MicroRNAs miR-142-5p, miR-150-5p, miR-320a-3p, and miR-4433b-5p in Serum and Tissue: Potential Biomarkers in Sporadic Breast Cancer

Abstract

Breast cancer (BC) is a heterogeneous disease, and establishing biomarkers is essential to patient management. We previously described that extracellular vesicle-derived miRNAs (EV-miRNAs) miR-142-5p, miR-150-5p, miR-320a, and miR-4433b-5p in serum discriminated BC from control samples, either alone or combined in a panel. Using these previously described markers, we intend to evaluate whether the same markers identified in EVs are also potential biomarkers in tissue and serum. Expression analysis using RT-qPCR was performed using serum of 67 breast cancer patients (BC-S), 19 serum controls (CT), 83 fresh tumor tissues (BC-T), and 29 adjacent nontumor tissue samples (NT). In addition, analysis from The Cancer Genome Atlas (TCGA) data (832 BC-T and 136 NT) was performed. In all comparisons, we found concordant high expression levels of miR-320a and miR-4433b-5p in BC-S compared to CT in both EVs and cell-free miRNAs (cf-miRNAs). Although miR-150-5p and miR-142-5p were not found to be differentially expressed in serum, panels including these miRNAs improved sensitivity and specificity, supporting our previous findings in EVs. Fresh tissue and data from the TCGA database had, in most comparisons, an opposite behavior when compared to serum and EVs: lower levels of all miRNAs in BC-T than those in NT samples. TCGA analyses revealed reduced expression levels of miR-150-5p and miR-320a-3p in BC-T than those in NT samples and the overexpression of miR-142-5p in BC-T, unlike our RT-qPCR results from tissue in the Brazilian cohort. The fresh tissue analysis showed that all miRNAs individually could discriminate between BC-T and NT in the Brazilian cohort, with high sensitivity and sensibility. Furthermore, combining panels showed higher AUC values and improved sensitivity and specificity. In addition, lower levels of miR-320a-3p in serum were associated with poor overall survival in BC Brazilian patients. In summary, we observed that miR-320a and miR-4433b-5p distinguished BC from controls with high specificity and sensibility, regardless of the sample source. In addition, lower levels of miR-150-5p and higher levels of miR-142-5p were statistically significant biomarkers in tissue, according to TCGA. When combined in panels, all combinations could distinguish BC patients from controls. These results highlight a potential application of these miRNAs as BC biomarkers.

Keywords: biomarkers; breast cancer; miR-142-5p; miR-150-5p; miR-320a; miR-4433b-5p.

FIGURE 1

Expression levels of miRNAs by RT-qPCR in serum samples and potential of cf-miRNAs to discriminate BC and subtypes (LA and TNBC) from CT. Expression levels of miRNAs (A) miR-320a-3p and (B) miR-4433b-5p in BC-S and subtypes (LA and TNBC) and CT. After outlier removal, each dot represents one sample. ROC curves for BC-S diagnosis and prognosis in Brazilian samples (C–F), comparing BC-S to CT (black), CT to LA (gray), and CT to TNBC (dotted). ROC curves were performed to evaluate miR-320a-3p (C) or miR-4433b-5p (D) individually (F) or in combination with all miRNAs in a completed panel (E). ROC = receiver operating characteristic; AUC, area under the curve; CT, serum controls; BC-S, breast cancer serum samples; LA, luminal A; and TNBC, triple-negative breast cancer. (**) p = 0.001 and (****) p < 0.0001.

FIGURE 2

Expression levels of miR-142-5p, miR-150-5p, miR-320a, and miR-4433b-5p on fresh tissue samples. miR-142-5p, miR-150-5p, miR-320a, and miR-4433b-5p evaluated by RT-qPCR in tissue samples (A–D), and the comparison between serum and tissue expression in matched samples (E,F). (A–D) Levels of expression were evaluated in NT and BC-T samples; BC-T samples comprised LA and TNBC subtypes, and the expression of all miRNAs was evaluated. After outlier removal, each dot represents one sample in the tissue group. NT samples showed overexpression of all miRNAs evaluated compared to BC-T (A) and LA samples. (E,F) BC-S and BC-T were evaluated as matched pairs from patients with both samples to all miRNAs (n = 16). NT and CT were also compared but not paired. Inverse directions were found between miRNA expression comparing BC-S (high) and BC-T (low). BC-T, breast cancer tissue samples; NT, non-tumor samples; BC-S, breast cancer serum samples; LA, luminal A; TNBC, triple-negative breast cancer; and CT, controls. (*) p = 0.01; (**) p = 0.001; (***) p = 0.0001; and (****) p < 0.0001.

FIGURE 3

Low expression of miR-320a-3p in serum is correlated with poor overall survival. The serum samples of breast cancer patients (BC-S) were divided into groups based on miR-320a-3p expression levels (high—straight line; low—dotted line). Time was evaluated in days and probability in percentage (%) of survival.

FIGURE 4

Expression levels of miR-142-5p, miR-150-5p, miR-320a-3p, and miR-4433b-5p in tissue samples according to the TCGA database. Expression levels of (A) miR-142-5p, (B) miR-150-5p, (C) miR-320a-3p, and (D) miR-4433b-5p in BC-T and subtypes (LA and BLBC) and NT. Fold change between the two groups compared and the order of comparisons were indicated before the FC and p values. NT, non-tumor samples; BC-T, breast cancer tissue samples; LA, luminal A; BLBC, basal-like breast cancer; and FC, fold change. (*) p < 0.01; (**) p = 0.001; (***) p = 0.0001; and (****) p < 0.0001.