Abstract

Background:

This study aimed to compare the dosimetric differences between the use of range shifter and bolus in spot-scanning proton arc (SPArc) therapy for postoperative radiotherapy after left-sided breast cancer modified radical mastectomy, with a primary focus on target volume dose coverage and organs-at-risk sparing. By comparison with VMAT, this study further explored the potential of SPArc to reduce the normal tissue complication probability (NTCP) and the effective dose to immune cells (EDIC).

Methods:

This retrospective study utilized the imaging datasets of 18 patients who treated with VMAT. Three SPArc plans were designed for each patient using the RayStation v2025: 1. SPArc with bolus (SPArcB); 2. SPArc with range shifter (SPArcR); 3. SPArc with both bolus and range shifter (SPArcRB). The target volumes included the supraclavicular target volume and the chest wall target volume. The Lyman–Kutcher–Birman NTCP model was adopted to evaluate pulmonary and cardiac toxicities and the EDIC model was applied to assess the immune cell dose across different plans. Robustness optimization accounting for a ± 3.5% range uncertainty and a 5 mm setup error was incorporated into all SPArc treatment plans.

Results:

All SPArc plans met robustness criteria and significantly reduced doses to the heart and left lung, as well as EDIC, compared to VMAT. The SPArcR plan demonstrated superior cardiac protection, achieving the lowest mean heart dose and the most significant reduction in NTCP for both the heart and lungs. While SPArcB showed a slight advantage in reducing the mean left lung dose, SPArcR yielded the highest conformity index for target coverage.

Conclusion:

This study confirms that, in postoperative radiotherapy for left-sided breast cancer after modified radical mastectomy, the standalone application of a range shifter can optimally reduce cardiac dose and the associated risk of complications while ensuring adequate target volume coverage, and it circumvents the issues of setup variability and hygiene concerns associated with the clinical use of bolus.

1 Introduction

According to the latest data from the World Health Organization (WHO), over 2.3 million women worldwide were diagnosed with breast cancer in 2022, with 670,000 dying from the disease. In 157 out of 185 countries, breast cancer is the most commonly diagnosed cancer among women. It is projected that by 2040, the annual number of new breast cancer cases globally will exceed 3 million (1). In the comprehensive management of breast cancer, radiotherapy is an indispensable component of post-operative treatment. It aims to maximise local disease control, minimise the risk of locoregional recurrence, and improve patients’ overall survival rates (2). However, for left-sided breast cancer, due to its anatomical proximity to the heart, there is an elevated risk of radio-induced cardiac injury and fatal cardiovascular events. Studies indicate that for every 1 Gy increase in the mean heart dose, the risk of major coronary events rises by 7.4% (3). Furthermore, radiotherapy may also induce complications such as radiation pneumonitis and radiation myelopathy, which can severely compromise the patient’s quality of life (4, 5). How to ensure adequate dose coverage to the target volume while minimising radiation dose to adjacent organs at risk, such as the heart and lungs, remains the central challenge in clinical practice.

Compared to photon radiotherapy, proton therapy offers a potential solution to achieve the aforementioned goal, leveraging its unique physical property known as the Bragg peak. As the proton beam deposits its maximum energy at a specific depth, its dose undergoes a rapid fall-off thereafter, thereby allowing for a significant reduction in the radiation dose to organs at risk such as the heart and lungs. The study by Tommasino et al. (6) demonstrated that, while achieving equivalent target coverage, intensity-modulated proton therapy (IMPT) significantly reduced the radiation dose to the heart and lungs compared to intensity-modulated radiotherapy (IMRT). Moreover, the predicted risks of skin toxicity and cardiopulmonary diseases were also lower. The spot-scanning proton arc (SPArc) therapy, first proposed by Ding et al. (7), is a novel, robust, and efficient optimization algorithm for proton therapy. In recent years, multiple studies have demonstrated that in the treatment of tumors such as oropharyngeal cancer (8), liver cancer (9), and esophageal cancer (10), SPArc exhibits significant dosimetric advantages compared to techniques like IMPT and volumetric modulated arc therapy (VMAT), and shows potential for further reducing normal tissue toxicity. The study by Chang et al. on left-sided breast cancer whole breast radiotherapy demonstrated that, compared to IMPT, SPArc can further reduce the radiation dose to healthy tissues and lower the normal tissue complication probability (NTCP) (11). These findings confirm the feasibility and potential clinical value of SPArc in breast cancer radiotherapy.

