Prognostic value of microRNAs in osteosarcoma: a meta-analysis

BACKGROUND Osteosarcoma is the most common primary bone malignancy. We meta-analyzed the prognostic value of altered miRNAs in patients with osteosarcoma. METHODS Sources from MEDLINE (from inception to August 2016) and EMBASE (from inception to August 2016) were searched. Studies of osteosarcoma with results of miRNA and studies that reported survival data were included and two authors performed the data extraction independently. Any discrepancies were resolved by a consensus. The outcome was overall survival and event-free survival assessed using hazard ratios (HRs). RESULTS After reviewing the full text of 65 articles, 25 studies including 2,278 patients were eligible in this study. The pooled HR for deaths was 1.40 (95% confidence interval [CI] 1.01-1.94, p=0.04) with random-effects model (χ2=113.08, p<0.00001, I2=79%) for patients of osteosarcoma with lower expression of miRNA. However, the pooled HR for events was not significant (HR 0.97, 0.63-1.48, p=0.87, χ2=72.65, p<0.00001, I2=79%). In pathway analysis of miRNAs, miRNA449a, 199-5p, 542-5p have common target genes. CONCLUSIONS Expression level of miRNA in patients of osteosarcoma is important as a prognostic factor.


INTRODUCTION
Osteosarcoma is the most common primary bony malignancies and first leading cause of sarcoma-related deaths in children and young adults. It mainly occurs in metaphyseal area of distal femur and proximal tibia. Osteosarcoma is highly aggressive tumor and most likely metastasizes to the lung [1]. Even though the survival rate has modestly increased over the last 2 decades by implicating radiotherapy and neo-adjuvant chemotherapy, five-year survival rate of metastatic osteosarcoma still remains in the range of 15% to 30%. For localized osteosarcoma which can be totally resectable, the five-year survival rate increases up to 70% [2]. Hence, excavating novel prognostic biomarkers of osteosarcoma is strongly needed to detect tumor at early stage. They will contribute to select better treatment options in earlier stages and predict the outcome more accurately.
MicroRNAs (miRNAs) are a class of small noncoding RNAs, which could interfere translation of many proteins after gene transcription [3]. Previous studies revealed that they are implicated on a variety of physiological processes including cellular differentiation, apoptosis, angiogenesis and cell proliferation [4]. Additionally, miRNAs play a pivotal role in tumorigenesis that can act either as tumor suppressor genes or www.impactjournals.com/oncotarget/ Oncotarget, 2017, Vol. 8, (No. 5), pp: 8726-8737 Research Paper oncogenes [3,5]. Level of miRNAs expressed in several malignancies, including lung, kidney, liver, and cervical cancer were significantly different from that of normal tissue [6,7]. Recent researches have demonstrated that various miRNAs are also altered in osteosarcoma tissue or blood sample [8]. These discoveries may indicate that miRNA could be potential prognostic biomarkers of osteosarcoma. Therefore, we meta-analyzed the prognostic value of altered miRNAs in patients with osteosarcoma.

Study characteristics
The electronic search identified 289 articles. Non-English language articles (n=3), non-human studies (n=13), conference abstracts (n=34), and 174 studies that did not meet the inclusion criteria based on their title and abstract were excluded. After reviewing the full text of 65 articles, 25 studies including 2,278 patients were eligible for inclusion in the study. The detailed procedure is shown in Figure 1. The studies included in this metaanalysis reported the prognostic value of 26 miRNAs in patients with osteosarcoma. Twenty-five studies assessing 26 miRNA expressions were included in the meta-analysis . The prognostic value of miRNA expressions was assessed by analyzing overall survival (OS) in 2,278 patients, event free survival (EFS) in 1,569 patients. Visual inspection of the funnel plot suggested no evidence of publication bias ( Figure 2B and 3B). The study characteristics are summarized in Table 1.

miRNA expression and prognosis of osteosarcoma
To analyze the prognostic value of low expression of miRNA in osteosarcoma, forest plots with OS and EFS are depicted in Figure 2  ). Due to heterogeneity in pooled analysis of miRNA, we performed a subgroup analysis according to the expression level of miRNA. The HRs were calculated on the basis of low expression of miRNA, which means HR > 1 and < 1 implied poor and good prognosis for patients with low miRNA expression.

