Publication-only abstracts (abstract number preceded by an "e"), published in conjunction with the 2018 ASCO Annual Meeting but not presented at the Meeting, can be found online only.
Accuracy of the prediction of the distant recurrence risk of patients with ductal luminal breast cancer based on clinical factors versus prediction based on oncogenic mutations or gene expression.
2018 ASCO Annual Meeting
J Clin Oncol 36, 2018 (suppl; abstr e12585)
Author(s): David Demanse, Dominique Pinet, Christian Eisen, Lellean Jebailey, Wolfgang Hackl; Novartis Pharma AG, Basel, Switzerland; NOVARTIS, Basel, Switzerland; Novartis Institute for BioMedical Research, Cambridge, MA; Novartis Pharma Corp, Basel, Switzerland
Background: Luminal breast cancer is not a uniform cancer type, but a heterogeneous group of disease subtypes characterized by own clinical, pathologic, biologic and genetic characteristics. Traditionally clinical-pathologic (CP) variables have been used to guide the use of endocrine treatment and/or chemotherapy to inform the choice of systemic adjuvant therapy for patients with early stage breast cancer. More recently gene expression profiling of tumours has been applied to aid doctors in personalizing treatment decisions. In the retrospective study presented here we worked with ER and/or PR positive, HER2 negative ductal breast cancers of the METABRIC collection (N = 1.104) to determine the prediction accuracy (PA) of different distant recurrence risk (DDR) prediction methods used in clinical practice. Methods: CP and genomic variables (oncogenic mutations/CNV, gene expression) were used to develop a prognostic model for disease free survival using a robust statistics approach combined with cross-validation methodology. The PA of the different sets of information was assessed using time-dependent ROC curves. Results: The average PA at 5 years was 68% for the model based on CP variables (tumour size, differentiation, nodal status, age) alone, whereas the PA values for genetic (Foundation Medicine T5-panel) or transcriptomic variables alone (20.000 genes of cBioPortal) were 55% and 60%, respectively. Amongst all mutations tested (286 oncogenic driver mutations, 19 select gene re-arrangements) only TP53 loss-of function (LoF) mutation turned out to be a covariate predictive of high DRR. Neither TP53 LoF mutation nor any of the other genetic or transcriptomic variables impacted the accuracy of DDR predictions based on CP parameters significantly. Conclusions: Collectively our study illustrates that predictions of DRR in patients with ductal luminal breast cancers based on CP parameters, mutations or transcriptomic profiles alone are highly discordant, suggesting that CP prediction models are more accurate and reliable than predictions based on mutations or by gene expression profiling.
1. TAILORx: Phase III trial of chemoendocrine therapy versus endocrine therapy alone in hormone receptor-positive, HER2-negative, node-negative breast cancer and an intermediate prognosis 21-gene recurrence score.