Clinical value of lncRNA ZFAS1 in various cancers: an updated meta-analysis of 2263 cancer patients

Background: LncRNA ZNFX1 antisense RNA 1 (ZFAS1) is largely reported to be abnormally expressed in various human cancers and associated with unfavorable prognosis. So this study aims to further investigate its clinical value in multifarious cancers. Methods: Databases of PubMed, Web of science, Cochrane library, Embase, Chinese National Knowledge Infrastructure and Wanfang was retrieved to collect eligible studies until October 22, 2020. The clinical role of ZFAS1 was comprehensively assessed by STATA 12.0. Additionally, the CeRNA network in which ZFAS1 involved was constructed. And function analysis of targeted mRNAs was performed by R package “cluster proler”. Results: 23 studies of 2263 cancer patients met the inclusion criteria in this meta-analysis. The comprehensive results suggested that patients with elevated ZFAS1 expression got poor overall survival (OS) (HR=1.80, 95%CI: 1.50-2.10, P ≤ 0.001), and they are prone to lymph node metastasis (OR=2.67, 95%CI: 1.73-4.14, P ≤ 0.001), distant metastasis (DM) (OR=3.45, 95%CI: 1.61-7.39, P=0.001), advanced TNM stage (HTS) (OR=2.99, 95%CI: 1.97-4.54, P ≤ 0.001), larger tumor size (LTS) (OR=1.54, 95%CI: 1.05-2.24, P ≤ 0.001), poor histologic grade (PHG) (OR=1.57, 95%CI: 1.09-2.26, P=0.015). The CeRNA network and function analysis revealed the potential mechanism which ZFAS1 may involve in cancers. Conclusion: The current analysis revealed that elevated ZFAS1 expression is common in various cancer and was signicantly associated with poor OS and clinical characteristics, thus it may function as prognostic biomarker in various cancer. Nevertheless, larger well-designed clinical research a is still warranted to verify our ndings. or odds ratios (ORs) with 95% condence interval (CI) were used to assess the correlation of prognosis and clinicopathological features with ZFAS1 expression, respectively. Heterogeneity test was performed by Cochran’s Q test and I 2 statistic. Random-effect model was used if the heterogeneity was signicant (I 2 ≥ 50%, and P<0.05), instead, xed-effect model will be chosen. The robustness and heterogeneity source of results was tested via sensitivity analysis and subgroup analysis, respectively. Potential publication bias was assessed by and Begg’s test(cid:0) and Pr > |z|(cid:0)0.05 indicates no publication bias existed. Statistically signicance was dened if P<0.05.


Background
Cancer, as one of the leading causes of death, with high morbidity, has continually imposed heavy economical burdens on society and families. According to the latest report, 1,806,590 new cancer cases and 606,520 cancer-related deaths will occur in the United States in 2020 [1]. This severe situation has made numerous researchers to concentrate on the study of early diagnosis and e cient treatment of cancer patients. Indeed, the death rate has gradually decline with great advancement made in diagnosis and therapy of cancer, but the 5-year survival rate of cancer patients is still disappointing [2]. Thus, it's of great need to seek useful therapeutic and prognostic biomarker for the improvement of cancer patients' prognosis.
Long non-coding RNA (lncRNA), with length of more than 200 nucleotides, is a classic kind of non-coding RNA which lacks of protein coding ability. Existing evidence have proved that lncRNA can act as a signi cant gene regulator and participate in various biological process, including epigenetic regulation, cell cycle, alternative splicing, chromatin modi cation and so on [3][4][5]. Importantly, it's also con rmed that lncRNA can function as oncogene or tumor-suppressing gene through control of tumorigenesis, invasion, progression and metastasis in multiple cancers [6][7][8].
LncRNA ZNFX1 antisense RNA 1 (ZFAS1, also known as Zinc nger antisense 1), which transcribed from the antisense orientation of zinc nger NFX1-type containing 1(ZNFX1), is located on chromosome 20q13.13 [9]. Growing studies have reported abnormal expression of ZFAS1 in diverse cancers, including pancreatic cancer [10], hepatocellular carcinoma [11], melanoma [12], esophageal squamous cell carcinoma [13] and so on. Moreover, elevated ZFAS1 expression is reported to be related to poor prognosis and clinical characteristics in various cancers [14][15][16], which indicate that ZFAS1 may serve as a promising biomarker for the prognosis and therapy of cancer patients. However, limited by insu cient sample size in single study and inconsistent ndings among studies, the clinical role of ZFAS1 in cancer deserve further study to con rm. So we performed this updated meta-analysis to comprehensively evaluate the association between ZFAS1 expression and OS and clinicopathological characteristics of cancer patients.

