Exploring methylation signatures for high de novo recurrence risk in hepatocellular carcinoma
Article information
Abstract
Background/Aims
Hepatocellular carcinoma (HCC) exhibits high de novo recurrence rates post-resection. Current post-surgery recurrence prediction methods are limited, emphasizing the need for reliable biomarkers to assess recurrence risk. We aimed to develop methylation-based markers for classifying HCC patients and predicting their risk of de novo recurrence post-surgery.
Methods
In this retrospective cohort study, we analyzed data from HCC patients who underwent surgical resection in Korea, excluding those with recurrence within one year post-surgery. Using the Infinium Methylation EPIC array on 140 samples in the discovery cohort, we classified patients into low- and high-risk groups based on methylation profiles. Distinctive markers were identified through random forest analysis. These markers were validated in the cancer genome atlas (n=217), Validation cohort 1 (n=63) and experimental Validation using a methylation-sensitive high-resolution melting (MS-HRM) assay in Validation cohort 1 and Validation cohort 2 (n=63).
Results
The low-risk recurrence group (methylation group 1; MG1) showed a methylation average of 0.73 (95% confidence interval [CI] 0.69–0.77) with a 23.5% recurrence rate, while the high-risk group (MG2) had an average of 0.17 (95% CI 0.14–0.20) with a 44.1% recurrence rate (P<0.03). Validation confirmed the applicability of methylation markers across diverse populations, showing high accuracy in predicting the probability of HCC recurrence risk (area under the curve 96.8%). The MS-HRM assay confirmed its effectiveness in predicting de novo recurrence with 95.5% sensitivity, 89.7% specificity, and 92.2% accuracy.
Conclusions
Methylation markers effectively classified HCC patients by de novo recurrence risk, enhancing prediction accuracy and potentially offering personalized management strategies.
Graphical Abstract
INTRODUCTION
Hepatocellular carcinoma (HCC), a prominent malignancy with high recurrence rates post-resection, presents a challenge in patient management [1-3]. Resection, often the primary treatment option, is chosen frequently because of the limited availability of organs for transplantation. However, nearly 70% of HCC patients experience recurrence within five years [1,4-7]. Although the cut-off may vary, HCC recurrence after curative resection is generally categorized as either early or de novo, with each type linked to distinct underlying mechanisms; early recurrence is commonly attributed to latent primary cancer spread, while de novo recurrence is possibly attributable to de novo HCCs or multiple lesions in the remaining liver [8].
Despite these associations, the mechanisms behind recurrence remain elusive, limiting the development of reliable predictive markers. Existing used prognostic markers, such as alpha-fetoprotein (AFP) and prothrombin induced by vitamin K absence or antagonist-II (PIVKA-II), show varied efficacy across cancer stages, indicating the need for more precise markers [1,5,9,10]. DNA methylation patterns, emerging early in cancer progression, have shown promise as clinical biomarkers in various cancers, including colorectal cancer [2,11-14]. However, the role of methylation markers in predicting HCC recurrence post-resection remains understudied [15,16].
In this study, we aimed to develop a DNA methylation-based prognostic model specifically targeting de novo recurrence in HCC.
