.01] | |
| Internal capsule | FA | −1.65 [−2.85, −0.44]* | −1.90 [−3.26, −0.55]* | 0.49 [−0.24, 1.23] |
| MD | 0.31 [0.16, 0.46]* | 0.20 [0.04, 0.36] | 0.01 [−0.08, 0.10] | |
| Genu of Corpus Callosum | FA | −2.88 [−4.60, −1.16]* | −2.65 [−4.68, −0.63]* | 0.93 [−0.13, 1.99] |
| MD | 0.35 [0.09, 0.60] | 0.25 [−0.06, 0.56] | −0.07 [−0.23, 0.08] | |
| Body of Corpus Callosum | FA | −2.28 [−4.11, −0.44] | −2.12 [−4.3, 0.07] | 0.56 [−0.63, 1.75] |
| MD | 0.28 [0.01, 0.54] | 0.15 [−0.17, 0.47] | −0.05 [−0.22, 0.11] | |
| Splenium of Corpus Callosum | FA | −2.27 [−3.72, −0.82]* | −2.7 [−4.53, −0.87]* | 0.47 [−0.41, 1.36] |
| MD | 0.29 [0.05, 0.52] | 0.33 [0.06, 0.6] | −0.04 [−0.18, 0.11] | |
| Forceps major | FA | −2.04 [−3.47, −0.60]* | −2.46 [−4.19, −0.72]* | 0.26 [−0.61, 1.14] |
| MD | 0.38 [0.05, 0.71] | 0.44 [0.07, 0.81] | −0.01 [−0.22, 0.20] | |
| Forceps minor | FA | −1.70 [−2.98, −0.43]* | −1.69 [−3.17, −0.22] | 0.68 [−0.11, 1.46] |
| MD | 0.27 [0.07, 0.47] | 0.15 [−0.09, 0.39] | −0.05 [−0.18, 0.07] | |
| Middle Cerebellar Peduncle | FA | −1.53 [−2.65, −0.41]* | −1.22 [−2.50, 0.07] | 0.49 [−0.19, 1.17] |
| MD | 0.07 [−0.12, 0.26] | 0.17 [−0.04, 0.38] | −0.08 [−0.20, 0.03] | |
| Inferior Cerebellar Peduncle | FA | −1.56 [−3.51, 0.39] | −2.14 [−4.41, 0.13] | 1.24 [0.03, 2.44] |
| MD | 0.31 [0.06, 0.56] | 0.11 [−0.18, 0.39] | −0.09 [−0.24, 0.07] | |
| Superior Cerebellar Peduncle | FA | −1.55 [−2.98, −0.12] | −1.28 [−2.92, 0.36] | 0.31 [−0.57, 1.20] |
| MD | 0.26 [−0.08, 0.60] | 0.30 [−0.08, 0.69] | −0.06 [−0.29, 0.17] |
FDR‐adjusted P‐value (q) < 0.05.
All models were adjusted for age, sex, education, hypertension, hyperlipidemia, stroke, and depressive symptoms.
FA, fractional anisotropy (10−2mm2/s); MD, mean diffusivity (10−4 mm2/s).
To further explore these patterns in greater detail, we examined associations between individual frailty components and DTI metrics (Fig. 1). Among the mobility‐related components, slowness was associated with widespread alterations, with significant reductions in FA and increases in MD observed in the SLF, ILF, IFOF, ATR, IC, genu, and splenium of the CC, and forceps minor. In contrast, weakness showed more limited associations, with significant FA and MD alterations confined to the ATR and ICP. These results suggest that the broader pattern observed in the mobility subtype may be primarily driven by slowness. Non‐mobility components showed greater heterogeneity. Exhaustion was associated with FA and MD alterations in only two regions, ATR and the splenium of the CC. In contrast, weight loss was linked to more extensive changes, with significant FA and MD differences observed across multiple tracts. No significant associations were found for low physical activity.
Fig. 1.
