Potentially functional variants of CYP2A6 and ABHD12 in the arachidonic acid metabolism pathway genes predict the survival of HBV-related hepatocellular carcinoma patients

Abstract

Arachidonic acid (AA) metabolism plays essential roles in inflammation, tissue regeneration, immune regulation, and tumorigenesis. However, the prognostic significance of genetic variants in AA metabolism genes for hepatitis B virus (HBV)-related hepatocellular carcinoma (HCC) remains unclear. We analyzed 56 AA metabolism-related genes in 866 HBV-HCC patients using Cox proportional hazards regression, with Bayesian false discovery probability and false positive report probability for multiple testing correction. Two independent SNPs, CYP2A6 rs28399433 A > C (HR = 1.30, 95% CI = 1.07–1.58, P = 0.008) and ABHD12 rs3827014 C > T (HR = 0.69, 95% CI = 0.52–0.93, P = 0.013), were identified. A genetic score combining protective alleles showed a dose-dependent association with improved overall survival (P_trend<0.001), with significant multiplicative interactions with smoking (P_interaction=0.016) and alcohol consumption (P_interaction=0.049), and additive interactions for smoking (RERI = 0.65, 95% CI = 0.10–1.19) and BCLC stage B/C (RERI = 1.25, 95% CI = 0.17–2.33). Functional validation with luciferase reporter assays demonstrated allele-specific effects on gene expression (P < 0.001). Lower CYP2A6 and higher ABHD12 levels in HCC tissues compared to normal were observed in the UALCAN database (P = 6.43 × 10⁻⁷ for CYP2A6; P < 1.00 × 10⁻¹² for ABHD12) and in 103 paired tumor/normal tissues. Survival analysis using the KMplot database indicated that decreased CYP2A6 and increased ABHD12 expression were associated with poorer survival (P < 0.001 and P = 0.008, respectively). These findings suggest that functional variants in CYP2A6 and ABHD12 may serve as novel survival biomarkers for HBV-HCC by regulating gene expression.

Supplementary Information

The online version contains supplementary material available at 10.1186/s12885-026-16212-x.

Introduction

Primary liver cancer (PLC) represents a major global health burden, ranking the third most common cause of cancer-related deaths worldwide [1]. The disease exhibits significant geographical disparities, with China accounting for approximately 316,500 PLC deaths in 2022 [2]. Hepatocellular carcinoma (HCC), the predominant histological subtype, comprises nearly 80% of PLC [3]. In China, chronic hepatitis B virus (HBV) infection is the primary etiological driver, contributing to 84.4% of HCC cases [4]. Despite therapeutic advances, the survival rate for HCC remains poor, with a five-year survival rate of only 12% [5]. HCC survival is influenced by multiple clinical factors, including tumor stage, histological subtype, differentiation grade, invasive growth pattern, and pathological classification of microvascular invasion [6]. Notably, marked interpatient heterogeneity in clinical outcomes persists even among individuals with similar disease stages [7], suggesting that host genetic factors may play an important role in HCC survival.

Arachidonic acid (AA) metabolism contributes to cancer development and progression by regulating inflammatory and immune responses [8]. This metabolic pathway is primarily governed by four key enzymatic cascades: (1) the cyclooxygenase (COX) pathway, (2) the lipoxygenase (LOX) pathway, (3) the cytochrome P450 (CYP450) pathway, and (4) the anandamide pathway [9]. These pathways generate bioactive lipid mediators that orchestrate diverse oncogenic processes across multiple malignancies, including colorectal cancer [10], cutaneous malignancies [11], and HCC [12]. In HCC, COX- and LOX-derived metabolites (e.g., prostaglandins and leukotrienes) promote tumor proliferation through activation of the extracellular signal-regulated kinase (ERK) signaling cascade [13]. Similarly, CYP450-mediated epoxyeicosatrienoic acids (EETs) contribute to HCC progression by enhancing tumor growth and metastatic potential [14]. In China, HBV-related HCC accounts for up to 84.4% of all HCC cases [4]. Notably, recent studies have also demonstrated that HBV infection itself may dysregulate the AA metabolic pathway [15].

Single-nucleotide polymorphisms (SNPs), one of the most common and stable forms of genetic variation [16, 17], can affect gene expression and have emerged as promising non-invasive biomarkers for predicting cancer susceptibility and survival, including HCC [18–20]. Increasing evidence from candidate gene studies has demonstrated that specific SNPs are associated with HCC survival outcomes [21–23], highlighting their potential as survival indicators. To date, the role of genetic variants in the AA metabolism pathway in HCC survival is unknown. Therefore, we conducted a comprehensive analysis of AA metabolism-related genetic variants in a retrospective cohort of patients with HBV-related HCC and explored their clinical utility as prognostic biomarkers.

