Peripheral regulatory T cells trajectory predicts progression-free survival in hepatocellular carcinoma patients during on-demand transarterial chemoembolization using joint model: a prospective study

Introduction

Hepatocellular carcinoma (HCC), the most common primary liver cancer and a leading cause of cancer-related death worldwide, poses a huge health burden (1). Currently, the advent of immunotherapy has revolutionized the management of HCC (2). Consequently, a combination of transarterial chemoembolization (TACE) and systemic therapies has increasingly yielded encouraging results (2-4). As reported, TACE, which delivers chemotherapeutic drugs and embolization agents into the tumor vesseles, has a role to modulate immunosuppressive tumor immune microenvironment (TIME), thereby enhancing antitumor immunity (5). Given a limited objective response to immune checkpoint inhibitor therapy (6), we are in urgent need of clinical insight from the exploration of the mechanisms of synergistic effects with TACE and immunotherapy.

Regulatory T cells (Tregs), sustaining immune tolerance and preventing immune overactivation, play a critical role in shaping the immunosuppressive TIME in HCC. This lymphocyte subset highly expresses cytotoxic T-lymphocyte antigen-4 and suppresses antigen-presenting cells’ function, and responses to other immunosuppressive factors or cells (7). TACE-mediated change of Tregs and its prognostic value have been previously suggested (8,9). Studies have reported that lower baseline Tregs and a greater magnitude of peripheral Tregs reduction following a single TACE are associated with lower tumor progression risk and improved survival in HCC patients (10-12).

However, nearly all relevant studies share a few common limitations. First, most investigations exclusively focused on changes in Tregs following a single TACE, leaving the Tregs dynamics during repeated TACEs and their accumulated prognostic significance underexplored. TACE is typically performed as per either an on-demand or an on-schedule pattern in patients with unresectable HCC, who often present with a substantial tumor burden. In this context, the accumulative regulatory effect of repeated TACE on TIME should be explored. Second, previous studies predominantly utilized static Tregs value or stratified patients based on the mean or median difference in Tregs. Consequently, these findings stand on the assumption that the impact of covariates on outcome remains constant over time. In contrast, these effects may vary due to follow-up interventions and clinical events (13).

During periodic treatment, TIME evolution is a dynamic process, characterized by increasing spatial and temporal heterogeneity. This complexity highlights the clinical importance of dynamic assessment. In recent years, joint modeling of longitudinal and time-to-event outcomes has been employed in various cancer types and chronic diseases to identify a high-sensitivity biomarker (14-16). It was hypothesized that longitudinal Tregs measurements may be more informative for tumor response compared with static measurements during repeated TACE. This study was designed to profile the longitudinal Tregs trajectory during on-demand TACE and evaluate its clinical relevance in predicting tumor progression in HCC patients. We present this article in accordance with the REMARK reporting checklist (available at https://jgo.amegroups.com/article/view/10.21037/jgo-2025-1-1035/rc).

Methods

Study population

All consecutive treatment-naïve HCC patients who were treated in the Interventional Radiology Department of Beijing Friendship Hospital were prospectively included in this study from March to December 2024. The inclusion criteria were as follows: (I) HCC confirmed with histology or radiology according to the standards of American Association for the Study of Liver Diseases (17); (II) Eastern cooperative oncology group performance status 0 or 1; (III) Child-Pugh A or B; (IV) Barcelona Clinic Liver Cancer (BCLC) stage B and A (patients who were unable or unwilling to undergo hepatectomy or ablation due to an multiple tumor located in different liver lobes, advanced age, or advanced cirrhosis). The exclusion criteria were as follows: (I) prior treatment for HCC; (II) incomplete clinical and imaging data; (III) informed consent was not acquired. Figure 1 illustrates a patient recruitment flowchart.

Figure 1 Patient inclusion flowchart. BCLC, Barcelona Clinic Liver Cancer; HCC, hepatocellular carcinoma.

Ethical statement

This study was conducted in accordance with the Declaration of Helsinki and its subsequent amendments. The study was approved by the Bioethics Committee of Beijing Friendship Hospital, Capital Medical University (No. 2024-P2-379-01) and informed consent was taken from all the patients.

