Investigation of Spot Size Variations in Multiroom Proton Therapy Systems and Their Clinical Significance: An In silico Study

Abstract

Background and Purpose:

In multiroom proton therapy facilities, maintaining beam consistency is critical, particularly when treatment interruptions occur due to machine downtime. Among beam parameters, spot size frequently exceeds tolerance limits, potentially compromising treatment accuracy. This study aims to assess the clinical implications of spot size variations and establish benchmarks for beam matching, with a specific focus on the spot size parameter in proton therapy using tool for particle simulation (TOPAS), Monte Carlo (MC) simulation.

Materials and Methods:

The study analyzed the effects of spot size deviations (±0.3 mm and ± 0.6 mm) on proton therapy treatment plans using TOPAS MC simulations. The five variable spot sizes models were created in the RayStation treatment planning system by simulating the known spot size shift error in the TOPAS for 33 proton energies ranging from 70.18 to 226.2 MeV. These models were evaluated using both homogeneous phantom fields and heterogeneous clinical fields targeting the pelvis, brain, and prostate. Key dosimetric metrics target, coverage (TC), conformity index, and homogeneity index, were assessed. In addition, two-dimensional gamma analysis was performed at tolerances of 1%/1 mm, 2%/2 mm, and 3%/2 mm to quantify the clinical impact of spot size variations on treatment delivery accuracy.

Results:

The evaluation indicated that Model −0.6 mm achieved the highest target coverage (TC), while Model 0.6 mm resulted in the lowest. Comparative analysis within the homogeneous phantom revealed marked variations in TC and conformity indices among the tested models. Clinical investigation revealed that TC in the pelvic area was constant, but that TC in the brain and prostate was more sensitive to changes in spot size. Gamma analysis showed superior passing rates for the model ± 0.3 mm, particularly at 2%/2 mm and 3%/2 mm criteria, confirming its suitability for optimal treatment accuracy.

Conclusion:

Spot size variations significantly influence the accuracy of proton therapy, with deviations of ± 0.3 mm yielding the most favorable results. Adhering to this tolerance ensures consistent beam matching and precise treatment delivery across various clinical sites, supporting the reliability of multiroom proton therapy systems.

INTRODUCTION

Proton therapy is a highly advanced and rapidly evolving modality in radiation oncology, offering a precise and effective treatment option for cancer. Its ability to deliver targeted radiation while sparing surrounding healthy tissue makes it particularly advantageous for treating tumors that are resistant to conventional radiation therapies.[1] The pencil beam scanning (PBS) technique employs a focused proton beam, with the spot size defined by a Gaussian distribution of particles at one sigma (1 σ). This approach enables precise dose delivery by accurately targeting tumor cells while minimizing radiation exposure to surrounding healthy tissues.

Modern proton therapy facilities are typically equipped with multiple 360-degree rotating gantries, all powered by a single cyclotron in a multi-room treatment configuration designed to optimize patient throughput. Globally, a single cyclotron can support treatments across three to five rooms.[2,3] However, the high costs of establishing and maintaining proton therapy facilities remain a significant challenge, particularly in middle-income and developing countries.[4,5] Improving access to these advanced therapies is crucial, especially for pediatric patients, tumors near critical structures, and recurrent cancers.[6,7] Despite its many benefits, proton therapy presents technological complexities and operational challenges. Machine downtime, cyclotron malfunctions, or individual room issues in multiroom setups can disrupt treatment schedules. Facilities equipped with room-matching capabilities can mitigate these disruptions by allowing patient transfers between rooms. However, this requires precise magnetic optic tuning to ensure beam spot sizes meet stringent range parameters.[8,9,10] Advanced features such as room-matching capabilities, while optimizing patient throughput, increase costs due to the precision engineering needed to align gantries to a shared baseline value. Beam optics tuning is essential for maintaining the spot size tolerances necessary for room matching.[9] For institutions without dedicated beam-matching capabilities, alternative strategies are employed to align beam parameters and enable patient transfers during equipment downtime.[11] Our center is equipped with three PBS 360° rotating gantries. We analyzed beam parameters across treatment rooms to assess the feasibility of beam matching. The distal range (R90%) derived from the integrated depth dose (IDD) curve showed high consistency within 0.5 mm across all gantries. Beam output, calibrated using the gantry treatment room 3 (GTR3) as the baseline, exhibited a variance of < 1%. However, achieving uniformity in spot size remained a challenge. The vendor-calibrated initial spot size is based on baseline values specified in the system requirement document (SRD).[8] Spot size analysis revealed difficulties in maintaining the ± 0.3 mm tolerance across all gantries, raising questions about the clinical impact of spot size variations during patient transfers in emergency scenarios.[12]

