Global untargeted and individual targeted plasma metabolomics of breast cancer recurrence modified by hormone receptors

Breast cancer is a heterogeneous and complex etiological disease. Understanding perturbations of circulating metabolites could improve prognosis.

We recruited breast cancer patients from Kaohsiung Medical University (KMU) to perform untargeted (case–control design) and targeted (patient cohort) metabolomics analyses in the discovery and validation phases to evaluate interaction effects between clinical factors and plasma metabolites using multivariable Cox proportional hazards model.

In the discovery phase, partial least squares-discriminant analysis (PLS-DA) showed that plasma metabolites were significantly different between recurrent and non-recurrent breast cancer patients. Metabolite set enrichment analysis (MSEA) and metabolomic pathway analysis (MetPA) showed that valine, leucine, and isoleucine degradation was the significant pathway, and volcano plot showed significant ten upregulated and two downregulated metabolites between recurrent and non-recurrent cases. Combined with receiver operating characteristic (ROC) curve and biological significance, creatine, valine, methionine, and mannose were selected for the validation phase. In this patient cohort with 41 new-recurrent vs. 248 non-recurrent breast cancer cases, followed for 720.49 person-years, compared with low level of valine, high valine level was significantly negatively associated with recurrent breast cancer (aHR: 0.36, 95% CI: 0.18–0.72, P = 0.004), especially in ER-negative and PR-negative status. There were interaction effects between valine and ER (P_interaction = 0.006) as well as PR (P_interaction = 0.002) on recurrent breast cancer. After Bonferroni correction, stratification effects between valine and hormone receptors were still significant.

Our study revealed that plasma metabolites were significantly different between recurrent and non-recurrent patients, proposing therapeutic insights for breast cancer prognosis.