Abstract

Background

   Hepatocellular carcinoma (HCC) remains a major global health challenge,
   with the pathogenesis of HBV-induced HCC incompletely understood. Due
   to HCC’s significant heterogeneity, more accurate prognostic models and
   reliable biomarkers are urgently needed. This study aimed to develop
   such tools for HBV-related HCC.

Methods

   Protein expression profiles and clinicopathological data were obtained
   from the CPTAC (Clinical Proteomic Tumor Analysis Consortium) database.
   Differential protein expression analysis between HBV-induced HCC and
   normal tissues were performed, followed by univariate and multivariate
   Cox regression to construct a prognostic protein signature. The
   four-protein signature’s prognostic power was assessed using
   time-dependent receiver operating characteristic (ROC) curves,
   Kaplan-Meier analysis, multivariate Cox regression, and a nomogram.
   Protein expression levels of the four proteins were validated in
   clinical specimens of HBV-related HCC. The signature was further
   validated in an independent cohort.

Results

   A novel four-protein signature (SMC2, SMC4, UBE2C, UHRF1) was
   established for prognostic prediction in HBV-related HCC. ROC curves
   demonstrated good survival prediction performance in both the CPTAC HCC
   cohort and the validation cohort. Transcriptomic and
   immunohistochemical (IHC) analyses revealed significant overexpression
   of these four proteins at both mRNA and protein levels in HCC tissues.
   The signature stratified patients into high-risk and low-risk groups
   with significantly different survival outcomes. Multivariate Cox
   regression analysis confirmed that the 4-protein signature
   independently predicted overall survival (OS). Furthermore, the
   signature showed strong discriminatory power between HBV-related HCC
   and normal tissue. Time-dependent ROC analysis indicated that the
   protein signature exhibited superior specificity and sensitivity
   compared to alpha-fetoprotein (AFP) for diagnosing HBV-associated HCC.

Conclusions

   Our study establishes a novel 4-protein signature and nomogram for
   predicting OS in HBV-related HCC, potentially aiding individualized
   clinical treatment decisions.

Supplementary Information

   The online version contains supplementary material available at
   10.1007/s12672-025-03693-8.

   Keywords: HBV, Hepatocellular carcinoma, Prognostic marker, Protein
   signature

Introduction

   Primary liver cancer ranks as the sixth most common malignancy and the
   third leading cause of cancer-related death globally [[32]1].
   Hepatocellular carcinoma (HCC) is the predominant form, accounting for
   75–85% of cases [[33]2]. HCC development is influenced by genetic
   predisposition, viral infections, and environmental factors [[34]3].
   Key risk factors include metabolic syndrome, alcohol consumption,
   aflatoxin exposure, and hepatitis viruses (HBV and HCV), with HBV and
   HCV being the most significant [[35]4].

   HBV infects over 250 million people chronically worldwide. Infection
   can range from asymptomatic to causing liver failure and cancer
   [[36]4]. HBV contributes to HCC development through integration into
   the host genome, proto-oncogene activation, and interactions between
   HBV proteins and host factors that dysregulate cell signaling pathways
   [[37]5, [38]6]. Alcohol consumption further increases HCC risk in HBV
   patients by more than two-fold, primarily by accelerating liver
   fibrosis [[39]4].

   While HBV is a well-established risk factor for HCC, its precise
   molecular mechanisms remain incompletely understood. Current HCC
   treatments, including surgery and chemotherapy, have limited efficacy
   and high recurrence rates [[40]7, [41]8]. Early detection and treatment
   are crucial for improving outcomes, particularly in HBV-related HCC,
   highlighting the urgent need for research into novel mechanisms and
   biomarkers.

   This study analyzed proteomic data from 159 paired HBV-related HCC
   tumor and adjacent liver tissue samples obtained from the CPTAC
   database. Data was normalized, grouped, and differentially expressed
   proteins (DEPs) were identified. Gene Ontology (GO) and Kyoto
   Encyclopedia of Genes and Genomes (KEGG) enrichment analyses of DEPs
   were conducted to guide mechanistic investigations. Subsequently,
   Kaplan-Meier survival analysis, univariate and multivariate Cox
   regression, and time-dependent ROC analysis were employed to identify
   proteins predictive of overall survival (OS) in HCC patients. A
   prognostic risk score model based on protein expression was developed
   for OS prediction. Ultimately, the identified prognostic biomarkers
   derived from DEPs may facilitate targeted treatments and improved
   patient outcomes.

