Cong R, Chen B, Cheng SN, Zhai YR, Wu F, Feng B, Wang LY, Chen ZW, Zhu YJ, Ma XH, Zhao XM. Early recurrence risk stratification of hepatocellular carcinoma receiving preoperative radiotherapy: Incorporating clinical and imaging features across treatment phases. World J Gastroenterol 2026; 32(23): 118308 [DOI: 10.3748/wjg.v32.i23.118308]
Corresponding Author of This Article
Xin-Ming Zhao, Professor, Department of Diagnostic Radiology, National Cancer Center/National Clinical Research Center for Cancer/Cancer Hospital, Chinese Academy of Medical Sciences and Peking Union Medical College, No. 17 Panjiayuan Nanli, Chaoyang District, Beijing 100021, China. zhaoxinming@cicams.ac.cn
Research Domain of This Article
Gastroenterology & Hepatology
Article-Type of This Article
research-article
Open-Access Policy of This Article
This article is an open-access article which was selected by an in-house editor and fully peer-reviewed by external reviewers. It is distributed in accordance with the Creative Commons Attribution Non Commercial (CC BY-NC 4.0) license, which permits others to distribute, remix, adapt, build upon this work non-commercially, and license their derivative works on different terms, provided the original work is properly cited and the use is non-commercial. See: http://creativecommons.org/licenses/by-nc/4.0/
World J Gastroenterol. Jun 21, 2026; 32(23): 118308 Published online Jun 21, 2026. doi: 10.3748/wjg.v32.i23.118308
Early recurrence risk stratification of hepatocellular carcinoma receiving preoperative radiotherapy: Incorporating clinical and imaging features across treatment phases
Rong Cong, Sai-Nan Cheng, Bing Feng, Le-Yao Wang, Zhao-Wei Chen, Yong-Jian Zhu, Xiao-Hong Ma, Xin-Ming Zhao, Department of Diagnostic Radiology, National Cancer Center/National Clinical Research Center for Cancer/Cancer Hospital, Chinese Academy of Medical Sciences and Peking Union Medical College, Beijing 100021, China
Bo Chen, Yi-Rui Zhai, Department of Radiation Oncology, National Cancer Center/National Clinical Research Center for Cancer/Cancer Hospital, Chinese Academy of Medical Sciences and Peking Union Medical College, Beijing 100021, China
Fan Wu, Department of Hepatobiliary Surgery, National Cancer Center/National Clinical Research Center for Cancer/Cancer Hospital, Chinese Academy of Medical Sciences and Peking Union Medical College, Beijing 100021, China
Zhao-Wei Chen, Department of Radiology, Chengdu First People’s Hospital, Chengdu 610000, Sichuan Province, China
Co-corresponding authors: Xiao-Hong Ma and Xin-Ming Zhao.
Author contributions: Cong R, Chen B, Ma XH, and Zhao XM contributed to the study conception and design; Cong R, Chen B, Cheng SN, Wu F, and Zhai YR were responsible for patient enrollment and data collection; Cong R, Cheng SN, Feng B, Wang LY, Chen ZW, Zhu YJ, Ma XH, and Zhao XM involved in imaging review, data analysis, and interpretation of data; Cong R and Ma XH involved in manuscript preparation; Chen B, and Zhao XM involved in critical revision of the manuscript; Cong R and Chen B contributed equally to this manuscript as co-first authors; Ma XH and Zhao XM contributed equally to this manuscript as co-corresponding authors. All authors reviewed and approved the final manuscript.
Supported by Beijing Hope Run Special Fund of Cancer Foundation of China, No. LC2022A14; and National High Level Hospital Clinical Research Funding, No. 80102022505.
Institutional review board statement: This study was approved by the Ethics Committee of the National Cancer Center/Cancer Hospital, Chinese Academy of Medical Sciences and Peking Union Medical College, No. NCC2023C-936.
Informed consent statement: The need for patient consent was waived due to the retrospective nature of the study.
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 this study can be obtained from the corresponding author.
Corresponding author: Xin-Ming Zhao, Professor, Department of Diagnostic Radiology, National Cancer Center/National Clinical Research Center for Cancer/Cancer Hospital, Chinese Academy of Medical Sciences and Peking Union Medical College, No. 17 Panjiayuan Nanli, Chaoyang District, Beijing 100021, China. zhaoxinming@cicams.ac.cn
Received: December 30, 2025 Revised: January 28, 2026 Accepted: March 9, 2026 Published online: June 21, 2026 Processing time: 161 Days and 1.2 Hours
Abstract
BACKGROUND
There are still large variations in survival among patients with hepatocellular carcinoma (HCC) who received preoperative radiotherapy (RT), necessitating early recurrence (ER) risk stratification for decision-making and more appropriate management of patients.
