Dosimetric Comparison between Two Dose Calculation Algorithms in SBRT Treatment of Lung Cancer in Ring-based and C-arm Radiation Therapy Equipment

Patient Selection

The study recruited 20 patients aged 18 or above with stage I-II primary lung tumours or metastases located peripherally or centrally in the middle or upper lobe treated with SBRT at KWH. Data, including planning CT images and contouring of targets and OARs, were imported to the Varian Eclipse TPS version 17.01. Patients were categorized according to tumour location, with treatment plans optimized based on RTOG-0813 and RTOG-0915 protocols. Table 1 summarized the collected patient and tumour characteristics, and the treatment plan conditions for 20 patients included in the study.

Data Collection

Twenty patients recruited in this study retrospectively, all received SBRT treatment for lung tumours in KWH between December 2020 and November 2023. The anonymized data, including the planning CT images and the associated contouring of the targets and OARs were imported to the TPS. After importing the data, the samples were categorized according to the tumour location. There were 12 cases with peripherally located tumours that were eligible to apply the dose prescription and dose constraints specified in RTOG- 0915 protocol. For the remaining 8 cases with centrally located tumours, RTOG-0813 protocol was applied. The dose prescription for an individual patient was based on the respective RTOG protocol according to the tumour location.

Structures and Target Contouring

Target and OARs structures delineation were done by oncologists in KWH. An extra 0.5 cm in the axial plane and 1 cm in the craniocaudal plane will be added to the GTV to form the PTV, aligning with the RTOG-0813 and 0915 protocols. All OARs contours including spinal cord, esophagus, brachial plexus, heart, trachea and proximal bronchial tree, proximal trachea, whole lung, great vessels, and skin were checked for any discrepancy and were revised according to the contouring guidelines in RTOG-0915 and RTOG-0813 protocols if necessary.

RT Equipment and Dose Calculation Algorithms

Two types of RT equipment (TrueBeam and Halcyon) and two different dose calculation algorithms (AAA and AXB) were included in this study. Varian Eclipse TPS version 17.01 at was used for treatment planning. The photon optimizer (PO) algorithm version 17.01 was used for dose optimization in all cases. Acuros XB-13.5 physical material table was chosen for AXB dose calculation. Four VMAT plans were computed for each patient, namely, TrueBeam-AAA (TB-AAA), TrueBeam-AXB (TB-AXB), Halcyon-AAA (Hal- AAA), and Halcyon-AXB (Hal-AXB).

Truebeam VMAT Planning Technique

For TB-AAA and TB-AXB plans, 6MV FFF mode was implemented with Millennium 120 MLC and a maximum dose rate of 1400 MU/min. After careful evaluation of treatment delivery efficiency and the achievement of desired target coverage, a 2-half arc coplanar field arrangement was implemented in all SBRT plans. The arc angles were fixed at 350° to 179° clockwise (CW) and 179° to 350° counterclockwise (CCW) for left lung cases, and at 181° to 10° CW and 10° to 181°CCW for right lung cases to optimize target coverage while sparing normal lung tissues, particularly the contralateral lung. Collimator angles of 30° and 330° were chosen to minimize overlapping interleaf leakage caused by the tongue-and-groove effect. Isocenters for each field were initially positioned at the PTV mass center, then the X and Y coordinates were rounded off to the nearest 1 decimal place and the Z coordinate was rounded off to the nearest multiple of 0.25 based on the CT scan slice thickness (2.5mm). The field sizes were set with approximately 5mm margins around the PTV.

In plan optimization, two standard templates were constructed for all treatment plans. With reference to RTOG- 0813 and RTOG-0915 protocols, all the treatment plans were optimized using the dose-volume constraints of the target volume and OARs set in the standard templates. The normal tissue objective (NTO) priority was set to 100. During optimization, fine adjustments of the planning objective priorities and construction of additional pseudo-structures were used on a case-by-case basis. After the generation of treatment plans with TB-AAA, plans were re-optimized using TB-AXB. The optimization criteria were identical for both TB and HAL plans, ensuring consistency in treatment planning and enabling a direct comparison between the two equipment and dose calculation algorithms.

