Prognostic Value of Circular RNA ciRS-7 in Various Cancers: A PRISMA-Compliant Meta-Analysis

Background Circular RNAs (circRNAs) have been shown to be involved in tumorigenesis. As a member of circRNAs, ciRS-7 is thought to be a negative prognostic indicator in multiple types of cancer. The present study aimed to comprehensively explore the value of ciRS-7 in tumor malignancy. Materials and Methods. A systematic review of PubMed, Web of Science, and the Cochrane library was carried out to examine the related studies. The pooled odds ratios (ORs) and hazard ratios (HRs) with 95% confidence intervals (95% CIs) were calculated from the available publications by STATA 12.0. Subgroup analysis, publication bias, sensitivity analysis, and meta-regression were conducted. Results This meta-analysis included 1,714 patients from 13 cohorts. The results suggested that high ciRS-7 expression was significantly associated with overall survival (OS) (HR = 2.17, 95% CI = 1.50–3.15, P < 0.001) in various cancers. Stratified analyses indicated that elevated levels of ciRS-7 appeared to be a powerful prognostic biomarker for patients with non-small-cell lung cancer (NSCLC) (HR: 2.50, 95% CI: 1.07–6.07, P < 0.001) in various cancers. Stratified analyses indicated that elevated levels of ciRS-7 appeared to be a powerful prognostic biomarker for patients with non-small-cell lung cancer (NSCLC) (HR: 2.50, 95% CI: 1.07–6.07, P < 0.001) in various cancers. Stratified analyses indicated that elevated levels of ciRS-7 appeared to be a powerful prognostic biomarker for patients with non-small-cell lung cancer (NSCLC) (HR: 2.50, 95% CI: 1.07–6.07, P < 0.001) in various cancers. Stratified analyses indicated that elevated levels of ciRS-7 appeared to be a powerful prognostic biomarker for patients with non-small-cell lung cancer (NSCLC) (HR: 2.50, 95% CI: 1.07–6.07, P < 0.001) in various cancers. Stratified analyses indicated that elevated levels of ciRS-7 appeared to be a powerful prognostic biomarker for patients with non-small-cell lung cancer (NSCLC) (HR: 2.50, 95% CI: 1.07–6.07, Conclusions High expression of ciRS-7 has a significant correlation with the high stage in various cancers, and ciRS-7 is intimately associated with an adverse OS in numerous cancers. Thus, ciRS-7 may act as a potential biomarker for the development of malignancies.


Introduction
Circular RNAs are a type of noncoding RNAs that have a unique single-stranded closed ring structure, which were discovered in viroid via electron microscopy as early as in 1976, and originally thought to be that with incorrect splicing of RNA [1,2]. Researchers have been focusing on circRNAs since the application of high-throughput RNA sequencing and bioinformatics analyses, and thousands of human circRNAs have been identified so far [3,4].

Materials and Methods
is meta-analysis was performed by following the PRISMA guidelines [31]. Systematic literature research was independently performed by two authors through electronic databases, including PubMed, Web of Science, and the Cochrane Library for eligible studies published until March 18, 2019. Studies were selected using the following terms: "CDR1-AS" or "cdr1as" or "ciRS-7" and "tumor" or "cancer" or "carcinoma". Citation lists of review articles were manually searched for finding potentially eligible studies.

Study Selection. Two authors (Guangwei Tian and Lin
Guan) independently evaluated all of the included studies, and any disagreement was resolved by consultation with the supervisor (Nan Li). e following inclusion criteria were considered: (a) studies investigating on patients with any type of cancers; (b) studies in which the patients were grouped by the expression of ciRS-7 via validating techniques; (c) studies reporting the sufficient information association between ciRS-7 expression and survival, in which the hazard ratio (HR) with 95% confidence interval (95% CI); (d) studies reporting the relationship between ciRS-7 expression and clinical stage of cancer patients. Exclusion criteria were as follows: (a) studies without sufficient or usable data; (b) duplicative publications; (c) reviews, case reports, editorials, and conference abstracts; (d) laboratory studies only investigating the molecular function of ciRS-7.

Data Extraction and Quality Assessment.
e primary outcome used in this meta-analysis was overall survival. e secondary outcome was clinical stage. e following information was extracted from each study: first author name, year, country, tumor type, sample size and type, expression of ciRS-7, detection technique, follow-up duration, and HRs of ciRS-7 expression for overall survival (OS) with 95% CIs. In all the articles included in our study, the expression of ciRS-7 was detected in tumor tissues. e Newcastle-Ottawa Score (NOS) quality assessment system adopted for casecontrol studies was utilized to examine the methodological quality of studies [32]. Included studies were scored according to three broad perspectives, selection, comparability, and exposure, through the star scoring method with a range of lowest quality (0 star) to the highest quality (9 stars). A score of ≥6 indicated high quality [33].

Statistical Analysis.
Meta-analysis was performed using Stata 12.0 software (Stata, College Station, TX, USA). e prognostic value of ciRS-7 expression in patients with various types of cancers was analyzed by the pooled HR associated with the 95% CI. If the HR and 95% CI were not directly reported, we extracted them from Kaplan-Meier curves [34]. ORs were extracted for evaluating the association between the expression of ciRS-7 and tumor stage. Higgins I 2 statistic was performed for evaluating the heterogeneity [35]. If heterogeneity was present (I 2 > 50% or P < 0.05), a random-effect model was applied [36]; otherwise, a fixed effect model was used [37]. Publication bias was conducted by Begg's test [38]. Sensitivity was also performed to identify the effect of the individual study data on summary for HRs and ORs.