Significant Associations Between Frailty Components and White Matter Tracts. This figure illustrates the significant associations between frailty components: (a) slowness, (b) weakness, (c) exhaustion, and (d) weight loss and white matter tracts. Significant associations are defined by participants with a specific frailty component showing lower fractional anisotropy (FA) and higher mean diffusivity (MD) in specific white matter tracts compared to those without the component. Significant group difference based on FDR‐adjusted P‐values (q < 0.05). ATR, anterior thalamic radiation; CC, corpus callosum; CST, corticospinal tract; CgC, cingulum (cingulate gyrus); CgH, cingulum (hippocampus); Fma, forceps major; Fmi, forceps minor; IC, internal capsule; IFOF, inferior fronto‐occipital fasciculus; ICP, inferior cerebellar peduncle; ILF, inferior longitudinal fasciculus; SLF, superior longitudinal fasciculus. Orientation: A, anterior; P, posterior; S, superior; I, inferior; L, left; R, right.
Notably, across both subtype‐ and component‐level analyses, the ATR and IC were among the most consistently implicated tracts, suggesting that these regions may represent common sites of vulnerability associated with frailty. (Detailed estimates are provided in Supplementary Table 4.)
Discussion
In this study, we extend prior research by demonstrating that WM microstructural alterations are associated with overall frailty status and display distinct patterns across frailty subtypes and individual components. We identified tract‐specific WM changes in community‐dwelling older adults, using FA and MD metrics derived from DTI. Consistent with our hypothesis, frailty subtypes were differentially associated with WM alterations. The mobility subtype, primarily characterized by slowness, showed widespread FA and MD changes predominantly in major association and interhemispheric fibers relevant to motor coordination. In contrast, the non‐mobility subtype exhibited a more selective pattern, primarily involving projection and limbic‐related tracts such as the ATR, IC, cingulum, and forceps major. Our findings underscore the heterogeneous nature of frailty‐related brain changes and highlight the need for subtype‐specific approaches in identifying neural targets for early detection and intervention in aging populations.
Distinct Association of Mobility Frailty Subtype with White Matter Tracts
Among the tracts associated with the mobility subtype, notable involvement was observed in the genu and splenium of the CC, as well as the MCP. The CC plays a key role in interhemispheric communication. Its anterior segment, the genu, has been linked to executive function, processing speed, and gait performance in older adults. 28 , 29 , 30 The splenium, which connects posterior cortical regions involved in visual processing and memory, has been associated with verbal fluency and global cognition. 28 , 31 A longitudinal study found greater decline in global cognition scores for older adults with splenium FA reduction compared to those with preserved WM. 31 Alterations in these segments may help explain the reduced motor function and cognitive impairment commonly observed in individuals with mobility‐related frailty.
At the component level, mobility‐related components, particularly slowness, were also associated with widespread alterations in major association fibers, including the SLF, ILF, and IFOF. These tracts support frontoparietal and fronto‐occipital connectivity, which is important for both motor control and cognitive performance. 29 , 31 , 32 Disruptions in these pathways have previously been linked to reduced muscle strength, slower processing speed, and executive function impairment. 29 , 31 , 33 Collectively, the findings suggest that widespread alterations in key WM tracts likely contribute to the poor performance observed across all cognitive domains in individuals with the mobility subtype in our study.
Our findings also extend prior structural MRI research showing cerebellar GM atrophy in the mobility subtype. 6 The association we observed between the mobility subtype and WM changes in the MCP, a major cerebellar projection tract, builds on this work by highlighting corresponding disruptions at the WM level. In summary, these findings suggest that disruptions in both interhemispheric and cerebellar WM pathways may jointly contribute to the neural mechanisms underlying mobility‐related frailty.
