The study, conducted by a surgical and oncology team at Chang-Gung Memorial Hospital and Chang-Gung University College of Medicine in Taoyuan, enrolled twenty-four patients with advanced hepatocellular carcinoma classified under the Barcelona Clinic Liver Cancer staging system. Five of the participants were in stage B with large or infiltrating tumors, while nineteen were in stage C, the most advanced category in the classification. The therapeutic strategy was deliberately two-phased. First, patients received between three and six courses of a checkpoint inhibitor, either nivolumab at 3 milligrams per kilogram of body weight or pembrolizumab at 2 milligrams per kilogram. Both agents block the PD-1 receptor, a checkpoint molecule that tumors exploit to disable attacking T cells. Then, after this priming phase, patients received three courses of autologous dendritic cell therapy, using immune cells harvested from their own peripheral blood monocytes, grown in the laboratory, and loaded with lysate prepared from the patient’s own tumor tissue.

The logic behind this sequence rests on a well-established principle of cancer immunology. Checkpoint inhibitors work best when the immune system has already recognized the tumor and mounted a T cell response against it, but many patients with hepatocellular carcinoma never generate such a response in the first place. Dendritic cells are the professional antigen-presenting cells of the immune system, the sentinels whose job is to capture foreign or abnormal proteins, process them into peptide fragments, and display those fragments to naive T cells in lymph nodes, thereby initiating targeted cytotoxic immunity. By pulsing laboratory-expanded dendritic cells with tumor lysate, the researchers aimed to teach each patient’s immune system to recognize the specific antigens carried by their own cancer. The preceding checkpoint blockade was intended to create a permissive environment, preventing newly activated tumor-specific T cells from being shut down by PD-1 signaling within the tumor microenvironment.

The results, published open access in the journal Cancer Immunology, Immunotherapy, are striking for a preliminary single-arm study. After the short course of checkpoint inhibitor therapy alone, the objective response rate, meaning the proportion of patients whose tumors shrank measurably by the modified Response Evaluation Criteria in Solid Tumors, was 16.7 percent, with a disease control rate of 70.8 percent. After the subsequent dendritic cell therapy, the objective response rate more than doubled to 41.7 percent, while disease control, which combines partial responses with stable disease, reached 75.0 percent. In other words, the addition of the dendritic cell phase converted a substantial fraction of patients whose disease had merely stabilized, or had not responded at all, into genuine responders.

Survival figures reinforced the impression of clinical benefit. Across all twenty-four patients, the median overall survival reached 36 months, an exceptional figure for advanced hepatocellular carcinoma, a disease in which historical median survival without effective systemic therapy was measured in months. The one-year overall survival rate was 87.5 percent, the two-year rate 75.0 percent, and the three-year rate 45.8 percent. Nearly half of the patients in this advanced-stage cohort were still alive three years after starting the sequential regimen, a result that, if confirmed in larger controlled trials, would represent a meaningful shift in the natural history of the disease.

One of the most intriguing findings of the study concerns a simple blood test that is routinely available in any hospital. The researchers found that patients who achieved an objective response after dendritic cell therapy had a significantly lower neutrophil-to-lymphocyte ratio before starting the dendritic cell phase than patients who did not respond, a difference that reached statistical significance with a p value of 0.028. The neutrophil-to-lymphocyte ratio, or NLR, is a widely used surrogate marker of systemic inflammation. A high ratio is thought to reflect an immune milieu dominated by neutrophils and inflammatory signaling, conditions that are associated with immunosuppression, tumor progression, and resistance to immunotherapy. A low ratio, by contrast, suggests a lymphocyte-rich environment in which adoptively transferred, tumor-primed dendritic cells are more likely to find receptive T cells to activate. If validated, this finding could give clinicians a cheap and rapid biomarker for selecting patients most likely to benefit from the dendritic cell phase of treatment.

Seven of the patients received nivolumab and seventeen received pembrolizumab before the dendritic cell infusions, and the study did not report a meaningful divergence between the two checkpoint inhibitors in this setting, which is consistent with their similar mechanisms of action. The dendritic cell product itself was manufactured in the Cell Processing Unit of the Chang-Gung Memorial Hospital Cell Therapy Center, a facility dedicated to producing patient-specific cellular therapeutics under controlled conditions. The use of tumor lysate rather than a single defined antigen is a deliberate design choice. Hepatocellular carcinoma is a heterogeneous disease, particularly in the context of chronic hepatitis B or C infection and cirrhosis, and no universal tumor antigen has been identified. Loading dendritic cells with the full repertoire of antigens from an individual patient’s tumor allows the immune system to mount a polyclonal response against multiple targets simultaneously, potentially reducing the chance that the cancer can escape by downregulating a single antigen.