AIM
To develop prognostic models integrating magnetic resonance imaging features and clinico-therapeutic-pathological variables across the treatment phases for predicting postoperative ER of HCC patients receiving preoperative RT.
METHODS
Consecutive 81 patients who received RT-based therapy and subsequent hepatectomy for HCC between November 2014 and December 2023 were retrospectively included. Magnetic resonance imaging features and clinico-therapeutic-pathological variables associated with ER-free survival (RFS) were identified using the least absolute shrinkage and selection operator Cox regression analysis. Multiple prognostic models across treatment phases were constructed by multivariate Cox regression analysis. Model performance was assessed with C-index, area under curve, calibration curve and decision curve analysis.
RESULTS
The baseline-preoperative model incorporated three baseline variables (cirrhosis, gamma-glutamyl transpeptidase level, arterial phase hyperenhancement portion ≥ 50%) and two preoperative variables (alpha-fetoprotein level, preoperative complete response). The postoperative combined model was developed by adding microvascular invasion and excluding alpha-fetoprotein level. The baseline-preoperative and postoperative combined model achieved C-indexes of 0.779 (95% confidence interval: 0.714-0.844) and 0.786 (95% confidence interval: 0.717-0.855), respectively, outperforming prognosis models based on other treatment phases and major staging systems (all P < 0.05). Moreover, the two models enabled the effective stratification of early RFS, overall RFS, and overall survival.
CONCLUSION
The baseline-preoperative and postoperative combined model could offer a simple tool to predict early postoperative recurrence of HCC patients receiving preoperative RT, which may assist with personalized treatment and surveillance.
Core Tip: This study developed two models (baseline-preoperative and postoperative combined models) that integrated multimodal factors across multiple treatment stages in patients with hepatocellular carcinoma who received radiotherapy-based therapy combined with subsequent surgery. These models exhibited comparable and satisfactory performance in predicting early recurrence and allowed risk stratification of early recurrence-free survival, overall recurrence-free survival, and overall survival. Including readily accessible variables, the findings have potential clinical relevance, particularly for preoperative decision-making regarding the timing and personalized planning of surgery after radiotherapy, as well as postoperative management involving adjuvant therapy and close surveillance for high-risk patients.
Citation: Cong R, Chen B, Cheng SN, Zhai YR, Wu F, Feng B, Wang LY, Chen ZW, Zhu YJ, Ma XH, Zhao XM. Early recurrence risk stratification of hepatocellular carcinoma receiving preoperative radiotherapy: Incorporating clinical and imaging features across treatment phases. World J Gastroenterol 2026; 32(23): 118308
Radical surgical resection remains the primary strategy for the long-term survival of patients with hepatocellular carcinoma (HCC); nevertheless, the low resection and high recurrence rates are still major obstacles to improving prognosis[1,2]. Locoregional therapy (LRT) has demonstrated its value in addressing such challenges, which could reduce or eliminate tumoral viability, lower the high risk of postoperative recurrence of resectable HCCs, facilitate curative resection of unresectable HCCs and ultimately offer better patient outcomes[3-5].
Advances in modern radiotherapy (RT) techniques have made RT a guideline-recommended LRT option for managing HCC, either alone or in combination, with favorable toxicity profiles and good local tumor control rates[6,7]. Adding RT before surgery has been reported to improve long-term survival of patients with HCC, particularly those with portal vein tumor thrombus, centrally located HCC, or a high risk of microvascular invasion (MVI)[8-11]. However, there is still large interindividual variation in survival. A prospective study by Wu et al[10] showed that 36.8% of patients who received neoadjuvant RT had early relapse within 2 years after surgery, accounting for 70% of patients with relapse. Early recurrence (ER) significantly affects patient survival and quality of life, and identifying its associated clinical and radiological markers is clinically imperative for decision making and more appropriate management of patients[12-14].
Tumor stage, serum markers such as alpha-fetoprotein (AFP) and the albumin-bilirubin (ALBI) ratio, and histopathologic characteristics including pathological subtype, tumor differentiation, and MVI have been reported as prognostic factors[15-17]. However, the use of these factors as therapeutic guidance in clinical practice remains limited because they are inadequate for fully characterizing the aggressiveness of HCC; some markers are also only accessible postoperatively. Growing evidence has indicated the potential of radiological markers, such as non-smooth tumor margins, corona enhancement, arterial phase hypovascular components, radiologic response after LRT, and markedly low apparent diffusion coefficient (ADC) values, to infer pathological features and survival outcomes[18-20]. Some clinical, radiological or incorporating prognostic models have been developed to predict survival for patients who received either surgery alone or RT alone[21-23]; however, there is a lack of models for predicting ER in patients receiving RT combined with surgical resection for HCC. Therefore, this study aimed to develop prognostic models based on readily accessible clinical, radiological and pathological factors from baseline, preoperative, and postoperative stages to predict postoperative ER in patients with HCC who received preoperative RT.