DISCUSSION
This study evaluated the prognostic value of miRNA in patients with osteosarcoma. As we described above, the combined HR of OS from the included studies was 1.40 (1.01-1.94, p=0.04), which provided that low expression of miRNAs was related with shorter OS in patients with osteosarcoma. However, the combined HR of EFS was not significant. Due to heterogeneity in pooled analysis of miRNA, we performed a subgroup analysis according to the expression level of miRNA. For a more detailed analysis, we performed a subgroup analysis according to miRNA expressions (high or low) related with poor prognosis. High expression of miRNA 210, 17-92 cluster, 128, 9, 214, 542-5p, 130b, 130a, and 199b-5p and low expression of miRNA 132, 145, 382, 133a, 26a, 340, 20a, 92a, 143, 451, 144, 22, 195, 124, 449a, 99a, 224 were associated with a poor prognosis of either deaths or events.
The possible role of miRNAs during osteosarcomagenesis is also investigated. Altered or aberrant expression of miRNA, its function during tumorigenesis, and potential target genes are summarized in Table 4. Nine miRNAs (miRNA 210, 17-92 cluster, 128, 9, 214, 542-5p, 130b, 130a, and 199b-5p) whose expression level correlated with poor prognosis function as an oncogene in osteosarcoma. Of that, miRNA 210 contributes its action in tumor initiation. miRNA 17-92 cluster, 214, 130b have been reported as they progress early stage osteosarcoma into advanced stage. And miRNA 9, 130a are associated with metastasis. On the contrary, 17 miRNAs (132, 145, 382, 133a, 26a, 340,  20a, 92a, 143, 451, 144, 22, 195, 124, 449a, 99a, 224) function as a tumor suppressor gene, which expressed low level when they were related with worse outcome. Among these miRNAs, miRNA 382, 133a, 26a, 340, 143, 144, 195 have shown an anti-metastatic activity. Especially, miRNA 133a, 143 could be utilized as a future therapeutic target for lung metastasis. Moreover, miRNA 132, 145, 92a might act as a therapeutic standard for chemotherapy. This is the first meta-analysis that evaluate the prognostic value of miRNAs in osteosarcoma. We demonstrated the association of the outcomes (OS, EFS) of osteosarcoma and level of miRNA expression. In addition, we measured HRs as the effectiveness endpoint in the current study. We did not measured the odds ratios or risk ratios whether these values represent only crosssectional cumulative estimate and do not including the time factor [35]. Hence, HRs are most appropriate for our report which analyze time-to-event outcomes. The type of samples was homogenous. All of the samples which were experimented in included studies were obtained from tumor tissue of osteosarcoma patients. This might be increased reliability and decreased heterogeneity of our study. We excluded studies using serum or plasma because there were a few studies about serum or circulating miRNAs in patients with osteosarcoma [36,  37]. However, there are increasing interest in circulating miRNAs due to their comfortable accessibility by drawing patient's blood compared to miRNAs of tissue from operation and they could be easily repeated and monitored through.
From the pathway analysis, we found that several common targets and pathways of miRNA199b-5p, 449a and 542-5p, for example, fatty acid biosynthesis (target gene: Fatty Acid Synthase (FASN)) ( Table 3). FASN is known as a therapeutic target in the treatment of A B   [38]. Pathways related to other malignancies were found in this analysis, such as breast cancer, and prostate cancer [38][39][40][41]. The target genes of miRNAs in various cancers may overlap, which suggest that miRNAs could be promising therapeutic targets in cancer treatment.
In this study, we investigated the prognostic value of miRNA in patients with osteosarcoma using a meta-analysis approach. Finally, we concluded that decreased miRNA expression in tumor tissue is associated with worse outcome of patients with osteosarcoma. However, we should also pay attention to increased expression of several miRNAs in tumor tissue of osteosarcoma. Furthermore, prospective studies with a large sample size are warranted to clarify the prognostic role of miRNAs in osteosarcoma.

Data search and study election
We performed a systematic search if MEDLINE (from inception to August 2016) and EMBASE (from inception to August 2016) for English-language publications using the keywords "osteosarcoma", "miRNA", "prognosis". All searches were limited to human studies. The inclusion criteria were studies of osteosarcoma that reported the results of miRNA expressions and survival data. Reviews, abstracts, and editorial materials were excluded. Two authors performed the searches and screening independently, and discrepancies were resolved by consensus.

Data extraction and statistical analysis
Data were extracted from the publications independently by two reviewers, and the following information was recorded: first author, year of publication, country, miRNA expression analyzed, number of patients, staging, and end points. The primary outcome was OS, which was defined as the time from the initiation of therapy until death from any cause. The secondary end point was EFS. Data regarding diseasefree survival (DFS), progression-free survival (PFS), recurrence-free survival (RFS) were obtained from the included studies, and were redefined as EFS, which was measured from the date of initiation of therapy to the date of recurrence or metastasis [42,43].
The effects of miRNA expressions on survival were assessed using hazard ratios. Survival data were extracted following a methodology suggested previously [44]. A univariate HR estimate and 95% CIs were extracted directly from each study, if provided by the authors. Otherwise, p values of the log-rank tests, 95% confidence intervals, number of events, and numbers of patients at risk were extracted to estimate the HR indirectly. Survival rates calculated from Kaplan-Meier curves were read using Engauge Digitizer version 3.0 (http://digitizer.sourceforge. net) to reconstruct the HR estimate and its variance, assuming that patients were censored at a constant rate during followup. The HRs were calculated on the basis of low expression of miRNA, which means HR > 1 and < 1 implied poor and good prognosis for patients with low miRNA expression. Heterogeneity among studies was assessed using χ 2 tests and I 2 statistics, as described previously [45]. Funnel plots were used to assess publication bias [46]. p values < 0.05 were considered to be statistically significant. The data from each study were analyzed using Review Manager (RevMan, Version 5.3 Copenhagen: The Nordic Cochrane Centre, The Cochrane Collaboration, 2014). miRNA target genes and related pathways were predicted by DIANA-mirPath v3.0 (http://www.microrna.gr/miRPathv3) based on Tarbase, KEGG and GO [34] to interpret functional value of miRNAs.

CONFLICTS OF INTEREST
None.