Abstract
Abstract: Biomarkers have gained an increased importance in the past years in helping physicians to diagnose Alzheimer's disease (AD). This study was designed to identify a blood-based, transcriptomic signature that can differentiate AD patients from control subjects. The performance of the signature was then evaluated for robustness in an independent blinded sample population. RNA was extracted from 177 blood samples (90 AD patients and 87 controls) and gene expression profiles were generated using the human Genome-Wide Splice Array™. These profiles were used to establish a signature to differentiate AD patients from controls. Subsequently, prediction results were optimized by establishing grey zone boundaries that discount prediction scores near the disease status threshold. Signature validation was then performed on a blinded independent cohort of 209 individuals (111 AD and 98 controls). The AclarusDx™ signature consists of 170 probesets which map to 136 annotated genes, a significant number of which are associated with inflammatory, gene expression, and cell death pathways. Additional signature genes are known to interact with pathways involved in amyloid and tau metabolism. The validation sample set, after removal of 45 individuals with prediction profile scores within the grey zone, consisted of 164 subjects. The AclarusDx™ performance on this validation cohort had a sensitivity of 81.3% (95% CI: [73.3%; 89.3%]); and a specificity of 67.1% (95% CI: [56.3%; 77.9%]). AclarusDx™ is a non-invasive blood-based transcriptomic test that, in combination with standard assessments, can provide physicians with objective information to support the diagnosis of AD.