Development and validation of lncRNAs–based Nomogram for prognosis prediction of hepatocellular carcinoma

doi:10.21203/rs.3.rs-1618673/v1

Download PDF

Research article

Development and validation of lncRNAs–based Nomogram for prognosis prediction of hepatocellular carcinoma

https://doi.org/10.21203/rs.3.rs-1618673/v1

This work is licensed under a CC BY 4.0 License

Version 1

posted

You are reading this latest preprint version

BACKGROUND

Hepatocellular carcinoma (HCC) is a common cancer with a poor prognosis. It is critical to develop effective tools for prognosis prediction of HCC and better understand its underlying molecular mechanism. Hence, the aim of this study is to construct a lncRNAs–based nomogram for prognosis prediction of HCC.

RESULTS

The expression data of long noncoding RNA (lncRNA) and messenger RNA (mRNA) of HCC patients were obtained from TCGA and GEO datasets. Differentially expressed lncRNAs between tumor and normal tissues were conducted in these datasets. A lncRNAs–based nomogram for prognosis prediction was then constructed. CeRNA network and Gene Set Enrichment Analysis (GSEA) were also performed to identify significantly enriched genes which share a common biological function or regulation based on our lncRNAs signature. A total of 415 patients were enrolled in this study, including 377 patients from TCGA dataset, 38 patients from GEO dataset (24 patients from GSE115019 and 14 patients from GSE101728). A two lncRNAs signature and lncRNAs–based nomogram was identified in discovery set (HR: 0.76, 95%CI: 0.63–0.91; P < 0.01;C-index = 0.71) and validated in test set (HR: 0.54, 95%CI: 0.30–0.96; P = 0.03;C-index = 0.73). In addition, ceRNA network and lncRNAs associated with pathways, such as metabolic pathways, chemical carcinogenesis, drug metabolism and carbon metabolism were further identified.

CONCLUSIONS

In summary, we constructed a lncRNAs–based nomogram that was a reliable prognostic tool for HCC patients. Additionally, our ceRNA network and GSEA results identified some potential mechanisms which were valuable research issue in future studies on the mechanism of HCC progression.

lncRNA

hepatocellular carcinoma

nomogram

Hepatocellular carcinoma (HCC) is a common cancer with a poor prognosis in the world.[1] The incidence and mortality of HCC rank the 7th and 3rd of all cancers, respectively.[2] Despite improvements in HCC diagnosis and treatment, lots of patients with HCC are diagnosed at an advanced stage with high rates of recurrence and metastasis. Hence, it is critical to develop effective tools for prognosis prediction of HCC and better understand its underlying molecular mechanism.

Long non-coding RNAs (LncRNAs) are ≥ 200 nucleotides and have no coding function.[3, 4] However, a growing number of studies have found that lncRNAs play major roles in various biologic processes such as tumorigenesis, cell migration, proliferation and tumor progression.[5, 6] Previous studies have shown that lncRNAs are dysregulated in HCC and act as proto-oncogene or anti-oncogene. For example, lncRNA UFC1 promotes the progression of HCC by enhancing β-catenin expression and stabilizing protein HuR.[7] LncRNAGAS5 suppresses metastasis by targeting miR-182 and ANGPTL1 in HCC. In addition, lncRNAs also had potential roles of promising biomarkers.[8]

Up to now, few studies established a lncRNAs–based Nomogram for prognosis prediction of HCC. Nomogram offers the advantage of visualizing important parameters and convenient for clinical application.[9–11] However, current nomograms most focused on clinical data not lncRNAs. Additionally, few studies constructed competing endogenous RNA (ceRNA) network in HCC based on differentially expressed lncRNAs. Therefore, the aim of our study was to develop a nomogram based on the lncRNAs signature which could predict the prognosis of HCC. Additionally, the analysis of the function and signal pathway of the included lncRNAs may reflect the underlying molecular mechanism of HCC.

