Plasma proteomics of sleep traits reveals systemic immune-metabolic pathways and genetically prioritized proteins.

The plasma proteomic signatures of sleep disturbance remain poorly characterized. Using data from 43,709 predominantly European-ancestry, middle-aged and older UK Biobank participants, we depict a large-scale atlas of plasma proteomic signatures of seven self-reported sleep traits (sleep duration, chronotype, insomnia symptoms, daytime napping, daytime sleepiness, snoring, and ease of getting up in the morning) and a derived sleep health score. We identify 935 proteins associated with at least one sleep trait, converging on lipid metabolism, immune function and inflammation, cell adhesion, and neurochemical signaling. Leveraging genomic structural equation modeling to define three latent sleep factors, namely circadian preference, daytime sleep burden, and nighttime sleep adequacy, bidirectional Mendelian randomization (MR) identifies one protein (LTA) with robust cis-instrument and strong colocalization support (PP.H4 = 0.98) for a putative causal effect on nighttime sleep adequacy. Sixteen additional genetically supported candidate proteins rely primarily on trans-pQTL instruments or weaker colocalization. These genetically supported candidates are prospectively associated with incident cardiovascular disease, stroke, type 2 diabetes, dementia, chronic kidney disease, depression, and mortality over a median 13.6-year follow-up, with the strongest per-SD hazard ratio (HR) associations observed for chronic kidney disease (e.g., BTN2A1: HR = 2.33) and type 2 diabetes (e.g., RBP5: HR = 1.58). Collectively, these findings highlight the potential of large-scale proteomics in elucidating sleep pathogenesis, and generate testable hypotheses for validation in independent cohorts and experimental models.