In proton therapy, to generate a proton beam suitable for treating superficial targets, a range shifter (RS) is typically employed to degrade the beam energy. An RS is a homogeneous plate, fixed to the snout at a certain distance from the patient’s body, commonly made of materials such as acrylonitrile butadiene styrene or polyethylene (12). For VMAT, the application of a bolus over the chest wall target is commonly employed to improve superficial dose uniformity. This ensures adequate dose delivery to the shallow layer of the chest wall target following modified radical mastectomy for breast cancer, thereby aiding in the prevention of postoperative recurrence (13). However, as a novel proton therapy optimization algorithm, whether SPArc requires the use of a bolus to modify the superficial dose distribution for the chest wall target following left-sided modified radical mastectomy, and whether employing a bolus or an RS can achieve a superior balance between target dose coverage and optimized distal organ sparing, remains a subject that has not been thoroughly investigated.

This study aims to systematically evaluate the dosimetric differences between using a bolus or an RS in SPArc for post-operative radiotherapy following left-sided modified radical mastectomy for breast cancer. These SPArc plans will be compared with clinically implemented VMAT plans, with a focus on target dose coverage, organs at risk (OARs) sparing, and specific dosimetric parameters. Furthermore, the study will investigate the potential of different planning strategies to reduce the NTCP and the effective dose to immune cells (EDIC). By elucidating the impact of these technical parameters on dose distribution, this research will provide a theoretical basis for the individualized clinical selection and protocol optimization of SPArc technology.

2 Materials and methods2.1 Selection of retrospective patient data

This study protocol was approved by the Institutional Review Board of West China Hospital, Sichuan University (Approval No.: 2025227). All procedures were performed in accordance with local laws and institutional requirements. Written informed consent was obtained from all participants. The datasets generated and analyzed during this study are not publicly available due to privacy restrictions, but are available from the corresponding author upon reasonable request. This retrospective study consecutively included imaging datasets of patients who underwent radiotherapy following modified radical mastectomy for left-sided breast cancer at our institution between 2024 and 2025. Eighteen patients were ultimately enrolled, with no eligible patients excluded, thereby minimizing selection bias. The inclusion criteria were as follows: 1. Pathologically confirmed left-sided breast cancer and receipt of VMAT; 2. High risk of skin recurrence assessed by physicians, requiring the use of a bolus with consistent coverage of the chest wall target volume as prescribed during treatment; 3. Target volume encompassing both the chest wall and supraclavicular fossa; 4. Patient immobilization using a breast vacuum cushion, with treatment delivered under free breathing at a prescribed dose of 5,000 cGy in 25 fractions. Detailed characteristics of the patients and tumours are presented in Table 1.

CharacteristicValuePatients18Age at RT (years)55 (42–70)Target volume (cc)CTVsc120.98 (66.26–426.84)CTVcw405.85 (123.87–719.55)LocationCentral breast8 (44.44%)Upper inner quadrant2 (11.11%)Upper outer quadrant4 (22.22%)Lower outer quadrant4 (22.22%)T stageT15 (27.78%)T28 (44.44%)T33 (16.67%)T42 (11.11%)N stageN04 (22.22%)N112 (66.67%)N21 (5.56%)N31 (5.56%)M stageM017 (94.44%)M11 (5.56%)LateralityLeft18

Patient and tumor characteristics.

The datas of age at RT and target volume are presented as the median and range of all patients.

