Dosimetric and biological comparison of treatment plans between EDGE and CyberKnife systems in stereotactic body radiation therapy for localized prostate cancer

Aims: The aim of this study was to make a quantitative comparison of plan quality between MLC-based EDGE system and the cone-based CyberKnife system in stereotactic body radiation therapy (SBRT) for patients with localized prostate cancer. Materials and methods: Ten patients with prostate volumes ranging from 34.65 to 82.16 cc were used for prostate SBRT. Treatment plans were created for both EDGE and CyberKnife G4 systems using the same dose-volume constraints. Dosimetric indices including Planning Tumor Volume (PTV) coverage, conformity index (CI), new conformity index (nCI), homogeneity index (HI), gradient index (GI) were applied for target, while the sparing of critical organs, including bladder, rectum, femoral heads, urethra, penile bulk and normal tissue outside PTV), were evaluated interms of various dose-volume metrics and integral dose (ID). Meanwhile, the required delivery time and number of monitor units (MUs) during irradiation were measured to estimate the treatment eciency. The radiobiological indices such as equivalent uniform dose (EUD), tumor control probability (TCP) and the normal tissue complication probability (NTCP) were also analyzed. Results: All dose constraints were achieved by both systems. It showed that the DEGE plans results were closest to the CK plans results in terms of PTV coverage, HI and GI. For EDGE, more conformal dose distribution in the target as well as reduced exposure of critical organs were obtained together with reduction of 91% delivery time and 72% monitor units. EDGE plans also got lower EUD for bladder, rectum, urethra and penile bulk, which associated with reduction of NTCPs. However, higher values of EUD and TCP for tumor were obtained with CK plans. Conclusions: Our study indicated that both systems were capable of producing almost equivalent plan quality and can meet clinical requirements. CyberKnife G4 system has higher target dose while EDGE system has more advantages based on the considerations of normal tissue sparing and delivery eciency. With abundant clinical experience, CK provides accurate SBRT treatment with high quality. EDGE system also can be considered to be an option for SBRT treatment for localized prostate cancer treatment.


Introduction
Stereotactic Body Radiation Therapy (SBRT), or Stereotactic Ablative Radiotherapy (SABR) has grown up to be a signi cant treatment modality for several years, as an alternative of the conventional radiotherapy in prostate cancer [1][2][3][4]. Especially, SBRT has been recognized as an appropriate option in cases of localized prostate cancer [5][6][7][8][9]. The radiobiological rational for prostate SBRT is due to its relatively lower α/β ratio (been estimated at 1.5 Gy) than adjacent organs at risk (OARs), which implies the gains in cost effectiveness and biologically equivalent dose to large fractionated radiotherapy [10][11][12]. Trials have reported superior biochemical control outcomes for patients with prostate cancer by hypo-fractionation [1][2][3][4]. SBRT for prostate cancer was recommended as an alternative to conventionally fractionated regiments according to ASTRO model policy update of 2013, as well as National Comprehensive Cancer Network (NCCN) guidelines on prostate version 2.2014. There remains however the technical limitations in the delivery of such high doses due to the proximity of sensitive normal tissues and organs. Therefore, more conformal radiation and sharper dose fall-off outside the targets are necessary in order to deliver such high dose safely.
Currently, multiple techniques available are developed for SBRT treatments [13,14], among which CyberKnife® (Accuray Inc., Sunnyvale CA) system has been known as one of the predominant SBRT facilities applied in the treatment of prostate cancer [14]. CyberKnife (CK) is a frameless image-guided radiotherapy system involving a 6-MV FFF (Flattening Filter Free) linear accelerator mounted on a exible robotic arm, which makes it capable of delivering radiation from hundreds of non-coplanar directions. Moreover, its ducial tracking technique allows for real-time tumor position and motion corrections during prostate SBRT treatment. These capabilities would make it produce improved conformal isodose with high precision [15].
Meanwhile, LINAC, using multileaf collimator (MLC), can also be used for SBRT by either intensitymodulated radiotherapy (IMRT) or volumetric modulated arc therapy (VMAT) [16]. EDGE® (Varian Medical Systems, Palo Alto, CA), an update version of TrueBeam, is one of the typical LINAC-based SBRT system. This dedicated machine is equipped with the HD (High de nition) 120 leaf MLC (Multi Leaf Collimator), with two modes of FF (Flattening Filter) and FFF beam delivery [17][18]. The MLC leaf resolution improvement with 2.5 mm leaf widths which allows more conformal dose delivery to the target. This system is equipped with multiple imaging modalities for treatment localization.
In order to make it clear which technique is superior, many comparative studies have been carried out between the LINAC and CK system for prostate SBRT [16,[19][20][21].. However, there is no study directly comparing the characteristics of dose distribution of treatment plans between EDGE and CK. Therefore, it is essential to make a further study on the properties about emerging treatment technology of EDGE system for making an appropriate option for individualized SBRT treatment.
In our study, we performed a comprehensive evaluation of plan quality with the dose performance of EDGE compared to CK SBRT plans for prostate cancer. These comparison results were implemented by adopting some physical and radiobiological indices according to the dose volume histograms (DVHs) calculated on the evaluation software framework developed by our group. The nal analyzed results can be used to nd out virtues and shortcomings in optimized plans of each technique for making the most appropriate choice in prostate SBRT treatment. Besides, the monitor units (MUs) used and the beam-on times were also compared to examine the delivery e ciency for both systems.