Materials and methods

Antibodies and reagents

   Antibodies targeting SMC2 (ab10399) and UHRF1 (ab213223) were obtained
   from Abcam (Cambridge, MA). The antibody specific to SMC4 (PA5-56150)
   was sourced from Thermo Fisher Scientific (Waltham, MA).
   Paraformaldehyde, hydrogen peroxide, and diaminobenzidine were acquired
   from ZSGB-BIO (Beijing, China).

Proteomics data and preprocessing

   Protein expression data and clinical information for HCC patients were
   retrieved from the CPTAC database ([42]https://pdc.cancer.gov/pdc/)
   [[43]9]. Data from 159 patients with complete information were combined
   for analysis (Supplementary Table 1) [[44]10–[45]13]. Proteomic data
   underwent quality control and normalization as described by Mertins et
   al. [[46]14]. Specifically, data normalization was achieved through the
   median centering method across the total proteins to account for
   variations in sample loading. K-nearest neighbor (k-NN) imputation, via
   the “impute” R package, addressed missing values, excluding proteins
   with over 50% missing data to ensure sufficient data for imputation
   [[47]9].

Identification of differentially expressed proteins (DEPs) and GO/KEGG
enrichment analysis

   The limma package was used to identify DEPs using thresholds of
   |log2FC| >1 and p < 0.05. The clusterProfiler package was used for GO
   and KEGG enrichment analysis, and GOplot for visualization. GO analysis
   categorized DEPs into Biological Processes (BP), Cellular Components
   (CC), and Molecular Functions (MF). KEGG analysis identified pathways
   associated with DEPs. A significance threshold of p < 0.05 was applied.

Protein-protein interaction (PPI) network, module analysis, and hub protein
identification

   The STRING database ([48]https://string-db.org/) was used to construct
   a PPI network of DEPs with an interaction score ≥ 0.9. The MCODE plugin
   identified functional modules within the PPI network using parameters:
   node score cutoff = 0.2, k-core = 2, max.depth = 100, degree cutoff = 2
   [[49]15]. The igraph package was used to visualize the core network of
   critical module genes.

Selection of prognostic and diagnostic proteins

   Kaplan-Meier analysis assessed the association between DEP expression
   and HCC patient prognosis. Univariate and multivariate Cox proportional
   hazards regression models were used to determine hazard ratios (HRs)
   and 95% confidence intervals (CI), evaluating the independent
   predictive value of clinicopathological parameters and DEP expression
   on survival (p < 0.05 considered significant). The area under the curve
   (AUC) of the time-dependent ROC for 1-, 3-, and 5-year OS was
   calculated for DEPs associated with prognosis. DEPs with prognostic
   relevance and AUC values > 0.7 were selected for further investigation.

Clinical validation of the 4-protein signature by immunohistochemistry (IHC)

   Twenty paired HCC tumor and adjacent normal tissue samples were
   collected from untreated patients undergoing surgery at Hangzhou TCM
   hospital affiliated to Zhejiang Chinese Medicial University between
   January 2024 and June 2025. Samples were divided: one part
   paraffin-embedded for histopathology, the other frozen at -80 °C. All
   patients had confirmed clinical and pathological HCC diagnoses; tumor
   stage and grade were defined according to UICC, AJCC, and WHO criteria.

   Paraffin-embedded tissues were sectioned at 4 μm. Sections were
   deparaffinized, rehydrated, and subjected to antigen retrieval. After
   PBS washing, endogenous peroxidase was blocked with 3% hydrogen
   peroxide for 10 min. Sections were then blocked with 5% BSA for 20 min.
   Primary antibodies (anti-SMC2 (1:50), anti-SMC4 (1:100), anti-UHRF1
   (1:50 or 1:100)) were applied overnight at 4 °C. Signal detection used
   the streptavidin-peroxidase method with diaminobenzidine (DAB) as the
   chromogen, followed by hematoxylin counterstaining. Sections were
   examined and photographed under light microscopy. Scoring was performed
   independently by two blinded observers.

