Clinical model for predicting overall survival outcomes in individuals with hepatocellular carcinoma: a retrospective cohort analysis

Abstract

Background

The prognostic factors for survival outcomes in patients with hepatocellular carcinoma (HCC) are not well defined. This study aimed to identify the prognostic factors for HCC and to construct a predictive nomogram model.

Methods

A total 165 patients with HCC were identified between 25 January 2010 and 10 November 2021. Independent prognostic factors were identified using univariable and multivariable Cox regression analyses. A nomogram was constructed to predict the patient survival rate. The concordance index (C-index), area under the curve (AUC), and calibration curves were used to assess the predictive accuracy and discrimination of the model. Decision curve analysis was used to confirm the clinical utility of the nomogram.

Results

A total of 165 patients were randomly selected retrospectively. Univariable and multivariable analyses revealed that body mass index, albumin, carbohydrate antigen 19-9 (CA19-9), tumor size, and tumor size, lymph node, metastasis (TNM) stage were independent factors for predicting patient survival. We constructed a 1-, 3-, and 5-year survival rate prediction clinical model by using these independent prognostic factors, which yielded C-indexes of 0.838, 0.798 and 0.725, respectively. On the basis of the AUCs and calibration curve and decision curve analyses, we concluded that the prognostic model for HCC exhibited excellent performance.

Conclusions

The clinical model demonstrated good calibration, discrimination, clinical utility, and practical decision-making effects for the outcomes of patients with HCC. These findings may help oncologists and surgeons make better clinical decisions.

Highlight box.

Key findings

• Nomograms are convenient and accurate methods for predicting survival outcomes in recent years.

• This model provides a valuable predictive tool for these patients.

• This model could serve as a reference for future clinical decision-making and therapeutic strategizing.

What is known and what is new?

• The prognostic prediction model clearly classified patients who underwent surgical resection according to their outcomes.

• Nomogram for predicting hepatocellular carcinoma (HCC) survival was constructed, which contributes to providing excellent guidance for patients with resect tumors.

What is the implication, and what should change now?

• A clinically practical nomogram for predicting survival in patients with HCC was successfully constructed.

• This model provides a valuable predictive tool for the patients with advance HCC.

• The model can be used to make clinical judgments in estimating patient outcomes and in early treatment strategies.

Introduction

Globally, hepatocellular carcinoma (HCC) is the sixth most common cause of cancer-related death, with a 5-year overall survival (OS) rate of less than 20% (1). In recent decades, however, liver transplantation and surgical resection have been used globally to treat only small numbers of HCCs in patients with cirrhosis (2,3). Even for patients with successful resection, the 5-year survival rate is less than 20%, and the median OS is only 2 years, which is significantly worse than the desired results (4,5). Furthermore, the number of elderly patients with HCC has increased with increasing life expectancy, and HCC is expected to become even more common in elderly patients over time (6). In clinical practice, HCC staging is important in determining the need for resection therapy and for predicting patient outcomes. All international HCC staging systems are limited by preoperative variables. However, postoperative pathological indicators and early recurrence patterns of tumors are also closely related to patient outcomes (4-7). Furthermore, the tumor size, lymph node, metastasis (TNM) staging system has several limitations because of the influence of various factors, including age, sex, tumor biomarkers, and therapy-related variables, on patient outcomes.

Nomograms are convenient and accurate methods for predicting survival outcomes in recent years (8). In the past decade, nomograms have been widely accepted by some investigators for predicting the prognosis of various tumors (9,10).

Clinical nomograms have been used worldwide to predict the outcomes of many cancers. In this study, clinical investigations were performed, and a nomogram for predicting HCC survival was constructed, which contributes to providing excellent guidance for patients with resettable tumors. We present this article in accordance with the TRIPOD reporting checklist (available at https://jgo.amegroups.com/article/view/10.21037/jgo-2025-645/rc).

