Alzheimer's Imaging Consortium.

Area of Science:

• Gerontology
• Proteomics
• Biomarkers

Background:

• Multi-organ aging and disease modeling is a growing research area.
• Plasma proteomics is established for predicting biological age (proteome-based biological age gap, ProtBAG).
• Cerebrospinal fluid (CSF) proteomics offers a novel window into organ-specific aging.

Purpose of the Study:

• To derive 11 organ-specific ProtBAGs using CSF proteomics data from the Alzheimer's Disease Neuroimaging Initiative (ADNI) study.
• To evaluate the performance of two machine learning models in predicting organ-specific biological age.
• To expand the multi-organ aging clock framework using CSF biomarkers.

Main Methods:

• CSF proteomics data from 736 ADNI participants were analyzed using the SomaScan 7k platform.
• Organ-enriched proteins were identified, and 11 organ-specific ProtBAGs were developed using Linear Support Vector Regression (SVR) and LASSO regression.
• Model performance was assessed using nested random holdout cross-validation, Mean Absolute Error (MAE), and Pearson's r.

Main Results:

• Imputation of missing proteomics values with a 2% missing rate yielded the best model performance (r²=0.54).
• Linear SVR and LASSO regression models demonstrated comparable performance across the 11 organ systems.
• The brain and hepatic ProtBAGs exhibited the lowest MAE and highest Pearson's r values, indicating superior accuracy in predicting biological age.
• MAE values ranged from 4.5 to 6, consistent with prior brain imaging studies.
• Endocrine and reproductive system ProtBAGs showed relatively lower model performance.

Conclusions:

• This study successfully developed 11 organ-specific ProtBAGs from CSF proteomics, enhancing the existing multi-organ aging clock framework.
• The findings highlight the potential of CSF proteomics for assessing organ-specific aging.
• Future research will explore the links between these novel ProtBAGs, cognitive function, and Alzheimer's disease progression.