Construction of the protein signature and risk score calculation

   Prognostic proteins (AUC > 0.7 in time-dependent ROC) were used to
   build a prognostic signature. An individual risk score for each HCC
   patient was calculated using a formula derived from the protein
   expression levels weighted by their regression coefficients in the
   multivariate Cox proportional hazards model. To identify the most
   prognostically relevant threshold that is robust to the underlying
   distribution of the risk score, the optimal cut-off value for
   stratifying patients was determined using the “surv_cutpoint” function
   (from the “survminer” R package), which maximizes the survival
   differences between groups based on the two-sided log-rank test
   statistic. This method is distribution-free and identifies the
   cut-point with the most significant association with survival outcomes.
   Subsequently, patients with HBV-related HCC were stratified into
   high-risk and low-risk cohorts based on this empirically derived
   threshold. Risk curve plots, survival status plots, and heatmaps
   visualized the patient risk distribution and protein expression
   patterns. The timeROC R package was used for time-dependent ROC curve
   analysis to evaluate the signature’s discriminatory capacity over time.

Construction of a predictive nomogram

   A nomogram incorporating independent prognostic factors identified by
   multivariate Cox regression was developed to predict 1-, 3-, and 5-year
   OS in HCC. Discriminatory ability was assessed using time-ROC curves.
   Calibration curves plotted predicted probabilities against observed
   survival rates to evaluate accuracy. The performance of the nomogram
   was compared to models using single prognostic factors.

Statistical analysis

   Statistical analyses were performed using R software version 3.6.3. The
   Wilcoxon test compared two groups of boxplot data. One-way ANOVA
   compared multiple groups of normalized data. Kaplan-Meier curves with
   log-rank tests and univariate Cox proportional hazards regression
   calculated p-values, HRs, and 95% CIs. Statistical significance was set
   at p < 0.05 unless otherwise stated.

Results

Identification of DEPs in HCC

   Compared to adjacent liver tissues, 418 DEPs were identified in HCC
   tissues: 109 upregulated and 309 downregulated (Supplementary Table 2).

Functional and pathway enrichment of DEPs

   GO and KEGG analyses revealed the functional roles of DEPs. The top 10
   enriched BPs were predominantly metabolic processes, including small
   molecule catabolic process, organic acid biosynthetic process,
   carboxylic acid biosynthetic process, and amino acid metabolic process.
   For CC, DEPs were significantly enriched in the mitochondrial matrix,
   collagen-containing extracellular matrix, peroxisome, microbody,
   collagen trimer, peroxisomal matrix, microbody lumen, CMG complex, DNA
   replication preinitiation complex, and MCM complex. Enriched MFs
   included oxidoreductase activity, iron ion binding, vitamin binding,
   monooxygenase activity, heme binding, tetrapyrrole binding, and
   arachidonic acid monooxygenase activity (Fig. [50]1A-B). KEGG pathway
   analysis showed enrichment in retinol metabolism, carbon metabolism,
   steroid hormone biosynthesis, amino acid biosynthesis and metabolism,
   fatty acid metabolism, pyruvate metabolism, and
   glycolysis/gluconeogenesis (Fig. [51]1C-D).

Fig. 1.

   [52]Fig. 1
   [53]Open in a new tab

   The differentially expressed proteins in HBV-related HCC were subjected
   to GO analysis and KEGG pathway analysis. A Conducting GO enrichment
   analyses on the proteins that were found to be differentially
   expressed. B The circular diagram depicting the enrichment of GO terms
   among the proteins that are differentially expressed. C KEGG pathway
   enrichment analysis on the proteins that exhibited differential
   expression. D The KEGG enrichment circular diagram illustrates the
   differentially expressed proteins

PPI network construction and hub gene identification

   The STRING database generated a DEP interaction network comprising 212
   nodes and 1038 edges (Fig. [54]2A). MCODE analysis identified 30 hub
   proteins (Fig. [55]2B). The top 10 DEPs with highest connectivity were
   CYP2E1, CYP3A4, AOX1, CYP1A1, CYP1A2, MCM2, MCM4, CDK1, TOP2A, and
   CYP2B6. CYP2E1, CYP3A4, AOX1, CYP1A1, CYP1A2, and CYP2B6 were
   predominantly downregulated (Fig. [56]2C-D), while TOP2A, CDK1, MCM4,
   and MCM2 were significantly upregulated (Fig. [57]2D).

