Stereotactic radiosurgery (SRS), a crucial treatment technique for intracranial brain metastases, involves delivering high radiation doses to an intracranial target in a single session. Since its introduction by Leksell, SRS has demonstrated substantial efficacy, achieving local control rates of 80%–95% over 5–10 years with minimal long-term toxicity [1-8].

Generally, precisely delivering high radiation doses to the intracranial target while minimizing exposure to surrounding critical organs represents a critical aspect of SRS. Hence, ensuring proper patient immobilization and positioning during SRS is crucial. Initially, frame-based SRS systems using fixed head frames were employed to ensure accurate target localization during planning and delivery [9-11]. However, these systems presented notable disadvantages, including patient discomfort, frame slippage, and a slight risk of complications such as bleeding and infections [11]. Consequently, linear accelerators (Linacs) with multileaf collimator (MLC) delivery systems emerged as the preferred alternatives.

Notably, advancements in radiotherapy, particularly volumetric modulated arc therapy (VMAT), have enabled the treatment of large, complex, or multiple brain lesions using multiple arcs. For instance, RapidArc (Varian Medical Systems) is a VMAT technique offering coplanar or noncoplanar, isocentric delivery. Through 1–3 rotations of a Linac gantry, this technique can deliver intensity-modulated radiation doses with high conformality, producing superior treatment plans compared to multi-field intensity-modulated radiation therapy, while reducing the treatment duration [12,13]. HyperArc (Varian Medical Systems) is another novel isocentric VMAT technique particularly designed for MLC-based noncoplanar stereotactic radiotherapy. This technique features automated optimization and delivery, focusing on minimizing planning workloads by automatically determining ideal isocenters, collimator angles, and arc arrangements. This approach ensures maximal conformality and minimal radiation spillage into surrounding tissues [14-16]. Notably, HyperArc combines up to four arcs: a full arc with a couch angle of 0° and three half arcs with couch angles of ±45° and ±90°. It utilizes standard immobilization devices and incorporates automated couch transitions to enhance delivery efficiency [17,18].

This study evaluates and compares the dosimetric performance of HyperArc and RapidArc in SRS for a single brain metastasis, intending to share the acquired clinical experiences and insights regarding the clinical implementation of HyperArc.

1. Patient and plan preparation

From January 2017 to December 2022, twenty patients diagnosed with intracranial brain metastases, each presenting a distinct target volume, were retrospectively selected at our institution. These patients had previously undergone SRS using the VMAT technique.

For diagnostic purposes, computed tomography scans of the treatment sites of all patients were recorded using an immobilization technique. These scans were performed using Discover RT (GE Healthcare), yielding images with a resolution of 512×512 pixels and a slice thickness of 1 mm.

An oncologist delineated the gross target volumes (GTVs) and normal organs based on T1-weighted and T2 magnetic resonance imaging. For this analysis, the selection of normal organs was exclusively confined to normal brain tissues, thus excluding other organs. The planning target volume (PTV) was defined by extending an isotropic margin of 1 or 2 mm from the GTV. To facilitate dosimetric evaluations and plan optimizations, two concentric rings were created around the PTV at distances of 3 and 13 mm, termed as Ring_3mm and Ring_13mm, respectively. Notably, any overlap of the PTV with the normal brain tissue was excluded from the normal brain structure to improve PTV coverage.

2. Treatment planning

Each VMAT plan in this study used 6 MV photon beams without flattening filter from a TrueBeam STx, equipped with a high-definition 120^TM MLC system (Varian Medical Systems). These plans were specifically designed as HyperArc plans (VMAT_HA) including four automatically arranged arcs: a single full arc (couch rotation: 0°) and three half arcs (couch rotations: ±45° and ±90°). To minimize the dispersion of low doses to normal brain tissues, a half noncoplanar arc was excluded based on the tumor location. The HyperArc software automatically determined the isocenter by centering it on the PTV. Additionally, ideal collimator angles were automatically configured.