Materials and methods

Study populations

This retrospective cohort included 866 patients diagnosed with HBV-related HCC confirmed by histopathology, who received curative hepatic resection at Guangxi Medical University Cancer Hospital between July 2007 and December 2017 [23, 24]. Five milliliters of peripheral blood were collected prior to surgery for genomic DNA extraction. The specific inclusion and exclusion criteria have been described in detail in previous studies [24]. Comprehensive baseline information was recorded through questionnaires and hospital record reviews, including patient age, sex, smoking, alcohol consumption, serum alpha fetoprotein (AFP) level, presence of cirrhosis, vascular tumor embolus, and Barcelona Clinic Liver Cancer (BCLC) stage. Overall survival (OS) was defined as the duration between the date of surgery and either death or the most recent follow-up. Postoperative follow-up was conducted via telephone at three-month intervals during the first two years and every six months thereafter until March 31, 2020. The research protocol received approval from the Institutional Review Board of Guangxi Medical University Cancer Hospital (KY2025629), and written informed consent was voluntarily signed by each participant before enrollment.

Genotyping, gene selection and SNPs

Genotyping was performed with the Illumina Infinium^® Global Screening Assay (Shanghai, China) [25]. Genes involved in AA metabolism were compiled from the MSigDB database (https://www.gsea-msigdb.org) using the query term “arachidonic acid metabolism”. After removing two X chromosome linked genes, FAAH2 and AWAT1, a set of 56 candidate genes was retained for analysis (Supplementary Table S1). All SNPs located within these genes and within 2 kilobases upstream or downstream were retrieved from the 1000 Genomes Project. The quality control pipeline implemented stringent thresholds, including a genotype call rate exceeding 95%, a minor allele frequency (MAF) greater than 5%, and adherence to Hardy-Weinberg equilibrium (HWE) with a p-value cutoff of 10 − 6. After QC filtering, 5,067 SNPs passed QC and were used in downstream analyses.

Expression Quantitative Trait Loci (eQTL) analysis and functional prediction

Normal liver (n = 261) and whole blood (n = 800) samples were obtained from the GTEx database (https://www.gtexportal.org) [26]. These samples were used to perform the eQTL analysis to identify potentially functional SNPs. Functional annotation of candidate SNPs and variants in strong linkage disequilibrium within the same loci was conducted using SNPinfo (https://manticore.niehs.nih.gov/snpinfo/snpfunc.html) [27] and HaploReg v4.2 (https://pubs.broadinstitute.org/mammals/haploreg/haploreg.php) [28]. We first identified SNPs with P < 0.001 in the eQTL analysis and then retained those annotated as transcription factor binding sites (TFBS) in SNPinfo. Tagging SNPs were recognized using LD analysis (r² ≥ 0.8) via the HaploView 4.2 software [29]. Additionally, somatic alteration frequencies of the corresponding genes were examined through cBioPortal for Cancer Genomics to investigate other mechanisms that may influence gene expression in tumor tissue (http://www.cbioportal.org) [30].

Differential gene expression analysis

The transcriptomic profiles of HCC and matched normal liver tissues were initially assessed via the UALCAN database (https://ualcan.path.uab.edu/) [31]. The subsequent verification of gene expression levels was performed using RNA sequencing data from 103 paired tumor and normal samples, which were obtained from patients who underwent hepatic resection at Guangxi Medical University Cancer Hospital. Furthermore, the correlation between the expression of target genes and the OS of HCC patients was evaluated using the UALCAN and KMplot online platforms (http://kmplot.com/analysis/) [32] .

Dual-luciferase reporter assay

To experimentally assess the influence of the identified variants on gene transcriptional activity, dual-luciferase reporter assays were conducted. Genomic DNA fragments of approximately 500 base pairs encompassing CYP2A6 rs28399433 and ABHD12 rs3827014 (Supplementary Table S2) were cloned into the pGL3-basic vector (Promega) to generate three constructs per SNP: empty vector control, major allele, and minor allele variants. Human embryonic kidney 293T cells (CCTCC NO GDC0187) were maintained in Dulbecco’s modified Eagle medium supplemented with fetal bovine serum and penicillin–streptomycin to final concentrations of 10% and 1%, respectively, at 37 °C in a humidified incubator containing 5% CO₂. We seeded cells into 96-well plates at a density of 1 × 10⁴ cells per well and performed transfection 24 h later with 100 ng of each plasmid using Effectene Transfection Reagent (QIAGEN). A Renilla luciferase plasmid (pRL-TK, 10 ng) was co-transfected for normalization. Subsequently, we measured the luciferase activity 24 h after transfection using the Dual-Luciferase Reporter Assay System (Promega, Wisconsin, USA). The authenticity of HEK-293T cells was verified by short tandem repeat (STR) profiling, and cultures were routinely confirmed to be mycoplasma-free. Each experimental condition was performed in triplicate to ensure reproducibility.