The SRD defines baseline spot sizes for each energy level, with biannual preventive maintenance (PM) adjustments made to align spot sizes with baseline tolerances. This study evaluates the clinical impact of spot size deviations ranging from ±0.3 mm to ±0.6 mm across various energies in diverse clinical contexts. Literature reports varying tolerance recommendations for spot size deviations. Langner et al. observed spot size variations exceeding the vendor-specified 15% limit,[12] while the American Association of Physicists in Medicine (AAPM) TG224 recommends tolerances of ± 10% or ±1 mm.[13,14] Symmetry is another critical parameter, with a two-dimensional (2D) symmetry tolerance of <10% or 1 mm, as suggested by ion beam application (IBA). Although limited studies have quantified the clinical impact of spot size deviations between 0.1 mm and 0.6 mm using modeled data, some notable efforts exist. Rana et al. modeled the effects of spot size deviations of ±3 mm and ±6 mm on lung stereotactic body radiation therapy plans using treatment planning system (TPS).[15] Liu et al. investigated the influence of spot size and spacing on small-cell lung cancer treatment plans but did not detail their method for creating variable spot size models.[16] Chanrion et al. evaluated spot size variations of ± 10%, ±25%, and ± 50% for skull-base and prostate tumors using Monte Carlo (MC) dose engines to recalculate treatment plans with increased spot sizes.[17] Langner et al. reported spot size differences of < 0.5 mm and 10% across gantries in multiple institutions, emphasizing the need for stringent beam matching.[12] Rana et al. recommended maintaining spot size variations within ± 5% for effective beam matching in multi-room setups.[11] Kraan et al. categorized spot sizes into small (2.5–5 mm), medium (5–10 mm), and large (10–20 mm) across energy ranges from 230 MeV to 69 MeV, highlighting their impact on treatment precision.[18]

Even though previous studies have investigated the impact of spot size errors on various clinical sites, the methods used to generate inaccurate spot size models often involve scaling the spot size profiles relative to a reference model.[15] This study introduces an innovative methodology utilizing a tool for particle simulation (TOPAS) a Geant4-based MC dose calculation engine to precisely model variable spot sizes with constant deviations of up to ± 0.6 mm.[19] These models were integrated into treatment plans for various clinical sites to systematically evaluate the impact of spot size errors on critical treatment parameters. The research aims to quantify the influence of spot size variations across diverse clinical scenarios and assess the feasibility of beam matching in multiroom proton therapy setups. By examining spot size and symmetry deviations, the study underscores their critical role in influencing treatment precision and delivery efficiency.

