Baishideng Publishing Group Inc, 7041 Koll Center Parkway, Suite 160, Pleasanton, CA 94566, USA
Share the Article
Liao MJ, Zhou Y, Wang ZL, Wang XX, Ai X, Song LN, Fei R, Cong X, Liu F, Zhu PP, Chen HS, Liao WJ, Rao HY. Development and validation of a prognostic prediction model for alpha-fetoprotein-negative hepatocellular carcinoma with microvascular invasion after surgery. World J Gastroenterol 2027; 33(2): 123092 [DOI: 10.3748/wjg.123092]
World J Gastroenterol. Jan 14, 2027; 33(2): 123092 Published online Jan 14, 2027. doi: 10.3748/wjg.123092
Development and validation of a prognostic prediction model for alpha-fetoprotein-negative hepatocellular carcinoma with microvascular invasion after surgery
Min-Jun Liao, Zi-Long Wang, Xiao-Xiao Wang, Xin Ai, Lin-Nan Song, Ran Fei, Xu Cong, Feng Liu, Hong-Song Chen, Hui-Ying Rao, Peking University Hepatology Institute, Infectious Disease and Hepatology Center of Peking University People’s Hospital, Beijing Key Laboratory of Innovative Drug and Device Translation in Endocrine and Metabolic Diseases, Peking University People’s Hospital, Beijing 100044, China
Yu Zhou, Department of Hepatopancreatobiliary Surgery, Taikang Tongji (Wuhan) Hospital, Wuhan 430000, Hubei Province, China
Yu Zhou, Wei-Jia Liao, Laboratory of Hepatobiliary and Pancreatic Surgery, The Affiliated Hospital of Guilin Medical University, Guilin 541001, Guangxi Zhuang Autonomous Region, China
Peng-Peng Zhu, National-Local Joint Engineering Research Center of Biodiagnosis and Biotherapy, The Second Affiliated Hospital, Xi’an Jiaotong University, Xi’an 710000, Shaanxi Province, China
Co-corresponding authors: Wei-Jia Liao and Hui-Ying Rao.
Author contributions: Liao MJ and Zhou Y contributed equally to this manuscript as co-first authors; Rao HY and Liao WJ put forward and designed the study, and they contributed equally to this manuscript as co-corresponding authors; Liao MJ, Zhou Y, and Zhu PP wrote this article and participated in the collection and analysis of data; Wang ZL, Wang XX, Ai X, Song LN, Fei R, Cong X, Chen HS, and Liu F participated in the analysis of data and helped revised the manuscript. All authors contributed to the article and approved the final manuscript.
AI contribution statement: The authors declare that generative AI was not used in the creation of this manuscript.
Supported by National Key R&D Program of China, No. 2022YFA1303804; Dongguan Social Development Technology Project, No. 20221800906062; Taikang Health Investment Youth Medical Research Initiation Fund, No. 2025030; and the Research Fund of the Second Affiliated Hospital, Xi’an Jiaotong University, No. YJ(YB)202518.
Institutional review board statement: The study protocol was conducted in accordance with the principles of the Declaration of Helsinki, and approved by the Ethics Committees of Peking University People’s Hospital (approval No. 2024PHB061-001) and the Affiliated Hospital of Guilin Medical University (approval No. 2021WJWZC14), and written informed consents were obtained from all individual participants.
Informed consent statement: All study participants, or their legal guardian, provided informed written consent prior to study enrollment.
Conflict-of-interest statement: All the authors report no relevant conflicts of interest for this article.
STROBE statement: The authors have read the STROBE Statement-checklist of items, and the manuscript was prepared and revised according to the STROBE Statement-checklist of items.
Data sharing statement: If reasonably necessary, the datasets used and analyzed in the current study are available from the corresponding author upon reasonable request.
Corresponding author: Hui-Ying Rao, Professor, Peking University Hepatology Institute, Infectious Disease and Hepatology Center of Peking University People’s Hospital, Beijing Key Laboratory of Innovative Drug and Device Translation in Endocrine and Metabolic Diseases, Peking University People’s Hospital, No. 133 Fuchengmennei Street, Xicheng District, Beijing 100044, China. rao.huiying@163.com
Received: May 18, 2026 Revised: July 16, 2026 Accepted: August 26, 2026 Published online: January 14, 2027 Processing time: 180 Days and 3.5 Hours
Abstract
BACKGROUND
Microvascular invasion (MVI) is widely recognized as a key factor associated with poor prognosis in hepatocellular carcinoma. Patients with alpha-fetoprotein (AFP)-negative hepatocellular carcinoma (AFP-NHCC) who also have MVI represent a distinct subgroup in which postoperative survival is particularly difficult to predict.
AIM
To develop and validate a postoperative prognostic scoring system for predicting survival in AFP-NHCC with MVI patients.
METHODS
This study included clinical data from patients with AFP-NHCC and MVI who underwent surgical treatment. Independent predictors of overall survival (OS) were identified using Cox proportional hazards regression analysis. A prognostic scoring model was constructed based on these independent factors. The model performance was assessed using the concordance index, calibration curves, and decision curve analysis.