On-demand TACE

Based on preoperative evaluation of liver function, in principle, each TACE procedure was performed to treat all viable lesions. Contrasted-enhanced multiple-phase magnetic resonance imaging (CE-MRI) was performed 4–8 weeks after the first TACE. Subsequent TACE was repeated 1–2 weeks after imaging evaluation if residual tumors were present. The tumor was supervised by CE-MRI every 3 months, and TACE was performed as appropriate. Either conventional TACE (cTACE) or drug-eluting beads (DEB) TACE was chosen at the patient level based on tumor characteristics and liver function reserve at the discretion of the board-certified clinicians. For each TACE, the dose of doxorubicin was 50–75 mg/m² and was adjusted according to the patient’s characteristics. Doxorubicin was mixed into 5–20 mL of Lipiodol (Lipiodol Ultra-Fluide, Laboratoire Guerbet, Villepinte, France) or HepaSphere™ microspheres (Merit Medical, Utah, USA) with a diameter of 30–60 µm. When tumor progression occurred, the combination with or conversion to systemic therapies was considered. The protocol of systemic therapy was determined by a multidisciplinary tumor board, including interventional radiologists, hepatologists, radiologists and immunologists in oncology. The TACE procedure is detailed in Appendix 1.

Tregs measurement

Four blood samples were collected at 1 day before initial TACE (T0), within 1 week after TACE (T1), 4–8 weeks after initial TACE (T2, or 1 day before second TACE if necessary) and 9–16 weeks after initial TACE (T3, or 1 day before third TACE if necessary). The blood collection procedure was illustrated in Figure S1. Tregs measurement were stopped at the timepoint of receiving systemic agents or withdrawal of consent. Linear mixed model allowed for missing data in the repeated measurements, but in this study, a missing Treg value was allowed at most one timepoint.

Peripheral blood mononuclear cells were isolated from fresh blood by Ficoll-Paque density gradient centrifugation. All monoclonal antibodies were from BioLegend company for flow cytometry, including FITC anti-human CD3, PerCP/Cyanine5.5 anti-human CD4, APC anti-human CD25, PE anti-human CD127 (IL-7Rα) (18). Before staining, human TruStain FcX™ was used to reduce non-specific binding. FACS Express 3.0 software was used for data analysis. Gate strategy and representative flow dots of CD4⁺CD25⁺CD127dim/⁻ Tregs were provided in Figure S2.

Outcome

The primary outcome was progression-free survival (PFS), defined as the time from TACE initiation to disease progression, or death from any cause in the absence of progression, irrespective of withdrawal from study. Tumor response on radiologic evaluation was determined by the modified Response Evaluation Criteria in Solid Tumor (19). Safety was evaluated using the National Cancer Institute Common Terminology Criteria for Adverse Events version 5.0.

Data collection

Child-Pugh score was assessed as a liver function index. The repeated measurement of time-varying variables, including Tregs, Child-Pugh score and alpha-fetoprotein (AFP), was recorded until the censor date. The demographic data and other laboratory tests and embolization techniques were recorded at baseline as time-independent variables. Given the time span of Tregs measurements, age was time-independent.

Joint model

To model the association between the longitudinal trajectory of time-variant variables and PFS, a joint model was fitted, which combined a linear mixed effect model with random intercept and slope for the longitudinal trajectory and a Cox proportional hazards model for the time to event data.

The linear mixed-effects model accounted for both fixed and random effects, accommodating unmatched timepoints and unequal time intervals across patients to address subgroup or patient-level variations. To sufficiently capture time-varying effects, the full model considered all time-varying variables and their interactions. Model simplification was then performed in a step-wise way. Non-significant interaction terms were removed first. Then, the random effects were refined by sequentially removing non-significant random effects based on the Bayesian Information Criterion. Parameters for the joint model were estimated simultaneously using a full maximum likelihood approach. The proportional hazards assumption for the Cox component was verified using Schoenfeld residual plots, with log-transformation applied if necessary. The unstandardized coefficients of the joint models were used to quantify the association between Treg changes over time and the risk of endpoint. The unstandardized coefficient was used for calculating hazard ratio (HR) and its and 95% confidence interval (CI). The detailed model construction, estimation and tuning is provided in Appendix 1.

Statistical analysis

Continuous variables expressed as the mean ± standard deviation or median [interquartile range (IQR)] were compared by Student t test or Mann-Whitney U test, according to the normality after the Shapiro test; Categorical variables expressed as number (percentage) were compared by χ² test or Fisher’s exact test. Values at different timepoints were compared using repeated-measured data ANOVA, with P values corrected for multiple testing by the false discovery rate via the Benjamini and Hochberg procedure (q-value <0.05). Additional HR and its 95% CI of risk factor were calculated in joint models. A variable with a P value <0.05 in the univariate joint model was the candidate for the multivariate joint model. The joint models fitting was performed using JMBayes package in R (version 4.3.2). P value <0.05 was indicative of a significant difference.

