Large Randomized Trial Finds Proton Therapy Offers No Survival Advantage in Advanced Liver Cancer

A closely watched randomized clinical trial designed to determine whether proton beam radiation can outperform conventional photon radiation in patients with advanced hepatocellular carcinoma has come back with a sobering answer: it does not. The Phase III NRG-GI003 study, led by the National Cancer Institute-funded clinical trials network NRG Oncology, was stopped at its planned interim analysis after crossing a predefined futility boundary, meaning that the accumulating data made it statistically implausible that proton therapy would ever demonstrate the overall survival advantage the trial was designed to detect. The findings were presented in the Plenary Session of the 2026 American Society for Radiation Oncology Annual Meeting in Boston, one of the most prominent stages in radiation oncology, underscoring how much weight the field had placed on this question.

Hepatocellular carcinoma is the most common form of primary liver cancer and a leading cause of cancer death worldwide, particularly in regions where viral hepatitis and other chronic liver diseases are prevalent. For patients whose tumors are caught early, surgery, transplantation, or ablation can be curative, but advanced disease presents a far more difficult problem. Many of these patients are treated with radiation, either to control the primary tumor or to manage vascular invasion, yet the surrounding liver tissue is often fragile, compromised by cirrhosis, and intolerant of high radiation doses. This clinical tension is precisely what made proton therapy so appealing in theory.

Proton therapy differs from conventional photon radiotherapy in a fundamental physical way. Photons, the high-energy X-rays used in standard linear accelerators, deposit radiation dose along their entire path through the body, entering the tissue, delivering a dose peak, and then exiting through healthy tissue on the far side. Protons, by contrast, carry a characteristic depth-dose profile known as the Bragg peak: they release most of their energy at a specific depth determined by their initial energy, and essentially stop there. In principle, this allows clinicians to concentrate a destructive dose inside a tumor while sparing the healthy liver parenchyma, bowel, and other adjacent organs that photons would inevitably pass through. For an organ as radiosensitive and functionally precious as a cirrhotic liver, the dosimetric argument for protons has long been compelling, and several single-arm and observational studies suggested meaningful reductions in liver toxicity.

NRG-GI003 was designed to convert that theoretical and dosimetric promise into a hard survival endpoint. The trial enrolled 115 eligible patients with advanced hepatocellular carcinoma and stratified them by two clinically important factors: the planned number of treatment fractions, either five or fifteen, and the presence of tumor vascular thrombosis, in which cancer cells invade the portal or hepatic veins and markedly worsen prognosis. Patients were then randomly assigned to receive either proton or photon therapy, delivered as five fractions of 30 to 50 Gy or fifteen fractions of 37.5 to 67.5 Gy. The trial was powered to detect a substantial survival benefit, with a target hazard ratio of 0.58 favoring protons, a bold assumption that reflected the optimism surrounding the technology.

That optimism did not survive contact with the data. At the planned interim analysis, conducted once 53 death events had been recorded, the trial crossed its futility boundary, defined as a hazard ratio greater than 1, and was released for reporting. Median follow-up across all patients was 20 months, with a range extending from just over two weeks to nearly 82 months. Far from showing the hypothesized benefit, proton therapy failed to improve overall survival at all: the one-sided p-value was 0.83, with a hazard ratio of 1.30 and a 90 percent confidence interval of 0.83 to 2.04. At 24 months, overall survival was 56.2 percent in the proton arm and 69.3 percent in the photon arm, numerically favoring the conventional approach, although the confidence intervals overlapped considerably.

The secondary endpoints told the same story. Progression-free survival, which measures how long patients live without their disease worsening, was essentially indistinguishable between the arms, with a hazard ratio of 1.02 and a two-sided p-value of 0.92; 24-month progression-free survival was 31.6 percent for protons versus 30.4 percent for photons. Local progression, a measure of how well the radiation controlled the treated tumor itself, also showed no significant difference, with a hazard ratio of 0.80 and a p-value of 0.54. Notably, 24-month local progression rates were low in both arms, at 19.8 percent with protons and 22.3 percent with photons. The investigators highlighted this as an encouraging signal in its own right, adding to accumulating evidence that hepatocellular carcinoma should not be considered a radioresistant cancer, a label that historically limited the use of radiation in liver tumors.

One area where protons did show a numerical edge was toxicity. Grade 3 or higher treatment-related adverse events, the severe side effects that most concern clinicians, occurred in 11 percent of patients receiving proton therapy compared with 24 percent of those receiving photon therapy. However, with a p-value of 0.093, this difference did not reach the threshold for statistical significance at alpha equal to 0.05. The trend is consistent with the expected dosimetric advantage of protons, but in a trial of this size, halted early for futility, the study lacked the power to confirm it as a genuine difference rather than a chance finding.

Lead author Theodore S. Hong, MD, of Dana-Farber Cancer Center at Harvard, framed the negative result as informative rather than definitive. He noted that although the findings were not positive, the trial will help shape the future of research in this disease, shifting attention toward identifying subgroups of patients with hepatocellular carcinoma who may be more likely to benefit from proton therapy. He also pointed beyond survival, suggesting that future studies should determine whether protons improve quality of life or reduce side effects compared with traditional photon therapy, even in the absence of a survival advantage. That reframing reflects a growing recognition in radiation oncology that endpoints such as patient-reported outcomes, liver function preservation, and cost-effectiveness matter enormously when two treatments deliver comparable survival.

The result also arrives at a moment of intensifying scrutiny for proton therapy more broadly. Proton facilities are dramatically more expensive to build and operate than photon centers, and payers and health systems have increasingly demanded randomized evidence that the technology translates its physical dose advantages into clinical benefit. Past randomized comparisons in other tumor types have sometimes shown that what looks superior on a dose-distribution map does not produce better outcomes in patients, for reasons that include tumor biology, interpatient variability in proton range, setup uncertainty, and the genuine effectiveness of modern photon techniques such as intensity-modulated radiotherapy and stereotactic body radiotherapy. NRG-GI003 now stands as one of the clearest examples of that gap between physics and clinical endpoints in gastrointestinal oncology.

The trial was conducted through the NCI National Clinical Trials Network and supported by National Cancer Institute grants funding NRG Oncology operations, its statistics and data management center, the NCORP research network, and the irradiation quality assurance center IROC. For patients with advanced hepatocellular carcinoma, the practical message is that conventional photon radiation remains the standard against which any new technology must be judged, and that it performs respectably: tumor control rates and two-year survival in both arms of this trial were encouraging for a population with advanced disease. For the field, the challenge now is precision of a different kind, using molecular and clinical characteristics to determine which patients, if any, stand to gain from the steeper price and logistical burden of proton beams, and whether the technology’s true value lies not in extending life but in softening the burden of treatment itself.

Subject of Research: A randomized Phase III trial comparing proton versus photon radiation therapy for advanced hepatocellular carcinoma

Article Title: NRG Oncology trial results show no survival benefit with the use of proton therapy for patients with advanced hepatocellular carcinoma

Article References: NRG Oncology trial results show no survival benefit with the use of proton therapy for patients with advanced hepatocellular carcinoma. (n.d.). Original publication

Image Credits: AI Generated

DOI: Not provided

Keywords: hepatocellular carcinoma, proton therapy, photon therapy, radiation oncology, NRG-GI003, NRG Oncology, clinical trial, overall survival, ASTRO, liver cancer, futility analysis, adverse events