Development of lncRNAs signature and lncRNAs–based nomogram

A total of 415 patients were enrolled in this study, including 377 patients from TCGA dataset, 38 patients from GEO dataset (24 patients from GSE115019 and 14 patients from GSE101728). The baseline clinical information of these patients was shown in Table S1.

Differentially expressed genes (DEGs) analysis was used to get differently expressed lncRNAs between tumor and normal tissues. There were 3046, 2613and 113differentially expressed lncRNAs in TCGA dataset (Fig. 1A), GSE101728 (Fig. 1B) dataset andGSE115019 dataset (Fig. 1C), respectively. Finally, 10 overlapped differentially expressed lncRNAs in these datasets were selected for construction of lncRNAs signature (Fig. 1D). Cox regression analysis was used to find the most significant prognostic lncRNAs (AL161668.5, DDX11-AS1) to construct a signature from the 10 lncRNAs in discovery set (Fig.2A). DDX11-AS1 was highly expressed in tumor than normal tissues. On the contrary, AL161668.5 was highly expressed in normal tissues. In addition, both DDX11-AS1 and AL161668.5 were associated with overall survival (OS) significantly (HR: 1.81, 95%CI: 1.29-2.54, P < 0.01; HR: 0.97, 95%CI: 0.94-0.99, P = 0.04, respectively). Subsequently, risk score of each patient based on the expression of lncRNAs were acquired automatically by R software (Fig. 2B). Using the median risk score, HCC patients could be divided into high or low risk groups (Fig. 2C). Low-risk patients experienced longer OS compared with those with higher risk scores (HR: 0.76, 95%CI: 0.63–0.91; P < 0.01, Fig. 2D).

A nomogram was constructed that integrated our lncRNAs signature and other clinical parameter in discovery set, such as age, gender, grade and stage (Fig. 3A).The C-index of the nomogram was 0.71. Table 1 showed that risk scores and stage were associated with OS significantly in univariate Cox analysis (HR: 1.32, 95%CI: 1.10-1.60, P < 0.01; HR: 1.69, 95%CI: 1.33-2.14, P < 0.01, respectively). Additionally, risk scores, age and stage were significantly correlated with prognosis of HCC in multivariate Cox regression model. (HR: 1.30, 95%CI: 1.07-1.59, P < 0.01; HR: 1.02, 95%CI: 1.00-1.04, P = 0.04; HR: 1.76, 95%CI: 1.37-2.26, P < 0.01, Figure S1 A, respectively).

Validation of lncRNAs–based Nomogram in test set

To confirm the prognostic value of this lncRNAs–based nomogram, we also did the Cox analysis in test set. Patients with HCC in low risk group also had longer survival time (HR: 0.54, 95%CI: 0.30–0.96; P = 0.03, Figure S1 C). Similarly, risk scores and stage were associated with OS significantly in univariate Cox analysis in test set (HR: 1.98, 95%CI: 1.20-3.27, P = 0.01; HR: 1.56, 95%CI: 1.12-2.16, P = 0.01, respectively, Table 1). In addition, risk scores and stage were also significantly correlated with prognosis of HCC in multivariate Cox regression model (HR: 1.68, 95%CI: 1.00-2.81, P = 0.05; HR: 1.53, 95%CI: 1.08-2.17, P = 0.02, Figure S1 B, respectively, Table 1). Besides, calibration plots revealed that our nomograms also did well in the test set (Fig 3B, C). The C-index of the nomogram was 0.73.

Construction of a ceRNA network and identification of associated biological signaling pathway

The mRNAs and miRNAs targeted by lncRNAs were predicted by circlncRNAnet, miRcode and TargetScan database. Only DDX11-AS1 had detailed information. There were 29 mRNAs and 5miRNAs correlated with DDX11-AS1, which were finally selected for constructing ceRNA network (Fig. 4A).