Safety considerations are central to any cellular therapy, and the sequential approach appears to have been tolerated acceptably in this cohort. Immune-related adverse effects, the characteristic toxicities of checkpoint blockade arising from immune attack on healthy tissues, are a known risk of the priming phase, and limiting the checkpoint inhibitor course to three to six cycles may help contain cumulative toxicity. The dendritic cell phase, which uses the patient’s own cells pulsed with autologous tumor material, is generally regarded as having a favorable safety profile compared with more aggressive cellular therapies such as engineered T cell products, because the cells are not directly cytotoxic and carry no risk of uncontrolled expansion. The study was supported by a Chang-Gung Research Grant, and the authors declared no conflicts of interest.

The authors themselves are careful to frame the work as preliminary. The study involved a small number of patients, lacked a randomized control arm, and was conducted at a single institution with an integrated cell therapy infrastructure that is not widely available. Selection factors, including the patients’ underlying liver function, prior treatments such as transcatheter arterial embolization or tyrosine kinase inhibitors, and tumor biology, may all have influenced the outcomes. The researchers explicitly state that a large-scale study is needed in the future before the sequential regimen can be considered a standard of care. Nonetheless, the magnitude of the improvement in response rate, from under 17 percent after checkpoint blockade alone to nearly 42 percent after the dendritic cell phase, provides a compelling signal that the two modalities act synergistically rather than redundantly.

The broader significance of the study lies in its timing. Immunotherapy for hepatocellular carcinoma has advanced rapidly, with checkpoint inhibitor combinations now front-line options, yet the majority of patients still fail to respond, and the liver’s uniquely immunosuppressive environment remains a stubborn barrier. Therapeutic dendritic cell vaccines have been pursued for decades with mixed results, and this trial suggests that the order of operations matters: priming the system with checkpoint release before delivering antigen-loaded dendritic cells may succeed where either approach alone falls short. As personalized cell manufacturing becomes more routine and biomarkers such as the neutrophil-to-lymphocyte ratio help identify likely responders, the strategy of sequencing checkpoint blockade with dendritic cell therapy could move from a promising experiment to a practical weapon against one of the world’s deadliest cancers.

Subject of Research: Sequential immunotherapy combining immune checkpoint inhibitors and autologous dendritic cell therapy for advanced hepatocellular carcinoma

Article Title: Sequential therapy with immune checkpoint inhibitors and dendritic cells for advanced hepatocellular carcinoma

Article References: Lee, W.-C., Wang, Y.-C., Wu, T.-H., Lai, Y., Hung, H.-C., Lee, J.-C., Cheng, C.-H., Lee, C.-F., Wu, T.-J., Chou, H.-S., & Chan, K.-M. (2026). Sequential therapy with immune checkpoint inhibitors and dendritic cells for advanced hepatocellular carcinoma. Cancer Immunology, Immunotherapy. https://doi.org/10.1007/s00262-026-04583-w

Image Credits: AI Generated

DOI: 10.1007/s00262-026-04583-w

Keywords: hepatocellular carcinoma, immune checkpoint inhibitors, dendritic cells, immunotherapy, nivolumab, pembrolizumab, cell therapy, neutrophil-to-lymphocyte ratio, BCLC staging, liver cancer, clinical trial, cancer immunology

The study, published in the open-access journal Cancer Reports, rests on two variables that are already measured in every hepatocellular carcinoma workup: the number of tumors visible on imaging and the Model for End-Stage Liver Disease score, a laboratory index calculated from bilirubin, creatinine, and the international normalized ratio of prothrombin time that quantifies the functional reserve of a cirrhotic or damaged liver. The researchers drew on the rationale of the Japanese Society of Hepatology guidelines, which recommend considering liver resection for patients with multiple tumors when the tumor number is three or fewer, and reserving non-surgical treatments for patients with more than three tumors. They also incorporated the findings of a previous Italian study showing that a MELD score above nine signals an inadequate liver function reserve in patients undergoing liver resection, and that such a score was independently associated with postoperative liver decompensation, with an odds ratio of 2.26, a 95 percent confidence interval of 1.10 to 4.58, and a p value of 0.02.

