Literature search and selection
The literature selection process is shown in Fig. 1. Preliminarily, 151 relevant studies in total were yielded from the search of the PubMed, Cochrane Library, EMBASE, CNKI, Weipu, and Wanfang electronic databases. Among these, 89 studies were excluded as duplicate articles. Then we further excluded 34 studies by reviewing the title and, abstract. Subsequently, 11 more studies were not able to be included because of insufficient data and being unrelated to our study. Finally, 17 studies containing 1788 patients were eligible for this meta anaylsis and were highly consistent with the inclusion criteria. All of the included studies were published between 2017 and 2020 and came from China. Multiple forms of cancers were analyzed in the present meta-analysis, including gastric cancer [15], ovarian cancer [16], glioma [17, 18], colorectal cancer [19, 20], hepatocellular carcinoma [21, 22], breast cancer [23], renal cell carcinoma [24], osteosarcoma [11, 12], lung cancer [9, 25], acute myeloid leukemia [26], papillary thyroid carcinoma [27, 28]. The detailed information obtained from the studies is summarized in table 1.
Table1: The main characteristics of the included studies in the meta-analysis.
![](https://myfiles.space/user_files/58653_1b1c6aeb34a62c68/58653_custom_files/img1602612857.jpg)
SNHG3 expression highly correlated with OS, RFS and DFS
Overall, 15 of the 17 studies investigated cancer prognosis. A total of 2072 patients were assessed for the HR and 95% CI of OS. The random-effects model was performed to analyze the pooled HR and its 95% CI depended on no obvious heterogeneity (P = 0.01,I2 = 51%). We further elucidated the relationship between SNHG3 expression and the overall survival, as illustrated in Fig. 2. The pooled results revealed that the high expression of SNHG3 was related to poor prognosis of cancers (HR = 2.15, 95%CI: 1.76–2.63, P < 0.00001, Fig. 2A). In the subgroup anaylsis stratified by tumor type, we found that elevated SNHG3 could act as a prognostic predictor for patients with digestive system tumors (HR = 2.34, 95%CI: 1.53–3.57, P = 0.003) or patients with non-digestive system tumors (HR = 1.95, 95%CI: 2.43–2.67, P = 0.0002, Fig. 2B). Thus, the prognosis of cancer patients with SNHG3 overexpression was worse than those with low expression of SNHG3. In terms of DFS, only 3 studies were included, and the pooled results indicated that patients with high expression of SNHG3 had poor DFS (HR = 2.04, 95%CI: 1.35–3.09, P = 0.0007, Fig. 3A), Only one focus on the relationship between SNHG3 and tumor recurrence (HR = 2.22, 95%CI: 1.04–4.76, P = 0.004, Fig. 3B).
Independent prognostic value of SNHG3 in cancers
Multivariate analysis and a fixed-effects model were used in 5 studies (P = 0.45,I2 = 0%) calculate the independent prognostic value of SNHG3 in cancer. The combined HRs showed that the elevated expression of SNHG3 could be an independent prognostic factor for OS in patients with cancer(HR = 1.90, 95%CI: 1.59–2.27, P < 0.00001, Fig. 4).
Relationship between SNHG3 expression and clinicopathological characteristics
The merged results from 11 studies with 1204 patients demonstrated that patients with SNHG3 overexpression have a more advanced stage (III/IV) cancer (III/IV vs. I/II, OR = 2.91, 95%CI: 1.60–5.29, P = 0.0005, Fig. 5A). Here we used a random-effects model because of obvious heterogeneity (P༜0.0001, I2 = 73%). In addition, these 5 studies contained 726 individuals showed correlation between SNHG3 and LNM in various cancers. A fix-effects model was utilized again because of obvious heterogeneity (P = 0.17, I2 = 37%), and the pooled results showed that lymph node metastasis was more susceptible to the upregulated SNHG3 expression group than the downregulated SNHG3 expression group (OR = 5.00, 95%CI:2.82–8.87, P༜0.00001, Fig. 5B). Only 4 studies provided information for distant metastasis (DM). The pooled results indicated that patients with high SNHG3 expression have more metastasis to distant organs or tissues (OR = 2.29, 95%CI: 1.52–3.47, P < 0.0001, Fig. 5C). Again, a fixed-effects model was used (P = 0.27,I2 = 24%). 9 studies provided information for tumor size, which showed that patients with high SNHG3 expression have larger size (OR = 1.80, 95%CI: 1.04–3.11, P = 0.04, Fig. 5D). Furthermore, we did an investigation on the relationship between SNHG3 expression and age, gender, and differentiation. However, the pooled results suggested that SNHG3 expression was not positively associated with these characteristics (Fig. 6A–C). The details are shown in Table 2.
Table 2:Summary of the relationship between SNHG3 over-expressed and clinicopathological parameters.
![](https://myfiles.space/user_files/58653_1b1c6aeb34a62c68/58653_custom_files/img1602612900.jpg)
Publication Bias And Sensitivity Analysis
The begg’s test was used to evaluate the publication bias in this meta-analysis. No significant publication bias for OS and independent factor for OS was found in this meta-analysis (Fig. 7A–B). As illustrated in Fig. 8A–B, we performed the sensitivity analysis to prove that the results were robust, and the summary HRs were not affected after removal of study one by one.