2.2 Patient setup, description of target volumes and OARs

All patients underwent patient immobilization using a vacuum cushion equipped with an oblique plate, with a bolus (Shenzhen Tongchuang Medical Technology Co., Ltd.) applied outside the chest wall target volume. The bolus coverage extended from the inferior border of the left clavicle to the inferior margin of the breast, medially to the anterior midline, and laterally to the midaxillary line. Target volumes and OARs were contoured slice by slice by physicians on simulation CT images. The clinical target volume (CTV) was defined as the supraclavicular lymph node CTV (CTVsc) and chest wall CTV (CTVcw), including residual glandular tissue, adjacent adipose tissue, surgical scars, skin, subcutaneous tissue, and intercostal muscles. The planning target volumes (PTVsc and PTVcw) for the VMAT plan were generated by expanding CTVsc and CTVcw by 0.5 cm, respectively. All PTVs and CTVs included the skin surface. The contoured OARs included the left lung, right lung, heart, left anterior descending (LAD) artery, spinal cord, contralateral breast, left humeral head, and thyroid gland. The OARs constraints for plan optimization are presented in Table 2.

StructureParameterOptimal criteriaPriorityHeartMean dose (cGy)<800 cGyHighV500 cGy (%)<40%HighLADMean dose (cGy)<2,500 cGyHighV4000 cGy (%)<20%HighLung_LMean dose (cGy)<1,400 cGyHighV500 cGy (%)<50%HighV1000 cGy (%)<35%HighV2000 cGy (%)<25%HighSpinal_CordMax dose (cGy)<2,500 cGyHighThyroidMean dose (cGy)<2,800 cGyMiddleLung_RV500 cGy (%)<20%MiddleBreast_RMean dose (cGy)<400 cGyLowHumerus_Head_LMax dose (cGy)<5,000 cGyLowV3000 cGy (%)<20%Low

Dose constraints and planning priority for organs at risk.

V500 cGy, V1000 cGy, V2000 cGy, V3000 cGy, and V4000 cGy refer to the percentage of the volume of the respective organs that receives at least the corresponding dose, respectively.

2.3 Dose prescription, treatment planning, and robust optimization

The prescribed dose to the target volume was 5,000 cGy in 25 fractions. The treatment plan was designed to ensure that at least 98% of the CTV receives 100% of the prescribed dose, while the maximum dose within the target volume does not exceed 107% of the prescribed dose. The VMAT treatment plan was optimised based on the PTV, whereas the SPArc treatment plan was optimised directly on the CTV. A constant relative biological effectiveness (RBE) value of 1.1 was adopted in the proton therapy plan (14). To minimise potential biases from different treatment planning systems (TPS), all treatment plans were calculated and generated using RayStation v2025 (RaySearch Laboratories AB, Stockholm, Sweden). For each patient, four radiotherapy plans were designed using SPArc and VMAT techniques, respectively (Table 3). The bolus was modeled as a water-equivalent material (density = 1.0 g/cm3) with a physical thickness of 0.5 cm, corresponding to a water-equivalent thickness of 0.5 cm. This thickness increases superficial dose to ensure adequate target coverage, consistent with routine clinical practice. In the no-bolus group, the corresponding region was assigned as air to simulate the clinical scenario without the use of a bolus. The RS was modeled as polymethyl methacrylate (PMMA) with a density of 1.19 g/cm3, a physical thickness of 3.65 cm, and a field size of 30 × 40 cm2. This thickness shifts the proton beam range forward to the required depth for adequate superficial target coverage. PMMA was selected as the RS material due to its near water-equivalent stopping power and well-characterized scattering properties, which help ensure dose calculation accuracy in the treatment planning system. Although PMMA introduces slightly greater scattering compared with materials such as paraffin or polyethylene, this effect is clinically acceptable (15). In addition, PMMA has stable density, high mechanical strength, and resistance to deformation, making it suitable for clinical use, and it is the default range shifter material for the IBA ProteusPlus proton therapy system.

GroupsBolus materialsRange shifterAbbreviationVMAT with BolusWaterNOVMATSPArc with Bolus without RSWaterNOSPArcBSPArc without Bolus with RSAirYESSPArcRSPArc with Bolus with RSWaterYESSPArcBR

Radiation therapy plan setup.

