Biomarkers

Area of Science:

• Neuroscience
• Biomarker Discovery
• Computational Biology

Background:

• Alzheimer's disease (AD) is characterized by amyloid-beta plaques (A), neurofibrillary tangles (T), and neuronal loss (N), collectively known as A/T/N.
• Understanding the spatial progression of neurodegeneration is crucial for predicting AD trajectories and outcomes.

Purpose of the Study:

• To explore the staging and pseudotime of AD patients using A/T/N biomarkers.
• To utilize Positron Emission Tomography (PET) and Magnetic Resonance Imaging (MRI) data from the Alzheimer's Disease Neuroimaging Initiative (ADNI).

Main Methods:

• Applied PHATE for dimensionality reduction to visualize disease progression trajectories for A/T/N modalities.
• Used Slingshot to translate PHATE embeddings into 1D pseudotime values.
• Employed SuStaIn, a machine learning algorithm, to predict patient stages and biomarker sequences.
• Compared SuStaIn stage predictions with PHATE/Slingshot pseudotime values to assess robustness.

Main Results:

• SuStaIn predicted stages closely aligned with PHATE/Slingshot pseudotime values across all modalities.
• Amyloid PET and tau PET showed stronger trajectory alignments than MRI-based volume.
• Biomarker event sequences from SuStaIn agreed with PHATE pseudotime-informed models.
• Amyloid and tau modalities exhibited moderate pseudotime correlation but distinct driving biomarker events.

Conclusions:

• Integrative analysis of computational methods enhances confidence in disease progression timing.
• Amyloid and tau biomarkers appear to follow distinct cortical trajectories.
• Different computational approaches independently yielded similar findings regarding disease progression.