Systematic analysis of the expression profiles and prognostic values of the FAM72 family in liver cancer

Abstract

Family with sequence similarity 72 (FAM72) plays a crucial role in the functions of neural stem cells, specifically in promoting the self-renewal capabilities of neural progenitor cells. The FAM72 family members FAM72A–D have been shown to play a role in tumorigenicity. However, their expression and prognostic significance in liver cancer remain unknown. In this study, we used bioinformatics analyses to assess the expression and prognostic relevance of FAM72A–D in liver cancer. The expression levels of FAM72A–D were markedly elevated in liver cancer tissues compared with normal tissues. FAM72A–D expression correlated with advanced clinical stage, high histological grade, and unfavorable prognosis in patients with liver cancer. Single-cell RNA sequencing results suggested that FAM72A–D are predominantly present in proliferative T cells. We examined genetic alterations in FAM72A–D and found that gene mutations were linked to low overall survival rates in patients with liver cancer. We further developed a robust FAM72 gene signature to predict the prognosis of patients with liver cancer, and the gene signature was associated with immune infiltration. Collectively, these findings indicate the involvement of the FAM72 family in liver tumorigenesis and suggest its potential utility as a biomarker for adverse prognostic outcomes.

Graphical abstract

Highlights

• •High FAM72A-D expression correlates with advanced stages, poor prognosis.
• •FAM72A-D mutations are associated with reduced overall survival in liver cancer patients.
• •FAM72A-D is predominantly found in proliferative T cells.
• •FAM72 gene signature can predict prognosis, immune infiltration of liver cancer patients.

1. Introduction

Hepatocellular carcinoma (HCC) is the most common primary liver neoplasm and the fourth leading cause of cancer-related mortality globally [1]. Over the past two decades, the incidence of liver cancer has increased by 53.7 % and the mortality rate has increased by 48.0 % [2]. Clinical treatments, including surgical resection, liver transplantation, and hepatic artery chemoembolization, offer opportunities for long-term survival for patients with early-stage HCC. However, because of poor patient compliance and the limitations of ultrasound examination [3,4], most patients are diagnosed with HCC at an advanced stage, when treatment is no longer an option [5]. Unlike the steadily increasing survival rates for other malignancies, the overall survival (OS) rate for HCC remains largely stagnant, mainly because of the scarcity of effective early diagnostic strategies and therapeutic targets. Consequently, the identification of novel molecular pathways and therapeutic biomarkers is critical to improve patient outcome.

Several prognostic biomarkers for survival and predictive markers for therapeutic response for liver cancer have been identified, including exosomes [6], microRNAs [7], and proteins such as TDO2 [8], APLN [9], GULP1 [10], and KK-LC-1 [11]. However, for better understanding and classification of liver cancer and to aid therapeutic decision-making, it is essential to identify more biomarkers and evaluate their prognostic and predictive abilities.

The family with sequence similarity 72 (FAM72) protein-encoding gene is specific to neural stem cells [12,13]. High expression of FAM72 induces DNA repair and promotes cell mutations [14]. Research has shown that FAM72 is associated with the development of numerous cancers such as breast cancer, glioblastoma, and prostate cancer [12,15,16]. High expression level of FAM72 is correlated with poor prognosis in glioblastoma, multiple myeloma, and lung adenocarcinoma [[16], [17], [18]]. These results indicate that FAM72 may be a potential biomarker for the diagnosis and prognosis of cancer. However, few studies have examined the expression and clinical significance of members of the FAM72 family in liver cancer.

Here, we evaluated the expression of FAM72 family members, FAM72A, FAM72B, FAM72C, and FAM72D, in liver cancer using multiple independent datasets. We investigated the relationship between the expression of FAM72 family members and clinical pathological parameters and the prognostic value of the expression of FAM72 family members in liver cancer. We also developed a FAM72 gene signature that reliably predicted the prognosis of patients with HCC in various cohorts and examined the mechanisms associated with the FAM72 gene signature.

