A total of 126 relevant publications were acquired after searching PubMed, Web of Science, Embase, and the Cochrane Library. Following the removal of duplicate publications, 67 studies were considered. Titles and abstracts were reviewed, and 16 non-relevant papers and 7 review articles were excluded. Further screening of the full texts of the 44 remaining articles led to the elimination of 18 reports due to a lack of study data regarding prognoses or clinicopathologic characteristics. Finally, 26 studies [8-33] from 2017 to 2020 met the criteria for the meta-analysis. Figure 1 describes the selection process for the included publications.
The characteristics of the 26 included articles are summarized in Table 1. A total of 1770 patients with sample sizes ranging from 36 to 132 were evaluated and divided into high and low TP73-AS1 expression groups according to the mean cut-off value. All studies were performed in China and published from 2017 to 2020. The studies examined a wide variety of cancers, including hepatocellular carcinoma (2 studies), osteosarcoma (2 studies), gastric cancer (5 studies), ovarian cancer (2 studies), clear cell renal cell carcinoma (1 study), bladder cancer (1 study), breast cancer (3 studies), cholangiocarcinoma (1 study), lung cancer (3 studies), brain glioma (1 study), pancreatic cancer (1 study), colorectal cancer (2 study) ,cervical cancer (1 study) and retinoblastoma (1 study). The expression of TP73-AS1 in tissues was through qRT-PCR analysis. OS data were reported in 19 studies, with 3 studies reporting DFS. The NOS assessment scores of the included studies ranged from 6 to 8.
Correlation between TP73-AS1 and clinicopathologic parameters.
Relationship between TP73-AS1 expression and age
As shown in Figure 2a, seven studies explored the correlation between TP73-AS1 expression and age. No noticeable heterogeneity was observed amongst the studies (I²=36.7%, P=0.148) so a fixed model was employed for analysis. No significant correlation was identified between TP73-AS1 overexpression and patient age (OR=1.12, 95% CI 0.77-1.64, P>0.05).
Relationship of TP73-AS1 expression with gender
Thirteen studies investigated potential associations between TP73-AS1 expression and gender. Upon assessment with the fixed model, no heterogeneity was observed amongst the studies (I²=0%, P=0.713). As shown in Figure 2b, expression of TP73-AS1 did not correlate with patient gender (OR=1.08, 95% CI 0.84-1.38, P>0.05).
Relationship of TP73-AS1 with TNM stage
TNM stage and TP73-AS1 expression were reported for 794 patients across twelve studies. The pooled results showed an OR=3.27 (95% CI: 2.43-4.39, P<0.00001) with no notable heterogeneity (I2=0%, P=0.526). The data were therefore analyzed using a fixed model (Figure 2c). The upregulation of TP73-AS1 was significantly associated with advanced TNM stage.
Relationship of TP73-AS1 expression with tumor size
The relationship between tumor size and TP73-AS1 expression was evaluated for 8 studies of 501 patients. Forest plots indicated no evident of heterogeneity amongst the studies (I2=0%, P=0.665). Subsequent analysis indicated that the overexpression of TP73-AS1 correlated with a tumor size ≧ 5cm (OR=3.00, 95% CI: 2.08-4.35, P<0.00001) (Figure 2(d)).
Relationship between TP73-AS1 expression and lymph node metastasis
Data for total OR and 95% CI of LNM were collected from 13 studies. Data analysis yielded a pooled OR of 2.77 (95% CI 1.42-5.38, P<0.00001) using a random model, owing to significant heterogeneity (I2=78.5%, P≤0.001) (Figure 2e). We concluded that TP73-AS1 overexpression was associated with lymph node metastasis of cancer.
Relationship between TP73-AS1 expression and distant metastasis
Five studies involving 341 patients were analyzed to evaluate the correlation between TP73-AS1 expression and distant metastasis. No significant heterogeneity was detected amongst the studies（I2=0%, P =0.604, fixed-model). Model results indicated that TP73-AS1 upregulation was related to distant metastasis (OR=4.50, 95% CI: 2.62-7.73, P<0.00001), (Figure 2f).