RESULTS
A total of 251 patients were assigned to training (n = 135) and validation (n = 116) cohorts. Univariate and multivariate Cox proportional hazards regression analyses identified tumor burden score, geometric mean of gamma-glutamyl transpeptidase and white blood cell, and neutrophil-to-lymphocyte ratio as independent prognostic factors for OS in the training cohort. A TGN scoring system based on tumor burden score, gamma-glutamyl transpeptidase and white blood cell and neutrophil-to-lymphocyte ratio was subsequently constructed and exhibited good discriminative performance in both the training and validation cohorts, with area under the curve values of 0.816 (95% confidence interval: 0.745-0.887) and 0.807 (95% confidence interval: 0.727-0.885), respectively. Patients were stratified into three risk groups based on TGN scores. Kaplan-Meier analyses revealed significant differences in OS and disease-free survival among the three risk groups.
CONCLUSION
The TGN model provides an effective tool for individualized postoperative risk stratification in patients with AFP-NHCC and MVI, and may be used to guide postoperative management and surveillance strategies.
Core Tip: Alpha-fetoprotein-negative hepatocellular carcinoma (AFP-NHCC) combined with microvascular invasion (MVI) constitutes a distinct subgroup, biomarkers or prognostic indicators for AFP-NHCC patients with MVI are not investigated and defined yet. TGN model provides an effective tool for risk stratification in patients with AFP-NHCC and MVI, and may be used to guide postoperative management and surveillance strategies.
Citation: Liao MJ, Zhou Y, Wang ZL, Wang XX, Ai X, Song LN, Fei R, Cong X, Liu F, Zhu PP, Chen HS, Liao WJ, Rao HY. Development and validation of a prognostic prediction model for alpha-fetoprotein-negative hepatocellular carcinoma with microvascular invasion after surgery. World J Gastroenterol 2027; 33(2): 123092
Hepatocellular carcinoma (HCC) represents a major global health burden. According to 2022 estimates, it ranked the sixth most common malignancy and the third leading cause of cancer-related death worldwide[1], with its incidence continuing to rise[2-4]. Owing to its insidious nature, fewer than 30% of patients with HCC are eligible for radical surgical resection at the time of initial diagnosis. Although radical resection for early-stage HCC is associated with a favorable prognosis and a significant reduction in mortality[5], postoperative recurrence and metastasis rates remain high within five years, ranging from 50% to 70%[6,7], severely impairing patient outcomes and long-term survival. As a result, the overall prognosis of HCC remains poor, with a five-year survival rate below 20%[8,9], highlighting the severity of this malignancy.
Serum alpha-fetoprotein (AFP) is a commonly used and important biomarker for the diagnosis, treatment monitoring, and prognosis assessment of HCC. However, approximately 30%-40% of patients with HCC present with normal AFP levels (≤ 20 ng/mL)[10-12] and are classified as AFP-negative HCC (AFP-NHCC), and AFP ≤ 20 ng/mL as the negative diagnostic threshold for HCC is the most commonly used threshold in clinical practice. These patients have specific tumor biological characteristics and clinical outcomes, and their normal AFP levels make early diagnosis and risk assessment more challenging. Therefore, understanding the relationship between AFP levels and key pathological features of HCC is essential for improving diagnostic and therapeutic strategies in AFP-NHCC. Elevated AFP levels correlate closely with several aggressive tumor features, including tumor burden score (TBS), microvascular invasion (MVI), tumor aggressiveness, and poor prognosis[13,14]. Among these, MVI reflects enhanced local invasion and distant metastatic capacity, which is associated with poor prognosis in HCC. Lauwers et al[15] reported an MVI incidence of 51.3% in patients undergoing surgical resection for HCC, whereas Kılcı et al[16] reported an MVI incidence of 28.6% in patients with AFP-NHCC. Studies have shown that MVI incidence is higher when AFP levels are elevated. Importantly, even in AFP-NHCC patients without elevated AFP, MVI remains an adverse pathological feature associated with postoperative recurrence and poor survival, highlighting the need for AFP-independent prognostic stratification. Collectively, the mechanisms promoting MVI in HCC result from multiple factors, rather than from AFP or any single factor.
Researchers have found several useful biomarkers for HCC diagnosis and monitoring, such as a detection kit consisting of a seven-microRNA panel[17], the GALAD model[18], and the GAAD model[19]. Predictors for HCC with MVI have also been reported[20,21]. However, there remains a lack of biomarkers or prognostic indicators for AFP-NHCC patients with MVI. Increasing evidence suggests that the clinical treatment outcomes of HCC depend on several key factors, such as TBS, baseline liver function, and patient physiological status. Among these, TBS is a promising non-invasive marker for evaluating disease status as well as predicting survival[13,22,23]. In addition, the geometric mean of gamma-glutamyl transpeptidase and white blood cells (GMWG) score[24] and the neutrophil-to-lymphocyte ratio (NLR) model[25-28] are significant prognostic tools for HCC. Based on these findings, we developed the TGN model using TBS, GMWG, and the NLR, which may serve as a reliable prognostic indicator in AFP-NHCC patients with MVI, thereby facilitating personalized management.