Methods

Patient selection

A total of 165 patients underwent surgery at Sir Run Run Shaw Hospital of Zhejiang University. All consecutive patients who underwent surgical resection for HCC at the Zhejiang University School of Medicine Sir Run Shaw Hospital between 25 January 2010 and 10 November 2021 were included in this study. The study was conducted in accordance with the Declaration of Helsinki and its subsequent amendments. The study was approved by the Institutional Review Board of Zhejiang University School of Medicine Sir Run Shaw Hospital (No. 2024-114). Informed consent was waived because of the retrospective study design and collection of readily available clinical data.

The inclusion criteria were as follows: (I) primary HCC; (II) HCC diagnosed pathologically after surgical removal; (III) hepatectomy margins confirmed by a pathologist; (IV) number of lymph node metastases or extrahepatic metastases; and (V) postoperative liver cancer diagnosed as HCC according to pathological examination of the liver tissue with abdominal magnetic resonance imaging (MRI) and/or computed tomography (CT) by the consensus agreement of experienced radiologists.

The exclusion criteria were as follows: (I) other malignant liver tumors confirmed by pathology; (II) suspected or confirmed distant metastasis before surgery; and (III) incomplete clinical data or missing data.

Diagnosis and postoperative care

All conservative patients underwent routine laboratory tests, such as liver function tests, hepatitis virus infection [hepatitis B virus (HBV), hepatocellular carcinoma (HCV)] tests, and tumor marker tests, as well as ultrasound (US), MRI and CT, prior to the operations, all of which were performed by experienced surgeons. The surgical methods included open surgery and minimally invasive surgery according to the location and distribution of the tumor.

Margin resection was confirmed by a pathologist to be free of tumor invasion.

Pathologic specimens were taken from the tumor edge and from inside the tumor.

Pathologists confirmed the diagnosis of the patients via hematoxylin and eosin (HE) staining and immunohistochemistry of the surgically resected specimens. The tumors were divided into stages I, II, III, and IV.

Postoperative care, including liver function and renal function tests, complete blood count tests, serum alpha-fetoprotein (AFP) level tests and CT, was performed routinely on days 1, 3, and 5 after surgical resection. Postoperative outcomes were assessed according to the Clavien-Dindo classification (11,12).

Follow-up procedure and laboratory tests

A regular follow-up strategy was employed for all patients (both inpatients and outpatients) after discharge: all patients were examined at 3- to 6-month intervals for the first 2 years, at 6- to 12-month intervals for the next 3 years, and then at 1-year intervals until 10 years. The follow-up evaluation routinely included a physical examination, blood examination, chest radiography, measurement of tumor marker (AFP) levels, and abdominal US, CT and/or MRI scans. Recurrence was diagnosed on the basis of the results of physical examination and diagnostic imaging. The date of recurrence was defined as the date of histological proof of identification on the basis of clinical medical imaging findings by a physician.

All patients were followed up until 10 November 2021.

Clinicopathological variables

Demographic and clinical data, including age at diagnosis, sex, body mass index (BMI), total bilirubin, direct bilirubin, albumin, AFP, carbohydrate antigen 19-9 (CA19-9), cancer antigen 125 (CA125), carcinoembryonic antigen (CEA), ferritin, tumor number, tumor location, tumor size, TNM stage, major metastases, lymph node metastases, vital status, cirrhosis, resection margin and survival time, were collected. Imaging results from contrast-enhanced MRI, contrast-enhanced CT, and US were analyzed to confirm the presence of cirrhosis, and data from perioperative tests, including complete blood count, liver function and tumor markers, were recorded.

Statistical analysis

Descriptive statistics of all the data from the included HCC patients were calculated. To explore the associations between variables and survival outcomes, both univariable and multivariable Cox regression analyses were performed, and relevant packages were used for correlation analysis. Least absolute shrinkage and selection operator (LASSO) regression was used to identify six independent factors of OS. LASSO regression decreases the assessed variance in the collected variables and offers an interpretable final model that may be more precise. We constructed a nomogram based on the prognostic factors identified from the univariable analysis to predict the 1-, 3-, and 5-year survival rates by the leave-one-out method. The performance of the model was measured with the concordance index (C-index), calibration curves, receiver operating characteristic (ROC) curves, area under the curve and decision curve analysis (DCA). ROC curves were used to assess the discriminant ability of the predictive model. Furthermore, survival curves were constructed using the Kaplan-Meier method and compared using the log-rank test. All the statistical analyses were performed using R language version 4.4.1. P<0.05 was considered to indicate statistical significance.

