Abstract
Purpose:
To investigate whether there is a volume threshold in target volume of brain metastases below which a small cone size and sharp penumbra in Gamma Knife (GK) may provide improved plan quality when compared to Volumetric Modulated Arc Therapy (VMAT)-based stereotactic radiosurgery (SRS).
Methods:
For patients treated on GK SRS for brain metastases in 2018-2019 in our institution, 121 patients with two and three targets were identified. Twenty-six patients with two or three brain metastases (total of 76 lesions) were selected for this study. Two VMAT plans, SmartArc (Pinnacle) and HyperArc (Eclipse), were generated retrospectively for each patient. Plan quality was evaluated based on RTOG conformity index (CI), Paddick gradient index (GI), normal tissue (NT) V12Gy and V4.5Gy. By using the receiver operating characteristic (ROC) curve for both VMAT plans (SmartArc and HyperArc) and metrics of RTOG CI and NT V12Gy, we compared GK plans to SmartArc and HyperArc plans separately to determine the threshold volume.
Results:
For SmartArc plans, both ROC curve analyses showed a threshold volume of 0.4 cc for both CI and NT V12Gy. For HyperArc plans, the threshold volumes were 0.2 cc for the CI and 0.5 cc for NT V12Gy. GK plans produced improved dose distribution compared to VMAT for targets ≤0.4 cc, but HyperArc was found to have competing results with GK in terms of CI and NT V12Gy. For targets > 0.4 cc, both SmartArc and HyperArc showed better plan quality when compared to the GK plans.
Conclusions:
Target volumes ≤0.4 cc may require a small cone size and sharp penumbra in GK while for target volumes >0.4 cc, VMAT-based SRS can provide improved overall plan quality and faster treatment delivery.
Keywords: Gamma Knife, VMAT, Linac SRS, radiosurgery, brain metastases, plan quality
INTRODUCTION
Stereotactic radiosurgery (SRS) has been increasingly used for improving local control of brain metastases. For patients with limited numbers (1-4) of brain metastases,[1,2] SRS is especially favored over whole brain radiation for better sparing of normal brain tissue. Gamma Knife (GK) is one of the most commonly used modalities for delivering SRS to intracranial metastases. However, one main drawback of GK is the prolonged treatment delivery, especially when the Co-60 source activity decays, which not only further increases treatment time but also has a radiobiological disadvantage, resulting in increased survival fraction.[3] Linear accelerator (Linac)-based SRS is another commonly used method for brain metastases, which can be carried out using circular cones, dynamic conformal arcs, and volumetric modulated arc therapy (VMAT). With increased dose rates in flattening filter free (FFF) beams, VMAT-based SRS can treat multiple brain metastases with significantly less time using 1-2 isocenters as opposed to treating each target individually as is done with GK or cone-based Linac SRS.[3-5] Some studies reported that, VMAT-based SRS achieved similar conformity and normal brain tissue sparing to GK SRS.[6-9] Other studies reported that VMAT-based SRS generally provides improved target conformity, while the normal brain receiving intermediate and low dose spillage may increase compared to GK SRS.[10,11] Given large variations in target volumes and target shapes, several previous studies have shown that VMAT was more favorable than GK SRS for large and irregular targets.[7,11-13] However, there is still a question about whether there is a threshold volume at which GK SRS with small cone sizes (smallest cone size is 4 mm) and sharp penumbra (the virtual source to focus distance is ranged from 43 cm to 61 cm for Leksell Gamma Knife® Icon™) may produce improved plan quality when compared to VMAT-based SRS. Our hypothesis was that GK SRS could outperform VMAT-based SRS for small targets. With the various commercial treatment planning systems available today, we evaluated VMAT planning using SmartArc auto-planning tool (v16.2, Philips Medical System, Cleveland, OH) and HyperArc (v15.1, Varian Medical System, Palo Alto, CA). The use of the automated optimization in both systems may have an advantage of mitigating user dependency. This study was designed to investigate and determine what the lower limit of the target volume for VMAT-based SRS of multiple brain metastases would be, below which the GK SRS could provide improved plan quality.
