Systematic review and meta-analysis of prognostic microRNA biomarkers for survival outcome in laryngeal squamous cell cancer

Background Laryngeal carcinoma is a primary malignant tumor originating from the laryngeal mucosa, and its pathogenesis is not fully understood. It is a rare type of cancer that shows a downward trend in the 5-year survival rate. In clinical practice, dysregulated microRNAs are often observed in patients with laryngeal cancer. In recent years, an increasing number of studies have confirmed that the strong biomarker potential of microRNAs. We conducted a systematic review and meta-analysis to identify and highlight multiple microRNAs as biomarkers for disease prognosis in patients with laryngeal cancer. Methods We actively searched the systematic reviews in PubMed, Embase, Web of Science and The Cochrane Library to select the studies that met the proposed guidelines. A total of 5307 patients with laryngeal cancer were included in this study to evaluate the association between microRNAs expression levels and patient outcomes. For overall survival in the clinical stage, a hazard ratio (HR) and corresponding 95% confidence interval (CI) are calculated to assess the effect of survival. Results A total of 36 studies on microRNAs and laryngeal cancer recovery were included in this meta-analysis. The selected endpoints for these studies included overall survival (OS) and disease-free survival (DFS).The comorbidities of overexpression and underexpression of microRNAs were 1.13 (95% CI 1.06–1.20, P < 0.05) and 1.10 (95% CI 1.00–1.20, P < 0.05), respectively. Conclusion MiRNA-100, miRNA-155, miRNA-21, miRNA-34a, miRNA-195 and miR-let-7 are expected to be potential noninvasive and simple markers for laryngeal cancer.


Introduction
Laryngeal cancer is the second most common head and neck cancer (HNC), accounting for 1 to 5% of systemic malignancies. A recent study showed that the 5-year accurate diagnoses, provide information for patients and guide treatment.
MicroRNAs are small endogenous RNA molecules containing approximately 22 nucleotides that regulate gene expression by binding to specific messenger RNAs [4]. MicroRNAs are often located in genomic regions associated with cancer and are involved in the regulation of gene function [5]. It is believed that microRNAs can be used as tumor suppressor genes as well as oncogenes, and play an important role in the occurrence and development of cancer [6,7]. MicroRNAs are shortstranded noncoding RNAs that play a very important role in malignant tumorigenesis, and they regulate the expression of tumor-related genes at the posttranscriptional level through specific binding to their target genes [8,9]. MicroRNAs are highly stable and can be measured in biological fluids, including serum, plasma, and saliva, allowing for quick results and repeated analysis during and after treatment. This property strengthens the role of microRNAs as biomarkers and their great potential in the field of oncology [10]. MicroRNAs have been widely researched, and the results indicate that they may provide a sensitive method for the detection, monitoring and prognosis of laryngeal squamous cell carcinoma [11,12]. In this study, we conducted a systematic review of the existing literature on this topic in PubMed, Embase, Web of Science and The Cochrane Library. Then, we conducted a meta-analysis of the survival rates (including overall survival [OS] and disease-free survival [DFS]) of patients expressing different levels of microRNAs.

Selection criteria
This meta-analysis was conducted based on an originally conceived protocol. To be qualified, studies had to meet all the following criteria: (1) the study discussed the expression of microRNAs in patients with laryngeal squamous cell carcinoma. (2) The study investigated the relationship between microRNAs expression levels and the survival status of cancer patients. (3) The study was published in full text and reported the risk ratio (HR) of DFS or OS based on microRNA status, with a confidence interval (CI) of 95%.
Studies that met one of the following criteria were excluded: (1) case reports, conference proceedings, letters, or reviews/meta-analyses; (2) thyroid and oral tumors; (3) animal studies; (4) laboratory studies; or (5) incomplete data (no NLR HR for OS/DSS). Incomplete study data (for example, studies that included only Kaplan-Meier curves or HRs that did not report a 95% confidence interval) were not initially ruled out. When this occurred, we contacted the authors and attempted to obtain the raw data.