Case selection and volume de nition
Ten patients with localized prostate cancer staged T1-T2b treated using CK SBRT at our institution between 2018 and 2019 were enrolled randomly. Each patient was scanned in head rst-supine position, with a full bladder and an empty rectum. Computed tomography (CT) simulation was performed on a Brilliance™ Big Bore 16-slice CT scanner (Philips, Amsterdam, the Netherlands) with a slice thickness of 1.5 mm. Clinical target volume (CTV) and critical structures were contoured jointly by oncologist and radiologist based on the fusion of CT and magnetic resonance (MR) images on the MultiPlan® system (Accuray Inc., Sunnyvale CA; version 4.02). CTV was de ned as the whole prostate gland, with sizes of 59.15 ± 15.63 cc (median, 61.48 cc). Planning Target Volume (PTV) was expanded from CTV with a 5 mm isotropic margin, except 3 mm posteriorly according to the literature [1,2], with sizes of 98.25 ± 23.65 cc (median, 106.47 cc). Organs at risk (OARs) including bladder, rectum, small bowel, femoral heads, penile bulb, and urethra were contoured. The planning CT together with contours mentioned above were transferred to the Varian Eclipse® system (Varian Medical Systems, Palo Alto, CA; version 13.5) for EDGE planning.

Treatment planning
Two sets of plans were produced with the same CT images and delineated structures. For the purpose of comparison, all the plans were required to prescribe the same dose of 36.25 Gy delivered in 5 fractions and the prescription dose corresponds 100% non-normalized isodose. Dose constraints were set based on the criteria of the RTOG-0938 and previous studies [1,3,7,22]. Required planning constraints are detailed in Table 1. The CK plans were carried out with Multiplan® version 4.0.2 using sequential optimization method. A 6 MV FFF photon beam was employed with a dose rate of 800 MU/min and one or two cones with size of 20 ~ 30 mm. The plans were optimized with sequential process based on the ray tracing algorithm (RTA). Besides, 5 'shells' expanded isotropically from PTV were used to make steep dose fall-off gradient. At the end of the optimization, beams and time reduction were used to make the plan clinically practical. The VMAT plans were produced for EDGE system with the Eclipse version 13.5 using two full 360• arcs with the same isocenter located at the geometric center of PTV. The 10MV FFF photon beams at a high dose rate of 2400 MU/min were used in the optimization [17,23]. The plans were optimized with progressive resolution optimizer (PRO) and calculated with the analytical anisotropic algorithm (AAA) with a grid size of 1.5 mm. As is listed in Table 2, the maximum, minimum and mean dose (D max , D min and D mean ) as well as coverage (V 100 ) of CTV and PTV were evaluated.. Meanwhile V 120 , V 125 and V 130 of PTV were also recorded to compare the details of hot spots in target volume. The volumes covered by 37 Gy, 100% and 50% of prescription isodose line (PIDL) for bladder, and that covered by 36 Gy, 100%, 90%, 80%, 75%, 50% of PIDL for rectum were categorized for plan evaluation. Meanwhile, D max and D mean were analyzed for all the OARs. To investigate the details of dose distribution outside PTV, V 20 ,V 50 and V 100 of normal tissue were also compared.