MATERIALS AND METHODS
Patients
This retrospective cohort study was approved by the Ethics Committee of the National Cancer Center/Cancer Hospital, Chinese Academy of Medical Sciences and Peking Union Medical College, No. NCC2023C-936, and the requirement for informed consent was waived. We identified consecutive adult patients (≥ 18 years) who received RT-based therapy and subsequent curative-intent surgical resection for HCC between November 2014 and December 2023. The inclusion criteria were as follows: (1) HCC diagnosed as LR-5 by Liver Imaging Reporting and Data System version 2018 or biopsy-proven; (2) Baseline magnetic resonance imaging (MRI) within 1 month preinitiation of treatment and available MRI within 1 month preoperatively; and (3) Available clinical, laboratory, and pathological information. The exclusion criteria were as follows: (1) Any locoregional treatment or systemic therapy before admission; (2) Surgical pathology diagnosis of combined hepatocellular-cholangiocarcinoma; and (3) No available MRI at baseline. Ultimately, 81 patients were included in this study (Figure 1A).
Figure 1 Study population and design.
A: Flowchart of patient selection; B: Study schema. HCC: Hepatocellular carcinoma; MRI: Magnetic resonance imaging.
Data collection
Clinical data across three treatment phases were collected from electronic medical records (Figure 1B): Baseline phase data including age, sex, Child-Pugh class, Barcelona Clinic Liver Cancer stage, liver cirrhosis, serum hepatitis status, and laboratory data [AFP, albumin, aspartic transaminase, alanine aminotransferase, total bilirubin, and gamma-glutamyl transpeptidase (GGT) levels, and platelet count]; preoperative phase data comprising the same laboratory data; and postoperative phase data consisting of pathological response and pathological information of viable tumors (tumor differentiation, liver capsule involvement, MVI status, and satellite nodules). The ALBI score was calculated using the following formula: ALBI score = (log10 total bilirubin μmol/L × 0.66) + (albumin g/L × -0.085); the ALBI grade was defined as follows: ≤ -2.60 (grade 1), > -2.60 to ≤ -1.39 (grade 2) and > -1.39 (grade 3)[24].
RT procedure
All patients received intensity-modulated RT or volumetric-modulated arc therapy preoperatively. Four-dimensional computed tomography (CT) was performed with chest-abdominal thermoplastic mask immobilization. CT and MRI were co-registered to optimize delineation of target and normal structure for treatment planning. Gross tumor volume included primary tumor and tumor thrombosis. Clinical target volume was defined as gross tumor volume plus a 5 mm margin in all directions and 1 cm along the vessels invaded by tumor thrombosis. Planning target volume was created as clinical target volume plus a 5 mm margin in the anterior-posterior and left-right directions and a 10 mm margin in the cranial-caudal direction. The prescribed dose to 95% of the planning target volume was 46-60 Gy in 23-30 fractions over 5-6 weeks, based on the dose constraints of organs at risk.
Image acquisition
All patients underwent contrast-enhanced MRI examinations using 3.0-T MR systems (GE Signa HDxt, GE Discovery MR 750, GE SIGNA™ Pioneer, GE HealthCare, WI, United States; or Siemens MAGNETOM Prisma, Siemens Healthineers AG, Germany) at baseline and preoperatively. The standard liver MRI protocol included in- and opposed-phase T1-weighted imaging (T1WI), axial fat-suppression T2-weighted imaging, diffusion-weighted imaging (b values of 0 and 800 second/mm2) with ADC maps reconstructed using the monoexponential model, and dynamic contrast-enhanced-T1WI. For dynamic contrast-enhanced-T1WI, un-enhanced, arterial phases (30 seconds), portal venous phase (60 seconds), delayed phase/transitional phase (180 seconds), and hepatobiliary phase (20 minutes, for gadoxetic acid) were acquired using a 3D T1WI breath-hold fat-suppressed spoiled gradient-recall echo sequence (LAVA or VIBE) pre- and post-injection of gadodiamide (Omniscan, GE Healthcare, WI, United States) or gadoxetic acid (Primovist, Bayer Healthcare, Germany). MRI sequence details are provided in Supplementary Table 1.
