Evaluating the polygenicity of brain structure features using Compositional Data Analysis

Abstract

Background: Imaging genetics (IG) studies aim to jointly analyse neuroimaging and genetic data with the objective of discovering new genetic variations related to brain features. Most IG studies focus on the individual analysis of brain structures. An alternative strategy is to incorporate compositional data analysis (CoDA) methods to assess the joint modulation of specific brain subregions. Objective: The aim of this project was to investigate whether the genetic predisposition to specific neurodegenerative disorders (quantified with polygenic risk scores, PRS) was associated with the joint modulation of hippocampal subfields volumes (target regions for neurological disorders) by assessing the performance of CoDA (Selbal algorithm). Methods: A total of 1,071 participants from the ALFA study with available information on genetics and neuroimaging data were included. Genetic predisposition to Alzheimer’s Disease (AD), Amyotrophic Lateral Sclerosis (ALS) and Progressive Supranuclear Palsy (PSP) was estimated by calculating PRS (PRSice v.2). Selbal algorithm was applied to find the hippocampal subregions whose joint volumetric variation was most closely related to a higher genetic risk of each neurodegenerative condition. Logistic regression models were assessed to test the association between the genetic predisposition of each condition and the volumetric combination of the hippocampal subfields. Models were adjusted by sex and we also performed sex- and hemisphere-stratified models. Results: Results showed that a compensatory increase in the average volume of CA3, CA4 and hippocampal fissure related to CA1 and hippocampal tail was significantly associated with a higher genetic risk of AD. Results also showed that a higher genetic risk of ALS was significantly related to a compensatory increase in the CA1 compared to the hippocampal fissure. Results for PSP showed that a compensatory increase in the subiculum in comparison to the parasubiculum was significantly associated with a higher genetic risk. Moreover, we found different joint volumetric modulation of hippocampal substructures associated with higher genetic risk of each condition between sex, as well as among hemispheres. Conclusion: To our knowledge, this is the first study analysing the relationship between cognitively healthy individuals at high genetic risk of AD, ALS, and PSP and the joint volumetric variation of hippocampal subfields. Therefore, this work provides a new and innovative perspective for IG studies with the aim of improving our understanding of the effects that the genetic predisposition to neurodegenerative disorders has on brain structure modulation.