The reciprocal predictive relationships between retinal thickness and brain volume over time in Alzheimer’s disease: A random-intercept cross-lagged panel model

Theme: Neuro-ophthalmology
What: Neuro-ophthalmology
Part of: Neuro-ophthalmology II / Neuro-ophtalmolgie II
When: 6/1/2024, 02:00 PM - 03:30 PM
Where: Room | Salle 714 B

Abstract

Purpose: To understand the directionality of the predictive relationships between retinal thickness and brain atrophy over time in Alzheimer’s disease.

Study Design: Retrospective longitudinal cohort study.

Methods: ADMCI subjects were recruited in the Ontario Neurodegenerative Research Initiative. They underwent spectral domain ocular coherence tomography (SD-OCT) and magnetic resonance imaging (MRI) of the brain annually. Subjects with SD-OCTs and MRIs available for at least one timepoint were included in the current study. Peripapillary retinal nerve fiber layer (pRNFL) and posterior pole OCT scans were obtained, and poor quality images and confounding retinal diseases (maculopathies and optic neuropathies) were excluded. Global pRNFL and mean macular total retinal thickness (mRT), were each summed across both eyes for analysis. Brain volumes were quantified with semi-automatic brain region extraction. The brain-parenchymal volume (BPV), an indicator of brain atrophy, was quantified by summing the brain tissue volume and dividing by the total intracranial volume. The random-intercept cross-lagged panel model (RI-CLPM) is a statistical model used to determine the direction in which selected longitudinal variables predict each other over time. Unlike other regression models, it does not assume the direction in which the variables influence each other. The model rigorously assesses the predictive relationship of each longitudinal variable on the other while controlling for all reciprocal predictions, cross-sectional relationships, and the variable’s own change over time. pRNFL, mRT, and BPV were assessed in a RI-CLPM over three timepoints spanning two years, controlling for baseline age, sex, and number of apolipoprotein e4 (APOE4) alleles.

Results: 85 ADMCI subjects met inclusion criteria. Mean age at baseline was 71.2 (SD=8.5), and subjects were 52% male. BPV, mRT, and pRNFL showed stable progression over time, strongly predicting their subsequent timepoints. pRNFL and mRT significantly correlated with each other at cross-sectional timepoints, but neither correlated with BPV. Longitudinally, pRNFL at baseline significantly predicted BPV at year one (B=-0.05, p=0.046), but BPV never reciprocally predicted retinal thicknesses. pRNFL and mRT did not significantly predict each other. Regarding covariates, having more APOE4 alleles predicted thicker pRNFL (B=0.220, p=0.032), but a smaller BPV (B=-0.306, p<0.001). Being older significantly predicted a thinner mRT (B=-0.185, p=0.049) and smaller BPV (B=-0.580, p<0.001). Females were likelier to have a thinner mRT (B=-0.220, p=0.029) and larger BPV (B=0.245, p<0.001).

Conclusions: In ADMCI subjects, retinal thickness predicted brain volume longitudinal changes, but the reciprocal did not occur, suggesting that there may by a directional relationship during ADMCI disease progression. This may be important for methods of early detection of ADMCI.

Presenter(s)

Presenting Author: Saffire Krance

Additional Author(s):

Walter Swardfager, Department of Pharmacology and Toxicology, University of Toronto, Toronto, Ontario, Canada

Hugo Cogo-Moreira, Department of Psychiatry, Universidade Federal de São Paulo, São Paulo, Brazil

Wendy Hatch, Department of Ophthalmology and Vision Sciences, University of Toronto, Toronto, Ontario, Canada

Christopher Hudson, University of Waterloo, School of Optometry and Vision Science, Waterloo, Ontario, Canada

Peter Kertes, Department of Ophthalmology and Vision Sciences, University of Toronto, Toronto, Ontario, Canada

Sandra Black, Division of Neurology, Department of Medicine, Sunnybrook Health Sciences Centre, University of Toronto, Toronto, Ontario, Canada

---