Results

Clinicopathological characteristics of the patients

A total of 165 patients who underwent liver resection from 25 January 2010 to 10 November 2021 was enrolled in this study. The baseline characteristics of the patients and their tumors are summarized in Table 1. The median age of the included patients was 64.7 years (range, 39–90 years), as shown in Figure 1; in total, 143 (86.7%) were male, and 22 (13.3%) were female. The majority of patients (n=130; 78.8%) had HBV infection, while only 2 (1.2%) had HCV infection. There were 69 cases of recurrence among all patients (41.8%). Postoperative pathology revealed that 123 patients had liver cirrhosis (74.5%). The median tumor diameter was 4.49 cm (range, 1.0–16 cm), and multiple tumors were present in 7 patients (4.2%). For the majority of patients, standardized routine testing revealed a median AFP concentration of 1,369 ng/mL, a median CA19-9 concentration of 40.1 ng/mL, and a median CA125 concentration of 20.1 ng/mL. Only 5 patients (3.0%) had stage IV disease, 15 (9.0%) had stage III disease, 28 (17.0%) had stage II disease, and 117 (71.0%) had stage I disease; the TNM stage was the most common for all patients. A total of 12 (7.3%) patients had portal vein thrombus. The OS rate of the present study was 85 months (95% confidence interval (CI): 64–102). Recurrence-free survival (RFS) for all patients was 65 months (95% CI: 35–86). HCC was the most common histological type and was detected in all patients (100%).

Table 1. Characteristics of patients with table HCC.

Variables	All patients (n=165)
Age (years)	64.7 [12.1]
Sex
Female	22 (13.3)
Male	143 (86.7)
BMI (kg/m2)	23.4 [3.22]
Alcohol use
No	99 (60.0)
Yes	66 (40.0)
Smoking history
No	91 (55.2)
Yes	74 (44.8)
Cirrhosis
No	42 (25.5)
Yes	123 (74.5)
Hypertension
No	119 (72.1)
Yes	46 (27.9)
Diabetes
No	151 (91.5)
Yes	14 (8.5)
HBc-IgM
Negative	163 (98.8)
Positive	2 (1.2)
HBc-IgG
Negative	8 (4.85)
Positive	157 (95.2)
HBV
Negative	35 (21.2)
Positive	130 (78.8)
HCV
Negative	163 (98.8)
Positive	2 (1.2)
ALB (g/L)	40.3 [6.97]
TBIL (μmoI/L)	21.0 [26.1]
Direct bilirubin (μmoI/L)	13.3 [9.78]
CA19-9 (U/mL)	40.1 [106]
CA125 (U/mL)	20.1 [38.7]
AFP (ng/mL)	1,369 [9,949]
CEA (ng/mL)	2.94 [1.73]
Ferritin (ng/mL)	351 [319]
TNM stage (%)	1.44 [0.78]
Tumor size (cm)	4.49 [2.99]
Tumor number	1.05 [0.24]
Recurrence
No	96 (58.2)
Yes	69 (41.8)
Hospital stays (days)	16.5
Resection margin	(8.5–25.0)
Negative	162 (98.2)
Positive	3 (1.8)
Type of resection
Left	43 (26.1)
Right	115 (69.7)
Bilobe	7 (4.2)
Lymph node metastasis
No	161 (97.6)
Yes	4 (2.4)
Metastasis
No	161 (97.6)
Yes	4 (2.42)
Portal vein thrombus
No	153 (92.7)
Yes	12 (7.3)

Data are presented as median [interquartile range] or n (%). AFP, alpha-fetoprotein; ALB, albumin; BMI, body mass index; CA125, cancer antigen 125; CA19-9, carbohydrate antigen 19-9; CEA, carcinoembryonic antigen; HBc-IgG, anti-hepatitis B core immunoglobulin G; HBc-IgM, anti-hepatitis B core immunoglobulin M; HBV, hepatitis B virus; HCC, hepatocellular carcinoma; HCV, hepatitis C virus; TBIL, total bilirubin; TNM, tumor size, lymph node, metastasis.