All VMAT_HA plans were optimized using a photon optimizer (version 16.1; Varian Medical Systems) with a uniform voxel resolution of 2.5 mm. Furthermore, the jaw-tracking function was employed to minimize leakage doses to normal organs. Prescription doses for the PTV varied from 14–25 Gy in one fraction, according to tumor characteristics and locations. Throughout the optimization process, we adhered to the planning constraints outlined in the report of the American Association of Physicists in Medicine Task Group 101 [19], focusing on sparing the normal brain tissue and thus preventing further complications. Table 1 lists the planning constraints for both the target volume and normal brain tissue in brain SRS. To optimize the target coverage and minimize dose leakage to surrounding normal tissues, the automatic lower dose objective and SRS normal tissue objective techniques were applied. To further enhance the dosimetric plan quality, each VMAT_HA plan was reoptimized using the existing dose distribution as its base distribution. Subsequently, dose calculations were performed using the Acuros XB advanced dose calculation algorithm (version 16.1; Varian Medical Systems) on a 2 mm calculation grid. Plan normalization ensured that 100% of the prescribed dose covered 95% of the PTV.

Table 1 . Planning constraints for the planning target volume, normal brain tissue, and ring structures surrounding the PTV at distances of 3 and 13 mm in brain stereotactic radiosurgery plans.

Structure	Planning constraints	Priority
PTV	V100% >95%	150
	D100% >100% of the prescription dose	150
Normal brain	V2Gy <20%	100
Ring3mm	Dmax <70% of the prescription dose	100
Ring13mm	Dmax <30% of the prescription dose	100

PTV, planning target volume; V_n%, volume receiving n% of the prescription dose; D_n%, dose received by n% of the target volume; V_nGy, volume receiving n Gy; D_max, maximum dose; Ring_nmm, ring structure surrounding the PTV at a distance of n mm..

For comparative analysis, each VMAT plan was also created as a RapidArc plan (VMAT_RA) using 6 MV photon beams without flattening filter from a Clinac iX equipped with a Millennium 120 MLC system (Varian Medical Systems). Each VMAT_RA plan comprised one full arc with a couch rotation of 0° and one half noncoplanar arc with a couch rotation of 90° (or 270°). In these plans, collimator angles were manually set to 30° and 330°. All VMAT_RA plans were optimized using the progressive resolution optimizer (version 8.9; Varian Medical Systems), following identical planning protocols and constraints as those applied for the VMAT_HA plans.

3. Evaluations of treatment plans

The evaluation of dosimetric quality in this study involved a comprehensive analysis of the dose-volumetric histogram (DVH) parameters for each plan, focusing on the target coverage and doses delivered to normal organs. The DVH parameters assessed for the PTV included minimum, maximum, and mean doses. Additionally, the conformity index (as proposed by Paddick et al. [CI_paddick]), gradient index (GI), and homogeneity index (HI) were computed as follows [20-22]:

where TV_{prescription dose} represents the target volume encompassed by the prescription dose, TV denotes the total target volume, V_{prescription dose} represents the volume receiving the prescription dose, and V_{prescription dose 50%} denotes the volume receiving 50% of the prescription dose.

Similarly, the DVH parameters analyzed for the normal organs included the volumes receiving 100% and 50% of the prescription dose (V_100% and V_50%, respectively), along with the volumes receiving 18 Gy (V_18Gy), 12 Gy (V_12Gy), and 2 Gy (V_2Gy).

Treatment efficiency and deliverability were evaluated using the total beam-on time, total monitor units (MUs), and modulation complexity score for VMAT (MCS_v). The MCS_v metric, developed by Masi et al. [23], evaluates the complexities in MLC motion and beam apertures in VMAT plans. A lower MCS_v value indicates increased modulation complexity. The foregoing metrics were computed using an inhouse software developed in MATLAB R2021a (MathWorks).

Statistical analysis was performed using the Shapiro–Wilk test to determine the normality of the datasets. Depending on the obtained results, either the Wilcoxon signed-rank test or a paired t-test was employed to evaluate the dosimetric differences in the DVH parameters, total MUs, beam-on times, and MCS_v between the VMAT_HA and VMAT_RA plans. Statistical significance was established at P<0.05. Overall analyses were conducted using the statistical program PRISM (version 8.4.3; GraphPad Software Inc.).

