Age-related amyloid beta dynamics modeled with the generalized additive model for location, scale, and shape (GAMLSS) across diverse populations: Cross-sectional trajectories and longitudinal validation

Abstract

Introduction: We developed and validated age-related amyloid beta (Aβ) positron emission tomography (PET) trajectories using a statistical model in cognitively unimpaired (CU) individuals.

Methods: We analyzed 849 CU Korean and 521 CU non-Hispanic White (NHW) participants after propensity score matching. Aβ PET trajectories were modeled using the generalized additive model for location, scale, and shape (GAMLSS) based on baseline data and validated with longitudinal data. Subgroup analyses examined apolipoprotein E (APOE) ε4 and sex effects.

Results: Age-related centile curves of Aβ PET Centiloid values showed stable distributions in the lower percentiles, increasing with age in the upper percentiles. NHWs exhibited steeper Aβ accumulation trajectories, particularly among APOE ε4 carriers. Calibration with longitudinal data confirmed the reliability of this cross-sectional method.

Discussion: We developed a statistical model of age-related Aβ PET trajectories using baseline data, validated with longitudinal data. NHWs exhibited steeper trajectories than Koreans, suggesting population-specific differences in Aβ accumulation.

Highlights: A generalized additive model for location, scale, and shape model was applied to examine age-related amyloid beta (Aβ) trajectories with baseline data. Trajectories were validated using longitudinal data, confirming model reliability. Non-Hispanic Whites exhibited steeper trajectories than Koreans, especially in apolipoprotein E ε4 carriers. Our approach enables scalable modeling of Aβ dynamics for Alzheimer's disease prevention strategies. Findings highlight the importance of multi-ethnic research in Aβ accumulation.