Data extraction
To ensure the accuracy of data extraction, two authors extracted data according to the inclusion and exclusion criteria, and a third author assisted in making the final decision on the disputed information. Data were extracted, including author name, year of publication, country, tumor type, total number of patients and test method, follow-up time, microRNAs expression, HR and 95% CI. Meta-analyses were performed as required by the item statement of the systematic review and meta-analysis priority report.

Quality assessment
Methodological quality was systematically reviewed and meta-analyzed based on the quality assessment template of the National Heart, Lung and Blood Institute (NHLBI). Based on the opinions of the two reviewers, each study was subjectively rated as having a "high", "moderate" or "low" risk of bias rating (Table 1).

Statistical methods
Meta-analysis was performed on data from multiple included studies. Comprehensive meta-analysis software (RevMan5.3 and Stata 15.1) was used. A combined HR > 1 indicates that the prognosis of the group with elevated microRNA expression was poor. The red square shows a combined effect estimate of survival in patients with laryngeal cancer arranged by microRNA expression. We attempted to explore the heterogeneity of the results through subgroup analysis.

Publication bias
We used Egger's bias assessment graph test to construct funnel plots (scatter plots constructed using standard error [Y axis] and log (HR) [X axis]) for all included studies. The symmetry of the study distribution on the regression line is inversely proportional to the size of the publication bias in the meta-analysis. In the process of data selection and quality assessment, any differences were resolved through discussion among the reviewers.

Study selection
We found a total of 299 records and placed them into the EndNote Web reference manager to delete duplicate articles. A total of 36 studies were included in the metaanalysis. The details are shown in Fig. 1.

Study characteristics
The studies included in this systematic review included a sample of 36 research articles. The average patient age in most studies was approximately 60. We found that these studies originated from eight countries around the world, including The Netherlands (n = 1), Italy (n = 2), Germany (n = 1), Brazil (n = 2), U.S. A. (n = 6), Athens (n = 1), Japan (n = 1), and China (n = 22). MicroRNA expression was found in preserved tissue samples (23 studies), serum (10 studies) and plasma (3 studies). Eight studies used TaqMan low-density arrays to detect micro-RNAs expression profiles, and 27 studies used RT-PCR to detect microRNAs expression. The average period for follow-up studies was 3-7 years. In all the studies included in the systematic review and meta-analysis, a total of 24 microRNAs were reported. Of the 24 micro-RNAs included, 10 microRNAs were upregulated, and 14 microRNAs were downregulated. Table 2 describes the research design and patient information for all 36 studies (Table 2).

Meta-analysis
The association of OS with at least one microRNA in laryngeal squamous cell carcinoma (LSCC) was reported in 36 studies. The expression of these molecules was detected primarily at the RNA level, and the upregulated expression was compared with that in normal tissues. The meta-analysis pooled HR and 95% CI (n = 36) from studies with OS as the endpoint, providing an overall effect of a downregulated microRNA size estimate (HR) of 1.10 (95% CI 1.00-1.20), while the estimated pooling effect size (HR) of upregulated microRNAs was 1.13 (95% CI 1.06-1.20). The effect size estimate (HR) for studies using DFS as their survival endpoint was 2.57 (95% CI 1.56-4.23). When OS was used as a survival endpoint, there was no significant difference between the upregulated and downregulated expression of these microR-NAs. This finding suggests that any change in microRNA expression is associated with a lower survival rate in patients with LSCC. The apparent overall heterogeneity between these studies using OS as a survival endpoint was found to be high (I 2 = 88.05;). Similar results were found in DFS (Figs. 2, 3 and 4).

