Preoperative prognostic model combining tumor burden score and tumor markers to predict long-term outcomes following hepatectomy for intrahepatic cholangiocarcinoma: a multi-institutional analysis

Background and aim: Intrahepatic cholangiocarcinoma (ICC) is an aggressive liver malignancy with limited prognostic tools to guide treatment strategies. This study aimed to develop and validate a preoperative prognostic model combining tumor burden score (TBS), carcinoembryonic antigen (CEA), and carbohydrate antigen 19-9 (CA19-9), termed the TCCA model, to predict outcomes in patients with ICC undergoing hepatectomy.

Methods: Patients who underwent curative resection for ICC between 2014 and 2020 were retrospectively identified from a multi-institutional database. The impact of the TCCA model on overall survival (OS) and recurrence-free survival (RFS) was evaluated in training and validation cohorts. Predictive performance was evaluated using the area under the Receiver Operating Characteristic curve (AUC), the Akaike Information Criterion (AIC), and the C-index.

Results: A total of 849 patients were included. Lower TCCA scores were associated with better median OS (score 0: 59.7 months; score 1: 31.3 months; score 2: 19.4 months; score 3: 11.5 months, respectively) and median RFS (28.8; 15.4; 9.7; 8.1 months, respectively). The TCCA model performed well in both the training cohort (AUC: 0.697 for OS and 0.649 for RFS) and the validation cohort (AUC: 0.672 for OS and 0.632 for RFS), outperforming the 8th edition TNM system and other models, with the highest C-index (0.734) and lowest AIC (3840). Subgroup analyses demonstrated that the TCCA model maintained good discriminative ability among patients with negative CEA or CA19–9 levels.

Conclusion: The TCCA model accurately stratifies ICC patients for OS and RFS after resection. It provides a simple and practical tool for preoperative risk assessment, supporting individualized surgical decision-making and individualized patient counseling.

Introduction

Intrahepatic cholangiocarcinoma (ICC), the second most common primary liver cancer, arises from the epithelium lining the peribiliary glands of secondary or higher-order bile ducts (1). Despite an increase in incidence over recent decades, ICC remains a rare malignancy, with most institutions performing few resections annually (2). Although radical resection is the most effective treatment for resectable ICC, the disease has a high propensity for recurrence and metastasis rate, leading to poor long-term outcomes, with reported 5-year overall survival (OS) rates ranging from 20% to 40% (3).

Even when standard criteria for resectable ICC are applied, it remains unclear which subgroup of patients truly benefits from surgical resection in terms of long-term survival (4). To better stratify patients undergoing resection, the American Joint Committee on Cancer (AJCC) regularly updates the Tumor−Node−Metastasis (TNM) classification system (5). However, the TNM classification system relies on postoperative pathological data, limiting its utility for preoperative decision-making (6). Several postoperative prognostic models have also been proposed, but they share similar limitations (7, 8). Given the aggressive nature of ICC, accurate preoperative prognostic evaluation is crucial for optimizing treatment strategies.

Various preoperative prognostic models have been introduced to improve treatment decision-making for ICC, incorporating clinical parameters such as radiologic features and inflammatory markers (9–12). While these models show promise, many lack sufficient accuracy or are too complex for routine clinical practice. For instance, radiomics-based models, although effective, require advanced imaging analysis tools that may not be readily available in routine clinical practice (13). Additionally, models based on inflammatory markers, such as Neutrophil−to−Lymphocyte Ratio (NLR), Prognostic Nutritional Index (PNI), however, may fail to capture the full oncologic complexity of ICC, underscoring the need for a more balanced yet comprehensive scoring system that is both practical and accurate in predicting patient outcomes (14, 15).

The tumor burden score (TBS), which combines tumor size and number, is a well-established prognostic factor for ICC and is now part of the AJCC-T classification. Moreover, elevated serum levels of carcinoembryonic antigen (CEA) or Carbohydrate antigen 19-9 (CA19-9) are associated with adverse outcomes and reflect underlying oncological behavior (16). However, previous studies have generally combined TBS with a single biomarker such as CA19–9 or CEA (17, 18). Relying on one serum marker may underestimate the biological heterogeneity of ICC, particularly in patients who are CA19-9- or CEA-negative.

To the best of our knowledge, this study is the first to integrate TBS, CEA, and CA19–9 into a single preoperative model—termed the TCCA model—to provide a more comprehensive and balanced assessment of both tumor burden and tumor biology, thereby improving individualized risk stratification and treatment planning for patients with ICC undergoing liver resection.

Methods

Patients and selection criteria

The retrospective study collected data from nine large tertiary medical institutions across China between 2014 and 2020. The training cohort data were obtained from Mengchao Hepatobiliary Hospital of Fujian Medical University, Eastern Hepatobiliary Surgery Hospital of Naval Medical University, and The Affiliated Hospital of Chuanbei Medical University. The validation cohort data were gathered from Cancer Hospital of the Chinese Academy of Medical Sciences, Tongji Hospital Affiliated to Tongji Medical College, Renji Hospital Affiliated to Shanghai Jiaotong University, The Second Hospital Affiliated to Zhejiang University, Tiantan Hospital Affiliated to Capital Medical University, and West China Hospital of Sichuan University. Informed consent was obtained from all patients before surgery, adhering strictly to the guidelines of the Declaration of Helsinki. Ethical approval was obtained from the Institutional Review Boards of all participating institutions (approval number 2023_017_01).

