(T-105) Generating data-driven insights from the Alzheimer’s Disease Neuroimaging Initiative (ADNI): Towards establishing a QSP model of Alzheimer’s disease

Angelia Wang, PhD – Scientist, Clinical Pharmacology & Pharmacometrics, Janssen R&D; Tao Peng, PhD – Senior Scientist, Janssen R&D; Lingjue Li, PhD – Senior Scientist, Janssen R&D; Weirong Wang, PhD – Senior Scientific Director, Janssen R&D; Damayanthi Devineni, PhD – Senior Director, Janssen R&D; Vivaswath Ayyar, PhD – Associate Director, Janssen R&D

Objectives: Alzheimer’s disease (AD) is a progressive neurodegenerative disease linked with cognitive decline. Two cardinal pathological hallmarks of AD include amyloid beta (Aβ) plaque buildup and hyperphosphorylated tau (ptau) protein, the latter well-correlating with cognitive decline. Recent published clinical trials on anti-Aβ therapies demonstrate reduction in both Aβ and tau, suggesting a mechanistic interaction between Aβ and tau pathologies. The Alzheimer’s Disease Neuroimaging Initiative (ADNI) provides expansive, publicly available datasets that enable analysis of AD progression and identification of specific biomarker predictors of cognitive decline. We aimed to comprehensively analyze these datasets in order to inform a QSP model that will 1) understand and quantify the putative mechanistic interactions between Aβ pathology (plasma/CSF soluble Aβ and Aβ PET) and tau pathology (plasma ptau, CSF ptau, and neurofibrillary tangle [NFT] burden measured via Tau PET), 2) relate these measurable biomarkers with cognitive tests (MMSE and iADRS), and 3) employ virtual population techniques to suitably recapitulate observed biomarker and efficacy profiles of clinical therapies.

Methods: Longitudinal changes in Aβ and Tau PET, along with CSF and plasma ptau181, were consolidated across different ADNI datasets and analyzed using MATLAB. These data were further stratified between patients categorized as cognitively normal (CN), mild cognitive impairment (MCI), and advanced AD. Since there has been substantial clinical interest in targeting early symptomatic AD, data from MCI patients were further categorized as Aβ+ (normalized Aβ SUVR > 0.78) and tau+ (CSF ptau181 > 24.5 pg/mL) using previously defined thresholds within ADNI datasets. Results from ADNI dataset analyses are being used to inform parametrization of a mathematical model for brain antibody disposition, Aβ and ptau dynamics, and inform plausible parameter distributions for relevant AD virtual populations.

Results: MCI Aβ+ patients showed higher CSF and plasma ptau181 than Aβ-, suggesting interrelationship between Aβ and tau pathologies consistent with previous findings linking amyloid burden to soluble ptau.3 Moreover, tau positivity in Aβ+/tau+ MCI patients, CSF ptau181 demonstrated a longitudinal annualized increase of 1.82 pg/mL in Aβ+tau+ MCI patients. NFT burden measured by tau PET (SUVR averaged over Braak 3 regions, weighted by volume) in MCI Aβ+ patients demonstrated longitudinal increase, except for those with high baseline NFT burden. These results are used to inform model development and parameterization.

Conclusions: Data from published experiments, reported trial readouts, and ADNI analyses will aid in establishing an integrated quantitative framework for AD. Innovative approaches providing data-driven insights from large and multidimensional data sources are critically important to increase credibility and confidence in mechanistic model-informed decision-making.

Citations: 1. McKhann GM et al. Alzheimers Dement 7 (2011)
2. Yadollahikhales G et al. Neurotherapeutics 20 (2023)
3. Ashton NJ et al. Nature Medicine 28 (2022).