2. Materials and methods

2.1. Datasets

The Cancer Genome Atlas (TCGA) initiative, spearheaded by the National Cancer Institute in collaboration with the National Human Genome Research Institute, was designed to conduct a comprehensive molecular characterization of more than 20,000 primary cancer specimens and the corresponding normal tissues from 33 cancer types. The mRNA expression levels of the FAM72 family were obtained from TCGA database (https://portal.gdc.cancer.gov). The external validation cohort included data from the International Cancer Genome Consortium (ICGC) (https://dcc.icgc.org/) and the datasets GSE54236, GSE144269, and GSE116174 from the Gene Expression Omnibus (GEO) database (https://www.ncbi.nlm.nih.gov/geo/).

2.2. Pan-cancer analysis of mRNA expression of the FAM72 family

We analyzed the mRNA expression of the FAM72 family in normal and tumor tissues across multiple cancer types in TCGA pan-cancer cohort. We also assessed the prognostic values of the mRNA expression of the FAM72 family in pan-cancer using the GEPIA database (http://gepia.cancer-pku.cn/). Univariate Cox regression analysis was used to examine the association of gene expression with OS in different tumor types.

2.3. Investigation of FAM72 family expression in liver cancer and the association with clinicopathological features

We compared the expression of the FAM72 family in liver cancer tissues and normal tissues using the Wilcoxon rank-sum test. We also examined FAM72 family expression in various subgroups stratified by age, gender, tumor stage, and degree of differentiation. The results were visualized using ggplot2. The survminer package's survival_cutpoint function was used to determine the best cut-off value for FAM72 family expression in TCGA cohort. We analyzed the difference in survival between high-risk and low-risk liver cancer patient groups using the log-rank test. We conducted univariate and multivariate Cox regression analyses to assess the prognostic value of FAM72 family members using TCGA cohort.

2.4. Single cell analysis

Single-cell analysis was conducted using GSE140228 and GSE166635 to investigate the distribution of FAM72 family members in cell subsets in liver cancer at the individual cell level. We examined the expression patterns of FAM72 family members in liver cancer cohorts using annotations sourced from the scCancer website (https://bianlab.cn/scCancerExplorer/explore/).

2.5. Correlation between mRNA expression of FAM72 family members and copy number variation (CNV) and methylation

We obtained CNV and methylation datasets related to liver cancer from the cBioPortal platform (http://www.cbioportal.org/) and analyzed the expression levels of the FAM72A–D genes across various CNV classifications. We also investigated the relationship between the mRNA expression of FAM72A–D and the corresponding promoter methylation levels. We assessed the prognostic implications of the FAM72 gene family on survival using the log-rank test.

2.6. Correlation of FAM72 family expression with immune infiltration

We analyzed TCGA cohort using the Estimate and ssGSEA algorithms available on the Xiantao website (https://www.xiantaozi.com/). These analyses focused on the relationship between FAM72 family expression and indicators of immune infiltration. A P-value below 0.05 indicated statistical significance.

2.7. Construction of the FAM72 gene signature

Using variables obtained from the multivariate Cox regression analysis, we developed the FAM72 risk score. This score was computed by summing the expression levels of the genes, each weighted by its respective coefficient. Time-dependent ROC curve analysis was used to assess the predictive capability of the FAM72 risk score. We analyzed the prognosis of patients with high- and low-risk scores using survival curves. The BEST website (https://rookieutopia.hiplot.com.cn/app_direct/BEST/) was used to further validate our model. We visually assessed the calibration curve by comparing the nomogram predictions to observed probabilities, where a 45° line indicated perfect calibration. We further analyzed clinicopathological features and the FAM72 risk score across TCGA, ICGC, GSE54236, GSE144269, and GSE116174 cohorts to determine if the FAM72 score is an independent risk factor influencing the prognosis of liver cancer patients. A decision curve analysis (DCA) was conducted to assess the clinical utility of the FAM72 risk score in relation to conventional clinicopathological factors across various patient cohorts. We assessed the FAM72 risk score by analyzing the time-dependent area under the curve (AUC) with respect to various clinicopathological features. We also compared the FAM72 risk score with 10 previously reported prognostic models for HCC (Table 1) [[19], [20], [21], [22], [23], [24], [25], [26], [27], [28]].

Table 1.

Summary of prognostic genes for hepatocellular carcinoma extracted from ten published studies.