Halcyon VMAT planning technique

For comparison, TB-AAA treatment plans were re- optimized in Eclipse TPS using Hal-AAA followed by Hal-AXB. 6MV FFF mode was implemented with dual-layered stacked- and-staggered MLC and a maximum dose rate of 800 MU/ min. Identical arc geometry, collimator rotation and isocenter placement were used as in TrueBeam VMAT plans. In addition, the procedures of plan optimization and the settings were all identical to TrueBeam plans as described above.

Dose Prescription and Dose constraints of OARs

All the treatment plans fulfilled the prescription dose constraints as per RTOG-0915 and RTOG-0813 requirements.

The prescription dose for RTOG 0813 was 50 Gy delivered in 5 fractions, while for RTOG 0915, it was 48 Gy delivered in 4 fractions as listed in Table 1. The plans in all samples were normalized to achieve 98% PTV volume receiving 100% of the prescription dose (PTV V98% = 100%) and maximum PTV dose within 125% of the prescription dose. The rationale for limiting the maximum dose to 125% in this study was to ensure the prescribed dose was delivered effectively to the PTV, maintain dose homogeneity, and help achieve the D2cm dose constraint of RTOG-0813 and 0915. The planning objectives were used with reference to the Advisory Committee on Radiation Oncology Practice (ACROP) Guideline established by faculty members of the European Society for Radiotherapy and Oncology (ESTRO) and previous literature [14]. In addition, 99% of PTV volume received at least 90% of the prescription dose (PTV V99% > 90%) in all plans [13, 15].

Note: RUL – right upper lobe; RML – right middle lobe; LUL – left upper lobe; LML – left middle lobe. Table 1: List of tumour size, locations, prescription dose and protocol of the selected 20 patients.

All plans either met RTOG-0813 or RTOG-0915 protocol criteria normal tissue dose constraints based on the tumour locations. OARs included spinal cord, skin, lung, esophagus, heart, great vessels, ribs, and trachea and ipsilateral bronchus. The details of RTOG-0813 and RTOG-0915 protocol criteria were listed in Table 2.

Note: No dose constraints and planning objectives for Ribs in RTOG-0813; Sourced from RTOG-0813 & RTOG-0915 [13, 15]. Table 2: Dose constraints and planning objectives of RTOG-0813 & 0915.

Dosimetric Evaluation and Statistical Analysis

Several dosimetric parameters were generated for evaluation and comparison between 4 plans. RTOG-0915 and RTOG-0813 protocols were employed to conduct a comparative analysis of the TrueBeam and Halcyon VMAT plans with the utilization of AAA and AXB dose calculation algorithms for lung SBRT treatment planning. The primary objectives of this study evaluated several key parameters derived from the dose-volume histograms (DVHs) of the treatment plans generated.

For evaluating the high and intermediate dose spillage parameters of the PTV, Conformity index (CI), Homogeneity index (HI), Maximum dose (Dmax), R50%, and D2cm, the parameters were generated and defined as:

2 PIV TV CI TV PIV = ×

D D HI D − =

2% 98%

50% Where TVPIV, TV and PIV represent the volume of the target covered by the prescription isodose the target volume and the prescription isodose volume respectively. The ideal value for CI is 1 and the clinical desirability. The ideal value of HI is 0. D2% and D50% were evaluated separately. R50% which quantifies the ratio of the 50% prescription isodose volume to the PTV volume. The acceptable range of R50% depends on the variations in PTV size, and the acceptable range of R50% based on the RTOG-0813 and 0915 protocols. D2cm refers to the maximum dose at any point 2 cm away from the PTV margin in any direction. Compliance with the criterion of D2cm depends on the specific dimensions of the PTV and based on the RTOG-0813 and 0915 protocols. For evaluating the treatment planning and delivery efficiency, total number of monitor units (MU) per fraction and dose calculation time were recorded. Also, estimated BOT is simply defined as:

BOT Total MU DoseRate =

While the dose rate is modulated throughout the treatment to meet the requirements of the treatment plan, each machine typically operates at the highest achievable dose rate within the limits of its technical specifications and clinical constraints. Therefore, maximum dose rates of TrueBeam (1400 MU/min) and Halcyon (800 MU/min) will be used for analysis. As two protocols were used in our project, all dosimetric parameters were expressed as ratios relative to the prescription dose to normalize for variation in the prescription dose across individual plans. All plans either met RTOG-0815 or 0915 criteria normal tissue dose constraints. Also, due to variations in prescription dose between two RTOG protocols, individual plan values for any parameter were expressed as a ratio relative to the prescription dose and * was used to indicate the ratio of the dose, where Ratio* Dosimtric parameter Prescriptiondose =

