Temporalis muscle biomarkers from routine brain MRI and risk of dementia in two independent cohorts.
Authors
Affiliations (9)
Affiliations (9)
- Russell H. Morgan Department of Radiology and Radiological Science, Johns Hopkins University School of Medicine, Baltimore, Maryland, USA.
- Optimal Aging Institute, Department of Population Health, New York University Grossman School of Medicine, New York, New York, USA.
- Wake Forest University School of Medicine, Wake Forest University, Winston-Salem, North Carolina, USA.
- Department of Biostatistics, Johns Hopkins Bloomberg School of Public Health, Baltimore, Maryland, USA.
- Mallinckrodt Institute of Radiology, Washington University in St. Louis, Saint Louis, Missouri, USA.
- Department of Radiology, University of California San Francisco, San Francisco, California, USA.
- Translational Gerontology Branch, National Institute on Aging Intramural Research Program, Baltimore, Maryland, USA.
- Department of Epidemiology, Johns Hopkins Bloomberg School of Public Health, Baltimore, Maryland, USA.
- Department of Neurology, Johns Hopkins University School of Medicine, Baltimore, Maryland, USA.
Abstract
Skeletal muscle loss is associated with cognitive decline, but whether neuroimaging-derived muscle characteristics predict incident dementia remains unclear. We evaluated associations of deep learning-derived temporalis muscle (TM) cross-sectional area (CSA) and radiomic texture features from baseline T1-weighted magnetic resonance imaging (MRI) with incident dementia in dementia-free participants from the Alzheimer's Disease Neuroimaging Initiative (ADNI) (n = 750) and the Atherosclerosis Risk in Communities (ARIC) study (n = 532). TM was segmented using a convolutional neural network trained in ADNI and externally validated in ARIC. Radiomic features were reduced using least absolute shrinkage and selection operator-penalized Cox models to generate a composite score. Multivariable Cox regression adjusted for demographics, apolipoprotein E ε4, baseline cognition, body mass index, and physical performance. Higher TM radiomic scores were associated with increased dementia risk in ADNI (hazard ratio per SD, 1.32) and ARIC (1.64). Smaller TM CSA predicted dementia in ADNI but not ARIC. TM texture patterns from routine brain MRI are associated with dementia risk, supporting TM phenotyping as a scalable marker of systemic biological vulnerability.