miRNA profiling identifies deregulated miRNAs associated with osteosarcoma development and time to metastasis in two large cohorts

Abstract

Osteosarcoma (OS) is an aggressive bone tumor primarily affecting children and adolescents. The etiology of OS is not fully understood. Thus, there is a great need to obtain a better understanding of OS development and progression. Alterations in miRNA expression contribute to the required molecular alterations for neoplastic initiation and progression. This study is the first to investigate miRNA expression in OS in a large discovery and validation cohort comprising a total of 101 OS samples. We established the signature of altered miRNA expression in OS by profiling the expression level of 752 miRNAs in 23 OS samples using sensitive LNA-enhanced qPCR assays. The identified miRNA expression changes were correlated with gene expression in the same samples. Furthermore, miRNA expression changes were validated in a second independent cohort consisting of 78 OS samples. Analysis of 752 miRNAs in the discovery cohort led to the identification of 33 deregulated miRNAs in OS. Twenty-nine miRNAs were validated with statistical significance in the second cohort comprising 78 OS samples. miRNA/mRNA targets were determined, and 361 genes with an inverse expression of the target miRNA were identified. Both the miRNAs and the identified target genes were associated with multiple pathways related to cancer as well as bone cell biology, thereby correlating the deregulated miRNAs with OS tumorigenesis. An analysis of the prognostic value of the 29 miRNAs identified miR-221/miR-222 to be significantly associated with time to metastasis in both cohorts. This study contributes to a more profound understanding of OS tumorigenesis, by substantiating the importance of miRNA deregulation. We have identified and validated 29 deregulated miRNAs in the - to our knowledge - largest discovery and validation cohorts used so far for miRNA analyses in OS. Two of the miRNAs showed a promising potential as prognostic biomarkers for the aggressiveness of OS.

Keywords: epigenetics; metastasis; miRNA; osteosarcoma.

Figure 1

A detailed outline of the different processes and analyses performed for the OS discovery and validation cohort. QC, quality control; diff., differential; btw., between.

Figure 2

Unsupervised hierarchical cluster analysis. The expression value (log₂) for each detectable miRNA (C _q < 37) was used to determine the clustering of the samples. Red: primary OS samples; yellow: OS cell lines; blue: OB cell lines.

Figure 3

Boxplots of the expression of nine differentially expressed miRNAs. Red: OS samples. *P < 0.05; **P < 0.001.

Figure 4

Network of validated target genes. Network of validated target genes associated with the terms ‘Cell Death and Survival, Cancer, Hematological System Development and Function’. The relation of 29 validated target genes (miRWalk2.0) assembled in a network associated with 11 miRNAs altered in OS in this study. Eight genes were downregulated and 21 upregulated. Red: downregulated; blue: upregulated. Gray lines: association between genes; Red lines: association between gene and miRNA; continuous line: direct association; dashed line: indirect association.

Figure 5

The prognostic value of the expression of miR‐221‐3p and miR‐222‐3p in function of time to metastasis. (A) Kaplan–Meier analysis of time to metastasis when subdividing the OS samples into high/low expression groups for miR‐221‐3p (based on the median expression). (B) Kaplan–Meier analysis of time to metastasis when subdividing the OS samples into high/low expression groups for miR‐222‐3p (based on the median expression). (C) Kaplan–Meier analysis of time to metastasis when combining the expression of both miR‐221‐3p and miR‐222‐3p for each patient. The discovery cohort and validation cohort have been combined for all three analyses. The Kaplan–Meier analysis of the high/low expression groups for the individual analyses of miR‐221‐3p and miR‐222‐3p contained 47 OS samples in each group. For the combined analysis of miR‐221‐3p/miR‐222‐3p expression, the number of OS samples in each group was as follows: 29 OS samples in high‐miR‐221‐3p/high‐miR‐222‐3p; 30 OS samples in low‐miR‐221‐3p/low‐miR‐222‐3p; 17 OS samples in high‐miR‐221‐3p/low‐miR‐222‐3p; and 18 OS samples in low‐miR‐221‐3p/high‐miR‐222‐3p. Data for the individual cohorts are shown in Fig. S4.