MATERIALS AND METHODS
Study design and patients
A retrospective analysis was conducted on consecutive patients who underwent surgical resection for HCC. We analyzed tissue samples from 266 patients, comprising 140 tissue samples from the discovery cohort, 63 from Validation cohort 1, and 63 from Validation cohort 2 at Seoul National University Hospital (SNUH) and the tissue samples were collected between 2011 and 2016. Patient information was obtained from electronic medical records. Additionally, we included 217 patients from the the cancer genome atlas (TCGA cohort) [17], conducting genome-wide DNA methylation analysis across all cohorts. Recurrence-free survival (RFS) was defined as the interval from the date of surgery to the date of recurrence detection. Utilizing a total of four distinct HCC cohorts, we deliberately excluded cases of early recurrence (within 1 year) to focus on de novo recurrence [8,18]. Patients with history of previous treatments for HCC, including resection, transarterial embolization, radiofrequency ablation, or percutaneous ethanol injection, were not considered for this study. The post-surgical follow-up of these patients entailed systematic monitoring for any signs of HCC recurrence, adhering to the hospital’s standard operating procedures, which included CT (computed tomography) and MRI (magnetic resonance imaging) scans every three months to meticulously track their health status. After curative resection, the enrolled patients were followed up until their demise to monitor their disease progression. During liver resection, tissue samples were collected from both cancerous and non-cancerous regions. The ‘tumor’ samples were obtained from areas within the tumor that appeared macroscopically malignant. In addition, ‘non-tumor’ samples were taken from adjacent liver tissue that was situated in close proximity to the tumor. Tissue specimens extracted during surgery were immediately frozen and stored at –80°C, in line with the standardized institutional protocols. The detailed demographic and clinical data of the participants are systematically presented in Table 1. The study protocol was approved by the Institutional Review Boards (IRB) of Seoul National University Hospital (IRB No. H-1602-016-739) as conforming to the ethical standards outlined in the Declaration of Helsinki and its subsequent amendments. All participants provided written, informed consent.
Characteristic of clinical factors in enrolled cohorts
RESULTS
Patient characteristics
Patient demographics and clinical characteristics are presented in Table 1. Briefly, the mean age of the 483 patients was 59.7 (95% confidence interval [CI] 58.7–60.8), and 364 were males (75.4%). Among these patients, 293 (60.2%) were HBsAg positive.
Methylation biomarkers for de novo recurrence risk were identified in the discovery cohort and the performance of the markers was validated in two Validation cohorts: Validation cohort 1 and TCGA. Furthermore, we experimentally confirmed the effectiveness of these markers in both Validation cohort 1 and 2 using methylation-sensitive high-resolution melting (MS-HRM) (Supplementary Fig. 1).
The median RFS was 53.4 months (95% CI 49.9–56.8) for the discovery cohort, 76.8 months (95% CI 67.9–85.6) for Validation cohort 52.2 months (95% CI 23.2–28.5) for the TCGA cohort, and 79.1 months (95% CI 69.2–89.1) for Validation cohort 2 (Supplementary Fig. 2).
Epigenetic profiling for postoperative prognostication in HCC
Principal component analysis effectively differentiated uniform methylation patterns in normal liver tissue from heterogeneous profiles observed in tumors (Supplementary Fig. 3). This underscores consistent methylation patterns observed in normal samples, even in regions with the most variable probes (Supplementary Fig. 4). From the tumor specimens, we selected the top 3,000 variable probes and utilized consensus clustering to categorize HCC into two distinct subgroups: Subgroup 1 (n=81) and Subgroup 2 (n=59), with unique methylation signatures (Fig. 1A). Differential methylation analysis identified 1,328 probes with substantial methylation differences between these subgroups (median methylation difference ≥0.5, FDR-adjusted P<0.01). Methylation group 1 (MG1) mirrored the methylation profile of normal liver tissue, whereas methylation group 2 (MG2) showed marked deviations, predominantly demethylated (96.8%) compared to normal samples (Supplementary Fig. 5).
HCC classification based on DNA methylation using consensus clustering. (A) Co-classification matrix for K=2 featuring the most variable 3,000 probes in tumor samples. The pink bar represents methylation group 1 (MG1), and the grey bar indicates methylation group 2 (MG2). In this heatmap of the clustered consensus matrix, both rows and columns represent tumor samples, with colors indicating the frequency of co-clustering across multiple iterations of k-means clustering. Dark blue represents samples that consistently cluster, while white indicates samples that rarely cluster. The intensity of the color reflects the frequency of co-clustering, with darker shades indicating higher clustering consistency. The color bar on top shows sample groupings within each methylation group. (B) Kaplan–Meier plots of RFS between the two HCC subgroups, highlighting the significance of P-values. (C) Box plot illustrating the methylation β-value of four candidate markers across normal liver (N), MG1, and MG2 samples. P-values are determined using an unpaired t-test. ****Represents P-value <0.0001. (D) T-distributed Stochastic Neighbor Embedding (t-SNE) analysis using four candidate markers. DMP, differentially methylated probe; HCC, hepatocellular carcinoma.