The inclusion criteria were as follows: (1) ECOG score of 0−2, (2) Child-Pugh score between A5 to B7, (3) Dynamic contrast-enhanced computed tomography (CT) or Multi-phase magnetic resonance imaging (MRI) assessment of tumor size and number within one month before surgery, (4) serum levels of CEA and CA19–9 were measured within one week prior to surgery, (5) R0 resection with postoperative pathological confirmation of ICC. The exclusion criteria were as follows: (1) receipt of preoperative therapies such as radiofrequency ablation, local interventional procedures, or chemotherapy, (2) recurrent ICC or other simultaneously malignancies, (3) died within 30 days or were lost to follow−up within 3 months postoperatively.

Variables of interested

Demographic and preoperative laboratory examination data were collected, including age, gender, Hepatitis B surface antigen (HBsAg), presence of cirrhosis, Child-Pugh grade, as well as CA19–9 and CEA levels. Inflammatory and nutritional markers such as NLR, platelet-to-lymphocyte ratio (PLR), PNI, andγ-glutamyl transferase to alanine aminotransferase ratio (GAR) were also calculated, and determine the optimal cutoff values based on ROC analysis. Number of tumors and the size of the largest tumor were assessed via CT or MRI. Additional surgical and pathological data included details on the anatomic resection, margin width, tumor differentiation, satellite nodular, macrovascular and microvascular invasion, as well as information on postoperative adjuvant chemotherapy. TNM staging followed the 8^th edition of the AJCC staging manual.

Definition of TBS and TCCA model

TBS is determined as the Euclidean distance on a Cartesian plane, based on two factors: the largest tumor size (x-axis) and the total number of tumors (y-axis). In cases with multiple lesions, the tumor size refers to the largest nodule. The formula for calculating TBS is TBS² = (tumor size)² + (number of tumors)² (19). For example, if the largest tumor measures 3 cm and there are four tumors in total, the TBS can be calculated as TBS = √(3² + 4²) = 5, resulting in a TBS value of 5 units.

The TCCA model is composed of three preoperative variables derived from multivariate analysis: TBS, CEA, and CA19-9. Each continuous variable is dichotomized into low and high groups. The optimal critical value of TBS was determined to be 6.0 units through ROC analysis. The cutoff values for CEA (5.0 ng/mL) and CA19-9 (37.0 U/mL) are based on the normal upper limit for laboratory examination. Patients are assigned a score of 1 for each high group and 0 for each low group, resulting in a total TCCA score ranging from 0 to 3 points.

Definition of other important clinical and follow−up related variables

Microvascular invasion (MVI) was defined as intraparenchymal vascular involvement, while macrovascular invasion involved major branches of the portal vein, hepatic artery, or hepatic veins, all based on histological examination. Adjuvant chemotherapy includes the administration of gemcitabine-based chemotherapy regimens or capecitabine-based monotherapy within 1 to 2 months postoperative, with or without TACE. The primary outcome was OS, defined as the interval from ICC resection to death or last follow-up, while the secondary outcome was recurrence-free survival (RFS), defined as the time from resection to recurrence or last follow-up. Recurrence was confirmed by tumor biopsy or follow-up imaging identifying suspicious lesions.

Statistical analysis

Continuous variables were summarized as mean ± standard deviation (SD) or median with interquartile range (IQR). Group differences were evaluated using one-way ANOVA or the Kruskal-Wallis tests. Categorical variables were reported as frequencies and percentages, with group comparisons conducted using the Chi-square test. The ROC curve, Harrell c-index, and Akaike information criterion (AIC) were used to assess the discriminative ability and predictive accuracy of the model, while Kaplan-Meier survival analysis was conducted to evaluate long-term outcomes. Model calibration was assessed using calibration plots and the Hosmer-Lemeshow test. Statistical analysis was executed using SPSS^® version 25.0 (IBM, Armonk, New York, USA) and R program version 3.2.0 (http://www.r-project.org/). A p-value of less than 0.05 was deemed statistically significant.

Results

Baseline characteristics of training and validation cohorts

A total of 849 ICC patients were retrospectively reviewed, including 635 in the training cohort and 214 in the validation cohort. As summarized in Table 1, the mean age of the entire cohort was 54.4 years, with the majority being male (n =576, 67.8%). Approximately half of the patients were HBsAg-positive (n =411, 48.4%). The mean maximum tumor size was 6.25 cm, and 16.6% of patients had multiple tumors. Compared to the validation cohort, the training cohort had higher proportions of liver cirrhosis (31.0% vs. 23.4%), elevated CEA levels (28.3% vs. 20.0%), MVI (14.2% vs. 8.4%), and poorer tumor differentiation (93.7% vs. 88.3%), while fewer patients received postoperative adjuvant chemotherapy (12.8% vs. 20.6%). There was no significant difference in CA19–9 levels between the two cohorts.

Table 1

Table 1. Baseline clinicopathological characteristics of the training and validation cohort.

