NRG-GI003 tested whether the liver-sparing dosimetric advantages of proton therapy translate into improved survival compared with photon radiotherapy in patients with advanced hepatocellular carcinoma (HCC). The phase III trial randomized 116 patients with unresectable or locally recurrent HCC (Child-Pugh A or B7, 1–3 tumors) to proton or photon radiotherapy in 5 or 15 fractions, with dose individualized according to normal-liver constraints. The study was powered to detect a substantial overall survival (OS) benefit (HR=0.58). At the planned interim analysis, the prespecified futility boundary was crossed (HR>1), and the trial was released for reporting.
At a median follow-up of 20 months, protons did not improve OS compared with photons (HR=1.30, 90% CI 0.83–2.04, one-sided p=0.83). At 24 months, OS was 56.2% with protons versus 69.3% with photons. This numerical difference should be interpreted cautiously: there were no significant differences in progression-free survival (HR=1.02, p=0.92) or local progression (HR=0.80, p=0.54). The broad confidence interval and interim nature of the analysis argue against interpreting the HR of 1.30 as evidence that proton therapy worsens survival.
Treatment-related grade 3 or higher adverse events were numerically lower with protons (11% versus 24%, p=0.09). Although not statistically significant, this 13-point absolute difference is notable and warrants further study. Sparing uninvolved liver is a principal rationale for proton therapy in HCC, and detailed reporting of hepatic toxicity and changes in liver function will help determine whether selected patients may derive a meaningful toxicity benefit despite the absence of an OS advantage.
Two contextual factors are important. Trial accrual spanned 2017–2026, a period during which systemic therapy for advanced HCC changed substantially. Randomization protects internal comparison, but information on systemic therapies received before and after radiotherapy will help determine how these findings apply to contemporary practice. In addition, 93% of patients had Child-Pugh A liver function and all had Zubrod performance status 0–1. Whether the findings extend to carefully selected patients with more limited hepatic reserve remains uncertain.
The NRG investigators should be congratulated for completing a challenging randomized phase III comparison of proton and photon therapy. NRG-GI003 provides high-level evidence that the dosimetric advantages of proton therapy do not automatically translate into improved survival. Routine selection of protons solely with the expectation of extending survival in patients resembling those enrolled is not supported by these results.
Importantly, as a superiority trial, NRG-GI003 should not be interpreted as establishing equivalence between protons and photons or demonstrating that proton therapy has no role in HCC. The numerical reduction in grade 3 or higher toxicity deserves further evaluation. Whether proton therapy provides clinically meaningful liver sparing for selected patients with limited hepatic reserve, larger or anatomically challenging tumors, or when adequate photon dose constraints cannot be achieved remains an important question.
Abstract 2, Initial Results of NRG Oncology NRG-GI003: A Phase III Randomized Trial of Protons vs. Photons for Hepatocellular Carcinoma (HCC), was presented during PL 01- Plenary 01, at ASTRO's 68th Annual Meeting.
Published on: September 29, 2026
