Biomarker significance of plasma and tumor miR-21, miR-221, and miR-106a in osteosarcoma

Abstract

Osteosarcoma is the most common malignant bone tumor in children and young adults. Despite the use of surgery and multi-agent chemotherapy, osteosarcoma patients who have a poor response to chemotherapy or develop relapses have a dismal outcome. Identification of biomarkers for active disease may help to monitor tumor burden, detect early relapses, and predict prognosis in these patients. In this study, we examined whether circulating miRNAs can be used as biomarkers in osteosarcoma patients. We performed genome-wide miRNA profiling on a discovery cohort of osteosarcoma and control plasma samples. A total of 56 miRNAs were upregulated and 164 miRNAs were downregulated in osteosarcoma samples when compared to control plasma samples. miR-21, miR-221 and miR-106a were selected for further validation based on their known biological importance. We showed that all three circulating miRNAs were expressed significantly higher in osteosarcoma samples than normal samples in an independent cohort obtained from the Children's Oncology Group. Furthermore, we demonstrated that miR-21 was expressed significantly higher in osteosarcoma tumors compared with normal bone controls. More importantly, lower expressions of miR-21 and miR-221, but not miR-106a, significantly correlated with a poor outcome. In conclusion, our results indicate that miR-21, miR-221 and miR-106a were elevated in the circulation of osteosarcoma patients, whereas tumor expressions of miR-21 and miR-221 are prognostically significant. Further investigation of these miRNAs may lead to a better prognostic method and potential miRNA therapeutics for osteosarcoma.

Keywords: biomarker; miRNA; osteosarcoma; plasma; prognosis.

Figure 1. Differential expression of circulating miRNAs in the discovery cohort

Heatmap and hierarchical clustering of the expressions (-dCq) of the 220 differentially expressed miRNAs in osteosarcoma (OS) and control samples (A). Red and green denote high and low expression, respectively. Scatter plots of the three miRNA candidates indicate relative miRNA expression, which is expressed as normalized linear expression values (2^-dCq) of each individual miRNA in the OS and control plasma samples (B). FC denotes fold change of OS/control samples.

Figure 2. Validation of the three miRNA candidates in the plasma samples of an independent cohort from the Children’s Oncology Group

Scatter plots to show higher miRNA expression levels of miR-21, miR-221 and miR-106a in the osteosarcoma (OS) plasma samples relative to the control samples of the validation set (A). Relative miRNA expression is expressed as normalized linear expression values (2^-dCq). Receiver Operating Characteristics analysis of the three miRNAs in discriminating the OS cases from the control cases (B). FC denotes fold change of OS/control samples.

Figure 3. Detection of the miR-21, miR-106a and miR-221 in osteosarcoma (OS) tissues

Scatter plots showing the tumor expression (-dCt) of miR-21, miR-106a and miR-221 in osteosarcoma tumor (n = 89) and normal bone (n = 5) samples as measured by qRT-PCR. (A) FC denotes fold change of OS/control samples. In-situ hybridization (ISH) 20× images showing the miR-21 expression (red) in OS tissues in formalin-fixed, paraffin embedded OS tissue microarray (B). The blue color indicates DAPI (4′,6-diamidino-2-phenylindole) stained nuclei. Left, middle and right panels show U6 (positive control), miR-21, and scrambled probe (negative control) images, respectively.

Figure 4. Kaplan-Meier analyses of miR-21 and miR-221 in the TARGET cases

miR-21 and miR-221 are significantly associated with overall (A) and event-free (B) survival when risk-stratified by 1st quartile of –dCt values. miR-21 and miR-221 are further independently prognostic from metastatic disease at diagnosis as well as significant in only the metastatic subpopulation for both overall (C) and event-free (D) survival. miR-21 alone also significantly stratifies patients presenting with no metastasis at diagnosis (D). All statistics presented were computed by log-rank tests.

Figure 5. Correlation of miRNAs with known prognostic factors and among themselves

The differential expression of miR-106a (A) and miR-21 (B) with metastasis at diagnosis and histologic response. FC denotes fold change of ME/NM and GR/PR samples. ME, NM, GR and PR denote metastasis, no metastasis, good response and poor response, respectively. The expressions of miR-21 and miR-221 were highly positively correlated to one another in osteosarcoma samples (C) whereas miR-106a exhibits negative correlation with both miR-21 (D) and miR-221 (E). R represents the Pearson correlation coefficient. Comparing the risk-stratified grouping with respect to miR-21 and miR-221 shows that most patients fall into the same risk group, and are statistically significant using Fisher’s exact test (F).