Model	PMID	Type	ENSEMBL	SYMBOL
Ou-ATM	32309377	mRNA	ENSG00000141867	BRD4
Ou-ATM	32309377	mRNA	ENSG00000168036	CTNNB1
Ou-ATM	32309377	mRNA	ENSG00000118058	MLL
Ou-ATM	32309377	mRNA	ENSG00000164362	TERT
Ou-ATM	32309377	mRNA	ENSG00000141510	TP53
Kong-BM	33021385	mRNA	ENSG00000149658	YTHDF1
Kong-BM	33021385	mRNA	ENSG00000198492	YTHDF2
Xu-CCI	34217320	mRNA	ENSG00000170801	HTRA3
Xu-CCI	34217320	mRNA	ENSG00000130558	OLFM1
Xu-CCI	34217320	mRNA	ENSG00000198523	PLN
Lu-FIO	34235075	mRNA	ENSG00000147889	CDKN2A
Lu-FIO	34235075	mRNA	ENSG00000083307	GRHL2
Lu-FIO	34235075	mRNA	ENSG00000258818	RNASE4
Lu-FIO	34235075	mRNA	ENSG00000131747	TOP2A
Pan-CM	33818013	mRNA	ENSG00000023445	BIRC3
Pan-CM	33818013	mRNA	ENSG00000213923	CSNK1E
Pan-CM	33818013	mRNA	ENSG00000141503	MINK1
Pan-CM	33818013	mRNA	ENSG00000136997	MYC
Pan-CM	33818013	mRNA	ENSG00000186575	NF2
Fang-FIO	33392087	mRNA	ENSG00000169252	ADRB2
Fang-FIO	33392087	mRNA	ENSG00000100342	APOL1
Fang-FIO	33392087	mRNA	ENSG00000176749	CDK5R1
Fang-FIO	33392087	mRNA	ENSG00000149557	FEZ1
Fang-FIO	33392087	mRNA	ENSG00000106617	PRKAG2
Fang-FIO	33392087	mRNA	ENSG00000189298	ZKSCAN3
Hong-BMC	34147074	mRNA	ENSG00000198099	ADH4
Hong-BMC	34147074	mRNA	ENSG00000165556	CDX2
Hong-BMC	34147074	mRNA	ENSG00000137869	CYP19A1
Hong-BMC	34147074	mRNA	ENSG00000069482	GAL
Hong-BMC	34147074	mRNA	ENSG00000184502	GAST
Hong-BMC	34147074	mRNA	ENSG00000213398	LCAT
Hong-BMC	34147074	mRNA	ENSG00000198670	LPA
Hong-BMC	34147074	mRNA	ENSG00000005421	PON1
Hong-BMC	34147074	mRNA	ENSG00000114113	RBP2
Hong-BMC	34147074	mRNA	ENSG00000157005	SST
Hong-BMC	34147074	mRNA	ENSG00000242366	UGT1A8
Zhao-FMB	33553243	mRNA	ENSG00000138798	EGF
Zhao-FMB	33553243	mRNA	ENSG00000126934	MAP2K2
Zhao-FMB	33553243	mRNA	ENSG00000171444	MCC
Zhao-FMB	33553243	mRNA	ENSG00000213281	NRAS
Yan-Ag	33049716	mRNA	ENSG00000119535	CSF3R
Yan-Ag	33049716	mRNA	ENSG00000169248	CXCL11
Yan-Ag	33049716	mRNA	ENSG00000158869	FCER1G
Yan-Ag	33049716	mRNA	ENSG00000115607	IL18RAP
Yan-Ag	33049716	mRNA	ENSG00000147168	IL2RG
Yan-Ag	33049716	mRNA	ENSG00000171522	PTGER4
Zhang-OL	32788937	mRNA	ENSG00000113249	HAVCR1
Zhang-OL	32788937	mRNA	ENSG00000134323	MYCN
Zhang-OL	32788937	mRNA	ENSG00000175793	SFN

2.8. Analysis of immune infiltration with respect to the FAM72 risk score

The MCPcounter package was used to assess immune cell infiltration levels using markers of 10 immune cell types in TCGA cohort. Spearman correlation analysis was used to evaluate the correlation between the FAM72 risk score and the immune infiltration levels estimated by MCPcounter. We also conducted a Spearman correlation analysis using the TIMER database (https://cistrome.shinyapps.io/timer/) to assess the association between the FAM72 gene signature and immune cell infiltration levels. Specifically, we analyzed the relative abundance of six distinct types of tumor-infiltrating immune cells and investigated variations in immune checkpoint expression across different FAM72 risk categories. The analysis included the following immune cell types: CD4 T cells, CD8 T cells, B cells, neutrophils, macrophages, and dendritic cells; the immune checkpoints included PDCD1, CD274, CTLA4, HAVCR2, ICOS, and TIGIT.

