Skin, Survivorship and Where the Beams Are: Particle Therapy in August 2026

The strongest evidence in August comes from a randomised trial, and it does not say what proton advocates might have hoped. In early breast cancer, protons caused more acute skin reaction than photons. That finding, alongside meningioma data that show excellent but modality-independent tumour control, frames a month in which the field's attention turns from whether protons work to what they cost patients in the short and long term, how secondary risks arise, and whether the people who would benefit can reach a centre at all.

Breast and head and neck: the skin question

The DBCG Proton Trial is a national phase III study randomising patients with invasive breast cancer or DCIS, who were expected to receive a high heart or lung dose with photons, between protons and photons. Its pre-planned acute side-effect sub-study included 167 patients and 186 treated breasts (84 proton, 83 photon), using moderate hypofractionation. Skin effects peaked in week five and were markedly higher after protons: grade 2–3 dermatitis in 71% versus 20%, pruritus in 20% versus 5%, and local pain in 17% versus 4%. Yet all grade 2–3 skin effects were transient, resolving to grade 0–1 within two to four weeks. Respiratory symptoms, dysphagia and fatigue were mild and similar between arms, arm and shoulder morbidity was rare, and no patient discontinued treatment because of side effects. The authors judge proton therapy feasible and clinically acceptable, while recognising the higher rate of transient skin toxicity.

In head and neck cancer the picture is different. A prospective, non-randomised cohort of 158 patients at Kaohsiung Chang Gung Memorial Hospital (113 VMAT, 45 IMPT) found maximum acute dermatitis comparable between modalities (grade 2: 30.0% versus 29.5%; grade 3: 7.0% versus 4.8%), and no difference in patient-reported skin quality of life on Skindex-16. Clinically meaningful worsening affected 37.0% of patients at the end of treatment but only 7.0% three months later. Advanced nodal disease and doses of 66 Gy or more, not modality, predicted worse skin outcomes. The authors acknowledge the non-randomised design and the absence of skin-specific dosimetry.

Meningioma: control is excellent, survivorship is the issue

A systematic review and meta-analysis of 24 studies and 4,673 patients compared radiotherapy options for WHO grade 1 meningioma. All achieved excellent local control: at 5 years, 95.9% for conventionally fractionated protons, 93.0% for conventionally fractionated photons, 95.3% for linac-based stereotactic radiosurgery or radiotherapy, and 89.8% for Gamma Knife. Meta-regression found no significant association between modality or tumour volume and control, and fractionated protons and photons maintained control despite being used for larger tumours than stereotactic techniques. Five-year overall survival in the proton cohort was 93.1%. The case for protons in benign meningioma, in other words, will not rest on tumour control.

It may instead rest on how patients live afterwards, and here a prospective five-year cohort offers sobering detail. Of 126 adults treated with proton therapy, mostly for meningioma, 77 (61%) completed the five-year assessment. Global health status did not change significantly, but the QLQ-C30 summary score declined modestly, fatigue and insomnia worsened to a clinically meaningful degree, physical and cognitive functioning fell, and emotional functioning improved. On the brain-tumour module, hair loss rose meaningfully while headaches decreased. The authors are careful not to attribute these changes to treatment, but argue for structured long-term follow-up and rehabilitation.

Secondary risks in children and the CSI question

At the Paul Scherrer Institute, a previously reported cohort of 119 children treated with pencil-beam scanning protons for intracranial ependymoma had four radiation-induced tumours (3.4%) after a median follow-up of 5.3 years. A follow-up analysis compared this with a cohort of 51 children treated on commercial-like scanning systems, in which no such tumours were reported after a median of five years, and used Monte Carlo simulation to show higher secondary neutron exposure from PSI's Gantry 1 than from Gantry 2. The authors frame the link between system-specific neutron dose and tumour risk as hypothesis-generating, given the small number of events and confounding by chemotherapy and treatment volume — but the question of whether delivery hardware influences late risk is one every paediatric centre will want answered.

