Development of a serum-based microRNA panel for Alzheimer's disease diagnosis

Abstract

MicroRNAs (miRNAs) are emerging as promising blood-based biomarkers for Alzheimer's disease (AD) because of their stability and regulatory roles in disease-related pathways. This study aimed to develop a serum-based miRNA panel for AD diagnosis. Serum samples from 550 participants were categorized into discovery (85 AD, 65 healthy controls [HCs]), training (73 AD, 53 HCs), and validation (99 AD, 99 HCs, 36 vascular cognitive impairment [VCI], 40 dementia with Lewy bodies [DLB]) cohorts. Through small RNA sequencing and qPCR validation, we identified key miRNAs and constructed a diagnostic panel via machine learning. The 7-miRNA panel achieved area under the curve (AUC) values of 0.970 and 0.928 in the training and validation cohorts, respectively. The panel effectively differentiated AD from VCI (AUC = 0.951) and DLB (AUC = 0.851). Diagnostic performance remained robust across subgroups defined by age, gender, disease severity, and comorbidities (AUC 0.758-0.988). The risk score derived from the 7-miRNA panel was significantly associated with cognitive impairment (Mini-Mental State Examination [MMSE] score, r = -0.72) and plasma amyloid pathology biomarkers (Aβ42/40 ratio, r = -0.25; p-tau217, r = 0.36). A serum-based 7-miRNA panel, which offers a minimally invasive and accessible approach, has strong diagnostic potential for AD.