LncRNA Taurine Upregulated Gene 1 as a Potential Biomarker in the Clinicopathology and Prognosis of Multiple Malignant Tumors: A Meta-Analysis

Background The lncRNA taurine upregulated gene 1 (TUG1) is a recently identified potential biomarker in cancer. However, its prognostic role in various cancers is inconsistent among published data. We conducted this meta-analysis to comprehensively confirm the prognostic effect of TUG1 in malignant tumors. Methods We systemically analyzed the prognostic-predictive capacity of TUG1 through amplifying sample sizes and cancer types. STATA 12.0 was applied for this meta-analysis. Results A total of 57 eligible studies were included in our meta-analysis. The pooled results suggested that overexpression of TUG1 was significantly correlated with unfavorable overall survival (OS) (HR = 1.70, p < 0.001), shorter recurrence-free survival (RFS) (HR = 2.40, p ≤ 0.001), and shorter event-free survival (EFS) (HR = 1.88, p < 0.001) in patients with cancer. In the subgroup analysis by cancer type, elevated TUG1 expression was associated with poorer survival in patients with gastrointestinal cancer, urinary tumors, gynecological tumors, hematological tumors, and osteosarcoma. However, high expression of TUG1 in respiratory tumors indicated a better prognosis. There was no correlation between high TUG1 expression and OS in patients with head and neck neoplasms or melanoma. Additionally, overexpression of TUG1 was found to be correlated with low-grade tumor differentiation, advanced tumor stage, positive lymphatic metastasis, and positive distant metastasis. Conclusions High TUG1 expression correlates with poor prognosis and advanced clinicopathological features, verifying the prognostic-predictive capacity of TUG1 in tumors, especially in gastrointestinal cancer, urinary tumors, gynecological tumors, hematological tumors, and osteosarcoma. Meanwhile, the prognostic role of TUG1 in respiratory tumor may be opposite to other tumors.


Introduction
Long noncoding RNAs (lncRNAs) are an emerging class of vital regulators participating in various biological functions and disease processes to different degrees [1,2]. Nextgeneration sequencing has revealed that specific lncRNAs are mutated or aberrantly expressed in cancers, and the specific role of lncRNAs in different tumors is yet to be annotated [3].
The lncRNA taurine upregulated gene 1 (TUG1) has been reported to exert oncogenic or tumor suppressive function in cancer through altering cancer-related gene expression at the transcriptional level. According to previous studies, TUG1 was found to be upregulated and oncogenic in a broad spectrum of cancers, including colorectal cancer, bladder cancer, esophageal squamous cell carcinoma, and osteosarcoma [4][5][6][7]. Meanwhile, in some types of breast cancer, non-smallcell lung cancer, and glioma, TUG1 was expressed at a low level when compared with noncancerous tissues and acts as a tumor suppressor [8][9][10]. Meta-analyses have attempted to demonstrate the potential diagnostic or prognostic role of TUG1 in cancer, but the conclusions were not consistent [11][12][13][14][15][16][17][18]. In recent years, studies examining TUG1 expression have been conducted through high-throughput wholegenome sequencing or quantitative real-time polymerase chain reaction (qRT-PCR), and we carried out the current meta-analysis to evaluate the role of TUG1 in tumors by expanding the number of samples and tumor types. This research may provide additional evidence for TUG1 in predicting the prognosis of tumors.

Materials and Methods
2.1. Literature Search Strategy. Until January 15, 2021, relevant literature concerning the expression of the lncRNA TUG1 in cancer was extracted from databases including PubMed, Embase, and Web of Science, together with three Chinese databases: China National Knowledge Infrastructure (CNKI), Wanfang, and Weipu. Key terms and all possible combinations were as follows: 'taurine upregulated gene 1 OR TUG1' AND 'cancer OR tumor OR neoplasm OR carcinoma.' The reference lists of all primary studies were also examined to identify additional eligible studies.

Study Selection.
All eligible literature included in our meta-analysis met the following criteria: (1) The expression of the lncRNA TUG1 was measured in human tumors, and patients were grouped according to the expression levels of TUG1. (2) Assessment of the relationship between TUG1 expression and overall survival (OS), progression-free survival (PFS), disease-free survival (DFS), recurrent-free survival (RFS), event-free survival (EFS), or clinicalpathological parameters such as tumor differentiation, tumor stage, and metastasis. (3) Sufficient information was provided to estimate the hazard ratio (HR) or odds ratio (OR) and their 95% confidence intervals (CIs).
The articles were excluded if they had the following characteristics: (1) letters, reviews, case reports, and conference abstracts without original data; (2) laboratory studies conducted at the cellular level only; (3) lack of available data or survival curves to compute HRs, ORs, or the corresponding 95% CIs, and (4) multiple duplicate publications with overlapping populations, excluding the smaller sample cohort. Wang) independently extracted data and assessed study quality using a standardized form. Any discrepancy was arbitrated by a third reviewer (Qi Huang). The following characteristics were retrieved: first author's name, publication year, country of patients' origin, tumor type, sample size, number of patients in the TUG1 level group, tumor stage, detection method of TUG1 expression, survival data (obtained directly or extracted from Kaplan-Meier survival curve), clinical-pathological data, and Newcastle-Ottawa Quality Assessment Scale (NOS) score.