GSEA were performed to identify the lncRNA associated biological signal pathway. We found that the lncRNAs were highly associated with 11 KEGG pathways (Fig. 4B)including metabolic pathways, retinol metabolism, chemical carcinogenesis, metabolism of xenobiotics by cytochrome P450, drug metabolism, bile secretion, cell cycle, carbon metabolism, cellular senescence, microRNAs in cancer and oocyte meiosis. The mRNAs in the ceRNA network were also enriched in these pathways. For the GO functional analyses, they were mainly enriched in 51 GO terms including 12 biological process terms (Fig. 4C), 18 molecular function terms (Fig. 4D), and 21 cellular component terms (Fig. 4E). The top three biological process, molecular function and cellular component were biological regulation, metabolic process, response to stimulus; protein binding, ion binding, nucleic acid binding; membrane, nucleus, cytosol, respectively.

HCC is a highly aggressive cancer with many pathogenic factors, such as hepatitis virus, diabetes, gene mutation, toxicants, and so on.[12, 13] The underlying molecular mechanism of HCC and development of effective prediction tool are still to be solved. Previous studies found that lncRNAs played major roles in biological processes in different cancers, such as prostate cancer, breast cancer, et al.[4, 14, 15] In addition, some lncRNAs could also been used as novel biomarker or adjunct to improve the existing biomarkers in different cancers including HCC.[15, 16] Nevertheless, few studies investigated the role of lncRNAs in nomogram models. In 2019, N Shao et al. suggested that their lncRNAs–based nomogram was a reliable prognostic tool for biochemical recurrence in patients with prostate cancer.[3] Therefore, the purpose of our study was to identify a lncRNAs–based nomogram associated with OS of HCC.

AL161668.5 and DDX11-AS1 were identified and included in our lncRNA signature and lncRNAs–based nomogram. The nomogram could predict prognosis of HCC well and identify patients into low or high risk groups. AL161668.5 was a lncRNA that was not investigated before. However, a few studies found thatDDX11-AS1 played an important role in progression of cancers, including HCC, colorectal cancer and gastric cancer. Liao HT reported that DDX11-AS1 was involved in HCC proliferation, differentiation and cell cycle.[17] Li Y suggested that DDX11-AS1 repressed LATS2 expression and acted as an oncogene in HCC, which may be a potential therapeutic target for future treatment.[18] Tian JB found that DDX11-AS1 induced by YY1 bound to miR-873 prevented CLDN7 from miRNAs-mediated degradations and accelerates colorectal cancer progression.[19] In summary, it was reasonable that DDX11-AS1 was included in our nomogram. Besides, the role of AL161668.5 in HCC progression was a valuable research issue in future.

In addition, our ceRNA network provided a new framework to study complex interactions among RNAs in HCC. The identified mRNAs and miRNAs may have a profound impact on HCC and warrant further research on mechanisms of ceRNA. Besides, GSEA analysis including GO enrichment and KEGG pathway analysis suggested lncRNA-related genes were mainly involved in metabolic pathways, retinol metabolism, chemical carcinogenesis, drug metabolism, bile secretion, carbon metabolism, cellular senescence, et al. This indicates that the metabolism-related genes may play important role in tumorigenesis and the development of HCC. Nowadays, tumor metabolism is one of the most popular research focuses.[20–22] Previous studies had reported that many genes or lncRNAs could affect tumor progression by regulation of tumor metabolism. Therefore, future study may focus on the two included lncRNAs and inquire their function in HCC. Additionally, metabolism related genes or lncRNAs could be expected targets for future treatment.

Overestimating the risks of patients with HCC in clinical experience implied a higher medical expenses and disease burden with overtreatment. Hence, our nomogram may help to choose appropriate treatments and clinical management in patients with HCC through identifying patients into low or high risk group. Identification of the high risk HCC patients is the key to precise treatment and reducing the burden of overtreatment. Therefore, an effective predicting tool for prognosis of HCC may have widespread prospects in clinical application. Besides, the potential mechanism may put forward the direction for further study on HCC progression.

