Prognostic value of long non-coding RNA MALAT1 in hepatocellular carcinoma: A study based on multi-omics analysis and RT-PCR validation

Abstract

Background: This study aimed to explore the relationship between MALAT1 and the prognosis of patients with hepatocellular carcinoma (HCC). Methods: We constructed a MALAT1 protein-protein interaction network using the STRING database and a network of competing endogenous RNAs (ceRNAs) using the StarBase database. Using data from the GEPIA2 database, we studied the association between genes in these networks and survival of patients with HCC. The potential mechanisms underlying the relationship between MALAT1 and HCC prognosis were studied using combined data from RNA sequencing, DNA methylation, and somatic mutation data from The Cancer Genome Atlas (TCGA) liver cancer cohort. Tumor tissues and 19 paired adjacent non-tumor tissues (PANTs) from HCC patients who underwent radical resection were analyzed for MALAT1 mRNA levels using real-time PCR, and associations of MALAT1 expression with clinicopathological features or prognosis of patients were analyzed using log-rank test and Gehan-Breslow-Wilcoxon test. Results: Five interacting proteins and five target genes of MALAT1 in the ceRNA network significantly correlated with poor survival of patients with HCC (p < 0.05). High MALAT1 expression was associated with mutations in two genes leading to poor prognosis and may upregulate some prognostic risk genes through methylation. MALAT1 was significantly co-expressed with various signatures of genes involved in HCC progression, including the cell cycle, DNA damage repair, mismatch repair, homologous recombination, molecular cancer m6A, exosome, ferroptosis, infiltration of lymphocyte (p < 0.05). The expression of MALAT1 was markedly upregulated in HCC tissues compared with PANTs. In Kaplan-Meier analysis, patients with high MALAT1 expression had significantly shorter progression-free survival (PFS) (p = 0.033) and overall survival (OS) (p = 0.023) than those with low MALAT1 expression. Median PFS was 19.2 months for patients with high MALAT1 expression and 52.8 months for patients with low expression, while the corresponding median OS was 40.5 and 78.3 months. In subgroup analysis of patients with vascular invasion, cirrhosis, and HBsAg positive or AFP positive, MALAT1 overexpression was significantly associated with shorter PFS and OS. Models for predicting PFS and OS constructed based on MALAT1 expression and clinicopathological features had moderate predictive power, with areas under the receiver operating characteristic curves of 0.661-0.731. Additionally, MALAT1 expression level was significantly associated with liver cirrhosis, vascular invasion, and tumor capsular infiltration (p < 0.05 for all). Conclusion: MALAT1 is overexpressed in HCC, and higher expression is associated with worse prognosis. MALAT1 mRNA level may serve as a prognostic marker for patients with HCC after hepatectomy.

Keywords: DNA methylation; competing endogenous RNAs (ceRNAs); hepatocellular carcinoma; interacting proteins; metastasis associated lung adenocarcinoma transcript 1 (MALAT1); multi-omics; mutation; prognosis.

FIGURE 1

Analysis of MALAT1-interacting proteins. (A) The protein–protein interaction network of MALAT1 was analyzed using the STRING tool. (B) Chord plot displaying the relationship between MALAT1-interacting proteins and Kyoto Encyclopedia of Genes and Genomes (KEGG) terms. (C) The outer circle shows a scatter plot for each term of the log (fold change) of the assigned genes. Orange circles indicate upregulation. The z-score is a crude measure of significance of enrichment.

FIGURE 2

Kaplan-Meier survival curves for patients showing high or low expression of genes encoding MALAT1-interacting proteins.

FIGURE 3

MALAT1-associated network of competing endogenous RNAs (ceRNAs). Blue rectangles represent microRNAs (miRNAs) that can bind to MALAT1; yellow triangles represent target genes that compete with MALAT1 for binding to miRNAs.

FIGURE 4

Kaplan-Meier survival curves for patients showing high or low expression of MALAT1 target genes.

FIGURE 5

Differentially mutated genes. (A) Forest plot of Fisher testing to identify differentially mutated genes. (B) Proportion of each type of mutation and the overall mutation percentage of differentially mutated genes. The left half of the graph shows samples expressing high MALAT1 and the right half shows samples expressing low MALAT1. (C) Red represents mutated genes enriched in samples expressing high MALAT1. Blue represents mutated genes enriched in samples expressing low MALAT1.