MATERIALS AND METHODS

Validated tool for particle simulation model to create variable spot sizes to build model in the treatment planning system

A TOPAS MC simulation (Geant4-based MC dose engine, version 3.9) was utilized to generate Gaussian spot size distributions at varying depths in air for 33 discrete energy levels. The PBS proton therapy system was accurately modeled within the TOPAS MC framework and rigorously validated against multiple beam parameters. The TOPAS modeling was performed by integrating experimentally measured beam data from the PBS system, including IDD, spot size, and output. This validated model was subsequently employed to simulate variable spot size profiles (±3 mm and ± 6 mm) relative to the reference spot size, ultimately facilitating the modeling of a new machine in the RayStation TPS. A total of 10⁸ proton particle histories were used to simulate the spot profiles, and the simulation uncertainties were below 1% in the simulation. The RayStation TPS version 12A (RaySearch Laboratories AB, Stockholm, Sweden) was utilized to incorporate these spot size variations into clinical models. Since the RayStation TPS does not allow direct modifications of spot size within its existing beam model, TOPAS was used to simulate variable spot size profiles. Spot size parameters such as “beam position spread and beam angular spread along X and Y axis perpendicular to the beam axis” were systematically adjusted by ± 0.3 mm and ± 0.6 mm relative to the reference spot size, creating a set of alternative spot profiles for each of the 33 proton energy levels. In addition, TPS modeling requires spot size data at a minimum of three different depths. However, to enhance accuracy, both measurements and simulations were conducted at six depths: Isocenter (0 cm) and ± 10 cm, ±20 cm, and 35 cm relative to the isocenter as illustrated in Figure 1. This ensured a comprehensive dataset for modeling the variable spot sizes.

Figure 1.

Tool for particle simulation geometry for assessing spot size at different depths. Simulations were performed at the isocenter (0 cm) and at depths of − 20 cm, −10 cm, 10 cm, 20 cm, and 35 cm relative to the isocenter

To integrate these spot profiles into the RayStation TPS, the reference beam model was duplicated, and the TOPAS-generated spot size profiles were incorporated one by one while keeping the IDD and output data unchanged. This process resulted in the creation of four new beam models corresponding to +0.3 mm, −0.3 mm, +0.6 mm, and −0.6 mm spot size variations, in addition to the existing reference model. As a result, a total of five beam models were available to evaluate the impact of spot size deviations on clinical dose distributions, ranging from increased to decreased spot sizes relative to the reference.

Following the creation of these modified beam models, the spot size data were analyzed and compared to the reference values. These findings are presented in detail in the results section. Figure 2 provides a step-by-step visual representation of the workflow, summarizing the methodology used for this study

Figure 2.

Workflow for spot size error simulation, treatment planning system modeling, and clinical evaluation in proton therapy

Beam matching parameters in a multiroom proton therapy setup

Beam matching in a multiroom proton therapy system is governed by three critical parameters: Absolute dose, energy range, and spot size. Literature recommends acceptance criteria for these parameters as ± 2% for absolute dose, ±0.5 mm for range, and ± 5% for spot size.[11] In this study, both proton output and range met these criteria, supported by technical assistance from IBA engineers. The proton beam spot size, defined as one standard deviation (σ) along a specified axis, is derived from the full width at half maximum of the Gaussian intensity distribution. This parameter depends on beam energy, ranging from 2.8 mm at 226.2 MeV to 6.5 mm at 70.18 MeV, and may vary with axial angle.[14,20,21] Spot size measurements were performed using a Lynx scintillation detector (IBA Dosimetry, Germany), which consists of a 0.4 mm-thick gadolinium-based scintillating screen, a detection area of 30 cm × 30 cm, and an effective spatial resolution of 0.5 mm.[22] Measurements were conducted over 5 days across 30 energy levels and 12 gantry angles, spaced at 30° increments. A five-spot plan was generated using proton layer definitions in IBA’s Adapt Prescribe software. The plan included one spot at the center and four at the corners of a quadrant.[14] These measurements aimed to evaluate spot size variations between gantries, focusing on potential differences between the central spot and off-center spots scanned farther from the central axis.

To assess the clinical impact of spot size variability, variations of ± 0.3 mm (approximately 5% across spot sizes of 2.8–6.5 mm) and ± 0.6 mm (approximately 13%) were analyzed in treatment plans. This analysis highlights the significance of maintaining precise beam spot size consistency to ensure effective treatment delivery.