MATERIALS AND METHODS
Patients
This study included HCC patients who underwent radical surgical resection at the Peking University People’s Hospital and the Affiliated Hospital of Guilin Medical University from May 2015 to December 2020. The inclusion criteria were as follows: (1) Age > 18 years; (2) AFP-NHCC (AFP ≤ 20 ng/mL) with pathological evidence of MVI; (3) Eligibility for radical surgical treatment; and (4) Child-Pugh class A or B liver function. Patients were excluded if they met any of the following criteria: (1) Diagnosis of cholangiocarcinoma or non-HCC malignancy; (2) Infectious diseases, immune disorders, or hematologic diseases, or use of medications affecting blood parameters; (3) Prior antitumor therapy; and (4) Incomplete clinicopathological data or loss to follow-up. The patient flowchart was shown in Figure 1. In accordance with requests from the ethics committee[29], the study protocol was approved by the Institutional Review Boards of participating organizations and conducted in accordance with the Declaration of Helsinki. All participants provided written informed consent before surgical intervention.
Figure 1 The flowchart of hepatocellular carcinoma patient enrollment.
HCC: Hepatocellular carcinoma; CCC: Cholangiocarcinoma.
Clinical data collection
The diagnosis of HCC was histologically confirmed. The clinical variables include age, gender, family history of cancer, liver cirrhosis, tumor size, tumor number, TBS, Child-Pugh class, Barcelona Clinic Liver Cancer (BCLC) stage, and early recurrence (defined as recurrence within 2 years after surgery). Laboratory data were derived from preoperative examinations, including hepatitis B surface antigen (HBsAg), white blood cells (WBC), neutrophils, lymphocytes, platelets, fibrinogen (FIB), and liver function biomarkers such as albumin, globulin, total bilirubin, direct bilirubin, alanine aminotransferase, aspartate aminotransferase (AST), and gamma-glutamyl transferase (GGT) (Table 1).
Table 1 Comparison of clinicopathological characteristics of two groups’ patients, mean ± SD/ median (interquartile range).
Patients underwent follow-up at two months postoperatively, then every three months with imaging surveillance (ultrasound, computed tomography or magnetic resonance imaging as needed) and tumor markers (AFP, des-gamma-carboxyprothrombin), extending to 3-6 months after two years. The mean follow-up period was 47.0 months (median: 41.0; range: 2.0-130.0). Disease-free survival (DFS) was defined as time from surgery to recurrence, metastasis, death, or last follow-up. Overall survival (OS) was calculated from the date of surgery until death or final follow-up. Follow-up ended in January 2026, and data were right censored.
Variables and definitions
The NLR was defined as the ratio of neutrophil count to lymphocyte count, while the GMWG score was calculated according to our previous report, GMWG2 = (GGT × WBC)2[24]. Tumor size was measured in centimeters (cm), TBS2 = (maximum tumor diameter)2 + (number of tumors)2[22,23]. Receiver operating characteristic curve analysis was used to determine the optimal cutoff values for NLR and GMWG in predicting OS. The following clinicopathological variables were categorized: Gender (male vs female), age (> 55 years vs ≤ 55 years), alcohol abuse (yes vs no), smoking (yes vs no), HBsAg (positive vs negative), liver cirrhosis (yes vs no), tumor number (multiple vs single), TBS (> 8.0 vs ≤ 8.0), BCLC (B-C vs 0-A), FIB (> 3.0 g/L vs ≤ 3.0 g/L), and AST (> 40 U/L vs ≤ 40 U/L). Subsequently, the prognostic value of these variables was evaluated.
Statistical analysis
Statistical analysis was performed using SPSS 24.0 and R software (version 4.3.1). Data are presented as mean ± SD for normally distributed data or median with interquartile range for non-normally distributed data. 95% confidence intervals (CIs) were provided when applicable. Continuous variables were compared using t-test, and categorical variables were analyzed using the χ2 test or Fisher’s exact test. Univariate Cox proportional hazards regression was performed to identify variables associated with OS, and multivariate Cox regression analysis was subsequently conducted to determine independent prognostic factors. The TGN model was constructed using coefficients derived from the multivariate analysis. The model performance was assessed using the concordance index (C-index). Survival curves were estimated using the Kaplan-Meier method, and compared using the log-rank test. Calibration was evaluated using 3-, 5-, and 8-year OS rates. Clinical utility was assessed by decision curve analysis. P < 0.05 was considered statistically significant.
RESULTS
Clinical and pathological characteristics of HCC patients
The analysis was performed on 251 patients meeting the inclusion criteria, including 135 in the training cohort from Peking University People’s Hospital and 116 in the validation cohort from Affiliated Hospital of Guilin Medical University. The clinicopathological characteristics of the patients were summarized in Table 1. All variables were comparable between the two cohorts. The mean age of HCC patients was 53 years and 55 years, respectively, with the majority of patients being male (91.9% and 87.9%, respectively). The majority of patients were HBsAg-positive (78.5% and 75.9%, respectively) and had cirrhosis (88.9% and 86.2%, respectively). The GMWG values were 24.20 ± 9.47 and 23.06 ± 9.74, respectively. The NLR values were 2.62 ± 1.32 and 2.52 ± 1.56, respectively. The TBS values were 7.74 ± 4.48 and 7.43 ± 4.57, respectively. The median OS was 49.0 months and 45.0 months, respectively.
