Biomarkers.

Area of Science:

• Neuroimaging and Proteomics
• Biomarker Discovery
• Machine Learning in Medicine

Background:

• White matter lesions (WMLs) are common in aging populations, linked to vascular risk factors but with poorly understood molecular mechanisms.
• Individual WML burden varies significantly, necessitating advanced diagnostic tools beyond traditional risk factor assessment.
• Plasma proteomics offers a non-invasive approach to identify biomarkers for WMLs.

Purpose of the Study:

• To develop and validate a molecular model using plasma proteomics and clinical data to assess WML severity.
• To explore the underlying molecular mechanisms of WMLs through integrated proteomic and clinical analysis.
• To establish a reliable, non-invasive tool for evaluating WMLs.

Main Methods:

• Two cohorts (n=181 and n=78) were used, classifying participants into mild and severe WML groups based on MRI.
• Plasma proteomic profiling (HPLC-MS/MS) identified differentially expressed proteins, followed by functional and pathway analyses.
• A predictive model integrating key plasma proteins and cognitive scores was developed using LASSO regression and validated in both cohorts.

Main Results:

• The severe WML group exhibited significantly poorer cognitive performance.
• 123 differentially expressed plasma proteins were identified, enriched in pathways related to vascular diseases and metabolism.
• A validated model incorporating four proteins (ABI1, ASS1, CXCL12, EIF2S1) and cognitive scores achieved high discriminatory performance (AUC 0.845 internal, 0.840 external).

Conclusions:

• A robust, non-invasive plasma proteomics-based model for assessing WML severity was successfully developed and validated.
• The model demonstrates high sensitivity, specificity, and stability across independent cohorts.
• Findings provide insights into WML pathogenesis and support personalized diagnosis and treatment strategies.