The Potential Diagnostic Value of Exosomal Long Noncoding RNAs in Solid Tumors: A Meta-Analysis and Systematic Review

Background Exosomes are defined as small membranous vesicles. After RNA content was discovered in exosomes, they emerged as a novel approach for the treatment and diagnosis of cancer. Long noncoding RNAs (lncRNA), a kind of specific RNA transcript, have been reported to function as tumor growth, metastasis, invasion, and prognosis by regulating the tumor microenvironment in exosomes. This study aims at exploring the potential diagnostic of exosomal lncRNA in solid tumors. Methods A meta-analysis conducted from January 2000 to October 2019 identified publications in the English language. We searched all relevant English literature from the Web of Science, EMBASE, and PubMed databases through October 1, 2019. The articles were strictly screened by our criteria and critiqued using the Preferred Reporting Items for Systematic Reviews and Meta-Analyses guidelines. Results There were 28 studies with 19 articles (4017 patients) identified, including studies on gastric cancer, laryngeal squamous cell carcinoma, colorectal cancer, cholangiocarcinoma, breast cancer, esophageal squamous cell carcinoma, hepatocellular carcinoma, nonsmall cell lung cancer, and prostate cancer. A meta-analysis showed that the combined value of sensitivity in 29 studies was 0.74 (95% confidence interval [CI], 0.7–0.78), and the combined value of specificity in the studies was 0.81 (95% CI, 0.78–0.83). This suggests the high diagnostic efficacy of liquid exosomes in cancer patients. It is statistically insignificant in terms of sex, ethnicity, and year. The diagnostic power of urinary system tumors was found to be higher than that of digestive system tumors by several subgroup analyses. Conclusions We performed a meta-analysis and literature review of 28 studies that included 4017 patients with 10 malignant cancer types. Mechanistically, our study demonstrated that lncRNAs in exosomes could be a promising bioindicator for the diagnosis and prognosis of solid tumors. INPLASY Registration Number: INPLASY202060083.


