Proton Therapy Enters a New Evidence Phase: Meningioma, HCC, and the Challenge of Randomised Data

The particle therapy literature this month delivers a mixed but instructive signal. On one hand, disease-specific evidence is accumulating across meningioma, hepatocellular carcinoma, lung, and breast — with several studies reaching randomised or prospective milestone status. On the other, a comprehensive analysis of 15 years of clinical trial registrations exposes the stubborn gap between dosimetric promise and definitive evidence. The field is generating more data, but not always the right kind.

Meningioma Evidence Matures

Two complementary studies advance the case for proton therapy in meningiomas. A systematic review and meta-analysis of 19 studies involving 1,431 patients provides the most comprehensive pooled estimate to date: a 5-year overall survival proportion of 91% and a pooled complication rate of 16%, with acceptable toxicity across WHO grades. The majority of patients had grade I tumours (70.6%), but the inclusion of grade II and III disease adds clinical breadth. Heterogeneity was high, reflecting the variability in dosing and institutional practice, but the meta-analysis offers a benchmark for future comparisons.

The more interesting signal comes from a prospective single-arm phase 1 trial of dose-escalated intensity-modulated proton therapy for high-risk meningiomas. Twenty-one patients with grade 2 (residual or recurrent) or grade 3 (after gross total resection) disease received 66 Gy(RBE) for gross disease and 63 Gy(RBE) for tumour beds — doses above conventional standards. No acute dose-limiting toxicity was observed. At a median follow-up of 3.4 years, 3-year progression-free survival was 88.2% and overall survival 95.0%. Five late grade 3 toxicities occurred in four patients, but no treatment-related deaths. These results are particularly encouraging given that high-grade meningiomas have poor outcomes with conventional radiotherapy. The phase 1 data justify progression to phase 2 and 3 studies to determine whether dose escalation using protons confers a meaningful advantage over standard-dose photon or proton therapy, and whether the toxicity profile holds in larger, multicentre cohorts.

Hepatocellular Carcinoma: Large Tumours, Promising Signals

Large, unresectable hepatocellular carcinoma remains one of the most challenging indications for radiotherapy, with the risk of radiation-induced liver disease constraining dose. Two retrospective studies — one for protons, one for carbon ions — report encouraging outcomes.

In the proton study, 51 Western patients with HCC ≥5 cm (median tumour size 9.2 cm, 39% with vascular invasion) received 45.0–67.5 Gy(RBE) in 15 fractions. One-year local control was 90%, with a 4% cumulative incidence of isolated local failure. Overall survival at 1, 2, and 3 years was 65%, 46%, and 30%, respectively — reasonable given that 45% were BCLC stage C and 26% had Child-Pugh B/C cirrhosis. The dominant failure pattern was out-of-field intrahepatic progression (50%), suggesting that beyond local control, the field needs better integration of systemic therapy. The authors identified tumour size >10 cm and Child-Pugh B cirrhosis as independent predictors of worse survival and higher hepatotoxicity risk.

The carbon ion study from Japan reports on 33 patients with HCC ≥10 cm (median 11 cm), treated predominantly with 60 Gy(RBE) in 4 fractions. One-year local control was 85.9%, progression-free survival 30.0%, and overall survival 74.8%. Grade ≥3 toxicity occurred in 15% of patients, including pleural effusion and hepatic decompensation, but no grade 4 or 5 events. The short median follow-up of 15 months limits assessment of late toxicity, and the 15% rate of radiation-induced liver disease (Child-Pugh score increase ≥2) warrants attention. Nonetheless, for tumours of this size — where surgery and transarterial approaches are often inadequate — both proton and carbon ion therapy appear to offer a meaningful option.

Randomised Data: Lung and Breast

Two randomised comparisons published this month highlight the nuanced trade-offs of proton therapy.

The post-hoc analysis of a prospective randomised trial comparing passive-scattering proton therapy (PSPT) with IMRT in 147 patients with locally advanced NSCLC confirms and extends earlier findings. With median follow-up of 28.5 months, the key new insight concerns cardiac toxicity: 9 of 10 radiation-related major adverse cardiac events occurred in the IMRT group, with doses to the coronary arteries implicated. Grade ≥3 radiation pneumonitis was similar between arms (10.9% vs 10.5%), but two IMRT patients had grade 5 pneumonitis versus none in the PSPT arm. No differences emerged in local failure or progression-free survival. The survival outcomes were comparable to RTOG 0617, but with a cardiac safety advantage that may become more important as NSCLC patients live longer on immuno-oncology combinations.