Figure 1.

Histogram depicting the age distribution of the patients who underwent surgery and were included in our analysis. The mean age was 64.7 years.

Identifying independent prognostic factors

After all the variables were included in the univariable and multivariable Cox regression analyses, several statistically significant independent prognostic factors were identified. The results of the univariable and multivariable analyses, as presented in Table 2, revealed the following significant independent prognostic factors for OS: age above 70 years (HR 0.689; 95% CI: 0.425–1.118), indicating that increasing age is associated with poor OS; BMI (HR 0.912; 95% CI: 0.845–0.985], suggesting that BMI is a significant independent prognostic factor for OS; anti-hepatitis B core immunoglobulin M (HBc-IgM) (HR 22.366; 95% CI: 4.913–101.821); TNM stage, grade I, II, III, or IV (HR 1.862; 95% CI: 1.449–2.394), indicating that higher TNM stage is linked to worse OS; metastasis, presence of metastasis (HR 3.998; 95% CI: 1.25–12.785), indicating a better OS; tumor size larger than 5 cm (HR 1.119; 95% CI: 1.052–1.191), which is associated with a better prognosis for OS; surgical resection margin, positivity (HR 2.534; 95% CI: 1.321–4.859), indicating a better OS; recurrence (HR 2.677; 95% CI: 1.642–4.366), indicating a OS; postresection tumor markers, AFP (HR 2.677; 95% CI: 1.642–4.366), CA19-9 (HR 1.004; 95% CI: 1.002–1.005), CA125 (HR 1.005; 95% CI: 1.001–1.008), and serum albumin (HR 0.933; 95% CI: 0.892–0.976); and portal vein thrombus (HR 4.184; 95% CI: 1.792–9.771). Multivariate Cox regression analysis revealed that BMI (HR 4.184; 95% CI: 1.792–9.771), smoking history (HR 4.184; 95% CI: 1.792–9.771), CA19-9 (HR 1.003; 95% CI: 1.001–1.005), tumor size (HR 1.08; 95% CI: 1.009–1.156), and TNM stage (HR 1.613; 95% CI: 1.216–2.139) were independent prognostic factors for patient OS. The results of the univariable and multivariable Cox regression analyses are shown in Table 2.

Table 2. Univariable and multivariable Cox regression analyses of predictive factors for OS.

