Abstract
Changes in β-amyloid (Aβ) and hyperphosphorylated tau (T) in brain and cerebrospinal fluid (CSF) precede Alzheimer's disease (AD) symptoms, making the CSF proteome a potential avenue to understand disease pathophysiology and facilitate reliable diagnostics and therapies. Using the AT framework and a three-stage study design (discovery, replication, and meta-analysis), we identified 2,173 analytes (2,029 unique proteins) dysregulated in AD. Of these, 865 (43%) were previously reported, and 1,164 (57%) are novel. The identified proteins cluster in four different pseudo-trajectories groups spanning the AD continuum and were enriched in pathways including neuronal death, apoptosis, and tau phosphorylation (early stages), microglia dysregulation and endolysosomal dysfunction (mid stages), brain plasticity and longevity (mid stages), and microglia-neuron crosstalk (late stages). Using machine learning, we created and validated highly accurate and replicable (area under the curve [AUC] > 0.90) models that predict AD biomarker positivity and clinical status. These models can also identify people that will convert to AD.