Proton Arc Therapy: Rotational Delivery in the Proton Era

• Proton arc therapy delivers radiation from dozens to hundreds of gantry angles in a continuous or step‑and‑shoot rotation, enabling substantially sharper dose conformality and reduced exposure to surrounding healthy tissue compared to conventional multi‑field proton therapy.
• Clinical translation is accelerating: the first patients have been treated with step‑and‑shoot proton arc for head‑and‑neck cancer, and comprehensive dosimetric studies have demonstrated meaningful reductions in predicted toxicity across oropharyngeal, nasopharyngeal, lung, liver, esophageal, and paediatric indications.
• Key barriers to routine adoption remain – including longer beam‑on times on existing cyclotron systems, sensitivity to anatomical changes that may require adaptive replanning, and unresolved trade‑offs in heart dose for some lung cancer cases – but rapid progress in planning algorithms, delivery sequencing, and hardware development is closing these gaps.

Proton therapy has long been regarded as a precision instrument, its Bragg‑peak advantage allowing dose to be deposited sharply at depth while sparing proximal tissues. Yet for all its elegance, conventional intensity‑modulated proton therapy (IMPT) typically uses only two to five static beam directions – a constraint inherited from the era of passive scattering and slow gantries. Proton arc therapy (PAT) breaks this limitation by rotating the beam around the patient, irradiating the tumour from dozens to hundreds of angles in a single or multiple arcs. The conceptual leap is small; the dosimetric and clinical implications are large.

The Rotational Advantage

The fundamental insight behind PAT is that increasing the number of beam angles gives the optimiser more degrees of freedom to shape the dose distribution. In the original demonstration of spot‑scanning proton arc therapy (SPArc), Ding et al., 2016 showed that a rotational plan could reduce integral dose by 8‑14% compared to standard IMPT for oropharyngeal and lung cancer, while simultaneously improving target conformity and reducing maximum skin dose by 60%. The mechanism is straightforward: with many overlapping beams, the high‑dose region becomes tightly confined to the target, while each individual beam contributes a small fraction of the total dose to any off‑target location, creating a diffused low‑dose bath that is much shallower than the sharp lateral penumbra of static‑field plans.

Subsequent work confirmed this principle across multiple disease sites. In a cohort of 223 oropharyngeal cancer patients, de Jong et al., 2023 found that PAT reduced integral dose by 18% relative to IMPT and by 54% relative to volumetric‑modulated arc therapy (VMAT). When normal‑tissue complication probability (NTCP) thresholds from the Dutch National Indication Protocol were applied, PAT increased the proportion of patients qualifying for proton therapy from 66% to 81% – a direct translation of dosimetric advantage into patient access.

Clinical Evidence and Early Translation

The dosimetric benefits of PAT are now supported by a rapidly growing body of evidence. For head‑and‑neck cancer, Tattenberg et al., 2024 demonstrated that implementing PAT (with current 3.5% range uncertainties) reduced mean NTCPs for xerostomia and laryngeal oedema by 3.2 percentage points in nominal scenarios and by 4.2 points in worst‑case scenarios – more than double the benefit achievable by reducing range uncertainties from 3.5% to 1% in static IMPT. For paediatric ependymoma, Henjum et al., 2025 reported that PAT significantly improved CTV coverage and robustness while reducing mean cochlear dose from 18.3 Gy to 9.3 Gy (RBE) compared to three‑field IMPT.

In esophageal cancer, Cong et al., 2026 found that PAT achieved superior dose conformity and reduced lung V20 from 13.4% to 6.9% compared to IMPT, with lower spinal cord maximum doses. For hepatocellular carcinoma, Liu et al., 2024 showed that PAT reduced mean liver dose from 14.7 Gy to 10.6 Gy and lowered the predicted probability of radiation‑induced liver disease in challenging cases with large, deep targets.

Most strikingly, clinical implementation has begun. In June 2024, the first patient with head‑and‑neck cancer was treated using a step‑and‑shoot proton arc technique at a single institution, with 33 fractions delivered over two months (Liu et al., 2025). Fracchiolla et al., 2025 subsequently reported the world's first patient treatments using static PAT (discrete beams at 30 angles) for complex head‑and‑neck targets, demonstrating significant out‑of‑target dose reduction without compromising target coverage or delivery time. These milestones move PAT from a purely computational exercise to a clinically feasible reality.

The Delivery Efficiency Challenge

Despite its dosimetric promise, PAT faces a fundamental efficiency problem. Rotating a multi‑ton gantry while simultaneously switching proton energies, moving scanning magnets, and delivering spots creates a complex choreography that, unless carefully orchestrated, can prolong treatment times beyond 15–20 minutes.