Introduction
In the past several decades, cancer therapy has been studied extensively, which has become the most lethal causes of worldwide [1]. During the past century, there has been a dramatic development in cancer treatments, including surgery, radiotherapy, and immunotherapy [2]. However, the overall survival of malignant tumor patients is still unsatisfactory, and the diagnostic efficiency remains lower than desired, especially for advanced tumor patients. With the development of early diagnosis and treatment of cancer, we observed that an ideal biomarker could play a pivotal role [3,4]. Thus, it is our primary goal to explore a new effective biomarker for early diagnosis, prediction of prognosis, and ideal therapeutic target for cancer patients.
Exosomes are defined as small membranous vesicles, which are 40 to 150 nm in diameter, with specific surface molecular markers such as CD9, CD63, Hsp70, and TSP101 [5,6]. After RNA content was discovered in exosomes in 2007, they emerged as a novel approach for the treatment and diagnosis of cancer [7]. During the past 2 decades, many researchers have demonstrated that exosomes could contain various kinds of RNAs, including messenger RNAs (mRNA) [8], microRNAs (miRNA) [9], long noncoding RNAs    [10], and circular RNAs (circRNA) [11]. Exosomes in tumor cells have been found to activate specific tumor resistance responses and promote the occurrence and development of tumors [12]. Despite that, a large number of studies have reported that exosomes are involved in angiogenesis and metastasis, drug resistance, immune related, cytokines secretion, apoptosis, cell proliferation, and oncogenic cell transformation [13]. According to an investigation by Mudgapalli et al., exosomes could act as a messenger between the microenvironment and tumor cells that support leukemia growth, inducing oncogenic factors such as c-Myc [14]. In addition, serum-derived exosomes in primary prostate or prostate metastasis increase a high level of PKM2 expression, suggesting that exosomes could be a potential clinical biomarker in prostate cancer [15].
lncRNAs are defined as a type of specific RNA transcript longer than 200 nucleotides, which were once considered unfunctional transcriptional noise [16]. However, it was later found that ncRNAs play a functional role in carcinogenesis, tumor regulation, and gene expression [17]. Furthermore, lncRNA expression always demonstrates poor prognosis in several cancers, including lung cancer [18], osteosarcoma [19], bladder cancer [20], ovarian cancer [21], liver cancer [22], gastric cancer [23], hepatocellular carcinoma [24], and breast cancer [25]. In these studies, there is an obvious connection between high-level expression of lncRNA and worse clinicopathological outcome, such as distant metastasis, tumor size, lymph node metastasis, clinical stage, and drug resistance. For instance, high SNHG11 expression promotes metastasis and proliferation in colorectal cancer by targeting the Hippo pathway [26]. A study by Fang et al. reported that miR-223 could be mediated, as silencing by DLX6-AS1 promotes the progression of bladder cancer through the upregulation of HSP90B1 [27]. In another study, Xu et al. examined lncRNA PVT1, an oncogene that has been found to result in a poor overall survival in esophageal adenocarcinoma though high-level expression [28]. In addition, in our precious research, we demonstrated that the expression of lncRNA SNHG1 was significantly associated with worse clinical outcome [29]. Mechanistically, it has been reported to function as tumor growth, metastasis, invasion, and prognosis by regulating the tumor microenvironment.
Recent studies have shown that biomarkers with tumors are a hot topic, with researchers focusing on the value of ncRNAs, mRNAs, and exosomes in the diagnosis and treatment of cancer. Most of the available literature on these biomarkers was focused mainly on ncRNAs in blood. In addition to these primary data, systematic studies on the prognosis biomarkers are still needed for further exploration. Now that researchers are aware of the significance of ncRNA in exosomes, the function of circRNAs in exosomes has already been reported as a suitable diagnostic biomarker for tumors, especially in hepatocellular carcinoma [30]. However, the diagnostic effectiveness of lncRNAs in exosomes is still not clear.
Therefore, our team is the first to summarize the key role functional lncRNAs play in different types of tumors by performing a meta-analysis. The specificity and sensitivity of exosomal lncRNAs were evaluated to assess their feasibility as biomarkers of cancer diagnosis.   Figure 1: Overall quality assessment of the included articles using the QUADAS-2 tool. Determination of serum exosomal H19 as a noninvasive biomarker for bladder cancer diagnosis and prognosis 10

Materials and Methods
Serum and exosome long non coding RNAs as potential biomarkers for hepatocellular carcinoma 11 Exosomal long noncoding RNA CRNDE-h as a novel serum-based biomarker for diagnosis and prognosis of colorectal cancer 12 Plasma long noncoding RNA protected by exosomes as a potential stable biomarker for gastric cancer 13 Exosome-mediated transfer of lncRNA PART1 induces gefitinib resistance in esophageal squamous cell carcinoma via functioning as a competing endogenous RNA

14
The diagnostic/prognostic potential and molecular functions of long noncoding RNAs in the exosomes derived from the bile of human cholangiocarcinoma 15 Exosome-mediated transfer of lncRNA-SNHG14 promotes trastuzumab chemoresistance in breast cancer 16 Exosome-transmitted long noncoding RNA PTENP1 suppresses bladder cancer progression 17 Serum long noncoding RNA MALAT-1 protected by exosomes is upregulated and promotes cell proliferation and migration in nonsmall cell lung cancer 18 Tumor-derived exosomal long noncoding RNAs as promising diagnostic biomarkers for prostate cancer 19 Identification of an exosomal long noncoding RNA SOX2-OT in plasma as a promising biomarker for lung squamous cell carcinoma 5 BioMed Research International 2.2. Selection of Studies. Studies that met these criteria were included: (1) include all types of solid tumor patients; (2) all solid tumor patients were diagnosed by clinical histopathology; (3) included healthy individuals as a control group; (4) provided the expression levels of exosomal lncRNA; (5) provided the method to testify to the existence of related exosomes; (6) evaluated a liquid sample type; (7) included data about the diagnostic significance of exosomes in all types of cancer patients; and (8) included sample size, control group size, and sensitivity and specificity. Studies with the following characteristics were excluded: (1) duplicate literature; (2) insufficient data; (3) meta-analyses, letters, animal experiments, and reviews; (4) noncancer research; (5) articles that did not verify the presence of exosomes; articles that did not have diagnostic significance data related to the exosomal lncRNA; and (7) studies with a control group that did not meet the requirements.
This study was independently implemented through 2 investigators. In the case of discordance, there was discussion with a third researcher to reach a consensus. The inclusion and exclusion of two investigators was established in Table S2. The k-agreement was calculated between two investigators when deciding to include exclude articles. The k-agreement was 0.942 (p < 0:005), which indicated the great inter-rater agreement between the two independent investigators and the included studies meet the selective standards.