Bolus materials and RS are both set up in the RayStation v2025. VMAT with bolus is abbreviated as VMAT; SPArc with bolus without RS is abbreviated as SPArcB; SPArc without bolus with RS is abbreviated as SPArcR; SPArc with bolus with RS is abbreviated as SPArcBR.

For the SPArc plan, a single arc discretized into 20 directions was used. Each plan consisted of 360 energy layers, and the spot positions were automatically determined by the treatment planning system based on the patient’s individual anatomy. Delivery constraints were applied, with the minimum and maximum monitor units per spot set to 0.01 MU/fraction and 15 MU/fraction, respectively. The SPArc plans were optimized using the Monte Carlo algorithm, with the proton beam model based on the IBA ProteusPlus system. The proton beam energy ranged from 70 to 227 MeV, with spot sizes ranging from 2.73 mm (at 227 MeV) to 7.0 mm (at 70 MeV). For the VMAT plan, dual arcs were generated using 6 MV photon beams. The gantry angle spacing was set to 2°, and the maximum delivery time was limited to 120 s. The collimator angle was set to 0° for all plans. All plans utilized a single-isocenter tangential field arrangement.

All SPArc plans in this study underwent robustness optimization accounting for ± 3.5% range uncertainty and ± 5 mm setup uncertainty in six directions (anterior–posterior, posterior–anterior, left–right, right–left, superior–inferior, and inferior–superior). A total of 21 uncertainty scenarios, including nominal and perturbed conditions, were evaluated using a scenario-based worst-case analysis approach. For robustness evaluation, target coverage was quantified using D95% for both CTVcw and CTVsc across all uncertainty scenarios. A plan was deemed robust if, in at least 80% of the scenarios, the D95% for each target volume was no less than 95% of the prescribed dose. In addition, worst-case scenario analysis was performed to ensure that no clinically unacceptable target underdosage occurred under extreme perturbations (16).

2.4 Dose–Volume Histogram analysis

Dosimetric parameters of CTVsc and CTVcw were obtained via Dose–Volume Histogram (DVH) analysis to assess and compare various radiotherapy plans. In accordance with ICRU Report 83, dosimetric parameters including D98%, conformity index (CI), and homogeneity index (HI) were adopted in this study to evaluate target volume coverage. The calculation formulas for CI and HI are as follows (Equations 1,2):

Where Vt,ref is the volume of the target volume covered by the prescribed dose, Vt is the target volume, and Vref is the volume of the reference isodose (5,000 cGy) within the target volume. D2%, D50%, and D98% are the minimum absorbed doses received by 2, 50, and 98% of the target volume, respectively. A higher CI indicates better dose conformity to the target volume, with an ideal value of 1; whereas a lower HI denotes a more uniform radiation dose distribution within the target, with an optimal value of 0 (17). To evaluate the protection of OARs, dosimetric parameters including mean dose, max dose, V500 cGy, and V2000 cGy were recorded. V500 cGy and V2000 cGy refer to the percentage of the volume of the respective organs that receives at least 500 cGy and 2000 cGy, respectively. Specifically: the heart was primarily evaluated by mean dose and V500 cGy; the left lung by mean dose, V500 cGy, and V2000 cGy; the right lung by mean dose and V500 cGy; the spinal cord by max dose as the core endpoint; the LAD by mean dose; the contralateral breast by mean dose; the left humeral head by max dose; and the thyroid gland by mean dose.

