https://orcid.org/0000-0003-4760-4293
Figures
Abstract
Previous work has demonstrated that the expression of microRNA-21 (miR-21) is implicated in cervical cancer (CC). However, little is known regarding its associations with clinical parameters. We first conducted a meta-analysis using data from Gene Expression Omnibus (GEO) microarrays and The Cancer Genome Atlas (TCGA). Then, enrichment analysis and hub gene screening were performed by bioinformatic methods. Finally, the role of the screened target genes in CC was explored. According to the meta-analysis, the expression of miR-21 in cancer tissues was higher than in adjacent nontumor tissues (P < 0.05). In addition, 46 genes were predicted as potential targets of miR-21. After enrichment analyses, it was detected that these genes were enriched in various cancer pathways, including the phosphatidylinositol signaling system and mammalian target of rapamycin (mTOR) signaling pathway. In this study, bioinformatic tools and meta-analysis validated that miR-21 may function as a highly sensitive and specific marker for the diagnosis of CC, which may provide a novel approach to the diagnosis and treatment of CC.
Citation: Deng Z-M, Chen G-H, Dai F-F, Liu S-Y, Yang D-Y, Bao A-Y, et al. (2022) The clinical value of miRNA-21 in cervical cancer: A comprehensive investigation based on microarray datasets. PLoS ONE 17(4): e0267108. https://doi.org/10.1371/journal.pone.0267108
Editor: Ritu Kulshreshtha, Indian Institute of Technology Delhi, INDIA
Received: February 2, 2021; Accepted: April 3, 2022; Published: April 29, 2022
Copyright: © 2022 Deng et al. This is an open access article distributed under the terms of the Creative Commons Attribution License, which permits unrestricted use, distribution, and reproduction in any medium, provided the original author and source are credited.
Data Availability: The datasets generated and/or analyzed during the current study are available in the GEO repository (GSE86100: https://www.ncbi.nlm.nih.gov/geo/query/acc.cgi?acc=GSE86100; GSE30656: https://www.ncbi.nlm.nih.gov/geo/query/acc.cgi?acc=GSE30656; GSE19611: https://www.ncbi.nlm.nih.gov/geo/query/acc.cgi?acc=GSE19611) and TCGA datasets (https://portal.gdc.cancer.gov/legacy-archive/search/f).
Funding: This work was supported by the Key Research and Development Program of Hubei Province in the form of a grant (number 2020BCB023) and general project of national Natural Science Foundation of China in the form of a grant (number 82071655) to YC.
Competing interests: The authors have declared that no competing interests exist.
1. Introduction
Cervical cancer (CC) is one of the most common gynecological carcinomas worldwide, resulting in an unacceptably high mortality rate. Despite continuous advances in the early screening and treatment of CC, its incidence has shown a younger trend in recent years [1]. In addition, the prognoses of patients with CC are still poor. Hence, revealing the molecular characteristics underlying CC and exploring potential therapeutic targets is imperative.
MicroRNAs (miRNAs), a group of small noncoding RNAs, are thought to regulate the expression of a large number of protein-coding genes and are implicated in a variety of biological processes [2]. MicroRNA-21 (miR-21), an oncogenic microRNA involved in angiogenesis, tumor invasion, and tumor metastasis, was proved to be altered in a variety of cancers, such as gastric cancer, non-small cell lung cancer, bladder cancer, prostate cancer, pancreatic cancer, glioblastoma, and breast cancer [3–9], etc. In CC, a substantial body of evidences had verified that miR-21 was differentially expressed and regarded as a potential therapeutic target [10, 11]. However, little is known regarding its association with clinical parameters.
Here, we first calculated the expression of miR-21 in CC based on data from Gene Expression Omnibus (GEO) microarrays and The Cancer Genome Atlas (TCGA). And then, diagnostic meta-analysis was used to clarify the diagnostic value of miR-21. What’s more, bioinformatic analyses, containing Kyoto Encyclopedia of Genes and Genomes (KEGG), Gene Ontology (GO), and protein-protein interaction (PPI) network analysis, were operated to mining the underlying mechanisms of miR-21. In a word, meta-analysis and further bioinformatic techniques were aimed to explore the role of miR-21 in CC.
