Abstract Background: Cyclin-dependent kinase 4/6 inhibitors (CKD4/6i) have demonstrated clinical utility extending progression-free survival (PFS) and overall survival (OS) for advanced hormone receptor positive and HER2 negative (HR+/HER2-) breast cancer patients. The efficacy in early-stage breast cancer (eBC) is unclear, with conflicting results from adjuvant CDK4/6i trials on invasive disease-free survival. Thus, there is a critical need to identify biomarkers of response (BoR) to determine which, if any, eBC patients could benefit from this treatment. This BoR could also stratify advanced BC patients for likelihood to respond to CDK4/6i. Metabolism is influenced by both genome and environment, and changes in the metabolome can be correlated with drug responsiveness. Thus, metabolite BoRs may serve to identify eBC patients for which CDK4/6i would offer a therapeutic benefit.Methods: Plasma and serum samples from 50 early-stage ER+/HER2- breast cancer patients, treated with neoadjuvant CDK4/6 inhibitor palbociclib (palbo) and aromatase inhibitor (AI) anastrozole on NeoPalAna trial (ClinicalTrials.gov identifier NCT01723774), were collected from treatment-naïve patients (BL) and 3 consecutive time points: anastrozole,1 mg daily for 4 weeks (C1D1), anastrozole plus palbo,125 mg daily, for 15 days (C1D15), and for 4-5 months before surgery (SURG). Metabolites were extracted from all samples via methanol and chloroform precipitation and quantified using an unbiased, non-destructive, nuclear magnetic resonance (NMR)-based profiling platform (Olaris®, Inc., Waltham, MA). Statistical analysis and machine learning was used to identify differential metabolites and generate predictive models. A separate validation set of samples was collected from a subset of patients (N=6) who received an additional cycle of palbo treatment prior to surgery to assess model accuracy. Results: Non-parametric differential expression analysis of BL/C1D1, BL/C1D15, and C1D1/C1D15 identified 53 ,97, and 90 differential NMR resonances in plasma (p<0.05) and 36, 34, and 25 differential NMR resonances in serum (p<0.05), respectively. Based on the proliferative marker Ki67 levels at C1D15, 37 patients were classified as responders (Ki67≤2.7%) and 6 patients as non-responders (Ki67>2.7%). Analysis of the responder (R) and non-responder (NR) groups identified that 13 plasma and 14 serum resonances (21 unique resonances and 6 overlapping) were differentially expressed (p<0.05) at C1D1. Many of the differential resonances could be mapped back to amino acid metabolites including several branched chain amino acids such as leucine, valine, and isoleucine, and positively charged amino acids such as lysine. A Olaris® BoR score was generated using 5 differential resonances that had an AUC of 0.931 (training set) and 100% accuracy when predicting palbo-response in a blinded test set (N=6).Conclusion: The differential metabolites identified from matching plasma and serum samples suggest that, compared to serum, plasma has a better representation of the metabolic changes associated with palbo treatment-response. While comparing samples from R and NR patients, amino acids were found to be consistently altered in both serum and plasma before palbo treatment. In addition, a BoR model based on select metabolites could precisely stratify palbo-response in a blinded dataset. A larger independent validation cohort is ongoing. Citation Format: Chen Dong, Shana Thomas, Chandrashekhar Honrao, Leonardo O. Rodrigues, Nathalie Tessier, Bo Zhang, Souzan Sanati, Kiran Vij, Brenda J. Ernst, Karen S. Anderson, Mateusz Opyrchal, Foluso Ademuyiwa, Lindsay L. Peterson, Matthew P. Goetz, Donald Northfelt, Elizabeth O'Day, Cynthia Ma. Identifying a metabolite signature that correlates with tumor proliferation in early-stage breast cancer patients treated with CDK4/6 inhibitors from matched plasma and serum samples [abstract]. In: Proceedings of the 2021 San Antonio Breast Cancer Symposium; 2021 Dec 7-10; San Antonio, TX. Philadelphia (PA): AACR; Cancer Res 2022;82(4 Suppl):Abstract nr P5-13-20.
Read full abstract