Variables	Categories	Univariable analyzed			Multivariable analyzed
		HR (95% CI)	P value		HR (95% CI)	P value
Age (years)	≤70 vs. >70	0.689 (0.425–1.118)	0.13		–	–
Sex	Male vs. female	1.688 (0.73–3.903)	0.22		–	–
BMI (kg/m2)	≤24.0 vs. >24.0	0.912 (0.845–0.985)	0.01		0.907 (0.836–0.985)	0.02
Smoking	Absent vs. present	0.727 (0.452–1.171)	0.19		0.949 (0.905–0.996)	0.03
Drinking	Absent vs. present	1.132 (0.7–1.831)	0.61		–	–
HBV	Absent vs. present	1.074 (0.596–1.937)	0.81		–	–
HCV	Absent vs. present	4.534 (0.052–392.0368)	0.50		–	–
HBc-IgM	Absent vs. present	22.366 (4.913–101.821)	0.051		–	–
HBc-IgG	Absent vs. present	1.231 (0.386–3.925)	0.72		–	–
Hypertension	Absent vs. present	1.511 (0.915–2.495)	0.10		–	–
Diabetes	Absent vs. present	1.3 (0.594–2.843)	0.51		–	–
Cirrhosis	Absent vs. present	0.992 (0.579–1.7)	0.97		–	–
Albumin (g/L)	≤40 vs. >40	0.933 (0.892–0.976)	0.002		–	–
Total bilirubin (μmoI/L)	≤20 vs. >20	1.001 (0.99–1.012)	0.91		–	–
Direct bilirubin (μmoI/L)	≤20 vs. >20	0.986 (0.95–1.024)	0.47		–	–
AFP (ng/mL)	≤250 vs. >250	1 (1–1)	0.052		–	–
CA19-9 (U/mL)	≤25 vs. >25	1.004 (1.002–1.005)	0.051		1.003 (1.001–1.005)	0.003
CA125 (U/mL)	≤25 vs. >25	1.005 (1.001–1.008)	0.009		–	–
Tumor size (cm)	≤5.0 vs. >5.0	1.119 (1.052–1.191)	0.051		1.08 (1.009–1.156)	0.02
Tumor number	Absent vs. present	2.01 (0.935–4.323)	0.07		–	–
TNM stage	I vs. II, III, IV	1.862 (1.449–2.394)	0.051		1.613 (1.216–2.139)	0.001
Resection margin	Negative vs. positive	2.534 (1.321–4.859)	0.051		–	–
Recurrence	Absent vs. present	2.677 (1.642–4.366)	0.051		–	–
Lymph node metastasis	Absent vs. present	2.034 (0.497–8.329)	0.32		–	–
Metastasis	Absent vs. present	3.998 (1.25–12.785)	0.01		–	–
Portal vein thrombus	Absent vs. present	4.184 (1.792–9.771)	0.001		–	–

Significant if P<0.05. AFP, alpha-fetoprotein; BMI, body mass index; CA-125, cancer antigen 125; CA19-9, carbohydrate antigen 19-9; CI, confidence interval; HBc-IgG, anti-hepatitis B core immunoglobulin G; HBc-IgM, anti-hepatitis B core immunoglobulin M; HBV, hepatitis B virus; HCV, hepatitis C virus; HR, hazard ratio; OS, overall survival; TNM, tumor size, lymph node, metastasis.

Univariable Cox proportional regression OS analysis of 26 potential factors revealed five independent factors (Figure 2A), including BMI, albumin level, CA19-9 level, tumor size and TNM stage, most of which were significantly different between the groups (P<0.05). Multivariate Cox proportional regression analysis revealed five potential factors in the forest curve (Figure 2B), namely, BMI, albumin level, CA19-9 level, tumor size and TNM stage, most of which were significantly different between the groups (P<0.05).

Figure 2.

Forest plots for univariable and multivariable Cox regression analyses indicating predictive factors for both overall survival (A) and recurrence-free survival (B). BMI, body mass index; CA19-9, carbohydrate antigen 19-9; CI, confidence interval; HR, hazard ratio; TNM, tumor size, lymph node, metastasis.

Selection of predictors and model construction

LASSO regression was used to identify the best predictive factors. In combination with Cox analysis (Figure 3A,3B), six risk predictors, namely, BMI, serum albumin, CA19-9, tumor size, and TNM stage, were selected as risk predictors for the development of the survival nomogram in the present study.

Figure 3.

Feature selection based on LASSO regression. (A) LASSO coefficient profiles of the candidate variables, plotted against log (λ).Curves showing coefficient shrinkage as the penalty increases; (B) cross-validation curve for LASSO model selection used to determine the optimal LESSO penalty value of parameter λ. LASSO, least absolute shrinkage and selection operator.

Establishment and verification of the survival nomogram

Univariable and multivariable analyses were conducted with a Cox proportional hazards model, and we identified five independent prognostic factors of OS, namely, BMI, albumin, CA19-9, tumor size and TNM stage. A nomogram constructed from these factors was used to predict the 1-, 3-, and 5-year OS rates by summing the scores of each independent factor (Figure 4A). The nomogram offers a personalized approach for prognostication by considering various clinicopathological characteristics of individuals (Figure 4B).

Figure 4.