2.9. Gene Set Enrichment Analysis (GSEA) and Gene Set Variation Analysis (GSVA)

GSEA identifies common biological pathways by focusing on gene sets with shared functions and potential regulatory mechanisms. In this study, the fast GSEA (fgsea) method was used to investigate the mechanisms underlying differential FAM72 expression and the implications for liver cancer pathogenesis and prognosis. GSVA was used to detect activated signaling pathways in the high- and low-risk groups.

2.10. Statistical analysis

Statistical analyses were performed using R software (version 4.2.1). Continuous variables were analyzed using the Wilcoxon rank-sum test or the Kruskal–Wallis test, based on suitability. Survival analyses were conducted by Kaplan–Meier curve analysis, with log-rank tests used for intergroup comparisons. Univariate and multivariate Cox proportional hazards models were used to evaluate prognostic factors, with outcomes expressed as hazard ratios (HR) and 95 % confidence intervals (CI). Correlation assessments were performed using Spearman's rank correlation coefficient. The predictive efficacy of the FAM72 risk score was assessed through time-dependent receiver operating characteristic (ROC) curve analysis. A two-sided P-value of less than 0.05 indicated statistical significance.

3. Results

3.1. Expression and prognostic value of FAM72 family in pan-cancer

Fig. 1 presents the flowchart of graphical abstract. We observed a striking difference in FAM72A–D gene expression between tumor tissues and normal controls in 33 types of cancers in TCGA (Fig. 2A–D). Fig. 2E–H shows the differences in FAM72A–D expression in tumor tissues compared with the corresponding adjacent non-tumor tissues. These results indicated that FAM72A–D shows a significant increasing trend in most tumor tissues compared with normal and adjacent tissues. The prognostic value of FAM72A–D mRNA in TCGA pan-cancer dataset is shown in Fig. 2I. The results revealed the significant prognostic relevance of FAM72A–D mRNA in various cancers including ACC, KIRC, KIRP, LGG, LIHC, LUAD, and UVM.

Fig. 1.

the flowchart of the research.

Fig. 2.

Expression and prognostic value of the FAM72 family in pan-cancer patients.

(A–D) Comparison of expression levels of FAM72 family members between normal and tumor tissues.

(E–H) Analysis of differential expression in tumor tissues relative to paired adjacent normal tissues.

(I) Assessment of the correlation between FAM72 expression and overall survival across pan-cancer cases.

Notes: P values below 0.05 are highlighted with red and blue frames. High-risk subjects are denoted by red blocks, while low-risk individuals are indicated by blue blocks.

3.2. FAM72 family expression in liver cancer and the association with clinicopathological features

The expression level of FAM72A–D was significantly increased in liver cancer tissues compared with normal tissues (P < 0.001) (Fig. 3A). The increased expression of FAM72A–D in tumor samples suggests that the FAM72 family may be involved in the occurrence and development of liver cancer. Analysis of the diagnostic accuracy of FAM72A–D revealed AUC values of 0.906 (CI: 0.872−0.940), 0.879 (CI: 0.844−0.914), 0.913 (CI: 0.884–0.942) and 0.938 (CI: 0.915–0.961), respectively (Fig. 3B).

Fig. 3.

the FAM72 Gene Family Expression in liver cancer and Its Association with Clinicopathological Features

(A) Differential expression analysis of FAM72 family members in liver cancer tumor tissues as opposed to normal tissues; (B) Evaluating the predictive accuracy of the FAM72 family members in differentiating normal from tumor tissues; (C) Patient characteristics and expression of FAM72A-D; (D) The prognostic value of FAM72 family in TCGA liver cancer cohort; (E) Univariate and multivariate analyses were conducted to identify prognostic factors associated with the FAM72 gene family.