Image analysis
Two radiologists (Readers 1 and 2, with 5 and 10 years of experience in liver MRI, respectively) independently evaluated the MRI features, treatment response, and ADC values without knowledge of the clinical, laboratory, pathological, and follow-up information. Any discrepancies were resolved by reaching a consensus that involved a third radiologist (Reader 3, with over 20 years of experience in liver MRI). MRI features were assessed at baseline, including tumor burden (e.g., number, size, satellite tumors, and tumor in vein), Liver Imaging Reporting and Data System features, and other previously reported prognostic MRI features [e.g., non-smooth tumor margin, intratumoral artery, and arterial phase hyperenhancement (APHE) portion]. For multiple HCCs, the largest lesion was selected for analysis. Treatment response was recorded at early follow-up after RT (within 3 months) and preoperatively (within 1 month) using the Response Evaluation Criteria in Solid Tumors (RECIST), and the modified RECIST (mRECIST). Detailed information regarding MRI features and treatment response evaluation is provided in Supplementary Table 2.
For quantitative image analysis, ADCpre and ADCpost were measured by delineating the maximum cross-sectional area of the tumor on baseline and preoperative ADC maps, respectively, including cystic, necrotic, and hemorrhagic regions, to comprehensively capture tumor heterogeneity. The ADC of normal liver parenchyma was also measured to determine the tumor-to-liver ADC ratio, by positioning regions of interest (200-300 mm2), as far as possible, from tumor, major vasculature, abnormal perfusion, and artifacts. The values measured by the two radiologists were averaged for analysis. ΔADC% was calculated according to the following equation: (ADCpost - ADCpre)/ADCpre × 100.
Follow up
After hepatectomy, all patients underwent regular follow-up every 3 months for the first 2 years, every 6 months for the following 3 years, and annually thereafter. Serum AFP, liver function, and imaging examination (contrast-enhanced CT or MRI) were assessed at each follow up. Patients were followed up until death or the date of final follow-up (March 25, 2025). Recurrence was defined as radiological and/or histopathological identification of intrahepatic HCCs, tumor in vein or extrahepatic metastases. Recurrence-free survival (RFS) was calculated from the date of radical hepatectomy to the earliest diagnosis of recurrence or last follow-up. Early RFS was defined as the time from hepatectomy to ER or death (occurring ≤ 2 years). For patients who remained event-free within 2 years after surgery, early RFS were was recorded as the time from hepatectomy to the last available follow-up (if ≤ 2 years) or 2 years (if > 2 years) with a negative endpoint. Overall survival (OS) was defined as the time from hepatectomy to all-cause death or the last follow-up.
Statistical analysis
Statistical analyses were performed using SPSS software (version 26.0, IBM, NY, United States) and R software (version 3.6.3). The ER and non-ER groups were compared using the χ2 test or Fisher’s exact test for categorical variables and Student’s t-test or Mann-Whitney U test for continuous variables. Categorical variables are expressed as n (%). Continuous variables are expressed as mean ± SD or median [interquartile range (IQR)]. Interobserver agreement was measured using the κ coefficient for categorical features and intraclass correlation coefficient for continuous variables.
For survival analysis, continuous laboratory variables were transformed into categorical variables according to the ranges of normality for laboratory values. Variables with P < 0.2 in the univariate Cox regression analysis were screened using the LASSO Cox regression analysis with 10-fold cross validation to select variables with non-zero penalty coefficients. The identified baseline, preoperative, and postoperative variables were independently or collectively entered into multivariate Cox regression models using the backward stepwise method. Five models were constructed to predict ER (Figure 1B): (1) The baseline model, using baseline variables only; (2) The baseline-preoperative model, using baseline and preoperative variables; (3) The preoperative model, using preoperative variables only; (4) The pre-postoperative model, combining pre- and post-operative variables; and (5) The postoperative combined model, which included all variables from all three phases.
Model discrimination was assessed using the C-index and time-dependent area under the curve. Calibration and decision curves were plotted to evaluate the agreement and net clinical benefits of models. The bootstrap resampling method (1000 replications) was used for internal validation. Survival curves were estimated using the Kaplan-Meier method and compared using the log-rank test. Statistical significance was set at P < 0.05.
RESULTS
Patient characteristics
Patient characteristics are summarized in Tables 1 and 2. A total of 81 patients [median age: 55 years; IQR: 46.5-62 years; 71 (87.7%) male] were included in this study. Of them, 18.5% (15/81), 9.9% (8/81), 1.2% (1/81), and 1.2% (1/81) had tumor thrombus involving the segmental branches or above of the portal vein, right or left portal vein, main portal vein, and inferior vena cava, respectively. Pre-RT transarterial chemoembolization, RT concurrent with systemic therapy, and post-RT transarterial chemoembolization were performed in 6.2% (5/81), 50.6% (41/81), and 9.9% (8/81) of the patients, respectively.
Table 1 Baseline clinical characteristics and imaging features of patients, n (%) or median (interquartile range).