Data Extraction.
We collected the true-negative, falsenegative, true-positive, and false-positive values from the selected literature. For literature that did not provide these data, we calculated these based on the sensitivity and specificity and the number of samples. Moreover, we collected the author name, publication year, country, and ethnicity of the study population, cancer type, lncRNA, sample type, number of the case-patients, and number of control patients, as well as the true positivity, false positivity, true negativity, false negativity, sensitivity, and specificity. Although some articles did not relate clear information, we were able to contact the authors for clarification.

Quality Assessment.
According to the Quality Assessment of Diagnosis Accuracy Studies (QUADAS-2) criteria (http://prisma-statement.org), we evaluated the quality of the included studies using RevMan 5.3 software [31].   BioMed Research International The QUADAS-2 included 14 questions about the risk of bias of the included article. Answers included yes, not clear, and no, which corresponded to scores of −1, 0, and 1, respectively. We eliminated low-quality studies from our analysis. The quality of the literature was independently assessed by 2 investigators.
2.5. Statistical Analysis. Statistical analysis was performed by Stata 14.0 software (Stata, College Station, TX). We applied the bivariate random-effects regression model to combine the effect values of all included studies. The analysis included sensitivity, specificity, diagnostic likelihood ratio negative (DLR-), diagnostic likelihood ratio positive (DLR+) with corresponding 95% confidence intervals (CI) [32].The Higgin's I 2 and Cochran's Q tests were also included in the analysis [33]. Moreover, we used the kappa statistic to analyse the concordance between diagnosis on exosomal lncRNA and clinical histopathology. Furthermore, we plotted the bivariate boxplot to roughly assess the heterogeneity of the study.
We also plotted the summary receiver operating characteristic curve to calculate the pooled under the curve (AUC) value [34].
To further explore the potential heterogeneity between the included studies, we performed a subgroup analysis by Stata software 14.0. The included studies are divided into 6 subgroups based on tumor type, number of samples, and exosome isolation. We performed pooled SEN, SPE, DLR+, DLR-, diagnostic score, and diagnostic odds ratio analysis. Furthermore, Deek's funnel plot asymmetry test was performed to show the publication bias in the study, with the unequally distributed visual funnel plot or a p value less than 0.5 indicating significant publication bias in the study [35].
To explore the more clinical significance of the study, we plotted the Fagan plot showing the relationship between previous probability, likelihood ratio, and posterior probability. We also performed a likelihood ratio scattergram to determine the application value of exosome lncRNA in the clinical diagnosis.