2. Materials and methods
2.1 Data collection and the expression of miR-21
A microarray search in CC was conducted in the GEO database (https://www.ncbi.nlm.nih.gov/geo/). Those datasets without expression of miR-21 were excluded. Next, the expression matrix and clinical parameter data for miR-21 were extracted from TCGA (https://cancergenome.nih.gov/). Correlations between the clinical parameters and miR-21 were calculated by IBM SPSS 22.0. The number, mean (M), and standard deviation (SD) of control and experimental group were obtained based on the expression profile information. The expression of miR-21 in each study were visualized by plotting scatter diagrams in GraphPad Prism 8.
2.2 Comprehensive meta-analysis
Meta-analysis was carried out under the R environment using Meta package. The chi-squared test of Q and the I2 statistic were adopted to assess the potential heterogeneity among the studies. Within groups, comparisons were carried out by t-tests. The ratio of the count data was expressed as percentage (%) and compared using χ2 tests. In addition, the model used in the statistical heterogeneity was decided according to the I2 and P value in forest plots. Generally, the random-effect model was adopted when I2 > 50% or P < 0.05; otherwise, the fixed-effects model was adopted [12]. To explore the potential sources of heterogeneity, sensitivity analysis was also conducted.
To assess the diagnostic value of miR-21 for CC, the pooled specificity, sensitivity, positive likelihood ratio (PLR), negative likelihood ratio (NLR), and diagnostic odds ratio (DOR) were computed by meta-Disc software. Furthermore, summary receiver operating characteristic (sROC) curves were also drawn in meta-Disc, whereas the ROC curves of each study were plotted in GraphPad Prism. P < 0.05 was considered to be meaningful.
2.3 Latent targets of miR-21 in CC and the enrichment analysis
On the one hand, the mRNA targets of miR-21 were predicted by the online database miRWalk2.0 (http://zmf.umm.uni-heidelberg.de/apps/zmf/mirwalk2/) [13]. Only those genes projected by more than 6 of the servers were included. On the other hand, the highly expressed genes in CC were downloaded from GEPIA2 (http://gepia2.cancer-pku.cn/) [14]. The intersection of the genes from the two sources were recognized as targeted genes of miR-21. This procedure was implemented in Vennny 2.1.0 (https://bioinfogp.cnb.csic.es/tools/venny/index.html).
Subsequently, the enriched pathways of those genes were analyzed using R software [15, 16]. The circular plot of the KEGG pathways was plotted by an online platform for data analysis and visualization (http://www.bioinformatics.com.cn). And then, String (http://string-db.org/cgi/input.pl) software was used to construct and visualize the protein-protein interaction (PPI) network between those genes [17]. Based on the PPI network, the hub nodes were investigated by the maximal clique centrality (MCC) algorithm in CytoHubba, a plug-in in Cytoscape [18]. According the value of MCC, the top 4 were defined as hub genes.
2.4 The expression of hub genes and further validation
Given that four genes were identified, their expression boxplots and disease-free survival (DFS) curves were downloaded from GEPIA2 initially. Subsequently, the correlations between miR-21 and the identified hub genes were verified via LinkedOmics (http://www.linkedomics.org/) [19]. Both these two steps were aimed to shrink the scope of genes and further obtain final targets.
We next validated the miR-21 binding site on targets via miRactDB database on the one hand [20], and GeneCards (http://www.genecards.org/#) were applied to connect the final targets with the interested KEGG enriched pathways on the other hand [21]. In detail, the genes involved in the interested pathways were collected, and then the STRING were performed to constructed the PPI network between those final targets and pathway genes.
3. Results
3.1 The expression of miR-21 in GEO microarrays
The workflow of the study is illustrated in Fig 1A, while Fig 1B displays the flow chart of the search and selection process. Eventually, 3 microarrays from GEO database (GSE86100 [22], GSE30656 [23], GSE19611 [24]) and TCGA data, including 376 cases in total, were included in this study. The specific information is displayed in Table 1. In addition, the expression data of miR-21 from the tumor tissues and adjacent noncancerous tissues (serving as control groups) are shown in Fig 2A–2D. The CC groups had a significantly higher miR-21 expression than the control groups in GSE86100, GSE30656, and TCGA (P = 0.0141, P < 0.0001, P < 0.001), while no notable distinction was detected in GSE19611 (P = 1.009).
(A) The flow diagram of this study. (B) Flow chart of the literature screening process and results. Note: miR-21, microRNA-21; GEO, Gene Expression Omnibus; TCGA, The Cancer Genome Atlas; GEPIA, Gene Expression Profiling Interactive Analysis; DFS, disease-free survival; GO, Gene Ontology; KEGG, Kyoto Encyclopedia of Genes and Genomes; PPI, protein-protein interaction network.