Study selection
The reporting of this meta-analysis complied with the Systematic Reviews and Meta-Analyses (PRISMA) guidelines [17]. Selection of eligible studies was performed in databases of PubMed, Web of Science, Chinese National Knowledge Infrastructure and Wanfang up to October 2020. The search terms were: ("lncRNA ZFAS1" or "lncRNA ZNFX1 antisense RNA 1") AND ("cancer" or "carcinoma" or "tumor" or "malignancy").

Inclusion and exclusion criteria
Inclusion criteria were as follows: (a) the ZFAS1 expression of human tumor tissues were detected via method of qRT-PCR; (b) cancer patients were divided into high expression group and low expression group according to median or average expression level of ZFAS1; (c) study provided survival information and clinicopathological parameters of cancer patients according to different ZFAS1 expression.
Studies will be excluded if they meet one or more criteria as following: (a) reviews, case reports, letters, meta-analysis, conference articles and laboratory study; (b) study without survival information or clinicopathological features of patients; (c) high or low ZFAS1 expression were not de ned.

Data extraction and quality assessment
In strict accordance with inclusion and exclusion criteria, quali ed original literature data assessment and related data extraction were independently completed by two investigators (Zhou Liu and Biyun Lu), and any disagreement encountered is resolved through consultation with third investigator (Kaiyuan Chen). The following information was extracted from each literature: rst author, publication year, country, tumor type, detection method, total number of cases, survival information and clinicopathological parameters. If there were both univariate and multivariate analysis in the include study, the results of multivariate analysis will be applied. If only the survival curve is provided in the study, then Engauge Digitizer v10.4 software will be employed to extract HR and 95% CI. Quality assessment was conducted through Newcastle-Ottawa-Scale (NOS) criteria. High-quality studies were de ned as the NOS score ≥6.

CeRNA network construction
By using starbase v2.0 (http://starbase.sysu.edu.cn/), the targeted miRNA of ZFAS1 was predicted. And the targeted mRNA mining of predicted miRNA was performed via miRWalk with the lter set as 0.95, 3UTR, TargetScan, miRDB, miRTarBase. Then the targeted mRNAs were applied to perform Gene Ontology (GO) and Kyoto Encyclopedia of Genes and Genomes (KEGG) pathway analysis by R package "cluster pro ler".
Statistical analysis Stata 12.0 software (StataCorp, College Station, Texas USA) was used to analyze the extracted data. Pooled hazards ratios (HRs) or odds ratios (ORs) with 95% con dence interval (CI) were used to assess the correlation of prognosis and clinicopathological features with ZFAS1 expression, respectively. Heterogeneity test was performed by Cochran's Q test and I 2 statistic. Random-effect model was used if the heterogeneity was signi cant (I 2 ≥50%, and P<0.05), instead, xed-effect model will be chosen. The robustness and heterogeneity source of results was tested via sensitivity analysis and subgroup analysis, respectively. Potential publication bias was assessed by and Begg's test and Pr > |z| 0.05 indicates no publication bias existed. Statistically signi cance was de ned if P<0.05.

ZFAS1 expression and OS
In this part, 16 studies of 1637 patients were included. Among them, 5 studies directly provided HR and 95%CI, and the survival data from the other 11 studies were indirectly extracted from survive curve. The results revealed that patients with high ZFAS1 expression underwent poor OS (OR=1.80, 95%CI: 1.50-2.10, P≤0.001, Figure 2). It's found that many different characteristics were existed among the study of OS, therefore, to exclude potential bias, subgroup analysis on the basis of HR methods, sample size, follow-up month and tumor type indicate that high ZFAS1 expression was correlated to poor OS in the above 8 subgroups (Figure 3), which indicates there exists no potential bias in study about OS. Additionally, sensitivity analysis and Begg's test further con rms the robustness of this results (Figure 4). ZFAS1 expression and clinical characteristics OR and 95%CI was applied to evaluate the relationship between ZFAS1 expression and various clinical characteristics. As manifested in  Figure S1a), but not to age (OR=1.06, 95%CI: 0.89-1.26, P=0.499, Figure S1b) and gender (OR=1.13, 95%CI: 0.94-1.36, P=0.209, Figure S1c).

CeRNA network
Growing studies have found out that CeRNA network plays a signi cant role in the regulation of lncRNA to cancer. To explore the potential mechanism of ZFAS1 in cancer development, the CeRNA network centered on ZFAS1 was constructed. As shown in Figure 6, this CeRNA network which contains 95 nodes and 103 edges manifested lots of ZFAS1/miRNA/mRNA axis that may participate in the regulation of occurrence and development of cancer. GO and KEGG analysis indicated the biological functions and pathway the targeted mRNAs may involve in. Those results above together provide meaningful clues for future mechanism research of ZFAS1 in cancer.

