Radiation After Liver Cancer Surgery Tops Rankings in Global Therapy Analysis

For patients who undergo curative-intent surgery for hepatocellular carcinoma, the operation itself is only half the battle. Even when surgeons remove all visible tumor tissue, the cancer returns in a substantial proportion of patients, making postoperative recurrence the single most important barrier to long-term cure. A new network meta-analysis published in Clinical Cancer Bulletin has now systematically compared 28 randomized controlled trials involving 4,830 patients to answer a question that has frustrated hepatobiliary oncologists for decades: which postoperative adjuvant therapy actually works best, and how do the competing strategies stack up against one another when they have never been tested head-to-head?

The study, led by Ashraf Nadeem, Yunfan Yang, Xinyan Li, and Kun Li of the Department of Hepatobiliary and Pancreatic Surgery at Zhongnan Hospital of Wuhan University, employed a frequentist network meta-analysis, a statistical framework that pools both direct comparisons and indirect evidence across a connected web of randomized trials. Because most adjuvant therapies for liver cancer have been tested only against observation or placebo rather than against each other, conventional pairwise meta-analysis cannot rank them. The network approach uses a shared common comparator—in this case, observation or placebo—as a statistical anchor, allowing the relative performance of treatments that have never faced each other in a trial to be estimated within a single model. Treatments were ranked using P-scores, which express the probability that one intervention outperforms another under the network model.

The evidence base was assembled through an exhaustive search of MEDLINE, Embase, CENTRAL, Web of Science, Scopus, regional databases, clinical trial registries, and conference proceedings covering trials published between January 2000 and December 2025. The investigators deliberately restricted their network to trials conducted exclusively in patients undergoing hepatic resection, excluding mixed populations that also included ablation or transplantation, in order to preserve clinical coherence and strengthen the transitivity assumption on which indirect comparisons depend. The review was prospectively registered in PROSPERO and reported according to the PRISMA 2020 statement and its network meta-analysis extension, with risk of bias assessed using the Cochrane Risk of Bias 2 tool.

The headline finding concerns overall survival. Three internal radiation and brachytherapy approaches dominated the survival hierarchy. Adjuvant iodine-125 brachytherapy showed the largest reduction in the hazard of death compared with observation, with a hazard ratio of 0.36 (95 percent confidence interval 0.17 to 0.78), followed by intensity-modulated radiation therapy, or IMRT, at a hazard ratio of 0.44 (0.23 to 0.86), and iodine-131-labeled metuximab, a radiolabeled antibody targeting HAb18G/CD147 on tumor cells, at 0.46 (0.28 to 0.77). In probabilistic terms, iodine-125 brachytherapy carried the highest P-score for overall survival at 0.86, ahead of IMRT at 0.78 and iodine-131-metuximab at 0.77. Conventional transarterial chemoembolization, long a mainstay of postoperative consolidation, retained activity with a hazard ratio of 0.71 (0.55 to 0.92) but was generally outperformed by the precision radiotherapy approaches.

The picture shifted when the outcome was recurrence-free survival, the primary endpoint of most adjuvant trials and typically the earliest signal of treatment failure. Here, the autologous formalin-fixed tumor vaccine, or AFTV, an immunotherapy prepared from a patient’s own resected tumor tissue, ranked first with a P-score of 0.92, followed by a combination of the traditional medicine Cidan capsule with transarterial chemoembolization at 0.90, iodine-125 brachytherapy at 0.84, and IMRT at 0.80. The divergence between the two hierarchies is not contradictory, the authors argue, but reflects differences in endpoint maturity and mechanism: recurrence events accrue earlier than deaths, and some strategies may suppress relapse long before any difference in survival fully matures.

The mechanistic reasoning behind these rankings is instructive. Early postoperative recurrence in hepatocellular carcinoma is frequently driven by microscopic residual disease, narrow resection margins, microvascular invasion, and venous dissemination—processes that concentrate relapse risk around the surgical bed and regional vascular pathways. IMRT is conceptually well suited to sterilizing this presumed microscopic disease in the postoperative field, whereas transarterial chemoembolization targets hypervascular visible tumor tissue through the arterial supply and is a less precise fit for the minimal residual disease setting. Meanwhile, the immune-based strategies operate systemically. Hepatocellular carcinoma typically arises in a chronically inflamed, immunologically dysregulated liver characterized by T-cell exhaustion and immune escape, and checkpoint inhibition with agents such as the PD-1 inhibitor sintilimab—which demonstrated a favorable benefit-risk profile in the network—aims to restore cytotoxic antitumor immunity against occult malignant clones throughout the remnant liver.

