Senior Scientific Advisor Certara, Otago, New Zealand
Disclosure(s):
Stephen Duffull, PhD: No financial relationships to disclose
Objectives: Alzheimer’s Disease (AD) is an irreversible, progressive neurodegenerative disorder that results in loss of cortical function, dementia and ultimately death. Currently there is no cure and at best a modest improvement in cognitive scores have been demonstrated in clinical trials. The hallmarks of the pathogenesis of AD are (1) amyloid-β (Aβ), accumulation of brain amyloid-plaques, and (2) hyperphosphorylated tau (pTau), accumulation of insoluble intraneuronal neurofibrillary tangles of tau (NFTs). These processes are not, however, mutually exclusive. The aim of this work was to develop a mechanistic model that incorporates the interplay of commonly measured biomarkers including, Aβ42, Aβ soluble oligomers (AβO), Aβ plaque (via amyloid-PET), plasma pTau217, CSF pTau217, and NFT accumulation (via Tau-PET) in spatially resolved brain regions defined by Braak staging [1].
Methods: Synthesis of available literature provided in-depth descriptions of the both the Aβ and pTau systems, but no information were available on links between these subsystems. Clinical data were available for the anti- Aβ compounds: donanemab, lecanemab, aducanumab, gantenerumab and verubecestat and the anti-pTau antisense compound BIIB040. Donanemab and verubecestat were used for calibration and the model was then used to predict into lecanemab, aducanumab and gantenerumab and BIIB040. Coding was performed using R (ver 4.3.0).
Results: The mechanistic model consisted of 46 ordinary differential equations that spanned 3 spatially resolved brain regions to describe the progression of pTau throughout the brain. Each region consisted of 5 catenary intracellular compartments linked with localised ISF regions. All ISF regions were in equilibrium with the CSF and subsequently plasma. The Aβ system consisted of three compartments corresponding to Aβ42, AβO and Aβ-plaque with positive gain to emulate the effects of seeding. AD was invoked by two independent processes (1) an increase in hyperphosphorylation of tau and (2) formation of a seed to increase in the formation rate of AβO from Aβ42. The Aβ-pTau link model that was most consistent with the data involved the Aβ-activation of hyperphosphorylation of tau from normal tau. The calibrated model was able to holistically describe clinically available data. Simulations from the model demonstrated that early long-term treatment with a very low dose anti- Aβ intervention may result in substantially delayed onset of NFT progression potentially lasting decades. Early treatment was defined as when amyloid-PET positivity is first detected. Delayed treatment, as has been the case for all current clinical studies was predicted to have marginal effects on NFT progression.
Conclusions: A mechanistic model for AD biomarkers was developed and found to have good predictive performance for various anti-AD compounds. Simulations from the model indicated that early treatment intervention may be critical to successful amelioration of AD.
Citations: [1] Braak et al. Acta Neuropathol. 2006;112(4):389-404