1. DVH parameters

Table 2 summarizes the average DVH parameters for the PTV in both the VMAT_HA and VMAT_RA plans. Our analysis revealed statistically significant differences in all the considered DVH parameters between VMAT_HA and VMAT_RA (P<0.05), except for CI_paddick (P=0.152). Notably, compared to VMAT_RA, VMAT_HA presented lower minimum dose values and considerably higher maximum and mean dose values. Additionally, compared to VMAT_RA, VMAT_HA presented significantly elevated HI values. Moreover, VMAT_HA presented significantly smaller GI values (P<0.001), suggesting a more effective dose fall-off outside the PTV, compared to VMAT_RA. Overall, the quality of the DVH parameters for the PTV was superior in VMAT_HA compared to VMAT_RA.

Table 2 . Average DVH parameters for planning target volumes in brain stereotactic radiosurgery plans.

DVH parameter	VMATRA	VMATHA	P-value
Volume (cm3)	9.77±7.42		-
Minimum dose (%)	95.28±1.97	90.81±4.08	<0.0001
Maximum dose (%)	112.77±11.12	142.60±6.60	<0.0001
Mean dose (%)	105.94±5.39	119.33±4.57	<0.001
CIpaddick	1.10±0.13	0.99±0.09	-
HI	1.13±1.11	1.43±0.66	<0.0001
GI	3.79±0.76	2.56±0.17	<0.001

Data are presented as mean±standard deviation..

DVH, dose-volumetric histogram; VMAT_RA, volumetric modulated arc therapy with RapidArc; VMAT_HA, volumetric modulated arc therapy with HyperArc; CI_paddick, conformity index suggested by Paddick et al.; HI, homogeneity index; GI, gradient index..

Table 3 summarizes the average DVH parameters for the normal brain tissue in both the VMAT_HA and VMAT_RA plans. Across all parameters, VMAT_HA exhibited significantly lower values than VMAT_RA. In particular, the differences in V_50% and V_2Gy for the normal brain tissue between VMAT_HA and VMAT_RA were substantial (23.37 cm³ and 37.29 cm³, respectively, with P<0.0001 for V_50% and 11.28% and 17.82%, respectively, with P<0.0001 for V_2Gy).

Table 3 . Average DVH parameters for normal brain regions in brain stereotactic radiosurgery plans.

DVH parameter	VMATRA	VMATHA	P-value
Volume (cm3)	1,460.82±296.47		-
V100% (cm3)	11.03±9.32	9.40±7.05	<0.001
V50% (cm3)	37.29±25.43	23.37±16.70	<0.0001
V18Gy (%)	0.28±0.32	0.08±0.08	0.021
V12Gy (%)	1.04±0.92	0.46±0.46	0.002
V2Gy (%)	17.82±11.18	11.28±9.49	<0.0001

Data are presented as mean±standard deviation..

DVH, dose-volumetric histogram; VMAT_RA, volumetric modulated arc therapy with RapidArc; VMAT_HA, volumetric modulated arc therapy with HyperArc; V_n%, volume receiving n% of the prescription dose; V_nGy, volume receiving at least n Gy..

Thus, employing HyperArc for generating VMAT plans appears to enhance the dose fall-off outside the PTV and deliver a higher maximum dose within the PTV, while more effectively minimizing the exposure to surrounding normal tissues in brain SRS. For illustrative purposes, the dosimetric evaluations included dose distributions from VMAT_HA and VMAT_RA for two representative patients, as depicted in Fig. 1. The DVHs for these patients are presented in Fig. 2.

Figure 1. Representative dose distributions of brain stereotactic radiosurgery cases (patient #3 and #14): Dose distributions of volumetric modulated arc therapy plans created using RapidArc volumetric modulated arc therapy (VMAT_RA) (a) and HyperArc (VMAT_HA) (b) for patient #3. Dose distributions of VMAT_RA (c) and VMAT_HA (d) for patient #14. Doses are depicted by color wash with 2 Gy (the lowest dose) in blue and 30 Gy (the highest dose) in red.

Figure 2. Representative dose-volumetric histograms of brain stereotactic radiosurgery cases (patient #3 and #14). Volumetric modulated arc therapy plans generated using RapidArc (RA) (a) and HyperArc (HA) (b) plotted using solid and dashed lines, respectively, for the planning target volume (PTV) and normal brain.