Profile modeling in RayStation treatment planning system

Spot profiles generated using TOPAS were employed to develop beam models in the RayStation TPS. The reference model was named Model Reference, whereas models with variations of ±0.3 mm in spot size were designated as Model 0.3 mm and Model −0.3 mm, and those with ±0.6 mm variations as Model 0.6 mm and Model −0.6 mm. Spot size profiles at depths of +35 mm, 0 mm, and −20 mm relative to the isocenter were imported into RayStation TPS, while other parameters, such as output and IDD, were kept constant across all models.

The modeling utilized the MC algorithm in RayStation, with the spot size values for each model plotted and presented in the results section. The developed models were used to assess the impact of spot size variations in both homogeneous phantom media and heterogeneous patient media. The analysis covered clinical sites including the brain, pelvis, and prostate, representing the full range of spot sizes listed in Table 1.

Table 1.

The tabulated values of volume and energy ranges in the site-specific plan

Plan	Volume (cc)			Minimum energy (MeV)			Maximum energy (MeV)
	Pelvis	Brain	Prostate	Pelvis	Brain	Prostate	Pelvis	Brain	Prostate
P1	786.1	89.6	41.0	95.2	81.5	161.9	192.6	133.9	198.8
P2	1460.8	19.0	27.4	89.1	98.6	158.2	200.2	128.3	195.0
P3	442.3	9.7	28.7	85.7	83.9	157.4	175.8	113.5	191.2
P4	577.1	25.3	33.9	84.8	80.2	167.1	160.0	134.3	198.8
P5	644.5	33.0	30.2	90.4	98.5	159.2	185.1	137.7	195.8

Treatment plans were initially created using the reference model (Model Reference) and then replicated for the other four models. The plans were recalculated with the newly developed beam models without any additional optimization, enabling a direct evaluation of the effects of spot size variability on treatment outcomes.

Assessment of spot size error in homogenous spread-out Bragg peak cube

The impact of spot size variations was initially assessed using a homogeneous spread-out Bragg peak (SOBP) cube. A 10 cm × 10 cm × 5 cm cube, with a 10 cm × 10 cm field size, 5 cm modulation, and 10 cm range, was created within a 40 cm × 40 cm × 40 cm water phantom in the RayStation TPS. This cube was designated as the clinical target volume (CTV) to evaluate the influence of spot size variations on dose–volume indices, including target coverage (TC), conformity index (CI), and homogeneity index (HI). Treatment plans were generated by varying the spot spacing and energy layer (EL) spacing to clinically relevant values of 0.5, 0.7, and 1. Spot placement was determined as 1.06 times the sigma value when the TPS parameter was set to 1, with reductions in this parameter decreasing the inter-spot distance. For EL spacing, 80% of the proximal falloff was used as the criterion to position the subsequent EL when the parameter was set to 1. A value <1 resulted in a reduced EL spacing, thereby increasing the total number of ELs. Adjustments to spot size and EL spacing were made based on the complexity of the treatment plan. Key evaluation metrics included TC, defined as the volume receiving 98% of the prescribed dose (V98%); CI at 98% of the prescribed dose; and HI at 98% of the prescribed dose. These metrics were analyzed across five models in both homogeneous and site-specific clinical scenarios.[23,24,25]

The CI is defined as the ratio of the volume receiving 98% of the prescribed dose within the CTV to the total volume of the CTV, expressed as:

CI = Volume of 98% PD received by the CTV/Total volume of the CTV      (1)

The HI is calculated as:

HI = D (V_{ctv, 98%})/D (V_{ctv, 2%})      (2)

where, D (V_{ctv, 98%}) represents the dose at which 98% of the target volume (CTV) is covered.

D (V_{ctv, 2%}) represents the dose at which 2% of the target volume (CTV) is covered,

The comparative results of this analysis, including variations in TC, CI, and HI across the models, are detailed in the results section.