Discussion
Cancer has always been a great concern to human health. Due to its indetectable onset, when diagnosed, most cancer patients tend to have advanced clinical stage, LNM and DM, and current cancer therapy can't effectively stop its development, which make them unfortunately miss the optimal treatment time and have poor prognosis. Therefore, the identi cation of prospective biomarker for the diagnosis, treatment and prognosis of cancer is urgent.
Recently, plenty of studies indicated that several lncRNAs may serve as a promising prognostic biomarker for cancer patients, such as SNHG1 [32], linc00152 [33], PCAT-1 [34], FAM83H-AS1 [35] and so on. And ZFAS1 is exactly one of them. In this meta-analysis, our results showed that increased ZFAS1 expression can predict poor prognosis in diverse cancer patients (OR=1.80, 95%CI: 1.50-2.10, P≤0.001). Furthermore, subgroup analysis, sensitivity analysis and Begg's test performed in this study showed that no potential heterogeneity exists among included studies, which further testi ed the stability of our results. Meanwhile, elevated ZFAS1 expression was also associated with clinical characteristics, including HTS (OR=2. Since ZFAS1 was rstly discovered by Askarian-Amiri in breast cancer [9], mounting evidence have demonstrated that the dysregulation of ZFAS1 expression plays a critical role in the proliferation, invasion and metastasis of various cancers. And the carcinogenic molecular mechanism of ZFAS1 has been partly clari ed that it usually functions as CeRNA of miRNA to regulate downstream mRNA expression to affect manifold biological behaviors of tumor cell. Duan et al. found that ZFAS1 was obviously overexpressed in HCC and connected with worse prognosis of HCC patients, and silencing of ZFAS1 can suppress cell proliferation, invasion, migration and epithelial-mesenchymal transition (EMT) through regulation of the miR-624/MDK/ERK/JNK/AKT signaling pathway [11]. Several researchers have demonstrated that ZFAS1 was upregulated in CRC and can serve as a prognostic factor for CRC patients. Among them, Chen [10], and knockdown of ZFAS1 can reduce tumor progression in ovarian cancer via sponging of miR-150-5p to modulate RAB9A expression [29]. Xiao et al. noted that ZFAS1 acts as CeRNA of miR-193a-3p to control HB growth by targeting RALY via HGF/c-Met Pathway. Moreover, the CeRNA network build in our study may lead to another novel direction on the path of cancer mechanism research about ZFAS1.
Notably, several researchers have explored the role of ZFAS1 in cancers before our meta-analysis, but obvious difference can be observed in our study. Firstly, a total of 23 studies of 2263 cancer patients were included in our study, the most sample size in studies of Dong et al. [37] is only 12 studies of 1075 patients, therefore, larger sample size in our study make our ndings more convincible. Secondly, the prognostic role of ZFAS1 in more cancer types including PAAD, ESCC, HB, CCA, PTC, PCa, ccRCC, NPC and bladder cancer were explored in current study, which make our results more applicable. Thirdly, we rstly revealed that high ZFAS1 expression is signi cantly with DM. Lastly, we rstly constructed a CeRNA network in which ZFAS1 participated, which contribute to the discovery of new regulation mechanism.
However, it's undeniable that some limitations can be found in our study. Firstly, patients included in our study are all Chinese, which may restrict the applicability of our ndings in other race. Secondly, some HRs were extracted from survive curve, which may affect the reliability of our results. Lastly, the cut-off value de nes high and low expression of ZFAS1 were not consistent among studies, which may cause potential heterogeneity. Consequently, more studies of high-quality and diverse ethnicity are required to further verify our ndings.

Conclusions
In short, our ndings indicated that high ZFAS1 expression is signi cantly associated with poor OS and clinicopathological features including advanced tumor stage, lymph node metastasis, larger tumor size, poor differentiation and distant metastasis in cancer patients, thus it may be quali ed as a prognostic biomarker for cancer patients.

Declarations
Ethics approval and consent to participate Not applicable.

Consent for publication
All the listed authors agree to publish.

Availability of data and materials
All data are included in this article.

Competing interests
The authors declare no underlying con icts of interest.   Forest plot of the association between ZFAS1 expression and overall survival.

Figure 3
Forest plot of association between ZFAS1 expression and overall survival analyzed by subgroup analysis. a Subgroup analysis by tumor type b Subgroup analysis by HR extraction methods; indirectly, HR extracted from survival curve; directly, HR extracted from paper. c subgroup analysis by sample size. d Subgroup analysis by follow-up months Analysis of stability and heterogeneity of included studies about overall survival. a Sensitivity analysis. b Begg's test.

Figure 5
Forest plot of relation between ZFAS1 expression and clinicopathologic features. a Relation between ZFAS1 expression and HTS. b Relation between ZFAS1 expression and LTS. c Relation between ZFAS1 expression and LNM. d Relation between ZFAS1 expression and PHG. CeRNA network of ZFAS1 participated in. Figure 8