Characteristics of Studies.
ere were 85 studies from a database search, and 74 irrelevant studies and duplicates were excluded after reviewing the titles and abstracts. On the basis of the inclusion and exclusion criteria, ultimately, 13 cohorts from 11 studies and 1,714 patients were included in this meta-analysis. A flowchart of the studies screening process is shown in Figure 1, and the detailed characteristics are summarized in Table 1.
When we regarded the sample size, analysis method, and different cutoff values, there was no significant change for pooled HR in each subgroup ( Table 2). Due to significant heterogeneity in analysis, we detected the source of heterogeneity by multivariate meta-regression analysis. It was found that sample size (P � 0.066), cancer type (P � 0.860), ethnicity (P � 0.111), analysis method (P � 0.772), cutoff value (P � 0.208), and study design (P � 0.924) were not the sources of heterogeneity (Table 2).
A sensitivity analysis was conducted to assess whether one study would affect the overall results and to confirm the stability of results. In Figure 3, the results were relatively stable. e funnel plot was asymmetry ( Figure 4). Additionally, Begg's test was used for quantitative analysis, and P value was 0.115, which indicated no publication bias in this meta-analysis.

Association between ciRS-7 and Clinical
Stage. 6 cohorts from 5 studies comprising 715 patients provided valid data for the correlation between ciRS-7 and clinical stage [19,22,24,29,30]. Among these 6 cohorts, there were 3 types of cancers (2 CRC, 2 NSCLC, and 2 ESCC). e result showed that an increased ciRS-7 expression was associated with a higher clinical stage (OR � 2.30, 95% CI: 1.69-3.13, P < 0.001). Cancer type by subgroup analysis indicated that the positive correlation between ciRS-7 and clinical stage was found in CRC, NSCLC, and ESCC ( Figure 5). Significant heterogeneity was not observed in this statistic due to I 2 � 40.6%. Begg's test was performed, and no significant publication bias was found for the clinical stage (P � 1.000) ( Figure 6). Sensitivity analysis was performed through the sequential omission of every single publication, and the results were not significantly altered (Figure 7).

Discussion
Despite the advancement in treatment modalities, cancer is still one of the leading causes of death worldwide. It is estimated that there will be 22.2 million new cancer cases in 184 countries by 2030 [39]. erefore, novel biologicalspecific biomarkers are urgently needed for cancer early detection. Besides prognosis, biomarker is an important priority, which can affect clinical decision-making and the overall results.
Increasing evidence demonstrated that circRNAs are attractive candidates for diagnostic and prognostic biomarkers [40,41], and among them, ciRS-7 stands out as an oncogenic circRNA at least for a subset of cancer types. Meanwhile, in some studies, ciRS-7 also exerts anti-oncogenic functions [42,43]. Although researches on ciRS-7 develop fast, the prognostic role of ciRS-7 in cancer is still uncertain. For that reason, the study was conducted, and to our knowledge, no meta-analyses evaluated the correlation     In subgroup analysis, ciRS-7 showed inconsistent prognostic effects in different cancer types. In NSCLC, CRC, and GC, high expression of ciRS-7 was found to be associated with worse prognosis. While in "others" subgroup comprising ESCC, BC, LSCC, and CAA, the number of studies for each cancer type was only one, and the result showed HR � 1.90 (1.00-3.61), P � 0.05; therefore, we cannot reach a conclusion statistically that ciRS-7 has a role in predicting in this group. More studies are needed for credible conclusion for these four cancer types. e ethnic difference in cancer has been identified and investigated in many studies [44][45][46]. As is the case with cancer type, no significant association was observed between ciRS-7 expression and Caucasian populations due to small sample size, which may be buttressed by meta-regression. According to the results of meta-regression, we found that sample size is most likely to be the greatest impact on heterogeneity among all factors, because although P value of the sample size is 0.066, it is the lowest among all the factors.
With the closed ring structure, circRNAs are more stable, not easy to be degraded by RNase and high abundance than mRNA [56], which make circRNAs have more advantages in the development and application as a new clinical diagnostic marker. In view of the current researches, there are many prospects for the clinical application of ciRS-7. Recent studies found that circRNAs can be detected in blood and exosomes of cancer patients [57][58][59]  future, it will better reflect its clinical application value, if ciRS-7 can be detected in a noninvasive way. Furthermore, ciRS-7 is expected to become an emerging drug's target. Studies show ciRS-7 overexpression can activate the EGFR pathway, which is one of the most important targets of NSCLC, to induce tumor cell growth [21,22]. ese findings will provide new opportunities for ciRS-7 to be developed as a targeted drug in the future.
Several limitations in this study should be carefully considered. Firstly, the number of included studies was limited which may weaken the power of the results. Secondly, not all different factors were analyzed in this metaanalysis because it was literature-based, and we only got the summarized data rather than individual patient data.
irdly, some studies had not reported the cutoff values of ciRS-7, and there was no consensus for calculating the cutoff value. Fourthly, we extracted HR and 95% CI from the survival curve rather than directly from the original study for one study, which might impact the results. Finally, although publication biases on OS and clinical stage were not found by Begg's test, there may be an underlying publication bias as studies with positive results could be more likely to be published than those with negative results. erefore, our meta-analysis still needs further investigation.
In conclusion, this meta-analysis showed that upregulation of ciRS-7 is associated with adverse survival and higher clinical stage in multiple types of cancers. ciRS-7 may be a new star of the future drug target for fighting against      tumor growth or used as a biomarker for predicting the prognosis of cancer patients. Larger sample sizes and welldesigned studies are needed in the future.

Conflicts of Interest
e authors report no conflicts of interest in this work.

Authors' Contributions
Nan Li and Guang Li designed the study. Guangwei Tian and Lin Guan performed the Statistics. Nan Li and Guangwei Tian wrote the paper. Zihui Wang edited the manuscript.