In summary, we constructed a lncRNAs–based nomogram including two lncRNAs (AL161668.5 and DDX11-AS1).The nomogram can divide HCC patients into low or high risk group. The nomogram could help clinicians provide reliable parameters for individual clinical management. Future RCTs could test this nomogram for prediction of OS in HCC and the efficacy and safety of adjuvant treatment. Additionally, ceRNA network and lncRNAs associated with pathways, such as metabolic pathways, chemical carcinogenesis, drug metabolism and carbon metabolism, put forward the direction for the future investigation on the mechanism of HCC progression.

Data preparation

The raw lncRNAs sequencing data of HCC samples were obtained from TCGA data portal (https://cancergenome.nih.gov/) and Gene Expression Omnibus (GEO) databases (https://www.ncbi.nlm.nih.gov/geo/). HCC datasets of GSE101728 and GSE115019 were identified and included in this study.

Differentially expressed lncRNAs between tumor and normal tissues in TCGA and GEO datasets were assessed by linear models for microarray data (LIMMA) method. The lncRNAs with adjusted P< 0.05 and log fold change ≥ 1.25 were defined as significantly expressed lncRNAs. The overlapped lncRNAs in these datasets were considered as candidate lncRNAs for further analysis. Patients with adequate clinical information from TCGA dataset were selected for discovery and test sets under a stratified-random sampling method at a ratio of 1/3. At last, the discovery set had 249 patients and the test set had 84 patients.

Development and validation of lncRNAs–based Nomogram

We used multivariate Cox regression model to select the most significant prognostic associated lncRNAs and construct a lncRNAs signature for prognosis prediction in our discovery set. Then, risk score of each patient were calculated by R software automatically. Using the median score, HCC patients could be divided into high or low risk groups. After that, a nomogram based on risk score of lncRNAs signature was formulated. Besides, gender, age, grade and stage of HCC were also included. Calibration curves were generated by plotting the observed rates against the nomogram predicted probabilities.

Construction of a competing endogenous RNA network

The interactions between included lncRNAs and miRNA or mRNA were predicted using circlncRNAnet, miRcode and TargetScan database. We then constructed a ceRNA network using lncRNA-miRNA interactions or miRNA-mRNA interactions. Cytoscape software was used to draw the network.

Statistical analysis

Statistical analyses were conducted using SPSS 19.0 and R software (version 3.4.4, www.r-project.org). The relationship between clinical features and lncRNAs expression/signature were analyzed using t test for continuous variables and c²test for categorical variables. Cox regression model was used to do the multivariable survival analysis. Survival curves of overall survival (OS) were assessed and compared using the Kaplan-Meier method and log-rank tests. Gene Set Enrichment Analysis (GSEA) was performed to identify significantly enriched genes which share a common biological function based on the lncRNAs signature. The Kyoto Encyclopedia of Genes and Genomes (KEGG) pathway enrichment and Gene Ontology (GO) analyses were also conducted when FDR < 0.05 and p-value < 0.001.

Competing interests: The authors declare no conflict of interest.

Ethics approval

No ethical approval is required.

Data Availability Statement: The results here are in whole based upon data generated by the TCGA Research Network (https://www.cancer.gov/tcga) and Gene Expression Omnibus (GEO) databases (https://www.ncbi.nlm.nih.gov/geo/).

Acknowledgements

This work was supported by“Two Hundreds” Young and Middle-aged Medical Key Talent Project affiliated to Taihu Talent Project(grant no. BJ2020027).

Author Contributions: Conceptualization, Ying-wei Zhu, Min Xie; Formal analysis, Hong Tang and Feifei Zhao; Project administration, Min Xie; Supervision, Feifei Zhao; Validation, Han Zhao; Preparing figures, Feifei Zhao; Writing original draft, Hong Tang, Han Zhao; Writing – review & editing, Ying-wei Zhu.