2. Total MU, total beam-on time, and MCS_v

Table 4 presents the average total MUs, beam-on times, and MCS_v values. The data indicate that the VMAT_HA plans are more complex than the VMAT_RA plans, with statistically significant differences (MCS_v values of 0.539 for VMAT_HA and 0.619 for VMAT_RA, P<0.001). This complexity is further emphasized by the mean total MUs and beam-on times. Specifically, the average total MUs for VMAT_HA significantly exceeded those for VMAT_RA (6,704.85 and 4,245.60, respectively, with P<0.0001). However, the total beam-on time for VMAT_RA exceeded that for VMAT_HA (424.56 seconds compared with 287.35 seconds, respectively, with P<0.0001). Notably, these beam-on times were calculated using different dose rates for each system: 600 MU/min for VMAT_RA and 1,400 MU/min for VMAT_HA. This variation in dose rates is a critical factor in understanding the differences in beam-on times between the two systems, reflecting the inherent complexities and operational characteristics of each technology.

Table 4 . Average total monitor units, beam-on time, and modulation complexity scores of volumetric modulated arc therapy plans for brain stereotactic radiosurgery.

	VMATRA	VMATHA	P-value
MU	4,245.60±750.58	6,704.85±1,497.61	<0.0001
Beam-on time (s)	424.56±75.06	287.35±64.18	<0.0001
MCSv	0.619±0.032	0.539±0.045	<0.001

Data are presented as mean±standard deviation..

VMAT_RA, volumetric modulated arc therapy with RapidArc; VMAT_HA, volumetric modulated arc therapy with HyperArc; MU, monitor unit; MCS_v, modulation complexity score for volumetric modulated arc therapy generated by Masi et al. (2013) [23]..

In this investigation, we compared the performance of HyperArc and RaipdArc in generating brain SRS VMAT plans. In particular, we focused on analyzing the DVH parameters for both the target volume and surrounding normal organs, total MUs, beam-on times, and MCS_v values. Compared with VMAT_RA, VMAT_HA demonstrated a superior dose gradient from the PTV and a higher maximum dose in the PTV, while effectively minimizing dose exposure to the normal brain tissue. However, these improvements in plan dosimetric quality corresponded to an increase in the overall modulation complexity of the plan and total MUs, potentially affecting treatment deliverability and efficiency. Although VMAT_HA required approximately 1.6 times more total MUs than VMAT_RA, using TrueBeam STx with HyperArc, operating with 6 MV flattening filter free beams at a 1,400 MU/min dose-rate, could reduce the total beam-on time by a factor of 1.5 times compared with Clinac iX.

Similar studies, such as the study conducted by Vergalasova et al. [16], have explored the dosimetric quality of HyperArc at SRS treatment sites, further comparing it with the dosimetric qualities of other popular treatment plans such as GammaKnife and RapidArc. Their findings indicate that both RapidArc and HyperArc achieve conformality and low-dose brain spillage levels comparable to those of GammaKnife, while significantly reducing the beam-on time. Notably, the quality of the VMAT_RA plans varies substantially based on optimization constraint settings and the planner, while the VMAT_HA plans consistently demonstrate better dosimetric performance.

Our study highlights that using a next-generation treatment planning system for VMAT_HA produces a significantly higher HI than VMAT_RA plans, particularly if a dose constraint for the maximum dose is not established during optimization. Plans with higher HI values—traditionally considered less desirable for conventional fractionated radiotherapy—may be acceptable for hypo-fractionated therapies such as stereotactic body radiation therapy (SBRT) and SRS, particularly if the maximum dose is constrained within the GTV. This concept aligns with GammaKnife units, often prescribed for PTV with a 50% isodose line, implying a maximum dose within the target volume that is twice the prescription dose. The VMAT_HA planning approach facilitated further improvements in plan quality, including sharp dose gradients and reduced exposure to surrounding brain regions. This technique also opens up possibilities for dose escalations in treating larger tumors while maintaining normal tissue tolerance. Studies, such as those by Dong et al. [24] and others, have demonstrated the feasibility of dose escalations in treating various cancers using noncoplanar dose delivery techniques such as 4π radiotherapy [25,26]. Future studies could explore the viability of dose escalations using VMAT_HA planning for brain metastases and other tumor sites.