Statistical analysis

Associations between genetic variants and OS were assessed with multivariable Cox proportional hazards models that included relevant clinical and demographic covariates. To reduce false positive discoveries arising from multiple comparisons, we implemented both Bayesian false discovery probability (BFDP) and false positive report probability (FPRP) methods, applying cutoffs of 0.80 and 0.20, respectively [33]. Independent prognostic SNPs were determined via stepwise multivariable Cox regression analysis. The cumulative effects of significant variants were further evaluated by cumulative allele analyses and visualized through Kaplan-Meier (KM) survival curves. We performed stratification analysis to evaluate potential effect modification by patient characteristics and tested multiplicative interactions. The additive interaction was evaluated by estimating the relative excess risk attributed to interaction (RERI), which reflects the combined effect of two factors beyond their individual contributions [34]. Model discrimination was appraised using receiver operating characteristic (ROC) and area under the curve (AUC) analyses. Regional association plots were visualized with LocusZoom (http://locuszoom.sph.umich.edu) [35]. All statistical procedures were implemented in R software versions 4.1.3 using the packages survival, survminer, gap, and GenABLE. Two-sided P values below 0.05 were considered statistically significant.

Result

Characteristics of the study populations

This investigation encompassed a cohort of 866 Chinese patients with HBV-related HCC treated with hepatectomy, who were followed for a median of 39.05 months. The population was chiefly composed of males (87.4%, n = 760) with a median diagnosis age of 47 years. Multivariable Cox regression revealed four independent prognostic factors for OS: age, AFP levels, cancer embolus, and BCLC stage. Advanced BCLC stages B/C carried a nearly twofold higher risk of death than stages 0/A (HR = 1.98, 95% CI = 1.56–2.52, P < 0.001). Elevated AFP (HR = 1.29, 95% CI = 1.05–1.57, P = 0.015) and cancer embolus (HR = 1.74, 95% CI = 1.38–2.52, P < 0.001) were also associated with poorer OS (Supplementary Table S3). Interestingly, older age at diagnosis was associated with a reduced risk of death (HR = 0.81, 95% CI = 0.66–0.99, P = 0.036), potentially reflecting a survivor effect.

The associations between candidate SNPs in AA pathway genes and HBV-HCC survival

As depicted in the workflow (Fig. 1), we began by selecting 56 genes involved in the AA metabolism pathway as candidates (Supplementary Table S1), after QC, 5,067 SNPs were retained for subsequent analyses. Subsequent single-locus analysis identified 224 SNPs that were significantly associated with OS in HBV-HCC patients (P < 0.05, BFDP < 0.80, FPRP < 0.20). To prioritize functionally relevant variants, we performed eQTL analysis using GTEx data, which retained 124 SNPs (P < 0.001). Functional annotation using SNPinfo identified 13 potential SNPs located within TFBS, comprising one in CYP2A6 and 12 in ABHD12 (Table 1). As CYP2A6 harbored only one SNP (rs28399433) associated with survival, it was directly selected for further analysis. For ABHD12, linkage disequilibrium (LD) analysis (Haploview; r²< 0.8) identified rs3827014 as the representative variant (Supplementary Figure S1). Multivariable stepwise Cox regression confirmed two SNPs (rs28399433 A > C: HR = 1.29, 95% CI = 1.06–1.57, P = 0.010; rs3827014 C > T: HR = 0.69, 95% CI = 0.52–0.93, P = 0.013) as independent prognostic markers (Table 2). The results of the selected SNPs are summarized in the regional association plot (Supplementary Figure S2).

Fig. 1.

Flowchart of the present study design

Table 1.

Associations of 13 significant SNPs with overall survival of HBV-HCC patients

SNP	Chr	Gene	Allelea	MAF	HR (95% CI)	P b	BFDP	FPRP
rs28399433	19	CYP2A6	A > C	0.139	1.30 (1.07–1.58)	0.008	0.570	0.075
rs6076339	20	ABHD12	T > C	0.139	0.69 (0.52–0.93)	0.013	0.650	0.185
rs2482940	20	ABHD12	C > T	0.074	0.70 (0.54–0.92)	0.010	0.591	0.130
rs4815411	20	ABHD12	A > G	0.087	0.69 (0.52–0.93)	0.013	0.650	0.185
rs4815412	20	ABHD12	G > A	0.074	0.69 (0.52–0.93)	0.013	0.650	0.185
rs2027003	20	ABHD12	A > G	0.074	0.70 (0.54–0.92)	0.010	0.591	0.130
rs2027004	20	ABHD12	C > T	0.087	0.70 (0.54–0.92)	0.010	0.591	0.130
rs6107031	20	ABHD12	C > T	0.087	0.69 (0.52–0.93)	0.013	0.650	0.185
rs6115159	20	ABHD12	T > C	0.074	0.69 (0.52–0.93)	0.013	0.650	0.185
rs6050566	20	ABHD12	G > A	0.074	0.69 (0.52–0.93)	0.013	0.650	0.185
rs4813562	20	ABHD12	G > A	0.074	0.69 (0.52–0.93)	0.013	0.650	0.185
rs3827014	20	ABHD12	C > T	0.074	0.69 (0.52–0.93)	0.013	0.650	0.185
rs6115168	20	ABHD12	C > A	0.074	0.69 (0.52–0.93)	0.013	0.650	0.185

