Objective: Identify genetic variants associated with amblyopia in African American (AFR) and Admixed American (AMR) ancestry groups, expanding upon a previous studies conducted in European ancestry.

Design: Retrospective ancestry-stratified genome-wide association study (GWAS) and gene-level rare variant association study (RVAS).

Participants: Participants in the All of Us Research Program from AFR and AMR ancestry groups with whole genome sequencing available. Cases and controls were distinguished based on presence of ICD-9/10/SNOMED diagnosis codes for amblyopia in electronic health records. This yielded ancestry-stratified subsets of 269 cases and 71,585 controls of AMR ancestry and 366 cases and 79,460 controls of AFR ancestry.

Methods: Stratified logistic regression models adjusted for age, sex, and the top 10 principal components of genomic ancestry. GWAS was limited to common variants (mean allele frequency or MAF > 1%) and RVAS was limited to rare variants with coding sequence-altering effects (MAF < 1%, exonic only, excluded synonymous variants) aggregated at the gene level using the SKAT algorithm. Downstream analyses of the significant variants were performed using KEGG and GO pathway analysis and STRING database queries for protein-protein interactions and gene-gene interactions.

Main outcome measures: Single-nucleotide polymorphisms (SNPs) were determined to have genome-wide significance if p < 5e-8 in the GWAS and genes were determined to have significant association with amblyopia in the RVAS if p < 8.0 x 10-4.

Results: In the AMR GWAS, 245 unique SNPs mapping to 97 distinct loci were identified, notably within neurodevelopmental and axonal guidance genes, including ROBO1, SEMA4B, PTPRD, NRXN1, and CAMK2D. The AFR GWAS identified 11 significant variants corresponding to 6 loci mapping primarily to long-noncoding RNAs and pseudogenes. The AMR RVAS identified 15 genes, including axonal transport genes (KIF1B, KIF7) and growth factor signaling genes (EGF, ERBIN, and AKAP17A). The AFR RVAS identified a single gene, DLG2, which encodes the postsynaptic protein PSD-93, which promotes the closure of the sensitive period of neuroplasticity for vision in early childhood.

Conclusions: Genetic risk architectures for amblyopia differ across ancestries but fundamentally converge on neurodevelopmental signaling, cortical synapse assembly, and sensitive period plasticity rather than ocular structural dynamics.