LINC00511 as a prognostic biomarker for human cancers: a systematic review and meta-analysis

The present study was performed according to the Preferred Reporting Items for Systematic Reviews and Meta-Analysis (PRISMA) [29]. Relevant studies published before December 22, 2019 were systematically searched online in PubMed, EMBASE, Web of Science and the Cochrane Library databases. The following search terms were used in different combinations: (LINC00511 OR “long intergenic noncoding RNA 00511” OR “long intergenic non-protein coding RNA 511”) AND (cancer OR tumor OR tumour OR carcinoma OR neoplasm OR adenoma OR sarcoma OR malignancy). The reference lists of the full-text articles were also checked to find relevant studies that pointed out the relation between LINC00511 expression and clinical outcomes. No language restriction was applied.

The eligibility criteria were as follows: (1) the study subjects were patients with any type of cancer; (2) the studies investigated association between LINC00511 expression levels and clinical outcomes in cancer; (3) patients were separated into LINC00511 high expression level and LINC00511 low expression level groups; (4) the LINC00511 expression levels were measured by quantitative method (e.g., real-time reverse transcription polymerase chain reaction (qRT-PCR)); and (5) sufficient information and data were provided to calculate a hazard ratios (HRs) or odd ratios (ORs) and their corresponding 95% confidence intervals (CIs). Exclusion criteria were as follows: (1) duplicate publications; (2) reviews and meta-analyses; (3) studies without patient samples; (4) studies not relevant to cancer, LINC00511 or prognosis.

The quality of included studies was assessed using the Newcastle-Ottawa Scale (NOS). Three main categories have been considered including selection, comparability and outcome, and the stars rating system has been used. The scores of NOS were ranged from 0 star (lowest score) to 9 stars (highest score). A study with a NOS score higher than 5 was considered as a high-quality study [30, 31].

Two investigators reviewed all eligible studies independently and extracted the following data: the first author’s name, year of publication, country of origin, the tumor type, sample size, cut-off value, detection method, HRs and corresponding 95% CIs, and clinicopathological features. Any controversial issue was resolved by discussion. In case HR value was not provided, Engauge Digitizer version 11.2 (http://​markummitchell.​github.​io/​engauge-digitizer/​) was used to extract data from Kaplan-Meier curve. The extracted data were then used to estimate HRs and 95% CIs by using the HR calculations spreadsheet provided by Tierney et al. [32].

We used GEPIA, a web server for cancer and normal gene expression profiling and interactive analyses, to perform bioinformatics analysis. GEPIA uses the data from both TCGA (The Cancer Genome Atlas) and GTEx (Genotype-Tissue Expression) to perform several analyses. The datasets were computed by the UCSC Xena project based on a standard pipeline [33, 34]. Therefore, we directly use GEPIA to perform our analyses. GEPIA used Kaplan-Meier method and log-rank test for the survival analysis and one-way ANOVA for the gene expression analysis. First, the expression levels of LINC00511 in the different cancer types involved in our meta-analysis was analyzed by matching TCGA and GTEx data. Next, overall survival and disease-free survival curves were retrieved.

All statistical analyses were performed with RevMan 5.3 software (Cochrane community, https://​community.​cochrane.​org/​help/​tools-and-software/​revman-5/​revman-5-download/​) and STATA 15.0. The pooled HRs and their 95% CIs were used to evaluate the association of LINC00511 expression with the cancer patients’ survival. Odd ratios (ORs) analyses were performed to assess the correlation between LINC00511 expression and clinicopathological parameters. The heterogeneity between studies results was examined by Cochran’s Q test and Higgins I-squared (I²) statistic. In the absence of heterogeneity (p ≥ 0.10 and/or I² < 50%), the fixed-effect model was used; otherwise the random-effect model was applied [35, 36]. Potential publication bias was evaluated using the funnel plot, Egger’s test and Begg’s test. Sensitivity analysis was also performed to assess the impact of individual study on the pooled effect [37]. A p-value less than 0.05 was considered to be statistically significant.