Fig. 2.

   [58]Fig. 2
   [59]Open in a new tab

   Identification of hub proteins in the protein-protein interactions
   (PPI) network. A Utilizing the STRING online database to generate a PPI
   network of the differentially expressed proteins in HBV-related HCC. B
   The top 30 nodes in the PPI network based on their degree centrality.
   C-D The correlation network of ten hub genes is depicted with
   correlation coefficients represented by various colors

Survival analysis, cox regression, and time-dependent ROC identify four
prognostic proteins in HBV-related HCC

   Survival analysis using the R package, along with univariate and
   multivariate Cox analysis, investigated DEP associations with OS in
   HBV-related HCC. Initially, 105 DEPs were associated with OS.
   Univariate analysis identified 95 DEPs significantly correlated with
   OS. Multivariate Cox regression identified 37 DEPs as independent
   prognostic factors (data not shown). Time-dependent ROC curve analysis
   revealed that only four DEPs (SMC2, SMC4, UBE2C, UHRF1) had AUC
   values > 0.7.

   Kaplan-Meier analysis demonstrated that high expression of SMC2, SMC4,
   UBE2C, and UHRF1 was significantly associated with poor prognosis in
   HBV-related HCC (Fig. [60]3A, E, I, M). Multivariate Cox regression
   including sex, age, tumor differentiation, tumor size, tumor thrombus,
   and the four proteins confirmed that SMC2, SMC4, UBE2C, and UHRF1 were
   independent prognostic factors (Fig. [61]3C, G, K, O). ROC analysis
   indicated significant diagnostic efficacy of these proteins for
   predicting 1-year, 3-year, and 5-year survival rates (Fig. [62]3D, H,
   L, P).

Fig. 3.

   [63]Fig. 3
   [64]Open in a new tab

   Identification of four prognostic proteins in HBV-related HCC using
   Kaplan-Meier survival analysis, univariate and multivariate Cox
   regression analysis, and time-dependent receiver operating
   characteristic (ROC) analysis. A, E, I, M Utilized Kaplan-Meier curves
   to conduct survival analysis of SMC2 (A), SMC4 (E), UBE2C (I), and
   UHRF1 (M) in HCC related to HBV infection. Prognostic forest maps of
   proteins SMC2 B-C, SMC4 F-G, UBE2C J-K, and UHRF1 N-O in HBV-related
   HCC using univariate and multivariate Cox regression analysis. D, H, L,
   P Time-dependent ROC analysis of SMC2 (D), SMC4 (H), UBE2C (L), and
   UHRF1 (P) in HBV-related HCC

Correlation between four prognostic protein levels and clinical
characteristics

   We examined the correlation between the expression levels of the four
   prognostic proteins and clinical parameters (histological grade, tumor
   thrombus, age). The highest expression levels of all four proteins were
   associated with grade 3 histological tumors. SMC2, SMC4, and UHRF1
   expression was higher in tumors with thrombus compared to those
   without. UHRF1 expression was also higher in patients under 55 years
   old (Fig. [65]4A-H).

Fig. 4.