The median follow-up time was 28.2 months (IQR: 15.5-70.3 months). During the follow-up period, 49 (60.5%) patients experienced recurrence, of whom 75.5% (37/49) developed ER. Compared to the non-ER group, the ER group had significantly elevated AFP and GGT levels, together with higher rates of preoperative complete response (CR) and MVI (all P < 0.05, Tables 1 and 2). The cumulative 1-, 3-, and 5-year RFS rates were 65%, 47%, and 28%, respectively. The 1-year, 3-year, and 5-year OS rates were 92%, 75%, and 71%, respectively.
Prognostic predictors of early RFS and model construction
Seven variables were selected as potential factors associated with early RFS through univariate Cox regression analysis (P < 0.2, Supplementary Table 3) and LASSO Cox regression analysis (Supplementary Figure 1): Cirrhosis [hazard ratio (HR) = 2.30; P = 0.115], GGT > 60 U/L (HR = 3.23; P = 0.001), tumor size > 5 cm (HR = 1.75; P = 0.184), APHE portion ≥ 50% (HR = 0.46; P = 0.025), preoperative AFP > 400 ng/mL (HR = 3.65; P = 0.001), preoperative CR by mRECIST (HR = 0.32; P = 0.030), and MVI (HR = 2.15; P = 0.024).
Based on the four baseline factors, two preoperative factors, and one postoperative factor detailed above, we developed five prognostic models using multivariate Cox regression analysis (Table 3). The postoperative combined and baseline-preoperative models are presented as intuitive nomograms in Figure 2. The interobserver agreement was excellent for APHE portion ≥ 50% [0.83; 95% confidence interval (CI): 0.69-0.95] and substantial for preoperative CR by mRECIST (0.73; 95%CI: 0.52-0.90). The interobserver agreement for the imaging features is summarized in Supplementary Table 4.
The discriminative abilities of the five models and major staging systems are shown in Table 4 and Figure 3A. The postoperative combined model showed good discrimination ability, with a C-index of 0.786 (95%CI: 0.717-0.855) and a bootstrap-corrected C-index of 0.758, which is comparable with the C-index of the baseline-preoperative model (0.779; 95%CI: 0.714-0.844; P = 0.704). These two models were significantly superior to the other models and major staging systems (all P < 0.05). Similarly, the time-dependent area under the curves of the postoperative combined (0.845, 0.861, and 0.866 at 6, 12, and 18 months, respectively) and baseline-preoperative models (0.826, 0.860, and 0.865 at 6, 12, and 18 months, respectively) were also higher than those of the other models at various time points (all P < 0.05).
Figure 3 Performance of the prognostic models.
A: Time-dependent areas under the curves of the prognostic models at various time points; B: Calibration curves for the baseline-preoperative model for predicting 6-, 12-, and 18-month early recurrence-free survival; C: Calibration curves for the postoperative combined model for predicting 6-, 12-, and 18-month early recurrence-free survival. AUC: Area under the curve.
Table 4 C-index of the prognostic models and major staging systems.
Calibration plots for the postoperative combined and baseline-preoperative models showed good agreement between the model predictions and the actual observations (Figure 3B and C). The decision curves demonstrated that the postoperative combined and baseline-preoperative models provided a larger net benefit than the other models and major staging systems (Supplementary Figure 2).
Survival risk stratification
Based on the baseline-preoperative nomogram, patients were stratified into a low-risk group (≤ 217 points) or a high-risk group (> 217 points). Kaplan-Meier curves showed a significant difference in early RFS, overall RFS and OS between the two groups (all P < 0.001; Figure 4A-C). Based on the postoperative combined nomogram, patients were stratified into a low-risk group (≤ 240 points) or a high-risk group (> 240 points). Similarly, Kaplan-Meier curves showed a significant difference in early RFS, overall RFS and OS between the two groups (all P < 0.01; Figure 4D-F).
Figure 4 Kaplan-Meier curves for survival of different risk groups.
A: Early recurrence-free survival (≤ 2 years) stratified by the baseline-preoperative nomogram; B: Overall recurrence-free survival stratified by the baseline-preoperative nomogram; C: Overall survival stratified by the baseline-preoperative nomogram; D: Early recurrence-free survival (≤ 2 years) stratified by the postoperative combined nomogram; E: Overall recurrence-free survival stratified by the postoperative combined nomogram; F: Overall survival stratified by the postoperative combined nomogram.
DISCUSSION
In the present study, we investigated the prognostic predictors of ER in patients with HCC who received RT-based therapy combined with subsequent liver radical resection and with heterogeneity in postoperative early RFS. By comprehensively considering clinical, multimodal imaging, therapeutic, and pathological variables across the entire treatment period, we developed and internally validated two models (baseline-preoperative and postoperative combined models) that enabled risk stratification of ER after resection following RT. These two models outperformed the prognostic models based on other treatment phases (baseline, preoperative, and pre-postoperative models) and three major staging systems.
