Proton Therapy Does Not Improve Survival in Hepatocellular Carcinoma

Proton therapy did not improve overall survival (OS) compared with photon therapy in patients with advanced hepatocellular carcinoma (HCC), according to initial results from the phase 3 NRG-GI003 trial (NCT03186898) presented during the plenary session of the 2026 American Society for Radiation Oncology (ASTRO) Annual Meeting.^1,2 The trial was released for reporting after crossing a prespecified futility boundary at a planned interim analysis. Proton therapy also failed to improve the secondary end points of progression-free survival (PFS) and local progression.

The presenting study author was Theodore S. Hong, MD, the vice chair of Integration and Collaboration in Radiation Oncology at Beth Israel Deaconess Medical Center and Dana-Farber Cancer Institute (DFCI), associate chief of Procedure-based Clinical Research at DFCI, a professor of Radiation Oncology at Harvard Medical School, and lead and presenting author of the NRG-GI003 study.

“The rationale for the study was straightforward: proton therapy can reduce radiation dose to the uninvolved liver compared with photons,” stated Hong during his presentation.¹ “Because these patients frequently have compromised underlying liver function, we hypothesized that reducing dose to the normal liver could reduce hepatic decompensation and ultimately improve survival.”

What were the key survival findings from NRG-GI003?

At the planned interim analysis, which occurred after 53 OS events, the observed HR crossed the prespecified futility boundary for protons vs photons of an HR greater than 1, with a value of 1.3 (90% CI, 0.83-2.04; P = .83). The data monitoring committee recommended the trial be released for reporting.¹ The median OS was 26.6 months (90% CI, 21.0-39.8) with proton therapy vs 54.9 months (90% CI, 27.0-63.0) with photon therapy.

“There was no survival benefit with proton therapy,” Hong stated. “Note that the proton arm actually performed similarly to the 24-month median survival projected in the study. However, the photon arm, where the median survival was nearly 55 months, dramatically exceeded the 14 months assumed in the original statistical design.”

Proton therapy also did not improve PFS, with a median of 16.9 months (95% CI, 8.0-19.9) with proton therapy vs 13.7 months (95% CI, 9.7-21.7) with photon therapy (HR, 1.02; 95% CI, 0.65-1.62; 2-sided log-rank P = .92). Local progression was not improved either (HR, 0.80; 95% CI, 0.36-1.77; 2-sided P = .54); 24-month LP rates were 19.8% with protons and 22.3% with photons.²

Did any subgroups show a differential treatment effect?

In a post hoc, univariate forest plot analysis across prespecified subgroups, most did not show a convincing treatment effect, but tumor size showed a significant interaction (P = .019).¹ Among patients with tumors smaller than 5 cm, the 24-month OS rate was 66.9% with photons vs 66.0% with protons, with an HR of 0.63 (95% CI, 0.27-1.45) favoring protons. Among patients with tumors 5 cm or larger, the effect reversed to favor photons, with respective values of 73.7% vs 47.5% (HR, 2.51; 95% CI, 1.17-5.37). In a multivariable model adjusting for Child-Pugh score and tumor vascular thrombosis, the treatment-by-tumor-size interaction remained significant (P = .046).

“For tumors less than 5 cm, outcomes numerically favored protons,” Hong stated.¹ “Surprisingly, for tumors 5 cm or larger, the direction of effect reversed and numerically favored photons.” He cautioned that the finding should be interpreted carefully. “Please remember this is a post hoc analysis, and this should be considered hypothesis generating rather than definitive.”

What safety differences were observed between proton and photon therapy?

Grade 3 or higher treatment-related adverse events (AEs) occurred in 11% of patients receiving proton therapy vs 24% receiving photon therapy, a difference that did not reach statistical significance (P = .093). Grade 3 or higher investigations-related AEs occurred in 9% of the proton group vs 22% of the photon group, driven primarily by lymphocyte decrease. There were no grade 5 treatment-related AEs in either arm.

A separate dosimetric analysis confirmed the trial's underlying premise. For patients treated with 5 fractions, the median dose to liver minus gross tumor volume was 8.5 Gy with protons vs 12.8 Gy with photons; for patients treated with 15 fractions, it was 12.7 Gy vs 20.3 Gy, respectively.

“This slide demonstrates that the fundamental dosimetric premise of the trial was correct: protons substantially reduced the mean dose to the uninvolved liver,” Hong stated.¹ “There was also a numerical reduction in grade 3 or higher treatment-related [AEs]...although this did not reach conventional statistical significance. Overall, severe toxicity was uncommon in both groups, and there were no grade 5 treatment-related events.”

How was the NRG-GI003 trial designed?

NRG-GI003 randomly assigned patients with unresectable or locally recurrent HCC 1:1 to proton or photon radiation therapy, delivered in 5 fractions (30-50 Gy) or 15 fractions (37.5-67.5 Gy) with individualized dosing, and stratified by planned fractionation and the presence of tumor vascular thrombus. Eligible patients had up to 3 tumors, Child-Pugh A or B7 liver function, and an ECOG performance status of 0 or 1; prior systemic or liver-directed therapy was permitted.¹ The trial originally planned to enroll 167 patients.

“Because of slower than expected accrual, the study was amended and closed to accrual in January 2026, with a revised target of 118 patients and 80% power using a 1-sided alpha of 0.05,” Hong stated.¹ The trial was powered to detect an OS HR of 0.58, corresponding to an improvement in median survival from 14 to 24 months.¹

Between June 2017 and January 2026, 117 patients were registered, 116 were randomly assigned, and 115 were eligible for analysis (n = 57 proton, n = 58 photon). The median patient age was 72 years, 93% had Child-Pugh A liver function, 25% had tumor vascular thrombus, and 79% had a single tumor. The median total tumor dimension was 5.1 cm and was balanced between arms. The PTV radiation dose received was per protocol or had an acceptable variation per central review in 90% of patients. “Thus, there was no major imbalance in protocol compliance,” Hong said.

What do these findings mean for future HCC radiotherapy research?

“Although these findings were not positive, NRG-GI003 will help us shape the future of research surrounding this disease,” Hong said in a press release on the findings.² “The data from this trial shifts our focus to identifying potential subgroups of patients with HCC who may be more likely to derive benefit from proton therapy. Additionally, outside of survival benefit, future studies should aim to determine whether protons improve quality of life or reduce [AEs] when compared to traditional photon therapy.”

Closing his presentation, Hong summarized 4 takeaways: “In this randomized phase 3 trial, proton therapy did not improve OS compared with photon therapy in patients with advanced HCC. Second, the dosimetric advantage of protons was real and substantial, with significantly lower exposure of uninvolved liver, and there is a numerical reduction in severe toxicity. Third, both arms performed favorably compared [with] hypothesized OS, with a photon arm performing noticeably better than expected. And finally, the post hoc tumor size interaction suggests that the value of protons may not be uniform across all patients.”¹ He added that further work is needed to determine whether tumor burden, liver reserve, or other factors can identify patients most likely to benefit from proton therapy, and that investigation of patient selection for protons or photons remains ongoing.

References

1. Hong TS, Winter K, Koay EJ, et al. Initial results of NRG Oncology NRG-GI003: a phase III randomized trial of protons versus photons for hepatocellular carcinoma (HCC) (NCT03186898). Presented at: 2026 American Society for Radiation Oncology Annual Meeting; September 28, 2026; Boston, MA.
2. NRG Oncology trial results show no survival benefit with the use of proton therapy for patients with advanced hepatocellular carcinoma. News release. NRG Oncology. September 28, 2026. Accessed September 29, 2026. https://tinyurl.com/3uksfm6p