The comparison of the expression levels of ZFPM2-AS1 and ZFPM2 genes in HCC and non-tumor tissues was performed using GEPIA version 2.0 (37), during which TCGA (https://portal.gdc.cancer.gov) HCC samples were compared with GETx (https://www.gtexportal.org/home) samples, which were used as controls. The associations of expression levels of ZFPM2-AS1 and the ZFPM2 gene with the prognosis of HCC and other digestive system tumors were evaluated using the Kaplan Meier plotter (http://kmplot.com/analysis), which presents overall survival, disease free survival, relapse free and progression free survival (38), and GEPIA.

The present study was approved by the Ethics Committee of the Army Military Medical University (Chongqing, China) and written informed consent was provided by all participants prior to the study start. A total of 53 HCC and paired adjacent normal tissues (>2 cm from tumor tissues) 45 men and 8 women; age range, 30–74 years; median age, 53 years) were collected from the Department of Hepatobiliary Surgery (Chongqing, China) between November 2017 and May 2019, following surgical resection. All diagnoses were blindly confirmed by at least two pathologists at The First Affiliated Hospital of Army Military Medical University, and patients who received radiofrequency ablation, chemoradiotherapy or other treatments prior to surgery were excluded from the present study. Samples were subsequently stored at −80°C, prior to subsequent experimentation.

Total RNA was extracted from HCC tissues using TRIzol^® reagent (Invitrogen; Thermo Fisher Scientific, Inc.). cDNA was synthesized using the PrimeScript RT reagent kit with gDNA Eraser (Takara Bio, Inc.), and qPCR was performed using TB Green Premix Ex Taq II (Takara Bio, Inc.). The following primer sequences were used for qPCR: ZFPM2-AS1 forward, 5′-GCTTCTATGCCTTCCTTCCCTT-3′, and reverse, 5′-CTCCATACTCTCCCTGGGTT-3′; ZFPM2 forward, 5′-GCTACCCTCCCGTCATTT-3′, and reverse, 5′-TTAGCCATCTGCTGCCAT-3′; and β-actin forward, 5′-CCACGAAACTACCTTCAACTCC-3; and reverse, 5′-GTGATCTCCTTCTGCATCCTGT-3′. The following thermocycling conditions were used for qPCR: Initial denaturation at 95°C for 30 sec; 40 cycles of denaturation at 95°C for 5 sec, annealing and elongation at 60°C for 30 sec; and a final extension at 72°C for 30 sec. Relative ZFPM2-AS1 and ZFPM2 mRNA levels were measured using the 2^−ΔΔCq method (39) and normalized to the internal reference gene β-actin.

To investigate the biological functions and pathways of ZFPM2-AS1 and the ZFPM2 gene, gene-gene and protein-protein interaction (PPI) network analysis of the ZFPM2 gene was conducted using the GeneMANIA (http://genemania.org) and Search Tool for the Retrieval of Interacting Genes/Proteins (STRING) database version 11.0 (40). Genes associated with ZFPM2 and ZFPM2-AS1 were initially identified using the COXPRESdb database (version 7.3; http://coxpresdb.jp). Subsequently, Kyoto Encyclopedia of Genes and Genomes (KEGG) pathway and Gene Ontology (GO) analyses of ZFPM2-AS1 and ZFPM2-associated genes were performed using Database for Annotation, Visualization and Integrated Discovery (DAVID) version 6.8 (david.ncifcrf.gov/home.jsp).

To investigate the underlying mechanisms relevant to mutation status of ZFPM2-AS1 and ZFPM2 gene, the cBioPortal database (cbioportal.org/) was utilized. Kaplan-Meier survival estimates for overall survival of HCC patients, with or without mutations of the ZFPM2 gene was also analyzed, using the log-rank test.