Abbreviations: SNPs single nucleotide polymorphisms, HBV hepatitis B virus, HCC hepatocellular carcinoma, OS overall survival, FPRP False-positive report probability, HR hazards ratio, 95% CI 95% confidence interval, BFDP Bayesian false-discovery probability, MAF Minor Allele Frequency

^a Referring allele/effect allele

^b Multivariate Cox proportional hazards regression analysis was adjusted for age, sex, smoking status, drinking status, AFP level, cirrhosis, embolus, and BCLC stage

Table 2.

Two independent SNPs in Cox proportional hazards regression analysis with adjustment for other covariates in 866 HBV-HCC patients

Characteristics	Category	No. of patients	HR (95% CI)	P a
Age (year)	≤ 47	434	1.00
	> 47	432	0.82 (0.67–0.99)	0.043
Sex	Female	106	1.00
	Male	760	1.27 (0.92–1.75)	0.154
AFP level (ng/ml)	≤ 400	522	1.00
	> 400	344	1.29 (1.06–1.59)	0.013
Embolus	No	636	1.00
	Yes	230	1.74 (1.37–2.20)	< 0.001
BCLC stage	0/A	427	1.00
	B/C	439	2.03 (1.60–2.59)	< 0.001
CYP2A6 rs28399433 A > C	AA/AC/CC	640/211/15	1.29 (1.06–1.57)	0.010
ABHD12 rs3827014 C > T	CC/CT/TT	742/119/5	0.69 (0.52–0.93)	0.013

Abbreviations: OS overall survival, HR hazards ratio, 95% CI 95% confidence interval, AFP alpha-fetoprotein, BCLC Barcelona Clinic Liver Cancer

^aStepwise Cox regression analysis was adjusted for age, sex, smoking status, drinking status, AFP level, cirrhosis, embolus, and BCLC stage

As shown in Table 3, we identified the CYP2A6 rs28399433 C allele as a risk factor for poorer survival and the ABHD12 rs3827014 T allele as a protective factor conferring better survival in HBV-HCC patients (P_trend = 0.008 and 0.013, respectively). Under the dominant genetic model, patients harboring the CYP2A6 rs28399433 AC/CC genotypes exhibited significantly worse survival outcomes compared to those with the AA reference genotype (HR = 1.36, 95% CI = 1.10–1.68, P = 0.005). Conversely, individuals carrying the ABHD12 rs3827014 CT/TT genotypes had a 31% decrease in mortality risk (HR = 0.69, 95% CI = 0.51–0.94, P = 0.018). These associations were further corroborated by KM survival curves (Supplementary Figure S3A-D). Notably, the homozygous variant genotypes (CC and TT) were not individually statistically significant, likely due to their low frequencies (n = 15 and n = 5, respectively), resulting in limited statistical power.

Table 3.

Associations between two identified SNPs and overall survival of HBV-HCC patients

Genotype	No. of patients	Death (%)	Univariate analysis		Multivariate analysis
			HR (95% CI)	P	HR (95% CI)	P a
CYP2A6 rs28399433 A > C
AA	640	300 (46.9)	1.00		1.00
AC	211	113 (51.1)	1.26 (1.02–1.57)	0.035	1.36 (1.10–1.70)	0.005
CC	15	6 (40.0)	0.84 (0.38–1.89)	0.681	1.26 (0.56–2.86)	0.574
Trend				0.117		0.008
AA	640	300 (46.9)	1.00		1.00
AC + CC	226	119 (52.7)	1.23 (0.99–1.52)	0.055	1.36 (1.10–1.68)	0.005
ABHD12 rs3827014 C > T
CC	742	371 (46.0)	1.00		1.00
CT	119	47 (39.5)	0.71 (0.52–0.97)	0.025	0.71(0.52–0.96)	0.029
TT	5	1 (20.0)	0.37 (0.05–2.63)	0.320	0.31 (0.04–2.23)	0.246
Trend				0.014		0.013
CC	742	371 (46.0)	1.00		1.00
CT + TT	124	48 (38.7)	0.69 (0.51–0.94)	0.017	0.69 (0.51–0.94)	0.018
NPAb
0–1	200	105 (52.5)	1.00		1.00
2	569	281 (49.4)	0.84 (0.67–1.05)	0.118	0.76 (0.61–0.95)	0.018
3–4	97	33 (34.0)	0.54 (0.37–0.80)	0.002	0.48 (0.33–0.71)	< 0.001
Trend test				< 0.001		< 0.001
0–1	200	105 (52.5)	1.00		1.00
2–4	666	314 (47.1)	0.79 (0.63–0.97)	0.037	0.71 (0.57–0.89)	0.003