Characteristics and Quality of the Included Studies.
The procedure for selecting the included literature in this study is shown in Figure S1. We obtained 863 potential related articles based on the need for the research and the search criteria of the electronic literature database. Because of repetition, 411 articles were excluded. We then excluded 424 articles by manually screening the title and abstract of the article. We screened the full text and excluded 9 articles that did not focus on the topic (n = 3) or had incomplete data (n = 6). Finally, we obtained 19 articles. Several studies have been carried out simultaneously in the literature. Yazarlou [41] reported the lncRNA LINC00152 in exosomes, Wang et al. [42] showed the lncRNA HOTAIR in exosomes, Liu et al. [43] showed the lncRNA CRNDE-h in exosomes, Ge et al. [44] illustrated the lncRNA ENST00000588480.1 in exosomes, Zhang et al. [45] reported the lncRNA MALAT-1 in exosomes, Xue et al. [46] reported the lncRNA UCA-1 in exosomes, Dong et al. [47] reported the lncRNA SNHG14 in exosomes, Kang et al. [48] reported the lncRNA PART1 in exosomes, Sun et al. [49] reported the lncRNA LINC00161 in exosomes, Wang et al. [50] reported the lncRNA H19 in exosomes, Xu et al. [51] reported the lncRNA ENSG00000258332.1 in exosomes, Zhang et al. [52] reported the lncRNA RP11-838N2.4 in exosomes, Zhao et al. [38] reported the lncRNA HOTTIP in exosomes, Zheng et al. [53] reported the lncRNA PTENP1 in exosomes, and Teng et al. [54] reported the lncRNA SOX2-OT in exosomes. Thus, 19 articles were selected that included 28 studies with a total of 2084 cancer patients and 1933 controls.
The basic information of these 28 studies is shown in Table 1. These studies were performed between 2014 and 2019. Four studies were conducted in Iran, and 25 studies were conducted in China. The sample size of the studies ranges from 60 to 320. The sample types were plasma in 5 studies, urine in 10 studies, and serum in 14 studies. In addition, there were 10 types of cancer, including bladder cancer (   BioMed Research International cholangiocarcinoma, breast cancer, esophageal squamous cell carcinoma, hepatocellular carcinoma, nonsmall cell lung cancer, and prostate cancer. We also collected the information of exosomes that were included in the studies in Table 2. This included the isolation method, exosome content, content number, exosomes diameter range (nm), median diameter (nm), evaluation (TEM and NET), and exosome proteins. Figure 1, the outcomes of the QUADAS-2 study quality assessment indicated that the quality of the included studies was convincing. The QUADAS score is shown in Table 1. The titles of all included documents are provided in Table 3. We assumed that a score greater than 4 indicated a high-quality study and a score lower than 4 indicated a study with low quality. We included 8 poor quality studies and 20 better quality studies.  Figure S2. Most of the research fell in the middle, but 4 studies were outside, which also indicated that there may be heterogeneity between the included studies. Furthermore, we assessed the concordance between diagnosis on exosomal lncRNA and clinical histopathology by using the kappa statistic. The kappa statistic of all studies was 0.513, which shows that exosomal lncRNA is moderately accurate to diagnose solid tumors. Moreover, to further explore the concordance between diagnosis on exosomal lncRNA and clinical histopathology, we analyzed the kappa statistic in different types of solid tumors and different sample types. The concordance of blood samples (kappa statistic, 0.523) was better than that of urine samples (kappa statistic, 0.494). The concordance of digestive system tumors (kappa statistic, 0.542) was better than that of urinary system tumors (kappa statistic, 0.511).

Diagnostic
Performance. Sensitivity and specificity are shown in Figure 2 as a forest plot. The combined value of sensitivity in the 29 studies was 0.74 (95% CI, 0.7-0.78), and the combined value of specificity was 0.81 (95% CI, 0.78-0.83). DLR+ and DLR-are shown in Figure 3 as a forest plot. The combined value of DLR+ in the 29 studies was 3.86 (95% CI, 3.31-4.51), and the combined value of DLR-0.32 (95% CI, 0.27-0.37). The pooled odds ratio of the 29 studies was 12.21 (95% CI, 9.31-16), as shown in Figure 4. Furthermore, Figure 5 shows that the AUC was 0.85 (95% CI, 0.81-0.88). These results demonstrated the high diagnostic efficacy of liquid exosomes in cancer patients. Table 4, Figure S3, and Figure S4, we found a difference between digestive system tumors and urinary system tumors in terms of sensitivity (0. 73  . In summary, our data illustrated that the diagnostic power is higher in digestive system tumors, the bivariate boxplot ( Figure S3E) showed that 6 studies fell in the middle, but 2 studies were outside. The bivariate boxplot ( Figure S4E) showed that 14 studies fell in the middle, but 1 study was outside, which also indicated that the heterogeneity between the digestive system tumor group was more significant than that of the urinary system tumor group. We also found a difference between the different sample types in Figure S5, Figure S6, and Table 4. Blood samples had higher sensitivity (0.74), specificity (0.80), and AUC (0.84), indicating that blood samples were more reliable than urine samples. The bivariate boxplot ( Figure S5E) showed that 15 studies fell in the middle, but 3 studies were outside. The bivariate boxplot ( Figure S6E) showed that 8 studies fell in the middle, but 2 studies were outside. This also indicated that the heterogeneity between the urine groups was more significant than the blood group. Moreover, we found that the group with larger sample sizes (>70) ( Figure S7G) and the subgroup with fewer samples (<70) ( Figure S8G) had  We also performed the subgroup analysis based on other basic information such as age, sex, ethnicity, and country. We found that there were no significant differences between the groups, so they were not discussed.