A total of 635 patients were classified into four TCCA score groups (0 to 3) in the training cohort as summarized in Table 2. The mean age increased from 53.0 to 56.1 years as rising TCCA score, though the differences were not statistically significant (p = 0.068). There was also a non-significant trend towards more female patients in the higher TCCA groups (p = 0.101). Higher TCCA scores correlated with a decreased prevalence of HBsAg-positive status, elevated NLR, GAR, CEA, CA19-9, TBS, and a higher prevalence of macrovascular invasion, satellite nodules, lymph node metastasis, and advanced TNM stage (p < 0.05 for all). Additionally, patients with higher TCCA scores had larger tumors, underwent more anatomical resections, but had narrower resection margins and received more adjuvant chemotherapy (p < 0.05 for all).

Table 2

Table 2. Demographic and clinical characteristics by TCCA score (0 to 3) in the training cohort.

Independent prognostic factors for OS and RFS in the training cohorts

In the multivariate Cox model for OS in the training cohort, significant independent predictors included GAR >6.46 (HR: 1.36, 95% CI: 1.05−1.78, p = 0.022), CEA >5.0 ng/mL (HR: 1.42, 95% CI: 1.13−1.78, p = 0.003), CA19-9 >37 U/mL (HR: 1.46, 95% CI: 1.16−1.83, p = 0.001), TBS >6.0 units (HR: 1.68, 95% CI: 1.35−2.11, p < 0.001), macrovascular invasion (HR: 1.54, 95% CI: 1.16−2.04, p = 0.003), satellite nodules (HR: 1.36, 95% CI: 1.08−1.73, p = 0.010), and lymph node status (N1 vs. N0/Nx: HR: 1.73, 95% CI: 1.34−2.23, p < 0.001) (Table 3). For RFS, the independent predictors were GAR >6.46 (HR: 1.30, 95% CI: 1.03−1.70, p = 0.028), CEA >5.0 ng/mL (HR: 1.25, 95% CI: 1.01−1.56, p = 0.042), CA19-9 >37 U/mL (HR: 1.47, 95% CI: 1.19−1.81, p < 0.001), TBS >6.0 units (HR: 1.55, 95% CI: 1.26−1.91, p < 0.001), and lymph node status (N1 vs. N0/Nx: HR: 1.63, 95% CI: 1.29−2.06, p < 0.001) (Table 4).

Table 3

Table 3. Univariable and multivariable analyses for OS in the training cohort.

Table 4

Table 4. Univariable and multivariable analyses for RFS in the training cohort.

Effects of TCCA score on OS and RFS

At the end of the last follow-up, a total of 350 (55.1%) patients died and 400 (63.0%) patients recurred in the training cohort. Patients with higher TCCA score had an incremental worse OS (1−, 3−, and 5−year OS: score 0, 89%, 68%, 50%; vs. score 1, 79%, 46%, 33%; vs. score 2, 63%, 28%, 18%; vs. score 3, 48%, 12%, 8%; respectively, p < 0.001) and RFS (1−, 3−, and 5−year OS: score 0, 70%, 45%, 41%; vs. score 1, 54%, 33%, 28%; vs. score 2, 38%, 23%, 13%; vs. score 3, 26%, 12%, 3%; respectively, p < 0.001) (Figures 1a, b, Supplementary Table S1). In the validation cohort, Kaplan-Meier survival curves demonstrated significant differentiation in each group, with all p < 0.001 (Figures 1c, d, Supplementary Table S1).

Figure 1

Figure 1. KM curves of overall survival (OS) and recurrence-free survival (RFS) of ICC patients stratified by TCCA model in the training cohort (a, b) and in the validation cohort (c, d). TCCA score, Tumor Burden Score, carcinoembryonic antigen, and carbohydrate antigen 19–9 combined score.

Comparison and validation of the TCCA model

The ROC curve of TCCA model showed an AUC value of 0.70 (95% CI = 0.67−0.73) for OS and 0.65 (95% CI = 0.61−0.69) for RFS in the training cohort (Figures 2a, b). In the validation cohort, TCCA model demonstrated moderate prognostic prediction capabilities (OS: AUC 0.67, 95% CI = 0.63−0.71; RFS: 0.63, 95% CI = 0.59−0.68) (Figures 2c, d). For calibration of the TCCA model, calibration plots depicted a good consistency between the predicted outcome and the observed outcome of the model in terms of 5−year OS and RFS in the training and validation cohorts (Supplementary Figure S1).

Figure 2

Figure 2. ROC curves of the TCCA model for overall survival (OS) and recurrence-free survival (RFS) in ICC Patients: Training (a, b) and Validation Cohorts (c, d). TCCA, combination of Tumor Burden Score, carcinoembryonic antigen, and carbohydrate antigen 19-9.

The TCCA model demonstrated superior discriminative ability for predicting outcomes, with the highest C-index of 0.734 (95% CI: 0.711−0.757), compared to the AJCC 8^th TNM staging system (C-index: 0.599), TBS-CEA score (C-index: 0.648), and TBS-CA19–9 score (C-index: 0.660). Additionally, the TCCA model yielded the lowest AIC value (3840.8), indicating a better model fit than the other systems. All comparisons were statistically significant (p < 0.001) (Supplementary Table S2).