Statistical Analysis

All the collected data were subjected to statistical analysis using the Statistical Package for the Social Sciences (SPSS) version 26.0 software (IBM, Armonk, NY). In terms of the aforementioned dosimetric parameters, a two-way ANOVA Test was performed to assess and compare the significance of differences between the various independent variables: AAA vs AXB, Halcyon vs TrueBeam, and the combination of both calculation algorithms and both treatment equipment, namely, Hal-AAA, Hal-AXB, TB-AAA, and TB-AXB as mentioned above. A p-value less than 0.05 was considered statistically significant among the dosimetric outcomes.The primary endpoints were to gauge and compare key dosimetric parameters such as the CI, HI, Dmax, R50%, and D2cm between both calculation algorithm (AAA and AXB) in TrueBeam and Halcyon by using dose-volume histograms (DVHs) derived from treatment plans. Secondary endpoints included evaluation of treatment planning and delivery efficiency by comparing the MUs, estimated BOT, and dose calculation time between AAA and AXB in both TrueBeam and Halcyon for SBRT lung cancer treatment.

AAA AXB p-value TrueBeam Halcyon p-value TrueBeam Halcyon p-value AAA AXB AAA AXB CI 0.909 ± 0.003 0.888 ± 0.004 <0.001 0.905 ± 0.003 0.892 ± 0.004 0.004 0.914 ± 0.004 0.896 ± 0.005 0.903 ± 0.005 0.880 ± 0.005 0.539 HI 0.075 ± 0.003 0.096 ± 0.003 <0.001 0.084 ± 0.004 0.087 ± 0.004 0.509 0.073 ± 0.005 0.094 ± 0.005 0.076 ± 0.004 0.098 ± 0.005 0.892 D2%* 1.079 ± 0.004 1.103 ± 0.004 <0.001 1.089 ± 0.004 1.093 ± 0.004 0.525 1.078 ± 0.005 1.101 ± 0.006 1.080 ± 0.005 1.105 ± 0.006 0.893 D50%* 1.052 ± 0.003 1.070 ± 0.003 <0.001 1.060 ± 0.003 1.062 ± 0.003 0.562 1.051 ± 0.004 1.069 ± 0.004 1.053 ± 0.004 1.071 ± 0.004 0.998 Dmax* 1.109 ± 0.004 1.134 ± 0.004 <0.001 1.122 ± 0.005 1.121 ± 0.004 0.928 1.110 ± 0.006 1.133 ± 0.006 1.108 ± 0.005 1.134 ± 0.006 0.84 R50% 4.599 ± 0.075 4.824 ± 0.081 0.047 4.726 ± 0.077 4.697 ± 0.083 0.797 4.603 ± 0.099 4.849 ± 0.115 4.595 ± 0.115 4.799 ± 0.118 0.853 D2cm* 0.591 ± 0.013 0.599 ± 0.014 0.68 0.597 ± 0.014 0.593 ± 0.013 0.82 0.596 ± 0.019 0.599 ± 0.020 0.587 ± 0.017 0.599 ± 0.019 0.792 Total MU* 62.235 ± 1.084 61.574 ± 1.222 0.653 58.741 ± 0.977 65.068 ± 1.098 <0.001 60.140 ± 1.377 57.342 ± 1.349 64.331 ± 1.571 65.806 ± 1.556 0.149 Beam-on time (mins) 3.001 ± 0.154 2.997 ± 0.168 0.954 2.041 ± 0.031 3.957 ± 0.062 <0.001 2.090 ± 0.043 1.992 ± 0.042 3.912 ± 0.089 4.002 ± 0.087 0.181 Dose cal. Time (s) 45.675 ± 2.623 18.350 ± 0.376 <0.001 22.900 ± 1.119 41.125 ± 3.329 <0.001 29.550 ± 0.659 16.250 ± 0.228 61.800 ± 0.659 20.450 ± 0.256 <0.001 Note: *Due to variations in prescription dose between 2 protocols, individual plan values for any parameter are expressed as a ratio relative to the prescription dose. Table 3: Dosimetric comparison between dose calculations of AAA and AXB and RT equipment of TrueBeam and Halcyon in target coverage, CI, HI, D2%*, D50%*, Dmax*, R50%, and D2cm*.