Subgroup Analysis. As shown in
3.6. Publication Bias. As shown in Figure 6, the p value of the Deek's funnel plot asymmetry test was 0.48, which is greater than 0.05, indicating that these studies did not have publication bias.

Clinical Significance of the Study.
To explore the clinical significance of this study, the Fagan plot (Figure 7) was performed to show the prior probability of 50%, likelihood ratio of 4, and the post probability of 73%. Furthermore, we plotted the likelihood ratio dot plot (Figure 8). The upper left limit LRP was greater than 10, and the LRN was less than 0.1, indicating that the diagnosis could be diagnosed and excluded. An upper right limit LRP greater than 10 and the LRN greater than 0.1 suggested that the diagnosis could be confirmed. A lower left limit LRP less than 10 and LRN less than 0.1 indicated that the diagnosis could be excluded. A lower right limit LRP less than 10 and LRN greater than 0.1 indicated that neither diagnosis nor diagnosis could be ruled out. In conclusion, it is undeniable that lncRNAs in exosomes could be promising biomarkers in the diagnosis of tumors.

Discussion
However, there was controversy with respect of the clinical value regarding whether lncRNA could be useful molecules in several cancers. These findings demonstrate that lncRNAs in exosomes can be a compelling indicator of prognosis in human solid tumors, but further confirmation by dependable analyses is required. Therefore, a meta-analysis was performed to account for whether it played a critical role in diagnosis for cancer patients.
This systematic literature review and meta-analysis included 19 articles with a total of 4017 patients with 10 types of malignant tumors. The combined value of sensitivity in 29 studies was 0.74 (95% CI, 0.7-0.78), and the combined value of specificity was 0.81 (95% CI, 0.78-0.83). The combined value of DLR+ in 29 studies was 3.86 (95% CI, 3.31-4.51), and the combined value of DLR-was 0.32 (95% CI, 0.27-0.37). The pooled odds ratio of the 29 studies was 12.21 (95% CI, 9.31-16). The AUC was 0.85 (95% CI, 0.81-0.88). The kappa statistic of all studies was 0.513, which shows that exosomal lncRNA has moderate concordance in diagnosing solid tumors. These results demonstrated the high diagnostic efficacy of liquid exosomes in cancer patients. In our report, the lncRNAs in exosomes were defined by an authoritative method, thus our results are credible. The outcome indicates that there may be heterogeneity between the included studies. And, the results of sensitivity, specificity, and DLR show they can be highly diagnostic efficacy biomarkers. In addition, we conducted subgroup analyses including sex, ethnicity, year, sample type, sample number, and cancer type, which showed    . Thus, the potential biological mechanisms of exosomal lncRNA in the digestive system should be studied further in future studies. The bivariate boxplot indicated that the heterogeneity of the digestive system tumor group was more significant than that of the urinary system group. Moreover, our outcome also suggested that the heterogeneity between the blood group was more significant than that of the urine group, suggesting that we could adopt a bloodexosomes isolation strategy in our further research to collect better data.
With the development of biomarkers in cancer therapy, more specific and high-efficiency molecules have been used clinically, such as circulating tumor DNA (ctDNA), circulating tumor cells, microRNA, lncRNA, and circRNA. For instance, Lee et al. illustrated that patients with detectable ctDNA, a promising prognostic biomarker, showed a trend toward higher risk for disease recurrence than those without detectable ctDNA [55]. Circulating tumor cells can enable early cancer detection, tumor dynamics assessment, minimal residual disease detection, and therapy monitoring [56]. Our study indicated that mir-24-3p, a type of miRNA, can promote cell migration and invasion by targeting TEL2 in nasopharyngeal carcinoma as a biomarker [57]. SNHG1, a novel lncRNA of the SNHG family, indicated that SNHG1 could be a compelling prognosis indicator in human solid tumors [29]. Compared with traditional biomarkers such as AFP [58], CEA [59], and CA199 [60], these nextgeneration molecules represent high potential in diagnosis value. Nonetheless, the novel biomarkers we mentioned previously still need to be further verified in the future. Therefore, as clinical researchers, it is our duty to explore novel biomarkers and cells that help doctors determine the clinical strategy for cancer patients.
Additionally, our team was also the first to summarize the decade of exosomal long RNA species, suggesting the significant role of exosomal lncRNA in cancer diagnosis and clinical therapy. Accumulating reports have illustrated that exosomal lncRNA is related to many processes of cancer deterioration, such as tumor cell-cell communication, drug resistance, invasion, migration, immune response, and angiogenesis [61]. Wu