TCCA model performance in CEA- and CA19-9-negative subgroups

In the validation cohort, 80.0% and 46.3% of patients were negative for CEA and CA19-9, respectively. In the CEA-negative subgroup, the median OS for patients with TCCA scores of 0, 1, and 2 was 50.3, 30.3, and 14.8 months, respectively, while the median RFS were 40.6, 17.1, and 8.3 months (all p < 0.001) (Figures 3a, b). In the CA19-9-negative subgroup, the median OS for TCCA scores of 0, 1, and 2 was 50.3, 30.7, and 17.7 months (p = 0.033), and the median RFS were 40.6, 13.4, and 4.9 months (p = 0.004) (Figures 3c, d).

Figure 3

Figure 3. KM curves of overall survival (OS) and recurrence-free survival (RFS) of ICC patients stratified by the TCCA model in biomarker-negative subgroups of the validation cohort: Carcinoembryonic antigen (CEA)-negative subgroup (a, b) and Carbohydrate antigen 19-9 (CA19-9)-negative subgroup (c, d). TCCA score, Tumor Burden Score, CEA, and CA19–9 combined score.

Discussion

The proposed TCCA model represents a novel and comprehensive integration of tumor burden and tumor biology. While previous studies have linked TBS with either CA19–9 or CEA, the TCCA model is the first to combine all three preoperative indicators (TBS, CEA and CA19-9), reflecting both tumor morphology and biomarker-derived aggressiveness. This integration enhances prognostic accuracy, particularly among patients with negative CA19–9 or CEA levels, and reduces the risk of underestimating tumor aggressiveness when relying on a single biomarker. The TCCA model demonstrated superior discrimination and calibration performance (C-index 0.734) compared with previously established scores (TBS−CA19-9: 0.660; TBS−CEA: 0.648) and the AJCC−TNM staging system (0.599). In addition to its good prognostic performance, the TCCA model is simple and clinically practical, as it can be calculated preoperatively using routine imaging and serum data, enabling surgeons to provide individualized prognostic counseling during preoperative discussions.

The TCCA model incorporates three preoperative factors: TBS, CEA, and CA19-9, all of which were identified as independent risk factors for OS and RFS in the training cohort. Many studies have confirmed that both tumor size and tumor number are important prognostic factors, as reflected in the 8^th edition of AJCC T-staging system (20). However, these two parameters are categorized dichotomously with arbitrary cutoffs, which may limit their ability to assess prognosis accurately and provide personalized treatment recommendations for resectable ICC patients (21). TBS, a composite metric of tumor morphology and a continuous variable, may better reflect total tumor burden in relation to survival outcomes. The concept of TBS originates from the ‘Metro-ticket’ system, initially applied to patients with colorectal liver metastases, and was later shown to have a significant inverse relationship with OS in HCC patients undergoing liver transplantation (19, 22). Multivariate analysis in this study showed that high TBS was associated with a 1.68-fold increased risk of death (HR 1.68, 95% CI 1.35−2.11, p < 0.001) and a 1.52-fold increased risk of recurrence (HR 1.52, 95% CI 1.23−1.88, p < 0.001). In addition to tumor morphology, serum CEA and CA19–9 are recognized as surrogate markers of tumor biology and robust predictors of long-term outcomes in ICC patients (23). These well-established ICC biomarkers are widely used in clinical practice and can be easily assessed preoperatively. In our multivariate analysis, elevated levels of both CEA and CA19–9 were independently associated with worse OS and RFS, with elevated CEA levels linked to a 1.42-fold increased risk of death (HR 1.42, 95% CI 1.13−1.78, p = 0.003) and a 1.25-fold increased risk of recurrence (HR 1.25, 95% CI 1.01−1.56, p = 0.042), while elevated CA19–9 levels were associated with a 1.46-fold increased risk of death (HR 1.46, 95% CI 1.16−1.83, p = 0.001) and a 1.45-fold increased risk of recurrence (HR 1.45, 95% CI 1.17−1.79, p < 0.001). Our results demonstrate that all three variables—TBS, CEA, and CA19-9—serve as important prognostic indicators, providing valuable insights for predicting outcomes in ICC patients.

Currently, various preoperative models based on TBS have been developed, and the combination of TBS with other clinical variables exhibited enhanced predictive efficacy for patients following ICC resection. Munir et al. demonstrated an interplay between TBS and ALBI grade, revealing that patients with both high TBS and ALBI grade experienced significantly higher 2-year recurrence (84.6%) and 5-year mortality (94.6%) compared to those with both low TBS and ALBI grade (24). Similarly, Wang et al. constructed a novel index combining TBS with the albumin-to-alkaline phosphatase ratio (AAPR), which effectively stratified postoperative survival outcomes in ICC patients undergoing curative resection, particularly in predicting postoperative recurrence, with AUC values of 0.653 for OS and 0.658 for RFS (25). Additionally, Zhang et al. integrating TBS with more inflammatory and nutritional markers, such as AAPR, albumin–globulin ratio (AGR), and monocyte-to-lymphocyte ratio (MLR), to construct the TIIN score can improve predictive accuracy—with a 3-year OS AUC of 0.728 in the training cohort and 0.695 in the validation cohort—this approach also introduces greater complexity (26). Our multivariate analysis also identified GAR as an independent prognostic factor for OS and RFS. However, given that inflammatory and nutritional markers are highly susceptible to variations in infection status and nutritional status, limiting their capacity to accurately reflect tumor biology (27).