Gene set enrichment analysis using gene expression data from the two subgroups revealed distinct pathway enrichments. In MG1, pathways like Notch, WNT, and TGFβ were notably enriched; meanwhile, MG2 showed enrichment in pathways related to peroxisomes and metabolism (Supplementary Fig. 6).
A comparison of clinical outcomes between these HCC subgroups revealed a poorer prognosis for MG2 characterized by a significantly higher recurrence rate (23.5 vs. 44.1%, MG1 vs. MG2). The median RFS for MG1 was 49.4 months (95% CI 44.8–54.2), for MG2 was 43.9 months (95% CI 39.1–48.7) (Fig. 1B and Supplementary Fig. 7; log-rank P=0.017).
Deciphering methylation markers for high-risk recurrence in HCC
To delineate methylation markers for high de novo recurrence risk in HCC, we employed a machine learning approach, specifically using the Random Forest algorithm, on two subgroups of patients with HCC. The discovery cohort (n=140) was randomly split into training and validation sets, with 5-fold cross-validation used to identify predictive markers for de novo recurrence risk based on differences in RFS between MG1 and MG2.
Our analysis revealed four key methylation markers: cg21325760 (MAGEL2), cg10544510, cg06702718, and cg15997204 (MYT1L) (Supplementary Table 1). MG1 (low risk) showed methylation patterns similar to those of normal liver, while MG2 (high risk) exhibited significant demethylation (Fig. 1C). T-distributed Stochastic Neighbor Embedding analysis emphasized the distinct separation of these groups (Fig. 1D).
Evaluating the predictive performance of these four markers, individual performances were strong, with area under the curve (AUC) ranging from 90.8% to 96.5%, sensitivities from 85.0% to 89.7%, and specificities from 79.1% to 88.6% in discovery cohort. Dual-marker panels with the combination of cg10544510 and cg06702718 achieving the highest performance (AUC: 96.8%, sensitivity: 91.0%, specificity: 89.7%). The average AUC increased with more markers, reaching 97.3% for triple-marker combinations and 97.7% for quadruple-marker panels (Supplementary Table 2).
Comprehensive validation of the markers in independent HCC cohorts
To validate the methylation markers identified for high-risk recurrence stratification in HCC, we extended our analysis to two independent Validation cohorts: Validation cohort 1 and TCGA, respectively, excluding early recurrence cases. Hierarchical clustering based on methylation profiles (microarray) successfully identified predicted-MG1 and predicted-MG2 (referred to as Pre-MG1 and Pre-MG2) based on their methylation profiles. In the Validation cohort 1, the markers separated patients into Pre-MG1 (n=31) and Pre-MG2 (n=32) groups (Fig. 2A). Similarly, in the TCGA cohort, despite the exclusion of certain markers (cg06702718), the remaining marker (cg10544510) effectively distinguished patients into Pre-MG1 (n=82) and Pre-MG2 (n=135) groups (Fig. 2B).
Validation of four HCC prognostic candidates in two independent HCC cohorts. (A, B) Methylation heatmap of prognostic candidates in the Validation cohort 1 (Pre-MG1: 31 samples, PreMG2: 32 samples) (A) and the TCGA cohort (Pre-MG1: 82 samples, PreMG2: 135 samples) (B). X-axis is liver tumor patients and y-axis is methylation probe. In the TCGA cohort (B), only one CpG site (cg10544510) was used due to the absence of two methylation panel members in the TCGA dataset, which utilized a different version of the methylation chip (TCGA: 450K) compared to Validation cohort 1 (850K). (C, D) Kaplan–Meier plots for recurrence-free survival in the Validation cohort 1 (C) and TCGA cohort (D), with P-value significance indicated. HCC, hepatocellular carcinoma; TCGA, the cancer genome atlas.
Kaplan–Meier analysis showed that Pre-MG2 was associated with lower RFS than Pre-MG1 in the Validation cohort 1. The median RFS for Pre-MG1 was 81.3 months (95% CI 64.6–98.0), for Pre-MG2 was 60.8 months (95% CI 43.9–77.7) (log-rank P=0.029; Fig. 2C). This trend was mirrored in the TCGA cohort, where Pre-MG1 had a median RFS of 67.6 months with 22 recurrences, compared to Pre-MG2’s median RFS of 39.8 months with 49 recurrences (log-rank P=0.068; Fig. 2D).