Results

Patient cohort

A total of 30 HCC patients with 109 Treg measurements were prospectively included in the study. Patient-specific Tregs trajectories in progressive disease (PD) and non-PD groups is illustrated in Figure 2. At the T1 time point, 11 Treg measurements were missing. Hepatitis B virus (HBV) was the most common cause of liver disease, followed by hepatitis C virus. The baseline and time-varying variables of the patients are summarized in Table 1 and Table S1, respectively. All complications following the included TACE procedures were tolerated, including post-embolization syndrome, mild fever and vomiting.

Figure 2 Patient-specific Tregs trajectories during treatment in progressive disease group (red line) and non-progressive disease group (blue line). T0, 1 day before initial TACE; T1, within 1 week after TACE; T2, 4–8 weeks after initial TACE (or 1 day before second TACE if necessary); T3, 9–16 weeks after initial TACE (or 1 day before third TACE if necessary). TACE, transarterial chemoembolization; Tregs, regulatory T cells.

No significant differences in baseline variables between PD and non-PD groups

During a median follow-up period of 13.3 (IQR, 10.8–15.3) months, PD was observed in 11 patients. Among them, two patients experienced local tumor progression, six developed intrahepatic distant metastases, two had macrovascular invasion, and one developed lung metastasis. Patients underwent an average of 3.5 (IQR, 3–4) TACEs, with a median PFS of 11.1 (IQR, 9.3–13.7) months. The 6-month and 1-year PFS rates were 79.7% and 55.0%, respectively. No significant differences in baseline variables were found between PD and non-PD groups.

Tregs trajectory and associated factors

Tregs value at T3 was significantly lower than that at T0 [5.84 (IQR, 4.97–8.40) vs. 6.96 (IQR, 5.28–10.37), P=0.007; Figure 3A], Tregs at T2 were lower than baseline value but did not achieve significance (P=0.057). Among 19 patients with available four Tregs measurements, no significant differences in Tregs were observed between T1 and other timepoints (Figure 3B). Numerically, in non-PD patients, Tregs decreased by an absolute value of 1.66 (95% CI: 0.67–2.65) or a percent of 19.20% (95% CI: 6.9–31.5) per month. In contrast, in PD patients, Tregs slightly increased by an absolute value of 0.0715 (95% CI: −1.49 to 1.64) or a percent of 5.52% (95% CI: −11.3 to 22.4) per month.

Figure 3 Comparisons of Tregs among multiple timepoints. (A) In all patients, Tregs value at T3 is significantly lower than it at T0. No statistically significant difference was observed between T0 and T2 timepoint. (B) In 19 patients with T1 measurements, no significant differences in Tregs between T1 and other measurements were found. T0, 1 day before initial TACE; T1, within 1 week after TACE; T2, 4–8 weeks after initial TACE (or 1 day before second TACE if necessary); T3, 9–16 weeks after initial TACE (or 1 day before third TACE if necessary). TACE, transarterial chemoembolization; Tregs, regulatory T cells.

Longitudinal changes in Tregs were primarily associated with time, while no other examined clinical factors demonstrated a statistically significant influence on their trajectory (Table 2). To compare the tumor response in Tregs trajectories between different levels of BCLC stage, sex, embolization technique, candidate covariates and their interaction terms with time were included in the linear mixed models (Figures S3-S5, Tables S2-S4). It was demonstrated that all variables and their interaction terms with time had no significant effects on the Tregs trajectory.

Association between Treg trajectory and tumor progression risk

The trajectories of Child-Pugh score, AFP, Tregs and other time-independent variables were considered as candidate variables for the joint model. As shown in Table 3, in multivariate joint model, the trajectory of Tregs was predictive of PFS. It was found that for each 1% increase in the Tregs, HR for the PD occurrence was 2.04 (95% CI: 1.31–3.92; P<0.001). Compared with BCLC A, BCLC B was associated with higher PD risk (HR: 3.80; 95% CI: 1.55–8.06; P=0.02).

Personalized prediction of the impact of Tregs evolvement on tumor progression

A representative case with a specific Tregs trajectory is illustrated in Figure 4. A Tregs trajectory with an increasing trend was observed in a 69-year-old male patient with BCLC stage B HCC. At the T3 timepoint, his estimated 10-month progression-free probability was approximately 55% (Figure 4A). With Tregs further elevating at T4 point, his estimated 10-month progression-free probability decreased to around 50%, and the whole curve had a steeper descending slope (Figure 4B). Eventually, this patient developed intrahepatic recurrence after 11.5 months. Additional representative cases with other specific Treg trajectories are provided in Figure S6. Detailed analytic code is provided in Appendix 2.

