Causal inference maps the sex-dimorphic genetic architecture and inflammatory-metabolic networks underpinning perceived aging in UK Biobank

Background
Self-perceived facial aging is an externally observable aging-related phenotype that may reflect both biological and psychosocial dimensions of aging. However, its genetic and molecular determinants, sex-specific heterogeneity, and relationship with systemic aging-related pathways remain incompletely understood.

Methods
Using multi-omics data from 435,078 UK Biobank participants, we performed bidirectional Mendelian randomization (MR), colocalization, and summary-data-based MR (SMR) integrating blood eQTLs, mQTLs, pQTLs (2,923 proteins), NMR metabolomics (256 traits), and clinical hematology to delineate causal molecular networks of perceived aging. We applied Differential Expression–Sliding Window Analysis (DE-SWAN) to evaluate lifespan dynamics and constructed a 23-feature Perceived Aging (PA) score validated against conventional biological clocks (PhenoAge and KDM-BA) and incident cardiovascular outcomes.

Results
We identified associated regulatory variants (eQTLs, pQTLs) and pinpointed a core network of nine inflammatory proteins (e.g., TNF, PD-L1) significantly associated with self-perceived facial aging. These inflammatory mediators causally influenced a distinct profile of 137 systemic aging-associated metabolites, predominantly lipids (e.g., LDL), revealing an inflammation-lipid metabolism axis driving self-perceived facial aging. Stratified analyses revealed distinct sex-specific trajectories: Low-density lipoprotein and phospholipid dysregulation predominated in midlife males (ages 55-65), whereas inflammatory signaling and lipid transport disruptions peaked in peri- and post-menopausal females (ages 45-55). Based on these observed associations, we developed a multi-omics PA score integrating immune-metabolic signatures. The PA score correlated with self-perceived facial aging and showed distinct patterns from conventional biological age indices, offering a complementary metric for facial aging assessment.

Conclusions
This study maps a sex-dimorphic genetic and immune-metabolic architecture underlying self-perceived facial aging in the UK Biobank, which support an inflammation–lipid metabolism axis as a molecular correlate of perceived aging and suggest that self-perceived facial aging may capture aspects of systemic aging that are partly distinct from conventional biological age measures.