Liver Protection, Paediatric Benchmarks and the LET Question: Particle Therapy in June 2026

June's literature returns again and again to a single question: where does the physical advantage of a particle beam become a measurable clinical one? In the liver, the answer is becoming clearer. In the paediatric brain, the field now has benchmarks against which to judge it, and a sharper view of the residual biological uncertainty at the end of the proton range. Around these clinical results sit a national consensus on who should be treated, and a set of preclinical and technical studies that are steadily moving FLASH and adaptive particle therapy towards the clinic.

The liver: a consistent protective signal

A meta-analysis comparing proton with photon radiotherapy for hepatocellular carcinoma pooled four retrospective cohort studies. Its most robust finding concerns toxicity rather than survival. Proton therapy significantly reduced the risk of radiation-induced liver disease (OR 0.26, 95% CI 0.15–0.46) and of non-classical RILD (OR 0.27, 95% CI 0.15–0.47), with no heterogeneity and no significant publication bias for either outcome, giving moderate certainty. The number needed to treat to prevent one case of liver injury was approximately five.

The survival result is more striking on paper but weaker in substance. The hazard ratio for overall survival was 0.45 (95% CI 0.34–0.60), but with moderate heterogeneity (I² = 48%) and significant publication bias the authors grade the certainty as low and call any survival benefit inconclusive. The honest reading is that proton therapy's value in HCC is, for now, best argued through liver preservation, while the survival question awaits prospective, multicentre data.

Paediatric and young-adult benchmarks

Two systematic reviews give clinicians something they have long lacked: structured expectations for young patients treated with protons. The first, covering paediatric central nervous system tumours, included 77 studies, 40 of which reported overall survival in a total of 3,798 patients. Statistically significant time-dependent survival models emerged for medulloblastoma, ependymoma and base-of-skull chordoma. Low-grade glioma and craniopharyngioma showed overall survival above 95% up to 10 years after treatment. The late-effects picture is varied: endocrinopathies ranged from 11.7% in ependymoma to 94% in craniopharyngioma, vasculopathy from 1% in low-grade glioma to 36% in craniopharyngioma, and hearing loss from 4% in ependymoma to 26.3% in medulloblastoma. The variation reflects tumour location and treatment burden as much as modality, and the authors present the figures as reference benchmarks for counselling families and for future comparative research.

The second review, of 35 primary studies in adolescents and young adults aged 15–39, takes a broader view of modern radiotherapy. For CNS and skull base tumours, proton therapy showed high local control, although anterior pituitary dysfunction was frequent at 36.3%. In testicular seminoma, protons significantly reduced acute gastrointestinal toxicity (0% versus 29.5% grade 1–2 diarrhoea) and were associated with a 50–60% lower model-estimated risk of second malignancy. The review is candid that particle therapy evidence remains scarce for clinically important AYA sites such as sarcoma and breast cancer, and that the gap between modelled second-cancer risk and observed latency will only be closed by long-term registries.

Dose, LET and the brainstem

Concern that elevated linear energy transfer at the distal edge of a proton beam might raise the risk of normal-tissue injury has hovered over paediatric practice for years. A voxel-wise case-control study drawn from 1,200 CNS tumour patients treated with double-scattering proton therapy between 2006 and 2023 examined 12 children who developed symptomatic brainstem necrosis against 36 matched controls. Mean dose in necrotic brainstem voxels was 54.1 Gy(RBE), against 42.6 Gy(RBE) in healthy voxels. In voxels receiving more than 54 Gy(RBE), mean LET was 2.9 keV/µm in necrotic tissue and 2.7 keV/µm in healthy tissue.

Mixed-effect modelling found a significant positive interaction between dose and LET, so elevated LET was associated with a higher probability of necrosis in high-dose regions, and LET-weighted dose was significantly higher in necrotic voxels. Yet heterogeneity between case-control groups was considerable, and dose-matched analyses found higher LET in necrotic voxels in only a subset of them. The authors' conclusion is measured: necrosis was primarily dose-driven, with LET a contributor in high-dose regions rather than a uniform effect. For planners, that supports attention to LET where dose to the brainstem is already high, without treating it as the dominant driver.

Selecting the right patients

A RAND/UCLA appropriateness process in Spain, run through 12 national working groups representing 118 specialists, rated 72 clinical statements across 13 oncological domains. After two rounds, 61 statements were retained, all reaching consensus: 53 judged appropriate for proton therapy, six uncertain and two inappropriate. Paediatric and adolescent oncology, skull base tumours, sarcomas, selected head and neck cancers, re-irradiation and anatomically complex cases led the appropriate list. The framework is explicitly intended to reduce variation in practice and support equitable access.