Figure 4 A specific case with a representative Tregs trajectory (asterisks, the left longitudinal axis) and its responding probability of progression-free survival (right longitudinal axis). A male aged 69 years with BCLC B stage HCC has an increasing trend of Tregs values. At T3, his estimated 10-month progression-free probability is around 55% (A), but further decreases to about 50% when Tregs value elevated at T4 (B). As illustrated, the decreasing slope of whole curve of progression-free survival became pronounced with Tregs increasing during treatment. *, P<0.05. BCLC, Barcelona Clinic Liver Cancer; HCC, hepatocellular carcinoma.

Discussion

This study profiled the longitudinal changes in Tregs, a key immunosuppressive marker, in HCC patients who underwent on-demand TACE, and furthermore unveiled the association between the Tregs trajectory and tumor progression risk.

Regarding baseline characteristics, there were no differences between PD and non-PD groups, likely highlighting the limitations of traditional statistical methods in identifying prognostic factors based on static information. Specifically, the relatively small sample reduced statistical power to detect differences between the two groups. In addition, the BCLC B stage is widely accepted to be heterogeneous, with varying reported clinical outcomes (20). Indeed, some BCLC stage A tumors also exhibit considerable heterogeneity, particularly those with a single large lesion or multiple nodules. In this study, half of the included cases were beyond the up-to-seven criteria. As a result, traditional prognostic factors, such as tumor characteristics, may not differ significantly between patients with and without PD.

Patients with intermediate-stage HCC often experience repeated TACE. The cumulative immunoregulatory effects of repeated TACE sessions are not equivalent to the additive effect of a single TACE session. In such clinical scenarios, focusing solely on perioperative Tregs changes for a single TACE session does not accord with the real-world practice. Our findings demonstrated that Tregs value at T3, after an average of two TACE sessions, was significantly lower than the baseline value at T0. Furthermore, the Tregs value at T2 showed a decreasing trend compared to baseline. These results suggest that a more pronounced downregulation of immunosuppressive function was achieved after two consecutive TACE sessions compared with a single session. However, within the overall cohort exhibiting a decreasing trend in Tregs, there was significant inter-individual variability in the Tregs trajectory. PD patients showed an increasing trend in the Tregs trajectory, whereas non-PD patients showed a decreasing trend. The joint model results indicated that for every 1% increase in the Tregs trajectory, the risk of tumor progression increased by approximately 1.04 times. The possible mechanism was that the cases that failed to show a decreasing Tregs trajectory after TACE may be associated with more immune exhaustion. A study demonstrated that the baseline peripheral Tregs frequency in the earlier HCC stage was lower than that in the advanced stage, and the reduction after a single TACE was associated with the increased risk of tumor progression (21). In addition, at the T1 timepoint, the Tregs value did not differ significantly from baseline, which was consistent with findings from an animal experiment (9).

Tregs’ trajectory was entirely time-dependent and unaffected by other potential variables, such as demographic or tumor characteristics and embolization technology. Interaction analyses were conducted to explore potential factors influencing Tregs changes. These findings further reinforce the robustness and validity of the linear mixed model used in this study.

To the best of our knowledge, this may be the first time to characterize Tregs’ trajectory using a joint model to explore its prognostic value in the HCC population. The model utilized repeated patient-specific data to fit a linear mixed model and incorporated survival data to investigate the prognostic role of biomarker dynamics. While the trajectory of AFP was also analyzed within the joint model, negative results were observed across different models. In the limited application of joint models in HCC domain, AFP has been the commonly used serum marker to predict tumor progression or stratify survival (22,23), primarily because it is easily and repeatedly measurable. However, its predictive accuracy is limited by the fact that a substantial proportion of HCC cases are AFP-negative (14/30 cases in this study). Currently, tumor response is primarily evaluated through radiologic imaging. For cases classified as “stable disease” or showing no significant “partial response”, it is challenging to interpret the mismatch between radiologic evaluations and the magnitude of AFP changes when using a simple cutoff value. In addition, in advanced HCC, the broad range of absolute AFP values and cases where AFP exceeds the upper limit of normal can reduce the sensitivity of predictive models. To address this, this study employed a log-transformed AFP value for analysis. Moreover, AFP response was heavily influenced by the pattern of tumor progression. For cases with intrahepatic progression or small-sized lung metastases, minor AFP responses may be difficult to detect, often easily affected by systematic measurement errors, and consequently masked.