MATERIALS AND METHODS
Patients
For patients treated on GK with single-fraction SRS for multiple brain metastases in 2018-2019 in our institution, 121 patients with two and three targets were identified. Twenty-six patients (76 lesions total) with selectivity of >0.5 for each target were selected for this study. Selectivity, which is defined as the ratio of the target volume receiving the prescribed dose to the volume of the prescribed dose, is one of the metrics for dose evaluation in radiosurgery planning. Our local institutional review board approved this study. Individual target volumes ranged from 0.01 to 17.68 cc (median, 0.75cc). Targets were prescribed doses ranging from 11-24 Gy. The prescribed radiation doses were modified under the discretion of radiation oncologists considering clinical conditions of each patient and the tolerance of critical structures such as maximum dose of 12 Gy to the brainstem and maximum dose of 10 Gy to the optic nerves. The target volumes and prescription dose of each patient are listed in Table 1. All patients received magnetic resonance imaging (MRI) with 1 mm slice thickness for target delineation as well as a CT scan with 1 mm or 1.5 mm slice thickness for treatment planning. The gross target volumes (GTVs) and normal tissue (NT) structures were contoured by a neurosurgeon based on MRI and were transferred to the planning CT after rigid image fusion.
Table 1.
Target volume and prescription per target for the 26 patients
| Patient | Target 1 | Target 2 | Target 3 | |
|---|---|---|---|---|
| 1 | Target volume (cc) Prescription (Gy) |
6.42 13 |
0.75 13 |
|
| 2 | Target volume (cc) Prescription (Gy) |
2.14 14 |
0.13 14 |
|
| 3 | Target volume (cc) Prescription (Gy) |
4.54 18 |
0.82 24 |
6.95 18 |
| 4 | Target volume (cc) Prescription (Gy) |
0.61 11 |
2.14 11 |
0.63 11 |
| 5 | Target volume (cc) Prescription (Gy) |
8.78 15 |
9.71 15 |
1.12 24 |
| 6 | Target volume (cc) Prescription (Gy) |
1.53 22 |
0.09 22 |
4.3 18 |
| 7 | Target volume (cc) Prescription (Gy) |
2.07 18 |
1.62 24 |
1.02 24 |
| 8 | Target volume (cc) Prescription (Gy) |
1.74 24 |
0.16 24 |
1.56 24 |
| 9 | Target volume (cc) Prescription (Gy) |
0.56 24 |
17.68 15 |
0.92 15 |
| 10 | Target volume (cc) Prescription (Gy) |
4.28 11 |
2.37 11 |
0.19 11 |
| 11 | Target volume (cc) Prescription (Gy) |
7.73 18 |
0.75 18 |
0.14 18 |
| 12 | Target volume (cc) Prescription (Gy) |
0.26 24 |
16.62 15 |
0.55 24 |
| 13 | Target volume (cc) Prescription (Gy) |
5.33 15 |
3.24 12 |
0.14 24 |
| 14 | Target volume (cc) Prescription (Gy) |
6.24 15 |
1.57 12 |
2.62 12 |
| 15 | Target volume (cc) Prescription (Gy) |
0.19 24 |
6.79 18 |
3.31 13 |
| 16 | Target volume (cc) Prescription (Gy) |
0.6 24 |
6.23 18 |
9.91 15 |
| 17 | Target volume (cc) Prescription (Gy) |
0.6 16 |
0.71 16 |
4.49 16 |
| 18 | Target volume (cc) Prescription (Gy) |
0.32 24 |
1.92 18 |
2.94 18 |
| 19 | Target volume (cc) Prescription (Gy) |
0.4 24 |
0.44 24 |
0.13 24 |
| 20 | Target volume (cc) Prescription (Gy) |
0.04 18 |
0.17 24 |
0.02 24 |
| 21 | Target volume (cc) Prescription (Gy) |
0.01 24 |
0.02 24 |
0.01 24 |
| 22 | Target volume (cc) Prescription (Gy) |
0.15 24 |
0.04 24 |
0.91 12 |
| 23 | Target volume (cc) Prescription (Gy) |
0.6 24 |
0.46 24 |
0.03 24 |
| 24 | Target volume (cc) Prescription (Gy) |
0.2 24 |
0.42 24 |
0.08 18 |
| 25 | Target volume (cc) Prescription (Gy) |
0.58 24 |
0.17 24 |
0.03 24 |
| 26 | Target volume (cc) Prescription (Gy) |
4.16 18 |
0.75 24 |
0.27 24 |
GK planning
GK plans were created using GammaPlan treatment planning system (version 10) for treatment on the Leksell Gamma Knife® Icon™, which contains 192 Co-60 sources. Each source has three available collimator sizes (4, 8 and 16 mm) as well as a blocked collimator. GK plans were generated by experienced neurosurgeons or medical physicists. Multiple GK shots were manually placed inside the target to deliver the prescription dose to >99% of the target volume. Further adjustments were made until clinically acceptable plan quality was achieved. A dose grid size of 0.5-1 mm was used in dose calculation depending on the size of the target.