The integration of multi-phase data yielded better prognostic prediction performance than the use of any single-phase data alone or conventional staging systems, thereby highlighting the dynamic and integrated aspects of treatment. Upon comparing models from different treatment phases and assessing the contribution of factors from various phases within the combined model, we found that baseline factors contributed more significantly to prognostic prediction. The baseline factors reflect the inherent aggressiveness of the tumor and the constitutional status of the patient. Cirrhosis, GGT > 60 U/L, and APHE portion ≥ 50% were selected and included in the baseline-preoperative and postoperative combined models, with cirrhosis and GGT > 60 U/L having a greater magnitude of effect on the outcome. Liver cirrhosis is a major established risk factor for the development and progression of HCC, and it is commonly associated with poor liver function and high perioperative risk[25,26]. As a key enzyme in glutathione metabolism and a biomarker of oxidative stress, elevated GGT levels also indicate liver dysfunction and play an important role in tumor formation and progression[27,28]. Numerous studies have shown the negative impact of cirrhosis and high GGT levels on HCC prognosis postoperatively or post-LRT[23,29-32], which is consistent with the findings in the present study. Despite achieving preoperative CR, all four patients with cirrhosis and elevated GGT in our study experienced ER, suggesting that preoperative RT may not completely eliminate micro-metastases or alter the tumor-prone liver microenvironment. In addition, we observed an association between the APHE portion and risk of postoperative ER. The arterial phase hypoenhancement portion is likely to reflect tumor necrosis, abundant fibrotic stroma, hemorrhage, and/or low microvascular density. Less than 50% APHE has been reported to be associated with macrotrabecular-massive subtype, vessels encapsulate tumor clusters, and advanced-stage recurrence[19,33,34], collectively indicating its prognostic value.
With the inclusion of preoperative factors, the baseline-preoperative model outperformed the baseline model. Preoperative AFP > 400 ng/mL was a significant risk factor for predicting ER of RT-treated HCC, consistent with previous studies on patients with HCC undergoing surgery, LRT, or LRT combined with surgery[20,35,36]. Preoperative CR according to mRECIST was another preoperative prognostic factor. Multiple studies conducted in diverse HCC treatment settings have demonstrated an association between treatment response and patient survival[22,37,38]. Wu et al[22] indicated that attaining CR according to mRECIST within 6 months following stereotactic body radiation therapy predicted better progression-free survival and OS, and Muglia et al[38] confirmed a strong correlation between the complete radiological response by mRECIST and OS in patients treated with Yttrium-90 transarterial radioembolization for HCC.
MVI was the only postoperative variable included in the postoperative combined model, while the preoperative AFP level was excluded, and the postoperative combined model showed a marginally better performance than the baseline-preoperative model. MVI status is routinely incorporated into prognostic stratification systems because of its validated association with more advanced tumor stage, aggressive recurrence and poorer survival[21,39]. Unlike complete radiological response, our findings indicated that major pathological response and pathological CR did not emerge as independent predictors of ER. Similarly, Youn et al[20] reported that pathologic viable tumor was not an independent prognostic factor for LRT-treated HCC. Presumably, the prognostic power of major pathological response and pathological CR is weakened by rigid cutoff for minimal residual tumor.
The baseline-preoperative and postoperative combined models also enabled risk stratification of overall RFS and OS after resection following RT. These findings have potential clinical relevance. First, preoperative CR by mRECIST was associated with improved RFS, and preoperative AFP > 400 ng/mL correlated with poor RFS. Preoperative treatment response and AFP level may indicate the timing of surgery. For example, an appropriate extension of the treatment course to achieve further tumor response and reduced AFP levels might benefit patients with objective response or a declining yet persistently elevated AFP level after RT, especially those with concurrent high-risk baseline factors. Second, high-risk patients stratified by the baseline-preoperative model may benefit from more aggressive surgical strategies such as a wider resection margin or anatomical hepatectomy. Third, the postoperative combined model may help identify high-risk patients likely to benefit from adjuvant therapies and closer surveillance (e.g., more frequent follow-up and more sensitive techniques). Additionally, the variables included in models were clinically readily accessible.
This study has some limitations. First, the single-center source and limited sample size are notable constraints, and the absence of external validation may have affected the generalizability of the results. Second, the retrospective study design inevitably results in selection bias. Further validation in a larger, multicenter, prospective cohort is warranted to confirm the reliability of our models, as well as their potential utility as clinical decision-making tools. Third, up to 80.2% of the enrolled patients had chronic hepatitis B. The validation and extrapolation of our findings also demand enrollment of patients with diverse chronic liver disease etiologies. Finally, contrast-enhanced MRI examinations were conducted using either extracellular or hepatobiliary contrast agents. Although Yoon et al[40] reported that the prognostic imaging features of HCC may be comparable in terms of ER between the two contrast agents, there remains potential variability in imaging assessment, and the potential additional value of hepatobiliary phase features will be further explored in a larger cohort with an increased proportion of patients undergoing MRI with hepatobiliary contrast agents.