Optimal cutoff values for NLR and GMWG
According to receiver operating characteristic curve analysis in the training cohort, the optimal cutoff values for preoperative NLR and GMWG were 2.3 and 22.7, respectively, with relatively high specificity and sensitivity. For NLR, the area under the curve (AUC) was 0.692 (95%CI: 0.600-0.784), with a sensitivity of 68.4% (95%CI: 56.9%-78.4%) and specificity of 70.2% (95%CI: 56.6%-81.6%). For GMWG, the AUC was 0.712 (95%CI: 0.629-0.801), with sensitivity of 60.3% (95%CI: 49.3%-70.8%) and specificity of 70.9% (95%CI: 57.1%-82.4%) (Figure 2). Comparatively, NLR and GMWG had higher sensitivity and specificity at the optimal cutoff values. The cutoff value of TBS determined as 8.0 was based on relevant studies[22,23].
Figure 2 The receiver operating characteristic curves for determining cutoff values of neutrophil-to-lymphocyte ratio and gamma-glutamyl transpeptidase and white blood cells for predicting overall survival in the training cohort.
A: Neutrophil-to-lymphocyte ratio; B: Gamma-glutamyl transpeptidase and white blood cells. ROC: Receiver operating characteristic; NLR: Neutrophil-to-lymphocyte ratio; GMWG: Gamma-glutamyl transpeptidase and white blood cells; AUC: Area under the curve; CI: Confidence interval.
Prognostic factors and model construction
In the training cohort, univariate Cox regression analyses revealed that TBS [hazard ratio (HR): 3.037; 95%CI: 1.962-4.539; P < 0.001], BCLC (HR: 2.015; 95%CI: 1.201-3.380; P = 0.008), FIB (HR: 1.591; 95%CI: 1.025-2.468; P = 0.038), GMWG (HR: 2.634; 95%CI: 1.677-4.138; P < 0.001), and NLR (HR: 2.844; 95%CI: 1.747-4.230; P < 0.001) were significantly associated with OS. However, variables including gender, age, alcohol use, smoking, HBsAg, liver cirrhosis, tumor number, and AST level showed no significant association with OS (P > 0.05) (Table 2). Multivariate Cox regression analyses revealed that TBS (HR: 2.724; 95%CI: 1.654-4.486; P < 0.001), GMWG (HR: 1.797; 95%CI: 1.112-2.906; P = 0.017), and NLR (HR: 2.170; 95%CI: 1.296-3.634; P = 0.001) were identified as independent predictors of OS (Table 3). Based on the multivariate Cox regression results, a postoperative survival prognostic model for AFP-NHCC patients with MVI was established and named the TGN score (TBS, GMWG, NLR). The formula for calculation of this score is detailed in Table 3.
Table 2 Univariate Cox regression analysis of overall survival in the training cohort.
The TGN scoring system demonstrated good discriminatory performance in both the training and validation cohorts. As shown in Figure 2, the AUC value of the TGN score in the training cohort was 0.816 (95%CI: 0.745-0.887), which was higher than those of its individual components as well as the BCLC stage (TBS: 0.715, GMWG: 0.693, NLR: 0.686, BCLC: 0.652). Moreover, the TGN scoring system demonstrated comparable accuracy in the validation cohort (AUC = 0.807, 95%CI: 0.727-0.885), which was higher than those of its individual components and BCLC stage (TBS: 0.694, GMWG: 0.710, NLR: 0.666, BCLC: 0.668) (Figure 3). These findings were supported by C-index analysis, with C-index values of 0.786 (95%CI: 0.671-0.894) and 0.778 (95%CI: 0.676-0.881) for the training and validation cohorts, respectively. Calibration accuracy was assessed by plotting calibration plots based on 3-, 5-, and 8-year OS rates, which demonstrated high concordance between predicted and observed survival. In addition, decision curve analysis confirmed its clinical utility (Figure 3 and Supplementary Figure 1).
Figure 3 Prediction performance of the TGN scoring model based on receiver operating characteristic curve analysis and calibration curve analysis.
Area under the curve values of the TGN scoring model, tumor burden score, gamma-glutamyl transpeptidase and white blood cells, neutrophil-to-lymphocyte ratio, and Barcelona Clinic Liver Cancer in the training and validation cohorts. Calibration curves demonstrated high concordance between TGN predicted and observed overall survival at 3, 5, and 8 years in the training and validation cohorts. A: Area under the curve of training cohort; B: Area under the curve of validation cohort; C: Calibration curve of training cohort; D: Calibration curve of validation cohort. ROC: Receiver operating characteristic; TBS: Tumor burden score; GMWG: Gamma-glutamyl transpeptidase and white blood cells; NLR: Neutrophil-to-lymphocyte ratio; BCLC: Barcelona Clinic Liver Cancer; AUC: Area under the curve.