   [66]Fig. 4
   [67]Open in a new tab

   Correlation between the expression of four prognostic proteins and
   clinicopathologic characteristics. A-D The scatter plot illustrates the
   relationship between SMC2/4 expression and clinicopathologic features,
   such as histologic grade and tumor thrombus. E The correlation between
   the expression of UBE2C and the histologic grade of HBV-related HCC.
   F-H Box plot analysis was conducted to assess the UHRF1 expression
   levels in patients with HBV-related HCC, stratified by various clinical
   characteristics including histologic grade, presence of tumor thrombus,
   and age. The group designated as grade I in Fig. 4A, C, E, and F is not
   presented, as it is considered invalid due to the inclusion of only a
   single patient

Construction and validation of the 4-protein prognostic signature

   Multivariate Cox analysis confirmed the prognostic significance of
   SMC2, SMC4, UBE2C, and UHRF1, associating high expression with
   increased risk (Fig. [68]5A). The final risk score incorporated four
   protein biomarkers weighted by their Cox regression coefficients: risk
   score = 1.9106*Exp[(SMC2)]
   − 2.0133*Exp[(SMC4)] + 0.8297*Exp[(UBE2C)] + 0.2065*Exp[(UHRF1)]. The
   risk score formula was applied to each patient. Using the median risk
   score as the cutoff (determined by survminer), patients were stratified
   into high-risk (n = 71) and low-risk (n = 72) groups (Fig. [69]5B).
   Higher risk scores correlated with increased patient mortality
   (Fig. [70]5C). A heatmap confirmed significant overexpression of the
   four proteins in the high-risk group (Fig. [71]5D). Kaplan-Meier
   analysis revealed significantly worse OS in the high-risk group
   compared to the low-risk group (Fig. [72]5E). Time-dependent ROC
   analysis for the signature predicting 1-, 3-, and 5-year OS yielded AUC
   values of 0.775, 0.670, and 0.836, respectively (Fig. [73]5F),
   indicating sensitive and specific prognostic capability.

Fig. 5.

   [74]Fig. 5
   [75]Open in a new tab

   The development of a prognostic protein signature, along with the
   assessment of its predictive capabilities and validation. A Development
   of a protein signature through prognostic analysis in HBV-related HCC.
   (B) The distribution and median value of risk scores within the cohort
   of patients with HBV-related HCC. C The survival outcomes of patients
   with HBV-related HCC were examined, with a positive correlation
   observed between mortality rates and higher risk scores. D The heatmap
   displays the expression profiles of the four prognostic proteins in
   both low- and high-risk groups. E The Kaplan-Meier survival analysis
   was conducted on the four-protein signature. F The analysis of the
   time-dependent ROC curve for the four-protein signature

   Analysis of TCGA data revealed significant mRNA overexpression of SMC4,
   UBE2C, and UHRF1 in HCC tissues (p < 0.05) (Fig. [76]6B-D), but not
   SMC2 (Fig. [77]6A). However, analysis of RNA-seq data from Gao et al.
   confirmed SMC2 mRNA overexpression in HBV-related HCC compared to
   adjacent tissues (Fig. [78]6E). IHC analysis demonstrated significant
   protein overexpression of SMC2, SMC4, and UHRF1 in HBV-related HCC
   tissues (Fig. [79]7A-I and Supplementary Fig. 1–10). Supporting data
   from The Human Protein Atlas also showed marked overexpression of SMC4
   and UBE2C proteins in HCC tissues (Fig. [80]7J-K).

Fig. 6.

   [81]Fig. 6
   [82]Open in a new tab

   To confirm the mRNA expression of prognostic protein signature in HCC.
   A-D The mRNA levels of SMC2, SMC4, UBE2C, and UHRF1 in HCC and normal
   tissues from the TCGA cohort were analyzed. E SMC2 mRNA levels in
   HBV-related HCC and adjacent non-tumor liver tissues were quantified
   using RNA-Seq (n = 159). The results mentioned above were derived from
   the study carried out by Gao et al. [[83]9]. Data are depicted as mean
   ± SE. *, p < 0.05

Fig. 7.

