Particle therapy intelligence briefing — April 2026

This month’s particle therapy literature brings granular clarity to an old question: which patients benefit most from which modalities, and at what cost? From the nasopharynx to the chest wall, new studies refine the risk-benefit calculus by linking dosimetric precision to patient-specific outcomes, while emerging tools in carbon-ion therapy begin to personalise the adaptive process itself.

The swallowing cost of sparing: proton therapy for nasopharyngeal carcinoma

The strongest clinical signal this month comes from a prospective comparison of intensity-modulated proton therapy (IMPT) and volumetric modulated arc therapy (VMAT) in 104 patients with nasopharyngeal carcinoma. The study by Huang and colleagues found that IMPT delivered significantly lower mean doses to the oral cavity, middle and inferior pharyngeal constrictors, and the supraglottic larynx. Early weight loss — a clinical surrogate for acute dysphagia — was substantially reduced in the proton group.

Yet the one-year swallowing-related quality of life, measured by the MD Anderson Dysphagia Inventory, did not differ between arms. Dose to the superior pharyngeal constrictor consistently predicted worse scores at both the end of treatment and one year, irrespective of modality. This dissociation between dosimetric advantage and patient-reported outcome underscores a principle now emerging across multiple sites: organ-at-risk dose reduction is necessary but not sufficient. The minimum clinically meaningful threshold for constrictor dose may need to be lower than current constraints, a question that prospective dose-escalation or de-escalation trials could answer.

A companion narrative review on skull base tumours, by Loganadane and colleagues, reinforces the theme that the relationship between physical dose and tissue injury is not linear. In proton therapy, elevated linear energy transfer at the distal Bragg peak may contribute to focal radiation-associated image changes in the temporal lobes, even when integral brain dose is reduced. The review synthesises evidence that risk increases when more than 1% of healthy brain receives >57.6 Gy (RBE) or when V67Gy exceeds 0.17 cc. Distinguishing transient radiation-induced contrast enhancement from tumour progression requires multimodal imaging and discipline: the authors recommend surveillance for asymptomatic lesions rather than premature intervention.

Biological dose as a differentiator: protons versus carbon ions in the head and neck

A technical comparison by Wakisaka and colleagues directly addresses the biological gap between beam species. By incorporating LET-dependent variable α/β ratios into EQD2 and normal tissue complication probability calculations for nine head and neck cancer patients, the study found that carbon ions produced lower EQD2 in organs such as the brain, skin, and oral mucosa compared with protons. The α/β ratio for protons remained stable at approximately 3 Gy across tissues, while for carbon ions it varied markedly along the beam path, reaching up to 20 Gy. In NTCP analysis, carbon-ion therapy showed a significant reduction in predicted acute toxicities, particularly dermatitis. Late toxicities were low for both modalities.

The clinical relevance of this advantage will depend on whether it translates in prospective series. But the methodological contribution is clear: direct comparisons between particle species that ignore LET-dependent biology risk misleading conclusions. Future comparative trials should plan for variable RBE modelling as a stratification or sensitivity analysis.

Chest wall caution: capsular contracture after proton post-mastectomy radiotherapy

The largest cohort comparison of pencil-beam scanning proton therapy versus IMRT in the post-mastectomy reconstruction setting, reported by Zerey and colleagues, introduces a note of caution. Among 175 patients, the two-year capsular contracture rate for protons with direct-to-implant reconstruction reached 50%, compared with 35% for photons with direct-to-implant, 23% for protons with tissue expander and implant, and 12% for photons with tissue expander and implant. In multivariable analysis, direct-to-implant reconstruction carried a hazard ratio of 3.0 for capsular contracture compared with two-stage reconstruction. The effect of proton therapy itself did not reach statistical significance in the adjusted model (HR 1.76, p=0.083), but the signal is consistent with prior concerns about biological dose distribution in the chest wall.

This study does not settle the question of whether proton therapy increases capsular contracture risk. But it does make a compelling case that the choice of reconstruction method — particularly the decision to irradiate a permanent implant versus a temporary expander — may matter more than the beam modality. For patients considering proton therapy and direct-to-implant reconstruction, a frank discussion of the 50% two-year contracture rate should be part of shared decision-making.

