Expression signature of p53 status (p53 signature) for the prediction of TP53 mutation and prognosis in breast cancer

Abstract

652 Background: The current risk evaluation criteria, such as the St. Gallen and NIH, are not sufficient for predicting which BC patients are at high risk for recurrence and mortality. The TP53 mutations are associated with worse RFS and OS, independent other risk factors in BC, but the p53 status is not simply determined and often determined incorrectly (Soussi et al. Nat Rev Cancer 2006). Our purpose was to identify the gene sets that determine expression signature of the p53 status and to correlate the gene expression profile (GEP) with clinical outcome. Methods: Comprehensive expression analysis (Agilent 41K human genome oligo microarray) and DNA sequencing of the TP53 gene in 38 Japanese BC (30 Stage I-II, 8 stage III-IV; 20 TP53 wt, 20 TP53 mt) were performed using RNA and DNA obtained from the microdissected frozen tumor samples. We determined a gene set consisted of 33 genes from differentially expressed genes depending on the p53 status in the 26 BC (training set) and validated the ability to predict p53 status in the remaining 12 BC (test set). Prognostic value of the gene set was analyzed in 29 BC (stage I or II), and were validated using publicly available dataset of the independent 295 BC (stage I or II), van de Vijver et al. NEJM 2002). Results: The GEP using the 33 genes, many of which related in cell cycle and cell division, predicted the p53 status (wt or mt p53 signature) accurately in the test set (accuracy: 83%, overall accuracy: 95%). The p53 signature has the ability to predict RFS of the 29 early BC (Log Rank, P=0.032). It divided the 295 early BC into 176 wt and 119 mt group, and also has the ability to predict both RFS and OS (Log Rank, both P<0.0001). This was also true when the cohort were stratified by ER status, LN status, St. Gallen criteria and NIH criteria, and retained the predicted value in ER+, LN+, LN-, St. Gallen (intermediate/high) and NIH (high) subgroups (Log Rank, P<0.0001). Conclusions: The p53 signature is a powerful and independent predictor of the outcome of disease in early breast cancer than standard systems based on clinical and histologic criteria. No significant financial relationships to disclose.