Abbreviations: SNPs single nucleotide polymorphisms, OS overall survival, HBV hepatitis B virus, HCC hepatocellular carcinoma, HR hazards ratio, 95% CI 95% confidence interval, NPA number of protective alleles

^a Multivariate Cox proportional hazards regression analysis was adjusted for age, sex, smoking status, drinking status, AFP level, cirrhosis, embolus, and BCLC stage

^b Protective alleles were CYP2A6 rs28399433 _A and ABHD12 rs3827014 _T

Combined and stratified analyses of independent SNPs associated with HBV-HCC survival and their interactions with risk factors and genetic risk score

To assess the cumulative prognostic impact of the identified SNPs, we integrated the protective alleles (CYP2A6 rs28399433 A and ABHD12 rs3827014 T) into a composite genetic risk score, defined as the number of protective alleles (NPAs). Multivariable-adjusted analysis revealed a significant inverse association between increasing NPA count and mortality risk (P_trend< 0.001). When stratifying patients by NPA count (0–1 vs. 2–4 alleles), those carrying ≥ 2 protective alleles demonstrated a 29% decreased in mortality risk (HR = 0.71, 95% CI = 0.57–0.89, P = 0.003), with clear survival differentiation evident in Kaplan-Meier analysis (Fig. 2A-B).

Fig. 2.

Two identified significant SNPs of the arachidonic acid metabolism pathway genes predict overall survival of HBV-HCC patients. Kaplan–Meier survival curves for OS in the HBV-HCC dataset for (A) the combined protective alleles and (B) dichotomized groups of the NPAs. C Three-year HCC OS prediction by ROC curve; (D) Time-dependent AUC curves for the clinical model and the clinical model plus protective alleles. Solid curves represent the estimated AUC values, and dashed curves represent the corresponding 95% confidence intervals. #Protective alleles were CYP2A6 rs28399433 A allele and ABHD12 rs3827014 T allele. Abbreviations: SNPs, single-nucleotide polymorphism; NPA, number of protective alleles; ROC, receiver operating characteristic curve; AUC, area under curve

Furthermore, we evaluated the predictive capability of the two SNPs at the 1-, 3-, and 5-year survival for HBV-HCC by comparing the AUC of models that included clinical variables with models incorporating the protective alleles. Adding the protective alleles to the clinical model resulted in a statistically significant but clinically modest improvement in 3-year survival prediction, with the AUC increasing from 72.72% to 74.06% (P = 0.043, Fig. 2C). However, no significant improvement was observed for 1-year survival (AUC: 71.07% to 71.69%, P = 0.442) or 5-year survival (AUC: 72.04% to 73.13%, P = 0.143) (Supplementary Figure S3E and F). The time-dependent AUC curves are displayed in Fig. 2D.

In subgroup analyses, patients carrying 2 to 4 NPAs showed better OS than those with 0 or 1 across most strata, except for younger individuals (≤ 47 years), females, non-smokers, non-drinkers, AFP levels > 400 ng/ml, without cirrhosis, with vascular embolus and BCLC stage 0/A (Supplementary Table S4). Notably, significant multiplicative interactions were observed between the genetic score and smoking status (P_interaction = 0.016) as well as drinking status (P_interaction = 0.049) (Supplementary Table S4), suggesting potential gene-environment interactions influencing HCC survival outcomes. Additive interactions were also evident for smoking and BCLC stage, with RERI (95% CI) values of 0.65 (0.10–1.19) and 1.25 (0.17–2.33), respectively (Supplementary Table S4). The combined effects of these factors with a lower protective allele count (0–1 NPA) surpassed the sum of individual effects.

Bioinformatics functional prediction

Regarding the two independent SNPs, bioinformatic analyses using HaploReg v4.2 predicted their potential functional implications. Specifically, the CYP2A6 rs28399433 A > C variant is located within DNA regions characterized by promoter and enhancer histone modifications, DNase hypersensitivity sites, and transcription factor binding motifs. Similarly, the ABHD12 rs3827014 C > T variant is situated in DNA regions associated with promoter histone signals, DNase hypersensitivity sites, and transcription factor binding motifs (Supplementary Table S5).