(A-D) Scatterplot of the miR-21 expression levels in the included cervical cancer studies. (A) GSE86100; (B) GSE30656; (C) GSE19611; (D) TCGA. (E) A forest plot of miR-21 expression between CC and adjacent nontumor tissues. (F) Sensitivity analysis of the selected four studies. Note: Normal, adjacent nontumor tissues; CESC, cervical squamous cell carcinoma and endocervical adenocarcinoma; TCGA, The Cancer Genome Atlas.
3.2 The meta-analysis of the data and the diagnostic value of miR-21
According to the results of our meta-analysis (Fig 2E), the random-effects model was applied since the degree of heterogeneity were high (I2 = 89%, P < 0.01). The combined standardized mean difference (SMD) greater than zero were considered as further supporting evidence of the higher expression of miR-21. The sensitivity analysis as showed in Fig 2F, after omitting GSE30656, the combined SMD became moot (95% CI = -0.12–5.16), which proved that our results were not stable enough.
Next, the results of diagnostic meta-analysis included pooled specificity, sensitivity, PLR, NLR and DOR were described in Fig 3A–3E, respectively. It can be seen from Fig 4A–4E, the area under the curve (AUC) of either ROC or sROC were very high, which implied the performance of our model is good.
(A) Sensitivity; (B) Specificity; (C) Positive likelihood ratio; (D) Negative likelihood ratio; (E) Diagnostic odds ratio.
(A-D) Receiver operating characteristic (ROC) curves based on 4 studies. (A) GSE86100; (B) GSE30656; (C) GSE19611; (D) TCGA. (E) Summary receiver operating characteristic (sROC) curve for the diagnostic value of miR-21 in cervical cancer.
Furthermore, the relationship between miR-21 and clinicopathological features were also explored in TCGA database based on 309 CC samples and 3 adjacent nontumor tissues. As illustrated in Table 2, miR-21 expression in CC was much higher than that in adjacent noncancerous tissues. In addition, in patients with distant metastasis (M1), it was significantly lower than that in patients without distant metastasis (M0) (P < 0.05). Nevertheless, there is no statistical significance between miR-21 and other clinical pathological features.
3.3 Identifying the promising miR-21 target genes in CC using bioinformatics
A total of 169 genes were obtained by more than six algorithms from miRWALK2.0, while 5758 overexpressed genes were collected from GEPIA2. After intersection, 46 predicted genes were screened (Fig 5A and S1 Table).
(A) The intersection of GEPIA2 and miRWalk2.0 to obtain predictive target genes. (B) Gene ontology terms categorization and distribution of the miR-21 targeted genes. (C) Chord plot of the KEGG pathways. (D) Bar graph of the KEGG pathways. (E) The PPI networks of the target genes of miR-21.
To further uncover the function of miR-21 in CC, KEGG and GO annotations were performed. Details were provided in Fig 5B–5D, S2 and S3 Tables. Of note, the KEGG enrichment analysis revealed that miR-21 can make a critical difference in CC through multiple pathways, including the phosphatidylinositol signaling system and mammalian target of rapamycin (mTOR) signaling pathway (Fig 5C and 5D). Fig 5E displayed the PPI network diagrams. What’s more, the top 4 hub genes (MAF, EPAS1, PIKFYVE, and SACM1 L) were calculated via Cytoscape.
3.4 The expression and prognostic condition of miR-21 targeted genes
Subsequently, the correlations between miR-21 and those identified hub genes were verified via LinkedOmics (http://www.linkedomics.org/).The expression of the four hub genes obtained in the previous step and their DFS curves are shown in Fig 6A–6D. From the figure, we found all four genes were significantly downregulated in the CC group compared to the control group. Notably, only EPAS1 was prognosis-related (P = 0.041), while the remaining 3 hub genes were not associated with an improved DFS (MAF: P = 0.35, PIKFYVE: P = 0.24, SACM1 L: P = 0.12). The correlation between the identified target genes and miR-21 is shown in Fig 7A–7D. Similarly, there were only EPAS1 hold a weak positive correlation (r = 0.1657) with miR-21. Given this, EPAS1 were regarded as the final target of miR-21 and used in following analyses.
