Basic Science and Pathogenesis.

Background: Over-representation of several health conditions (such as diabetes, hearing loss, etc) have been identified up to 15 years before Alzheimer's Disease (AD) diagnosis through the study of electronic health records [1]. Mechanisms underlying these associations remain elusive. We propose to study the associations between these co-pathologies (proxied by genetic risk scores), and the physiological and clinical evolution of AD patients.

Method: Longitudinal and genetic data was obtained on 1120 AD patients from the ADNI cohort. Genetic risks for AD and selected health conditions (type 2 diabetes, hypertension, hearing loss, depression, ldl cholesterol, and colon cancer) are computed using state-of-the-art Polygenic Risk Scores. We utilize a non-linear mixed-effects model [2] to estimate a multimodal progression profile of the disease (jointly modelling structural and PET imaging biomarkers as well as cognitive scores - fig. 1). Interpretable parameters describing each patient's progression are estimated (Fig. 2). Pearson correlations between random-effects and genetic risk scores were computed, with presented P-values Bonferroni-corrected for multiple testing. Conditioning the fixed-effects of the model by the genetic risk scores (as introduced in [2]) allows studying associations between these genetic risks and the disease dynamics at a biomarker-by-biomarker level.

Result: Higher AD genetic risk is associated with all longitudinal aspects of the disease: earlier onset (P<0.001), faster disease pace (P<0.001), and inter-marker patterns (P = 0.042). Elevated type-II diabetes genetic risk is associated with earlier onset of AD (p = 0.031). Increased hypertension genetic risk is associated with a faster progression pace (P = 0.012). We then exhibit (Fig. 3) finer links between these genetic risks and the progression patterns of the disease.

Conclusion: This study unveils the relationship between genetic risk scores for health conditions and nuanced aspects of AD progression. We show that despite being established on case/control cohorts, these genetic risk scores offer valuable insights into the finer dynamics of disease progression. References: [1] Nedelec et al., Identifying health conditions associated with Alzheimer's disease up to 15 years before diagnosis: an agnostic study of French and British health records, in Lancet Digital Health, 2022 [2] Fournier and Durrleman, A Multimodal Disease Progression Model for Genetic Associations with Disease Dynamics, in MICCAI 2023.