Organ-specific proteomic aging and cognitive performance: Implications for risk prediction of Alzheimer's disease and related dementias in older adults

Abstract

Background and objectives: Biological aging, characterized by cellular and molecular changes, may play a key role in neurodegenerative diseases. While recent proteomic advancements have introduced new aging clocks, widespread validation remains necessary. This study evaluated organ-specific and cognition-enriched proteomic clocks in relation to chronological age and cognitive change.

Methods: We analyzed plasma proteomic data from the CHARIOT PRO SubStudy (N = 409), measured using the SomaScan assay (version 4.1) at four time points over three years (months 0, 12, 24, and 36). Using published proteomic organ age weights, we calculated conventional, organ-specific, and cognition-enriched biological ages and compared them with chronological age. Adjusted multilevel regression analyses assessed associations between baseline proteomic AgeGaps (biological-chronological age differences) and cognitive performance over 54 months.

Results: The cohort (mean age: 71.8 ± 5.5 years; 50.1 % female) showed moderate to strong correlations between proteomic ages and chronological age (r = 0.37-0.80; MAE = 4.2-2.7). Over three years, AgeGaps increased across the conventional, organismal, muscle, liver, artery, and immune systems, ranging from 2.1 ± 1.9 to 1.0 ± 2.3 years. The artery AgeGap was most strongly associated with cognitive decline, with conventional and organismal AgeGaps showing similar patterns. Higher baseline AgeGap z-scores (i.e., greater biological age) in the artery and brain were associated with poorer cognition, as measured by the Repeatable Battery for the Assessment of Neuropsychological Status Total Scores (Coeff. -3.0, 95 % CI: -3.4, -2.5; and -1.1, 95 % CI: -1.5, -0.6) and the Preclinical Alzheimer's Cognitive Composite (Coeff. -0.5, 95 % CI: -0.6, -0.4; and -0.14, 95 % CI: -0.3, -0.03).

Conclusions: These findings highlight the interplay between neurological function and cardiovascular aging in cognitive decline. Organ-specific biological age assessments may aid in the early detection of age-related changes, informing personalized interventions. Our study underscores the importance of proteomic aging signatures in elucidating Alzheimer's disease mechanisms and other neurodegenerative conditions, advocating for an integrated approach to brain and cardiovascular health.

Keywords: Biological age; Cognitive changes; Longitudinal validation; Multilevel models; Organ-specific aging; Proteomics.

Fig. 1

(a) Organ-specific Biological Age. We estimated the 11 organ-specific aging models, along with conventional and organismal (non-organ-specific proteins) models, in the CHARIOT PRO SubStudy (N = 409) using plasma proteomic data measured over three years; (b) Cross-sectional predicted age versus chronological age (all months, N = 1274); (c) Change in predicted age versus change in chronological age over longitudinal follow-up (N = 384).

Fig. 2

Coefficient of the baseline AgeGap z-score regarding (a) RBANS total score (N = 409) and (b) PACC (N = 409), using multilevel mixed-effects linear regression with 500 bootstrap replications, adjusted for time point, chronological age, and sex; (c) RBANS (N = 400), additionally adjusted for Aβ, APOE ε4, family history of dementia, education, hypertension, BMI, hemoglobin, LDL, triglycerides, and sodium; and (d) PACC (N = 400), additionally adjusted for Aβ, APOE ε4, family history of dementia, education, marital status, BMI, hemoglobin, and LDL.

Fig. 3

Coefficient of the baseline AgeGap z-score associated with the decline in the RBANS total score for Aβ and APOE ε4 status. Models were adjusted for time point, chronological age, and sex. (a) and (b) represent subgroup analyses, while (c) and (d) present interaction term analyses with the baseline AgeGap z-score and Aβ+ or APOE ε4 carrier status.

Fig. 4

Coefficient of the baseline AgeGap z-score associated with the decline in the RBANS Language Index for (a) BMI$\ge$30, (b) hypertension, and (c) education level below a bachelor's degree. Models were adjusted for time point, chronological age, and sex.

Fig. 5

Coefficient of (a) ADCS-ADL-Participant and (b) ADCS-ADL-Study Partner, using multilevel mixed-effects linear regression with 500 bootstrap replications; coefficient of the baseline AgeGap z-score with respect to (c) ADCS-ADL-Participant and (d) ADCS-ADL-Study Partner, adjusted for time point, chronological age, and sex.