Assessment of gray matter features in reflecting advanced brain age in PCDS: A cross-national study
Poster No:
1149
Submission Type:
Abstract Submission
Authors:
Chen-Yuan Kuo1,2, Li-Ning Peng3,4, Chih-Ping Chung1,3, Pei-Lin Lee3, Liang-Kung Chen3,5, Ching-Po Lin2,3,6, Kun-Hsien Chou2,7
Institutions:
1Department of Neurology, Neurological Institute, Taipei Veterans General Hospital, Taipei, Taiwan, 2Institute of Neuroscience, National Yang Ming Chiao Tung University, Taipei, Taiwan, 3Center for Healthy Longevity and Aging Sciences, National Yang Ming Chiao Tung University, Taipei, Taiwan, 4Center for Geriatric and Gerontology, Taipei Veterans General Hospital, Taipei, Taiwan, 5Taipei Municipal Gan-Dau Hospital (managed by Taipei Veterans General Hospital), Taipei, Taiwan, 6Department of Education and Research, Taipei City Hospital, Taipei, Taiwan, 7Brain research center, National Yang Ming Chiao Tung University, Taipei, Taiwan
First Author:
Chen-Yuan Kuo
Department of Neurology, Neurological Institute, Taipei Veterans General Hospital|Institute of Neuroscience, National Yang Ming Chiao Tung University
Taipei, Taiwan|Taipei, Taiwan
Department of Neurology, Neurological Institute, Taipei Veterans General Hospital|Institute of Neuroscience, National Yang Ming Chiao Tung University
Taipei, Taiwan|Taipei, Taiwan
Co-Author(s):
Li-Ning Peng
Center for Healthy Longevity and Aging Sciences, National Yang Ming Chiao Tung University|Center for Geriatric and Gerontology, Taipei Veterans General Hospital
Taipei, Taiwan|Taipei, Taiwan
Center for Healthy Longevity and Aging Sciences, National Yang Ming Chiao Tung University|Center for Geriatric and Gerontology, Taipei Veterans General Hospital
Taipei, Taiwan|Taipei, Taiwan
Chih-Ping Chung
Department of Neurology, Neurological Institute, Taipei Veterans General Hospital|Center for Healthy Longevity and Aging Sciences, National Yang Ming Chiao Tung University
Taipei, Taiwan|Taipei, Taiwan
Department of Neurology, Neurological Institute, Taipei Veterans General Hospital|Center for Healthy Longevity and Aging Sciences, National Yang Ming Chiao Tung University
Taipei, Taiwan|Taipei, Taiwan
Pei-Lin Lee
Center for Healthy Longevity and Aging Sciences, National Yang Ming Chiao Tung University
Taipei, Taiwan
Center for Healthy Longevity and Aging Sciences, National Yang Ming Chiao Tung University
Taipei, Taiwan
Liang-Kung Chen
Center for Healthy Longevity and Aging Sciences, National Yang Ming Chiao Tung University|Taipei Municipal Gan-Dau Hospital (managed by Taipei Veterans General Hospital)
Taipei, Taiwan|Taipei, Taiwan
Center for Healthy Longevity and Aging Sciences, National Yang Ming Chiao Tung University|Taipei Municipal Gan-Dau Hospital (managed by Taipei Veterans General Hospital)
Taipei, Taiwan|Taipei, Taiwan
Ching-Po Lin
Institute of Neuroscience, National Yang Ming Chiao Tung University|Center for Healthy Longevity and Aging Sciences, National Yang Ming Chiao Tung University|Department of Education and Research, Taipei City Hospital
Taipei, Taiwan|Taipei, Taiwan|Taipei, Taiwan
Institute of Neuroscience, National Yang Ming Chiao Tung University|Center for Healthy Longevity and Aging Sciences, National Yang Ming Chiao Tung University|Department of Education and Research, Taipei City Hospital
Taipei, Taiwan|Taipei, Taiwan|Taipei, Taiwan
Kun-Hsien Chou
Institute of Neuroscience, National Yang Ming Chiao Tung University|Brain research center, National Yang Ming Chiao Tung University
Taipei, Taiwan|Taipei, Taiwan
Institute of Neuroscience, National Yang Ming Chiao Tung University|Brain research center, National Yang Ming Chiao Tung University
Taipei, Taiwan|Taipei, Taiwan
Introduction:
Neuroimaging-derived brain biological age serves as a widely accepted health indicator, reflecting the potential advancement or delay in brain aging compared to chronological age. Various gray matter (GM) features have been utilized for this purpose. However, the comparative sensitivity and feasibility of GM feature-derived brain age models in reflecting advanced brain aging within neurodegenerative disorders using multi-center, cross-national cohort datasets remain understudied. Physio-cognitive decline syndrome (PCDS), characterized by concurrent mobility and cognitive impairment without dementia, represents a preclinical phenotype of dementia syndrome in older individuals [1]. Our recent single-cohort study uncovered an advanced brain age in PCDS using a brain age model based on GM volume (GMV) [2]. This current study focuses on utilizing PCDS as the disease group to assess the feasibility of two GM feature-based brain age models, specifically GMV and GM density (GMD), across multi-center, cross-national studies.
