Investigate relationship between retinal layer thicknesses and polygenic risk of neurodegenerative disease and cognitive decline.

Optical coherence tomography (OCT) allows for non-invasive retinal visualization. Polygenic risk scores (PRS) involving genetic variants of individual small effects, are associated with disease risk/outcomes.

We utilized OCT images from 50,342 UK Biobank participants to calculate retinal layer thicknesses. We associated these thicknesses with ICD-based diagnosis of Alzheimer’s disease (AD) or Parkinson’s disease (PD) as well as a PRS for each disease using the Lassosum method. Multivariate models were developed to predict future cognitive decline.

AD diagnosis was associated with a thinner OPL (aOR=1.69, p=0.027). AD PRS was associated with thicker INL, CSI, IPL, and ISOS (p<0.05 for all) and thinner RNFL and GCL (p<0.05 for all). Diagnosed PD was associated with a thinner ISOS (aOR=2.38 , p=0.005) and PD PRS was associated with thinner OPL and CSI (p<0.05 for both). 

Worse baseline cognitive performance was associated with thinner RNFL (aOR=1.067, p<0.001), OPL (aOR=1.017, p<0.001) and ISOS (aOR=1.071, p<0.001), and thicker INL (aOR=0.948, p< 0.001), RPE (aOR=0.977, p=0.033), and GCL (aOR=0.982, p=0.012). In a stepwise Cox model, thinner RNFL (aHR=0.99, p=0.007) and OPL (aHR=0.99, p=0.02) and thicker IPL (aHR=1.04, p=0.002), as well as higher AD PRS predicted future cognitive decline. The model containing retinal thicknesses and PRS improved prediction of future cognitive decline vs. demographic factors alone.

Measurements from noninvasive retinal OCTs are associated with diagnosis and genetic risk of neurodegenerative disease and may serve as biomarkers for future cognitive decline.