Model nomogram and prediction of survival for patients with HCC. (A) Nomogram constructed on the basis of independent factors influencing survival in patients with diffuse HCC. (B) Nomogram generated according to the results of multivariable logistic regression (*, P<0.05; **, P<0.01; ***, P<0.001). BMI, body mass index; CA19-9, carbohydrate antigen 19-9; HCC, hepatocellular carcinoma; sur.prob, survival probability; TNM, tumor size, lymph node, metastasis.

Here, we constructed calibration charts for the probability of survival at 1, 3 and 5 years on the basis of the independent factors and the actual survival of patients. The results revealed that the survival predictions of the nomogram were very close to the actual survival observations, except for the 5-year curve, whose values differed more from the actual values than did the curves predicted at 1 and 3 years (Figure 5). The calibration plots indicated good model calibration.

Figure 5.

Calibration curves of the prognostic nomogram for 1-year (A), 3-year (B), and 5-year (C) survival rates.

ROC analysis was used to assess the accuracy of the nomogram, and the AUCs for predicting the 1-, 3- and 5-year OS rates in the entire study were 0.838, 0.798 and 0.725, respectively (Figure 6). The results indicated good efficacy in predicting patient outcomes.

Figure 6.

ROC curve of the prognostic nomogram for 1-, 3-, and 5-year survival rates. AUC, area under the ROC curve; ROC, receiver operating characteristic.

Furthermore, the clinical application value of the nomogram was determined with DCA by calculating the net benefits at different risk threshold probabilities. The area under the DCA curve for predicting the 1-, 3-, and 5-year OS rates reflects the utility of the nomogram in guiding clinical decision-making (Figure 7A-7C).

Figure 7.

Overall decision curves for 1-year (A), 3-year (B), and 5-year (C) survival rates.

Additionally, the area under the C-index curve of the risk score demonstrated the consistently high accuracy of the prognostic model (Figure 8A), confirming the reliable clinical predictive power of the nomogram.

Figure 8.

Nomogram building curves analysis on the predictive model for HCC patients. (A) C-index curves of the risk score and clinical features. (B) Decision curve analysis of the clinical benefits of the model for predicting 1-, 3- and 5-year survival rates. AUC, area under the receiver operating characteristic curve; BMI, body mass index; CA19-9, carbohydrate antigen 19-9; DCA, decision curve analysis; HCC, hepatocellular carcinoma; TNM, tumor size, lymph node, metastasis.

Finally, a comparison of the DCA results of the nomogram at 1, 3, and 5 years is shown in Figure 8B.

Assessment of the risk stratification ability of the nomogram model

The above analyses demonstrate the better predictive effect of the survival nomogram model. We calculated total scores for each patient with the nomogram for risk stratification; specifically, the patients were classified into two groups (low risk and high risk) on the basis of the median risk score. The results of the Kaplan-Meier survival analysis are shown in Figure 9. These data indicate that the nomogram can discriminate between the two risk groups well, with a p value indicating statistical significance. Overall, these results demonstrate the good predictive value of the risk stratification system for patients with HCC.

Figure 9.

Evaluation of model prediction performance. (A) Kaplan-Meier analysis of the OS rate for the high- and low-risk groups for all patients. (B) RFS rate according to the prognostic prediction model in the present study. OS, overall survival; RFS, recurrence-free survival.

Discussion

This was a population-based, single-center retrospective study with the goal of identifying prognostic factors that could aid in risk stratification to predict the survival probability of individuals with HCC. This model could serve as a reference for future clinical decision-making and therapeutic strategizing.

In the present study, univariable and multivariable analyses revealed prognostic factors for patients with HCC. Individual survival nomograms were then constructed on the basis of these factors. Previous studies have reported prognostic factors for survival in patients with HCC. By employing LASSO Cox regression, we identified five variables (BMI, albumin, CA19-9, tumor size, and TNM stage) that served as independent predictors of survival. Some of the variables we reported are consistent with those of previous studies. In addition, we developed a nomogram for survival that included five of these prognostic factors (BMI, albumin, CA19-9, tumor size, and TNM stage). Our nomogram performed well in predicting survival at 1 and 3 years in patients with HCC, but this performance could be further improved by integrating more types of data and increasing the number of patients. Clinical models provide personalized predictions by incorporating easily accessible variables that are routinely obtained in clinical practice.

