Abstract
PURPOSE: To comprehensively characterize the clinical and genomic landscapes of PIK3CA, AKT1, and PTEN alterations and examine their functional implications in AKT-driven breast cancer. EXPERIMENTAL DESIGN: Comprehensive genomic profiling of 51,767 breast tumors was performed with FoundationOne (®) CDx or FoundationOne (®) . We examined the genomic landscape of PIK3CA , PTEN , and AKT1 alterations and their distribution across clinical variables of interest. Prior deep mutational scanning (DMS) data was used to functionally characterize clinical PTEN variants. RESULTS: There were 29,157 total variants across PIK3CA, AKT1, and PTEN , including pathogenic variants and VUS. The most frequently altered gene was PIK3CA (37.4% of cases), followed by PTEN (13.5%), then AKT1 (5.4%). The most common alterations in each gene were PIK3CA H1047R (35.6% of PIK3CA -altered cases), E545K (19.7%), and E542K (11.7%); AKT1 E17K (69.7%); and PTEN homozygous copy number deletion (37.3%). PIK3CA alterations were less prevalent in patients of African genetic ancestry (27.1% vs 38.6% in European genetic ancestry), while AKT1 and PTEN alterations were balanced across ancestries. PIK3CA , AKT1 , and PTEN pathogenic alterations were all mutually exclusive to each other. Using available DMS data on missense PTEN mutations, we found that 32.5% showed discordant effects on protein stability and phosphatase activity, underscoring the need for functional validation beyond predicted loss-of-function. CONCLUSIONS: Here we present the landscape of PIK3CA , AKT1 , and PTEN alterations in the largest clinical cohort examined to date. The functional implications of lesser-known variants in each gene warrant further investigation by tools such as deep mutational scanning.