Experimental validation of the methylation markers using MS-HRM
To validate our candidate methylation markers for HCC, MS-HRM analysis was employed [19]. This technique discerns methylation levels in clinical samples via distinctive melting curves, assigning scores from 0 to 100 based on area under melting curves.
Concentrating on the most effective marker pair, cg10544510 and cg06702718, we measured their methylation status against established DNA methylation controls, classifying HCC samples into MG1 or MG2. In the Validation cohort 1 (n=51), using a 90th percentile cut-off value for MG2 (94.23), we achieved high accuracy in subgroup determination. The sensitivity, specificity, and accuracy for cg10544510 were 81.82%, 89.66%, and 86.27%, respectively, and 90.91%, 89.66%, and 90.20% for cg06702718. Combining both markers increased sensitivity to 95.45% with an accuracy of 92.16% (Supplementary Fig. 8).
Applied to an independent Validation cohort 2, this method effectively segregated 24 MG1 and 39 MG2 patients (Supplementary Fig. 9). The recurrence-free survival rate was about two times higher in MG2 compared to that in MG1. The mean RFS for MG1 was 84.6 months (95% CI 68.9–100.3), and for MG2 was 75.8 months (95% CI 62.9–88.8) (Wilcoxon P=0.035; Fig. 3).
Analysis of methylation in Validation cohort 2. (A) Kaplan–Meier plot for recurrence-free survival in the Validation cohort 2. (B) Bar graph showing recurrence rates in the Validation cohort 2, with black indicating recurrence and grey representing non-recurrence, comparing the outcomes between MG1 and MG2. MG, methylation group.
Assessing clinical factors in HCC recurrence risk prediction
Next, we evaluated the efficacy of various clinical factors including sex, age, TNM stage, serum AFP, PIVKA-II, MoRAL score, resection margin, tumor nodules, Milan criteria, and microvascular invasion in predicting the risk of recurrence in HCC. In the univariate analysis for RFS, the methylation group was identified as a significant factor across multiple cohorts: discovery cohort (hazard ratio [HR] 2.02; 95% CI 1.12–3.66; P=0.020), Validation cohort 1 (HR 2.06; 95% CI 1.06–4.01; P=0.033), TCGA cohort (HR 1.70; 95% CI 1.02–2.81; P=0.041), and Validation cohort 2 (HR 3.04; 95% CI 1.02–9.06; P=0.045) (Table 2). To reduce the influence of individual cohort characteristics, variables with P-values <0.01 in at least three cohorts were selected for multivariate analysis, including the methylation group and tumor stage. In the multivariate analysis, the methylation group remained a significant predictor of RFS in the discovery cohort (HR 1.87; 95% CI 1.02–3.43; P=0.042) and Validation cohort 1 (HR 2.04; 95% CI 1.05–3.96; P=0.036) and Validation cohort 2 (HR 2.99; 95% CI 1.01–8.92; P=0.049). Additionally, tumor stage was identified as a significant factor in the TCGA cohort (HR 1.87; 95% CI 1.05–3.33; P=0.034) and Validation cohort 2 (HR 6.94; 95% CI 1.43–33.63; P=0.016) (Table 3).
Univariable Cox’s proportional hazards regression of prognostic factors that influence recurrence-free survival after resection of hepatocellular carcinoma
Multivariate analysis of recurrence-free survival in 4 enrolled HCC cohorts
To further validate the robustness of the methylation marker, we merged the three Validation cohorts (Validation cohort 1, Validation cohort 2, and TCGA cohort). We evaluated its prognostic value using Kaplan–Meier analysis for RFS in combination with clinically significant factors, including tumor size, serum AFP levels, etiology, vascular invasion, and the methylation group. The results demonstrated that the methylation group stratified patients into distinct prognostic categories based on RFS. In the MG2, which is associated with poor prognosis, a combination of high AFP levels, positive alcohol consumption history, and the presence of vascular invasion was linked to significantly worse outcomes. Conversely, the MG1, characterized by a normal-like methylation profile, exhibited a more favorable RFS curve, with no significant differences observed across varying clinical factor combinations, indicating a stable prognostic outcome (Supplementary Fig. 10).