The bottleneck is energy‑layer switching time (ELST). On cyclotron‑based systems, changing the beam energy can take several seconds – especially for upward switches – and PAT plans typically require far more energy layers than conventional IMPT plans. Burford‑Eyre et al., 2024 emulated sawtooth PAT delivery on a Varian ProBeam system and found that continuous arc delivery increased beam‑on and gantry movement time by 4.9–7.5 minutes for single arcs and 7.3–12.4 minutes for dual arcs compared to static IMPT, with upward ELSTs contributing approximately 60% of the total energy‑switching time. The authors concluded that "no clear benefit was observed regarding beam‑on time for sawtooth PAT relative to static IMPT" on current hardware.

This has spurred intense optimisation work. Several strategies have emerged to shorten delivery time without sacrificing plan quality. Wuyckens et al., 2024 developed a hybrid method combining fast geometry‑based energy‑layer pre‑selection with dose‑based filtering, achieving up to 22% reduction in delivery time for oropharyngeal cases and 15% for lung cases. Zhao et al., 2022 introduced an evolutionary algorithm that directly optimises spot sparsity with beam‑delivery time as an input, enabling planners to specify a desired treatment duration. More recently, the multi‑arc constant‑energy approach (Ma et al., 2025) eliminated energy switching entirely within each arc by using block orthogonal matching pursuit to select beam energies – reducing energy‑switching time from over 15 minutes to under 3 seconds for a liver case while improving target conformity.

Liu et al., 2024 demonstrated that on the IBA ProteusONE platform with a short ELST of 0.2 seconds, dynamic arc delivery could actually be faster than IMPT, with average delivery times of 160 versus 182 seconds. The practical conclusion is that delivery efficiency is not an inherent limitation of PAT but rather a system‑specific challenge that can be solved with fast energy switching, smarter plan optimisation, and gantry velocity scheduling.

Hard Trade‑Offs: Not All Sites Benefit Equally

The enthusiasm for PAT must be tempered by the observation that dosimetric superiority is not universal. The most instructive counterexample comes from lung cancer. In a planning study of 14 unresectable lung cancer patients, Chocan et al., 2024 found that while PAT improved dose conformality by 66% compared to IMPT, it also increased heart mean dose by 23% (1.4 Gy), translating to a statistically significant 2% increase in predicted 2‑year mortality risk. The interplay effect – the interaction between breathing motion and the rotating beam – was also more detrimental to PAT plans, with wider DVH bandwidths for both target coverage and hotspots.

This finding highlights a critical principle: PAT's advantage depends on anatomy. For tumours that lie close to critical organs such as the heart, the increased number of beam directions may include more trajectories that pass through that organ, offsetting the conformality gain. Wei et al., 2025 found the opposite for early‑stage lung tumours with motion >10 mm, showing that PAT actually provided better robustness against interplay effects than IMPT, with median GTV D95 deviation of only −0.6% versus −2.4% for single‑fraction stereotactic body radiotherapy. The discrepancy likely reflects differences in target size, location, and fractionation, underscoring that site‑specific evaluation remains essential.

Other disadvantages include sensitivity to rotational setup errors, which Chang et al., 2022 addressed by developing a multi‑CT robust optimisation framework that mitigates dose perturbations from patient roll. Liu et al., 2025 showed that systematic gantry position errors can degrade gamma passing rates, particularly for brain stereotactic cases, while random errors had minimal impact – suggesting that quality assurance tolerances for PAT should be tighter than for static‑field delivery.

The Biological Dimension: RBE and LET

Proton arc therapy introduces new opportunities and challenges related to relative biological effectiveness (RBE). Because RBE increases with linear energy transfer (LET) at the distal edge of the Bragg peak, distributing dose from many angles can concentrate high‑LET regions in the tumour while spreading low‑LET dose across healthy tissue. Reiners et al., 2025 demonstrated that dynamic arc delivery increased RBE enhancement to the optic chiasm from 1.05 (static PBS) to 1.18, but that LET optimisation with a maximum LETd constraint of 2.5 keV/µm applied above 80% of the maximum structure dose could reduce RBE enhancement back to 1.02–1.03. For prostate cancer, Vieceli et al., 2026 showed that four‑beam and arc LET‑optimised plans achieved 6–8% higher RBE‑weighted dose to the target while sparing the urethra by exploiting its lower α/β ratio.

These results suggest that PAT can be used for biological dose shaping, but also that without explicit LET optimisation, the complex interplay of many beams may inadvertently increase RBE‑weighted dose to organs at risk. Integrating LET constraints into the optimisation loop is an active area of research and will likely be a prerequisite for safe clinical deployment in sites adjacent to neurological structures.

Clinical Implementation and the Road Ahead

The first‑in‑human treatments reported by Liu et al., 2025 and Fracchiolla et al., 2025 provide proof of principle that PAT can be delivered safely within existing clinical workflows. The step‑and‑shoot approach used in the head‑and‑neck case (9 static fields at 20° intervals) required 16.7 minutes average delivery time per fraction, and daily synthetic CT monitoring detected target coverage degradation of 3–5% due to weight loss and setup changes, triggering a replan after 12 fractions. This experience underscores that PAT's sharper dose gradients increase sensitivity to anatomical changes, making robust image guidance and adaptive capabilities essential.