The eQTL analysis and dual-luciferase reporter assay

The eQTL analysis revealed a significant association between the rs28399433 A allele and elevated expression of CYP2A6 in both liver (P = 2.59 × 10^− 3, Fig. 3A) and whole blood (P = 6.92 × 10^− 4, Fig. 3B). Additionally, the rs3827014 T allele exhibited a significant association with reduced ABHD12 expression in liver (P = 6.33 × 10^–17, Fig. 3C), but no such significant association was observed in whole blood samples (Fig. 3D).

Fig. 3.

eQTL analysis of CYP2A6 rs28399433 and ABHD12 rs3827014 and dual-luciferase reporter assay. The correlation of rs28399433 genotypes and CYP2A6 mRNA expression in (A) normal liver tissue (n=261, P=2.59 × 10-3) and (B) whole blood (n=800, P=6.92 × 10-4). The correlation of rs3827014 genotypes and ABHD12 mRNA expression in (C) normal liver tissue (n=261, P=6.33 × 10-17) and (D) whole blood (n=800, P=0.314); (E, F) Normalized reporter gene activity from the constructed fragments of CYP2A6 rs28399433 A > C and ABHD12 rs3827014 C > T in the HEK-293T cell lines. Abbreviations: eQTL, expression quantitative trait; SNPs, single-nucleotide polymorphism. ****P < 0.0001

To further validate the functional effect of rs28399433 and rs3827014, we conducted luciferase reporter assays in HEK-293T cells. The rs28399433 C allele significantly decreased reporter activity relative to the A allele (P < 0.0001), suggesting that the rs28399433 A > C variant may suppress the transcriptional activity of CYP2A6 (Fig. 3E). Similarly, the rs3827014 T allele significantly decreased reporter activity compared to the C allele (P < 0.0001), indicating that the rs3827014 C > T variant may inhibit the transcriptional activity of ABHD12 (Fig. 3F).

Differential mRNA expression analysis and survival of HCC

As shown in Fig. 4A, in UALCAN database, CYP2A6 mRNA levels were significantly reduced in HCC compared with normal tissues (P = 6.43 × 10^− 7), while ABHD12 showed a significant increase in tumor tissues (P < 1.00 × 10^–12, Fig. 4D). A similar pattern was observed in our collection of 103 paired tissue samples (CYP2A6: P = 3.80 × 10^–15; ABHD12: P < 2.22 × 10^–16) (Fig. 4B and E). Most critically, survival analysis using KMplot database demonstrated that patients with lower CYP2A6 or higher ABHD12 expression exhibited significantly poorer OS in HCC patients (P = 6.90 × 10^− 4 and P = 4.70 × 10^− 3, Fig. 4C and F). Supporting this, UALCAN survival analysis indicated that higher ABHD12 expression was associated with a poor survival in patients with HCC (P = 0.00044) (Supplementary Figure S3H). Although reduced CYP2A6 expression did not reach statistical significance (P = 0.29), a similar trend toward worse survival was observed (Supplementary Figure S3G).

Fig. 4.

Differential mRNA expression analysis and overall survival analysis of CYP2A6 and ABHD12 using KMplot database. A, D Higher CYP2A6 and lower ABHD12 mRNA expression levels were found in the normal tissues compared to LIHC tissues from UALCAN database; (B, E) Higher CYP2A6 and lower ABHD12 mRNA expression levels were found in the normal tissues compared to HCC tissues in the 103 paired tumor and liver tissue samples. C Lower expression levels of CYP2A6 was correlated with poorer survival. F Higher ABHD12 mRNA expression levels were correlated with poorer survival. Abbreviations: LIHC, liver hepatocellular carcinoma

Mutation analyses

Finally, we assessed somatic alteration frequencies of CYP2A6 and ABHD12 in HCC using data retrieved from cBioPortal for Cancer Genomics. As shown in Supplementary Figure S4, CYP2A6 exhibited low mutation prevalence across multiple cohorts, with observed rates of 1.75% in the MERiC Basel cohort, 0.87% in the AMC cohort, 0.80% in the TCGA Firehose Legacy cohort, 0.55% in the TCGA PanCancer Atlas cohort, and 0.41% in the INSERM cohort. ABHD12 showed similarly low frequencies, with rates of 1.23% in INSERM, 0.80% in TCGA Firehose Legacy, 0.55% in TCGA PanCancer Atlas, and 0.40% in the CLCA cohort. These findings indicate that somatic mutation of these genes is rare in HCC. Therefore, regulatory germline variants may partially contribute to CYP2A6 and ABHD12 mRNA expression, representing one of several regulatory factors involved in their transcription.

