Abstract
Breast cancer (BRCA) affects one in eight women, and while past studies have identified susceptibility genes and variants, those driving early-onset disease remain opaque. We introduce EA-Pathways, a control-free association method that aggregates ultra-rare germline coding variants weighted by Evolutionary Action to detect functional impact distribution biases in genes and pathways. In UK Biobank women with BRCA, EA-Pathways prioritized candidate risk genes across pathways enriched in known familial genes. Ultra-rare, high-impact variants in these genes were BRCA specific, and variants from eight pathways were associated with 2-year-earlier disease, including homology-directed DNA repair, TP53, and pexophagy pathways. The early-onset effect replicated in All of Us with diagnoses up to 6, 9, and 3 years earlier in European, admixed American, and African genetic ancestries, highlighting risk alleles with differing frequencies between ancestry groups. These findings demonstrate EA-Pathways as a sensitive pathway association method capable of identifying cancer-predisposing genes and ancestry-informed BRCA screening opportunities.</p>