2.5 Evaluation of potential clinical benefit for OARs based on the NCTP model and hematological toxicity based on the EDIC model

The SPArc and VMAT plans developed for each patient will be evaluated for potential clinical benefits using the NTCP model. The Lyman–Kutcher–Birman NTCP model was adopted to calculate pulmonary and cardiac toxicity of different plans based on the mean dose. The definition of this model is as follows (Equations 3,4):

NTCP is determined by two parameters: TD₅₀ (tolerance dose for 50% complication), defined as the dose that results in a 50% complication rate when the entire organ is uniformly irradiated; and m, the slope parameter of the sigmoid dose–response curve, where a smaller value indicates a steeper curve. For assessing the risk of radiation pneumonitis, the equivalent uniform dose (EUD) was set as the mean lung dose, with TD₅₀ = 30.8 Gy and m = 0.37, based on the NTCP model proposed by Seppenwoolde et al. (18). For evaluating the risk of any grade of pericardial effusion, the EUD was adopted as the mean heart dose, with TD₅₀ = 34.3 Gy and m = 0.75, in accordance with the NTCP model developed by Fukada et al. (19). To compare the risk differences between each plan and the VMAT plan, the NTCP ratio was used, calculated as follows: NTCP ratio = NTCPSPArc/NTCPVMAT. This ratio was computed separately for SPArcB, SPArcR, and SPArcRB plans.

We adopted the EDIC model proposed by Jin et al. (20) to evaluate the dose to immune cells across different treatment plans. This model quantifies the cumulative radiation dose received by systemically circulating lymphocytes during radiotherapy. A study by Chen et al. has confirmed that the EDIC is an important factor influencing radiation-induced lymphopenia (RIL) in breast cancer radiotherapy, and can be used for individualised risk prediction via NTCP modellin (21). The definition of this model is as follows (Equation 5):

Where MLD refers to the mean lung dose; MHD denotes the mean heart dose; n represents the number of radiotherapy fractions; and ITDV stands for the integral total dose volume, defined as the total integral dose within the planning CT images divided by the average volume to obtain the mean dose.

2.6 Statistics

Statistical analyses were performed using IBM SPSS 27.0 (IBM Corp., Armonk, NY, USA). First, the Friedman test was employed to compare the overall differences among the four radiotherapy plans, with a p-value <0.05 indicating statistically significant differences. Where a significant overall difference was identified, pairwise comparisons were conducted using the Wilcoxon signed-rank test, supplemented by Bonferroni multiple testing correction. The corrected significance level was set at α’ = 0.00833.

3 Results

SPArc was used to design radiotherapy plans for 18 patients undergoing post-radical mastectomy radiotherapy for left-sided breast cancer, resulting in the generation and analysis of 72 treatment plans. The SPArcB plan (with the minimum air gap) had a mean air gap of 31.18 cm, whereas the SPArcR plan had a mean air gap of 27.74 cm, representing an average reduction in air gap of 11.03% (95% confidence interval: 10.03–12.03%). The SPArcRB plan had a mean air gap of 27.68 cm, corresponding to an average reduction in air gap of 11.01% (95% confidence interval: 10.07–11.96%).

3.1 Robustness analysis of different SPArc plans

Table 4 summarizes the robustness evaluation results of target coverage for CTVcw and CTVsc under all uncertainty scenarios. All three SPArc plans (SPArcB, SPArcR, and SPArcRB) satisfied the robustness evaluation criteria under 3.5% range uncertainty and 5 mm setup uncertainty, defined such that in at least 80% of the scenarios, the D95% for each target volume was no less than 95% of the prescribed dose. For CTVcw, the mean D95% values were comparable across plans, ranging from 5022.1 ± 47.6 cGy (SPArcB) to 5052.2 ± 31.7 cGy (SPArcRB), with SPArcR showing an intermediate value of 5038.8 ± 24.5 cGy. The worst-scenario D95% values followed a similar pattern: SPArcB had the lowest mean worst-scenario value (4954.9 ± 57.7 cGy), while SPArcR and SPArcRB achieved slightly higher values (4988.6 ± 32.0 cGy and 4982.5 ± 32.3 cGy, respectively). For CTVsc, SPArcR demonstrated the most favorable robustness, with the highest mean D95% (5039.1 ± 15.4 cGy) and the highest worst-scenario D95% (4994.8 ± 22.5 cGy). In contrast, SPArcB exhibited the lowest mean D95% (4989.9 ± 39.4 cGy) and notably the lowest worst-scenario D95% (4867.6 ± 93.1 cGy), accompanied by the largest standard deviation, indicating greater sensitivity to uncertainties. SPArcRB showed intermediate robustness for CTVcw. Overall, while all plans remained within clinically acceptable limits, SPArcR provided the most robust target coverage, particularly for the CTVcw structure.