(A) SACM1 L; (B) PIKFYVE; (C) MAF; (D) EPAS1. Notes: Expression of the hub genes was detected in 306 cervical cancer tissues (T) and 13 adjacent nontumor tissues (N) based on the GEPIA database, with the cutoff criteria of |log2FC| ≥ 1.0 and adj. p < 0.05. SACM1 L, suppressor of actin mutations 1-like; PIKFYVE, phosphoinositide kinase, FYVE-type zinc finger containing; MAF, macrophage activating factor; EPAS1, endothelial PAS domain protein 1.
(A-D) Spearman’s analysis was used to show the correlations between the miR-21 expression levels and the validated target genes. (A) SACM1 L; (B) PIKFYVE; (C) MAF; (D) EPAS1. (E) The PPI network between EPAS1 and genes within the PI3K/AKT/mTOR signaling pathway. Notes: SACM1 L, suppressor of actin mutations 1-like; PIKFYVE, phosphoinositide kinase, FYVE-type zinc finger containing; MAF, macrophage activating factor; EPAS1, endothelial PAS domain protein 1.
According to previous GO enrichment analysis, those candidate targets are enriched in DNA-binding transcription activator activity and phosphatidylinositol phosphate phosphatase activity, both of which are associated with oncology. The KEGG analysis results revealed that mi-21 may involve in the phosphatidylinositol and the mTOR signaling pathways. One of the most prestigious phosphatidylinositol pathways is the phosphatidyl-inositol 3-kinase/serine-threonine kinase (PI3K/AKT) signaling pathway, which is implicated in the development and progression of cancers. Therefore, PI3K/AKT/mTOR was considered as a candidate signaling pathway.
In the process of validation, Table 3 illustrated that there were miR-21 binding sites both on the promotor and coding region of EPAS1 were predicted by miRactDB. Moreover, a total of 31 genes, including EPAS1 and 30 genes within the PI3K/AKT/mTOR signaling pathway downloaded from the PathCards module of GeneCards database, were applied to constructed PPI network. In Fig 7E, it’s clearly apparent that EPAS1, the final target of miR-21, were strongly interlinked with the PI3K/AKT/mTOR signaling pathway. Along this process, miR-21 and the cancer-involved pathways were contacted via EPAS1.
Discussion
A growing body of evidence suggested that miR-21 plays a fundamental role in migration, invasion, metastasis, and proliferation of breast cancer, head and neck squamous cell carcinoma, gastric cancer, colorectal cancer, etc. [25–29]. In this study, the tissues type (adjacent noncancerous tissue or CC samples) and the occurrence of distant tumor metastasis (M0 or M1) were associated with miR-21. This also indicated that miR-21 may be involved in the occurrence and development of CC.
Studies have examined that miR-21 may be a suitable diagnostic biomarker for colorectal cancer, with moderate sensitivity and specificity [30], while hold good diagnostic value in breast cancer and colorectal cancer [31, 32]. However, the potential diagnostic value of miR-21 in CC and its correlation with clinical features had seldom been studied. The present study focused on investigating the role and latent mechanism of miR-21 in CC by co-applying meta-analysis and bioinformatic approaches.
Here, increased expression of miR-21 was found in GSE86100, GSE30656 and TCGA, but in GSE19611, no significant differences were observed. A possible reason for this may be the limited number of tumor samples in GSE19611. Furthermore, a meaningless combined SMD emerged when sensitivity analysis was conducted (Fig 2F). We can see from Fig 2E, although the sample size of GSE30656 was not the largest one, its weighting ratio was the highest, reaching 25.7%. Given this, once GSE30656 was removed, the analysis results were severely fluctuated. In the future, higher quality and larger sample size studies need to be performed to obtain more instructive conclusions.
From Fig 2E, forest plot of the meta-analysis, the combined SMD was 2.50 (95% CI: 0.72–4.27, random-effects model). Additionally, the AUC value of either ROC or sROC were more than 0.7 (Fig 4A–4E). In a word, our meta-analysis demonstrated that miR-21 may serves as a highly sensitive and specific biomarker in CC. Subsequently, we performed a series of bioinformatic assays.
On the one hand, PI3K/AKT/mTOR was thought as a candidate signaling pathway after enrichment analyses [33–35]. What’s more, the effect of miR-21 on PI3K/AKT/mTOR pathway has also been demonstrated in other kinds of diseases. Based on the above information, it can be reasonably speculated that miR-21 may affect the invasion and migration of CC cells by regulating the PI3K/AKT/mTOR pathway.
On the other hand, EPAS1 were screened to be the final target of miR-21. EPAS1 is the only one that has statistically significant correlation with DFS and holds the largest Spearman correlation coefficients with miR-21 among the four targeted genes. Endothelial PAS domain protein 1 (EPAS1), which is also known as hypoxia-inducible factor-2α (HIF-2α), belongs to the family of hypoxia-inducible factors (HIFs) [36]. Hypoxia is a common feature of many solid tumors, including CC. The mTOR pathway, which is essential for cell proliferation, is repressed under hypoxia. As early as 2012, report showed that activation of the HIF-2α pathway increases the activity of mammalian target of rapamycin complex 1 (mTORC1) [37]. Moreover, a recent paper reported that HIF-2α could promote the apoptosis of breast cancer cells via PI3K/AKT/mTOR signaling pathway [38]. Although the function of EPAS1 in the tumorigenesis of CC is yet clear, we have demonstrated that EPAS1 were tightly linked with the PI3K/AKT/mTOR pathway. Thus, we speculated that EPAS1 plays an important role in CC through the PI3K/AKT/mTOR pathway.
It is undeniable that this study has several limitations. First, the composition of the samples differed somewhat due to the differences between data sources. For example, in the TCGA database, there were 309 cases of CC tissues and 3 cases of adjacent nontumor samples (Table 2 and Fig 2D), while the corresponding figures in the GEPIA2 database were 306 and 13, respectively (Fig 6). There is a risk that a mistake will occur, but not to the extent to which it could influence the reliability of the data, as the large data volumes from TCGA database were included. Next, because only four independent datasets were incorporated, we did not assess the risk of publication bias. Finally, due to the HPV infection status in TCGA are unavailable, we failed to identify the relationship of HPV-related CC and miR-21.
Taken together, the results of this study confirm that miR-21 is upregulated and plays a vital role in the occurrence and development of CC. More importantly, miR-21 has the potential to be a novel, highly sensitive, highly specific, noninvasive biomarker for the diagnosis of CC, but additional large-scale studies are required to verify its diagnostic value. The results of bioinformatic researches present a new method for exploring the pathogenesis of CC and will guide our experimental thinking.
Supporting information
S1 Table. Promising 41 target genes of miR-21 in CC obtained from the intersection of GEPIA2 and miRWalk2.0.
https://doi.org/10.1371/journal.pone.0267108.s001
(DOCX)
S2 Table. The GO enrichment analysis of predicted target genes.
https://doi.org/10.1371/journal.pone.0267108.s002
(DOCX)
S3 Table. The KEGG enrichment analysis of predicted target genes.
https://doi.org/10.1371/journal.pone.0267108.s003
(DOCX)
References
- 1. Vu M, Yu J, Awolude OA, Chuang L. Cervical cancer worldwide. Current problems in cancer. 2018;42(5):457–65. pmid:30064936.
- 2. Cai L, Wang W, Li X, Dong T, Zhang Q, Zhu B, et al. MicroRNA-21-5p induces the metastatic phenotype of human cervical carcinoma cells by targeting the von Hippel-Lindau tumor suppressor. Oncology letters. 2018;15(4):5213–9. pmid:29552160.
- 3. Li Q, Li B, Li Q, Wei S, He Z, Huang X, et al. Exosomal miR-21-5p derived from gastric cancer promotes peritoneal metastasis via mesothelial-to-mesenchymal transition. Cell Death Dis. 2018;9(9):854. Epub 2018/08/30. pmid:30154401.
- 4. Zhang HH, Huang ZX, Zhong SQ, Fei KL, Cao YH. miR21 inhibits autophagy and promotes malignant development in the bladder cancer T24 cell line. Int J Oncol. 2020;56(4):986–98. Epub 2020/04/23. pmid:32319564.
- 5. Petrovic N. miR-21 Might be Involved in Breast Cancer Promotion and Invasion Rather than in Initial Events of Breast Cancer Development. Mol Diagn Ther. 2016;20(2):97–110. Epub 2016/02/20. pmid:26891730.
- 6. Zhao Q, Chen S, Zhu Z, Yu L, Ren Y, Jiang M, et al. miR-21 promotes EGF-induced pancreatic cancer cell proliferation by targeting Spry2. Cell Death Dis. 2018;9(12):1157. Epub 2018/11/23. pmid:30464258.
- 7. Masoudi MS, Mehrabian E, Mirzaei H. MiR-21: A key player in glioblastoma pathogenesis. J Cell Biochem. 2018;119(2):1285–90. Epub 2017/07/21. pmid:28727188.
- 8. Bica-Pop C, Cojocneanu-Petric R, Magdo L, Raduly L, Gulei D, Berindan-Neagoe I. Overview upon miR-21 in lung cancer: focus on NSCLC. Cell Mol Life Sci. 2018;75(19):3539–51. Epub 2018/07/22. pmid:30030592.
- 9. Ribas J, Lupold SE. The transcriptional regulation of miR-21, its multiple transcripts, and their implication in prostate cancer. Cell Cycle. 2010;9(5):923–9. Epub 2010/02/18. pmid:20160498.
- 10. Zhang L, Zhan X, Yan D, Wang Z. Circulating MicroRNA-21 Is Involved in Lymph Node Metastasis in Cervical Cancer by Targeting RASA1. International journal of gynecological cancer: official journal of the International Gynecological Cancer Society. 2016;26(5):810–6. pmid:27101583.
- 11. Yao T, Lin Z. MiR-21 is involved in cervical squamous cell tumorigenesis and regulates CCL20. Biochimica et biophysica acta. 2012;1822(2):248–60. pmid:22001440.
- 12. Li J, Liu K, Zhang T, Yang Z, Wang R, Chen G, et al. A comprehensive investigation using meta-analysis and bioinformatics on miR-34a-5p expression and its potential role in head and neck squamous cell carcinoma. American journal of translational research. 2018;10(8):2246–63. pmid:30210668.
- 13. Dweep H, Gretz N. miRWalk2.0: a comprehensive atlas of microRNA-target interactions. Nat Methods. 2015;12(8):697. Epub 2015/08/01. pmid:26226356.
- 14. Tang Z, Kang B, Li C, Chen T, Zhang Z. GEPIA2: an enhanced web server for large-scale expression profiling and interactive analysis. Nucleic acids research. 2019;47(W1):W556–W60. pmid:31114875.
- 15. Ashburner M, Ball CA, Blake JA, Botstein D, Butler H, Cherry JM, et al. Gene ontology: tool for the unification of biology. The Gene Ontology Consortium. Nat Genet. 2000;25(1):25–9. Epub 2000/05/10. pmid:10802651.
- 16. Kanehisa M, Goto S, Sato Y, Furumichi M, Tanabe M. KEGG for integration and interpretation of large-scale molecular data sets. Nucleic acids research. 2012;40(Database issue):D109–D14. pmid:22080510.
- 17. Szklarczyk D, Franceschini A, Wyder S, Forslund K, Heller D, Huerta-Cepas J, et al. STRING v10: protein-protein interaction networks, integrated over the tree of life. Nucleic Acids Res. 2015;43(Database issue):D447–52. Epub 2014/10/30. pmid:25352553.
- 18. Chin C-H, Chen S-H, Wu H-H, Ho C-W, Ko M-T, Lin C-Y. cytoHubba: identifying hub objects and sub-networks from complex interactome. BMC systems biology. 2014;8 Suppl 4:S11. pmid:25521941.
- 19. Vasaikar SV, Straub P, Wang J, Zhang B. LinkedOmics: analyzing multi-omics data within and across 32 cancer types. Nucleic acids research. 2018;46(D1):D956–D63. pmid:29136207.
- 20. Tan H, Kim P, Sun P, Zhou X. miRactDB characterizes miRNA-gene relation switch between normal and cancer tissues across pan-cancer. Brief Bioinform. 2021;22(3). pmid:32436932.
- 21. Safran M, Dalah I, Alexander J, Rosen N, Iny Stein T, Shmoish M, et al. GeneCards Version 3: the human gene integrator. Database (Oxford). 2010;2010:baq020. pmid:20689021.
- 22. Gao D, Zhang Y, Zhu M, Liu S, Wang X. miRNA Expression Profiles of HPV-Infected Patients with Cervical Cancer in the Uyghur Population in China. PLoS One. 2016;11(10):e0164701. Epub 2016/10/21. pmid:27764149.
- 23. Wilting SM, Snijders PJ, Verlaat W, Jaspers A, van de Wiel MA, van Wieringen WN, et al. Altered microRNA expression associated with chromosomal changes contributes to cervical carcinogenesis. Oncogene. 2013;32(1):106–16. Epub 2012/02/15. pmid:22330141.
- 24. Pereira PM, Marques JP, Soares AR, Carreto L, Santos MA. MicroRNA expression variability in human cervical tissues. PLoS One. 2010;5(7):e11780. Epub 2010/07/30. pmid:20668671.
- 25. Masoudi MS, Mehrabian E, Mirzaei H. MiR-21: A key player in glioblastoma pathogenesis. Journal of cellular biochemistry. 2018;119(2):1285–90. pmid:28727188.
- 26. Irimie-Aghiorghiesei AI, Pop-Bica C, Pintea S, Braicu C, Cojocneanu R, Zimța A-A, et al. Prognostic Value of MiR-21: An Updated Meta-Analysis in Head and Neck Squamous Cell Carcinoma (HNSCC). Journal of clinical medicine. 2019;8(12). pmid:31766478.
- 27. Li P-F, Chen S-C, Xia T, Jiang X-M, Shao Y-F, Xiao B-X, et al. Non-coding RNAs and gastric cancer. World journal of gastroenterology. 2014;20(18):5411–9. pmid:24833871.
- 28. Nedaeinia R, Avan A, Ahmadian M, Nia SN, Ranjbar M, Sharifi M, et al. Current Status and Perspectives Regarding LNA-Anti-miR Oligonucleotides and microRNA miR-21 Inhibitors as a Potential Therapeutic Option in Treatment of Colorectal Cancer. Journal of cellular biochemistry. 2017;118(12):4129–40. pmid:28401648.
- 29. Pardini B, De Maria D, Francavilla A, Di Gaetano C, Ronco G, Naccarati A. MicroRNAs as markers of progression in cervical cancer: a systematic review. BMC cancer. 2018;18(1):696. pmid:29945565.
- 30. Shan L, Ji Q, Cheng G, Xia J, Liu D, Wu C, et al. Diagnostic value of circulating miR-21 for colorectal cancer: a meta-analysis. Cancer biomarkers: section A of Disease markers. 2015;15(1):47–56. pmid:25524942.
- 31. Jinling W, Sijing S, Jie Z, Guinian W. Prognostic value of circulating microRNA-21 for breast cancer: a systematic review and meta-analysis. Artificial cells, nanomedicine, and biotechnology. 2017;45(6):1–6. pmid:27684463.
- 32. Moridikia A, Mirzaei H, Sahebkar A, Salimian J. MicroRNAs: Potential candidates for diagnosis and treatment of colorectal cancer. Journal of cellular physiology. 2018;233(2):901–13. pmid:28092102.
- 33. Xu J, Zhu W, Chen L, Liu L. MicroRNA‑433 inhibits cell growth and induces apoptosis in human cervical cancer through PI3K/AKT signaling by targeting FAK. Oncology reports. 2018;40(6):3469–78. pmid:30272334.
- 34. Haddadi N, Lin Y, Travis G, Simpson AM, Nassif NT, McGowan EM. PTEN/PTENP1: ’Regulating the regulator of RTK-dependent PI3K/Akt signalling’, new targets for cancer therapy. Mol Cancer. 2018;17(1):37. Epub 2018/02/20. pmid:29455665.
- 35. Bahrami A, Hasanzadeh M, Hassanian SM, ShahidSales S, Ghayour-Mobarhan M, Ferns GA, et al. The Potential Value of the PI3K/Akt/mTOR Signaling Pathway for Assessing Prognosis in Cervical Cancer and as a Target for Therapy. J Cell Biochem. 2017;118(12):4163–9. pmid:28475243.
- 36. Albadari N, Deng S, Li W. The transcriptional factors HIF-1 and HIF-2 and their novel inhibitors in cancer therapy. Expert Opin Drug Discov. 2019;14(7):667–82. pmid:31070059.
- 37. Elorza A, Soro-Arnáiz I, Meléndez-Rodríguez F, Rodríguez-Vaello V, Marsboom G, de Cárcer G, et al. HIF2α acts as an mTORC1 activator through the amino acid carrier SLC7A5. Molecular cell. 2012;48(5):681–91. pmid:23103253.
- 38. Bai J, Chen W-B, Zhang X-Y, Kang X-N, Jin L-J, Zhang H, et al. HIF-2α regulates CD44 to promote cancer stem cell activation in triple-negative breast cancer PI3K/AKT/mTOR signaling. World journal of stem cells. 2020;12(1):87–99. pmid:32110277.