VMAT planning
SmartArc plans
All patients were retrospectively planned using the Pinnacle SmartArc auto-planning tool with further manual adjustments in plan objective and continuous optimization. The advanced auto-planning settings were created according to the planning method described by Balik et al.[9] One- or two- isocenter technique was used based on the separation between targets, ensuring a distance from the center of each target to the isocenter of <4 cm. Regarding the two-isocenter technique, the first isocenter was placed in the geometric center of the two targets that are closer to each other, while the second isocenter was placed at the center of the third target. Most plans were optimized using two to four arcs, one being coplanar and the remaining arcs being non-coplanar, per isocenter. In a few cases, where two targets were associated with one isocenter, two to three arcs were assigned to each target by optimizing one target at a time. Arc configurations were determined based on the planning guideline published by Clark et al.[5] and modified by the planner if necessary to further reduce normal brain dose. Three ring structures were created with maximum dose constraints to improve the dose fall-off.[5] For targets with a volume <0.5 cc, smaller ring structures, with a 0.2 cm, 0.5 cm and 1.5 cm expansion from GTV for inner, middle and outer control, respectively, were used. All plans were created using 6MV-FFF beam with high-definition multileaf collimator (HD-120 MLC) on an EDGE linear accelerator (Varian, Palo Alto, CA). All plans were normalized such that >99% of each tumor volume received the prescription dose. A grid resolution of 1-2 mm, depending on the target volume, was used for the final dose calculation using collapsed cone convolution algorithm with heterogeneity correction. For small targets (<0.5 cc), the final dose was calculated with a grid of 1 mm or 1.5 mm.
HyperArc plans
Eclipse HyperArc plans were generated for all patients using the same machine and beam energy as that for the SmartArc planning. The isocenter positioning followed the same approach described above. In HyperArc, the selection of the arc geometry is one full or half coplanar arc at a couch angle of 0° and three half non-coplanar arcs with couch angles of 45°, 315°, and 90° (or 270°) per isocenter. One coplanar arc and one to three non-coplanar arcs were selected per isocenter. The collimator angle for each arc was automatically optimized by the system. SRS normal tissue objective (NTO) and automatic low dose objective (ALDO) were selected in optimization. Maximum dose constraints and additional ring structures were not used in plan objectives. Final dose calculation was performed using AAA algorithm and a grid of 1.25 mm.
Plan Evaluation
RTOG conformity index (CI), Paddick gradient index (GI), NT V12Gy and 4.5 Gy isodose volume (V4.5Gy) of each target were used for plan evaluation. The RTOG CI is defined as CI=VRx/VTV, where VRx is the tissue volume receiving the prescription dose and VTV is the target volume. The GI is defined as GI=V50% Rx/VRx, where V50% Rx is the tissue volume receiving 50% of the prescription dose. NT V12Gy is obtained by subtracting the target volume from the volume of tissue receiving 12 Gy. Plan quality evaluation for both GK and VMAT plans were performed in MIM (MIM Software Inc., Cleveland, OH, USA).[8,9,11] The beam-on times of the VMAT plans were calculated from the MU and the dose rates.