Looking ahead, the ultimate form of PAT is continuous gantry rotation with dynamic dose delivery – true SPArc – which would eliminate the dead time between static fields and reduce treatment times to 4–5 minutes. Liu et al., 2024 showed that dynamic SPArc could increase daily patient throughput by 30% compared to IMPT in a single‑room centre. However, full dynamic delivery requires significant engineering: the gantry must maintain a commanded velocity profile with minimal lag, the beam control system must synchronise spot delivery with gantry position, and safety systems must handle rotational motion. Several manufacturers are now developing commercial arc modules, and the first prototype deliveries have been achieved on an IBA ProteusONE system.

The debate over whether PAT or FLASH radiotherapy represents the more promising next step for proton therapy (Kang and Ding, 2023) is not a zero‑sum contest. PAT is an engineering evolution that is ready for clinical deployment now, potentially expanding the patient population that benefits from protons. FLASH is a radiobiological revolution that will require years of systematic investigation. The two are complementary rather than competitive: the ultimate system may be a FLASH‑capable proton arc – SPLASH – that combines the dosimetric elegance of rotational delivery with the normal‑tissue sparing of ultra‑high dose rates.

For now, proton arc therapy stands as one of the most consequential technical advances in particle therapy since the introduction of pencil beam scanning. It does not solve every clinical problem, and for some patients it may even be counterproductive. But the evidence accumulated over the past decade makes a compelling case that for the right indication – head‑and‑neck cancers, paediatric tumours, deeply located liver and oesophageal lesions – PAT offers a degree of normal‑tissue sparing that static‑field IMPT cannot match. The remaining tasks are to prove its benefit in prospective trials, to refine its efficiency on commercial platforms, and to ensure that the routine clinical workflow can handle the demands of rotational, adaptive, and potentially FLASH‑capable proton therapy.

References

Ding X et al. Spot‑Scanning Proton Arc (SPArc) Therapy: The First Robust and Delivery‑Efficient Spot‑Scanning Proton Arc Therapy. · PMID: 27869083

de Jong BA et al. Proton arc therapy increases the benefit of proton therapy for oropharyngeal cancer patients in the model based clinic. · PMID: 37059337

Tattenberg S et al. Impact of and interplay between proton arc therapy and range uncertainties in proton therapy for head‑and‑neck cancer. · PMID: 38324904

Henjum H et al. Dosimetric comparison of intensity‑modulated proton therapy and proton arc therapy for pediatric ependymoma. · PMID: 40356085

Cong X et al. Advancing proton therapy: Dosimetric superiority of proton arc therapy over IMPT in esophageal cancer treatment. · PMID: 42086367

Liu P et al. Investigate potential clinical benefits and linear energy transfer sparing utilizing proton arc therapy for hepatocellular carcinoma. · PMID: 39326286

Liu P et al. First Clinical Implementation of Step‑and‑Shoot Proton Arc Therapy for Head and Neck Cancer Treatment. · PMID: 40474940

Fracchiolla F et al. Static proton arc therapy: Comprehensive plan quality evaluation and first clinical treatments in patients with complex head and neck targets. · PMID: 39939287

Burford‑Eyre S et al. Emulating the Delivery of Sawtooth Proton Arc Therapy Plans on a Cyclotron‑Based Proton Beam Therapy System. · PMID: 39409935

Wuyckens S et al. Efficient proton arc optimization and delivery through energy layer pre‑selection and post‑filtering. · PMID: 38742774

Zhao L et al. An evolutionary optimization algorithm for proton arc therapy. · PMID: 35878608

Ma J et al. A multi‑arc and constant‑energy‑per‑arc treatment planning method for efficient proton arc therapy. · PMID: 40983879

Liu G et al. Development of a standalone delivery sequence model for proton arc therapy. · PMID: 38064634

Chocan MS et al. A dosimetric and robustness analysis of proton arc therapy with early energy layer and spot assignment for lung cancer versus conventional intensity modulated proton therapy. · PMID: 39473175

Wei WG et al. Comparing the Robustness of Intensity‑modulated Proton Therapy and Proton‑arc Therapy Against Interplay Effects of 4D Robust‑optimised Plans for Lung Stereotactic Body Radiotherapy. · PMID: 39847967

Chang S et al. Introduce a rotational robust optimization framework for spot‑scanning proton arc (SPArc) therapy. · PMID: 36546347

Liu G et al. The first investigation of spot‑scanning proton arc (SPArc) delivery time and accuracy with different delivery tolerance window settings. · PMID: 37774715

Reiners K et al. RBE‑Guided Treatment Planning, LET Optimization, and Implications of Proton Arc Therapy for the Sparing of Nervous Tissue in Head and Neck Proton Therapy. · PMID: 41374925

Vieceli M et al. Robust Four‑Field and Arc LET‑Optimized Proton Prostate Cancer Treatment Plans With Urethral Sparing. · PMID: 42063694

Kang M, Ding X. FLASH instead of proton arc therapy is a more promising advancement for the next generation proton radiotherapy. · PMID: 37431574