A flowchart of the literature search strategy is shown in Fig. 1. According to the described search strategy, 118 articles were retrieved. Sixty articles were excluded based on duplication criteria. The 58 articles remained, underwent an initial screening. On the basis of title and abstract, 35 articles (non-cancer publications, non-LINC00511 publications, reviews and meta-analyses, retracted article, articles without prognosis and clinical data) were excluded. A total of 23 full-text articles were eventually assessed for eligibility. After reading full-text articles, 9 articles were excluded for reasons: articles without prognosis and clinical data in respect to LINC00511, not available full-text article and non-patient samples. At the end, 14 articles were used for the quantitative synthesis [16‐22, 24‐28, 38, 39]. The 14 studies involved a total of 1883 patients and all came from China. The sample size of included studies ranged from 36 to 412 patients and the type of cancer included non-small-cell lung cancer, pancreatic cancer, breast cancer, cervical cancer, hepatocellular carcinoma, ovarian cancer, and brain tumors. The articles included in the present study were all written in English and published between 2016 and 2019. The expression level of LINC00511 was mainly determined by qRT-PCR. The cut-off values included mean, median and the P25 value. In our study, the quality of all included studies is high (NOS score > 5). The overall characteristics of included studies are summarized in Table 1.

Elevated LINC00511 expression was significantly predictive of poor overall survival (HR = 2.62; 95% CI: 2.00–3.45; p <  0.001). A high heterogeneity was observed among the studies (P_H < 0.001; I² = 85.3%), so the random-effect model was used (Fig. 2a). In addition, high LINC00511 expression has been proven to have an independent prognostic value in four studies by processing the multivariate analysis [18, 20, 22, 25]. As previously reported [40], we conducted the meta-analysis of these studies and found that high LINC00511 expression may serve as an independent predictor for poor overall survival prognosis in cancers (HR = 3.37; 95% CI: 1.96–5.79; p < 0.001; I² = 82.5%; P_H = 0.001; Fig. 2b).

The relationship between LINC00511 expression and cancer progression was investigated by combining progression-free survival (PFS) studies. The results, from only two studies with available related data, demonstrated that high LINC00511 expression was predicted to be associated significantly with worse PFS (HR = 1.80; 95% CI: 1.29–2.51; p = 0.001; I² = 0%; P_H = 0.44; Fig. 2c). In addition, one study found that high LINC00511 may be associated with worse relapse-free survival (HR = 2.90; 95% CI: 1.04–8.12; p = 0.04).

Subgroup meta-analyses based on cancer types (according to NCBI’s medical subject headings (MeSH) [41]) (Fig. 3a), sample size (Fig. 3b), cut-off value (Fig. 3c) and data extraction method (Fig. 3d) were conducted. The results showed that high LINC00511 expression was significantly associated with poor overall survival in lung neoplasms (HR = 3.86; 95% CI: 1.74–8.54; p = 0.001; I² = 80.5%; P_H = 0.02), digestive system neoplasms (HR = 3.06; 95% CI: 2.76–3.40; p < 0.001; I² = 0%; P_H = 0.45), breast neoplasms (HR = 2.20; 95% CI: 1.50–3.23; p < 0.001; I² = 0%; P_H = 0.99) and urothelial neoplasms (HR = 3.15; 95% CI: 2.60–3.80; p < 0.001; I² = 0%; P_H = 0.80). Regarding the brain neoplasms subgroup, despite the relatively high HR, the relationship cannot be considered robust because the p-value is higher than 0.05 (HR = 1.50; 95% CI: 0.87–2.60; p = 0.15; I² = 68.6%; P_H = 0.07). In respect to sample size subgroups, there was significant correlation between high LINC00511 expression and poor overall survival prognosis of cancer patients in both large sample size (n ≥ 95) (HR = 2.65; 95% CI: 1.81–3.89; p < 0.001; I² = 92.9%; P_H < 0.001) and small sample size (n < 95) (HR = 2.63; 95% CI: 2.02–3.43; p < 0.001; I² = 0%, P_H = 0.79). Concerning the cut-off value subgroups analysis, whether the median was used as cut-off value (HR = 2.86; 95% CI: 2.43–3.38; p < 0.001; I² = 0%; P_H = 0.87) or the mean as cut-off value (HR = 2.24; 95% CI: 1.38–3.64; p = 0.001; I² = 0%, P_H = 0.99) or other cut-off values were used (HR = 2.80; 95% CI: 1.60–4.90; p < 0.001; I² = 95.2%; P_H < 0.001), high LINC00511 expression correlated with poor overall survival in cancer patients. Similarly, the subgroup meta-analysis based on the data extraction method showed that whether the data were extracted directly from the literature (HR = 3.21; 95% CI: 2.73–3.76; p < 0.001; I² = 37.4%; P_H = 0.14) or indirectly (HR = 1.80; 95% CI: 1.30–2.47; p < 0.001; I² = 41.6%; P_H = 0.11), high LINC00511 expression was significantly associated with poor overall survival in cancer patients. Of note, the heterogeneity is likely to come from diverse sources.