Data Extraction and Assessment of Study Quality. Two investigators (Jingjing Wu and Hui
The quality assessment of eligible studies was in accordance with the NOS. Our quality score was judged on three sections: selection, comparability, and exposure or outcome. With a mean score of 6.9 from enrolled studies, we defined studies scored 7 or above as high quality. 2.4. Statistical Analysis. All data analysis was performed using STATA software version 12.0 (Stata Corporation, College Station, TX, USA). We calculated the pooled HRs and the 95% CIs of the included articles to assess the impact of TUG1 on patient prognosis and clinical-pathological characteristics. OS, PFS, DFS, RFS, and EFS were all included in outcome analyses. HRs and their corresponding 95% CIs described in the literature were adopted directly. Otherwise, they were extracted from Kaplan-Meier curves by Engauge Digitizer version 4.1 (http://digitizer.sourceforge.net/). Additionally, we computed the ORs and their 95% CIs to explore the correlation between TUG1 expression and the clinicalpathological parameters of all tumors. In our analysis, an HR > 1 indicated that a high expression of TUG1 was an unfavorable factor in cancer, and an OR > 1 implied a worse parameter correlated with elevated TUG1 expression. Heterogeneity assessment was conducted by a Chi-squarebased Q statistical test and Higgins I-squared statistic. When the inconsistency index ðI 2 Þ ≥ 50% or p < 0:10, this indicated that there was substantial heterogeneity among the studies, and a random effects model was applied. When I 2 < 50% or p < 0:10, a fixed effects pattern was used. Sensitivity analysis was performed to assess the robustness of the overall results. Begg's test was conducted to determine the potential publication bias, with p < 0:05 indicating a clear publication bias. p values less than 0.05 were considered statistically significant.

Correlation of TUG1 Expression with Survival.
A total of 8405 patients were enrolled to assess the association between TUG1 level and OS. A random effect model was employed due to clear heterogeneity (I 2 = 86:9%, p < 0:001). A significant correlation was found between high TUG1 expression and unfavorable OS, and the pooled HR was 1.70 (95% CI: 1.48-1.95, p < 0:001) ( Figure 2). Twelve cohorts were included to investigate the relationship between TUG1 expression and PFS, DFS, RFS, and EFS. The results showed that TUG1 expression was not significantly correlated with PFS (p = 0:648) or DFS (p = 0:437), but a tendency toward worse RFS (p = 0:001) or EFS (p < 0:001) was revealed in patients with high level of TUG1 expression, although the number of included studies was extremely small ( Figure 3).

Subgroup Analyses of the Correlation between High TUG1
Expression and OS in Cancer. To address the heterogeneity among OS datasets, we performed subgroup analyses according to patients' origin, cancer type, sample size, and detection method (     Figure 5(b)), suggesting that a positive association between high TUG1 level and the prognosis of cancer patients existed no matter which study was removed.
3.6. Publication Bias. The potential for publication bias was assessed by funnel plots and Begg's test. The shape of the funnel plots for OS or PFS/DFS/RFS/EFS were asymmetric ( Figure 6), but the p value of the Begg's test for OS (Pr > |z | = 0:61) and DFS/RFS/EFS (Pr > |z | = 0:95) indicated that there was no severe publication bias in our present meta-analysis.

Discussion
lncRNAs participate in regulating tumoral biological processes by competitively interacting with certain micro-RNAs, altering the expression of key component proteins in the gene regulatory system [72,73]. To date, numerous studies have confirmed that mutation or misregulation of lncRNAs may promote tumorigenesis and metastasis and show that they are novel biomarkers and therapeutic targets for cancer [3,74,75]. TUG1, a 7.1 kb lncRNA, was first identified as a transcript upregulated in response to taurine treatment, which affects mouse retinal development [76]. Increasing numbers of studies have revealed that TUG1 is related to tumors. In most malignancies, TUG1 has been reported to be overexpressed and be involved in regulating of multiple processes in tumor progression, invasion and angiogenesis [77]. It has been confirmed by immunoprecipitation that TUG1 may recruit and bind to polycomb repressive complex 2 (PRC2) to regulate gene expression involved in tumorigenesis and tumor development [78,79]. Additionally, TUG1 can also exert its oncogenic role via sponging tumoral suppressor microRNAs or modulating cancer-related signaling pathways like Wnt, MAPK, or Notch1 [80][81][82]. However, TUG1 was found to be downregulated and acted as a tumor

