Kay Hoong Chow1, Tongrong Wang2, Ilke Tunali2, Samantha Burnham2, Sergey Shcherbinin2, Kevin Biglan2, Yvonne Vandenburg2, Matan Dabora2, Nieves Velez de Mendizabal2, Ivelina Gueorguieva1
1Eli Lilly and Company, 2Eli Lilly and Company
Background
Brain β-amyloid neuritic plaque accumulation is associated with an increased risk of progression to Alzheimer’s disease (AD). Several studies estimate rates of change in amyloid plaque over time and factors impacting amyloid plaque accumulation using ADNI [1], [2]. Change in natural amyloid plaque levels over a 10-year period was found to follow an exponential growth model with a population mean intrinsic rate of 3 centiloid units/year in the continuum of AD disease with age, gender, APOE4 genotype and disease stage being important factors. Preventing amyloid accumulation may delay the onset of Stage 1 AD [3], downstream irreversible pathological changes, and ultimately cognitive and functional impairment due to AD. Understanding on the natural rate of amyloid plaque accumulation and timely treatment with amyloid targeting agents will be of benefit. The aim of this analysis is to establish plaque accumulation model across >10 years using natural history cohorts and identify significant factors across a large number of participants.
Methods
A natural plaque accumulation model was established across the continuum of AD. Non-linear mixed effects analyses with covariate identification used stepwise covariate method and included participants from ADNI (N=1745), BIOFINDER (N=265) and LEARN (N=4492). Age was modelled as the independent variable to potentially characterize the typical long-term amyloid profile and the distribution of disease state parameters e.g. onset age of amyloid accumulation, rather than the typical study inclusion criteria determined parameters such as baseline amyloid at time of enrollment/inclusion. Linear, and non-linear (in particular, sigmoidal) models were evaluated to estimate population mean and between-participant variability on the accumulation rate, age of onset and other model parameters.
Results
An Emax model with mixture model identifying two sub-populations (amyloid accumulators and non-accumulators) was fitted to the data. Age of onset of amyloid accumulation was estimated at 71.5 (0.789%) years, EAge50 (the number of years post-start of accumulation at which 50% amyloid accumulation occurs) at 20.8 (2.64%) years and Emax at 185 (1.64%) centiloids. Covariates identified included APOE4 genotype on onset age, Emax and probability of being an amyloid-accumulator; race on Emax; and sex on EAge50. APOE4 genotype was particularly significant with increasing homogeneity being associated with earlier onset age and probability of being an amyloid accumulator. Non-carriers and homozygotes had an estimated onset age and of 71.5 years and 53.3 years respectively. Non-carriers and homozygotes had an estimated probability of being an accumulator and of 0.466 and 1 respectively. These findings are consistent with the known association between APOE4 genotype and disease progression.
Conclusions
This natural amyloid plaque model predicting the time course over decades may be used in designing and interpreting novel primary AD prevention trials. Preventing amyloid accumulation would prevent the onset of Stage 1 AD, downstream irreversible pathological changes and ultimately cognitive and functional impairment due to AD. At present, amyloid modifying therapies have typically been described by indirect turnover models [4]; these models have limitations when used for simulation of different patient populations due to the progression of amyloid being constrained by being in equilibrium at study baseline. Implementation of this long-term characterization of amyloid progression over participants’ age into these models may allow for better ability to predict the pharmacodynamic effect of these amyloid modifying therapies in different patient populations, such as in preclinical AD populations where baseline amyloid distribution is lower.
[1] Elhefnawy et al. J Pharmacokin. Pharmacodyn, Jan 2025
[2] Jaguest WJ etal., Neurology 96(9):e1347, 2021
[3] Jack CR Jr et al., Alzh Dementia 20: 5143-5169, 2024
[4] Gueorguieva I., Clin Pharmacol Ther. 2023 Jun;113(6):1258-1267
Reference: PAGE 33 (2025) Abstr 11741 [www.page-meeting.org/?abstract=11741]
Poster: Drug/Disease Modelling - CNS