Incorporating tumor-specific biomarkers into prognostic models may be a better choice than relying on inflammatory and nutritional markers. CA19–9 and CEA are recommended as tumor markers for the early detection and diagnosis of ICC by the Chinese expert consensus on management of intrahepatic cholangiocarcinoma (2022 edition) and the Liver Cancer Study Group of Japan Clinical Practice Guidelines (28). Previous studies have developed prognostic models by combining TBS with CA19-9, consistently demonstrating that Elevated TBS and CA19–9 levels was associated with poor prognosis (18, 29). However, approximately 10% of the population are genotypically negative for Lewis blood group antigen and therefore unable to synthesize CA19-9, limiting its utility as a tumor marker in all patients (30). Integrating preoperative CEA into the prognostic models may enhance the stratification of postoperative prognosis in ICC patients.

The TCCA model, which integrates three preoperative variables—TBS, CEA, and CA19-9—was developed without the inclusion of commonly used postoperative pathological variables. Although our study confirmed that variables such as macrovascular invasion, satellite nodules, and lymph node metastasis are closely related to OS in multivariate analysis. Lymph node metastasis, in particular, is widely regarded as the strongest independent prognostic factor for long-term outcomes following ICC resection (31). Nevertheless, we believe that the TCCA model, by combining preoperative tumor morphology and biological markers, offers a sufficiently comprehensive reflection of the biological behavior of ICC. As shown in Table 2, higher TCCA scores were associated with a lower prevalence of HBsAg positivity, a known protective factor for ICC prognosis (32). Additionally, as TCCA scores increased, inflammatory markers such as NLR and GAR also increased incrementally, while the nutritional markers PNI decreased—trends that were all associated with poorer prognosis. Furthermore, higher TCCA scores were correlated with increased rates of macrovascular invasion, microvascular invasion, satellite nodules, and lymph node metastasis, suggesting more advanced tumor biology and worse outcomes. Thus, the TCCA model demonstrates correlations not only with inflammatory and nutritional markers but also with common postoperative pathological variables, underscoring its potential as a valuable tool for prognostic assessment.

The TCCA model offers several key advantages. First, it allows for the straightforward preoperative assessment of each parameter, as imaging and tumor biomarker testing are standard components of preoperative evaluation, and these markers are also routinely monitored during postoperative follow-up in ICC patients in China. Second, the cutoff values for TBS, CEA, and CA19–9 are clinically relevant and practical. The TBS cutoff was set at 6 units, and based on the Pythagorean theorem, it can be quickly deduced that any solitary tumor with a diameter of 6 cm or more will exceed this threshold, classifying the patient into the high TBS group. The cutoff values for CEA and CA19–9 were aligned with the upper limits of normal laboratory reference ranges, minimizing variability and potential errors associated with differing cutoff values across studies. Third, the TCCA model effectively stratifies patients based on OS and RFS, offering clinicians valuable guidance for preoperative decision-making. For patients with lower TCCA scores, more aggressive resection may be recommended to improve long-term outcomes, while those with higher scores could benefit more from postoperative adjuvant chemotherapy. This underscores the importance of preoperatively identifying high-risk patients to initiate timely adjuvant chemotherapy.

Several limitations must be considered when interpreting the data from this study. First, as with all retrospective studies, there is potential for selection bias in selecting patients for surgical resection. Second, different etiologies of ICC may lead to variations in CEA and CA19–9 levels. Given the study population was limited to China, where HBV infection rates are higher, further research is necessary to assess the applicability of this model to Western populations. Additional, CA19–9 is particularly influenced by biliary obstruction, and monitoring dynamic changes in these biomarkers may provide more accurate and reliable information (33). Finally, the AUC of the TCCA model ranged from 0.63 to 0.70, indicating that there remains room for improvement. Future studies may further enhance the model by integrating additional preoperative data, such as selected radiomic features or circulating tumor DNA, to better capture tumor heterogeneity and biological behavior (34).

Conclusion

The TCCA model, which integrates tumor morphology and biological markers (TBS, CEA, and CA19-9), provides a reliable and practical preoperative tool for predicting survival outcomes in patients with ICC. Its simplicity and accessibility make it suitable for routine clinical application, assisting clinicians in risk stratification and optimization of treatment strategies for ICC patients before hepatectomy.

Data availability statement

The datasets used and/or analyzed during the current study are available from the corresponding author on reasonable request. Requests to access these datasets should be directed to Yongyi Zeng, bGFtcDE5NzMxMUAxMjYuY29t.

Ethics statement

The studies involving humans were approved by Institutional Review Boards of Mengchao Hepatobiliary Hospital of Fujian Medical University. The studies were conducted in accordance with the local legislation and institutional requirements. Written informed consent for participation was not required from the participants or the participants’ legal guardians/next of kin in accordance with the national legislation and institutional requirements.

Author contributions

JF: Writing – review & editing, Writing – original draft. TH: Formal Analysis, Writing – review & editing. QL: Writing – review & editing, Methodology. JLi: Writing – review & editing, Data curation. XB: Writing – review & editing, Data curation. JMW: Data curation, Writing – review & editing. FL: Writing – review & editing, Data curation. JW: Writing – review & editing, Data curation. KW: Writing – review & editing, Data curation. JLo: Writing – review & editing, Data curation. SC: Data curation, Writing – review & editing. YZ: Writing – original draft, Writing – review & editing, Conceptualization.