   [84]Fig. 7
   [85]Open in a new tab

   Immunohistochemistry (IHC) confirmed the expression of prognostic
   protein signature in HCC. A, D, G Representative IHC images
   illustrating SMC2 expression in thyroid cancer (A), SMC4 in nephridial
   tissue (D), and UHRF1 in colorectal cancer (G), serving as a positive
   control, are presented. B, E, H IHC staining assessed SMC2 (B), SMC4
   (E), and UHRF1 (H) in HBV-related HCC and adjacent normal tissues, with
   images provided at 100× and 400× magnifications. C, F, I H-score
   statistics were calculated for 20 (C, F) or 5 (I) paired specimens. J-K
   Compared to normal liver tissue, SMC4 (J) and UBE2C (K) expression is
   higher in HCC, according to The Human Protein Atlas database

Construction and validation of a predictive nomogram

   Univariate and multivariate Cox regression confirmed that the 4-protein
   signature risk score, alongside tumor grade, tumor thrombus, and AFP
   level, were independent prognostic variables for OS in HBV-related HCC
   (Fig. [86]8A-B). A composite nomogram incorporating these factors was
   developed to predict 1-year, 3-year, and 5-year OS probabilities
   (Fig. [87]8C). Calibration plots showed good agreement between nomogram
   predictions and actual survival outcomes (Fig. [88]8D). The AUC values
   for the nomogram predicting 3-year and 5-year OS surpassed those of the
   individual risk score, AFP, tumor thrombus, or grade alone
   (Fig. [89]8F-G), indicating superior predictive performance.

Fig. 8.

   [90]Fig. 8
   [91]Open in a new tab

   The identification of independent prognostic parameters and the
   development and validation of a protein-based prognostic model. A
   Forrest plot depicting the results of univariate Cox regression
   analysis in HBV-related HCC. B A forest plot depicting the results of a
   multivariate Cox regression analysis in HBV-related HCC. (C) The
   nomogram incorporates four protein-based risk score, AFP levels,
   presence of tumor thrombus, and tumor grade. D The calibration plot of
   the nomogram was utilized to assess the agreement between predicted 1-,
   3-, and 5-year OS rates and actual outcomes in patients with
   HBV-related HCC. E-G The time-dependent ROC curves for the nomogram,
   risk score, AFP, tumor thrombus, and grade in predicting 1-, 3-, and
   5-year OS in the HBV-related HCC

Discussion

   HCC remains a significant global public health burden. While
   demographic and clinical factors (age, sex, BMI, TNM stage, vascular
   invasion, tumor thrombus, AFP) offer some prognostic value [[92]16],
   the marked heterogeneity of HCC underscores the need for novel
   biomarkers and more precise prognostic models. For example, Li et al.
   [[93]17] utilized portal venous and hepatic arterial coefficients to
   predict overall and recurrence-free survival following hepatectomy in
   HCC patients. Similarly, Xiang et al. [[94]18] conducted an extensive
   review on the application of the PANoptosis signature (a novel
   inflammatory cell death pathway) in forecasting therapeutic response
   and prognosis in HCC. The integration of molecular signatures with
   clinical parameters appears promising as a strategy to address the
   heterogeneity of HCC, potentially offering enhanced predictive
   capabilities compared to individual biomarkers [[95]19]. Furthermore,
   the analysis of liver cancer biopsy specimens through
   immunohistochemistry (IHC) has revealed aberrant protein expression
   profiles as promising tools for cancer prognosis [[96]20–[97]22].

   This study identified a novel 4-protein signature (SMC2, SMC4, UBE2C,
   UHRF1) for prognostic prediction in HBV-related HCC. Importantly, high
   expression of these proteins constitutes significant risk factors
   associated with poor patient outcomes. Furthermore, a nomogram
   integrating this signature with traditional prognostic factors proved
   highly effective for predicting both short-term (1-year, 3-year) and
   long-term (5-year) survival. Collectively, these findings indicate that
   a risk model based on this 4-protein signature is a valuable prognostic
   tool for survival assessment in HBV-related HCC.