Distinct Association of Non‐Mobility Frailty Subtype with White Matter Tracts
The non‐mobility subtype showed a more restricted and selective pattern of WM alterations, with prominent involvement of projection and limbic‐related tracts such as the ATR, IC, cingulum, and forceps major. Among these, the cingulum includes dorsal (cingulate gyrus) and ventral (hippocampal) segments, connecting key limbic regions involved in memory and emotional regulation. 34 Alterations in the cingulate segment have been previously reported in frailty, 35 while reduced FA in the hippocampal cingulum has been linked to higher amyloid burden and poorer working memory in older adults. 36 The forceps major, connecting the occipital lobes, has also been implicated in visual processing and emotional regulation. 37 Collectively, these findings suggest that non‐mobility frailty may involve distinct limbic–subcortical pathways, in contrast to the pattern observed in the mobility subtype.
At the component level, however, marked differences emerged between exhaustion and weight loss. Exhaustion was associated with FA and MD alterations in only a limited set of tracts, most notably in the ATR and splenium of the CC. In contrast, weight loss was linked to more extensive changes across a broader range of regions. Although both belong to the same frailty subtype, this divergence may indicate that exhaustion and weight loss reflect distinct stages or mechanisms along the frailty trajectory. While our data are cross‐sectional, prior studies have suggested that exhaustion often appears earlier, whereas weight loss may signal more advanced systemic deterioration. 2 Moreover, some overlap in WM involvement was also observed between weight loss and slowness, which may reflect shared structural vulnerabilities across frailty components.
In summary, WM alterations in non‐mobility frailty differ from those in mobility subtypes and show distinct patterns across individual components. These findings highlight the subtype's heterogeneity and the need to consider temporal dynamics when investigating neural mechanisms and designing future assessments in frailty‐related brain aging.
Lack of Association of Low Physical Activity Subtype with WM Alterations
The absence of significant associations between the low physical activity subtype and WM alterations may be due to insufficient physical activity levels among participants producing detectable differences. Previous studies suggest that positive effects on WM integrity may require higher intensity or longer duration of physical activity. 38 Additionally, age‐related declines in WM alterations might overshadow the subtle effects of low physical activity levels in older adults.
Clinical implications
Our findings highlight important clinical implications. Firstly, the distinct WM patterns observed across frailty subtypes highlight the need for subtype‐specific and individualized approaches to assessment and intervention. For example, targeting association fiber alterations in mobility subtypes, or limbic and projection pathway disruptions in non‐mobility subtypes, may help tailor strategies to specific structural targets. This individualized framework may enhance early detection and prevent progression to more severe frailty or cognitive impairment.
In addition, the ATR and IC emerged consistently across subtypes and components, suggesting these tracts may reflect common structural vulnerabilities. Both are involved in motor‐sensory integration, executive function, and thalamocortical communication, and play a significant role in both frailty and cognitive impairment. 12 , 33 , 39 Monitoring microstructural changes in these tracts may therefore help identify individuals at elevated risk of frailty and cognitive decline.
Therefore, our work supports the use of tract‐specific imaging markers to enhance early recognition of frailty‐related brain changes and to inform more personalized and effective clinical strategies.
Strengths and Limitations
This study is the first to comprehensively examine the distinct associations between frailty subtypes and WM alterations using DTI in community‐dwelling older adults. By analyzing specific subtypes rather than treating frailty as a unified construct, we revealed meaningful differences in how these subtypes relate to brain structure and cognitive function. Additionally, using advanced neuroimaging techniques enabled the detection of early microstructural changes across a wide range of WM tracts, offering novel insights into the neural mechanisms underlying frailty‐related brain aging. Several limitations of our study should be acknowledged. Firstly, the cross‐sectional design prevents causal inferences, and reverse causality cannot be excluded. In particular, complex interactions may exist among depression, brain microstructure, and frailty. While we adjusted for depressive symptoms in our statistical models, we cannot clarify potential causal pathways within the current cross‐sectional design. Longitudinal studies are needed to establish the temporal relationships between frailty subtypes and WM alterations. Secondly, the relatively small number of frail participants in our sample may limit the generalizability of findings, particularly for detecting structural changes specific to frailty. To preserve statistical power, we combined frail and pre‐frail individuals; however, this approach may have attenuated associations that are more pronounced in the frail subgroup. Future studies with larger and more diverse samples, including sufficient numbers of frail individuals, are needed to determine whether distinct WM alterations are uniquely associated with frailty and how these patterns evolve across the frailty continuum.