Trachea and ipsilateral bronchus Dmax* 0.410 ± 0.030 0.421 ± 0.031 0.797 0.413 ± 0.030 0.419 ± 0.031 0.893 0.407 ± 0.043 0.418 ± 0.043 0.413 ± 0.044 0.425 ± 0.045 0.985 Esophagus Dmax* 0.201 ± 0.010 0.206 ± 0.011 0.75 0.200 ± 0.010 0.207 ± 0.011 0.664 0.198 ± 0.015 0.202 ± 0.015 0.203 ± 0.015 0.210 ± 0.016 0.912

Dosimetric Comparison of CI, HI, D50%, and Dmax

AAA vs AXB CI was found to be higher for AAA (0.909 ± 0.003) compared to AXB (0.888 ± 0.004) and HI was lower in AAA (0.075 ± 0.003) than in AXB (0.096 ± 0.003), with both differences being statistically significant (p < 0.001), indicating better conformity and homogeneity with AAA.

The dose to 50% of the volume (D50%*), and maximum dose (Dmax*) of the PTV were slightly higher for AXB (1.070 ± 0.003 and 1.134 ± 0.004, respectively) when compared with AAA (1.052 ± 0.003 and 1.109 ± 0.004, respectively) and the differences were statistically significant (p < 0.001).

Halcyon vs TrueBeam When comparing both RT equipment, TrueBeam (0.905 ± 0.003) had better CI than Halcyon (0.892 ± 0.004) with p = 0.004. No significant differences were observed in HI, D50%*, and Dmax* (p-values > 0.05).

Calculation Algorithms vs RT Equipment (Hal-AAA, Hal-AXB, TB-AAA and TB-AXB) When comparing the dosimetric outcomes of the AAA and AXB in Halcyon and TrueBeam, all p-values were greater than 0.539 and no significant differences were observed.

Dosimetric Comparison of R50% and D2cm

AAA vs AXB The dose fall-off (R50%) was sharper in AAA (4.599 ± 0.075) than in AXB (4.824 ± 0.081), and the difference was significant (p = 0.047). No significant difference was found in D2cm* (p > 0.05). Therefore, AAA provided better dose fall- off than AXB while maintaining similar levels of dose spillage compared to the AXB.

Halcyon vs TrueBeam Similarly, no significant difference was observed between Halcyon and TrueBeam in R50% and D2cm* with both p > 0.05. This suggested that both RT equipment had comparable dosimetric precision for R50% and D2cm*.

Calculation Algorithms vs RT Equipment (Hal-AAA, Hal-AXB, TB-AAA and TB-AXB) The analysis of the R50% and D2cm*, comparing the combinations of AAA and AXB in Halcyon and TrueBeam, indicated no significant differences with p > 0.05. The similar performance of different combination of calculation algorithms and RT equipment potentially allows for flexibility in clinical practice without compromising dosimetric accuracy.

Dosimetric Comparison of OARs

AAA vs AXB There were no statistically significant differences in the maximum dose (Dmax*) delivered to the trachea and ipsilateral bronchus, esophagus, great vessels, heart, lungs, ribs, skin, and spinal cord in the comparison between AAA and AXB, which p-values were greater than 0.05 consistently. The V20Gy(%), V10Gy(%), and V5Gy(%) of lung were also comparable between the two algorithms (p-values > 0.05).

Halcyon vs TrueBeam Similarly, the dosimetric comparison between the Halcyon and TrueBeam revealed no significant differences in the OARs. All parameters of the evaluated OARs were comparable, with p-values > 0.05, indicating no significant variation between both RT equipment.

Calculation Algorithms vs RT Equipment (Hal-AAA, Hal-AXB, TB-AAA and TB-AXB) The comparative analysis between Hal-AAA, Hal-AXB, TB-AAA, and TB-AXB indicates that there are no statistically significant differences in the dosimetric parameters including Dmax, Dmean, V20Gy(%), V10Gy(%), and V5Gy(%) evaluated for the OARs (p-values > 0.05).

This suggests that both AAA and AXB can be used interchangeably on either the Halcyon or TrueBeam systems without significant variations in dose delivery to the OARs. These findings could potentially support the flexibility of using either calculation algorithm in clinical practice, although treatment planning should always be carefully tailored to the individual patient’s needs.