PlanStructureMean D95% (cGy)Worst scenario D95% (cGy)SPArcBCTVcw5022.1 ± 47.64954.9 ± 57.7CTVsc4989.9 ± 39.44867.6 ± 93.1SPArcRCTVcw5038.8 ± 24.54988.6 ± 32.0CTVsc5039.1 ± 15.44994.8 ± 22.5SPArcRBCTVcw5052.2 ± 31.74982.5 ± 32.3CTVsc5026.4 ± 16.34965.6 ± 24.1

Robustness analysis of target coverage (D95%) for CTVcw and CTVsc across all uncertainty scenarios.

Values are presented as mean ± standard deviation (SD). “Mean D95%” is the overall mean of each patient’s average D95% across scenarios; “Worst scenario D95%” is the overall mean of each patient’s minimum D95% across scenarios. CTVsc: supraclavicular lymph node CTV; CTVcw: chest wall CTV.

3.2 Comparison of target volume dosimetry

The study demonstrated that different radiotherapy techniques (SPArc vs. VMAT) and parameter configurations (with or without bolus or RS) exerted a significant impact on the dose distribution of target volumes (p < 0.05). Figure 1 displays representative CT slices of one patient, presenting a comparison of the dose distributions of the CTVcw across the four radiotherapy plans. Among these plans, the SPArc plans exhibited markedly smaller low-dose regions (500 cGy, 1,000 cGy) than the VMAT plan, and the prescribed dose (5,000 cGy) was more conformal compared with that of the VMAT plan. According to Table 5, pairwise comparisons of the CI values of the four radiotherapy plans were performed using the Wilcoxon signed-rank test for the CTV as a whole. The results indicated that statistically significant differences existed between all pairs of plans. The CI values were ranked in descending order as follows: SPArcR > SPArcBR > SPArcB > VMAT. Specifically, SPArcR yielded the highest CI value, with the most significant difference observed relative to VMAT (p < 0.001). For the CTVsc, the D98% and D95% values of the VMAT plan were significantly higher than those of the SPArc plans, whereas no significant differences were observed among the SPArcB, SPArcR, and SPArcRB groups. The median D98% values of all treatment plans reached the prescribed dose. In terms of the HI, the values of the SPArcB, SPArcR, and SPArcRB plans were significantly higher than that of the VMAT plan, with the most significant difference detected between the SPArcRB and VMAT plans. The target volume homogeneity of the VMAT plan was slightly superior to that of the SPArc plans; however, the HI values of all groups remained at a low level overall. For the CTVcw, the D98% and D95% values of the VMAT plan were lower than those of the SPArc plans, with a statistically significant difference observed between the VMAT plan and the SPArcRB plan. Specifically, the SPArcRB plan delivered the highest D98% and D95% values for the chest wall target volume, whereas no significant differences were detected among the SPArcB, SPArcR, and SPArcRB groups. In terms of the HI, no significant differences were found in pairwise comparisons of all the plans. The SPArcB group exhibited relatively high HI values, while the HI values of the SPArcR and SPArcRB groups remained at a low level.

A representative of the radiation treatment plan. The comparison of patient dose distribution of CTVcw among four treatment plans (VMAT, SPArcB, SPArcR, SPArcRB).