Statistics
To evaluate the threshold target volume for VMAT-based SRS that produces better plan quality than GK, we compared RTOG CI and NT V12Gy between clinically treated GK plans and VMAT plans. More specifically, we compared GK plans to SmartArc and HyperArc plans separately. The Receiver Operating Characteristic (ROC) curve was employed for both VMAT plans to determine the threshold target volume. ROC curve is a statistical tool, which is commonly used for evaluating binary classification. It is obtained by comparing the predicted outcomes to the observed outcomes which we address below. A predicted positive condition was viewed as the measurement being less than the threshold value, while a predicted negative condition was viewed as the measurement being greater than the threshold value. The two true conditions, positive and negative, were defined as the GK plan quality being better than VMAT plan quality and the GK plan quality being no better than VMAT plan quality, respectively. Four possible outcomes, true positive (TP), false positive (FP), true negative (TN) and false negative (FN), were obtained when the true condition was compared to the predicted condition. One of four possible outcomes for each target was thus classified. True positive rate (sensitivity) is defined as TP/(TP+FN) and true negative rate (specificity) is defined as TN/(FP+TN). The ROC curve was obtained by plotting the true positive rate against the false positive rate (1 – specificity) for all possible threshold values varying from 0.1-10 cc.
The threshold value was selected from the point on the ROC curve with the minimum distance to the point (0, 1). Area under the Curve (AUC) represents the degree or measure of separability, which measures the two-dimensional area underneath the ROC curve. The Paired Wilcoxon signed rank test was used to compare the plan quality in GK vs. SmartArc and GK vs. HyperArc, with p<0.05 considered as statistically significant.
RESULTS
ROC curve analysis
The GK, SmartArc and HyperArc plans had an average of 100%, 99.62% and 99.73% target coverage, respectively. Figure 1a shows the distribution of RTOG CI against target volume for the GK, SmartArc and HyperArc plans. Certain targets were not included due to the inability to separate the isodose volumes from the two closely adjacent targets. A total of 71 targets are shown in Figure 1a. The ROC curve associated with RTOG CI for SmartArc and the data table for the selected thresholds were shown in Figure 1b. The threshold value of 0.4 cc was obtained with a sensitivity of 81.8%, specificity of 87.8%, positive predictive value of 75.0%, negative predictive value of 91.5%, and an accuracy of 85.9%, corresponding to the blue dot on the ROC curve. As shown in Figure 1b, specificity decreases as the target volume increases. The ROC curve associated with RTOG CI for HyperArc was plotted in Figure 1c. The threshold value of 0.2 cc was obtained with a sensitivity of 100.0%, specificity of 79.7%, positive predictive value of 35.0%, negative predictive value of 100.0%, and an accuracy of 81.7%, corresponding to the red dot on the ROC curve. The insert of Figure 1a showed an enlarged view of target volumes between 0 and 1 cc with the threshold volumes for SmartArc (0.4 cc) and HyperArc (0.2 cc) indicated by the blue and red dotted lines, respectively.
Figure 1.
a) RTOG CI as a function of target volume for GK, SmartArc and HyperArc plans. The insert shows an enlarged view of target volumes between 0 and 1 with cut-off volumes of 0.4 cc (SmartArc) and 0.2 cc (HyperArc) using GK as a baseline. ROC curve associated with RTOG CI and a table of statistics for selected thresholds for b) SmartArc and c) HyperArc. The dots on the ROC curves correspond to the cutoff target volumes.
Figure 2a shows the distribution of NT V12Gy against target volume for GK, SmartArc and HyperArc plans. A total of 60 targets were presented due to the inability to separate the 12 Gy isodose volumes from two closely adjacent targets. Targets having prescriptions less than 12 Gy were excluded from this analysis. The threshold value performed by the ROC curve analysis associated with NT V12Gy for SmartArc was 0.4 cc, with a sensitivity of 79.2%, specificity of 94.4%, positive predictive value of 90.5%, negative predictive value of 87.2%, and an accuracy of 88.3%, as seen in Figure 2b. The ROC curve analysis for HyperArc plans gave a threshold value of 0.5 cc with a sensitivity of 89.5%, specificity of 82.9%, positive predictive value of 70.8%, negative predictive value of 94.4%, and an accuracy of 85.0%, as shown in Figure 2c.
