Abstract P2-11-02: Understanding the biology and prognosis of PIK3CA gene mutations in primary breast cancer using gene expression profiling: A pooled analysis

Abstract

Abstract BACKGROUND: PIK3CA mutations (mt) represent one of the most common gene aberrations occurring in breast cancer (BC). Despite their high prevalence, the mechanisms underlying prognosis in PIK3CA mutated BC remain largely unknown. METHODS: Seven datasets of primary BC patients were compiled comprising 1030 patients with available PIK3CA genotype and whole genome gene expression data. Overall survival data were available for 669 patients, 502 were ER-positive patients. We evaluated whether genes of interest (ESR1, HER2, MET, MYC, TP53, FGFR1) and previously published gene signatures (GS) of proliferation (AURKA, CIN70, GGI, Gene70), stroma (PLAU, DCN), and kinase signalling pathways (IGF1, SRC, PTEN, RAS, MAPK), were associated with PIK3CA genotype using ROC curve and t-tests as well as clinicopathologic parameters using chi-square and Mann-Whitney test. Survival and interactions with genotype was evaluated using Cox regression analyses, stratified by dataset. RESULTS: A total of 1030 cases were pooled for analysis. PIK3CA mt were found in 30.4% (314/1030) of cases in our cohort. PIK3CA mutations were significantly associated with positive ER status (p<0.001) and older age (p = 0.002). There were no significant associations found with tumour size (p = 0.06), grade (p = 0.7), nodal status (p = 0.8) or HER2 status (p = 0.5). PIK3CA genotype was not significantly associated with survival overall: hazard ratio (HR):0.9; (95%CI: 0.56-1.14);p = 0.6. A previously defined PIK3CA-gene signature (PIK3CA-GS) was significantly associated with the PIK3CA genotype in 875 samples not previously used for training the signature (ROC AUC = 0.71, p<0.001). Expression levels of MET, MYC, SRC, PTEN, MAPK, RAS, and PTEN GS were significantly lower in mt vs wild-type (WT) cases (all p<0.05). ESR1, ERBB2, TP53, IGF1, PLAU, DCN were significantly higher in mt (all p<0.05). Levels of prognostic proliferation GS (GGI, Gene70, CIN70, AURKA) were significantly lower in mt vs WT cases (all p<0.05). With regards to overall survival, there was no significant interaction between PIK3CA genotype and proliferation signatures (interaction p values AURKA = 0.45, CIN70 = 0.41, GGI = 0.41, Gene70 = 0.34). In contrast, high levels of MET was prognostic in PIK3CA mt (1.7, 95%CI: 1.1-2.7) but not WT (0.7, 95%CI: 0.85-1.4) BC (p interaction = 0.013). CONCLUSIONS: This is the largest analysis to combine gene expression and PIK3CA genotype. PIK3CA genotype was not significantly associated with survival in this cohort. There was no interaction between known prognostic GS reflecting proliferation and genomic instability and genotype implying that additional hits are necessary for PIK3CA mutated tumors to develop an aggressive, proliferative phenotype. In contrast, upregulation of MET was associated with a poor prognosis in only PIK3CA-mutated tumours suggesting a possible biomarker and rationale combination with PI3K inhibitors. Citation Information: Cancer Res 2013;73(24 Suppl): Abstract nr P2-11-02.