Abstract 5573: GC-TOF mass spectroscopy reveals strong dependence of breast cancer metabolome on estrogene receptor, but not on HER2 status

Abstract

Abstract Breast cancer is a biologically inhomogeneous disease that has been extensively studied using molecular high-throughput platforms like microarrays. Molecular methods have been shown to be useful for therapy selection, prediction of disease outcome or complications such as metastases. Worldwide, immunhistological determination of estrogene receptor (ER) and HER2 status is part of the everyday routine diagnostics. A bunch of new biomarkers and biomarker signatures is currently under development and expected to further individualize and improve breast cancer treatment. Here, we present results of a GC-TOF mass spectrometry (GC-MS) metabolite study conducted by the METACancer consortium. In this project, 275 breast cancer tissues collected as fresh-frozen samples in the METACancer tumor bank were analyzed using GC-MS. Prior to metabolic profiling, tumors were divided in a training (187 tumors) and a validation cohort (88 tumors) with comparable clinicopathological characteristics. Both cohorts were profiled at the Fiehn lab (UC Davis, CA), the training cohort at the end of 2008, the validation cohort at the beginning of 2009. Analysis of the training cohort led to the identification of 468 metabolites that are abundant in breast cancer tissues. 161 out of these could be mapped to known chemical structures and metabolite names. Metabolite-by-metabolite analysis of the training cohort revealed 70 metabolites with significantly (p < 0.05, Welch's t-test) different concentrations between between ER+ and ER- tumors. Many of these differences (59%) could be affirmed by analysis of the validation cohort. Only are few changes (9 metabolites), possibly false positives, could be detected between HER2+ and HER2- tumors. None of these changes could be reproduced in the validation cohort. Next, we asked if the tumor cells exhibit metabolite patterns that are specific for ER status. To this end, a metabolic index (MI) was constructed as linear combination of 15 metabolites. The MI was constructed and optimized using only tumors of the training cohort. ROC analysis showed an excellent perfomance of the MI for prediction of the ER status in the training cohort (AUC = 0.91, leave-one-out cross-validation). Validation of the MI in the validation cohort affirmed the excellent performance (AUC = 0.97). A similar approach for prediction of the HER2 status failed (AUC not significantly better than 0.50). In this project, we have shown that metabolic profiling of fresh-frozen breast cancer samples with GC-MS is feasible. Interestingly, we detected a strong dependence of the metabolite patterns on ER status, but no dependence on HER2 status. Analysis of the changes in metabolic pathways between ER+ and ER- breast cancers may contribute to a better understanding of estrogen driven tumor growth. Citation Format: {Authors}. {Abstract title} [abstract]. In: Proceedings of the 101st Annual Meeting of the American Association for Cancer Research; 2010 Apr 17-21; Washington, DC. Philadelphia (PA): AACR; Cancer Res 2010;70(8 Suppl):Abstract nr 5573.