Figure 2.
a) NT V12Gy as a function of target volume for GK, SmartArc and HyperArc plans. The insert shows an enlarged view of target volumes between 0 and 1 with cut-off volumes of 0.4 cc (SmartArc) and 0.5 cc (HyperArc). ROC curve associated with NT V12Gy and a table of statistics for selected thresholds for b) SmartArc and c) HyperArc. The dots on the ROC curves correspond to the cutoff target volumes.
Dose distribution comparison
Figure 3 shows representative isodose distributions for a patient (#1) with two targets. Both Target 1 (6.42 cc) and Target 2 (0.75 cc) had a prescription dose of 13 Gy. The center-to-center distance between Target 1 and Target 2 was 10.47 cm. A two-isocenter technique was used where the isocenter was placed at the center of each target. Both VMAT plans for both targets showed more conformal dose distributions, particularly in the high dose region. As shown in Table 2, VMAT plans produced improved RTOG CI and lower NT V12Gy for both targets. Lower GI and better low-dose spilage were achieved in VMAT plans for Target 1, whereas for Target 2, which has a smaller target volume, the GI and V4.5Gy were slightly higher in VMAT plans compared to GK plans.
Figure 3.
Representative isodose distributions from GK, SmartArc and HyperArc plans for patient (#1) with two targets. Target volumes for Target 1 and Target 2 are 6.42 cc and 0.75 cc, respectively. Prescription is 13 Gy for both targets. Two-isocenter technique was used for the VMAT plans.
Table 2.
Plan quality endpoints of GK, SmartArc and HyperArc plans shown in Figure 3
| Target 1 | Target 2 | |||||
|---|---|---|---|---|---|---|
| GK | SmartArc | HayperArc | GK | SmartArc | HyperArc | |
| RTOG CI | 1.46 | 1.18 | 1.10 | 1.83 | 1.57 | 1.29 |
| NT V12Gy (cc) | 4.68 | 2.05 | 1.62 | 0.84 | 0.68 | 0.43 |
| GI | 2.92 | 2.43 | 2.75 | 2.77 | 3.31 | 3.48 |
| V4.5Gy (cc) | 46.2 | 30.59 | 34.43 | 6.24 | 6.72 | 6.33 |
Figure 4 shows representative isodose distributions for a patient (#19) with three targets. The prescription for all targets was 24 Gy. One isocenter was placed at the center of Target 1 (0.4 cc) and the other isocenter was placed at the geometric center of Target 2 (0.44 cc) and Target 3 (0.13 cc). For Target 1 and Target 2, the dose distributions in VMAT plans were more conformal than those for GK plans, which was reflected in the dosimetric metrics shown in Table 3. There, VMAT plans showed improved conformity and low-dose spillage for both targets. For Target 3, whose volume was smaller than the suggested threshold volume, GK had an advantage to SmartArc with improved conformity and normal brain sparing, whereas lower CI and NT V12Gy in HyperArc plans were achieved compared to those in GK.
Figure 4.
Representative isodose distributions of GK, SmartArc and HyperArc plans for patient (#19) with three targets. Target volumes for Target 1, Target 2 and Target 3 are 0.40 cc, 0.44 cc and 0.13 cc, respectively. Prescription was 24 Gy for all the targets. Two-isocenter technique was used for the VMAT plans, where Target 1 was associated with one isocenter and the other isocenter was placed at the geometric center of Target 2 and Target 3.
Table 3.