Dosimetric Comparison of Treatment Delivery and Dose Calculation Time AAA vs AXB Both total MU* and BOT in AAA and AXB had p-value > 0.05. There was no significant difference in dosimetric delivery efficiency between AAA and AXB based on total MU* and BOT. AAA and AXB demonstrated comparable dosimetric delivery efficiency. The dose calculation time for the AAA was significantly longer, averaging 45.675 ± 2.623 seconds, compared to the AXB with 18.350 ± 0.376 seconds. The difference in calculation times between the two algorithms was statistically significant (p < 0.001).

Halcyon vs TrueBeam When comparing Halcyon and TrueBeam, a significant difference was found in both total MU* per fraction and BOT. Halcyon required more total MUs* per fraction (65.068 ± 1.098 MU*) compared to TrueBeam (58.741 ± 0.977 MU*), with p-value < 0.001. Additionally, the BOT of Halcyon was longer (3.957 ± 0.062 mins) than TrueBeam (2.041 ± 0.031 mins), with p-value < 0.001.

The dose calculation time for Halcyon was 41.125 ± 3.329 seconds, which was longer than the time of 22.900 ± 1.119 seconds calculated by TrueBeam. Significant differences were observed in dose calculation time between Halcyon and TrueBeam with p < 0.001.

Calculation Algorithms vs RT Equipment (Hal-AAA, Hal-AXB, TB-AAA and TB-AXB) The combination of calculation algorithms in Halcyon and TrueBeam revealed no significant difference in total MU* and BOT (p > 0.05).

Significant differences were observed in the dose calculation time in the combination of both calculation algorithms in Halcyon and TrueBeam with p < 0.001. Within Halcyon, AAA had a dose calculation time of 61.800 ± 0.659s, while the AXB had a significantly shorter time of 20.450 ± 0.256s. For TrueBeam, the AAA had a calculation time of 29.550 ± 0.659s, and AXB had a time of 16.250 ± 0.228s. The TB-AXB demonstrated a significantly reduced dose calculation time compared to the other three combinations (Hal-AAA, Hal-AXB, TB-AAA), with p < 0.001 indicating that the differences are statistically significant.

Treatment Delivery

The statistically significant difference in total MU* and BOT between the Halcyon and TrueBeam can be attributed to their design and technology differences. Halcyon is designed as a streamlined workflow with fewer customizable options, and different dose rates, MLC, and field size, which might lead to a higher total MU and longer BOT to achieve a similar dose distribution [4]. As the stacked and staggered MLC system is used in Halcyon, it provides less interleaf leakage in nominal 6MV-FFF transmission, which is 0.01% in Halcyon MLC and 1.36% in Millennium 120. Therefore, it may require more MUs for Halcyon [4].

Shorter BOT and lower total MU in TrueBeam might be influenced by its more advanced beam modulation capabilities, such as a higher dose rate (1400MU/min) and larger field size for modulation, allowing for more efficient treatment delivery. The significant difference in BOT, with a p-value of less than 0.001, underscores a potential advantage of TrueBeam in reducing patient treatment time, which can reduce the dose uncertainties due to the patient’s motion during the treatment.

Besides, although Halcyon provides comparable dosimetric differences to TrueBeam, TrueBeam has the potential for dual-energy (photon and electron) irradiation for various applications in radiation therapy while Halcyon has only single 6MV photon.

Dose Calculation Time

AXB demonstrated significantly shorter dose calculation times (18.35 ± 0.376 s) compared to AAA (45.68 ± 2.62 s; p < 0.001). This notable dose calculation time reduction by AXB could be associated with the utilization of Graphics Processing Unit (GPU). Similarly, TrueBeam showed superior dose calculation times (22.90 ± 1.12 s) than Halcyon (41.13 ± 3.33 s; p < 0.001), attributed to TrueBeam’s higher dose rate and simpler single-layer MLC design compared to Halcyon’s stacked-and-staggered dual-layer MLC.

When combining both calculation algorithms and RT equipment, AXB consistently outperformed AAA on both TrueBeam and Halcyon (p-value < 0.001), with a more pronounced improvement on Halcyon, suggesting better optimization of AXB for Halcyon. These findings highlight the clinical efficiency of GPU-based AXB, allowing faster treatment planning and adjustments, particularly beneficial in high-volume centers.