Methods:
Participants of clinical datasets
Two cohort datasets, I-Lan Longitudinal Aging Study (ILAS) from Taiwan (N=1193, 566M/627F) and the National Institute for Longevity Sciences, Longitudinal Study of Aging (NILS-LSA) study from Japan (N=2302, 1173M/1129F), were used in this study. Using consensus criteria, participants aged 65 and above were categorized as robust or PCDS based on sex-stratified MMSE scores, grip strength, and walking speed in the bottom or top 20%. The study comprised 163 robust (age = 71.36±5.08 years, 103M/60F) and 103 PCDS (age = 74.68±6.19 years, 48M/55F) participants from ILAS, along with 137 robust (age = 70.57±4.08 years, 61M/76F) and 74 PCDS (age = 78.28±4.99 years, 43M/31F) participants from the NILS-LSA study.
Brain age model construction and evaluation
We gathered 1482 healthy individuals (age range: 18-92; 681M/801F) from five sites in Taiwan, training brain age prediction models with their T1-weighted MRI scans. Following our previous brain age model construction pipeline [3], we constructed two SVR-RBF models based on gray matter GMV and GMD respectively. Model performance was assessed using minimum mean absolute error (MAE) and maximum coefficient of determination (R2) between chronological and predicted age. The optimized brain age estimators were applied to ILAS and NILS-LSA datasets for individual predicted brain age and brain age gap (BAG) calculation, representing the difference between chronological age and predicted brain age.
Statistical analysis
Two ANCOVA models, incorporating age, age², sex, education years, and total intracranial volume as confounders, compared the BAG between robust and PCDS groups in each dataset. A significance level of <0.05 was considered statistically significant.
Two cohort datasets, I-Lan Longitudinal Aging Study (ILAS) from Taiwan (N=1193, 566M/627F) and the National Institute for Longevity Sciences, Longitudinal Study of Aging (NILS-LSA) study from Japan (N=2302, 1173M/1129F), were used in this study. Using consensus criteria, participants aged 65 and above were categorized as robust or PCDS based on sex-stratified MMSE scores, grip strength, and walking speed in the bottom or top 20%. The study comprised 163 robust (age = 71.36±5.08 years, 103M/60F) and 103 PCDS (age = 74.68±6.19 years, 48M/55F) participants from ILAS, along with 137 robust (age = 70.57±4.08 years, 61M/76F) and 74 PCDS (age = 78.28±4.99 years, 43M/31F) participants from the NILS-LSA study.
Brain age model construction and evaluation
We gathered 1482 healthy individuals (age range: 18-92; 681M/801F) from five sites in Taiwan, training brain age prediction models with their T1-weighted MRI scans. Following our previous brain age model construction pipeline [3], we constructed two SVR-RBF models based on gray matter GMV and GMD respectively. Model performance was assessed using minimum mean absolute error (MAE) and maximum coefficient of determination (R2) between chronological and predicted age. The optimized brain age estimators were applied to ILAS and NILS-LSA datasets for individual predicted brain age and brain age gap (BAG) calculation, representing the difference between chronological age and predicted brain age.
Statistical analysis
Two ANCOVA models, incorporating age, age², sex, education years, and total intracranial volume as confounders, compared the BAG between robust and PCDS groups in each dataset. A significance level of <0.05 was considered statistically significant.
Results:
Constructing two brain age estimators based on GM features, our findings revealed superior prediction performance from the GMD brain age estimator (MAE=5.33; R2=0.81) compared to the GMV brain age estimator (MAE=6.46; R2=0.73) within the training dataset (Figure 1). Moreover, the utilization of both the optimized GMV and GMD brain age estimators consistently demonstrated advanced brain age within the PCDS group across two distinct cohort datasets (ILAS: GMV, p=0.011, η2=0.025; GMD, p=0.01, η2=0.025; NILS-LSA: GMV, p=0.027, η2=0.024; GMD, p=0.029, η2=0.023) (Figure 2).
Conclusions:
In summary, our study reveals the intricate relationship between diverse GM features and the effectiveness of brain age estimators. While the GMD brain age estimator shows superior predictive performance during training, both GMD and GMV estimators indicate advanced brain age within the PCDS group during subsequent case-control comparisons. This underscores the potential of neuroimaging-derived brain age as a reliable indicator of advanced brain aging, particularly within the context of PCDS. These findings provide insights for similar multicenter, multinational cohort studies exploring brain health status across diseases.
Disorders of the Nervous System:
Neurodegenerative/ Late Life (eg. Parkinson’s, Alzheimer’s)
Lifespan Development:
Aging 1
Modeling and Analysis Methods:
Classification and Predictive Modeling 2
Novel Imaging Acquisition Methods:
Anatomical MRI
Keywords:
Aging
Machine Learning
MRI
STRUCTURAL MRI
Other - brain age, frailty, reproducibility
1|2Indicates the priority used for review
Provide references using author date format
2. Kuo CY, (2023), ‘Advanced brain age in community-dwelling population with combined physical and cognitive impairments’, Neurobiology of Aging, pp. 114-123.
3. Lee PL, (2022), ‘Regional rather than global brain age mediates cognitive function in cerebral small vessel disease’, Brain Commum., pp. 1-14.