Plan quality endpoints of GK, SmartArc and HyperArc plans shown in Figure 4
| Target 1 | Target 2 | Target 3 | |||||||
|---|---|---|---|---|---|---|---|---|---|
| GK | SmartArc | HyperArc | GK | SmartArc | HyperArc | GK | SmartArc | HyperArc | |
| RTOG CI | 1.38 | 1.35 | 1.25 | 1.41 | 1.39 | 1.25 | 1.46 | 2.85 | 1.31 |
| NT V12Gy (cc) | 3.7 | 1.64 | 1.62 | 3.62 | 2.41 | 2.02 | 1.49 | 2.28 | 0.98 |
| GI | 7.45 | 3.78 | 3.24 | 6.55 | 4.70 | 3.67 | 8.53 | 6.51 | 5.76 |
| V4.5Gy (cc) | 17.2 | 10.47 | 9.28 | 16.70 | 15.94 | 11.81 | 7.29 | 16.26 | 7.60 |
Plan quality comparison
By averaging the threshold values of 0.4, 0.2, 0.4 and 0.5 cc obtained from our analyses, we proposed a threshold target volume of 0.4 cc for SRS using VMAT technique. Table 4 shows statistical comparison of plan quality metrics of GK vs. SmartArc and GK vs. HyperArc, which is to evaluate whether there exits statistically significant differences between GK and VMAT plans when using the proposed threshold volume of 0.4 cc. For tumor volumes ≤0.4 cc, all dosimetric endpoints considered in this study showed that GK plans were favored over SmartArc plans with p<0.05. For tumor volumes >0.4 cc, RTOG CI, NT V12Gy and V4.5Gy in SmartArc plans were lower than those of the GK plans while no statistically significant difference was observed in the median GI. HyperArc plans showed improved plan quality compared with GK plans for tumor volumes >0.4 cc, which is consistent with SmartArc plans. On the other hand, for tumor volumes ≤0.4 cc, HyperArc plans had higher low-dose spread while maintaining comparable conformity as well as NT V12Gy compared to GK.
Table 4.
Statistical comparisons of plan quality endpoints of GK vs. SmartArc and GK vs. HyperArc for this study†
| GK | SmartArc | p-value | GK | HyperArc | p-value | ||
|---|---|---|---|---|---|---|---|
| Target ≤ 0.4 (cc) | RTOG CI | 1.68 | 1.85 | 0.0013 | 1.68 | 1.46 | 0.032 |
| NT V 12Gy (cc) | 0.70 | 1.25 | 0.0019 | 0.7 | 0.79 | 0.1923 | |
| GI | 3.75 | 6.62 | 0.0041 | 3.75 | 4.96 | 0.0284 | |
| V 4.5Gy (cc) | 2.58 | 6.82 | 0.0050 | 2.58 | 6.26 | 0.0065 | |
| Target > 0.4 (cc) | RTOG CI | 1.59 | 1.36 | <0.0001 | 1.59 | 1.23 | <0.0001 |
| NT V 12Gy (cc) | 3.87 | 3.04 | <0.0001 | 4.06 | 3.57 | <0.0001 | |
| GI | 3.19 | 3.30 | 0.5297 | 3.19 | 3.30 | 0.2969 | |
| V 4.5Gy (cc) | 17.55 | 16.04 | 0.0152 | 17.55 | 13.52 | <0.0001 |
†Median values were shown.
The median beam-on time of GK plans was 56.25 minutes. The median beam-on time was 8.33 minutes for SmartArc plans and 13.58 for HyperArc plans, where both plans showed statistical differences compared to GK plans with p<0.0001.