Survival analysis based on the TGN scoring system
Based on the TGN score, patients were divided into three risk groups, low (TGN < 0.6), intermediate (0.6 ≤ TGN < 1.7), and high (TGN ≥ 1.7). In the training cohort, the low TGN group demonstrated more favorable survival outcomes, with a median OS not evaluable (NE) months (95%CI: NE-NE) and a median DFS of 66.0 months (95%CI: NE-NE), significantly higher than the intermediate-risk group (OS: 67.0 months, 95%CI: NE-NE; DFS: 35.0 months, 95%CI: 17.4-52.6) and the high-risk group (OS: 24.0 months, 95%CI: 17.4-30.5; DFS: 14.0 months, 95%CI: 9.9-18.1). This stratification pattern was confirmed in the validation cohort, with median OS for the low-, intermediate-, and high-TGN groups of NE months (95%CI: NE-NE), 40.0 months (95%CI: 22.0-57.9), and 30.0 months (95%CI: 20.6-39.3) (log-rank P < 0.001), respectively; and median DFS was NE months (95%CI: NE-NE), 23.0 months (95%CI: 9.2-36.8), and 13.0 months (95%CI: 9.1-17.0) (log-rank P < 0.001) (Figure 4).
Figure 4 Prediction of overall survival and disease-free survival based on the TGN scoring model.
Kaplan-Meier curves illustrated overall survival and disease-free survival differences stratified by the TGN model, which classified patients into low-, intermediate-, and high-risk groups. Median survival and log-rank P values between groups were shown. A and B: Training cohort; C and D: Validation cohort. NE: Not evaluable; M: Months.
Prognostic performance of TGN scoring system in subgroup analyses
In the training cohort, the TGN score demonstrated favorable predictive performance across various subgroups, including tumor number (single or multiple), BCLC stage (0-A or B-C), and early recurrence (no or yes) (Figure 5). Moreover, similar prognostic effects were observed in the validation cohort (Supplementary Figure 2).
Figure 5 Subgroup analyses of survival based on clinical characteristics in the training cohort.
The TGN score demonstrated excellent discriminative performance across all subgroups. A: Single tumor [log-rank, overall survival (OS): P < 0.001, disease-free survival (DFS): P = 0.001]; B: Multiple tumors (log-rank, OS: P < 0.001, DFS: P < 0.001); C: Barcelona Clinic Liver Cancer stage (0-A) (log-rank, OS: P < 0.001, DFS: P < 0.001); D: Barcelona Clinic Liver Cancer stage (B-C) (log-rank, OS: P < 0.001, DFS: P = 0.001); E: Without early recurrence (log-rank, OS: P < 0.001, DFS: P < 0.001); F: Early recurrence (log-rank, OS: P < 0.001, DFS: P < 0.001). TN: Tumor number; OS: Overall survival; DFS: Disease-free survival; BCLC: Barcelona Clinic Liver Cancer.
DISCUSSION
Current prediction models typically consider AFP as a key variable, with high AFP level serving as an independent risk factor or biomarker for MVI[13,30,31] and poor prognosis[14,32] in HCC. However, these models are not applicable for AFP-negative cases. MVI refers to clusters of cancer cells within vasculature lined by endothelial cells[33], which exhibits high specificity but low sensitivity on imaging findings. Currently, the presence of MVI can only be confirmed through histopathological examination of resected surgical specimens. Thus, preoperative identification of MVI in HCC patients is of great value for formulating appropriate treatment strategies. In recent years, increasing research has focused on the pretreatment prediction of MVI, and preoperative prediction based on artificial intelligence (including radiomics and deep learning models) may be more clinically meaningful[34-36].
AFP-NHCC represents a distinct subgroup characterized by unique biological behavior and limited biomarker availability. Given the lack of prognostic predictors for AFP-NHCC with MVI, individualized postoperative management is severely restricted. In this study, we developed a novel prognostic model for AFP-NHCC combined with MVI, the TGN scoring system, consisting of TBS, GMWG, and NLR. TBS, which is based on tumor size and tumor number, has been validated as a progressive and prognostic parameter in HCC[13,22,23,37]. GMWG, a recently identified prognostic marker for HCC, consists of GGT and WBC[23]. Specifically, GGT is a key enzyme closely associated with oxidative stress and hepatic metabolism, with elevated level promoting liver injury and tumor cell invasiveness[38]. Additionally, WBC reflects systemic inflammatory status. The combination of GGT and WBC into a geometric mean (GMWG) effectively enhances prognostic prediction accuracy for HCC patients. NLR is calculated as the absolute neutrophil count divided by the absolute lymphocyte count, representing the dual roles of pro-inflammatory (neutrophils) and anti-tumor immune responses (lymphocytes). On one hand, neutrophils promote tumor proliferation and angiogenesis through secretion of cytokines and vascular endothelial growth factor; on the other hand, lymphocytes play a critical role in tumor immune surveillance[39]. Previous studies have confirmed higher survival rates among HCC patients with low postoperative NLR[24-27]. Comparatively, a higher NLR reflects the enhanced inflammatory response and inhibition of anti-tumor immunity; and a higher GMWG reflects the comprehensive condition of hepatic metabolic damage and systemic inflammation. These two variables have been proved to be associated with the prognosis of HCC.
The TGN scoring model serves as a promising prognostic tool for AFP-NHCC patients with MVI. The model uses readily available variables (tumor size and number, GGT, WBC count), all of which can be easily obtained from peripheral blood and routine clinical data, ensuring favorable reproducibility and cost-effectiveness. While single biomarkers offer limited prognostic discrimination, our findings demonstrate that an integrated multi-parameter predictive system offers superior clinical utility. In the training cohort, the TGN score achieved a significantly higher AUC value than any single indicator (TBS, GMWG or NLR) as well as the clinically used BCLC staging system, and comparable performance was also observed in the validation cohort. The TGN scoring system exhibited good discriminative stratification performance for all enrolled patients and key subgroups (stratified by tumor number, BCLC stage, or early recurrence), offering novel strategies for clinical management.