Conclusions
By distinguishing between mobility and non‐mobility subtypes, we identified unique WM alterations in brain regions involved in motor coordination, cognition, and emotional regulation. These findings highlight the heterogeneous nature of frailty and suggest that different subtypes may shape brain aging through distinct mechanisms. Clarifying these subtype‐specific WM patterns supports the need for personalized strategies to prevent cognitive decline and promote brain health in older adults. Future longitudinal studies are warranted to validate these associations and evaluate the impact of subtype‐specific interventions on WM alterations.
Funding Information
This study was supported by the National Science and Technology Council, previously the Ministry of Science and Technology, Taiwan (NSTC 113‐2321‐B‐418‐003, 112‐2314‐B‐A49‐047‐MY3, and MOST 106‐2628‐B‐010), and by internal funding from the School of Medicine, National Yang Ming Chiao Tung University, supported in part by the Yin Yen‐Liang Foundation Development and Construction Plan.
Disclosure statement
The authors declare that they have no known competing financial or non‐financial interests or personal relationships that are directly or indirectly related to the work reported in this paper.
Author contributions
Chen‐Hua Lin: Writing – review & editing, Writing – original draft, Visualization, Formal analysis, Conceptualization. Yah‐Ting Wu: Writing – original draft, Formal analysis. Jun‐Ying Wei: Writing – original draft, Formal analysis. Yao‐Chia Shih: Methodology. Yi‐ Ping Chao: Methodology. Yen‐Jun Lai: Data curation. Yen‐Ling Chiu: Funding acquisition, Data curation, Conceptualization. Yi‐Fang Chuang: Writing – review & editing, Writing – original draft, supervision, Methodology, Conceptualization. All authors contributed to the manuscript revision and approved the final submitted version.
Supporting information
Supplementary Table S1. Justification for the selection of white matter tracts included in this study.
Supplementary Table S2. Associations between frailty subtypes and cognitive function.
Supplementary Table S3. Associations between individual frailty components and cognitive function.
Supplementary Table S4. Associations between frailty components and diffusion metrics of white matter tracts.
Acknowledgments
We appreciate all participants' contributions to the Epidemiology of Mild Cognitive Impairment in Taiwan (EMCIT) and the Taiwan Precision Medicine Initiative on Cognitive Impairment and Dementia (TPMIC) study.
Data availability statement
The data that support the findings of this study are available from the corresponding author upon reasonable request.
References
- 1. Fried LP, Tangen CM, Walston J et al. Frailty in older adults: Evidence for a phenotype. J. Gerontol. A Biol. Sci. Med. Sci. 2001; 56: M146–M156. [DOI] [PubMed] [Google Scholar]
- 2. Stenholm S, Ferrucci L, Vahtera J et al. Natural course of frailty components in people who develop frailty syndrome: Evidence from two cohort studies. J. Gerontol. A Biol. Sci. Med. Sci. 2019; 74: 667–674. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 3. Xue QL, Bandeen‐Roche K, Varadhan R, Zhou J, Fried LP. Initial manifestations of frailty criteria and the development of frailty phenotype in the Women's health and aging study II. J. Gerontol. A Biol. Sci. Med. Sci. 2008; 63: 984–990. [DOI] [PubMed] [Google Scholar]
- 4. Robertson DA, Savva GM, Kenny RA. Frailty and cognitive impairment: A review of the evidence and causal mechanisms. Ageing Res. Rev. 2013; 12: 840–851. [DOI] [PubMed] [Google Scholar]
- 5. Buchman AS, Boyle PA, Wilson RS, Tang Y, Bennett DA. Frailty is associated with incident Alzheimer's disease and cognitive decline in the elderly. Psychosom. Med. 2007; 69: 483–489. [DOI] [PubMed] [Google Scholar]
- 6. Chen WT, Chou KH, Liu LK et al. Reduced cerebellar gray matter is a neural signature of physical frailty. Hum. Brain Mapp. 2015; 36: 3666–3676. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 7. Del Brutto OH, Mera RM, Cagino K et al. Neuroimaging signatures of frailty: A population‐based study in community‐dwelling older adults (the Atahualpa project). Geriatr. Gerontol. Int. 2017; 17: 270–276. [DOI] [PubMed] [Google Scholar]