Combination treatments should be implemented in a time-conscious manner, and cellular and molecular changes should be monitored to optimally exploit the interactions between the individual therapies (24). In this context, the accumulative regulatory effect of repeated TACE on TIME is of paramount importance. Our findings provided evidence on longitudinal Tregs trajectory as a novel monitoring tool for tumor progression prediction, a concept that may be integrated into treatment allocation strategies.

Several limitations of this study should be noted. First and primarily, the robustness of this study was limited by the small sample size, which reduced the generalizability of the findings. Future studies with a larger sample size and longer follow-up are needed to validate and extend these results. Second, specific causes of chronic liver disease are associated with a distinct immune landscape (2). Despite dominance of HBV-related HCC in this study, further investigations are needed in cohorts with an exclusive liver disease etiology. Lastly, the study lacked control groups to compare Treg trajectories in healthy populations or those with liver disease. Future research, including animal experiments, is warranted to address these issues.

Conclusions

Longitudinal trajectory of peripheral Tregs independently predicted PFS in HCC patients who underwent on-demand TACE. This study profiled the cumulative immunoregulatory effects of repeated TACE procedures and established the prognostic relevance of tracking Tregs trajectory.

Acknowledgments

None.

Footnote

Reporting Checklist: The authors have completed the REMARK reporting checklist. Available at https://jgo.amegroups.com/article/view/10.21037/jgo-2025-1-1035/rc

Data Sharing Statement: Available at https://jgo.amegroups.com/article/view/10.21037/jgo-2025-1-1035/dss

Peer Review File: Available at https://jgo.amegroups.com/article/view/10.21037/jgo-2025-1-1035/prf

Funding: This project was supported by the Seed Fund of Beijing Friendship Hospital, Capital Medical University in China under Grant YYZZ202248.

Conflicts of Interest: All authors have completed the ICMJE uniform disclosure form (available at https://jgo.amegroups.com/article/view/10.21037/jgo-2025-1-1035/coif). The authors have no conflicts of interest to declare.

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. This study was conducted in accordance with the Declaration of Helsinki and its subsequent amendments. The study was approved by the Bioethics Committee of Beijing Friendship Hospital, Capital Medical University (No. 2024-P2-379-01) and informed consent was taken from all the patients.

Open Access Statement: This is an Open Access article distributed in accordance with the Creative Commons Attribution-NonCommercial-NoDerivs 4.0 International License (CC BY-NC-ND 4.0), which permits the non-commercial replication and distribution of the article with the strict proviso that no changes or edits are made and the original work is properly cited (including links to both the formal publication through the relevant DOI and the license). See: https://creativecommons.org/licenses/by-nc-nd/4.0/.