Subgroup analysis
MicroRNAs of six subgroups were included in Stata 15.1 to obtain the combined forest plot (Fig. 5). The heterogeneity between groups was large (I 2 = 94.3%), so we conducted a subgroup analysis. We divided microRNA subgroups according to the survival end points (OS and  . 11).

miRNA-195
Three studies revealed the expression of miR-195 in patients with laryngeal cancer (Fig. 6). All three studies showed that the expression of the miR-195 gene was downregulated in patients with LSCC. The combined effect size estimate (HR) was 0.33 (0.20-0.55; P < 0.05).
All three studies showed that downregulation of miR-195 expression resulted in higher survival rates. The heterogeneity of the studies was 0 (I 2 = 0%), indicating that the three studies supported the same finding (Fig. 6).

miRNA-100
Two studies investigated miR-100 expression in LSCC patients (Fig. 7). Both studies showed that miRNA-100 expression was upregulated in patients with laryngeal squamous cell carcinoma. The combined effect size estimate (HR) was 1.88 (95% CI 0.83-4.26). As shown in     Fig. 7, there was significant heterogeneity between the two groups (I 2 = 96%). The work of Owen et al. [43] contradicts the work of another author. To identify the source of the difference, we searched existing databases and found that high miR-100 expression levels were associated with lower survival rates in oral squamous cell carcinoma (46 cases) and esophageal carcinoma (47 cases). By removing this study, the merger heterogeneity effect of the upregulated group will be reduced (Fig. 7).

miRNA-21
Two studies reviewed the expression of miR-21 in patients with laryngeal squamous cell carcinoma (Fig. 8).
The two studies demonstrated similar results regarding miR-21 expression in laryngeal squamous cell carcinoma (I 2 = 0%). This similarity in results indicated that the two studies agreed that overexpression of miR-21 in laryngeal cancer tissue leads to poor survival. The combined effect estimate (HR) was statistically significant, with a value of 1.78 (95% CI 1.23-2.57; P < 0.05) (Fig. 8).

miRNA-155
Two studies reviewed miR-155 expression in patients with laryngeal squamous cell carcinoma (Fig. 9). The combined heterogeneity of the two studies was very low (I 2 = 40%) and thus could be ignored. Both studies showed that overexpression of miR-155 in laryngeal squamous cell carcinoma resulted in reduced survival.

miRNA-let-7
Three studies reviewed the expression of miR-let-7 in patients with laryngeal squamous cell carcinoma (Fig. 10). The three studies demonstrated similar results regarding miR-let-7 expression in laryngeal squamous cell carcinoma (I 2 = 0%). This similarity in results indicated that the three studies agreed that overexpression of miR-let-7 in laryngeal cancer tissue leads to poor survival. The combined effect estimate (HR) was statistically significant, with a value of 2.74 (95% CI 1.94-3.86; P < 0.05) (Fig. 10).

Disease-free survival group miRNA-34a
Two studies reviewed the expression of miR-34a in LSCC patients (Fig. 11). Both studies showed that miR-34a was downregulated in laryngeal squamous cell carcinoma. The combined effect size (HR) was 2.93 (95% CI 1.24-6.91; P < 0.05). In Fig. 11, we can clearly see that when the two studies were combined, the overall heterogeneity was high (I 2 = 90%). Based on a review of published studies, it has been demonstrated that downregulation of miR-34a promotes laryngeal cancer cell proliferation and migration by targeting cyclin D1 [49]. The prognostic results obtained by Ogawa et al. contradicted the experimental phenomenon. Therefore, we consider that the study of Ogawa et al. led to a high degree of heterogeneity (Fig. 11). Figures 12 and 13 indicate that in terms of survival results, the funnel chart shows slight asymmetry between different studies. In the studies included in the analysis, visual observation of the funnel shape was the first step we used to assess publication bias. The evidence of asymmetry revealed the potential publication bias, and the publication bias of OS and DFS was detected by Begg and Egger methods. This suggests that a few studies may have been missing from our search, including results related to increased survival (Figs. 12,  13).