From these two parameters the investigators constructed a two-tier classification of intermediate-stage disease that they labeled BCLC B1 and BCLC B2. Patients in the B1 category had to meet both criteria simultaneously, meaning a tumor number of three or fewer and a MELD score of nine or below, and these were defined as the ideal candidates for liver resection. Patients in the B2 category needed to satisfy only one of the two criteria, and these were defined as nonideal candidates for surgery. The cohort consisted of patients diagnosed with hepatocellular carcinoma between 2011 and 2021 who were managed at the study institution, all of whom had Child-Pugh class A liver function, the mildest functional class, and who had undergone either liver resection or transarterial chemoembolization. The protocol was approved by the Institutional Review Board of Kaohsiung Chang Gung Memorial Hospital under reference number 202201189B0, conformed to the Declaration of Helsinki, and the requirement for individual informed consent was waived.

The enrollment process captured a rich set of baseline variables for each patient, including age, sex, alpha-fetoprotein level, hepatitis B surface antigen status, antibody test results for hepatitis C virus, total bilirubin, creatinine, international normalized ratio, Child-Pugh classification, and MELD score. Under the staging definitions used in the study, BCLC stage B patients were those with Child-Pugh A or B liver disease and multifocal hepatocellular carcinoma beyond the Milan criteria, who did not have cancer-related symptoms and did not show macrovascular invasion or extrahepatic spread. The treatment algorithm at the institution reflected the realities of clinical practice in Taiwan: liver transplantation was an option for patients whose tumor burden fell within the University of California San Francisco criteria with a low alpha-fetoprotein level and clinically significant portal hypertension, the latter defined by a platelet count below ten to the ninth per liter with splenomegaly, the presence of varices, or portosystemic collateral vessels. Lobectomy was considered for tumors confined to one lobe, single-needle radiofrequency ablation was applied to multiple tumors smaller than three centimeters, and multipolar radiofrequency ablation was used for multiple tumors in which at least one measured between three and five centimeters. External beam radiation, systemic chemotherapy, or targeted therapy with sorafenib or lenvatinib was reserved for infiltrative hypovascular tumors or a tumor burden too high for embolization. Notably, targeted therapies and selective internal radiation therapy were not reimbursed by Taiwan’s National Health Insurance for patients with BCLC stage B disease, which constrained how often these modalities could be deployed.

When the researchers compared the baseline characteristics of the liver resection and chemoembolization subgroups, most variables were well balanced. Age, sex, the diameter of the largest nodule, alpha-fetoprotein level, and hepatitis B surface antigen positivity did not differ significantly between the groups. One exception stood out: the proportion of patients testing positive for hepatitis C virus antibodies was lower in the resection subgroup than in the embolization subgroup, a difference that reached statistical significance with a p value of 0.046. In one comparison of patient subgroups, the two arms also did not differ in alpha-fetoprotein levels of at least 400 nanograms per milliliter, which were recorded in 25 of 84 patients, or 29.8 percent, in one arm and 12 of 64 patients, or 18.8 percent, in the other, with a p value of 0.125.

The survival analysis produced the study’s central finding. Among patients classified as BCLC B1, the ideal surgical candidates, the median overall survival of patients undergoing liver resection was not reached during the observation period, while patients receiving transarterial chemoembolization survived a median of 3.9 years, with an interquartile range of 2.8 to 4.9 years. The difference was statistically significant, with a p value of 0.007. Strikingly, no patients in this group died within 90 days of treatment, underscoring that carefully selected surgical candidates not only lived longer but did so without an early postoperative mortality penalty. The Kaplan-Meier estimator and the log-rank test were used for these comparisons, all p values were two-tailed, and a p value below 0.05 was considered statistically significant.

The picture was different in the BCLC B2 group, the nonideal surgical candidates. There, two patients, or 2.5 percent, of those undergoing liver resection and four patients, or 1.9 percent, of those receiving chemoembolization died within 90 days of treatment, a difference that was not statistically significant with a p value of 0.671. The authors concluded from the combined evidence that patients with BCLC B1 disease are ideal candidates for liver resection, whereas those with BCLC B2 disease are not. This conclusion challenges the reflexive exclusion of surgery from intermediate-stage disease and, equally, challenges the assumption that all intermediate-stage patients are alike.