FIGURE 6

Kaplan-Meier survival curves comparing patients carrying mutant or wild-type versions of TP53, IRX1, and LRP1B.

FIGURE 7

The upper plot is a heatmap of the top principal components correlating with MALAT1, tumor stage and tumor grade. “MALAT1” represents continuous variables, while “MALAT1 group (median)” and “MALAT1 group (quantile)” represent categorical variables grouped by median and quartiles. The color represents the degree of significance. The lower plot shows how much of the observed methylation variation was explained by the principal components.

FIGURE 8

Analysis of differentially methylated positions (DMPs). (A) Volcano plot of DMPs. Green dots represent the positions that are hypomethylated. Red dots represent the positions that are hypermethylated. (B) Clustering heatmap of DMPs. Column annotations provide information about samples expressing high or low MALAT1. Row annotations provide information about DMPs. (C) Forest plot showing genes upregulated by MALAT1-associated methylation that were significant in Cox univariate analysis. (D): Bar plot of pathways with areas under curves (AUCs) > 0.75 in the Gene Set Enrichment Analysis (GSEA).

FIGURE 9

Boxplot of the results of Immuno-Oncology-Biological-Research (IOBR) analysis comparing samples expressing high or low MALAT1 in terms of (A) tumor-associated signatures and (B) cells within the tumor microenvironment. *p < 0.05, **p < 0.01, *** < 0.001, ****p < 0.0001.

FIGURE 10

Prognosis analysis based on the TCGA LIHC cohort. (A) Overall survival and (B) Disease-free survival. Survival curves of patients with MALAT1 high expression or low expression were compared using the log-rank test. (C) A multivariate Cox regression model for disease-free survival containing MALAT1 and baseline clinical information. *p < 0.05; **p < 0.01; ***p < 0.001.

FIGURE 11

Kaplan-Meier survival curves of HCC patients after radical resection, stratified by low or high MALAT1 expression. (A) MALAT1 expression in 19 paired HCC and adjacent non-cancerous tissues from patients who underwent curative resection. (B,C) Progression-free survival (PFS) and overall survival (OS) of all patients. (D,E) PFS and OS of 59 patients with vascular invasion. Curves were compared using the log-rank test.

FIGURE 12

Multivariate analysis of 179 HCC patients to identify independent risk factors of progression-free survival and overall survival. The dots and bars represent the HR and 95% CI, respectively. CI, confidence interval; HR, hazard ratio; MALAT1, metastasis-associated lung adenocarcinoma transcript 1; OS, overall survival; PFS, progression-free survival.

FIGURE 13

Progression and survival prediction models. (A): Nomogram of the progression prediction model. (B): Nomogram of the survival prediction model. (C): Calibration curve for the progression prediction model. (D): Calibration curve for the survival prediction model.

FIGURE 14

Evaluation of model performance using time-dependent receiver operating characteristic curves. (A) Predictive power of the progression prediction model. (B) Predictive power of the survival prediction model. (C) Comparisons of areas under curves (AUCs) for progression prediction model, other prognostic parameters and MALAT1 alone. (D) Comparisons of areas under curves (AUCs) for survival prediction model, other prognostic parameters and MALAT1 alone. (E): Decision curves for the progression prediction model, other prognostic parameters and MALAT1 alone. (F): Decision curves for the survival prediction model, other prognostic parameters and MALAT1 alone.

FIGURE 15

Predictive value of MALAT1 for diagnosing clinicopathological characteristics, based on the area under the time-dependent receiver operating characteristic curve (AUC). (A) AUC was 0.631 for cirrhosis (95% CI: 55.6%–70.2%, p = 0.006). (B) AUC was 0.676 for vascular invasion (95% CI: 60.2%–74.4%, p < 0.001). (C) AUC was 0.605 for tumor capsular infiltration (95% CI: 52.9%–67.7%, p = 0.039). (D) AUC was 0.593 for AFP positive (95% CI: 51.8%–66.6%, p = 0.037).