Deep-learning-derived glaucoma-related endophenotypes enable novel genome-wide genetic and functional discovery.

Area of Science:

• Ophthalmology
• Genetics
• Artificial Intelligence

Background:

• Primary open-angle glaucoma (POAG) is a major cause of irreversible blindness.
• Previous genome-wide association studies (GWAS) were limited by imprecise phenotypes from electronic health records.
• The genetic architecture of POAG remains largely unexplained.

Purpose of the Study:

• To leverage a machine learning (ML) framework with optical coherence tomography (OCT)-derived endophenotypes to identify novel genetic loci associated with POAG.
• To expand the understanding of POAG genetic architecture and its underlying mechanisms.
• To discover potential therapeutic targets for POAG.

Main Methods:

• Developed a disease-trained, task-transfer ML framework to analyze glaucoma-related damage patterns from clinical OCT scans.
• Applied ML-derived OCT endophenotypes to GWAS in UK Biobank participants across European, African, and Asian ancestries.
• Conducted extensive functional analyses, including colocalization, gene-based tests, Mendelian randomization, and single-cell enrichment.

Main Results:

• Identified 36 and 43 genome-wide significant loci in European and cross-ancestry meta-analyses, respectively.
• Discovered 21 novel loci associated with POAG, expanding known genetic associations.
• Converged on 11 high-confidence effector genes, five novel to glaucoma, implicating Wnt-mediated outflow dysfunction and retinal ganglion cell vulnerability.

Conclusions:

• The ML-driven approach successfully identified novel POAG genetic loci and potential causal genes.
• Findings provide mechanistic insights into POAG pathogenesis at a cell-type resolution.
• The study presents a powerful, generalizable strategy for discovering disease mechanisms and therapeutic targets for complex conditions.