   This study demonstrated that the expression levels of the proteins
   SMC2, SMC4, UBE2C, and UHRF1 significantly increase as the degree of
   tumor differentiation decreases. However, among the 159 cases of
   HBV-related HCC detailed in Supplementary Table 1, only one patient was
   classified as Grade I. Consequently, it is currently not feasible to
   ascertain whether these four proteins can serve as early diagnostic
   markers for HBV-related HCC. Furthermore, Supplementary Table 1 lacks
   data on the TNM staging of HBV-related HCC patients. Future research
   should involve classifying HBV-related HCC patients according to their
   clinical data, including tumor size, number, vascular invasion, liver
   function, and the patient’s physical condition, using the BCLC
   (Barcelona Clinic Liver Cancer) and CNLC (Chinese Liver Cancer) staging
   systems. By analyzing the correlation between the expression levels of
   SMC2, SMC4, UBE2C, and UHRF1 proteins and the staging of HBV-related
   HCC patients, we can provide scientific evidence to determine whether
   these proteins can be utilized as early diagnostic markers for
   HBV-related HCC and to elucidate their association with the occurrence
   and progression of HBV-related HCC.

   Tumor thrombus represents a prevalent complication associated with
   cancers, significantly impacting both the survival duration and quality
   of life of affected patients [[98]23]. When tumor thrombus develops in
   critical tissues and organs, it poses a heightened risk to patient
   survival, potentially leading to rapid mortality [[99]23–[100]25].
   Consequently, the accurate diagnosis of tumor thrombus in patients with
   liver cancer, along with precise localization, is essential for
   ensuring safe and effective treatment. For example, a systematic review
   and single-arm meta-analysis by Wang et al. [[101]26] demonstrated that
   local ± systemic therapy offers superior long-term OS and manageable
   complications for HCC with hepatic vein tumor thrombus (HVTT), inferior
   vena cava tumor thrombus (IVCTT), and/or right atrium tumor thrombus
   (RATT), compared to surgery with adjuvant therapy or surgery alone.
   Globally, treatment strategies for HCC with portal vein tumor thrombus
   (PVTT) exhibit considerable variation. Although guidelines recommend
   systemic therapies such as sorafenib and immune checkpoint inhibitors,
   their efficacy remains limited in cases involving PVTT [[102]27].
   Emerging evidence increasingly supports the use of aggressive local
   treatments, including liver resection, transplantation, radiation, and
   transarterial chemoembolization (TACE), for select patient populations
   [[103]27]. Notably, this study identified elevated expression levels of
   SMC2, SMC4, and UHRF1 in HBV-related HCC patients with tumor thrombus
   compared to those without. Moreover, multivariate Cox regression
   analysis demonstrated that tumor thrombus serves as an independent
   prognostic factor for patients with HBV-related HCC. Consistent with
   our findings, Yang et al. [[104]28] demonstrated that CCL20 is
   overexpressed in HCC with bile duct tumor thrombus and is inversely
   correlated with surgical outcomes. Furthermore, Li et al. [[105]29]
   reported higher expression of CXCR4 in PVTT tissue relative to HCC
   tissue; however, the invasion capacity of PVTT cells was significantly
   reduced (p < 0.05) following the downregulation of CXCR4 expression
   [[106]29]. These findings highlight the need to investigate if SMC2,
   SMC4, and UHRF1 overexpression increases HCC invasiveness, potentially
   serving as diagnostic markers and therapeutic targets for HBV-related
   HCC with tumor thrombus.

   SMC proteins, highly conserved ATPases, are essential for chromosome
   structure, dynamics, and processes like mitosis, gene regulation, and
   DNA repair, and are implicated in tumorigenesis [[107]30, [108]31]. The
   SMC2-SMC4 heterodimer plays a crucial role in chromosome dynamics
   [[109]32, [110]33]. SMC4 is overexpressed in lung adenocarcinoma and
   functions as an independent prognostic marker; its knockdown
   significantly inhibits A549 cell proliferation and invasion [[111]34].
   Consistent with Zhou Bo et al.., we observed significant SMC2 and SMC4
   protein overexpression in HCC tissues by IHC (Fig. [112]7B-C, E-F).
   Zhou et al.. also demonstrated SMC4’s association with tumor size,
   differentiation, TNM stage, and vascular invasion in HCC, and its role
   in modulating the miR-219-SMC4-JAK2/Stat3(Tyr705) signaling axis to
   promote hepatoma cell proliferation and invasion [[113]35]. Aberrant
   SMC4 expression is also reported in prostate cancer [[114]36], and it
   serves as a potential indicator for breast cancer chemotherapy response
   [[115]37] and prognosis [[116]38].