CI, HI and GI
Additionally, conformity index (CI), new conformity index (nCI), homogeneity index (HI) and gradient index (GI) were also used to quantify the plan quality. The conformity index (CI) and new conformity index (nCI) describes how well the dose conforms to the boundary of the target volume and was de ned as follows [25,26] : (see Equations 2 and 3 in the Supplementary Files) where V Rx is the prescription isodose volume while V PTV and are the volume of PTV and that covered by the PIDL. Smaller CI and nCI imply a more conformal plan and the ideal values for both indices are 1.0.
The homogeneity index (HI) evaluates the degree of uniformity of dose inside the target volume [27]. Mathematically, the index was calculated according to the following equation: (see Equation 4 in the Supplementary Files) where D 2 (D 98 ) is the dose that covers 2% (98%) of the PTV, and D P is prescription dose. Usually, HI > 0, and HI = 0 means each voxel of target volume receives the same dose. The gradient index (GI) is implemented to assess the degree of the dose fall-off outside the target [28]. This index was expressed as follows: (see Equation 5 in the Supplementary Files) where V 50 and V 100 are the volumes covered by 50% and 100% prescription dose, respectively. A smaller value of GI indicates steeper dose fall-off.

EUD
The equivalent uniform dose (EUD), obtained with the DVH reduction method, is used to convert the inhomogeneous dose distribution into a simple uniform dose [29,30]. The EUD calculation was based on the phenomenological model suggested by Niemierko [29] and was de ned as: (see Equation 6 in the Supplementary Files) where v i is the percentage of voxels receiving dose d i . The v i and d i values are acquired from the DVHs and the sum of v i over all voxels equals to 1. a is a parameter which re ects the dose response property of distinct organs, and in some literatures the parameter n is used with a = 1/n. In clinical practice, a large negative value is employed to tumor, while large positive and small positive values are used for serial and parallel organs, respectively. a or n values in Table 3 were used here for tumor [30], bladder [31], rectum [32], femoral head [28,29], urethra [33] and penile bulk [34]. DVH of different doses per fraction is converted into biologically equivalent physical dose of 2 Gy per fraction (EQD 2 ) using the linear quadratic (LQ) model according to reference [29]. In the formula of EQD 2 , n f is the number of fractions. The α/β is a parameter from the issue-speci c LQ model of the certain organ, determining the fractionation sensitivity. α/β values in Table 3 were used here for tumor [10][11][12], bladder [35], rectum [36], femoral head [37]. Since there was no clinical data of α/β values for urethra and penile bulk, α/β = 3.0 was applied here as was usually used for most of OARs,

NTCP
The normal tissue complication probability (NTCP) were calculated based on the Lyman-Kutcher-Burman (LKB) model [29,30], in which NTCP for an organ to equivalent uniform dose (EUD) is given by (see Equation 9 in the Supplementary Files) where (see Equation 10 in the Supplementary Files) m is a dimensionless parameter and TD 50 is the whole organ dose for which NTCP is 50%. TD 50 and m for bladder [31], rectum [32], femoral head [29,30], urethra [33] and penile bulk [34] with de nitive clinical endpoints were listed in Table 3.

Statistical analysis
All the parameters were calculated from the DVHs with an in-house program based on C++. Statistical analyses were carried out using IBM SPSS Statistics version 21 (SPSS Inc.Armonk, NY). A paired t-test was performed to analyze the difference between EDGE and CK plans, and a p value < 0.05 was considered to reveal statistical signi cance.