DISCUSSION
In this study, we conducted a comprehensive genome-wide DNA methylation analysis in HCC, uncovering two distinct methylation patterns. These patterns enabled the classification of HCC patients into two subgroups: MG1, resembling normal liver tissue methylation with a low de novo recurrence risk, and MG2, characterized by extensive demethylation indicating a higher degree of epigenetic disruption and associated with a high de novo recurrence risk. Utilizing Random Forest analysis, we identified key methylation markers distinguishing these groups with high accuracy. Validation in three independent cohorts, including diverse patient populations, confirmed the robustness of our methylation markers of the original HCC tissue in predicting de novo recurrence risk. The MS-HRM assay provided experimental validation for the distinct methylation patterns between MG1 and MG2, reinforcing the clinical relevance of these methylation-defined HCC clusters. Our findings demonstrated superior performance compared to traditional prognostic indicators including AFP, PIVKA, and Milan criteria, highlighting the potential of methylation profiles in HCC management, particularly in guiding post-surgical care.
DNA methylation alterations are well-established in early cancer stages, with their stability making them ideal for early detection and monitoring of cancer progression [20-22]. For instance, the US FDA-approved colorectal cancer diagnostic test utilizes hypermethylation of SEPT9 in blood samples. Similarly, commercially viable DNA methylation test, such as GSTP1 for prostate cancer, MGMT for glioblastoma, and SHOX2 for lung cancer, have shown promise in detecting these cancers [23]. The level of methylation can have a significant impact on the carcinogenicity of the liver. HCC-affected livers are not in a normal state and often show varying degrees of liver cirrhosis, which is similar to an early stage of cancer. Additionally, it is well-established that this altered liver state is associated with de novo recurrence of HCC [24]. Our study extends these findings by demonstrating the utility of methylation markers in a novel clinical setting. Furthermore, the integration of our methylation markers into current clinical practice could enhance the precision of HCC prognosis and guide personalized treatment approaches, underscoring the importance of molecular diagnostics in the field of cancer.
DNA methylation can serve as a key regulatory mechanism in HCC recurrence. To evaluate its role, we analyzed methylation levels in not only tumor but also background liver tissues. Consistently stable methylation patterns were observed across normal samples, even in regions with high variability, suggesting that background liver tissue is epigenetically distinct from tumor tissue. Interestingly, our findings indicate that DNA methylation reflects not only the tumor’s aggressiveness but also the immune response in the tumor microenvironment, potentially influencing immunity and recurrence risk. Specifically, DNA methylation plays a critical role in the maturation, polarization, and differentiation of immune cells, as well as their function, thereby influencing tumor immune evasion [25,26]. Notably, prior colorectal cancer subtyping studies have shown that hypomethylated subtypes are linked to immune exhaustion [27]. A similar immune response, including immune cell distribution, may occur in the background liver tissue during de novo carcinogenesis. In a favorable immune microenvironment, small, early-stage tumors may fail to evade immune surveillance and subsequently disappear. Conversely, in an unfavorable immune microenvironment, de novo recurrence may occur more easily as the tumor evades immune defenses. Although de novo tumors are clonally independent from the primary HCC, the immune response to these tumors may share similar characteristics. Thus, methylation patterns in primary tumor tissue may reflect the anticipated immune microenvironment during carcinogenesis and could help predict the risk of de novo recurrence.