   UBE2C, encoding an ubiquitin-conjugating enzyme, plays a key role in
   mitotic cyclin degradation and cell cycle progression, contributing to
   cancer development [[117]39, [118]40]. UBE2C overexpression is observed
   in 27 tumor types, including breast, colorectal, lung, and liver
   cancer, and is strongly associated with poor patient prognosis
   [[119]41, [120]42]. Its dysregulation promotes chromosomal instability
   and facilitates progression in gastric cancer, NSCLC, and melanoma
   [[121]43–[122]45]. Wei et al.. confirmed UBE2C overexpression in HCC
   tissues and cell lines correlating with poor prognosis [[123]46]; our
   study corroborates UBE2C mRNA and protein overexpression in HCC (Figs.
   [124]6C and [125]7K). These findings underscore UBE2C’s significance as
   a critical oncogene in HCC and other cancers.

   UHRF1, a key epigenetic regulator, maintains promoter methylation of
   tumor suppressor genes, suppressing their expression and influencing
   tumorigenesis, treatment response, and prognosis [[126]47, [127]48].
   Dysregulated UHRF1 expression, characterized by elevated levels
   throughout the cell cycle, significantly contributes to tumor
   development, metastasis, and recurrence [[128]49]. IHC and microarray
   analyses confirm UHRF1 upregulation in lung, gastric, prostate, and
   colorectal cancers [[129]50–[130]52]. Consistent with this, our IHC
   analysis showed marked UHRF1 overexpression in the nuclei of hepatoma
   cells compared to adjacent normal liver (Supplementary Fig. 1). UHRF1
   thus represents a potential pan-cancer biomarker; reliable detection
   methods could benefit cancer diagnosis and prognosis evaluation.

   To our knowledge, prognostic models and nomograms based on protein
   expression signatures in HBV-related HCC are rarely reported. Our novel
   model demonstrated efficacy in predicting HBV-related HCC outcomes. The
   roles of SMC2, SMC4, UBE2C, and UHRF1 in tumorigenesis are
   well-supported in the literature, as summarized above. While we have
   not deeply explored the mechanisms here, potential regulatory pathways
   can be inferred from their known functions. Further experiments are
   warranted to investigate their specific roles in HBV-related HCC
   pathogenesis.

Limitations of the study

   Our study is subject to several limitations. Firstly, the CPTAC cohort
   predominantly consists of White or Asian patients, necessitating
   caution when generalizing findings to other ethnic groups. Secondly, it
   is imperative to candidly acknowledge that the lack of Bootstrap
   internal validation to evaluate the variability of the optimal cutoff
   point for the risk score constitutes a limitation of this exploratory
   study. Thirdly, it is crucial to perform external validation of the
   signature and nomogram in an independent cohort. Fourthly, the current
   analyses are primarily descriptive. Thus, functional experiments are
   required to elucidate the underlying mechanisms of these four proteins
   in HBV-related HCC. For instance, techniques like CRISPR knockout can
   assess the impact on cell line sensitivity to treatments like sorafenib
   and PD-1 inhibitors. Additionally, experiments involving nude mouse
   tumor formation or tail vein injection could be conducted to
   investigate the effects of gene silencing or overexpression of these
   proteins on the initiation, progression, and metastasis of HCC. For
   clinical application, future research should use fluorescence in situ
   hybridization (FISH) or IHC to measure protein levels in biopsies, and
   enzyme-linked immunosorbent assay (ELISA) or chemiluminescence
   immunoassay (CLIA) for blood samples. Additionally, findings should be
   validated with a larger, independent cohort of HCC patients, including
   both HBV-related and non-HBV-related cases.

Conclusions

   Our study establishes a novel 4-protein signature and nomogram for
   predicting OS in HBV-induced HCC, offering potential guidance for
   personalized clinical treatment decisions.

Supplementary Information

   Below is the link to the electronic supplementary material.
   [131]Supplementary Material 1.^ (21.9KB, xls)
   [132]Supplementary Material 2.^ (34.3KB, xls)
   [133]Supplementary Material 3.^ (28.4MB, ppt)

Acknowledgements