Sun Q, Zhang Z, Lu Y, Liu Q, Xu X, Xu J, Liu Y, Yu H, Yu D, Sun B: Loss of Xanthine Oxidoreductase Potentiates Propagation of Hepatocellular Carcinoma Stem Cells. Hepatology 2019.
Bray F, Ferlay J, Soerjomataram I, Siegel RL, Torre LA, Jemal A. Global cancer statistics 2018: GLOBOCAN estimates of incidence and mortality worldwide for 36 cancers in 185 countries. CA Cancer J Clin. 2018;68(6):394–424.
Shao N, Tang H, Qu Y, Wan F, Ye D. Development and validation of lncRNAs-based nomogram for prediction of biochemical recurrence in prostate cancer by bioinformatics analysis. J Cancer. 2019;10(13):2927–34.
Shao N, Zhu Y, Wan FN, Ye DW. Identification of seven long noncoding RNAs signature for prediction of biochemical recurrence in prostate cancer. Asian J Androl 2019.
Xie R, Wang M, Zhou W, Wang D, Yuan Y, Shi H, Wu L. Long Non-Coding RNA (LncRNA) UFC1/miR-34a Contributes to Proliferation and Migration in Breast Cancer. Med Sci Monit. 2019;25:7149–57.
Aalijahan H, Ghorbian S. Long non-coding RNAs and cervical cancer. Exp Mol Pathol. 2019;106:7–16.
Cao C, Sun J, Zhang D, Guo X, Xie L, Li X, Wu D, Liu L. The long intergenic noncoding RNA UFC1, a target of MicroRNA 34a, interacts with the mRNA stabilizing protein HuR to increase levels of beta-catenin in HCC cells. Gastroenterology. 2015;148(2):415–26. e418.
Chen F, Li Y, Li M, Wang L. Long noncoding RNA GAS5 inhibits metastasis by targeting miR-182/ANGPTL1 in hepatocellular carcinoma. Am J Cancer Res. 2019;9(1):108–21.
Borque A, Rubio-Briones J, Esteban LM, Sanz G, Dominguez-Escrig J, Ramirez-Backhaus M, Calatrava A, Solsona E. Implementing the use of nomograms by choosing threshold points in predictive models: 2012 updated Partin Tables vs a European predictive nomogram for organ-confined disease in prostate cancer. BJU Int. 2014;113(6):878–86.
Song K, Shi X, Wang H, Zou F, Lu F, Ma X, Xia X, Jiang J. Can a Nomogram Help to Predict the Overall and Cancer-specific Survival of Patients With Chondrosarcoma? Clin Orthop Relat Res. 2018;476(5):987–96.
Gandaglia G, Ploussard G, Valerio M, Mattei A, Fiori C, Fossati N, Stabile A, Beauval JB, Malavaud B, Roumiguie M, et al. A Novel Nomogram to Identify Candidates for Extended Pelvic Lymph Node Dissection Among Patients with Clinically Localized Prostate Cancer Diagnosed with Magnetic Resonance Imaging-targeted and Systematic Biopsies. Eur Urol. 2019;75(3):506–14.
Huang S, Zhu P, Sun B, Guo J, Zhou H, Shu Y, Li Q. Modulation of YrdC promotes hepatocellular carcinoma progression via MEK/ERK signaling pathway. Biomed Pharmacother. 2019;114:108859.
Shimada S, Mogushi K, Akiyama Y, Furuyama T, Watanabe S, Ogura T, Ogawa K, Ono H, Mitsunori Y, Ban D, et al. Comprehensive molecular and immunological characterization of hepatocellular carcinoma. EBioMedicine. 2019;40:457–70.
Le K, Guo H, Zhang Q, Huang X, Xu M, Huang Z, Yi P. Gene and lncRNA co-expression network analysis reveals novel ceRNA network for triple-negative breast cancer. Sci Rep. 2019;9(1):15122.
Ozgur E, Ferhatoglu F, Sen F, Saip P, Gezer U. Circulating lncRNA H19 may be a useful marker of response to neoadjuvant chemotherapy in breast cancer. Cancer Biomark 2019.
Chen QF, Huang T, Si-Tu QJ, Wu P, Shen L, Li W, Huang Z. Analysis of competing endogenous RNA network identifies a poorly differentiated cancer-specific RNA signature for hepatocellular carcinoma. J Cell Biochem 2019.
Liao HT, Huang JW, Lan T, Wang JJ, Zhu B, Yuan KF, Zeng Y: Identification of The Aberrantly Expressed LncRNAs in Hepatocellular Carcinoma: A Bioinformatics Analysis Based on RNA-sequencing. Sci Rep 2018, 8(1):5395.
Li Y, Zhuang W, Huang M, Li X. Long noncoding RNA DDX11-AS1 epigenetically represses LATS2 by interacting with EZH2 and DNMT1 in hepatocellular carcinoma. Biochem Biophys Res Commun. 2019;514(4):1051–7.
Tian JB, Cao L, Dong GL. Long noncoding RNA DDX11-AS1 induced by YY1 accelerates colorectal cancer progression through targeting miR-873/CLDN7 axis. Eur Rev Med Pharmacol Sci. 2019;23(13):5714–29.
Biancur DE, Paulo JA, Malachowska B, Quiles Del Rey M, Sousa CM, Wang X, Sohn ASW, Chu GC, Gygi SP, Harper JW, et al. Compensatory metabolic networks in pancreatic cancers upon perturbation of glutamine metabolism. Nat Commun. 2017;8:15965.
Polson ES, Kuchler VB, Abbosh C, Ross EM, Mathew RK, Beard HA, da Silva B, Holding AN, Ballereau S, Chuntharpursat-Bon E, et al: KHS101 disrupts energy metabolism in human glioblastoma cells and reduces tumor growth in mice. Sci Transl Med 2018, 10(454).
Ortmayr K, Dubuis S, Zampieri M. Metabolic profiling of cancer cells reveals genome-wide crosstalk between transcriptional regulators and metabolism. Nat Commun. 2019;10(1):1841.