- 8. Kant IMJ, de Bresser J, van Montfort SJT et al. The association between brain volume, cortical brain infarcts, and physical frailty. Neurobiol. Aging 2018; 70: 247–253. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 9. Chung CP, Chou KH, Chen WT et al. Cerebral microbleeds are associated with physical frailty: A community‐based study. Neurobiol. Aging 2016; 44: 143–150. [DOI] [PubMed] [Google Scholar]
- 10. Maltais M, de Souto Barreto P, Hooper C, Payoux P, Rolland Y, Vellas B. Association between brain β‐amyloid and frailty in older adults. J. Gerontol. A Biol. Sci. Med. Sci. 2019; 74: 1747–1752. [DOI] [PubMed] [Google Scholar]
- 11. Ducca EL, Gomez GT, Palta P et al. Physical frailty and brain white matter abnormalities: The atherosclerosis risk in communities study. J. Gerontol. A Biol. Sci. Med. Sci. 2023; 78: 357–364. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 12. Maltais M, de Souto BP, Perus L et al. Prospective associations between diffusion tensor imaging parameters and frailty in older adults. J. Am. Geriatr. Soc. 2020; 68: 1050–1055. [DOI] [PubMed] [Google Scholar]
- 13. Pelletier A, Periot O, Dilharreguy B et al. Age‐related modifications of diffusion tensor imaging parameters and white matter hyperintensities as inter‐dependent processes. Front. Aging Neurosci. 2015; 7: 255. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 14. López‐Sanz D, Suárez‐Méndez I, Bernabé R et al. Scoping review of neuroimaging studies investigating frailty and frailty components. Front. Med. (Lausanne) 2018; 5: 284. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 15. Chuang YF, Fan KC, Su YY et al. Precision probiotics supplement strategy in aging population based on gut microbiome composition. Brief. Bioinform. 2024; 25: bbae351. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 16. Chen LJ, Chen CY, Lue BH, Tseng MY, Wu SC. Prevalence and associated factors of frailty among elderly people in Taiwan. Int. J. Gerontol. 2014; 8: 114–119. [Google Scholar]
- 17. Liu LK, Guo CY, Lee WJ et al. Subtypes of physical frailty: Latent class analysis and associations with clinical characteristics and outcomes. Sci. Rep. 2017; 7: 46417. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 18. Chuang YF, Liu YC, Tseng HY et al. Urban‐rural differences in the prevalence and correlates of mild cognitive impairment in community‐dwelling older adults in Taiwan: The EMCIT study. J. Formos. Med. Assoc. 2021; 120: 1749–1757. [DOI] [PubMed] [Google Scholar]
- 19. Cui Z, Zhong S, Xu P, He Y, Gong G. PANDA: A pipeline toolbox for analyzing brain diffusion images. Front. Hum. Neurosci. 2013; 7: 42. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 20. Mori S, Oishi K, Jiang H et al. Stereotaxic white matter atlas based on diffusion tensor imaging in an ICBM template. Neuroimage 2008; 40: 570–582. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 21. Hua K, Zhang J, Wakana S et al. Tract probability maps in stereotaxic spaces: Analyses of white matter anatomy and tract‐specific quantification. Neuroimage 2008; 39: 336–347. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 22. Fortin JP, Parker D, Tunç B et al. Harmonization of multi‐site diffusion tensor imaging data. Neuroimage 2017; 161: 149–170. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 23. Shaikh S, Kumar A, Bansal A. Diffusion tensor imaging: An overview. Neurol. India 2018; 66: 1603–1611. [DOI] [PubMed] [Google Scholar]
