Integrated machine learning identifies biomarkers for bilirubin-induced Alzheimer’s disease-like lesions in neonates and adults

Neurological impairments resulting from bilirubin encephalopathy represent a hallmark of bilirubin's neurotoxic effects. Earlier research suggests that bilirubin may contribute to Alzheimer's disease (AD) pathology by inducing neuronal necrosis and abnormal tau phosphorylation. Nevertheless, the precise mechanisms linking bilirubin to neurodegeneration in both neonates and adults remain unclear. To address this, we established two complementary models: a pathological jaundice model using UGT1A1^⁻/⁻ neonatal mice and an adult bilirubin exposure model via lateral ventricle injection, followed by brain tissue sequencing. Integrating bioinformatics analyses with multiple AD datasets, we uncovered regulatory effects of bilirubin exposure on neurodegenerative processes across age groups. Machine learning approaches identified two key genes, BCL2-binding component 3 (Bbc3) and Mitogen-activated protein kinase kinase kinase 10 (Map3k10), as central mediators of bilirubin-induced neuroinjury in pathological jaundice. In adults, bilirubin exposure also promoted neuroinflammation. Notably, effector memory CD8⁺ T cells emerged as critical drivers of AD-associated neuroinflammation, and four additional biomarkers were identified to construct a high-performance diagnostic model. Together, these findings highlight potential biomarkers for diagnosing and monitoring bilirubin-induced neurological damage and provide a basis for developing targeted diagnostic and therapeutic strategies for AD.