The validation of the expression levels of the ZFPM2 gene in HCC tissues and adjacent normal tissues was further conducted with the GEO dataset GSE14520 (41). The receiver operating curve (ROC) with the area under the curve (AUC) value for assessing the predictive accuracy and discriminative ability of ROC was drawn to identify the diagnostic significance of expression level of the ZFPM2 gene.

SPSS version 22.0 (IMB Corp.) and GraphPad Prism version 7.0 (GraphPad Software, Inc.) were used for statistical analyses. P<0.05 was considered to indicate a statistically significant difference. All results are presented as mean ± standard deviation (unless otherwise shown). One-way ANOVA tests were used to evaluate the differences in ZFPM2-AS1 and ZFPM2 expression in clinical stages of HCC, while Wilcoxon's test was used for paired continuous variables. The χ² test was used to evaluate differences in categorical variables. All expression data were log transformed for differential analysis.

First the associations between the expression levels of ZFPM2-AS1 and the ZFPM2 gene and the clinical characteristics of HCC in TCGA and GTEx samples were analyzed. Table I presents the clinical characteristics of the patients from TCGA and GTEx databases (only sex available for GTEx), including sex, age at diagnosis, Child-Pugh score (42), creatinine value, HCC risk factor, family cancer history (data for 326 samples available), neoplasm histological grade (43) (data for 372 samples available) and metastasis status. The expression levels of lncRNA ZFPM2-AS1 were higher in HCC tissues compared with normal liver tissues (Fig. 1A), whereas the expression levels of the ZFPM2 gene were significantly lower in HCC tissues compared with normal liver tissues (Fig. 1C). No significant difference between the clinical stages of HCC and ZFPM2-AS1 (Fig. 1B; P=0.136) and ZFPM2 (Fig. 1D; P=0.935) expression levels were observed. For the survival of patients with HCC it was observed that higher expression levels of ZFPM2-AS1 were significantly associated with a less favorable prognosis (Fig. 2A), whereas higher expression levels of the ZFPM2 gene were significantly associated with better prognosis of HCC (Fig. 3). These bioinformatic results were also verified using clinical samples. The expression levels of lncRNA ZFPM2-AS1 were significantly higher in HCC tissues compared with adjacent normal tissues (Fig. 4B; P<0.001), whereas the expression levels of the ZFPM2 gene were significantly lower in HCC tissues compared with adjacent normal tissues (Fig. 4A; P<0.001).

According to the results obtained from COXPRESdb, lncRNA ZFPM2-AS1 was associated with the ZFPM2 gene. Thus, gene-gene and PPI network analysis of the ZFPM2 gene were conducted using the GeneMANIA and STRING tools and it was demonstrated that ZFPM2 primarily associated with GATA factors, including GATA1, GATA3 and GATA4 (Figs. 5 and 6).

Using the cBioPortal database, 9% (93/1,052) of samples were identified as harboring a mutated ZFPM2 gene. From Kaplan-Meier survival analysis, the overall survival rate demonstrated statistical differences, which means patients with HCC with ZFPM2 mutations had a less favorable prognosis compared with those without ZFPM2 mutations (P=0.0331; Fig. 7).

The top 200 ZFPM2 and ZFPM2-AS1 associated genes are presented in Table SI, which were identified using the COXPRESdb database. KEGG pathway and GO term analysis of the ZFPM2 associated genes were performed using DAVID. The GO term results demonstrated that these genes may be involved in the ‘integral component of plasma membrane’, ‘protein binding’ and ‘plasma membrane’ (Fig. 8).

As shown in Fig. 9, the expression levels of the ZFPM2 gene in HCC and non-tumor tissues were consistent in the GSE14520 dataset, in stages I and II. Consistent with TCGA data, ZFPM2 gene expression were significantly decreased in HCC tissues compared with the non-tumor tissues in both stage I and II (P<0.001; Fig. 9A and C). The ROC analysis of the ZFPM2 gene demonstrated a high accuracy of ZFPM2 in distinguishing between HCC tissues and non-tumor tissues (AUCs, >0.8; Fig. 9B and D)