miRNA expression changes during the course of neoadjuvant bevacizumab and chemotherapy treatment in breast cancer

Abstract

One of the hallmarks of cancer is sustained angiogenesis. Favorable results have been reported in some breast cancer (BC) patients receiving antiangiogenic therapy with bevacizumab (Bev) in combination with chemotherapy, and further knowledge on how Bev can be optimally combined with conventional treatment to increase efficacy is strongly needed. In this randomized, neoadjuvant phase II clinical trial, 132 patients with HER2-negative, nonmetastatic BC were treated with Bev in combination with sequential chemotherapy. Biopsies were sampled before treatment, after 12 weeks with anthracycline and after taxane therapy at week 25. MicroRNA (miRNA) expression profiling was performed on biopsies from each time point. Altogether, 241 biopsies were analyzed with the aim of identifying miRNA-based biomarkers of response to therapy. Results from the miRNA analyses were reported for the ER-positive cohort, which were previously demonstrated to benefit from antiangiogenic therapy in this study. For both treatment arms of this cohort, significantly different expression was observed for 217 miRNAs between objective responding and nonresponding patients before treatment initiation. These miRNAs have been linked to regulation of epithelial-mesenchymal transition, metastasis, and tumor growth, among other processes. Bev in combination with chemotherapy resulted in similar miRNA changes to chemotherapy alone. However, the deregulation of miRNA expression occurred earlier in the Bev arm. In both arms, tumor suppressor miRNAs were found upregulated after treatment, while oncogenic miRNAs were downregulated in the Bev arm. Patients responding to Bev showed a strong correlation between deregulated miRNAs and decreased proliferation score during the course of treatment, with downregulation of miR-4465 as the strongest indicator of reduced proliferation. Integrative analyses at miRNA-, gene-, and protein expression further indicated a longitudinal decrease in proliferation. Altogether, the results indicate that proliferation might represent a predictive factor for increased Bev sensitivity, which may aid in the identification of patients who could potentially benefit from Bev.

Keywords: angiogenesis; bevacizumab; breast cancer; miRNA; neoadjuvant.

Figure 1

Unsupervised hierarchical clustering of miRNA expression from all patients with biopsies before treatment (week 0; A); after 12 weeks of treatment; (B) and at the end of treatment (week 25); (C). miRNAs are shown in rows and patients in columns. Clusters were identified using the PART method (Nilsen, et al., 2013) with Pearson correlation‐based distance and average linkage. P‐values for associations between identified clusters and patient clinical parameters based on Fisher`s exact tests are indicated. Patients clustered significantly according to OR, proliferation, and PAM50 subtypes at week 0, while only relapse was significantly associated with patient clustering at week 25.

Figure 2

Changes in miRNA expression over time, regardless of treatment response, within (A) the chemotherapy arm and (B) the Bev arm. (C) 126 miRNAs were found differentially expressed in both treatment arms from week 0 to week 25, while 125 miRNAs were exclusively differentially expressed in the chemotherapy arm. Thirteen miRNAs were differentially expressed only in the Bev arm.

Figure 3

(A) miRNA–mRNA–protein integration model showing the miRNA–protein associations being kept most constant over time. On the x‐ and y‐axis, the beta value is plotted for a miRNA–protein pair at week 0 (x‐axis) and week 25 (y‐axis). (B) Venn diagram showing the proteins with most associations to miRNAs at each time point. Percentage indicates the number of associated miRNAs to each protein within the total number of significant miRNA–protein associations. Histone H3 ranked highest with most associations at all three time points.

Figure 4

Significant changes in miRNA expression over time in responding patients only within (A) the chemotherapy arm and (B) the Bev arm. (C) 54 miRNAs were found overlapping in the two treatment arms from week 0 to week 25, while 140 miRNAs were found exclusively in the chemotherapy arm. (D) The five miRNAs found exclusively in the Bev arm.

Figure 5

The 10 most significantly associated pathways associated with mRNAs with a significant correlation to four of the five miRNAs found differentially expressed between week 0 and week 25 in Bev‐responding patients.