- 24. Madden DJ, Bennett IJ, Burzynska A, Potter GG, Chen NK, Song AW. Diffusion tensor imaging of cerebral white matter integrity in cognitive aging. Biochim. Biophys. Acta 2012; 1822: 386–400. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 25. Winston GP. The physical and biological basis of quantitative parameters derived from diffusion MRI. Quant. Imaging Med. Surg. 2012; 2: 254–265. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 26. Soares JM, Marques P, Alves V, Sousa N. A hitchhiker's guide to diffusion tensor imaging. Front. Neurosci. 2013; 7: 31. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 27. Jellison BJ, Field AS, Medow J, Lazar M, Salamat MS, Alexander AL. Diffusion tensor imaging of cerebral white matter: A pictorial review of physics, fiber tract anatomy, and tumor imaging patterns. AJNR Am. J. Neuroradiol. 2004; 25: 356–369. [PMC free article] [PubMed] [Google Scholar]
- 28. Ghanavati T, Smitt MS, Lord SR et al. Deep white matter hyperintensities, microstructural integrity and dual task walking in older people. Brain Imaging Behav. 2018; 12: 1488–1496. [DOI] [PubMed] [Google Scholar]
- 29. Hoagey DA, Lazarus LTT, Rodrigue KM, Kennedy KM. The effect of vascular health factors on white matter microstructure mediates age‐related differences in executive function performance. Cortex 2021; 141: 403–420. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 30. Kerchner GA, Racine CA, Hale S et al. Cognitive processing speed in older adults: Relationship with white matter integrity. PLoS One 2012; 7: e50425. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 31. Shafer AT, Williams OA, Perez E et al. Accelerated decline in white matter microstructure in subsequently impaired older adults and its relationship with cognitive decline. Brain Commun. 2022; 4: fcac051. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 32. Herbet G, Zemmoura I, Duffau H. Functional anatomy of the inferior longitudinal fasciculus: From historical reports to current hypotheses. Front. Neuroanat. 2018; 12: 77. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 33. Tamura Y, Shimoji K, Ishikawa J et al. Association between white matter alterations on diffusion tensor imaging and incidence of frailty in older adults with cardiometabolic diseases. Front. Aging Neurosci. 2022; 14: 912972. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 34. Kollenburg L, Arnts H, Green A, Strauss I, Vinke S, Kurt E. The cingulum: Anatomy, connectivity and what goes beyond. Brain Commun. 2025; 7: fcaf048. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 35. Gutiérrez‐Zúñiga R, Davis JRC, Ruddy K et al. Structural brain signatures of frailty, defined as accumulation of self‐reported health deficits in older adults. Front. Aging Neurosci. 2023; 15: 1065191. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 36. Hirschfeld LR, Deardorff R, Chumin EJ et al. White matter integrity is associated with cognition and amyloid burden in older adult Koreans along the Alzheimer's disease continuum. Alzheimer's Res Ther 2023; 15: 218. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 37. Jenkins LM, Barba A, Campbell M et al. Shared white matter alterations across emotional disorders: A voxel‐based meta‐analysis of fractional anisotropy. Neuroimage Clin. 2016; 12: 1022–1034. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 38. Maleki S, Hendrikse J, Chye Y et al. Associations of cardiorespiratory fitness and exercise with brain white matter in healthy adults: A systematic review and meta‐analysis. Brain Imaging Behav. 2022; 16: 2402–2425. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 39. Tamura Y, Shimoji K, Ishikawa J et al. Associations between sarcopenia and white matter alterations in older adults with diabetes mellitus: A diffusion tensor imaging study. J. Diabetes Investig. 2021; 12: 633–640. [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
Supplementary Table S1. Justification for the selection of white matter tracts included in this study.
Supplementary Table S2. Associations between frailty subtypes and cognitive function.
Supplementary Table S3. Associations between individual frailty components and cognitive function.
Supplementary Table S4. Associations between frailty components and diffusion metrics of white matter tracts.
Data Availability Statement
The data that support the findings of this study are available from the corresponding author upon reasonable request.