Background

Mutations in the catalytic domain of phosphatidylinositol 3-kinase (PIK3CA) are among the most frequently observed activating mutations in breast cancer. We used the I SPY 1 TRIAL, a group of biologically and clinically high risk patients molecularly profiled and treated with neoadjuvant chemotherapy, to determine the frequency of mutations and their relationship to pathologic complete response (pCR) and outcomes, within the entire cohort and within subtypes defined by growth and hormone receptor (HR) expression.

Methods: Patients enrolled in the I-SPY 1 TRIAL had a tumor size ≥3.0cm and were administered a doxorubicin-containing regimen, followed by a taxane, prior to surgery. Sequenom single nucleotide polymorphism (SNP) profiling was performed on breast tumor genomic DNA isolated from a subset of patients (n=152). A total of 149 SNPs covering 16 genes (including PIK3CA and AKT1/2/3) were analyzed. Mutations were tested for association with estrogen receptor (ER), progesterone receptor (PgR), and HER2 status, as well as pCR, using Fisher's exact test; associations between mutations and recurrence-free survival (RFS) were measured by log-rank tests. pCR was defined as no invasive tumor present in either the breast or axillary lymph nodes following neoadjuvant treatment.

Results: Of 149 mutations profiled in the cohort, 13 of the SNPs were observed. PIK3CA mutations were the most frequently observed in the panel (15.1%), followed by AKT1(E17K; 2.7%), CTNNB1 (D32; 1.4%), NRAS(Q61; 0.7%), and FGFR2(N549; 0.7%). Mutations in PIK3CA or AKT1 was associated with ER-positivity (p=0.0047) and PgR-positivity (p=0.044). Within receptor subtypes, the frequencies of PIK3CA/AKT1 mutations were also significantly different (HR+HER2−: 27%:(18/68); HER2+ 20% (8/40) [HR+HER2+: 26%, HR-HER2+: 14%]; HR-HER2−: 0% (0/36), p<0.0008). Unlike previous reports (Loi et al, PNAS 2010), no significant association between PIK3CA/AKT mutation status and RFS was observed when we restricted our analysis to the adjuvant endocrine treated subset of the HR+HER2− patients (n=49; log rank p = 0.369). In contrast, and similar to cell line reports (Junttila et al, Cancer Cell 2009), PIK3CA mutations appears to associate with worse RFS within the small subset of trastuzumab treated HER2+ patients (n=22, 13 HR-HER2+, 9 HR+HER2−; log rank p=0.001), suggesting mutations may influence response. Similar analyses of a larger cohort are planned to confirm these observations.

Conclusions: Within the I-SPY 1 TRIAL cohort, PIK3CA and AKT1 mutations are much more frequent in the HR+ and HER2 subsets but are not predictive of response to therapy or outcome except potentially within the HER2+ subset. The potential link observed between activating PIK3CA/AKT mutations and trastuzumab resistance merits further investigation, as it may provide a clinical rationale for testing PIK3CA mutation status in HER2+ patients and investigating combinational therapies targeting this pathway, particularly in the HER2+HR+ subset which have an elevated risk for recurrence despite pCR and trastuzumab therapy.

Citation Information: Cancer Res 2011;71(24 Suppl):Abstract nr P5-01-05.