Previous studies have confirmed that nomograms are widely used to predict prognosis in many cancers (13,14). Nomograms have several benefits, such as the ability to increase the accuracy of prediction, and could help clinicians make better decisions and guide the selection of better treatment options. Recently, machine learning analysis was used to create a nomogram for predicting the outcomes of different patient cohorts (15).

Moreover, the C-index, ROC curves, and calibration plots were used to assess the performance of the nomograms, and the results were satisfactory. DCA revealed that the nomograms offered net clinical benefits in the prediction of survival in patients with HCC. These results demonstrated the promising accuracy and optimal consistency of the nomogram.

Furthermore, the risk stratification performed on the basis of the nomogram demonstrated strong discriminatory ability, as shown by the Kaplan‒Meier curves.

The results suggested that BMI is an independent prognostic factor for patients with HCC. A previous study has revealed that a BMI of 21.5 to 23.1 kg/m² is beneficial for patients with HCC, potentially leading to a longer survival time (16).

In the present study, tumor size was a significant factor for the outcomes of patients with solitary HCC and was shown to play a role in the survival of patients with HCC.

However, in an international study, a large tumor size was the key factor related to early HCC recurrence after resection surgery, according to the results of a preoperative model for RFS in the present study (low risk: 2-year RFS 64.8%; intermediate risk: 2-year RFS 42.5%; and high risk: 2-year RFS 20.7%) (17).

Furthermore, the 8^th edition of the American Joint Committee on Cancer (AJCC) TNM guidelines was used to determine tumor stage, lymph node status, and metastasis (18). According to the results of the present study, the TNM staging method is the most common prognostic evaluation technique but may restrict the accuracy of survival prediction. In the past decade, two clinical prediction models that offer superior predictive power over the AJCC TNM staging system have expanded the options for the accurate prediction of tumor prognosis (19,20).

Serum CA19-9 was initially identified as a tumor-associated antigen; in this study, it was identified as a significant predictive factor and was used to construct the nomogram model. These results are also consistent with clinical observations and support the clinical significance of CA19-9 as a prognostic marker in the diagnosis of HCC.

Albumin has been recognized as an independent prognostic factor for survival in patients with HCC. In this study, albumin was identified as a significant prognostic factor and was used to develop a comprehensive nomogram model.

This study has several limitations. First, this was a retrospective study conducted at a single institution. A prospective study with a greater number of patients is needed to validate our results. Second, the prognostic prediction model clearly classified patients who underwent surgical resection according to their outcomes. Third, the outcomes of patients with HCC are far from satisfactory even after curative surgery; multidisciplinary treatment may be important for improving the outcomes of patients with HCC.

Conclusions

A clinically practical nomogram for predicting survival in patients with HCC was successfully constructed. This model provides a valuable predictive tool for these patients. Owing to its reliability, the model can be used to make clinical judgments in estimating patient outcomes and in early treatment strategies. Future studies could incorporate molecular data and explore the role of emerging treatments to refine prognostic models and improve patient outcome rates, as these factors are increasingly recognized as important in the era of precision medicine.

Supplementary

The article’s supplementary files as

Ethical Statement: The authors are accountable for all aspects of the work in ensuring that questions related to the accuracy or integrity of any part of the work are appropriately investigated and resolved. The study was conducted in accordance with the Declaration of Helsinki and its subsequent amendments. The study was approved by the Institutional Review Board of Zhejiang University School of Medicine Sir Run Shaw Hospital (No. 2024-114). Informed consent was waived because of the retrospective study design and collection of readily available clinical data.

Data Sharing Statement

Available at https://jgo.amegroups.com/article/view/10.21037/jgo-2025-645/dss

jgo-17-01-24-dss.pdf ^{(87.5KB, pdf)}

DOI: 10.21037/jgo-2025-645