In this study, univariate analysis identified BCLC stage and FIB as predictors of poor OS; however, neither remained an independent predictor in multivariate analysis, partly explained by their lower prognostic risk coefficients compared with the other three indicators. Nevertheless, this does not undermine their prognostic relevance. The BCLC staging system is widely used in HCC clinical practice and research, in terms of treatment decision and prognostic assessment[40,41]. Similarly, FIB, a key coagulation protein, typically indicates hypercoagulable and inflammatory status when elevated, which may influence postoperative recovery and prognosis. In addition, FIB may play a key role in tumor initiation and progression in AFP-NHCC[42]. Compared with BCLC staging, the TGN model offers superior predictive performance and enables personalized risk stratification, which can guide adjuvant therapy and surveillance intensity. Low-risk patients may follow the standard schedule for postoperative follow-up to avoid excessive treatment; intermediate-risk patients may require closer monitoring; while high-risk patients may undergo intensified follow-up, multidisciplinary assessment, adjuvant therapy, or be candidates of clinical trials after radical resection. Moreover, the TGN model integrates tumor burden (TBS), hepatic metabolism, systemic inflammatory condition (GMWG), and immune-inflammatory balance (NLR), which may improve the accuracy of prognosis assessment and is more suitable for predicting the prognosis of AFP-negative patients with MVI, compared with other established hepatic function or inflammation-related scoring models, such as albumin-bilirubin score[43], systemic inflammation score[44] and FIB-4[45], as the application of model based on AFP-positive HCC is limited in AFP-NHCC subgroup. The combined application of TBS, GMWG and NLR has a synergistic effect mainly manifested in: Increasing the degree of liver damage, promoting systemic inflammation and tumor inflammatory responses, inhibiting anti-tumor immune responses, cooperatively shaping an environment conducive to tumor proliferation, invasion and MVI occurrence of HCC, and enhancing the malignancy and poor prognosis of HCC. Overall, the TGN model demonstrates good discriminative and calibration capabilities, highlighting the value of integrating tumor burden with systemic inflammatory markers for predicting postoperative survival in AFP-NHCC patients with MVI.
This study has limitations due to its design and small sample size, which may have resulted in selection bias. In addition, binary classification of continuous variables (TBS, GMWG, and NLR) may lead to information loss, therefore future studies should evaluate the variables as continuous variables and adopt multiple cutoff values when constructing prediction models to achieve more precise risk calibration. Moreover, surgical-related indicators such as surgical approach, operative time, resection extent, surgical margins and intraoperative blood transfusion were not included, however, the lack of these surgical variables may reduce the credibility of the model. In this study, the majority of patients had hepatitis B virus-related HCC, and antiviral therapy may have influence on prognosis; therefore, future prospective studies should verify the model’s applicability in hepatitis C virus and non-viral HCC populations, and meanwhile further studies should incorporate antiviral therapy and the dynamic changes of hepatitis B virus DNA as candidate prognostic variables. And future studies with larger sample sizes should collect related variables in a standardized manner, and more comprehensive indicators as well as prospective designs are required to validate these findings.
CONCLUSION
The variables of the TGN scoring model (based on TBS, GMWG and NLR) can be easily obtained from peripheral blood tests and daily data, with good repeatability and low cost. The TGN model provides an effective tool for individualized postoperative risk stratification of AFP-NHCC patients with MVI. Subsequent prospective studies can optimize model performance and expand its application scope to further promote precision treatment for HCC.
Zeng H, Cao M, Xia C, Wang D, Chen K, Zhu Z, Fu R, Zhang S, Zhou J, Wang H, Qi X, Dai S, Chen Y, Sun Z, Ding H, Li Q, Zhao H, Zhang X, Morze J, Ji JS, Sun F, Yu X, Qu C, Chen W. Performance and effectiveness of hepatocellular carcinoma screening in individuals with HBsAg seropositivity in China: a multicenter prospective study.Nat Cancer. 2023;4:1382-1394.
[RCA] [PubMed] [DOI] [Full Text][Cited by in Crossref: 11][Cited by in RCA: 29][Article Influence: 9.7][Reference Citation Analysis (0)]
Hassanipour S, Vali M, Gaffari-Fam S, Nikbakht HA, Abdzadeh E, Joukar F, Pourshams A, Shafaghi A, Malakoutikhah M, Arab-Zozani M, Salehiniya H, Mansour-Ghanaei F. The survival rate of hepatocellular carcinoma in Asian countries: a systematic review and meta-analysis.EXCLI J. 2020;19:108-130.
[RCA] [PubMed] [DOI] [Full Text] [Full Text (PDF)][Cited by in RCA: 41][Reference Citation Analysis (0)]
Chang YS, Tsai MJ, Tsai CJ, Wang CC, Lin CC, Yen YH, Hung CH, Kuo YH, Huang DS, Tai WC, Hu TH, Tsai MC. A new model based on preoperative AFP, albumin, and tumor burden score for predicting microvascular invasion in early-stage HCC.Am J Cancer Res. 2024;14:4979-4988.