Two clinical series show what careful selection looks like in practice. In a prospective study of definitive proton therapy for non-small cell lung cancer in patients with idiopathic pulmonary fibrosis — a group notoriously vulnerable to thoracic radiotherapy — 41 of 55 enrolled patients were analysed. Grade 3 or higher pulmonary toxicity occurred in 17.1%, including grade 4 or higher events in 9.8%. The split by stage is the key result: 3 of 32 patients (9.4%) with stage I–II disease versus 4 of 9 (44.4%) with stage III. One-year overall survival was 67.6% and locoregional control 93.2%. Protons made early-stage treatment acceptable; they did not make stage III disease safe.

From the Proton Collaborative Group registry, 79 patients with non-metastatic adenoid cystic carcinoma of the head and neck treated between 2012 and 2023 achieved 3-year locoregional control of 98%, progression-free survival of 80.5% and overall survival of 89.2% at a median follow-up of 3 years. Advanced T stage, positive margins and definitive-intent treatment were not associated with worse locoregional control. Acute grade 3 toxicity occurred in 16 patients and late grade 3 toxicity in two, with no grade 4 or higher events.

Skin and blood: where protons do not differ

Two studies are useful precisely because they find little difference between modalities. In 176 women receiving post-mastectomy radiotherapy (93 intensity-modulated proton therapy, 83 photon), acute grade 2 or higher dermatitis occurred in 47% and 48% respectively, and grade 3 dermatitis in 3% and 7%, neither difference significant. At 12 months, photon patients reported more skin colour change (26% versus 6%), the only patient-reported skin outcome to differ. The authors attribute the comparable results to deliberate skin constraints in proton planning — a reminder that the skin dose of IMPT is a planning choice.

A prospective study of 35 primary brain tumour patients treated with radiotherapy alone found no significant change in total circulating lymphocytes or their subpopulations, with modelled total blood doses of around 0.1–0.2 Gy. Proton therapy gave greater dosimetric sparing of blood and lymph nodes, but in limited intracranial volumes there was little lymphopenia to prevent.

Carbon ions and BNCT: outcomes and open questions

A single-centre series of 48 patients with unresectable primary or recurrent soft tissue sarcoma treated with carbon ions, 83.3% of them with recurrent disease, reported 5-year local control of 60.3% and overall survival of 54.7% at a median follow-up of 48.6 months. Patients receiving a biologically effective dose of 129 Gy or more had significantly better local control and progression-free survival. Late grade 3 toxicities were few, though the authors caution about lesions near critical nerves.

In accelerator-based boron neutron capture therapy, a retrospective analysis of 30 head and neck cancer patients found an objective response rate of 90% and a complete response rate of 50%, with complete response predicting better two-year survival and control. But neither the 18F-BPA PET SUVmax nor the conventional individual-model doses showed a significant association with complete response. That challenges a cornerstone of current BNCT dose planning, and the authors conclude that SUVmax-based dosimetry needs refinement.

FLASH and adaptive workflows move closer

The most consequential preclinical result of the month comes from a bioRxiv preprint on synchrotron-based proton FLASH. Mice received 10 Gy whole-brain irradiation on a clinical synchrotron using single-spill pencil-beam scanning, multi-spill scanning with a roughly two-second inter-spot delay, or passive scattering. Multi-spill delivery qualified as FLASH only under the dose-averaged dose rate definition, yet recognition memory was preserved across all three FLASH techniques compared with conventional delivery, and FLASH-treated mice had fewer Iba1-positive microglia. Because multi-spill delivery is what clinical-scale paediatric targets would require, the finding bears directly on whether synchrotron centres can offer FLASH, and on how dose rate should be reported in trials.

For carbon ions, a second facility reproduced the in vitro cell-sparing effect of ultra-high dose rate irradiation using matched beam parameters (7 Gy, 100 Gy/s versus 1 Gy/s). Fibroblasts and bronchial epithelial cells showed higher survival and less DNA damage, enhanced at higher LET, while salivary gland cells showed a smaller effect. The authors interpret this as evidence that the effect depends on beam parameters rather than on the device.

Finally, a volunteer study of 20 women tested a shuttle-based workflow for daily MRI-guided particle therapy, in which patients remain immobilised between scanner and beam. The external pelvic outline was highly stable (median Dice 0.99 across transfers), and physiological organ motion, correlated with bladder volume, remained the main source of variability, only marginally affected by transport. It is a practical step towards online adaptive particle therapy without an in-room MRI.

What June adds up to

The month's evidence strengthens the case for particle therapy where organ preservation is the goal — the liver, the developing brain, the young adult with decades of survivorship ahead — while repeatedly showing its limits: stage III lung disease in fibrotic lungs, skin outcomes that depend on planning rather than modality, BNCT dosimetry that does not yet predict response. The LET analysis and the FLASH preprint share a theme: biological effect depends on how a beam is delivered and measured, not only on dose. Turning that understanding into prospective trials, consistent reporting standards and fair selection frameworks such as Spain's is the work that remains.