The meta-analysis of the correlation between LINC00511 expression and clinicopathological parameters are summarized in Table 2. The pooled odds ratio (OR) values revealed significant association between high LINC00511 expression and large tumor size (OR = 3.10; 95% CI: 1.97–4.86; p < 0.00001; I² = 50%; P_H = 0.03), lymph node metastasis (OR = 3.11; 95% CI: 2.30–4.21; p < 0.00001; I² = 42%; P_H = 0.08), advanced clinical stage (OR = 3.95; 95% CI: 2.68–5.81; p < 0.00001; I² = 0%; P_H = 0.48), distant metastasis (OR = 2.39; 95% CI: 1.16–4.93; p = 0.02; I² = 21%; P_H = 0.28), and disease recurrence (OR = 4.62; 95% CI: 2.47–8.65; p < 0.00001; I² = 0%; P_H = 0.66). Meanwhile, the results showed no statistically significant correlation between LINC00511 expression and age, gender, and histological grade.

Funnel plot analysis, Egger’s test and Begg’s test were performed to evaluate potential publication bias. There was no obvious asymmetry on the funnel plot of OS (Fig. 4) and the tests of publication bias (Begg’s test: p = 0.161; Egger’s test: p = 0.630) indicated that no publication bias existed in studies reported OS.

The sensitivity analysis was performed by omitting each individual study to test the stability of the pooled result of the association between LINC00511 expression and OS. As shown on Fig. 5, when “Li C. (2019)” was omitted, the pooled result oscillated. Subsequently, we evaluated the pooled HR after removing “Li C. (2019)” and found that elevated LINC00511 expression still predicted worse OS (HR = 3.06; 95% CI: 2.80–3.33; p < 0.001; I² = 14.8%; P_H = 0.30). Therefore, the influential study didn’t alter the significance of the pooled result, sustaining the reliability of our pooled result.

To strengthen the results of our meta-analysis, we used GEPIA web server which contain a larger amount of sample size and performed bioinformatics analysis. The expression level of LINC00511 was analyzed in eleven different type of cancers (cancer types included in the meta-analysis). As shown on the Fig. 6, LINC00511 was highly expressed in almost all the related cancers, including lung adenocarcinoma (LUAD), lung squamous cell carcinoma (LUSC), pancreatic adenocarcinoma (PAAD), breast invasive carcinoma (BRCA), cervical squamous cell carcinoma (CESC), liver hepatocellular carcinoma (LIHC), kidney renal clear cell carcinoma (KIRC), kidney renal papillary cell carcinoma (KIRP), kidney chromophobe (KICH), glioblastoma multiforme (GBM), and brain lower grade glioma (LGG) (log₂FC value > 1 and p-value < 0.01). The Fig. 7 shows the Kaplan-Meier curves of the overall survival (Fig. 7a) and disease-free survival (Fig. 7b) analyses. A total of 4410 patients were divided into LINC00511 low expression and high expression groups based on the median LINC00511 expression. The patients in high expression group showed a worst overall survival (HR = 2.00; log-rank p-value < 0.001) and disease-free survival (HR = 1.8; log-rank p-value < 0.001) than those in the low expression group. Furthermore, we explored the possibility for any association of the LINC00511 expression to overall cancer survival in other cancer types that are not involved in the meta-analysis. It was found that LINC00511 overexpression was significantly associated with worse OS in adrenocortical carcinoma (ACC) (HR = 4.5; log-rank p-value = 0.00047; Fig. 7c), and thymoma (THYM) (HR = 7.00; log-rank p-value = 0.035; Fig. 7d). These results confirmed that LINC00511 is overexpressed in different types of cancers and this is correlated with poor survival. Of note, the GEPIA web server was accessed on December 28, 2019.