StructureValueVMATSPArcBSPArcRSPArcRBpPairwise comparisonsMedian (IQR)Median (IQR)Median (IQR)Median (IQR)CTVCI0.574 (0.531–0.602)0.608 (0.589–0.651)0.718 (0.675–0.771)0.676 (0.648–0.729)<0.05a,b,c,d,e,fCTVscD98% (cGy)5065.72 (5053.52–5075.85)5028.31 (5014.05–5043.28)5023.45 (5007.67–5033.43)5021.58 (5018.12–5039.62)<0.05a,b,cD95% (cGy)5092.62 (5072.90–5101.43)5059.51 (5045.83–5079.71)5062.32 (5041.76–5076.53)5064.82 (5048.37–5074.98)<0.05a,b,cHI0.052 (0.047–0.055)0.059 (0.055–0.063)0.062 (0.057–0.065)0.063 (0.060–0.066)<0.05a,b,cCTVcwD98% (cGy)5022.94 (4987.54–5040.15)5041.83 (5027.43–5057.33)5031.00 (5003.57–5042.03)5060.08 (5035.27–5069.95)<0.05cD95% (cGy)5065.44 (5056.04–5074.30)5081.80 (5073.05–5093.83)5079.01 (5054.45–5085.12)5101.78 (5085.29–5111.58)<0.05cHI0.060 (0.054–0.066)0.061 (0.058–0.067)0.057 (0.054–0.062)0.055 (0.050–0.060)<0.05N/AHeartMean dose (cGy)426.04 (362.84–443.96)145.19 (76.55–171.29)80.17 (61.56–87.77)98.61 (59.68–122.11)<0.05a,b,c,d,eV500 cGy (%)14.32 (11.31–16.05)6.53 (3.44–8.67)3.89 (2.79–5.01)5.19 (2.19–7.14)<0.05a,b,c,dLung_LMean dose (cGy)1298.84 (1277.94–1345.74)874.28 (792.98–972.49)877.92 (829.09–913.57)910.41 (867.07–966.73)<0.05a,b,cV500 cGy (%)48.03 (46.93–50.32)31.82 (29.67–34.84)33.89 (32.14–35.40)35.15 (33.67–35.97)<0.05a,b,c,eV2000 cGy (%)23.45 (23.00–24.47)17.75 (15.73–20.22)18.38 (17.16–19.16)18.86 (17.98–20.42)<0.05a,b,c,eLung_RMean dose (cGy)197.96 (182.49–228.90)17.64 (10.27–47.64)20.29 (19.43–21.16)33.97 (18.37–45.81)<0.05a,b,c,fV500 cGy (%)4.80 (2.66–6.52)0.43 (0.11–1.05)0.19 (0.07–0.45)0.84 (0.14–1.40)<0.05a,b,c,fSpinal_CordMax dose (cGy)804.76 (748.82–970.68)98.39 (63.81–291.28)152.78 (90.70–206.27)328.18 (119.37–386.22)<0.05a,b,c,e,fLADMean dose (cGy)1701.11 (1547.73–1978.14)801.66 (529.25–995.39)618.76 (538.29–701.98)742.87 (486.20–850.49)<0.05a,b,c,eBreast_RMean dose (cGy)458.17 (288.12–473.83)131.83 (91.64–161.31)129.70 (101.22–168.51)156.77 (124.82–184.13)<0.05a,b,c,fHumerus_Head_LMax dose (cGy)4125.32 (3656.58–4412.14)1798.56 (1516.40–2679.15)2452.30 (1983.75–2777.64)2397.37 (2028.59–3033.52)<0.05a,b,c,d,eThyroidMean dose (cGy)2820.58 (2638.48–2941.30)2092.50 (1983.77–2148.13)2053.27 (2026.26–2174.85)2275.15 (2145.27–2400.19)<0.05a,b,c,e,f

The dosimetric parameters and statistical results of target volumes and OARs.

Data were presented as median values with interquartile ranges (IQR, Q1–Q3). The p-value is the result of the Friedman test between the four radiotherapy plans. p < 0.05 means there is statistically significant difference among the groups. “Pairwise comparisons” in the table use the Wilcoxon signed-rank test with Bonferroni multiple testing correction, and the corrected significance level was set at α’ = 0.00833. Pairwise comparison groups represented by the different letters. a: VMAT vs SPArcB;b: VMAT vs SPArcR;c: VMAT vs SPArcRB;d: SPArcB vs SPArcR;e: SPArcB vs SPArcRB; f: SPArcR vs SPArcRB.

3.3 Dosimetric analysis of organs at risk