Large-scale cross-species proteomics and machine learning identify biomarkers and peripheral therapeutic targets for delirium
Year: 2026
Type: preprint
Abstract: Delirium is a frequent neurocognitive disorder in critically ill patients. We aimed to identify circulating biomarkers and therapeutic targets for delirium. We conducted a multi-tiered study integrating: (1) UK Biobank plasma proteomics (52,936 participants, 2,911 proteins, 17.5-year follow-up) with machine learning for risk prediction; (2) two-sample Mendelian randomization using protein QTL (N = 35,559) and delirium GWAS (4,381 cases vs. 469,981 controls) for causal inference; (3) cross-species proteomic profiling in human and mouse postoperative delirium (POD) cohorts; and (4) functional validation in POD mice via liver-specific knockdown of candidate proteins or systemic supplementation, with cognitive tests, histopathology, and cytokine assays. We identified 1,470 plasma proteins associated with incident delirium. A 21-feature predictive panel (20 proteins + age) achieved consistent performance (AUC 0.77–0.82) for delirium risk identification. Cross-species proteomic profiling converged on three peripheral proteins C2 and SPON2 (risk protein) and APLP1 (protective protein) which were validated across human POD, mouse POD, and UK Biobank cohorts. In POD mice, liver-specific knockdown of C2 or SPON2, as well as systemic supplementation of recombinant human APLP1, significantly ameliorated cognitive deficits, attenuated hippocampal neuroinflammation, reduced glial activation, and preserved synaptic integrity, without overt organ toxicity. Our findings identify a series of peripheral blood biomarkers for delirium risk prediction, and demonstrate that peripheral modulation of circulating C2, SPON2, and APLP1 confers therapeutic benefit. This study highlights the peripheral-to-central axis as a critical node in delirium risk and provides insight into the etiology and therapeutic strategies in delirium patients.
Source: Research Square
Institutions: Tongji University, South China University of Technology
Language: English
Field: Medicine
Domain: Health Sciences
Open Access status: green
OpenAlex ID: W7225757980