Dose-volume metrics
All planning constraints detailed in Table 1 were met by both EDGE and CK plans. The comparison of isodose lines from 20-120% of the prescription dose for a selected case is illustrated in Fig. 1. Obviously, both plans are very conformal and provide adequate coverage of PTVs. Besides, we can nd that the 100% PIDL (with red color) of EDGE plan is closer to PTV boundary than that of CK plan.
The averaged DVHs of CTV, PTV, bladder, rectum, left and right femoral heads, urethral as well as penile bulk are displayed in Fig. 2(a)-(h), respectively. The values of dose-volume parameters of target and OARs are detailed in Table 2. From both Fig. 2(a)-(h) and Table 2, CTV and PTV coverage of EDGE and the CK plans were found to be of similar levels and showed no obvious difference. The mean dose (D mean ) of CTV and PTV are higher for CK, indicating larger ablation effect within target.
The bladder DVH indices (D max , D mean , V 37Gy , V 100% and V 50% ) from the EDGE plans were also statistically lower than the CK plans, presenting a distinct reduction of irradiation. The EDGE plans achieved slightly better rectum protection with respect to D max , V 36Gy , V 100% , V 90%, V 80% and V 75%. The irradiation dose of right and left femoral heads for both systems were very low and showed no signi cant difference in terms of D max and D mean . Moreover, D max and D mean of urethra and penile bulk were much lower for EDGE plans. The volumes normal tissue covered by 20%, 50% and 100% PIDL were all lower for EDGE plans, which were associated with better conformity and steeper dose fall-off gradient. Meanwhile, the integral dose of target volumes were a little larger for CK plans. Otherwise, the ID of OARs were much lower for bladder, urethral, penile bulk as well as normal tissure outside PTV for EDGE plans, while there were no much signi cant difference of ID for rectums and femoral heads.

Dosimetric indexes and delivery e ciency
The average of dosimetric indexes including CI, nCI, HI and GI are listed in

Radiobiological comparison
The radiobiological parameter EUD extracted from DVHs for CTV, bladder, rectum, left and right femoral heads, urethral and penile bulk, as well as TCP of CTV and NTCP of all these OARs were compared between the EDGE and the CK plans. The average values, standard deviation (SD), and p values were detailed in Table 5. The CK plans provided a slightly greater EUD and comparatively higher TCP than the EDGE plans. However, the larger EUD for bladder, rectum, urethral and penile bulk in CK plans were obtained, which indicated dramatically increasing NTCP of CK compared to EDGE plans for the four organs, respectively. The NTCP of femoral heads were too small to be considered, and showed no signi cant difference.