References

1. Singal AG, Lampertico P, Nahon P. Epidemiology and surveillance for hepatocellular carcinoma: New trends. J Hepatol 2020;72:250-61. [Crossref] [PubMed]
2. Llovet JM, Castet F, Heikenwalder M, et al. Immunotherapies for hepatocellular carcinoma. Nat Rev Clin Oncol 2022;19:151-72. [Crossref] [PubMed]
3. Jin ZC, Chen JJ, Zhu XL, et al. Immune checkpoint inhibitors and anti-vascular endothelial growth factor antibody/tyrosine kinase inhibitors with or without transarterial chemoembolization as first-line treatment for advanced hepatocellular carcinoma (CHANCE2201): a target trial emulation study. EClinicalMedicine 2024;72:102622. [Crossref] [PubMed]
4. Zhang N, Liu K, Yu Y, et al. Efficacy and safety of atezolizumab and bevacizumab with or without TACE as first-line therapy for unresectable HCC: a multicenter cohort study. Hepatol Int 2025;19:1382-96. [Crossref] [PubMed]
5. Karimi A, Yarmohammadi H, Erinjeri JP. Immune Effects of Intra-Arterial Liver-Directed Therapies. J Vasc Interv Radiol 2024;35:178-84. [Crossref] [PubMed]
6. Rimassa L, Finn RS, Sangro B. Combination immunotherapy for hepatocellular carcinoma. J Hepatol 2023;79:506-15. [Crossref] [PubMed]
7. Tanaka A, Sakaguchi S. Regulatory T cells in cancer immunotherapy. Cell Res 2017;27:109-18. [Crossref] [PubMed]
8. Han JW, Yoon SK. Immune Responses Following Locoregional Treatment for Hepatocellular Carcinoma: Possible Roles of Adjuvant Immunotherapy. Pharmaceutics 2021;13:1387. [Crossref] [PubMed]
9. Tischfield DJ, Gurevich A, Johnson O, et al. Transarterial Embolization Modulates the Immune Response within Target and Nontarget Hepatocellular Carcinomas in a Rat Model. Radiology 2022;303:215-25. [Crossref] [PubMed]
10. Li F, Guo Z, Lizée G, et al. Clinical prognostic value of CD4+CD25+FOXP3+regulatory T cells in peripheral blood of Barcelona Clinic Liver Cancer (BCLC) stage B hepatocellular carcinoma patients. Clin Chem Lab Med 2014;52:1357-65. [Crossref] [PubMed]
11. Park H, Jung JH, Jung MK, et al. Effects of transarterial chemoembolization on regulatory T cell and its subpopulations in patients with hepatocellular carcinoma. Hepatol Int 2020;14:249-58. [Crossref] [PubMed]
12. Ren Z, Yue Y, Zhang Y, et al. Changes in the Peripheral Blood Treg Cell Proportion in Hepatocellular Carcinoma Patients After Transarterial Chemoembolization With Microparticles. Front Immunol 2021;12:624789. [Crossref] [PubMed]
13. Holder AM, Dedeilia A, Sierra-Davidson K, et al. Defining clinically useful biomarkers of immune checkpoint inhibitors in solid tumours. Nat Rev Cancer 2024;24:498-512. [Crossref] [PubMed]
14. Rizopoulos D, Molenberghs G, Lesaffre EMEH. Dynamic predictions with time-dependent covariates in survival analysis using joint modeling and landmarking. Biom J 2017;59:1261-76. [Crossref] [PubMed]
15. Johnson M, Serra Traynor C, Vishwanathan K, et al. Longitudinal Circulating Tumor DNA Modeling to Predict Disease Progression in First-Line Mutant Epidermal Growth Factor Receptor Non-Small Cell Lung Cancer. Clin Pharmacol Ther 2024;115:349-60. [Crossref] [PubMed]
16. Lam L, Soret PA, Lemoinne S, et al. Dynamics of Liver Stiffness Measurement and Clinical Course of Primary Biliary Cholangitis. Clin Gastroenterol Hepatol 2024;22:2432-2441.e2. [Crossref] [PubMed]
17. Marrero JA, Kulik LM, Sirlin CB, et al. Diagnosis, Staging, and Management of Hepatocellular Carcinoma: 2018 Practice Guidance by the American Association for the Study of Liver Diseases. Hepatology 2018;68:723-50. [Crossref] [PubMed]
18. Yang L, Jia S, Shao X, et al. Interleukin-35 modulates the balance between viral specific CD4(+)CD25(+)CD127(dim/-) regulatory T cells and T helper 17 cells in chronic hepatitis B virus infection. Virol J 2019;16:48. [Crossref] [PubMed]
19. Lencioni R, Llovet JM. Modified RECIST (mRECIST) assessment for hepatocellular carcinoma. Semin Liver Dis 2010;30:52-60. [Crossref] [PubMed]
20. Raoul JL, Forner A, Bolondi L, et al. Updated use of TACE for hepatocellular carcinoma treatment: How and when to use it based on clinical evidence. Cancer Treat Rev 2019;72:28-36. [Crossref] [PubMed]
21. Wang F, Jing X, Li G, et al. Foxp3+ regulatory T cells are associated with the natural history of chronic hepatitis B and poor prognosis of hepatocellular carcinoma. Liver Int 2012;32:644-55. [Crossref] [PubMed]
22. An C, Wei R, Yao W, et al. Association of serum AFP trajectories and hepatocellular carcinoma outcomes after hepatic arterial infusion chemotherapy: A longitudinal, multicenter study. Cancer Med 2024;13:e7319. [Crossref] [PubMed]
23. Yang H, Lu L, Guo W, et al. A Longitudinal Study of AFP Trajectories and Clinical Outcomes in Intermediate-Stage Hepatocellular Carcinoma After Hepatectomy. J Hepatocell Carcinoma 2024;11:219-28. [Crossref] [PubMed]
24. Zemek RM, Anagnostou V, Pires da Silva I, et al. Exploiting temporal aspects of cancer immunotherapy. Nat Rev Cancer 2024;24:480-97. [Crossref] [PubMed]