A systematic review of 32 studies of radiotherapy for leptomeningeal disease from solid tumours found median reported overall survival of 10.7 months for proton craniospinal irradiation against 4.1 months for photon CSI and 4.5 months for whole-brain or non-CSI photon treatment. Crude grade 3 or higher haematological toxicity was 59.4% with proton CSI and 80.0% with photon CSI. The authors stress that heterogeneity, era, patient selection and systemic therapy prevent any claim of superiority, and call for prospective comparison.

Carbon ions: numbers on RBE, and a gynaecological map

A propensity-matched study of 244 patients with locally advanced non-small cell lung cancer receiving radiotherapy with systemic therapy found significantly better locoregional control with carbon ions than photons. Fitting tumour control probability models to 3-year locoregional control, the authors derived a D₅₀ of 15.44 Gy for carbon ions and 59.94 Gy for photons, giving a clinical RBE estimate of 3.88 — described by the authors as preliminary and model-dependent, and, to their knowledge, the first clinical RBE estimate for tumour control in this disease derived this way.

The PTCOG Gynaecologic Subcommittee published a scoping review of 41 studies. Most (63.4%) came from Japanese centres and only 12 were phase I or II trials. Carbon ions were mostly used for radical treatment of radioresistant tumours such as melanoma and locally advanced cervical cancer, and protons mainly in postoperative and recurrent settings. The subcommittee calls for standardised protocols, centralisation of rare gynaecological cancers within the PTCOG network, and adequately powered prospective studies.

Measuring what cannot be seen: BNCT and FLASH dosimetry

Two BNCT papers address the modality's persistent weakness: knowing what dose is actually delivered. At the Institut Laue-Langevin, a Compton camera array imaged the 478 keV prompt gamma rays from boron-10 at a clinically relevant 65 ppm under neutron irradiation for the first time, localising the emission with a spatial resolution of 22 mm and a signal-to-background ratio of 6.4. High count rates and back-scattered events remain limitations, which the authors expect improved event filtering and new detectors to address. Separately, a silicon-diode fast-neutron detector placed at the end of the beam duct of an 8 MeV proton–beryllium BNCT system gave real-time neutron monitoring with 124.6 counts per mC of proton charge and a relative standard deviation of 1.2%, stable regardless of the phantom at the patient position — a complement to the indirect proton-charge measurement now used to infer neutron dose.

For FLASH, a study of TLD-100 dosimeters on synchrotron proton beams found response independent of dose between 3 and 18 Gy, but dependent on in-pulse dose rate: over-response of 1–7% below 7 Gy/s, near-unity response between 7 and 300 Gy/s, and 1–7% under-response above 300 Gy/s. For remote audits of ultra-high dose rate programmes, that is a correction worth knowing.

Looking further ahead, a review of laser-plasma accelerators for protons, carbon ions and very high energy electrons sets out their promise of extremely high dose rates in a compact footprint, recent in vitro and in vivo results, and the many technical challenges that remain before clinical use.

Access: the map Europe needs

Perhaps the month's most policy-relevant paper modelled proton therapy access across 31 European countries and 69,918 subregions. Estimated demand in 2025 was 78,500 patients, and over 56,000 lived in areas with no access or critical shortages. To 2030, planned expansions should improve spatial equity in Southern and Eastern Europe, while Northern and Western Europe face falling capacity-to-demand ratios. Adding one strategically placed single-room facility reduced the national weighted Gini index by a mean of 35.4% across the ten countries with the largest modelled gains. The authors conclude that large high-throughput centres suit small, densely populated countries, but geographically extensive nations may need decentralised networks including single-room units.

What August adds up to

August's randomised breast data are a useful corrective: protons can increase some acute toxicities, and only longer follow-up will show whether the trade is worthwhile. Meningioma data suggest protons will be judged on survivorship rather than control, and the ependymoma analysis raises the possibility that how a centre delivers protons affects late risk. Better dosimetry for BNCT and FLASH, and an honest map of who can reach treatment, are the foundations on which those judgements will be made.