Blood Biomarkers Predict Conversion from Cognitively Stable to Mild Cognitive Impairment or Alzheimer's Disease in Down Syndrome at 16-Month Follow-Up in ABC-DS.

INTRODUCTION:Individuals with Down syndrome (DS) face high risk for Alzheimer's disease (AD), yet presymptomatic detection of cognitive decline is hindered by lifelong intellectual disability. METHODS:Using data from the Alzheimer's Biomarker Consortium-Down Syndrome (ABC-DS), blood samples from 246 participants were analyzed, yielding 404 longitudinal observations (45 Converters, 359 Stable) collected at 0, 16, and 32 months were analyzed. A Support Vector Machine was trained on 25 plasma biomarkers spanning neurodegeneration, inflammation, and vascular health, along with demographic factors (age, sex, ethnicity, karyotype, apolipoprotein E [APOE ε4]). Batch-effect correction and feature selection were applied, resulting in 13 key markers. RESULTS:The refined model achieved 92.4% sensitivity, 59.9% specificity, and an area under the curve (AUC) of 77.9%, accurately identifying individuals at risk of cognitive decline up to 16 months before clinical progression. DISCUSSION:This multi-domain, blood-based machine learning approach demonstrates that plasma biomarkers are valuable non-invasive tools for early detection and risk stratification of cognitive decline in DS.