Alzheimer’s Disease: Biomarkers and Progression

Biomarker-informed prediction and trial methodology for Alzheimer’s disease and mild cognitive impairment

2026-06-07 18:16 PDT

Overview

Plasma and cerebrospinal fluid biomarkers have transformed Alzheimer’s disease research: p-tau 217, amyloid-beta 42/40, GFAP, and NfL now provide biological staging that supplements and in some cases supplants clinical rating scales as trial endpoints and eligibility criteria. This program examines biomarker-informed prediction of disease progression, the statistical modeling of biomarker trajectories, and the design of trials that exploit biomarker stratification to improve power.

The empirical foundation is access to data from the Alzheimer’s Disease Cooperative Study (ADCS) – a network of more than 70 research sites that has conducted landmark trials of donepezil, vitamin E, and other agents – together with the Alzheimer’s Disease Neuroimaging Initiative (ADNI) and the National Alzheimer’s Coordinating Center (NACC).

Current work

  • p-tau 217 in MCI to AD progression. Reanalysis of the ADCS Mild Cognitive Impairment trial incorporating newly assayed plasma p-tau 217 (Lilly platform) from stored baseline samples, examining its prognostic value and APOE4 interaction for time-to-progression from MCI to clinical AD.

  • Medications and MCI-to-AD progression. Ensemble machine learning analysis (gradient boosting, random forests, LASSO) examining whether concomitant medication exposure at baseline predicts MCI-to-AD conversion, replicated across ADCS, ADNI, and NACC cohorts.

  • The age paradox in Alzheimer’s disease. AUROC-based cross-cohort analysis of why age – the strongest epidemiological risk factor for AD – fails to predict MCI-to-AD conversion in clinical cohorts. Analysis uses mediation, collider-bias reasoning, and nonlinear age splines across the ADCS and ADNI databases.

  • GFAP trajectories in blast-exposed veterans. Parametric and nonparametric analysis of plasma GFAP longitudinal trajectories in the prazosin blast-exposure RCT, comparing linear mixed models, robust regression, and permutation tests.

  • Alzheimer’s disease literature synthesis. A living, versioned review of recent AD research spanning biomarker development, prevention trials, and clinical outcomes, updated iteratively across the research cycle.

Methods

Survival analysis (Cox regression); time-to-event modeling with biomarker interaction terms; AUROC estimation and cross-cohort comparison; ensemble machine learning (gradient boosting, random forests); multiple imputation via mice; LASSO variable selection via glmnet; nonlinear spline effects; ggplot2-based visualization of AD-specific outcomes.

Software

  • zzlongplot – Longitudinal plotting for biomarker trajectories and clinical endpoints in AD trials.
  • zztable1 – Cohort descriptive tables for multi-site AD studies.

Publications

The full publications list filtered by alzheimers-disease or biomarkers provides the relevant record, which includes methodological and applied work conducted through ADCS spanning more than two decades.