Objective:

Background:

Proteomics is enabling the characterization of brain aging biomarkers and the discernment of changes during brain aging at the molecular level.

Design/Methods:

Here, we leveraged multi-modal brain imaging data from 10,949 healthy adults in the UK Biobank and machine learning model to estimate brain age gap (BAG), which was generated by the deviation between brain age and chronological age and considered as an indicator of brain aging.

Results:

Plasma proteomics data of 2,922 proteins were further integrated and a protein-wide association analysis across 4,696 participants identified 13 proteins significantly associated with BAG (p < 1.71×10^-5), implicating stress, regeneration, and inflammation pathways. Brevican (BCAN: β = -0.838, p = 2.63×10^-10) and growth differentiation factor 15 (GDF15: β = 0.825, p = 3.48×10^-11) showed the most significant associations with BAG as well as multiple associations with dementia, stroke, depression, and movement functions. Additionally, dysregulation of BCAN influenced multiple cortical volumes, surface areas, and subcortical volumes. Mendelian randomization supported the causal association between plasma BCAN level and BAG (β = -0.473, p = 0.038). We revealed undulating changes in plasma proteome across brain aging in middle-aged adults, and further profiled brain age-related change peaks at 57, 70, and 78 years of life, implicating distinct biological pathways and implications during brain aging and suggesting pivotal periods to monitor and intervene brain aging.

Conclusions: