miR-9-5p Expression is Associated With Vascular Invasion and Prognosis in Hepatocellular Carcinoma

Hepatocellular carcinoma (HCC) is a common liver malignancy. Early vascular invasion (VI) has been associated with poor prognosis in HCC patients. MicroRNAs (miRNAs) play a signi�cant role in the emergence and development of many tumor types. In this study, we identi�ed miR-9-5p that could predict VI and prognosis in HCC patients based on TCGA database. Further, we explored the possible mechanism of miR-9-5p by target gene prediction, functional enrichment analysis, and protein ‐ protein interaction analysis. Univariate and multivariate analysis revealed that miR-9-5p was an independent risk factor for HCC. Finally, the nomogram based on miR-9-5p showed a good predictive value of HCC survival. In summary, miR-9-5p is associated with VI in HCC, and higher expression of miR-9-5p indicates poor prognosis in HCC.


Introduction
Hepatocellular carcinoma (HCC) is the fth most common cancer, and the third leading cause of cancerrelated deaths, worldwide [1]. The 5-year survival rate of HCC patients in Asia is less than 20% [2]. HCC is characterized by a high degree of malignancy, no obvious early symptoms, and intrahepatic and extrahepatic metastasis [3]. Currently, surgery remains the most e cient treatment in HCC patients, however, a great deal of patients have lost the opportunity of surgical treatment at the time of diagnosis and can only receive palliative care [4]. In the past few years, the global burden of liver cancer has been increasing [1,5]. Therefore, early diagnosis of HCC is particularly important.
MicroRNA (miRNA) is a non-coding RNA composed of 18-25 nucleotides. Besides, they are highly expressed in different types of body uids and are remarkably stable. They also target messenger RNAs (mRNAs) and inhibit their translation or induce degradation, thus providing a rapid and sensitive mechanism for regulating gene expression [6]. MiRNAs can also be used as a link to transmit information between cells [7]. Over the past few years, numerous studies have shown that miRNAs play a signi cant role in the occurrence and development of tumors. For example, up-regulation of miR-342-3p can inhibit HCC progression[8], down-regulation of miR-4521 inhibits gastric cancer progression [9], and miR-136-5p promotes the proliferation and metastasis of renal cell carcinoma [10]. With the recent development of liquid biopsy technology, some miRNAs have been used as biomarkers in different tumor types due to their high sensitivity and diagnostic e ciency [11]. Numerous studies have also demonstrated that various miRNAs are promising diagnostic and prognostic biomarkers for HCC. [12][13][14][15].
Vascular invasion (VI) is an important step in tumor invasion and metastasis. Studies have shown that VI is associated with poor prognosis in many tumor types [16][17][18][19]. But, there are few studies on the prediction of VI in HCC by miRNAs.
In this study, we investigated the expression and signi cance of miR-9-5p in predicting VI and prognosis in HCC patients using data from the TCGA database. Besides, we also determined the relationship between miR-9-5p expression and survival of HCC patients, as well as the possible biological function of miR-9-5p.

Subjects and clinical characteristics
The clinical data, read counts information of miRNA-Seq, and RNA-Seq of HCC patients were derived from the TCGA database (https://portal.gdc.cancer.gov/). The information on gene length was obtained from GENCODE [20]. MiRNAs annotation information was obtained from miRbase [21]. To standardize the sequencing data and reduce errors, the read counts were standardized using the following formula [22].
The TCGA-LIHC data set contained 377 cases, and after sequentially excluding cases without complete clinical information (n = 56), histological type of non-hepatocellular carcinoma (n = 10), with a previous history of liver cancer (n = 21), with overall survival (OS) < 30 days (n = 41) and lacking both mRNAs and miRNAs data (n = 8), the nal sample size was 241.

Screening of differentially expressed miRNAs (DEMs) related to VI in HCC
Patients were split into two groups according to whether or not they had VI. The DESeq2 package was used to analyze for any differences between groups, while the ggplot2 package was used to present the results in a volcano plot in R (Version 4.0.0). The cut-off value of | log 2 FC | >1 and p-value < 0.05 were determined.

Identi cation of DEMs
To verify the relationship between DEMs and survival, GraphPad Prism (Version 8.0.2) was used to draw the Kaplan-Meier associated survival curves. MiRNAs related to survival were expressed and selected for further analysis.
We identi ed the common prediction results in the three databases and considered them as potential target genes of miRNAs. A Venn diagram was used to visually represent the results(bioinformatics.psb.ugent.be/webtools/Venn/).

GO and KEGG Pathway Enrichment Analysis
Gene Ontology (GO) and Kyoto Encyclopedia of Genes and Genomes (KEGG) pathway enrichment analysis were used to determine the possible enrichment functions and important pathways of identi ed miRNAs. R package was used to analyze and visualize the results. P < 0.05 was considered statistically signi cant.

Protein-protein interactions (PPI)
Relational data on PPI was downloaded from the STRING database (string-db.org). The genes with a minimum interaction score of 0.7 (high con dence) were selected and the PPI networks were established in Cytoscape (Version 3.7.2).

Nomogram Model Construction
To evaluate the prognosis in HCC patients, the miRNAs expression data were combined with clinical features (gender, age, grade, stage, and T), and used to construct a novel nomogram for the prediction of 1-, 2-, and 3-year OS in HCC patients. The "rms" package and the "survival" package were used to build and validate the model in R. The C-index and calibration plots were used to evaluate the predictive accuracy of the nomogram.

Statistical Analysis
R software (version 4.0.0) was used to perform all statistical analyses in this study. An unpaired t-test was used to determine the differences in expression levels of miRNAs between groups. The t-test or the chi-square test was used to analyze the clinical characteristics of patients in each group.
Univariate/multivariate analysis was performed using the "survival" package and the results were visualized using the "survminer" package. The receiver operator characteristic (ROC) curve was used to assess the diagnostic accuracy of miRNAs in patient survival.

Identi cation of DEMs associated with VI in HCC
A total of 241 cases were obtained from the TCGA-LIHC data set, and the clinicopathological characteristics of the patients included in the analysis are shown in Table 1. The patients were divided into VI (+) group (n = 81) and VI (-) group (n = 160) based on the presence or absence of VI. The survival curves indicated that HCC patients with VI (+) had signi cantly worse survival compared with VI (-) (p = 0.0052, Fig. 1A) HCC patients. Further, 7 DEMs, including 5 up-regulated miRNAs (miR-9-5p, miR-9-3p, miR-1270, miR-196a-5p and miR-2114-5p) and 2 down-regulated miRNAs (miR-375-3p and miR-483-3p) were identi ed (Fig. 1B). Next, Kaplan-Meier survival analyses of the DEMs were performed to determine their prognostic signi cance. As shown in Fig. 2, the high expression levels of miR-9-5p and miR-9-3p were signi cantly correlated with poorer outcomes in HCC patients. Besides, we found that the expression level of miR-9-3p was relatively lower in HCC patients compared with the expression level miR-9-5p ( Fig. 3). Consequently, miR-9-5p was selected for further studies. 3.2. The relationship between miR-9-5p and clinical parameters.
The included samples were divided into the low expression group (n = 120) and high expression group (n = 121) according to the median expression level of miR-9-5p. Heatmap was used to visualize the expression patterns of miR-9-5p in HCC (Fig. 4A). Figure 4B shows the distribution of mir-9-5p expression, and the survival time and survival status of HCC patients were presented as a scatterplot (Fig. 4C).

Functional enrichment analysis of the target genes of miR-9-5p
A total of 245 mRNAs were identi ed from TargetScan, microT-CDS, and miRmap network databases, as the target genes of miR-9-5p (Fig. 6A). To further explore the function of miR-9-5p in HCC, GO and KEGG pathway analysis was performed. KEGG pathway analysis revealed the pathways in which the target genes are implicated. The results showed that the prolactin signaling pathway, transcriptional misregulation in cancer, and hepatocellular carcinoma were the main pathways regulated by the target genes (Fig. 6B). GO analysis revealed the function of the target genes based on the three GO terms. In the biological process (BP), the top three enriched terms were urogenital system development, hormonemediated signaling pathway, and cellular response to steroid hormone stimulus. At the level of cellular component (CC), the target genes were mainly enriched in the basement membrane, messenger ribonucleoprotein complex, and SWI/SNF complex. Lastly, in terms of molecular function (MF), DNAbinding transcription factor binding, DNA-binding transcription repressor activity, RNA polymerasespeci c, and DNA-binding transcription repressor activity were the top 3 most enriched terms (Fig. 6C).

PPI network analysis
The STRING online tool was used to construct the interaction network between miR-9-5p and the target genes (Fig. 7A). The top hub genes ranked by degree were selected for further analysis. These hub genes, including SIRT1, FOXO1, AR, SMARCD2, BCL6, SMARCE1, NCOR2, VCAN, CNTN4, and CNTN3, might play a critical role in cancer (Fig. 7B).

Construction of a nomogram model
We established and validate a nomogram model in the TCGA-LIHC cohort to predict the survival of HCC patients. This model predicted the 1-, 2-and 3-years overall survival, and age, grade, stage, T, and miR-9-5p expression variables were included in constructing the nomogram (Fig. 8A). The C-index of the nomogram was 0.697 (95%CI = 0.609-0.785). The calibration plot shows good consistency between the nomogram prediction probabilities and observed probability in predicting the 1-, 2-and 3-year survival rates in the TCGA-LIHC cohort (Fig. 8B-D).

Discussion
In 2020, there were over 906,000 new cases and 830,000 deaths of primary liver cancer, and HCC accounts for 75-85% of all primary liver cancers [1]. The global burden of HCC is high, and the prognosis of HCC is poor, which is related to early invasion and metastasis [26]. MiRNAs are members of the RNA family and play a key role in proliferation, invasion, metastasis, and immune escape of HCC [27][28][29][30]. Therefore, it is necessary to establish a prognostic model that includes the expression of miRNAs molecules related to HCC prognosis.
In previous studies, miR-9-5p has been reported to play an important role in a variety of diseases. In cervical cancer, miR-9-5p promotes tumor angiogenesis by targeting SOCS [31]. In prostate cancer, mir-9-5p regulates the expression of QKI-5, StarD, and other genes, thus affecting tumor invasion and metastasis [32,33]. However, Wang et al revealed that miR-9-5p inhibits tumor cell proliferation and promotes apoptosis by down-regulating PAK4 in colorectal cancer [34]. In HCC, several experimental studies have shown that miR-9-5p can promote the proliferation, invasion, and metastasis of tumor cells by down-regulating the expression of ESR1 [35], Klf4[36], CNNM1 [37], PPARA [38], and so on. In a previous study, high expression of miR-9-5p was associated with a poor prognosis of osteosarcoma [39]. However, patients with high expression of miR-9-5p are reported to have a better prognosis in ovarian cancer [40].
The present study focused on identifying miRNAs related to VI in HCC. Unlike other database-based studies [41][42][43], all the study subjects were HCC patients, to more accurately screen for differentially expressed miRNAs related to VI. Patients who met the inclusion criteria were divided into two groups based on whether or not they had VI, and the differences in miRNA expression levels analyzed. Seven differentially expressed miRNAs were identi ed, however, only miR-9-5p was highly expressed in HCC patients and correlated with OS. The potential targets of miR-9-5p were identi ed from three target gene prediction databases. GO and KEGG enrichment was used to predict the potential mechanism of miR-9-5p in HCC. PPI analysis was used to select the core target genes, and miR-9-5p was used in constructing a nomogram that had a signi cant prognostic value for predicting survival in HCC patients.
However, this study had some limitations. On the one hand, validation of the nomogram was not performed, due to lack of VI, survival, and miRNAs expression data using data from other databases. On the other hand, although many experimental studies have shown that miR-9-5p is related to the proliferation, invasion, and metastasis of HCC, the relationship between miR-9-5p expression and VI of HCC needs further experimental studies.
In conclusion, miR-9-5p is associated with VI in HCC and can be used as a potential diagnostic and prognostic signature for predicting the overall survival of HCC patients.  Risk score analysis of miR-9-5p in the TCGA cohort. A. The expression heatmap of miR-9-5p in HCC. B.
The miR-9-5p distribution in the TCGA-LIHC cohort. C. The survival time and survival status with increasing miR-9-5p expression.

Figure 7
Protein-protein interaction analysis. A. PPI network of the target genes. B. Identi ed hub target genes of the PPI network.