Discussion
In this study, we compared the plan quality of EDGE and CK in terms of dosimetric properties, delivery e ciency and predicted biological outcomes for prostate SBRT treatment. Both of the two techniques were able to produce clinically acceptable plans with adequate target irradiation and normal tissue sparing. Despite both systems were able to achieve excellent dose distribution according to the results above, EDGE had a little better performance in dosimetric results of conformity of PTV and better OAR sparing.
The EDGE plans were optimized using high de nition HD120 MLCs (with minimum spatial resolution of 2.5 mm) on the X axis with even littler size of gap on the Y axis, while the CK plans were made by 1-2 circular cones with size of 20∼30 mm. The high resolution of MLCs make it easier to reach more conformal dose distribution of PTV for EDGE, which will largely reduce the number of sub-elds.
The main reasons for the normal tissue sparing differences were due to the different characteristics of the two systems, which could be explained in two aspects. First and foremost, the plan optimization processes of the Multiplan version 4.0.2 and Eclipse 13.5 are very different. In the Multiplan, we could only set the maximum doses of OARs as constraints and optimize the mean doses of OARs, while in the Eclipse, several constraints could be set on the DVH curves of each OAR. This is one of the major reasons for superior OARs sparing of EDGE system. Further improvement for CK plan is feasible, if the optimization algorithm of Multiplan® evolves. Secondly, the beam arrangements in the process of planning optimization may play important roles for the dose distribution. CK offers superiority of highly exible angles, which delivered noncoplanar beams from all directions moved by the robotic arm while EDGE rarely used noncoplanar beams in the region of abdomen due to mechanical and geometrical limitations. However, the CK did not bene t from this advantage in this study because the beams of CK were mainly distributed in directions perpendicular to cranio-caudal (CC) direction in these plans, as the nal results of beam-angle optimization in light of the anatomical position of the prostates. The most bene cial beam angles were similar to those from two full 360 rotation arcs (178 segments for each plan) of EDGE which were rotated around CC direction.
As noted above, EDGE had the shortened average delivery time and the fewer MUs largely, as displayed in Table 4. Lessening treatment time means less scatter dose, which may lower the probability of secondary malignancies. On the other hand, decreased delivery time of EDGE can potentially reduce the effects of intra-fractional motion, and make the patients more comfortable. The VMAT technique, which delivers from a large number of angles with fewer control points, has been showed to decrease the number of MUs signi cantly, along with even lower MUs for dual-arc VMAT plans under the same condition as reported by Quan et al [40]. Moreover, EDGE system has 10FFF mode delivering the maximum high dose rate of 2400 MU per minute which severely shortens the beam-on time [18,23].
The radiobiological parameters in terms of EUD and TCP (NTCP) were calculated from DVHs, as showed in Table 5. The results indicated that the EDGE plans have slightly reduced CTV EUD than the CK plans, the results of which were in agreement with these of dosimetric evaluation. The mean EUD were lower for the OARs such as bladder, rectum, urethral and penile bulk in the EDGE plans in accordance with the calculated lower NTCP values consequentially. Both groups of plans were able to maintain high EUD to the tumors and yield good tumor TCP while low NTCP of normal structures were obtained in relation to late toxicity effects. For predicted clinical bene ts, this two treatment modalities can be considered to be safe.
Additionally, there also exists a concern for tumor and adjacent organs position variations throughout the course of treatment after the online match per fraction [41][42][43]. The intra-fraction prostate displacements were reported to be > 3 mm and > 5 mm were 24% and 5% of fractions respectively [43]. In this case, the target localization and real-time tracking systems are necessary to improve con dence in radiation dosimetry. Previous studies showed that CK has the competitive in light of target localization to deliver accurately in comparing conventional linear accelerator [44]. For the CK, two kilovoltage x-ray generators and two hereafter cameras are incorporated to nish ducial tracking for prostate motion [45]. Very small set-up errors were observed with 1.8 mm in the anterior posterior direction and 1.4 mm in the superior inferior direction [46]. However, EDGE system, designed for SBRT or SRS, has been improved to integrate Calypso 4D system capable of monitoring target position on the basis of radiographic transponder locations. Calypso system was reported to present a treatment accuracy of average 3D difference of 1.5 mm in dose delivery [47]. Thus EDGE has similar performance against motion uncertainties. For this reason, we delineated target margins for EDGE system according to the same protocols used for the CK. For all that, EDGE is lack of practical experience clinically by Calypso 4D system compared to CK.
Several limitations should be recognized in this investigation. Firstly, because the representative version of CyberKnife G4 system with the xed cone is most commonly used, it was selected to compare to the latest EDGE system in our study. The latest generation of CK system M6™, with IRIS collimator and InCise MLC, may increase the output rate and conformal dose distribution as well as to reduce delivery time.
Otherwise, the radiobiological parameters presented in this study are highly dependent on the model and related parameters. Therefore, the radiobiological responses could only be regarded as references when making clinical decisions. Further studies on clinical trials are required to collect practical experience and nd out which is the valuable option for localized prostate cancer.

Conclusion
A comparative quantitative assessment of the dosimetric and radiobiological indices of plans for both CybkerKnife and EDGE systems was made in this study. We con rm that radiotherapy systems with different characteristics should be investigated and utilized to help radiation oncologists choose a proper SBRT method for each individual patient to get better therapeutic effects. EDGE system can be used as an option for prostate cancer, especially for patients who cannot remain lying in bed for a long time.

Declarations
Ethics approval and consent to participate The study was approved by the institutional review board of our hospital.

Consent for publication
The consents for publication of data have been obtained from patients.

Supplementary Files
This is a list of supplementary les associated with this preprint. Click to download.