Abstract Background: Cancer cells have altered metabolism, which contributes to their ability to proliferate, survive in unusual microenvironments, and invade other tissues. Measuring the complete set of metabolites in an individual (i.e. the metabolome) provides a functional readout for cellular pathways. Further, changes in the metabolome can be correlated with disease status, prognosis and progression. Using a metabolomics platform and machine learning algorithms, biomarker signatures can be identified to predict response to therapy. Metabolite analysis of pretreatment plasma has shown promise for predicting response in a small retrospective study of the CDK 4/6 inhibitors palbociclib and ribociclib in hormone receptor-positive metastatic breast cancer (Zhang B et al, ASCO 2019 Abstr 3043). HER2-positive metastatic breast cancer patients have significant heterogeneity in response and progression-free survival (PFS) to HER2-targeted therapy. Here we retrospectively evaluated the pretreatment serum metabolome for association with PFS in a cohort of trastuzumab-treated metastatic breast cancer patients from a single institution. Methods: Pretreatment serum from 26 HER2-positive trastuzumab-naive metastatic breast cancer patients who were treated with first-line trastuzumab and chemotherapy were included in this exploratory analysis. Metabolites were extracted from previously frozen serum (1 mL) using ice-cold methanol and chloroform. The resulting metabolites were isolated and quantified using an unbiased, non-destructive, nuclear magnetic resonance (NMR)-based profiling platform (Olaris, Inc., Cambridge, MA). The serum was analyzed via 1D 1H NMR and 2D 13C-1H heteronuclear single quantum coherence spectroscopy (HSQC) using customized non-uniform sampling (NUS) techniques and processed with proprietary Olaris software. Supervised and unsupervised machine learning algorithms were used to identify patients with shorter and longer PFS to trastuzumab-based therapy. Results: The median PFS for this cohort was 301 days. Patients were subdivided into early progressors (PFS < 301 days) and late progressors (PFS ≥ 301 days). 23 metabolite resonance levels were statistically different between the two groups (KW test p<0.05, 11 metabolites expected by chance). Using advanced machine learning we constructed a model based on 5 metabolite resonances that showed significant discriminatory ability with an AUC of 0.964 using receiver-operating curve (ROC) analysis. 21 of the 26 patients received trastuzumab and chemotherapy as a first-line therapy, while the remaining 5 patients received first-line trastuzumab with subsequent chemotherapy, after 1 or more lines of previous chemotherapy for metastatic disease. We repeated our analysis using only the first-line therapy subgroup and identified a partially overlapping set of metabolite resonances that could nearly perfectly discriminate early and late progressors with an AUC of 0.973. Further efforts are underway to confirm the identity of these metabolites. Conclusions: Metabolic profiling of pretreatment serum using NMR was successful in identifying a biomarker signature that predicted PFS to trastuzumab in HER2-positive metastatic breast cancer. Expanded metabolome analysis is warranted in larger cohorts and clinical trials to confirm that this serum biomarker signature predicts PFS to trastuzumab therapy, particularly in the first-line setting. Further, by identifying the metabolites and metabolic pathways that differ between early and late progressors, it may be possible to identify novel targets and/or suggest combination treatments in the HER2-positive metastatic breast cancer setting. Citation Format: Elizabeth O'Day, Kim Leitzel, Suhail M Ali, Bo Zhang, Chen Dong, Haiwei Gu, Xiajian Shi, Joseph J Drabick, Leah Cream, Monali Vasekar, Hyma V Polimera, Vinod Nagabhairu, Prashanth Moku, Ashok Maddukuri, Harry Menon, Neha Pancholy, Walter P Carney, Wolfgang Koestler, Allan Lipton. Pretreatment serum metabolome predicts PFS in first-line trastuzumab-treated metastatic breast cancer [abstract]. In: Proceedings of the 2019 San Antonio Breast Cancer Symposium; 2019 Dec 10-14; San Antonio, TX. Philadelphia (PA): AACR; Cancer Res 2020;80(4 Suppl):Abstract nr P4-10-25.
Read full abstract