We analyzed the association of FAM72A–D gene expression in patients from TCGA cohort with age, histological grade, gender and TNM stage and found that high expression levels of FAM72A, FAM72B, FAM72C, and FAM72D were associated with advanced TNM stage and tumor grade (Fig. 3C). Survival analysis of patients with liver cancer in TCGA database indicated that increased expression levels of FAM72A, FAM72B, FAM72C, and FAM72D were associated with unfavorable outcomes (Fig. 3D). Univariate and multivariate Cox regression analyses of data from TCGA HCC cohort identified FAM72A and FAM72D as independent factors for liver cancer (Fig. 3E).

3.3. Distribution of FAM72A–D in different cell subpopulations

Single-cell analysis results indicated that in the GSE140228 cohort, FAM72A is present in mono/macrophages and dendritic cells, while FAM72B–D is mainly found in proliferating T cells. In the GSE166635 cohort, FAM72A is present in proliferating T cells and endothelial cells, whereas FAM72B–D is predominantly found in proliferating T cells (Supplementary Fig. 1).

3.4. Analysis of genetic alterations

Patients with liver cancer exhibited a high frequency of FAM72A–D gene alterations, as shown in Fig. 4A and B. The FAM72A–D mutation frequencies were 9 %, 15 %, 10 %, and 17 %, respectively. Significant correlations were identified among the expression levels of the FAM72A–D gene family (Fig. 4C). Additionally, genetic alterations in FAM72A–D were associated with a poor prognosis in patients with HCC (Fig. 4D). These findings indicate that high expression of the FAM72 gene family is related to poor OS in liver cancer.

Fig. 4.

Genetic Alterations analysis of FAM72 family

(A-B) Genetic alteration in FAM72A-D in liver cancer; (C) Correlation analysis among FAM72A-D members; (D) The connection between OS of HCC patients and Genetic alteration in FAM72A-D; (E) Correlation between FAM72 A, B, and D expression and CNV; (F) Correlation between FAM72 A, B, and D expression and methylation levels.

mRNA expression levels are influenced by CNV and methylation. Analysis using the cBioPortal platform revealed notable variations in the mRNA expression levels of FAM72A, FAM72B, and FAM72D across various CNV groups (Fig. 4E). We also analyzed the promoter methylation levels of the FAM72A, FAM72B, and FAM72D genes in relation to their expression. Promoter methylation showed a minimal effect on the mRNA expression of FAM72A, FAM72B, and FAM72D (Fig. 4F). In contrast, CNV significantly affected FAM72 expression levels. FAM72C was not examined as the corresponding data were absent in the database.

3.5. Correlation of FAM72A–D expression with immune infiltration

Analysis of TCGA cohort using the Estimate algorithm identified a significant link between FAM72A–D expression and the stromal score (Fig. 5A–D). Analysis using the ssGSEA algorithm identified associations between FAM72A–D expression with various immune cells. Th2 cells, T helper cells, and aCD cells were found to be positively correlated with the expression of the FAM72 gene family (Fig. 5E–H). The expression of the FAM72 gene family was strongly negatively correlated with Th17 cells, dendritic cells, neutrophils, CD8 T cells, and cytotoxic cells in other subgroups.

Fig. 5.

Correlation between FAM72 A-D expression and immune infiltration.

3.6. Construction of the FAM72 gene signature

Time-dependent ROC curve analysis indicated that the predictive accuracy of the FAM72 risk score varied across cohorts: 0.630 to 0.744 for TCGA, 0.715 to 0.732 for ICGC, 0.655 to 0.702 for GSE54236, 0.620 to 0.674 for GSE144269, and 0.621 to 0.670 for GSE116174 (Fig. 6A–E). In TCGA cohort, patients in the high-risk group had a poor prognosis (P < 0.001); similar results were obtained in ICGC (P < 0.001), GSE54236 (P < 0.001), GSE144269 (P = 0.005), and GSE116174 (P = 0.015) (Fig. 6F–J). FAM72A and FAM72D exhibited significant survival difference across TCGA, ICGC, GSE54236, GSE144269, and GSE116174 cohorts in BEST website (Supplementary Fig. 2). The calibration curve results demonstrated that the FAM72 risk score exhibited moderate reliability in forecasting survival rates at the 1-year, 2-year, and 3-year time points (Fig. 6K–O).

Fig. 6.

