Subtype-specific PI3K-pathway and TP53 alteration patterns in breast cancer: A cross-cohort comparison of TCGA-BRCA and AACR GENIE.

Abstract

e13011 Background: The PI3K pathway is among the most actionable signaling axes in breast cancer, yet its alteration frequency and genomic context may differ between research-grade primary tumor datasets and real-world clinico-genomic cohorts. We compared PI3K-pathway alteration prevalence and TP53 co-mutation patterns between TCGA-BRCA and AACR GENIE (breast). Methods: We analyzed TCGA-BRCA (n = 971) and GENIE breast cancer (n = 1221) using a harmonized gene set: PIK3CA, AKT1, PTEN, TP53. “PI3K-pathway altered” was defined as any mutation in PIK3CA and/or AKT1 and/or PTEN. Within-pathway relationships (mutual exclusivity/co-occurrence) were assessed using Fisher’s exact tests (odds ratios [OR]). TP53 co-mutation was evaluated across PI3K alteration classes (PIK3CA-only, AKT1-only, PTEN-only, ≥2 PI3K genes, PI3K-wildtype). Subtype-stratified analyses were limited to subtype-annotated samples. Results: Overall, PI3K-pathway alterations were common in both cohorts but higher in TCGA than GENIE (40.3% vs 35.7%, p = 0.0298). TP53 alterations were more frequent in GENIE than TCGA (46.3% vs 34.4%, p = 2.12×10⁻⁸). Within the PI3K pathway, PIK3CA and AKT1 were mutually exclusive in both cohorts (GENIE OR = 0.23, p = 0.00688; TCGA OR = 0.15, p = 0.00172).  In subtype-annotated samples, PI3K alteration prevalence was highest in HR+/HER2− disease (TCGA 48.8%) and lower in TNBC (TCGA 17.7%, GENIE 12.2%). TP53 co-mutation was enriched in TNBC (TCGA 82.3%, GENIE 89.9%) compared with HR+/HER2− (TCGA 20.9%), and TP53 rates varied across PI3K alteration classes within each subtype.  Conclusions: PI3K-pathway alterations are frequent across both TCGA and GENIE breast cancer cohorts, but their prevalence and TP53 genomic context differ, including in a subtype-dependent manner. These findings highlight how real-world and research cohorts can yield distinct actionable-landscape estimates and support interpreting PI3K-targeted strategies within subtype-specific, co-mutational contexts.