Both the 4π and VMAT_HA planning approaches represent noncoplanar planning techniques on conventional C-arm-type Linacs. The 4π optimization method begins with a vast array of more than 1,162 beams distributed across the 4π angle space with a 6° gap between adjacent beams [26], requiring the machine to traverse several noncoplanar beams. In contrast, the VMAT_HA optimization integrates noncoplanar beams with VMAT using four couch angles and one isocentric irradiation, potentially achieving shorter dose delivery times for multiple targets compared to the 4π technique.

An essential aspect of VMAT_HA planning is the SRS normal tissue objective, aimed at achieving the most closely packed dose possible while minimizing the dose spread across targets. In this regard, our findings indicated that compared to the VMAT_RA plans, the VMAT_HA plans involved more intricate MLC patterns with smaller segments (lower MCS_v values). However, as reported by Ohira et al. [27], lower MCS_v values in VMAT plans could lead to reduced dosimetric accuracy, as evidenced by gamma pass rates, highlighting the need for meticulous dosimetric validation before clinically implementing VMAT_HA plans.

In this study, we utilized two different MLC systems: high-definition 120^TM MLCs and Millennium MLCs from TrueBeam STx and Clinac iX, respectively. Notably, these MLCs exhibit evident differences in their geometric and dosimetric characteristics. Numerous studies have explored the effects of different MLC systems on planning and deliverability. In planning comparisons, the dosimetric impacts of different MLC systems on plan quality have been observed to be minimal [28-30]. Subramanian et al. [30] demonstrated that MLCs with a 5 mm width completely satisfied the RTOG-0813 treatment planning criteria for lung SBRT. The dosimetric superiority of VMAT_HA resulted from the delivery technique and optimization algorithm for planning rather than the different types of MLC systems.

Increased exposure of nontarget tissues to low radiation doses represents a crucial concern when using single-isocentric VMAT_HA for multiple targets. This is attributed to shared MLC leaf pairs between two or more targets, preventing the effective blocking of radiation exposure to normal tissues surrounding multiple metastases [31]. Wu et al. [32] demonstrated that employing a collimator optimization algorithm could significantly reduce the dose exposure to normal brain tissues while maintaining similar target coverages. This algorithm, developed for VMAT_HA planning, solves the “island blocking” issue, and our results demonstrate that compared to multi-isocentric irradiation, VMAT_HA provides a lower volume of brain tissues receiving a dose >4 Gy. Despite the slightly higher V_2Gy of the VMAT_HA plan, this low-dose spread is not deemed clinically significant, specifically considering that it remains below the levels typically used for whole-brain irradiation in a single fraction.

Compared with VMAT_RA, VMAT_HA demonstrated superior performance in achieving rapid dose fall-offs (as indicated by the GI) and a higher HI at the PTV, while more effectively sparing normal tissues in brain SRS. The enhanced dosimetric plan quality achieved by HyperArc highlights its potential as a new and preferred option for delivering SRS doses, particularly in high-precision treatments for brain SRS. This advancement presents HyperArc as a significant development in the field of radiation therapy, offering optimized treatment outcomes for patients undergoing brain SRS.

This study was supported by the National Research Foundation of Korea (NRF) grant funded by the Korea government (MSIT) (No. RS-2023-00253604).

The authors have nothing to disclose.

All relevant data are within the paper and its Supporting Information files.

Conceptualization: So-Yeon Park, Na Young Jang. Data curation: So-Yeon Park, Noorie Choi. Formal analysis: So-Yeon Park. Funding acquisition: So-Yeon Park. Investigation: So-Yeon Park. Methodology: So-Yeon Park. Na Young Jang. Project administration: So-Yeon Park. Resources: Noorie Choi, Na Young Jang. Software: So-Yeon Park. Supervision: Na Young Jang. Validation: Noorie Choi. Visualization: So-Yeon Park, Na Young Jang. Writing – original draft: So-Yeon Park. Writing – review & editing: So-Yeon Park, Na Young Jang.

This study was approved by the Institutional Review Board of Veterans Health Service Medical Center (IRB No. 2023-01-001). The requirement for written informed consent was waived by the IRB due to the retrospective nature of the study.