Funding

The author(s) declared that financial support was received for this work and/or its publication. This study was supported by the National Key Research and Development Program of China (2022YFC2407304); key Clinical Specialty Discipline Construction Program of Fuzhou (20230101); Major Research Projects for Young and Middle-aged Researchers of Fujian Provincial Health Care Commission (2021ZQNZD013); and the Health Science and Technology Innovation Platform Program of Fuzhou (2021-S-wp1).

Conflict of interest

The author(s) declared that this work was conducted in the absence of any commercial or financial relationships that could be construed as a potential conflict of interest.

Generative AI statement

The author(s) declared that generative AI was not used in the creation of this manuscript.

Any alternative text (alt text) provided alongside figures in this article has been generated by Frontiers with the support of artificial intelligence and reasonable efforts have been made to ensure accuracy, including review by the authors wherever possible. If you identify any issues, please contact us.

Publisher’s note

All claims expressed in this article are solely those of the authors and do not necessarily represent those of their affiliated organizations, or those of the publisher, the editors and the reviewers. Any product that may be evaluated in this article, or claim that may be made by its manufacturer, is not guaranteed or endorsed by the publisher.

Supplementary material

The Supplementary Material for this article can be found online at: https://www.frontiersin.org/articles/10.3389/fonc.2026.1720482/full#supplementary-material

Supplementary Figure 1 | Calibration curves for TCCA model to predict overall survival (OS) and recurrence-free survival (RFS) in training cohorts (a, b). TCCA, combination of Tumor Burden Score, carcinoembryonic antigen, and carbohydrate antigen 19-9.

Supplementary Table 1 | OS and RFS data stratified by TCCA score in the training and validation cohort. Abbreviation:TCCA score, Tumor Burden Score, carcinoembryonic antigen, and carbohydrate antigen 19–9 combined score. OS, Overall survival; RFS, Recurrence-free survival.

Supplementary Table 2 | Comparison of the TCCA score with AJCC staging system and other scoring systems in the training cohort. Abbreviation: TCCA score, Tumor Burden Score (TBS), carcinoembryonic antigen (CEA), and carbohydrate antigen 19-9 (CA19-9) combined score; TBS−CEA Score, TBS and CEA combined score; TBS−CA19–9 Score, TBS and CA19–9 combined score; CEA−CA19–9 Score, CEA and CA19–9 combined score; AJCC, American Joint Committee on Cancer; AIC: Akaike information criterion.

References

1. Moris D, Palta M, Kim C, Allen PJ, Morse MA, and Lidsky ME. Advances in the treatment of intrahepatic cholangiocarcinoma: An overview of the current and future therapeutic landscape for clinicians. CA Cancer J Clin. (2023) 73:198–222. doi: 10.3322/caac.21759

PubMed Abstract | Crossref Full Text | Google Scholar

2. Bragazzi MC, Venere R, Ribichini E, Covotta F, Cardinale V, and Alvaro D. Intrahepatic cholangiocarcinoma: Evolving strategies in management and treatment. Dig Liver Dis. (2024) 56:383–93. doi: 10.1016/j.dld.2023.08.052

PubMed Abstract | Crossref Full Text | Google Scholar

3. Wilbur HC, Soares HP, and Azad NS. Neoadjuvant and adjuvant therapy for biliary tract cancer: Advances and limitations. Hepatology. (2024) 82(5):1287–302. doi: 10.1097/HEP.0000000000000760

PubMed Abstract | Crossref Full Text | Google Scholar

4. Dong L, Lu D, Chen R, Lin Y, Zhu H, Zhang Z, et al. Proteogenomic characterization identifies clinically relevant subgroups of intrahepatic cholangiocarcinoma. Cancer Cell. (2022) 40:70–87.e15. doi: 10.1016/j.ccell.2021.12.006

PubMed Abstract | Crossref Full Text | Google Scholar

5. Amin MB, Greene FL, Edge SB, Compton CC, Gershenwald JE, Brookland RK, et al. The Eighth Edition AJCC Cancer Staging Manual: Continuing to build a bridge from a population-based to a more “personalized” approach to cancer staging. CA Cancer J Clin. (2017) 67:93–9. doi: 10.3322/caac.21388

PubMed Abstract | Crossref Full Text | Google Scholar

6. Sa-Ngiamwibool P, Aphivatanasiri C, Sangkhamanon S, Intarawichian P, Kunprom W, Thanee M, et al. Modification of the AJCC/UICC 8th edition staging system for intrahepatic cholangiocarcinoma: proposal for an alternative staging system from cholangiocarcinoma-prevalent Northeast Thailand. HPB (Oxford). (2022) 24:1944–56. doi: 10.1016/j.hpb.2022.06.004

PubMed Abstract | Crossref Full Text | Google Scholar

7. Chen Y, Liu H, Zhang J, Wu Y, Zhou W, Cheng Z, et al. Prognostic value and predication model of microvascular invasion in patients with intrahepatic cholangiocarcinoma: a multicenter study from China. BMC Cancer. (2021) 21:1299. doi: 10.1186/s12885-021-09035-5