Relationship between TP73-AS1 expression and differentiation
As shown in Figure 2g, eight studies were used to evaluate the correlation between TP73-AS1 expression and histological tumor differentiation. The random model was performed owing to heterogeneity amongst the studies (I2=72.0%, P≤0.001) with a pooled OR=1.39 (95% CI: 0.71-2.70, P=0.340). No marked differences were detected in differentiation status between the two groups.
Association between TP73-AS1 and prognostic indicators
Relationship of TP73-AS1 expression with OS
A total of 19 studies with data from 1315 patients were used to determine the utility of TP73-AS1 as a prognostic biomarker of cancer based on OS data. Pooled HR = 1.85 (95% CI: 1.53-2.22, P<0.00001) (Figure 3a). A fixed-effects model was used to estimate the HR of the studies, that showed no apparent heterogeneity (I2=0%, P=0.952). These data indicated that the upregulation of TP73-AS1 was associated with a poorer OS amongst multiple types of malignancies.
Relationship of TP73-AS1 expression with DFS
Only three studies reported DFS data that could be used to assess the prognostic effects of TP73-AS1. Our analysis suggested that TP73-AS1 overexpression was associated with DFS (pooled HR: 1.57, 95% CI: 1.03-2.42, P<0.05), (Figure 3b). No obvious heterogeneity was observed amongst the studies (I2=48.4%, P=0.144).
Sensitivity analysis and publication bias
A sensitivity analysis was used to evaluate the robustness of the pooled results. The data were deemed reliable without the removal of any studies (Figure 3c). No publication bias was observed through Begg’s tests (Figure 3d, P=0.368).
Verification of TP73-AS1 expression and its prognostic value based on the TCGA
To validate the expression of TP73-AS1 in diverse cancers, we used the GEPIA online tool for gene analysis. As shown in Figure 4, the expression of TP73-AS1 was dramatically upregulated in three malignancies including cholangiocarcinoma, lymphoid neoplasm diffuse large B-cell lymphoma and thymoma (|Log2fold change (FC)| cutoff >1 and P<0.01). In addition, log-rank analysis and Kaplan–Meier curves were used to verify the association between TP73-AS1 expression and the prognostic index of patients with different malignancies. Similar to our meta-analysis, the overexpression of TP73-AS1 correlated to a poorer OS in adrenocortical carcinoma (ACC) and low grade glioma (LGG) (log-rank P<0.05) (Figure 5a–b). Moreover, the upregulation of TP73-AS1 was associated with a poorer DFS in adrenocortical carcinoma (ACC), low grade glioma (LGG), colon adenocarcinoma (COAD), prostate adenocarcinoma (PRAD) and stomach adenocarcinoma (STAD) (log-rank P<0.05) (Figure 5c-g). These results indicated that TP73-AS1 serves as a novel prognostic biomarker for cancer malignancies.
Prediction of TP73-AS1 function
We predicted potential biological functions and the molecular mechanisms of TP73-AS1 in cells using public databases. We first explored ceRNA modulations for TP73-AS1 using targetscan, mirdb, mirtarbase, and mircode databases. TP73-AS1-miRNA-mRNA networks containing 8 miRNAs and 448 mRNA were constructed using cytoscape (Figure. 6). Figure 7 shows the top 12 KEGG and Go pathways. TP73-AS1 was predicted to participate in tumor signaling, including Kaposi sarcoma-associated herpesvirus infection, signaling pathways regulating the pluripotency of stem cells and TGF-beta signaling (Figure. 7a-b). GO functional enrichment analysis indicated that the molecular functions of TP73-AS1 included RNA polymerase II proximal promoter sequence-specific DNA binding, proximal promoter sequence-specific DNA binding, and core promoter binding (Figure. 7c-d).