CONCLUSION
In conclusion, we developed two models (baseline-preoperative and postoperative combined models) that integrate multimodal factors across multiple treatment stages in patients with HCC who received RT-based therapy combined with subsequent surgery. The models had comparable and satisfactory performance in predicting early HCC recurrence. These two models enabled effective stratification of early RFS, overall RFS, and OS, potentially guiding personalized treatment and surveillance.
Chen LC, Lin HY, Hung SK, Chiou WY, Lee MS. Role of modern radiotherapy in managing patients with hepatocellular carcinoma.World J Gastroenterol. 2021;27:2434-2457.
[PubMed] [DOI] [Full Text]
Wei X, Jiang Y, Zhang X, Feng S, Zhou B, Ye X, Xing H, Xu Y, Shi J, Guo W, Zhou D, Zhang H, Sun H, Huang C, Lu C, Zheng Y, Meng Y, Huang B, Cong W, Lau WY, Cheng S. Neoadjuvant Three-Dimensional Conformal Radiotherapy for Resectable Hepatocellular Carcinoma With Portal Vein Tumor Thrombus: A Randomized, Open-Label, Multicenter Controlled Study.J Clin Oncol. 2019;37:2141-2151.
[RCA] [PubMed] [DOI] [Full Text][Cited by in Crossref: 237][Cited by in RCA: 223][Article Influence: 31.9][Reference Citation Analysis (3)]
Wu F, Chen B, Dong D, Rong W, Wang H, Wang L, Wang S, Jin J, Song Y, Liu Y, Fang H, Tang Y, Li N, Zhu X, Li Y, Wang W, Wu J. Phase 2 Evaluation of Neoadjuvant Intensity-Modulated Radiotherapy in Centrally Located Hepatocellular Carcinoma: A Nonrandomized Controlled Trial.JAMA Surg. 2022;157:1089-1096.
[RCA] [PubMed] [DOI] [Full Text][Cited by in Crossref: 5][Cited by in RCA: 32][Article Influence: 8.0][Reference Citation Analysis (0)]
Wei X, Jiang Y, Feng S, Lu C, Huo L, Zhou B, Meng Y, Lau WY, Zheng Y, Cheng S. Neoadjuvant intensity modulated radiotherapy for a single and small (≤5 cm) hepatitis B virus-related hepatocellular carcinoma predicted to have high risks of microvascular invasion: a randomized clinical trial.Int J Surg. 2023;109:3052-3060.
[RCA] [PubMed] [DOI] [Full Text][Cited by in Crossref: 13][Cited by in RCA: 13][Article Influence: 4.3][Reference Citation Analysis (0)]
Bruix J, Takayama T, Mazzaferro V, Chau GY, Yang J, Kudo M, Cai J, Poon RT, Han KH, Tak WY, Lee HC, Song T, Roayaie S, Bolondi L, Lee KS, Makuuchi M, Souza F, Berre MA, Meinhardt G, Llovet JM; STORM investigators. Adjuvant sorafenib for hepatocellular carcinoma after resection or ablation (STORM): a phase 3, randomised, double-blind, placebo-controlled trial.Lancet Oncol. 2015;16:1344-1354.
[RCA] [PubMed] [DOI] [Full Text][Cited by in Crossref: 896][Cited by in RCA: 837][Article Influence: 76.1][Reference Citation Analysis (6)]
Yao LQ, Chen ZL, Feng ZH, Diao YK, Li C, Sun HY, Zhong JH, Chen TH, Gu WM, Zhou YH, Zhang WG, Wang H, Zeng YY, Wu H, Wang MD, Xu XF, Pawlik TM, Lau WY, Shen F, Yang T. Clinical Features of Recurrence After Hepatic Resection for Early-Stage Hepatocellular Carcinoma and Long-Term Survival Outcomes of Patients with Recurrence: A Multi-institutional Analysis.Ann Surg Oncol. 2022.
[RCA] [PubMed] [DOI] [Full Text][Cited by in Crossref: 64][Cited by in RCA: 60][Article Influence: 15.0][Reference Citation Analysis (0)]
Jiang H, Yang C, Chen Y, Wang Y, Wu Y, Chen W, Ronot M, Chernyak V, Fowler KJ, Bashir MR, Song B. Development of a Model including MRI Features for Predicting Advanced-stage Recurrence of Hepatocellular Carcinoma after Liver Resection.Radiology. 2023;309:e230527.