Discussion

In this study, we comprehensively examined the relationship between SNPs within 56 genes involved in the AA metabolism pathway and OS among patients with HBV-HCC, and identified two functionally significant variants, CYP2A6 rs28399433 A > C and ABHD12 rs3827014 C > T, associated with the OS. A combined genotype analysis revealed that patients carrying more protective alleles (NPAs) tended to have improved OS. Intriguingly, our findings demonstrated that the NPAs exhibited significant multiplicative interaction with both smoking and drinking in modulating OS. Moreover, we observed additive interaction effects for smoking status and BCLC stage. The eQTL analysis showed that the rs28399433 A allele was associated with higher CYP2A6 mRNA expression in both normal liver and blood samples, while the rs3827014 C allele was associated with elevated ABHD12 levels expression in normal liver tissue. These transcriptional effects were further validated through dual-luciferase reporter assays in HEK-293T cells. Comparative expression analyses indicated that CYP2A6 expression was markedly downregulated, whereas ABHD12 expression was upregulated in hepatocellular carcinoma tissues relative to adjacent normal liver, as observed in both the UALCAN dataset and 103 paired tumor/normal tissues. Moreover, data from UALCAN and KMplot confirmed that lower CYP2A6 or higher ABHD12 expression predicted worse survival in HCC patients.

In addition, the mutation analyses provide important context for interpreting our findings. The consistently low somatic alteration frequencies of CYP2A6 and ABHD12 across multiple independent HCC cohorts indicate that these genes are unlikely to serve as classical mutation-driven drivers in hepatocarcinogenesis. Instead, their aberrant expression patterns observed in tumor tissues are more likely attributable to transcriptional dysregulation rather than recurrent somatic events. In this context, our eQTL analyses and dual-luciferase assays offer complementary evidence supporting a regulatory role of germline variants. Collectively, these findings suggest that functional germline variants in CYP2A6 and ABHD12 may contribute, at least in part, to HCC prognosis through modulation of gene expression, highlighting a potential non-mutational regulatory mechanism underlying their clinical relevance.

CYP2A6 (cytochrome P450 family 2 subfamily A member 6), located on chromosome 19q13.2, serves as a crucial metabolic enzyme for various xenobiotics including nicotine, coumarin, and bilirubin [36, 37]. Beyond its metabolic role, emerging evidence implicates CYP2A6 in antitumor immunity and inflammation regulation [38]. Intriguingly, epidemiological studies have revealed tissue-specific associations between CYP2A6 polymorphisms and cancer risk, with variants conferring increased susceptibility to pancreatic and colorectal cancers but showing protective effects against lung and esophageal malignancies [39–43]. Particularly relevant to HCC, accumulating evidence supports a tumor-suppressive role for CYP2A6. Mechanistic studies demonstrate that CYP2A6 deficiency compromises macrophage maturation and phagocytic capacity, disrupts p53-mediated hepatic regeneration, and ultimately promotes HCC progression [38, 44, 45]. At the molecular level, CYP2A6 forms a complex with SRC, inhibiting the SRC/Wnt/β-catenin pathway and downstream oncogenes such as DVL2 and c-Myc, thereby suppressing tumor proliferation and migration [46]. Furthermore, CYP2A6 modulates AA metabolism by regulating the metabolic balance between 20-HETE and EETs, with consequent effects on the renin-angiotensin-aldosterone system (RAAS) pathway activation [38, 47, 48].

In this study, CYP2A6 rs28399433 A > C was associated with poorer OS among patients with HBV-related HCC and was also associated with reduced CYP2A6 mRNA expression levels in both normal liver and whole blood tissues. However, we observed significantly decreased CYP2A6 expression levels in HCC tissues compared with normal liver tissues. Analysis using the KMplot database further revealed that higher CYP2A6 expression was associated with improved overall survival. These observations indicate that CYP2A6 may play a potential tumor-suppressor role in the progression of HCC.

ABHD12 (abhydrolase domain-containing protein 12), located on chromosome 20p11.21, encodes a transmembrane serine hydrolase involved in lipid metabolism, immune regulation, and neural function [49, 50]. It has also been reported that ABHD12 may influence the levels of AA by hydrolyzing 2-arachidonoylglycerol (2-AG) [51]. Supporting evidence indicates that ABHD12 is overexpressed in HCC and other malignancies; its silencing inhibits tumor cell proliferation, migration, and invasion [52, 53]. In this study, ABHD12 rs3827014 C > T was associated with better survival of HBV-related HCC and was also associated with reduced ABHD12 mRNA expression levels in normal liver. However, we found significantly increased ABHD12 expression levels in HCC tissues compared with normal liver tissues, and higher ABHD12 expression levels were associated with poorer OS. We speculate that the functional SNP rs3827014 may alter transcription factor binding or chromatin architecture in the ABHD12 promoter, leading to dysregulated gene expression. Finally, the exact molecular mechanisms underlying the observed associations between the two SNPs and patient survival remain incompletely understood. Although rs3827014 was selected for functional validation, other linked variants within the same LD block may also contribute to transcriptional regulation. Therefore, further fine-mapping and functional studies are needed to disentangle their individual contributions.

