Large-scale machine-learning-based phenotyping significantly improves genomic discovery for optic nerve head morphology

Abstract

Genome-wide association studies (GWASs) require accurate cohort phenotyping, but expert labeling can be costly, time intensive, and variable. Here, we develop a machine learning (ML) model to predict glaucomatous optic nerve head features from color fundus photographs. We used the model to predict vertical cup-to-disc ratio (VCDR), a diagnostic parameter and cardinal endophenotype for glaucoma, in 65,680 Europeans in the UK Biobank (UKB). A GWAS of ML-based VCDR identified 299 independent genome-wide significant (GWS; p % 5 3 10 À8 ) hits in 156 loci. The ML-based GWAS replicated 62 of 65 GWS loci from a recent VCDR GWAS in the UKB for which two ophthalmologists manually labeled images for 67,040 Europeans. The ML-based GWAS also identified 93 novel loci, significantly expanding our understanding of the genetic etiologies of glaucoma and VCDR. Pathway analyses support the biological significance of the novel hits to VCDR: select loci near genes involved in neuronal and synaptic biology or harboring variants are known to cause severe Mendelian ophthalmic disease. Finally, the ML-based GWAS results significantly improve polygenic prediction of VCDR and primary open-angle glaucoma in the independent EPIC-Norfolk cohort.

Glaucoma Heritability: Molecular Mechanisms of Disease

“…They suggested that genetic variation in VCDR may be mediated by pathophysiologic processes involving IOP and the anterior segment of the eye. In fact, they found that there was a 13% overlap between genes in an IOP GWASA meta-analysis and their new VCDR GWAS [65]. Similarly, proposed molecular mechanisms linking genes to IOP and/or VCDR may have similar pathways, whether through glial activation or inactivation or through alterations in aqueous humor dynamics.…”

“…For example, the AI-enhanced GWAS performed by Alipanahi et al was able to use the AI-enhanced models to calculate genetic/polygenic risk scores (GRS/PRS), a method of quantifying risk based on individualized genetic characteristics. Alipanahi et al calculated a PRS for VCDR that they proposed could be applied to POAG, and they demonstrated that patients with a higher VCDR PRS had a higher prevalence of POAG [65].…”

“…For example, the interplay of microglial activation and nitric oxide, as described previously, can be a beneficial response to elevations in IOP but may also lead to the downstream loss of aquaporin channels, leading to optic nerve degeneration [46]. In the case of ASAP1, Alipanahi et al suggested that this association may indicate a mechanism by which glial cells mediate VCDR (Table 2) [65].…”

“…Then, using the machine learning-generated VCDR, they performed a GWAS to evaluate for genetic associations with VCDR. They uncovered 93 novel loci associated with VCDR [65].…”

“…Among the novel genetic loci they uncovered was a SNP near ASAP1, a locus previously associated with glioma [65,68]. Recent research has demonstrated that glaucomatous retinal ganglion cell loss, and therefore clinical measurements of VCDR, may in some cases be mediated by glial cells [66].…”