The prognostic values of FAM72 gene signature risk score

(A-E) Area under the Receiver Operating Characteristic (ROC) curve for the FAM72 risk score predicting 1, 2, and 3-year OS (F–J) The prognostic value of FAM72 risk score in liver cancer patients (K–O) Calibration curves for predicting OS of liver cancer patients utilizing the FAM72 risk score at 1, 2, and 3 years.

The prognostic significance of the FAM72 risk score was assessed using univariate and multivariate Cox regression analyses. Cox regression analysis results revealed that the FAM72 risk score was an independent prognostic factor across multiple cohorts, including TCGA (P < 0.001), ICGC (P < 0.001), GSE54236 (P = 0.001), GSE144269 (P = 0.006), and GSE116174 (P = 0.018) (Fig. 7A–E). The time-dependent AUC curve results indicated that compared with other clinical pathological features, the FAM72 risk score exhibited a higher predictive value in TCGA, ICGC, GSE54236, GSE144269, and GSE116174 cohorts (Fig. 7F–J). The results of DCA curve analysis yielded similar results (Supplementary Fig. 3).

Fig. 7.

The independent prognostic role of FAM72 risk score in the liver cancer cohort

(A-E) Univariate and multivariate regression analysis based on FAM72 risk score after incorporating clinical pathological features (F–J) Time dependent AUC curve analysis of the predictive value of FAM72 risk score and clinical pathological features (K) Comparison of the C-index for the FAM72 risk score and 10 previously published models across multiple cohorts.

Compared with the models from 10 prior studies, the FAM72 risk score exhibited the highest concordance index (C-index) across TCGA, GSE54236, GSE144269, GSE116174, and meta-cohorts. Additionally, it secured the third position within the ICGC cohort (Table 2). Therefore, the FAM72 gene signature exhibits a notable degree of robustness (Fig. 7K).

Table 2.

Comparison of the C-index of the FAM72 score and ten published hepatocellular carcinoma prognostic gene signatures, across different HCC cohorts.

Model	TCGA	ICGC	GSE54236	GSE144269	GSE116174	MetaCohort	Gene number
FAM72score	0.660	0.664	0.639	0.616	0.611	0.616	2
Fang-FIO	0.578	0.592	0.581	0.496	0.531	0.541	6
Hong-BMC	0.534	0.518	0.449	0.530	0.495	0.485	11
Kong-BM	0.647	0.526	0.597	0.556	0.549	0.550	2
Lu-FIO	0.575	0.667	0.552	0.598	0.525	0.562	4
Ou-ATM	0.598	0.490	0.507	0.545	0.507	0.496	5
Pan-CM	0.582	0.631	0.617	0.520	0.556	0.583	5
Xu-CCI	0.595	0.591	0.490	0.612	0.578	0.563	3
Yan-Ag	0.611	0.527	0.539	0.585	0.544	0.527	6
Zhang-OL	0.601	0.492	0.527	0.520	0.531	0.506	3
Zhao-FMB	0.573	0.708	0.633	0.537	0.555	0.594	4

3.7. Correlations between the FAM72 risk score and immune infiltration

The findings derived from the MCP algorithm indicated that the FAM72 risk score displayed significant positive associations with several immune cell types, including CD8T (r = 0.219, P < 0.001), fibroblasts (r = 0.131, P = 0.010), cytotoxic lymphocytes (r = 0.147, P = 0.005), B lineage cells (r = 0.209, P < 0.001), natural killer cells (r = 0.157, P = 0.003), monocytic lineage cells (r = 0.420, P < 0.001), T cells (r = 0.380, P < 0.001), and myeloid dendritic cells (r = 0.282, P < 0.001) (Fig. 8A). The correlation analysis results derived from the TIMER database revealed significant associations between the FAM72 risk score and several immune cell types. Significant correlations were found with CD4⁺ T cells (r = 0.142, P = 0.006), B cells (r = 0.375, P < 0.001), CD8⁺ T cells (r = 0.161, P = 0.002), macrophages (r = 0.316, P < 0.001), neutrophils (r = 0.322, P < 0.001), and dendritic cells (r = 0.390, P < 0.001) (Fig. 8B). Additionally, the expression levels of immune checkpoints, including PDCD1, CD274, CTLA4, HAVCR2, ICOS, and TIGIT, exhibited significant differences between the high-risk and low-risk groups (Fig. 8C).

