Low Expression of miR-20a-5p Predicts Benefit to Bevacizumab in Metastatic Breast Cancer Patients Treated within the TANIA Phase III Trial

Background: In metastatic breast cancer (MBC) patients, no biomarker predicting benefit to a bevacizumab-containing therapy has been established yet. MicroRNAs (miRNAs) are involved in angiogenesis and treatment resistance and therefore could be of predictive value.

Methods: Profiling of 754 miRNAs was performed in tumor samples of 58 MBC patients treated with a bevacizumab-containing first-line regimen (learning set). Based on progression-free survival (PFS), patients were divided into responders (R) and non-responders (NR). Differentially expressed miRNAs between R and NR were analyzed in a cohort of 57 patients treated with first-line chemotherapy without bevacizumab (control set), to exclude miRNAs providing prognostic information. MiRNA candidates significantly associated with PFS in multivariate analysis were further validated in tumor samples of 203 patients treated within the phase III trial TANIA randomizing between chemotherapy either alone or with bevacizumab after progression on first-line bevacizumab.

Results: Low expression of miR-20a-5p (multivariate p = 0.035) and miR-21-5p (multivariate p = 0.004) were significantly associated with longer PFS in the learning set, but not in the control set. In samples from the TANIA trial, low expression of miR-20a-5p was also significantly associated with longer PFS (hazard ration (HR) 0.60; 95%-CI 0.37-0.89; p = 0.012) and longer overall survival (OS; HR 0.54; 95%-CI 0.32-0.83; p = 0.007) in the bevacizumab arm but not in the chemotherapy-only arm (PFS: HR 0.73, p = 0.119; OS: HR 1.01; p = 0.964). For miR-21-5p no significant association with PFS or OS in both treatment arms was observed.

Conclusion: MiR-20a-5p expression in breast cancer tissue was predictive for a greater benefit from bevacizumab-containing therapy in two independent cohorts.

Keywords: bevacizumab; breast cancer; miR-20a-5p; microRNA; predictive biomarker.