[RCA] [PubMed] [DOI] [Full Text][Cited by in RCA: 5][Reference Citation Analysis (1)]
Xia D, Bai W, Wang Q, Chung JW, Adhoute X, Kloeckner R, Zhang H, Zeng Y, Sripongpun P, Nie C, Kim SU, Huang M, Hu W, Ding X, Yin G, Li H, Zhao H, Bronowicki JP, Li J, Li J, Zhu X, Wu J, Zhang C, Gong W, Li Z, Lin Z, Xu T, Yin T, Anty R, Song J, Shi H, Shao G, Ren W, Zhang Y, Yang S, Zheng Y, Xu J, Wang W, Zhu X, Fu Y, Liu C, Kaewdech A, Ding R, Zheng J, Liu S, Yu H, Zheng L, You N, Fan W, Zhang S, Feng L, Wang G, Zhang P, Li X, Chen J, Zhang F, Shao W, Zhou W, Zeng H, Cao G, Huang W, Jiang W, Zhang W, Li L, Feng A, Wang E, Wang Z, Han D, Lv Y, Sun J, Ren B, Xia L, Li X, Yuan J, Wang Z, Luo B, Li K, Guo W, Yin Z, Zhao Y, Xia J, Fan D, Wu K, Bettinger D, Vogel A, Han G; China HCC-TACE study group. Tumor burden with AFP improves survival prediction for TACE-treated patients with HCC: An international observational study(☆).JHEP Rep. 2025;7:101216.
[RCA] [PubMed] [DOI] [Full Text] [Full Text (PDF)][Cited by in Crossref: 4][Cited by in RCA: 8][Article Influence: 8.0][Reference Citation Analysis (0)]
Lauwers GY, Terris B, Balis UJ, Batts KP, Regimbeau JM, Chang Y, Graeme-Cook F, Yamabe H, Ikai I, Cleary KR, Fujita S, Flejou JF, Zukerberg LR, Nagorney DM, Belghiti J, Yamaoka Y, Vauthey JN; International Cooperative Study Group on Hepatocellular Carcinoma. Prognostic histologic indicators of curatively resected hepatocellular carcinomas: a multi-institutional analysis of 425 patients with definition of a histologic prognostic index.Am J Surg Pathol. 2002;26:25-34.
[RCA] [PubMed] [DOI] [Full Text][Cited by in Crossref: 176][Cited by in RCA: 155][Article Influence: 6.5][Reference Citation Analysis (3)]
Zhu Y, Feng B, Cai W, Wang B, Meng X, Wang S, Ma X, Zhao X. Prediction of Microvascular Invasion in Solitary AFP-Negative Hepatocellular Carcinoma ≤ 5 cm Using a Combination of Imaging Features and Quantitative Dual-Layer Spectral-Detector CT Parameters.Acad Radiol. 2023;30 Suppl 1:S104-S116.
[RCA] [PubMed] [DOI] [Full Text][Cited by in Crossref: 1][Cited by in RCA: 24][Article Influence: 8.0][Reference Citation Analysis (0)]
Vitale A, Lai Q, Farinati F, Bucci L, Giannini EG, Napoli L, Ciccarese F, Rapaccini GL, Di Marco M, Caturelli E, Zoli M, Borzio F, Sacco R, Cabibbo G, Virdone R, Marra F, Felder M, Morisco F, Benvegnù L, Gasbarrini A, Svegliati-Baroni G, Foschi FG, Missale G, Masotto A, Nardone G, Colecchia A, Bernardi M, Trevisani F, Pawlik TM; Italian Liver Cancer (ITA. LI.CA) group. Utility of Tumor Burden Score to Stratify Prognosis of Patients with Hepatocellular Cancer: Results of 4759 Cases from ITA.LI.CA Study Group.J Gastrointest Surg. 2018;22:859-871.
[RCA] [PubMed] [DOI] [Full Text][Cited by in Crossref: 45][Cited by in RCA: 47][Article Influence: 5.9][Reference Citation Analysis (3)]
Yang X, Chen B, Wang Y, Wang Y, Long J, Zhang N, Xue J, Xun Z, Zhang L, Cheng J, Lei J, Sun H, Li Y, Lin J, Xie F, Wang D, Pan J, Hu K, Guan M, Huo L, Shi J, Yu L, Zhou L, Zhou J, Lu Z, Yang X, Mao Y, Sang X, Lu Y, Zhao H. Real-world efficacy and prognostic factors of lenvatinib plus PD-1 inhibitors in 378 unresectable hepatocellular carcinoma patients.Hepatol Int. 2023;17:709-719.
[RCA] [PubMed] [DOI] [Full Text] [Full Text (PDF)][Cited by in Crossref: 27][Cited by in RCA: 57][Article Influence: 19.0][Reference Citation Analysis (0)]
Chen J, Fang Y, Tang Z, Dong E, Gao J, Zhu G, Kwangwari P, Feng S, Qu W, Wu X, Mao S, Zhao Q, Wang Y, Yang R, Guan Z, Chu T, Bu Y, Zhou J, Fan J, Fu X, Liu W, Ding Z, Shi Y. Predictive value of neutrophil-to-lymphocyte ratio in recurrent HCC after repeat hepatectomy or salvage liver transplantation.Hepatol Int. 2025;19:856-865.