11
BioMed Research International epithelial ovarian cancer-derived exomes in the coculture system, by transferring lncRNAs, restored the migration of endothelial cells which had been inhibited by TAM-derived exosomes through targeting the miR-146b-5p/TRAF6/NF-κB/MMP2 pathway, which suggests that exosomal lncRNAs play a powerful role in the regulation of the tumor microenvironment [62]. Moreover, some studies have reported that exosomal lncRNA can regulate the tumor microenvironment to influence the invasion, growth, metastasis, and prognosis of tumors [63,64]. Other research has also found that exosomal lncRNA can induce angiogenesis in cancer cells and affect tumor prognosis, growth, and invasion [65,66]. Pan et al. also found that exosomal lncRNA was additionally related to lymphatic metastasis in gastric cancer patients [67], which showed that tumors can indirectly enhance cellular migration and invasion abilities through exosomal lncRNA. Moreover, some studies have revealed that exosomal lncRNA plays a key role in the acquired drug resistance in some cancers [68,69]. Takahashi et al. found that LINC-ROR, derived from tumor cells, was enriched in the extracellular vesicles, which can lead to an increased level of chemoresistance in hepatocellular cancer cells. This suggests that hepatocellular cancer cells may use exosomal lncRNA to improve chemoresistance [70]. The potential biological mechanisms of exosomal lncRNA are summarized in Figure 9. Moreover, one of the most significant previous findings in exosomes for our report is that exosomes demonstrated a cargo biological function to deliver a number of molecules, such as drug, ncRNAs, and mRNAs [61]. Therefore, this special biological feature has led to many questions and a need for further investigation. Further work needs to be done to establish whether exosomes are involved in malignant tumor invasion as well as migration. Moreover, because of the specificity [71], stability [72], and accessibility [73], we can not only use exosomes to diagnose solid tumors but we can also apply them to gene therapy.

Conclusions
As a type of special molecule, lncRNA in exosomes could be a promising bioindicator for the diagnosis and prognosis of cancers. Our study provides convincing evidence through meta-analysis. However, further works are required in the future. Finally, we hope our results encourage more researchers to examine the prognostic and diagnostic role of lncRNA in exosomes as well as explore the underlying biomechanisms in different cancers.

Data Availability
The data supporting the research are included in the article.

Conflicts of Interest
The authors declare no conflicts of interest in this work.

Acknowledgments
Zilong Wu thanks the doctors for fighting the outbreak of new coronavirus infection in China. Figure S1: the study selection procedure for meta-analysis was conducted to the PRISMA statement. Supplementary 2.  Figure 9: The potential biological mechanisms of exosomal lncRNA. Exosomes can directly promote tumor growth, invasion, migration, and drug resistance. And indirectly promote tumor growth, invasion, migration by promoting angiogenesis, and regulate microenvironment.