Liver and oesophagus: defining the incremental benefit

Two large retrospective studies from Asia address the question of when particle therapy alone is sufficient and when combination therapy adds value. In hepatocellular carcinoma, Cheng and colleagues analysed 917 patients treated with proton beam therapy alone or after transarterial chemoembolisation. After propensity score matching, there was no difference in overall survival. Local control showed a trend favouring the combination (97.4% vs 94.8% at two years, p=0.068), and a modest benefit emerged in the subgroup with good liver function and no portal vein tumour thrombus. The authors conclude that the absolute benefit of adding TACE is small and should be weighed against patient factors — a conclusion supported by the excellent local control achieved with PBT alone.

The Japanese prospective registry study of particle therapy for oesophageal cancer, led by Ono and colleagues, compared outcomes with historical Japan Clinical Oncology Group photon trials. Three-year overall survival rates were 94.7% for stage I and 84.9% for stage II/III — comparable to the photon benchmarks. No grade 3 or higher cardiopulmonary toxicities were observed. The study adds to the evidence that particle therapy for oesophageal cancer is safe and effective, but does not demonstrate superiority over modern photon techniques. Without randomised comparison, the argument for particle therapy rests on the toxicity profile rather than survival advantage.

Personalising adaptation: a delta-omic model for carbon-ion therapy

One of the most operationally useful papers this month comes from Ju and colleagues, who developed a decision-support tool for adaptive radiotherapy during carbon-ion therapy for NSCLC. Among 69 patients, 46.4% triggered an adaptive replan based on anatomical or dosimetric changes. The model, combining one dose-volume histogram change metric, two delta-radiomic features, and two delta-dosiomic features, achieved a validated AUC of 0.897. At a high-sensitivity threshold, it identified all patients who would need adaptation while potentially reducing weekly CT evaluations in 21.4% of patients. At the Youden-optimal threshold, accuracy reached 78.6%, classifying half of patients as low risk for adaptation.

In a resource-constrained environment where each adaptive replan consumes physics and therapist time, a pre-treatment prediction of who will need adaptation — and who will not — could streamline workflows without compromising safety. The model requires prospective validation but represents the kind of practical tool that could make carbon-ion therapy more scalable.

Verification through activation: F-18 PET for proton dose mapping

Finally, a technical advance from Balcerzyk and colleagues demonstrates that F-18 activation during proton therapy can serve not only for range verification but for full three-dimensional dose mapping. Using a brain phantom irradiated to 3 Gy, the team showed that F-18 activity distribution reproduced the delivered dose within 5% and matched the planned dose fall-off within 1 mm. Among positron-emitting isotopes produced by proton irradiation, F-18 offered the closest spatial correspondence to planned dose. The activity levels were detectable on modern BGO-based PET systems approximately one hour after treatment.

The technique moves beyond prompt gamma-based range assessment toward true dose verification. If translated to the clinic, it could provide daily quality assurance of delivered dose — particularly valuable for sites where anatomical change or set-up uncertainty challenge the assumption that what was planned is what was delivered.

What the evidence demands

Across eight papers this month, a consistent message emerges: particle therapy’s dosimetric advantages are real, but they do not automatically translate into improved patient outcomes. The nasopharyngeal carcinoma study shows that organ-sparing can coexist with unchanged quality of life if dose to specific substructures is not adequately controlled. The breast reconstruction study suggests that surgical decisions may eclipse beam modality in determining complication risk. The carbon-ion head and neck work highlights that biological dose modelling must be integrated into comparisons between particle species. And the HCC and oesophageal studies remind us that adding other treatments or choosing particle therapy without randomised evidence may offer marginal gains in well-selected subgroups.

At the same time, the tools to make particle therapy better are arriving. Delta-omic models can triage adaptive workflows. F-18 PET can verify dose delivery. And variable RBE models can sharpen the difference between protons and carbon ions into something clinically actionable. The field is moving from the question “is particle therapy better?” to the more useful question “better for whom, and at what threshold?”