The breast cancer trial, by contrast, reveals a less straightforward picture. This randomised phase II study compared ultrahypofractionated proton versus photon radiotherapy (26 Gy in 5 fractions) in 72 non-metastatic breast cancer patients (interim analysis). The incidence of acute radiation dermatitis was significantly higher in the proton arm (97.2% vs 75%), with grade 2 dermatitis occurring in one patient. Photon-treated patients had better cosmetic outcomes (41.7% vs 13.9% excellent/good) and higher satisfaction scores. No locoregional recurrences were observed at a median of 9 months. The finding that protons — with their superior skin-sparing in standard fractionation — may increase acute skin toxicity in ultrahypofractionated schedules is a reminder that the Bragg peak's advantages are dose-fractionation dependent. Longer follow-up and the full cohort will determine whether the acute differences translate into cosmetic or functional consequences.

Technology Enablers

Several technical contributions this month address the practical challenges of proton delivery. A study on brass apertures for single-energy Bragg peak FLASH radiotherapy demonstrates that apertures can reduce lateral penumbra by up to 9.4 mm at large pullback distances, improving target conformity and OAR sparing in a re-irradiation glioblastoma plan, while maintaining FLASH dose rate in the high-dose region. The trade-off is a modest reduction in ultra-high dose rate coverage in low-dose regions — clinically acceptable given the dosimetric gains.

On the imaging side, a dual-energy CT study shows that advanced slice-wise patient-thickness estimation methods improve stopping-power ratio prediction accuracy, reducing proton range uncertainties. For obese patients, deviations of up to 0.3% between new and old methods were observed — small but relevant when range margins are already tight.

Proton lattice radiotherapy — a spatially fractionated technique delivering high-dose vertices within large tumours — was shown to be technically feasible using pencil-beam scanning, with plan-specific quality assurance meeting clinical tolerances. For targets >100 cm³, proton LRT produced more lattice vertices than photon VMAT, offering potential for immunomodulatory effects.

A dosimetric comparison of IMPT versus photon therapy for pituitary adenomas found that protons significantly reduced dose to the hippocampi (median D40% of 0.7 Gy vs 5.0 Gy for coplanar photons) at the cost of slightly higher doses to brainstem and optic structures. The overall adverse event burden, estimated using the ROCOCO performance scoring system, favoured protons.

Finally, NRG Oncology published comprehensive technical guidance for implementing proton therapy in multi-institutional trials, addressing setup uncertainties, motion management, RBE uncertainties, and robustness evaluation. The document is a pragmatic response to the inconsistency that has historically plagued multi-centre proton trials — a necessary step if level 1 evidence is to be generated at scale.

The Trial Landscape Reality Check

A cross-sectional analysis of 416 particle beam therapy trials registered over 15 years provides the sobering context for these advances. Despite steady growth, more than 70% of trials remain phase I/II, and over half enroll fewer than 50 participants — 67.9% for carbon ion trials versus 49.5% for protons. Only 5.3% of proton and 3.8% of carbon ion trials reached phase III, compared with 9.5% for photon trials. Early termination occurred in 26% of completed or terminated trials, predominantly due to slow accrual. Carbon ion trials remain geographically concentrated in Asia, while proton trials show wider but still limited dispersion.

These structural barriers are not new, but the data quantify them with authority. The Norwegian neuro-oncology consensus guidelines published this month offer one model for bridging the gap: a three-tiered system that distinguishes established indications, dose comparison candidates requiring individualised plan evaluation, and patients eligible for study inclusion. In the first six months of operation, 28 adult neuro-oncology patients received proton therapy under these guidelines, with 71% of dose comparison candidates selected based on quantitative OAR sparing criteria. The approach is pragmatic, transparent, and designed for iterative revision as trial data emerge.

The Collective Signal

This month's literature suggests that particle therapy is emerging from its exploratory phase — but unevenly. Meningioma and HCC evidence is strengthening, with proton and carbon ion data converging on consistent local control rates. Randomised lung data confirm a cardiac advantage that may shape guideline recommendations. The breast data inject a necessary note of caution: dosimetric superiority does not guarantee clinical superiority for all endpoints, and fractionation-specific effects matter.

The technology pipeline — FLASH delivery, improved range prediction, lattice techniques — promises to expand the therapeutic window further. But the trial landscape analysis is a reminder that better technology alone will not generate the evidence that payers, guideline committees, and patients require. The field needs more pragmatic randomised trials, standardised robustness reporting, international collaboration on RBE modelling, and funding mechanisms that support phase III studies in common indications. The Norwegian consensus model offers a template for how clinical practice and evidence generation can proceed in parallel — treating patients today while building the infrastructure for tomorrow's trials.