[RCA] [PubMed] [DOI] [Full Text][Cited by in RCA: 34][Reference Citation Analysis (0)]
Chan AWH, Zhong J, Berhane S, Toyoda H, Cucchetti A, Shi K, Tada T, Chong CCN, Xiang BD, Li LQ, Lai PBS, Mazzaferro V, García-Fiñana M, Kudo M, Kumada T, Roayaie S, Johnson PJ. Development of pre and post-operative models to predict early recurrence of hepatocellular carcinoma after surgical resection.J Hepatol. 2018;69:1284-1293.
[RCA] [PubMed] [DOI] [Full Text][Cited by in Crossref: 460][Cited by in RCA: 436][Article Influence: 54.5][Reference Citation Analysis (4)]
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: 2235][Article Influence: 203.2][Reference Citation Analysis (3)]
Corti A, Franzini M, Paolicchi A, Pompella A. Gamma-glutamyltransferase of cancer cells at the crossroads of tumor progression, drug resistance and drug targeting.Anticancer Res. 2010;30:1169-1181.
[PubMed] [DOI]
Wang L, Feng B, Li D, Liang M, Wang S, Wang S, Ma X, Zhao X. Risk stratification of solitary hepatocellular carcinoma ≤ 5 cm without microvascular invasion: prognostic values of MR imaging features based on LI-RADS and clinical parameters.Eur Radiol. 2023;33:3592-3603.
[RCA] [PubMed] [DOI] [Full Text][Cited by in RCA: 16][Reference Citation Analysis (0)]
Qu Q, Liu Z, Lu M, Xu L, Zhang J, Liu M, Jiang J, Gu C, Ma Q, Huang A, Zhang X, Zhang T. Preoperative Gadoxetic Acid-Enhanced MRI Features for Evaluation of Vessels Encapsulating Tumor Clusters and Microvascular Invasion in Hepatocellular Carcinoma: Creating Nomograms for Risk Assessment.J Magn Reson Imaging. 2024;60:1094-1110.
[RCA] [PubMed] [DOI] [Full Text][Cited by in Crossref: 1][Cited by in RCA: 12][Article Influence: 6.0][Reference Citation Analysis (0)]
Park SH, Kim B, Kim S, Park S, Park YH, Shin SK, Sung PS, Choi JI. Estimating postsurgical outcomes of patients with a single hepatocellular carcinoma using gadoxetic acid-enhanced MRI: risk scoring system development and validation.Eur Radiol. 2023;33:3566-3579.
[RCA] [PubMed] [DOI] [Full Text][Cited by in RCA: 4][Reference Citation Analysis (0)]
Sheng Y, Wang Q, Liu HF, Chen WH, He ZM, Wang Q. Preoperative Nomogram Incorporating Clinical Factors, Serological Markers and LI-RADS MRI Features to Predict Early Recurrence of Hepatocellular Carcinoma Treated with Transarterial Chemoembolization.Acad Radiol. 2023;30:1288-1297.
[RCA] [PubMed] [DOI] [Full Text][Cited by in RCA: 14][Reference Citation Analysis (0)]
Scheiner B, Kang B, Balcar L, Radu IP, Reiter FP, Adžić G, Guo J, Gao X, Yuan X, Cheng L, Gorgulho J, Schultheiss M, Peeters F, Hucke F, Ben Khaled N, Piseddu I, Philipp A, Sinner F, D'Alessio A, Pomej K, Saborowski A, Bathon M, Schwacha-Eipper B, Zarka V, Lampichler K, Nishida N, Lee PC, Krall A, Saeed A, Himmelsbach V, Tesini G, Huang YH, Vivaldi C, Masi G, Vogel A, Schulze K, Trauner M, Djanani A, Stauber R, Kudo M, Parikh ND, Dufour JF, Prejac J, Geier A, Bengsch B, von Felden J, Venerito M, Weinmann A, Peck-Radosavljevic M, Finkelmeier F, Dekervel J, Ji F, Wang HW, Rimassa L, Pinato DJ, Bouattour M, Chon HJ, Pinter M. Outcome and management of patients with hepatocellular carcinoma who achieved a complete response to immunotherapy-based systemic therapy.Hepatology. 2025;81:1714-1727.
[RCA] [PubMed] [DOI] [Full Text][Cited by in Crossref: 3][Cited by in RCA: 28][Article Influence: 28.0][Reference Citation Analysis (0)]
Muglia R, De Giorgio M, Marra P, Carbone FS, Dulcetta L, Prussia C, Loglio A, Ghirardi A, Grikke LA, Bianchi C, Poli GL, Gerali A, Erba PA, Sironi S, Fagiuoli S, Viganò M. Long-term outcomes of Yttrium-90 transarterial radioembolization for patients with hepatocellular carcinoma.Eur J Nucl Med Mol Imaging. 2025;52:3114-3124.
[RCA] [PubMed] [DOI] [Full Text][Cited by in RCA: 5][Reference Citation Analysis (2)]