PubMed Abstract | Crossref Full Text | Google Scholar

8. Wang XY, Zhu WW, Lu L, Li YT, Zhu Y, Yang LY, et al. Development and validation of a mutation-annotated prognostic score for intrahepatic cholangiocarcinoma after resection: a retrospective cohort study. Int J Surg. (2023) 109:3506–18. doi: 10.1097/JS9.0000000000000636

PubMed Abstract | Crossref Full Text | Google Scholar

9. Bo Z, Chen B, Yang Y, Yao F, Mao Y, Yao J, et al. Machine learning radiomics to predict the early recurrence of intrahepatic cholangiocarcinoma after curative resection: A multicentre cohort study. Eur J Nucl Med Mol Imaging. (2023) 50:2501–13. doi: 10.1007/s00259-023-06184-6

PubMed Abstract | Crossref Full Text | Google Scholar

10. Fiz F, Masci C, Costa G, Sollini M, Chiti A, Ieva F, et al. PET/CT-based radiomics of mass-forming intrahepatic cholangiocarcinoma improves prediction of pathology data and survival. Eur J Nucl Med Mol Imaging. (2022) 49:3387–400. doi: 10.1007/s00259-022-05765-1

PubMed Abstract | Crossref Full Text | Google Scholar

11. Fu J, Chen Q, Lai Z, Lin K, Fang G, Ding Z, et al. A novel preoperative inflammation score system established for postoperative prognosis predicting of intrahepatic cholangiocarcinoma. BMC Cancer. (2023) 23:188. doi: 10.1186/s12885-023-10668-x

PubMed Abstract | Crossref Full Text | Google Scholar

12. Zhu J, Wang D, Liu C, Huang R, Gao F, Feng X, et al. Development and validation of a new prognostic immune-inflammatory-nutritional score for predicting outcomes after curative resection for intrahepatic cholangiocarcinoma: A multicenter study. Front Immunol. (2023) 14:1165510. doi: 10.3389/fimmu.2023.1165510

PubMed Abstract | Crossref Full Text | Google Scholar

13. Horvat N, Papanikolaou N, and Koh DM. Radiomics beyond the hype: A critical evaluation toward oncologic clinical use. Radiol Artif Intell. (2024) 6:e230437. doi: 10.1148/ryai.230437

PubMed Abstract | Crossref Full Text | Google Scholar

14. Cui H, Li Y, Li S, and Liu G. Prognostic utility of preoperative inflammatory markers in patients with intrahepatic cholangiocarcinoma after hepatic resection: A systematic review and meta-analysis. Cancer Med. (2023) 12:99–110. doi: 10.1002/cam4.4935

PubMed Abstract | Crossref Full Text | Google Scholar

15. Huang G, Xi P, Yao Z, Zhao C, Li X, Chen Z, et al. The clinical association between the inflammation-nutritional condition and prognosis of locally advanced intrahepatic cholangiocarcinoma after R0 resection: evidence from competing risk and propensity matching analysis. J Inflammation Res. (2024) 17:2787–99. doi: 10.2147/JIR.S460103

PubMed Abstract | Crossref Full Text | Google Scholar

16. Kobayashi K, Ono Y, Kitano Y, Oba A, Sato T, Ito H, et al. Prognostic impact of tumor markers (CEA and CA19-9) on patients with resectable colorectal liver metastases stratified by tumor number and size: potentially valuable biologic markers for preoperative treatment. Ann Surg Oncol. (2023) 30:7338–47. doi: 10.1245/s10434-023-13781-0

PubMed Abstract | Crossref Full Text | Google Scholar

17. Fu J, Zheng L, Tang S, Lin K, Zheng S, Bi X, et al. Tumor burden score and carcinoembryonic antigen predict outcomes in patients with intrahepatic cholangiocarcinoma following liver resection: a multi−institutional analysis. BMC Cancer. (2024) 24:358. doi: 10.1186/s12885-024-12091-2

PubMed Abstract | Crossref Full Text | Google Scholar

18. Moazzam Z, Alaimo L, Endo Y, Lima HA, Ruzzenente A, Guglielmi A, et al. Combined tumor burden score and carbohydrate antigen 19–9 grading system to predict outcomes among patients with intrahepatic cholangiocarcinoma. J Am Coll Surg. (2023) 236:804–13. doi: 10.1097/XCS.0000000000000557

PubMed Abstract | Crossref Full Text | Google Scholar

19. Sasaki K, Morioka D, Conci S, Margonis GA, Sawada Y, Ruzzenente A, et al. The tumor burden score: A new “Metro-ticket” Prognostic tool for colorectal liver metastases based on tumor size and number of tumors. Ann Surg. (2018) 267:132–41. doi: 10.1097/SLA.0000000000002064

PubMed Abstract | Crossref Full Text | Google Scholar

20. Yoon SJ, Park S, Kim H, Shin SH, Heo JS, Rhu J, et al. The oncologic implications of tumor multiplicity in intrahepatic cholangiocarcinoma: its prognostic value might be underestimated. Cancer Res Treat. (2023) 55:948–55. doi: 10.4143/crt.2023.290