Our prognostic methylation markers include MAGEL2 (MAGE Family Member L2; cg21325760), MYT1L (Myelin transcription factor 1-like; cg15997204), along with two non-coding region markers (cg10544510, cg06702718). MAGEL2, a cancer-testis antigen, influences cancer progression and immune signaling [28-30], demonstrating its potential as a prognostic factor in cancers such as breast and prostate [31,32]. Similarly, MYT1L, relevant in gastric and glioma progression, plays a crucial role in cell proliferation and invasion, with its overexpression correlating to tumor aggressiveness [33-35]. Both MAGEL2 and MYT1L were confirmed to have negligible gene expression (TPM<0.01), suggesting that their functional roles may not rely on transcriptional regulation. The correlation between methylation levels (differential methylation) and gene expression (log2 fold change) was minimal, with Pearson correlation coefficients of 0.05 and –0.18, respectively (Supplementary Fig. 11). These findings imply that these markers may act through alternative regulatory mechanisms such as histone modifications, chromatin remodeling, or non-coding RNA activity.
Further genomic context analysis revealed that these markers are located in repetitive elements and regulatory regions. Specifically, cg21325760 is located in a simple repeat, cg10544510 in a short interspersed nuclear element element, and cg06702718 in a DNA transposon. Additionally, two markers (cg06702718 and cg15997204), identified in the super-enhancer and Genehancer databases as cis-regulatory elements [36,37]. Super-enhancers are critical regulators of gene expression in adjacent genes, particularly in cancer, where their dysregulation can drive tumorigenesis [38]. Specifically, cg06702718 is hypothesized to regulate critical cancer progression genes, including P53 [39]. The mechanistic link between hypomethylation and increased recurrence risk in the MG2 lies in its multifaceted impact on genomic stability and tumor biology. Global hypomethylation, a hallmark of early carcinogenesis, contributes to chromosomal instability and loss of imprinting, leading to abnormal genomic rearrangements and enhanced tumor heterogeneity [40,41]. Hypomethylation also reactivates oncogenes and repeat elements, such as transposable elements, inducing chromatin structural changes that promote genomic instability and tumor progression [42-44]. Together, these processes contribute to the genomic instability, aggressive tumor behavior, and increased recurrence risk observed in the MG2.
While these findings differentiate HCC subgroups, the exact roles in cancer progression and clinical utility need further exploration. Our methylation panel was validated in three independent cohorts with diverse etiologies (Validation cohort 1: 82.6% hepatitis B virus [HBV]-HCC, TCGA: 33.5% HBV-HCC, 13% HCV-HCC, and 37.7% alcohol abuse) and racial backgrounds (Validation cohort 1 and Validation cohort 2: 100% Asian, TCGA: 50% White, and 42.3% Asian), confirming the universal applicability of the identified markers. Upon examining variables with marginal significance (P<0.10) in the univariate analysis, a notable statistical disparity was observed in three cohorts within the methylation group. In the subsequent multivariate analysis, which included clinically relevant variables such as age, gender, AFP, and PIVKA-II, the methylation group remained significantly differentiated in two cohorts. Although statistical significance was not observed in the TCGA cohort or Validation cohort 2, Validation cohort 2 showed a high HR (3.08). These results demonstrate the consistency of the identified markers across multiple cohorts and highlight their potential for clinical application. No statistical significance was found for the remaining variables. The MSHRM technique’s application to the Validation cohort 2 further supported the stratification of HCC subgroups. The significant difference in de novo recurrence rates between the MG2 and MG1 highlights the potential of our identified markers in monitoring HCC recurrence informing subsequent treatment strategies, including the consideration of preemptive or salvage liver transplantation (LT). Specifically, in MG2 with high de novo risk, the short-term monitoring interval is recommended. Furthermore, the preemptive LT can be considered regardless of the degree of cirrhosis. Also, the salvage LT needs to be strongly suggested to patients with de novo recurrence. These findings emphasize the potential of our methylation markers to guide personalized treatment strategies and improve HCC management outcomes.
A notable strength of our study is the homogeneity of the early-stage HCC patient cohort. The majority of these patients were in Child–Pugh class A (discovery: 100%; Validation cohort 1: 100%, TCGA: 65.2%, Validation cohort 2: 98.4%) and predominantly in tumor stages 1 and 2 (discovery: 90%; Validation cohort 1: 93.6%, TCGA: 82.5%, Validation cohort 2: 93.6%). This homogeneity provides a clear picture of the potential of our identified markers in early-stage HCC, a crucial aspect in the landscape of cancer prognosis and management. The absence of standardized post-resection adjuvant treatments for HCC underscores the significance of our findings, suggesting potential personalized therapy paths based on methylation profiles. Moreover, combining these methylation panels with clinical data promises a more comprehensive prognostic tool, potentially improving treatment and disease management. Additionally, the long-term follow-up in most cohorts (discovery cohort: average 4.75 years; Validation cohort 1: 8.71 years; Validation cohort 2: 7.69 years) except for the TCGA cohort (2.23 years) adds to the robustness of our findings.