Fig. 8.

Correlation of immune infiltration and FAM72 risk score.

3.8. FAM72-related signaling pathways obtained by GSEA and GSVA

GSEA results revealed that the top five pathways with significant enrichment in the high FAM72 risk score group were G2M checkpoint, E2F targets, mitotic spindle, MYC targets, and DNA repair (Fig. 9A). Analysis using GSVA revealed that the top five significantly enriched pathways in the high-risk group were the cell cycle, mismatch repair, DNA replication, base excision repair, and homologous recombination; these results are illustrated in a volcano map and heatmap (Fig. 9B and C).

Fig. 9.

FAM72-Related Signaling Pathways Identified via Gene Set Enrichment Analysis (GSEA) and Gene Set Variation Analysis (GSVA)

(A) Pathway Enrichment Analysis of FAM72A gene signature; (B) GSVA analysis of high and low FAM72A risk score groups; (C) Clustering heatmap of GSVA scores between high and low FAM72 risk score groups.

4. Discussion

Liver cancer is the third leading cause of cancer-related mortality globally. HCC accounts for approximately 90 % of all primary liver cancers [29]. Most patients are diagnosed with liver cancer at an advanced stage, with a median survival rate of less than 2 years [30]. The most common clinical treatments for advanced HCC are surgical resection, transplantation, ablation, and trans-arterial chemoembolization; however, these treatments have a limited ability to improve survival rates [[31], [32], [33]]. While several biomarkers for HCC have been studied, most have not been widely accepted in clinical practice because of their association with poor prognosis in both early and advanced HCC, highlighting the need for additional biomarkers for early detection [34].

FAM72 is a novel self-renewal protein in neuronal progenitor cells and is expressed at a low level in other tissues under normal physiological conditions [13]. The FAM72 gene family contains four homologous genes (FAM72A–D). Through TCGA pan-cancer database analysis, we found that FAM72A–D expressions were significantly up-regulated in liver cancer tissues. Previous studies showed that FAM72 promotes cancer cell proliferation in glioblastoma and multiple myeloma [16,17]. FAM72A–D is significantly overexpressed in lung adenocarcinoma, and its expression level is related to immune infiltration. FAM72 is a potential biomarker for predicting the effect of immunotherapy in lung adenocarcinoma and a prognostic monitoring indicator for lung adenocarcinoma patients receiving immunotherapy [18]. The expression of FAM72 is increased in renal clear cell carcinoma and influenced by CpG island hypomethylation; the expression level of FAM72 is negatively correlated with patient prognosis [35]. Moreover, FAM72A may control cell growth by altering the metabolism of cellular reactive oxygen species, especially in EBV-induced tumors [36]. Together, this evidence suggests that FAM72 is closely related to tumor development.

In our study, we found that FAM72 family members (FAM72A–D) are potential diagnostic and prognostic biomarkers and are correlated with advanced TNM stages and tumor grades of liver cancer. Furthermore, patients with liver cancer harboring a high mutation frequency in FAM72 genes had a poorer prognosis. We constructed a robust FAM72 gene signature. Compared with previously published models [[19], [20], [21], [22], [23], [24], [25], [26], [27], [28]], the FAM72 gene signature demonstrated a higher mean C-index while using fewer genes. Our model thus achieved improved predictive accuracy with enhanced simplicity and potential clinical applicability. Validation using multi-center data supported its robust generalizability, providing a reliable foundation for broader implementation.

We performed a comprehensive literature review aimed at exploring the fundamental mechanisms by which the FAM72 family influences cancer. UNG is a key enzyme in normal cells that is essential for DNA repair in the BER pathway. Elevated FAM72A levels antagonize UNG2 activity, leading to UNG2 degradation through a proteasome-dependent mechanism, and promote error-prone DNA repair pathways, mutation, and neoplasia [37]. Therefore, the increased expression of FAM72 that leads to degradation of UNG2 and subsequent gene mutation affecting tumor development and prognosis may be one of the pathways by which FAM72 mediates the development of liver cancer. Inhibiting FAM72 expression may be a strategy to suppress tumors and improve the prognosis of patients. Studies showed that FAM72 homologs may trigger centrosome and mitotic spindle formation through mitotic cell cycle genes KIF23, SGO1, ASPM, KIF14, CEP55, CENPE and BUB1 to promote tumorigenesis [16]. Our GSEA results suggest that the FAM72 gene is associated with the cell cycle and mitotic spindle. FAM72 has been demonstrated to be a target for tumor therapy in various studies [12,[15], [16], [17], [18],38]. Together, these findings indicate that the FAM72 family may be a potential target for liver cancer.