Discussion
MicroRNAs are a class of nonprotein coding small RNAs 21-25 nucleotides in length which play a key role in gene regulation [50]. Circulating biomarkers are often used by clinicians to diagnose the early stages of LSCC. In contrast, invasive examination is not conducive to observing the progress of LSCC, and patients often reject invasive procedures because the examination may cause them pain. MicroRNAs are very stable, and clinical studies investigating their role have provided reproducible and consistent results [51]. In addition, we can easily detect microRNAs by RT qPCR for LSCC diagnosis. Circulating microRNAs can resist RNase A digestion and other adverse conditions, including repeated freeze-thaw cycles and long-term storage under unsuitable pH conditions [52]. MicroRNAs are promising biomarkers for early diagnosis and prognosis [53].
Our meta-analysis examined 36 articles including a patient population of 5307 individuals. In the different studies, there were 10 upregulated microRNAs and 14 downregulated microRNAs. Then, according to the expression level, they were divided into two groups, which were put into the tool to obtain the forest map.
It is generally believed that if I 2 is less than 50% this means that the heterogeneity between studies is small, and if it is greater than 50% then the heterogeneity is large. It does not make sense to combine HR of all microRNAs because their heterogeneity is too high. For  prognostic biomarkers, it is more appropriate to focus on the prognostic efficiency of each microRNA. In this study, six microRNAs (miR-34a, miR-195, miR-100, miR-21, miR-155 and miR-let-7) had high prognostic value. The remaining microRNAs were of low value.
The two groups that showed great heterogeneity were miRNA-34a and miRNA-100. We tried to find the source of the heterogeneity by exploring the basic biomolecular experimental findings. We conclude that high expression of five microRNAs (miR-34a, miR-100, miR-21, miR-155 and miR-let-7) indicates poor overall survival, while high expression of miRNA-195 indicates improved overall survival. DFS is only reported only in miR-34a.
In 2008, Thian-Sze et al. conducted the first study to detect the expression level of miR-195 in head and neck tumors. Real-time quantitative PCR was used to evaluate the upregulation of miR-195 expression in four types of tongue cancer cells [54]. In 2019, Tian et al. studied two human laryngeal squamous cell carcinoma cell lines with different in vitro radiosensitivity. It was found that miR-195 is a direct downstream target of DGCR5. Moreover, the overexpression of miR-195 enhanced the radiosensitivity of Hep-2R cells [55].
Lidia et al. isolated extracellular vesicles from the plasma of 60 patients with thyroid cancer in 2020 and analyzed vesicle microRNAs. The area under the curve (AUC) value of miR-let-7 was 0.814 [56]. This result shows that miR-let-7 has a high diagnostic ability.
Yao et al. confirmed that the miR-34a-5p/Axl axis plays an aggressive role in oral cancer cells through the Akt/ GSK-3 and β/β-catenin/Snail signal transduction pathways and may be a therapeutic target for oral squamous cell carcinoma [57].
The latest study by Weina et al. found that in the epithelial-mesenchymal transition of LSCC cells induced by TGF-β, miR-155HG can regulate epithelial-mesenchymal transition markers through the miR-155/SOX10 axis [58]. The author concluded that the miR-155HG/ miR-155-5p/SOX10 axis is a key part of the process in promoting LSCC.
The studies in this analysis were drawn from all currently available LSCC studies that investigate microRNA expression and prognosis. Compared with other studies,  the results of this study show certain advantages in accuracy. One of the shortcomings of our analysis is that the sample size of this study is relatively small. Because our study was retrospective, we were limited by the existing literature. In addition, we only chose the study endpoints of OS and DFS because of the lack of other survival outcomes, such as relapse-free survival. In addition, many risk factors contribute to the development of laryngeal squamous cell carcinoma. In the future, we can also collect data on early glottic diseases that suggest the occurrence of laryngeal cancer.

Conclusions
Overall, despite the limitations of our review, our data still provide evidence suggesting that miRNA-100, miR-155, miR-21, miR-34a, miR-195 and miR-let-7 are potential noninvasive and simple humoral tumor markers for laryngeal cancer.