Despite these strengths, our study is limited by its focus on post-surgical liver tissues and its retrospective nature. To improve the assay’s efficacy and applicability, future research should aim to validate these markers in larger, multicenter independent HCC cohorts and assess their performance in preoperative biopsy samples and circulating tumor DNA. These efforts will enable broader use in noninvasive diagnostics and real-time disease monitoring. Additionally, further validation across diverse ethnic groups, including Asian, European, and African populations, and in different liver diseases, such as nonalcoholic steatohepatitis and cirrhosis, will strengthen the clinical relevance and generalizability of our findings.
In conclusion, our study leverages genome-wide methylation profiling to distinguish HCC patients into two prognostically significant subgroups, MG1 and MG2. This novel methylation marker, validated across various etiologies and racial backgrounds, holds substantial prognostic significance, especially in early-stage HCC. Our findings not only deepen the understanding of HCC progression but also open avenues for more personalized and effective management strategies in clinical settings.
Notes
Authors’ contribution
DW Kim, JH Park, SK Hong, KW Lee, and YJ Kim were instrumental in conceptualizing and designing the study. JH Park, SK Hong, KS Suh, YR Choi, and NJ Yi played a pivotal role in patient enrollment and the compilation of clinical information. The experimental work and data collection were primarily conducted by DW Kim, MH Jung, and JO Pyeon, with MH Jung and JY Lee contributing significantly to the data analysis. The interpretation of the data was a collaborative effort involving DW Kim, JH Park, SK Hong, KW Lee, and YJ Kim. All authors engaged in a critical review of the manuscript, providing substantial intellectual input, and gave their final approval for the manuscript’s publication.
Acknowledgements
This research was supported by the Bio & Medical Technology Development Program of the National Research Foundation (NRF) funded by the Ministry of Science & ICT (grant number: NRF-2017M3A9A7050614).
Conflicts of Interest
The authors have no conflicts to disclose.
Abbreviations
AFP
alpha-fetoprotein
DM
differential methylation
HBV
hepatitis B virus
HCC
hepatocellular carcinoma
HCV
hepatitis C virus
MS-HRM
methylation-sensitive high-resolution melting
OS
overall survival
RFS
relapse-free survival
TCGA
the cancer genome atlas
SUPPLEMENTAL MATERIAL
Supplementary material is available at Clinical and Molecular Hepatology website (http://www.e-cmh.org).
Characteristics of four methylation signatures
Performance variation of markers based on probe combination in the discovery cohort
Cohort diagram of enrolled HCC cohorts.
Recurrence-free survival after surgical resection.
DNA Methylation Distribution in normal and tumor samples of liver.
Clustering analysis of the most variable probes in normal samples.
Distribution of differentially methylated probes.
Gene set enrichment analysis of differentially methylated probes in patients with HCC.
Recurrence rates in the discovery cohort.
MS-HRM methylation score analysis in the Validation cohort 1.
MS-HRM methylation score analysis in the Validation cohort 2.
Subgroup analyses of recurrence-free survival based on clinical characteristics in a Merged Validation cohort.
Correlation between gene expression and methylation for prognostic markers.
References
Article information Continued
Notes
Study Highlights
• In this cohort study involving 140 HCC patients, we identified two methylation markers using a machine learning method that are significantly associated with an increased risk of de novo recurrence. The performance of these methylation markers was validated in two cohorts: Validation cohort 1 and TCGA, and experimentally confirmed in both Validation cohort 1 and Validation cohort 2 using the MS-HRM assay. This novel methylation signature presents a potential tool for stratifying the risk of de novo recurrence in HCC patients post-resection and guiding postoperative management.