Recently, emerging studies have explored the possible link between expression of lncRNA NNT-AS1 and human tumors. Compared with adjacent noncancerous tissue and normal cell, upregulated NNT-AS1 expression was identified in most cancer tissues or cell lines and therefore indicated poor survival outcome, such as osteosarcoma [27], breast cancer [19], cervical cancer [26], gastric cancer [28], hepatocellular carcinoma [18], colorectal cancer [21], and non-small cell lung cancer (NSCLC) [2]. On the contrary, another study performed by Huang et al, claimed that NNT-AS1 was markedly downregulated in patients with ovarian cancer and ovarian cell lines [29]. However, results from above-mentioned studies should be interpreted with caution because of the limited sample size and discrete outcomes. Therefore, we designed and carried out this meta-analysis to further elucidate the correlation between NNT-AS1 and clinicopathologic outcomes and prognostic values in cancers.
Ten studies with eight cancer types containing 699 patients were pooled together in this study, and the results suggested that promoted NNT-AS1 expression was significantly associated with unfavorable prognosis of OS in patients with cancers. Subgroup stratified analysis further demonstrated that the tumor type, sample size, follow-up months, and survival analysis method did not alter the correlation between TTN-AS1 and OS. Thus, it seems like that differences among study design characteristics have no obvious effect on NNT-AS1 correlation with OS, suggesting that our pooled results were reliable. No publication bias regarding NNT-AS1 expression for OS was observed, indicating the credibility of our results. Furthermore, pooled data from TCGA dataset showed that NNT-AS1 was obviously correlated with OS (HR=1.1, p=0.018), which was consistent with our results. Since tumor recurrence is also an imporant prognosis parameter for cancer patients, we next explored the association between NNT-AS1 expression and DFS via TCGA datasets. We showed that NNT-AS1 overexpression was correlated with unfavorable DFS (HR=1.1, p=0.033) in the cancer types involved in this meta-analysis (Figure 5E). Moreover, one study [15] included in our analysis also reported that an elevated expression of TTN-AS1 predicted poor DFS in gastric cancer. However, interpretation regarding DFS should be careful since relevant studies were comparatively limited, and large cohort datasets are still needed for further validation the relationship between NNT-AS1 and tumor recurrence or other parameters. In addition, elevated NNT-AS1 level dramatically predicted worse clinical stage, vascular invasion, LNM, and DM. No significant association between NNT-AS1 and other clinicopathologic parameters including age and gender. Thus, even though NNT-AS1 expression has divergent association with clinicopathological parameters among included studies, our results still provided a reliable correlation via comprehensive meta-analysis. Consistent with our findings, result from the TCGA indicated that NNT-AS1 expression was significantly associated with clinical stage of human cancers. Taken together, our study firstly clarified the relationship between NNT-AS1 and cancer prognosis with a comprehensive evaluation on study heterogeneity and bias. TCGA dataset was explored to validate the role of NNT-AS1 in carcinomas, and the results showed that NNT-AS1 expression may serve as a potential indicator for cancer prognosis.
Previous studies have investigated the underlying mechanisms of NNT-AS1 in carcinogenesis. Overexpression of NNT-AS1showed positive association with poorer OS, advanced tumor stage, LNM, depth of invasion [30], vessel invasion and differentiation in numerous cancers. Functional assays revealed that NNT-AS1 could promote proliferation, weaken cell cycle arrest and alleviate apoptosis by competing with CDK6 for miR-363 binding in hepatocellular carcinoma [18]. High expression of NNT-AS1 facilitates cholangiocarcinoma prognosis via promoting epithelial-mesenchymal transition (EMT) [20]. While knockdown or inhibition of NNT-AS1 could suppress cancer cell colony formation and invasion, arrested the cell cycle and promoted apoptosis both in vitro and in vivo [21]. Additionally, when silencing NNT-AS1 in colorectal cancer, EMT and MAPK/Erk pathway were inhibited [21]. Moreover, other pathways including PI3K/Akt/mTOR and Wnt/β-catenin signaling pathway were also found involved in the tumorigenesis and progression [26, 15]. Besides, NNT-AS1 was capable of serving as a competing endogenous RNA (ceRNA) by sponging miR-485/BCL9 or miR-203 in cholangiocarcinoma [31, 32], miR-1301-3p/PODXL or miR-496/HMGB1 in bladder cancer [3, 33], miR-142-3p/ZEB1 in breast cancer [19], miR-424/E2F1 or miR-363 in gastric cancer [28, 8], miR-22-3p/YAP1 or miR-129-5p in non-small cell lung cancer [34, 30], and miR-320a in osteosarcoma [27], therefore alteration in cancer cell function resulting from NNT-AS1 downregulation may be rescued by miRNA inhibition. Notably, NNT-AS1 also showed a high expression level in drug-resistant NSCLC, which promoted the cisplatin resistance of cancer cells via the MAPK/Slug pathway [35]. All these studies suggested that NNT-AS1 could serve as an oncogenic biomarker in cancer progression. The schematic diagram of various molecules and signaling pathways associated with NNT-AS1 in human cancers were displayed in Figure 6.
Several deficiencies exist in this meta-analysis and they should be acknowledged. In the first place, our meta-analysis used the summarized data instead of raw data from the specific patients, and most of the HRs and 95% CIs were indirectly calculated by reconstructing survival curves instead of extracted from the original data, which inevitably could cause heterogeneity. Second, the cut-off value for NNT-AS1 expression differed across eligible studies due to the difficulty in reaching a consensus value, thus may introduce possible bias. Third, all enrolled studies were from China, which may cause biased results because of geographical differences. Thus, it should be cautious when applying our conclusions to the population in other regions. Fourth, data regarding NNT-AS1 expression levels with other prognostic outcomes, such as PFS, DFS were limited and thus unable to calculate the pooled value. Fifth, other factors such as different classification system of clinical stage, follow-up time, and analysis methods will also lead to possible bias. Sixth, the regulatory mechanisms of NNT-AS1 in cancer progression still remain largely unexplored. For instance, somatic mutation has been reported to affect lncRNA by regulation of methylation or expression of gene, miRNA, and transcriptional factor [36]. Currently, data regarding the role of somatic mutation on NNT-AS1 is lacking, which may merit future elucidation by using bioinformatics and in vitro assay [36, 37]. Last, in order to further clarify the correlation between NNT-AS1 expression with age or gender, more confounding variables, such as BMI, smoking history, or comorbidities, should be eliminated in order to minimize the possible bias [38]. Therefore, on the basis of the above limitations, well-designed comprehensive studies containing a large sample size, broader regions and countries, and more credible indicators are still warranted to further confirm our results.