Cancer is a complex, multifactorial disease influenced by both extrinsic and intrinsic factors [54, 55]. In HCC, gene-environment interactions may play a critical role in disease progression [56]. In this study, we observed significant interactions between genetic scores and smoking, as well as alcohol consumption, regarding the OS of HCC. In addition, additive interactions were observed between genetic scores and smoking, as well as BCLC stage B/C, indicating synergistic effects beyond the expected additive contributions. Our findings indicated that smoking and drinking were independent risk factors for HCC [57, 58]. Cigarette smoke contains multiple hepatotoxic and carcinogenic substances, such as tar and vinyl chloride, that can damage hepatic DNA and trigger mutagenesis [59]. Smoking also promotes oxidative stress, hepatic stellate cell activation, and fibrosis. Genetic polymorphisms in CYP2A6, such as *1 × 2 A, *1 × 2B (gene duplications), and *1B, that enhance enzymatic activity accelerate nicotine metabolism, leading to heavier smoking behavior and increased addiction susceptibility [60–62]. Although this has not been reported in HCC, multiple studies in lung cancer have demonstrated an interaction between CYP2A6 activity and cigarette consumption in influencing lung cancer susceptibility [63, 64]. Alcohol contributes to hepatocarcinogenesis likely through immune dysregulation and cytokine signaling alterations [65, 66]. These exposures may synergize with genetic susceptibility to worsen HCC outcomes. These findings suggest that smoking and alcohol use may represent modifiable risk factors, particularly relevant to individuals harboring rs28399433 or rs3827014 survival-associated variants, potentially influencing long-term outcomes in HCC.

While our study provides some valuable insights, several limitations should be acknowledged. First, all 866 HBV-HCC patients were of Chinese ethnicity and recruited recruited from the Cancer Hospital of Guangxi Medical University, which may limit the generalizability of our findings to populations. Multicenter studies with larger and more diverse cohorts are therefore warranted. Second, some clinically relevant variables, such as nutritional status, post-hepatectomy treatments, antiviral therapy, and des-gamma-carboxy prothrombin (DCP)—a marker associated with tumor aggressiveness, were not fully available. Future studies incorporating more comprehensive clinical and biomarker data are warranted to further validate our findings. Third, the functional assays were conducted in non-hepatic cells; thus, further validation in HCC-derived cell lines (e.g., HepG2 or Huh7) is necessary to better characterize the regulatory effects of these variants in a liver-specific context. Finally, the exact molecular mechanisms underlying the observed associations between the identified SNPs and HBV-HCC survival remain to be elucidated and require further investigation.

Taken together, our study identified two functional SNPs (CYP2A6 rs28399433 A > C and ABHD12 rs3827014 C > T) that were significantly associated with survival of HBV-HCC patients, potentially through regulation of the corresponding mRNA expression levels. These findings suggest that genetic variants in the AA metabolism pathway genes could serve as survival biomarkers for HBV-HCC patients, once further validated by other investigators.

Authors’ contributions

JP, QL, ZZ, HY, and XW conceived the study design, and supervised the whole project. JP performed the experiments. XW, JP, QL, and ZZ contributed to the data interpretation, data analysis, and writing of the manuscript. YJ, PC, YL, QM and QW contributed to the study design, and data interpretation of the present analysis. All the authors approved the submitted draft.

Funding

This work was supported by the Youth Science Foundation of Guangxi Medical University (GXMUYSF202503), the Joint Project on Regional High-Incidence Diseases Research of Guangxi Natural Science Foundation (Grant Number 2023GXNSFBA026049, 2025GXNSFBA069420, 2025GXNSFBA069580), the Project on Regional High-Incidence Diseases Research of Guangxi Natural Science Foundation (2024GXNSFDA010040), the Youth Program of Scientific Research Foundation of Guangxi Medical University Cancer Hospital (YQJ2024-09), the 111 Projects (D17011), and the Promoting Project of Basic Capacity for Young and Middle-aged University Teachers in Guangxi (2021KY0099).

Data availability

The original contributions presented in the study are described in the Materials and Methods, further information is available from the corresponding author upon request.

Declarations

Ethics approval and consent to participate

The study was conducted according to the guidelines of the Declaration of Helsinki and was approved by the Institutional Review Board of Guangxi Medical University Cancer Hospital (approval number: KY2025629), and written informed consent was obtained from each participant before data collection.

Consent for publication

Not applicable.

Competing interests

The authors declare no competing interests.