Disease Markers
suppressing gene in some types of breast cancer, NSCLC, and glioma [8][9][10]. For example, TUG1 can promote tumor cell apoptosis and inhibit the growth of glioma by activating caspase 3-and caspase 9-mediated proapoptosis, inhibiting bcl-2 mediated antiapoptosis [10,83]. Thus, the prognosticpredictive value of TUG1 in cancer is still uncertain and needs further evaluation. The expression level of TUG1 in tumors and the correlation of TUG1 with patients' survival and clinicopathological characteristics have been previously assessed. In this study, we collected specific data of TUG1 involvement in tumor progression and survival of patients with different types of tumors, and we analyzed and summarized whether TUG1 is suitable as a prognostic marker for these tumors.
Although multiple meta-analyses have suggested that TUG1 could be used as a tumor-related prognostic marker, most studies were conducted before 2017 [12][13][14][15][16][17][18]. The prognostic value of TUG1 in some particular types of tumors is   In our study, the results suggest that high TUG1 expression is significantly associated with worse OS in patients with malignant tumors, which is consistent with the conclusions drawn from previous studies [13,14,[16][17][18]. In addition, the correlation between TUG1 expression and PFS/DFS/RFS/EFS in patients with tumors was analyzed for the first time. No significant association between TUG1 expression and PFS/DFS was found. Meanwhile, since only one included paper reported RFS and two    In the subgroup analysis, we found that the high expression of TUG1 was related to poor OS of patients with gastro-            9 Disease Markers studies for each tumor type. In urinary tumors, only one study on UC showed that a high expression of TUG1 was significantly related to better prognosis (HR = 0:62, 95% CI: 0.39-0.97, p = 0:012). BLC included five eligible studies. One study enrolling 406 patients from the TCGA database suggested that patients with high expression of TUG1 tended to have better prognosis, but no statistical difference was found [40]. Another 4 studies included 256 bladder cancer patients, 3 studies from China and 1 from the Czech Republic, with the results suggesting that patients with higher TUG1 expression have a lower survival rate. Although the overall analysis of urinary system tumors confirmed that the high expression of TUG1 has a prognostic value for patients, it is still necessary to expand the sample size to evaluate whether TUG1 plays a role in UC that can be distinguished from other urinary tumors, and the prognostic value of TUG1 in BLC also needs to be further verified. In respiratory tumors, 4 NSCLC studies indicated that patients with upregulated TUG1 levels have better prognosis (HR = 0:46, 95% CI: 0.27-0.80, p = 0:061), which was opposite to other tumor types. While only one SCLC article showed that patients with upregulated TUG1 expression tend to have worse prognosis, no statistical significance was found. The difference in the role of TUG1 in NSCLS and SCLC must be clarified by expanding the included studies in the future, and the specific lung tissue and pathological type may determine the prognostic role of TUG1 in lung cancer. In head and neck neoplasms and malignant melanoma, the expression of TUG1 was not significantly correlated with the survival of tumor patients. Head and neck neoplasm analysis included 3 GBM, 1 NPC, and 1 OSCC study. Of the 3 GBM articles, 2 were from China, indicating that patients with high TUG1 level have better prognosis. One TCGA data analysis was contrary to the previous two studies, suggesting that patients with low TUG1 expression have a longer survival time. Two melanoma studies displayed an association between upregulated TUG1 and poor prognosis, with no statistical significance. All these tumors need to be further studied to clarify whether TUG1 has prognostic value. The subgroup analysis based on patients' origin, sample size, and TUG1 detection method suggested that the prognostic value of TUG1 was not affected by these factors.
Furthermore, we analyzed the clinicopathological parameters related to TUG1. Different from previous analyses, the high expression of TUG1 was positively associated with tumor TNM stage, tumor differentiation, lymph node metastasis, and distant metastasis, which further confirmed the meaningful prognostic value of TUG1 in various tumors.
There are some limitations in the current study, and it should be interpreted with caution. First, the survival analysis data were not provided directly in some studies and needed to be extracted from Kaplan-Meier curves. However, some Kaplan-Meier curves were relatively difficult to extract due to the low pixel count [69] or the large sample size [40,63], and the data obtained may contain errors. For example, the p value displayed in Zhou et al.' work suggested that no statistical difference was found between TUG1 expression and the survival of patients in gastric cancer or breast cancer, whereas a statistical difference appeared in these two tumors through repeated extraction and calculation [40]. Thus, we utilized the data acquired by actual extraction for statistical analysis. Second, the prognostic role of TUG1 in head and neck neoplasms and malignant melanoma has not been confirmed. This may be due to the limitation of the study samples, and more sample analysis is needed in the future. Third, there was statistical heterogeneity among the studies included in this research. This may be due to differences in tumor types, the number of cases, the patient sources, the 10 Disease Markers detection methods, and the cut-off values of TUG1. Of note, the difference in TUG1 cut-off values and units may influence the application of TUG1 in the clinic.

Conclusions
Despite the limitations described above, our meta-analysis still showed that elevated TUG1 is significantly related to favorable prognosis of respiratory tumors and poor prognosis of gastrointestinal cancers, gynecological tumors, hematological tumors, urinary tumors, and osteosarcoma. No definite conclusion has been reached for head and neck neoplasms and malignant melanoma, and further analysis with a larger sample size is needed. Furthermore, the high expression of TUG1 is significantly associated with late tumor stage, poor differentiation, more lymph node metastases, and distant metastasis of tumors.

Data Availability
All data generated or analyzed during this study are included in this article.

Conflicts of Interest
All authors declare that no conflict of interest exists.