Our analysis also demonstrated that the FAM72 family may be a promising target for immunotherapy for tumor treatment. Tumor infiltration levels of T cells are linked to the effectiveness of immunotherapy, making these cells a key focus of current cancer research [39]. Studies showed that the expression levels of FAM72A–D are negatively correlated with Th17 cells, dendritic cells, neutrophils, cytotoxic T lymphocytes, CD8⁺ T cells, B cells, mast cells and eosinophils and positively correlated with Th2 cells and helper T cells, indicating that FAM72 is widely involved in humoral and cellular immunity. Th2 cells are an independent risk factor for HCC proliferation and progression and associated with immunosuppression and cancer cell migration [40]. Th17 cells improve the survival of HCC patients by secreting IL-17 [41]. Our results suggest that high FAM72A expression may lead to high Th2 and low Th17 and a Th2/Th17 cytokine spectrum imbalance in tumors, resulting in the induction of tumor development. Immune checkpoints such as PDCD1, CD274, CTLA4, HAVCR2, ICOS, and TIGIT were also found to be positively correlated with FAM72A–D expression. A previous study showed that FAM72A–D expression level correlated with levels of eosinophils, NK cells, monocytes, and mast cells [18]. One study showed that patients with PD-L1 overexpression have better clinical outcomes after anti-PD-1 therapy compared with patients with low expression [42]. These results suggest that the expression level of FAM72 is associated with prognosis in patients with liver cancer and may indicate the level of immune infiltration, providing a reference for the application of immunotherapy in patients with liver cancer.

We further investigated the underlying cause of the aberrant expression of the FAM72 family in HCC. Methylome profiling of HCC tumors has revealed significant distinctions between liver tumor tissue and non-tumor liver tissue. Numerous studies have reported hypermethylation of tumor suppressor genes in liver cancer [43,44]. Prior research has indicated a correlation between elevated FAM72 expression and promoter hypomethylation in cancer patients, which is linked to enhanced clinical outcomes [16,17]. Notably, our results indicate that promoter methylation had a minimal impact on the mRNA expression of FAM72 A/B/D. However, CNV was found to influence the mRNA expression levels of the FAM72 family. Therefore, we speculate that the expression levels of the FAM72 gene family are mainly influenced by CNV.

This study has several limitations. First, the prognostic value of FAM72A needs to be further verified in real-world clinical trials. The database used in this study mainly included mRNA data, so it was not possible to fully elucidate the function of FAM72A in liver cancer. Moreover, the oncogenic mechanism of FAM72A needs to be further explored in vitro and in vivo. Further research and experimental validation are required to evaluate the potential efficacy of interfering with FAM72 as a therapeutic strategy for liver cancer.

5. Conclusion

This study systematically analyzed the role of FAM72A–D in HCC. Our results suggest the potential of FAM72A–D as a therapeutic and prognostic biomarker in HCC.

Author contributions

Weihao Kong, Long Teng, and Kangjie Zhang contributed to the original draft writing, methodology development, investigation, formal analysis, data curation, and conceptualization of the study.

Yajun Zou was responsible for visualization, methodology, formal analysis, data curation, and conceptualization.

Xingyu Wang and Jianlin Zhang participated in writing reviews and editing, original draft writing, visualization, supervision, formal analysis, data curation, and conceptualization.

Ethics statement

Not applicable.

Funding

This research was supported by The University Natural Science Research Project of Anhui Province (2023AH053322) and the Basic and Clinical Collaboration Enhancement Program Foundation of Anhui Medical University (2023xkjT038).

Declaration of competing interest

The author declares that there are no potential conflicts of interest.

Appendix A. Supplementary data

The following are the Supplementary data to this article.

figs1.

figs2.

figs3.

Data availability

Data will be made available on request.