The study also engaged directly with competing frameworks for subdividing stage B disease. Spanish and Japanese investigations have used the up-to-seven criteria, in which the sum of tumor number and the size of the largest tumor in centimeters must not exceed seven, combined with Child-Pugh scores of 5 to 9, to stratify candidates. One cited study found a five-year recurrence rate of 25 percent for liver resection versus 60 percent for chemoembolization in a subgroup defined as Child-Pugh score 5 to 7 and within up-to-seven, with survival outcomes not differing between other subgroups, and concluded that resection provided better overall survival for that subgroup while survival was comparable for the others. A Korean study took a different tack, defining B2 disease as two to four oligonodular tumors with a largest tumor smaller than 10 centimeters, and found that resection produced better overall survival than embolization in that group, with a p value of 0.014, while survival did not differ significantly in the other groups. The Taiwanese authors argued that the up-to-seven criteria are not suitable for selecting patients for resection, because the very advantage of surgery is the removal of large tumors; in other words, tumor size should not disqualify a patient from resection. They illustrated the point with a concrete contrast: a patient with multiple small tumors, up to four in number with the largest measuring three centimeters and confined to one lobe, would fall within up-to-seven and be deemed an ideal resection candidate by the Spanish and Japanese criteria, yet in the authors’ experience such a patient would not be recommended for surgery. Conversely, their own tumor-number-based approach aligns with the Korean finding that resection suits patients with two to four tumors but not those with five or more.

The authors acknowledged that a proportion of BCLC B2 patients in their cohort may have had diffuse infiltrative hepatocellular carcinoma or a tumor burden too high for embolization, meaning that neither resection nor chemoembolization was truly suitable for them. Because systemic therapies for BCLC stage B disease were not reimbursed by Taiwan’s National Health Insurance, very few patients at this stage received such treatments, leaving a therapeutic gap that the study could not fill. The limitations of the work were stated plainly: it was retrospective and monocentric; it lacked data on severe comorbidities; cirrhosis data were not presented because the registry defined cirrhosis histologically for surgical patients and by imaging for non-surgical patients, and image-based definitions can be vague and subjective; and postoperative complication data, which are important for treatment decisions, were not available. The investigators also did not perform propensity score matching between the two treatment arms, because the groups had similar characteristics except for the lower prevalence of hepatitis C positivity among resection patients.

Even with those caveats, the study’s contribution is a practical one. By using just two variables that are available before any treatment begins, the image-defined tumor number and the MELD score, clinicians can preoperatively determine which subgroup of BCLC stage B patients should undergo liver resection and which should be directed toward transarterial chemoembolization. In an era when treatment options for intermediate-stage liver cancer are multiplying, from radioembolization to multipolar ablation to systemic therapy, a simple, reproducible two-parameter rule offers a way to bring order to a contested clinical space, and it gives surgeons and interventional radiologists a shared, evidence-based language for one of hepatocellular carcinoma’s most persistent decisions.

Subject of Research: Selection criteria using tumor number and MELD score for liver resection versus transarterial chemoembolization in intermediate-stage hepatocellular carcinoma

Article Title: Tumor Number and Model for End‐Stage Liver Disease Score as Selection Criteria for Resection or Embolization of Intermediate‐Stage Hepatocellular Carcinoma

Article References: Yen, Y.-H., Yong, C.-C., Liu, Y.-W., Li, W.-F., Wang, C.-C., & Lin, C.-Y. (2026). Tumor Number and Model for End‐Stage Liver Disease Score as Selection Criteria for Resection or Embolization of Intermediate‐Stage Hepatocellular Carcinoma. Cancer Reports, 9(9), Article e70697. https://doi.org/10.1002/cnr2.70697

Image Credits: AI Generated

DOI: 10.1002/cnr2.70697

Keywords: hepatocellular carcinoma, liver resection, transarterial chemoembolization, MELD score, BCLC staging, tumor number, overall survival, Child-Pugh class A, Taiwan, liver function reserve, up-to-seven criteria, retrospective study