PubMed Abstract | Crossref Full Text | Google Scholar

21. Kamarajah SK. Evaluation of the AJCC 8th edition staging system for pathologically versus clinically staged intrahepatic cholangiocarcinoma (iCCA): a time to revisit a dogma? A surveillance, epidemiology, and end results (SEER) analysis. J Gastrointest Cancer. (2019) 50:392–9. doi: 10.1007/s12029-018-0084-5

PubMed Abstract | Crossref Full Text | Google Scholar

22. Mazzaferro V, Sposito C, Zhou J, Pinna AD, De Carlis L, Fan J, et al. Metroticket 2.0 model for analysis of competing risks of death after liver transplantation for hepatocellular carcinoma. Gastroenterology. (2018) 154:128–39. doi: 10.1053/j.gastro.2017.09.025

PubMed Abstract | Crossref Full Text | Google Scholar

23. Moro A, Mehta R, Sahara K, Tsilimigras DI, Paredes AZ, Farooq A, et al. The impact of preoperative CA19–9 and CEA on outcomes of patients with intrahepatic cholangiocarcinoma. Ann Surg Oncol. (2020) 27:2888–901. doi: 10.1245/s10434-020-08350-8

PubMed Abstract | Crossref Full Text | Google Scholar

24. Munir MM, Endo Y, Lima HA, Alaimo L, Moazzam Z, Shaikh C, et al. Albumin-bilirubin grade and tumor burden score predict outcomes among patients with intrahepatic cholangiocarcinoma after hepatic resection: a multi-institutional analysis. J Gastrointest Surg. (2023) 27:544–54. doi: 10.1007/s11605-023-05578-z

PubMed Abstract | Crossref Full Text | Google Scholar

25. Wang S, Liu M, Xiang L, Qiu H, Cheng L, Huang Z, et al. Tumour burden score combined with albumin-to-alkaline phosphatase ratio predicts prognosis in patients with intrahepatic cholangiocarcinoma. J Cell Mol Med. (2024) 28:e18530. doi: 10.1111/jcmm.18530

PubMed Abstract | Crossref Full Text | Google Scholar

26. Zhang H, Huang G, Li Q, Wang Y, Yang Z, Chen P, et al. Construction and validation of a novel tumor morphology immune inflammatory nutritional score (TIIN score) for intrahepatic cholangiocarcinoma: a multicenter study. BMC Cancer. (2024) 24:630. doi: 10.1186/s12885-024-12375-7

PubMed Abstract | Crossref Full Text | Google Scholar

27. Lee SC, Kim SJ, Yu MH, Lee KJ, and Cha YS. Uses of inflammatory markers for differentiation of intrahepatic mass-forming cholangiocarcinoma from liver abscess: case-control study. J Clin Med. (2020) 9:3194. doi: 10.3390/jcm9103194

PubMed Abstract | Crossref Full Text | Google Scholar

28. Kubo S, Shinkawa H, Asaoka Y, Ioka T, Igaki H, Izumi N, et al. Liver cancer study group of Japan clinical practice guidelines for intrahepatic cholangiocarcinoma. Liver Cancer. (2022) 11:290–314. doi: 10.1159/000522403

PubMed Abstract | Crossref Full Text | Google Scholar

29. Huang G, Zhang H, Yang Z, Li Q, Yuan H, Chen P, et al. Predictive value of HTS grade in patients with intrahepatic cholangiocarcinoma undergoing radical resection: a multicenter study from China. World J Surg Oncol. (2024) 22:17. doi: 10.1186/s12957-023-03281-6

PubMed Abstract | Crossref Full Text | Google Scholar

30. Parra-Robert M, Santos VM, Canis SM, Pla XF, Fradera J, and Porto RM. Relationship between CA 19.9 and the lewis phenotype: options to improve diagnostic efficiency. Anticancer Res. (2018) 38:5883–8. doi: 10.21873/anticanres.12931

PubMed Abstract | Crossref Full Text | Google Scholar

31. Zhang XF, Xue F, Dong DH, Weiss M, Popescu I, Marques HP, et al. Number and station of lymph node metastasis after curative-intent resection of intrahepatic cholangiocarcinoma impact prognosis. Ann Surg. (2021) 274:e1187–95. doi: 10.1097/SLA.0000000000003788

PubMed Abstract | Crossref Full Text | Google Scholar

32. Li Z, Gao Q, Wu Y, Ma X, Wu F, Luan S, et al. HBV infection effects prognosis and activates the immune response in intrahepatic cholangiocarcinoma. Hepatol Commun. (2024) 8:e0360. doi: 10.1097/HC9.0000000000000360

PubMed Abstract | Crossref Full Text | Google Scholar

33. Singh S, Tang SJ, Sreenarasimhaiah J, Lara LF, and Siddiqui A. The clinical utility and limitations of serum carbohydrate antigen (CA19-9) as a diagnostic tool for pancreatic cancer and cholangiocarcinoma. Dig Dis Sci. (2011) 56:2491–6. doi: 10.1007/s10620-011-1709-8

PubMed Abstract | Crossref Full Text | Google Scholar

34. Wen N, Peng D, Xiong X, Liu G, Nie G, Wang Y, et al. Cholangiocarcinoma combined with biliary obstruction: an exosomal circRNA signature for diagnosis and early recurrence monitoring. Signal Transduct Target Ther. (2024) 9:107. doi: 10.1038/s41392-024-01814-3

PubMed Abstract | Crossref Full Text | Google Scholar