A recent study of the mutation landscape of >3000 cancers across 12 major cancer types from the Cancer Genome Atlas (TCGA) program revealed PIK3CA as the second most commonly mutated gene, occurring at >10% frequency in 8 types of cancer. Limited preclinical evidence suggested that mutations affecting PIK3CA catalytic vs. non-catalytic domains can produce different phenotypic consequences; however, whether domain-specific PIK3CA mutations results in distinct pathway consequences across multiple cancer types remain unclear. Thus, we used the PARADIGM algorithm, which integrates gene expression and copy number data into a superimposed pathway structure, to infer the activities of ∼13K pathway features and compared the signaling consequences associated with different domain-specific PIK3CA mutations within the TCGA Pan-Cancer dataset. Restricting to tumors harboring missense mutations in the coding region of a single PIK3CA domain resulted in 447 unique cases. PIK3CA mutations are distributed across the domains as follows: adaptor binding domain (ABD) = 23, Ras-binding domain (RBD) = 1, C2 = 50, helical = 199, and kinase = 174. Interestingly, the distribution of PIK3CA mutations among the domains is significantly different across cancer types (chi-square test p < 0.0001); of note, kinase domain mutations account for >50% of breast cancer PIK3CA mutations, while mutations in the helical domain predominate in head-and-neck and lung squamous carcinomas. Employing logistic regression and adjusting for cancer type, we identified 711 pathway features associated with kinase domain mutations (p < 0.05). As expected for gain-of-function mutations, the inferred activity of the PI3K catalytic subunit appeared elevated among kinase domain mutations; and pathway enrichment analysis identified G-alpha (12/13) signaling events as the top enriched pathway among all interconnected pathway features (FDR-corrected EASE score = 0.0001). Interestingly, although the PI3K complexes appeared more activated in the kinase domain mutants, other constituents of the G-alpha (12/13) pathway, such as RHO GTPase complexes, showed lower activity. Similarly, 1115 pathway features were identified as associated with helical domain PIK3CA mutations. Pathway enrichment and sub-network analysis independently identified lower activation of proliferation-related (FOXM1, MYC and PLK1) pathways as a key characteristic of the helical domain relative to other PIK3CA mutants. Altogether, these findings demonstrate that somatic mutations in different PIK3CA domains across 12 different cancer types produce distinct patterns of pathway activation, suggesting that domain-specific mutants likely have different oncogenic consequences, perhaps even within the same tumor type, and that targeted PIK3CA inhibitors may show differential anticancer activity depending on the domain location of the PIK3CA mutation.

Citation Format: Christina Yau, Stephen Benz, Charles Vaske, Sam Ng, Josh Stuart, Christopher C. Benz. Differential pathway activation associated with domain-specific PIK3CA mutations. [abstract]. In: Proceedings of the 105th Annual Meeting of the American Association for Cancer Research; 2014 Apr 5-9; San Diego, CA. Philadelphia (PA): AACR; Cancer Res 2014;74(19 Suppl):Abstract nr 4165. doi:10.1158/1538-7445.AM2014-4165