[RCA] [PubMed] [DOI] [Full Text][Cited by in RCA: 4][Reference Citation Analysis (0)]
Matono T, Tada T, Kumada T, Hiraoka A, Hirooka M, Kariyama K, Tani J, Atsukawa M, Takaguchi K, Itobayashi E, Fukunishi S, Nishikawa H, Tanaka K, Tsuji K, Ishikawa T, Tajiri K, Koshiyama Y, Toyoda H, Ogawa C, Hatanaka T, Kakizaki S, Kawata K, Ohama H, Tada F, Nouso K, Morishita A, Tsutsui A, Nagano T, Itokawa N, Okubo T, Arai T, Nishimura T, Imai M, Kosaka H, Naganuma A, Aoki T, Kuroda H, Yata Y, Tamai H, Matsuura T, Komatsu S, Ueda Y, Nakamura Y, Yoshida O, Nakamura S, Enomoto H, Kaibori M, Fukumoto T, Hiasa Y, Kudo M; Real‐life Practice Experts for HCC (RELPEC) Study Group and the Hepatocellular Carcinoma Experts from 48 clinics in Japan (HCC 48) Group. Neutrophil-Lymphocyte Ratio Predicts Overall Survival in Patients With HCC Treated With Durvalumab Plus Tremelimumab.Hepatol Res. 2025;55:1285-1295.
[RCA] [PubMed] [DOI] [Full Text][Cited by in Crossref: 2][Cited by in RCA: 7][Article Influence: 7.0][Reference Citation Analysis (0)]
Rocco A, Sgamato C, Pelizzaro F, Simeon V, Coccoli P, Compare D, Pinto E, Palano G, Foschi FG, Raimondo G, Missale G, Svegliati-Baroni G, Trevisani F, Caturelli E, Brunetto MR, Vidili G, Masotto A, Magalotti D, Campani C, Gasbarrini A, Azzaroli F, Rapaccini GL, Stefanini B, Sacco R, Mega A, Giannini EG, Cabibbo G, Di Marco M, Guarino M, Chiodini P, Farinati F, Nardone G. Systemic inflammatory response markers improve the discrimination for prognostic model in hepatocellular carcinoma.Hepatol Int. 2025;19:915-928.
[RCA] [PubMed] [DOI] [Full Text] [Full Text (PDF)][Cited by in Crossref: 4][Cited by in RCA: 18][Article Influence: 18.0][Reference Citation Analysis (0)]
Johnson PJ, Berhane S, Kagebayashi C, Satomura S, Teng M, Reeves HL, O'Beirne J, Fox R, Skowronska A, Palmer D, Yeo W, Mo F, Lai P, Iñarrairaegui M, Chan SL, Sangro B, Miksad R, Tada T, Kumada T, Toyoda H. Assessment of liver function in patients with hepatocellular carcinoma: a new evidence-based approach-the ALBI grade.J Clin Oncol. 2015;33:550-558.
[RCA] [PubMed] [DOI] [Full Text][Cited by in Crossref: 2296][Cited by in RCA: 2348][Article Influence: 213.5][Reference Citation Analysis (7)]
Inokuchi S, Itoh S, Yoshizumi T, Morinaga A, Toshima T, Takeishi K, Nagao Y, Harada N, Ikegami T, Shimokawa M, Mori M. Prognostic significance of systemic inflammation score in patients who undergo hepatic resection for hepatocellular carcinoma.Langenbecks Arch Surg. 2021;406:773-779.
[RCA] [PubMed] [DOI] [Full Text][Cited by in RCA: 9][Reference Citation Analysis (1)]
Nahon P, Lusivika-Nzinga C, Merle P, Zoulim F, Decaens T, Ganne-Carrié N, Pageaux GP, Leroy V, Alric L, Bronowicki JP, Bourlière M, Gournay J, Tran A, Pol S, Mathurin P, Loustaud-Ratti V, Métivier S, De Ledinghen V, Abergel A, Thabut D, D'Alteroche L, Bouattour M, Asselah T, Ouzan D, Cales P, Chazouillères O, Gelu-Simeon M, Roulot D, Boursier J, Cagnot C, Tamazirt S, Pascale A, Nilusmas S, Lewin M, Ziol M, Carrat F, Duclos-Vallée JC; ANRS CO12 CirVir, CO22 Hepather, and LICAVIR groups. Value of non-invasive test dynamics in guiding HCC surveillance decisions after HCV cure in patients with cirrhosis.J Hepatol. 2025;83:701-711.
[RCA] [PubMed] [DOI] [Full Text][Cited by in Crossref: 1][Cited by in RCA: 6][Article Influence: 6.0][Reference Citation Analysis (1)]
Footnotes
Peer review: Externally peer reviewed.
Peer-review model: Single blind
Specialty type: Gastroenterology and hepatology
Country of origin: China
Peer-review report’s classification
Scientific quality: Grade A, Grade B, Grade B
Novelty: Grade B, Grade B, Grade B
Creativity or innovation: Grade B, Grade B, Grade C
Scientific significance: Grade B, Grade B, Grade B
P-Reviewer: Augustin G, Associate Professor, MD, PhD, Croatia; Yan X, PhD, China S-Editor: Hu XY L-Editor: A P-Editor: Lei YY