Assessment of spot size error in clinical cases

To evaluate the impact of spot size variations, 15 clinical cases were analyzed, comprising five cases each of pelvic, brain, and prostate tumors. These cases were categorized based on tumor characteristics: Large-volume tumors (pelvis), small-volume tumors (brain), and high-range targets (prostate), as summarized in Table 1. These clinical scenarios span the entire beam range and spot sizes typically treated at our institute. Robust intensity-modulated proton therapy plans were initially developed using a reference beam model. The plans were optimized using a MC dose calculation engine in RayStation TPS to ensure adequate TC while minimizing doses to surrounding normal tissues. The grid size was kept at 1 mm for the dose calculation in the planning. Robust optimization incorporated density uncertainty of ± 3.5% and setup margins of 3 mm for brain cases and 5 mm for pelvis and prostate cases. The robust worst scenarios were derived from the 28 independent dose distributions created for the 3 mm for the brain and 5 mm for the pelvis setup, and 3.5% density uncertainty in robust evaluation. Subsequently, the final plans were recalculated using each of the five beam models, including the reference model, for all sites.

For further analysis, dose-volume histograms (DVHs) of each plan were exported for gamma evaluation. Comparisons were conducted between the reference model and models with spot size variations of ± 0.3 mm and ± 0.6 mm to assess gamma passing rates. The myQA Patient PSQA tool from myQA software (IBA Dosimetry) was used to compare the different planes to evaluate the gamma value. 2D gamma evaluations were conducted on various planes for each clinical site. Random dose planes from each plan were selected for gamma evaluation, with criteria set at 1%/1 mm, 2%/2 mm, and 3%/2 mm, following guidelines by Farr et al.[26] Results of the gamma evaluation, including model-to-model comparisons, are presented in the results section. This study was focused solely on analyzing the effects of spot size variations on clinical cases, the plans were not intended for use in actual treatments.

RESULTS

Beam matching parameters in a multiroom proton therapy setup

Spot size adjustments were based on the spot reference data (SRD) provided by IBA, with a tolerance of ±0.3 mm. Across 30 energy levels, three gantries, and 12 gantry angles, most spot sizes remained within this limit. However, deviations were noted, with the Y-axis spot size reaching ±0.43 mm and three gantry angles in the X-axis exceeding the ±0.3 mm threshold, violating the acceptance criteria. Figure 3 compares beam-matching parameters among GTR1, GTR2, GTR3, and SRD, showing spot size deviations as a function of energy and gantry angles. Biannual preventive maintenance (PM) of the cyclotron, performed per IBA protocols, can influence beam parameters. While proton range remains stable within 1 mm and relative room-specific deviations are limited to ±0.5 mm, the absolute dose can be recalibrated to baseline values. However, spot size is more susceptible to fluctuations, with deviations of up to ±0.3 mm from SRD specifications.

Figure 3.

The plot shows the spot size deviation as a function of gantry angle and between rooms: (a) X-axis spot size deviation between gantry treatment room and SRD, (b) Y-axis spot size deviation between gantry treatment room and SRD

Modeled spot sizes in RayStation treatment planning system

Spot profile data obtained from TOPAS simulations, representing spot sizes across different planes, were incorporated into the RayStation TPS to create a new machine model. The modeled spot sizes were plotted for various constant spot size values, as illustrated in Figure 4. The RayStation TPS modeled spot size value was observed well within 0.1 mm of the spot size of the reference model considered in the TOPAS simulation, as depicted in Figure 4.

Figure 4.

The modeled spot sizes with respect to the energy plot for different models in the RayStation treatment planning system

Evaluation of spot size effect using homogenous spread-out Bragg peak cube

The homogeneous field SOBP cube was evaluated under various spot and EL spacing conditions. Results indicate that TC percentages generally decrease as spot and EL spacing increase. Among the models, Model −0.6 mm consistently demonstrated the highest coverage across all spacing scenarios, while Model 0.6 mm exhibited the lowest.

A statistically significant difference (P = 0.01) was observed in TC when comparing models, particularly for models with higher spot sizes. However, no significant differences were detected for varying spot and EL spacing, as shown in Figure 5. The CI analysis reveals that models diverging further from the reference achieved higher CI values across all spacing conditions. Statistical analysis confirmed a significant difference (P < 0.05) between models for CI, whereas spot and EL spacing exhibited no significant impact on CI (P > 0.05). In addition, the HI showed no significant variations across models or spacing conditions.

Figure 5.

Comparison of spot size models with varying spot and energy layer spacing for target coverage (TC), conformity index (CI), and homogeneity index (HI) in cube data: (a) TC (V98%), (b) HI (98%), (c) CI (98%), and (d) CI (95%)

Assessment of spot size error in clinical cases

TC (V98%) was evaluated across the pelvis, brain, and prostate cases under nominal and robust worst-case scenarios. In the Nominal case, the V98% coverage consistently reached 100% for all models, while robust worst-case coverage varied. Models with reduced spot size (Model −0.6 mm) showed improved worst-case coverage compared to higher spot size (Model 0.6 mm). Statistical analysis found no significant differences in TC (V95%) between models (P > 0.05). Brain cases followed similar trends in TC, CI, and HI, as shown in Figure 6, while dose distributions and dose differences were visualized for a representative case from each site in Figure 7.

Figure 6.

Model-to-model variation in target coverage (TC), conformity index (CI), and homogeneity index (HI) for pelvis, brain, and prostate cases: (a-c) show TC (V98%) for nominal and robust worst-case scenarios; (d-f) show the CI (98%) for nominal and robust worst-case scenarios; and (g-i) show the HI (98%) for nominal and robust worst-case scenarios

Figure 7.

Impact of spot size variation (±0.6 mm) on dose distribution and dose-volume histogram (DVH) across different anatomical sites (pelvis, brain, and prostate), relative to the Model reference. (a) Dose distribution in color wash for the Model − 0.6 mm, (b) dose distribution for the Model reference, (c) dose distribution for the Model + 0.6 mm, (d) dose difference between the Model − 0.6 mm and the Model reference, (e) DVH comparison of the − 0.6 mm, reference, and Model 0.6 mm, and (f) dose difference between the Model 0.6 mm and the Model reference

Increasing model margins (−0.6 mm to + 0.6 mm) reduced robust worst-case coverage across all cases. Nominal coverage remained stable in the pelvis cases, but robust worst-case coverage declined significantly with larger margins. The brain cases showed the highest sensitivity, with coverage dropping sharply and nearing zero at larger margins. Prostate cases maintained nominal coverage but showed significant robust worst-case reductions, highlighting the impact of margins on robustness, particularly for brain and prostate cases.

The dose distribution and DVH analysis [Figure 7] further supported these observations. The color wash dose distributions illustrate that the Model −0.6 mm resulted in a sharper dose gradient, leading to improved target conformity. The Model +0.6 mm, however, exhibited increased dose spread, potentially affecting normal tissue sparing. The dose difference maps (D) and (F) reveal localized deviations in dose deposition when compared to the reference model, with higher dose differences seen in the prostate case, indicating its higher sensitivity to spot size changes. The difference is observed <3% in pelvis and brain cases whereas the difference is observed more than 3% in prostate. The DVH comparisons (E) show that increasing the spot size leads to a shift in dose coverage, reinforcing the trend observed in TC and CI results.

Conformity and homogeneity index

The CI improved with increasing margins across all cases. Pelvis cases remained stable, the brain showed significant gains with larger margins, and prostate cases improved slightly but remained lower overall. The HI remained stable for nominal values across all margins but declined slightly for robust worst-case values. Prostate cases were most sensitive to larger margins, showing a more pronounced reduction. Nominal homogeneity was less affected by margins, whereas robust worst-case homogeneity declined gradually.

Gamma evaluation result

The gamma analysis of 2D dose distributions for pelvis, brain, and prostate cases was performed across six planes for each patient. The results reveal that the Model 0.3 mm consistently achieved the highest performance, with more than 98% gamma passing rates for both 2%/2 mm and 3%/2 mm criteria and minimal standard deviation, as summarized in Table 2. Conversely, the Model −0.6 mm exhibited the lowest mean passing rates for the stricter 1%/1 mm criterion, raising concerns about its reliability. While all models displayed generally consistent performance, the Model 0.3 mm emerged as the most reliable choice for accurate dose assessments

Table 2.

Model-to-model comparison of gamma passing results for pelvis, brain, and prostate cases

Patient group	Spot size model	1%/1 mm (mean±SD)	2%/2 mm (mean±SD)	3%/2 mm (mean±SD)
Pelvis	Model −0.6 mm	98.60±1.68	99.93±0.23	99.89±0.31
	Model −0.3	99.18±1.25	99.97±0.12	99.99±0.03
	Model 0.3	99.88±0.22	100.00±0.00	100.00±0.00
	Mode l 0.6 mm	99.08±1.38	100.00±0.00	100.00±0.00
Brain	Model −0.6 mm	94.62±2.15	99.45±0.62	99.91±0.12
	Model −0.3	98.59±1.32	99.98±0.09	99.99±0.03
	Mode l 0.3	98.71±0.91	99.89±0.15	99.98±0.05
	Model 0.6 mm	92.74±4.52	99.28±1.29	99.94±0.12
Prostate	Model −0.6 mm	72.01±8.62	93.44±4.62	96.28±2.93
	Model −0.3	95.69±3.28	99.89±0.26	100.00±0.00
	Model 0.3	92.80±3.61	99.33±0.86	99.88±0.37
	Model 0.6 mm	89.36±4.15	98.27±1.36	99.73±0.59

SD: Standard deviation

Tolerance levels

Higher tolerance criteria (2%/2 mm and 3%/2 mm) yielded consistently higher gamma passing rates across all patient groups, demonstrating robustness to model variations.

Anatomical sensitivity

Prostate cases exhibited the greatest sensitivity to spot size variations, particularly at the strictest 1%/1 mm criterion. However, increasing the tolerance reduced this sensitivity, with gamma passing rates approaching those of pelvis and brain cases.

Model performance

In pelvis and brain cases, gamma passing rates were largely unaffected by model variations within ± 0.6 mm, especially at tolerances of 2%/2 mm or higher. The brain group demonstrated high robustness to spot size changes, maintaining high passing rates even at stricter tolerances.

DISCUSSION

This study highlights the critical role of spot size variations in ensuring effective clinical outcomes, particularly for inter-room patient transfers in proton therapy. Our findings align with the tolerance recommendations from previous studies, which suggest maintaining spot size variations within ± 0.15 to ± 0.3 mm, approximately 5% of the spot sizes for 226.2 MeV and 70.18 MeV beams.[11] However, the spot size tolerance of 5% is very stringent for the engineers to the beam optics tuning and is only possible when the beam matching option is loaded in the contract.[8,9] The ±10% spot size tolerance proposed by AAPM TG 224 is significantly larger for higher spot sizes, potentially exceeding the clinical tolerance range of ±0.3 mm for critical scenarios.[13] A machine performance study by Shamurailatpam et al. reported dose output and range accuracy within 0.5% and 0.5 mm, respectively, while the relative spot position accuracy between off-center and central spots was ±0.6 mm. However, spot size variations along the x and y axes were observed to be ±0.82 mm.[10] For instance, at 70.18 MeV (6.5 mm Spot size), a ±15% variation results in a spot size range of 5.85–7.15 mm, which exceeds the ±0.3 mm tolerance. Similarly, for 226.2 MeV (2.8 mm spot size), the spot size range is 2.52–3.08 mm, again surpassing the recommended tolerance. These discrepancies emphasize the need to evaluate spot size variations more rigorously, especially within the ±0.3 mm to ± 0.6 mm range, to understand their dosimetric impact. Spot size variations are particularly critical during patient transfers between gantries due to differences in gantry-specific spot sizes, as shown in this study. Our institution observed a Y-axis spot size variation of up to +0.43 mm between two gantries, GTR1 and GTR3, which exceeded the variations relative to the SRD baseline. These differences often arise from periodic cyclotron maintenance (PM), during which engineers adjust the magnet settings to align the system with SRD reference limits. While the SRD baseline variations are typically within ±0.3 mm, inter-gantry differences can approach ±0.5 mm, underscoring the need for robust clinical evaluations to mitigate potential dosimetric inconsistencies.

Spot size variations in the treatment planning system model

This study investigated the effect of spot size variations of ±0.3 mm and ±0.6 mm within the RayStation TPS across 33 proton energy levels for various anatomical sites, including the pelvis, brain, and prostate. The analysis focused on dosimetric parameters such as TC, CI, and HI. Results indicated that spot size variations up to ±0.3 mm had an insignificant impact on these parameters. However, increasing the variation to ±0.6 mm led to a notable reduction in TC, particularly in prostate treatments, while having a minimal effect on conformity and homogeneity indices.

The decline in TC with increasing spot size is attributed to the expansion of the lateral penumbra, causing a greater dose spread beyond the target boundaries. Conversely, a smaller spot size results in penumbra contraction, improving dose localization.[15] This effect is inherently linked to the Gaussian distribution of the proton beam, where the lower tail significantly influences dose falloff and dose-volume indices.

Clinical impact and recommendations

The findings reveal that gamma passing rates remain consistently high for pelvis and brain treatments, even at strict tolerances of 1%/1 mm, and further improve with 2%/2 mm or higher tolerances. Prostate treatments, however, displayed greater sensitivity to spot size variations, with lower gamma passing rates at 1%/1 mm. Increasing the tolerance to 2%/2 mm or 3%/2 mm mitigated this sensitivity, achieving gamma passing rates comparable to those observed in pelvis and brain cases. The heightened sensitivity of prostate targets can be attributed to their deeper location and the smaller spot sizes required for high-energy beams. In prostate cases, the combination of high-energy beams (resulting in smaller spot sizes), sharp dose gradients, and proximity to critical structures magnifies the impact of spot size variations. As a result, even small spot size errors significantly affect dose distribution, leading to lower gamma passing rates at stricter tolerances. In contrast, superficial pelvis targets and mid-depth brain targets were less affected by spot size variations, demonstrating higher robustness.

In conclusion, spot size variations within ±0.3 mm are clinically acceptable for most anatomical sites, maintaining consistent gamma passing rates across all evaluated clinical targets. For variations extending up to ±0.6 mm, a tolerance level of 2%/2 mm or greater is recommended to preserve treatment accuracy while accommodating intergantry transfers and emergency scenarios. Model ±0.3 mm is identified as the most suitable option in cases requiring patient transfer between gantries, as it ensures that all gantry spot sizes remain within ±0.3 mm of the SRD baseline, particularly in facilities equipped with the IBA Proteus Plus system. To ensure optimal treatment outcomes, institutions should prioritize regular monitoring and adjustment of spot size parameters following maintenance activities to minimize inter-gantry variability.

CONCLUSION

This study investigated the influence of spot size variations on clinical outcomes in proton therapy, with a focus on patient transfers between gantries. The findings demonstrated that nominal TC remained stable across different spot size models, and variations up to ± 0.6 mm had minimal impact on dosimetric metrics such as conformity and homogeneity for most clinical sites, including pelvis and brain targets. However, prostate treatments exhibited greater sensitivity to spot size fluctuations, emphasizing the need for precise monitoring in such cases. This comprehensive analysis evaluated the effects of spot size variations across a range of clinical sites and energies, providing valuable insights for optimizing proton therapy delivery.

Conflicts of interest

There are no conflicts of interest.

Funding Statement

Nil.