Table 1 Univariatecox regression and multivariable cox regression analysis of clinicalvariables inthe discovery set andtest set.

			Discovery set			Test set
Variable		HR	(95% CI)	P value	HR	(95% CI)	P value
univariate Cox
gender
	Female		reference			reference
	Male	0.71	0.46-1.10	0.12	1.81	1.00-3.26	0.05
grade
	G1-G2		reference			reference
	G3-G4	1.15	0.88-1.52	0.31	1.07	0.72-1.60	0.74
stage
	I-II		reference			reference
	III-IV	1.69	1.33-2.14	0.00	1.56	1.12-2.16	0.01
age
	<60		reference			reference
	≥60	1.02	1.00-1.03	0.07	1.00	0.98-1.02	0.69
risk score
	Low risk		reference			reference
	High risk	1.32	1.10-1.60	0.00	1.98	1.20-3.27	0.01

multivariate Cox
gender
	Female		reference			reference
	Male	0.89	0.57-1.41	0.63	1.85	0.97-3.50	0.06
grade
	I-II		reference			reference
	III-IV	1.24	0.92-1.67	0.15	1.06	0.70-1.61	0.77
stage
	I-II		reference			reference
	III-IV	1.76	1.37-2.26	0.00	1.53	1.08-2.17	0.02
age
	<60		reference			reference
	≥60	1.02	1.00-1.04	0.04	1.01	0.99-1.03	0.47
risk score
	Low risk		reference			reference
	High risk	1.30	1.07-1.59	0.01	1.68	1.00-2.81	0.05

supplement.docx

Download PDF

Version 1

posted

You are reading this latest preprint version

Development and validation of lncRNAs–based Nomogram for prognosis prediction of hepatocellular carcinoma

Status:

Version 1

Abstract

BACKGROUND

RESULTS

CONCLUSIONS

Figures

Introduction

Results

Discussion

Conclusion

Methods

Declarations

References

Tables

Supplementary Files

Status:

Version 1