AFTV represents a different immunological logic altogether. Rather than amplifying pre-existing antitumor immunity, the autologous vaccine is designed to generate antigen-specific immune priming de novo, using formalin-fixed tumor tissue as a source of the full repertoire of tumor antigens in each individual patient. The authors note that this strategy is conceptually well matched to the adjuvant setting, where disease burden is minimal and immune containment is more achievable than in advanced disease. They are careful, however, to frame the finding as hypothesis-generating rather than definitive: the AFTV ranking rests on a single small phase II trial, mature overall survival data are lacking, and personalized vaccine platforms face practical challenges around standardization, scalability, and reproducibility across centers.

Safety signals across the 28 trials were reassuring. Grade 3 or higher adverse events were reported infrequently. Immune-related hepatitis of grade 3 occurred in 2 to 5 percent of patients receiving sintilimab, with no treatment-related deaths. Radiation-induced liver disease of grade 3 or worse was rare, affecting fewer than 1 percent of Child-Pugh A patients treated with IMRT. Post-embolization syndrome affected 5 to 10 percent of patients undergoing transarterial chemoembolization, and no grade 3 or higher events were reported for AFTV. The analysis also revealed that the strongest efficacy signals frequently came from clinically enriched high-risk populations: the sintilimab trial enrolled a cohort with 100 percent pathological microvascular invasion, the IMRT trial enrolled patients with 100 percent portal vein tumor thrombus, and the AFTV trial arose from a population with very high hepatitis B prevalence—features that matter for generalizability, since most included patients had hepatitis B virus-related disease and preserved liver function.

The authors are candid about the limitations of their work. The evidence network was largely star-shaped, with observation as the dominant comparator, meaning most comparisons between active treatments were estimated indirectly. Because the network contained no closed loops, formal consistency testing was not possible, and coherence rests entirely on the transitivity assumption. Hepatitis B prevalence ranged from 36 to 100 percent and microvascular invasion rates from 0 to 100 percent across trials, and meta-regression was infeasible given the small number of trials per covariate category. Landmark mixed-modality trials such as IMbrave050, which showed that adjuvant atezolizumab plus bevacizumab can reduce recurrence risk in high-risk patients after resection or ablation, could not be included because resection-specific estimates were not extractable. Evolving salvage therapies over the 26-year study window may also have influenced survival endpoints in older trials.

Nevertheless, the analysis points toward a future in which postoperative management of liver cancer is risk-adapted and mechanism-based rather than empiric. Precision radiotherapy may be most rational when relapse risk is concentrated in the surgical bed or vascular territory, while immune-based therapy may better serve patients whose recurrence risk is diffuse or systemic. The authors emphasize that not every patient should automatically receive adjuvant treatment; the strongest case exists for those with clearly elevated risk, such as microvascular invasion or tumor multiplicity. Large active-comparator randomized trials, standardized treatment-initiation windows, and biomarker-informed selection will be needed before a definitive standard emerges. Until then, this synthesis provides the clearest comparative map yet of a therapeutic landscape long defined by uncertainty.

Subject of Research: Comparative effectiveness of postoperative adjuvant therapies for hepatocellular carcinoma after curative resection, evaluated by network meta-analysis of randomized controlled trials

Article Title: Comparative efficacy and safety of postoperative adjuvant therapies after curative resection for hepatocellular carcinoma: a network meta-analysis

Article References: Nadeem, A., Yang, Y., Li, X., & Li, K. (2026). Comparative efficacy and safety of postoperative adjuvant therapies after curative resection for hepatocellular carcinoma: a network meta-analysis. Clinical Cancer Bulletin, 5(1), Article 14. https://doi.org/10.1007/s44272-026-00066-2

Image Credits: AI Generated

DOI: 10.1007/s44272-026-00066-2

Keywords: hepatocellular carcinoma, network meta-analysis, adjuvant therapy, liver resection, iodine-125 brachytherapy, IMRT, iodine-131 metuximab, recurrence-free survival, overall survival, immunotherapy, tumor vaccine, TACE