DISCUSSION
This study suggests a threshold target volume for multiple-target Linac-based SRS considering GK versus VMAT technique. We proposed a threshold target volume of >0.4 cc, for which Linac-based SRS can produce clinically equivalent or better plan quality when compared with GK plans. The aim of this study was to evaluate the performance of GK and VMAT techniques regarding individual target volume but not designed to provide patient selection criteria in clinical planning. Our results were consistent with the results from a previous study reported by Liu et al.[8], where VMAT-based SRS achieved similar or improved CI and NT V12Gy compared with GK for a median tumor volume of 0.5 cc. The use of HD-120 MLC and FFF beams in our study, including a total 76 targets, may overcome the limitations of the previous treatment planning study by Kim et al., where 5-mm MLC leaves without FFF beams were used.[7] In their dosimetry comparison study, Paddick CIs for VMAT plans were lower than GK with no significant difference in GI for tumor volumes >0.5 cc. The focus of the study by Kim et al. was on vestibular schwannoma in contrast to a multiple-target study in the present work, where dose bridging between targets may increase the dose to the normal brain tissue due to the transmitted leakage of the MLC leaves. For multiple-target studies, Potrebko et al.[11] showed that when using a mean tumor volume 1.16 cc as a threshold volume between small and large targets, a significantly lower V12Gy for GK plans as compared to VMAT plans was observed in the small target group, while for the large target group, there was no significant difference in V12Gy. Vergalasova et al.[13] reported that VMAT plans resulted in better conformity when compared to GK plans for target diameters >1 cm. VMAT plans achieved similar CI as GK plans for a target diameter of <1 cm. A target diameter of 1 cm can be approximated to a target volume of 0.52 cc if a spherical tumor was assumed. It is difficult to compare results of the present study to the two studies mentioned because previous studies divided the targets into several target diameter bins or stratified the targets according to the mean target volume. In our study, the threshold volume of 0.4 cc, below which GK achieved better plan quality, was determined by the ROC curve. Potrebko et al.[11] and Vergalasova et al.[13] studied plans with a median of eight and seven targets per patient, respectively, whereas in the present study, we have a median of three targets per patient. With an increased number of targets, dose spillage to the normal tissue between targets due to MLC leakage may deteriorate plan quality. In our study, patients with mostly three lesions were selected since SRS in patients with >4 lesions still remains clinically debatable. We utilized CI and NT V12Gy for ROC curve analysis because SRS plan evaluation guidelines proposed by RTOG focused on target dose coverage and normal brain tissue sparing in the high dose region.[14-16]
GI was not an ideal metric for such analysis since GI is dependent on VRx. Instead, NT V12Gy, an absolute value, has been a predictor for radionecrosis following SRS treatment. [17-19] V4.5Gy was chosen to evaluate the low-dose spill to normal brain tissue. The suggested threshold target volume of 0.4 cc in the present study was determined based on ROC curve analysis, predicting the probability of a binary outcome. Hence, it is possible that VMAT can produce improved plan quality compared with GK below the threshold target volume for a specific case. Moreover, the threshold value obtained from the ROC curve could be affected by the analyzed target volume range. We reanalyzed by restricting the target volumes from 0.1-3 cc because the CI and NT V12Gy in GK plans were typically worse than those in VMAT plans for target volumes larger than this range. The threshold values of SmartArc plans were the same as 0.4 cc for both CI and NT V12Gy. For HyperArc plans, the threshold values were 0.2 cc and 0.4 cc for CI and NT V12Gy, respectively. The averaged threshold value based on the CI and NT V12Gy metrics remained the same as 0.4 cc.
When treating multiple targets using the single-isocenter technique, the increased sensitivity to geometric uncertainties, such as rotational errors, can be induced.[20-23] When the distance from the target to the nearest isocenter is increased, rotational errors may result in target under-dosage and compromised OAR sparing. Therefore, a two-isocenter technique was implemented in this study when the target is >4 cm from the isocenter. Keeping the target within 4 cm from the isocenter can be beneficial for smaller targets since the 2.5 mm width of MLC leaves are within 4 cm of the isocenter when the treatment is planned using a Linac equipped with Varian HD-120 MLC leaves.[24,25] An additional isocenter added to the treatment plan, may increase the delivery time. Previous studies using a single-isocenter technique had beam-on times ranging from 2.2 to 11.2 minutes.[6,8,10,11,13] Treating multiple targets using the VMAT technique for SRS showed greatly improved efficiency of delivery when compared to GK. It is worthy to note that our institution’s GK planning principle emphasizes conformity over shorter treatment times. However, while maintaining clinically accepted plan quality, large shot sizes are sometimes preferred for large targets to improve efficiency. This may result in a slightly degraded CI. Consequently, one limitation of this retrospective study is that clinical GK plans were created in a time constrained manner and without inverse optimization. In contrast, VMAT plans were generated with intent to achieve similar or better plan quality to GK without clinical time constraints. Furthermore, VMAT plans were created using automatic planning algorithm with inverse planning. It is possible to further improve the quality of GK plans if there is no time constraint on planning and treatment delivery. Typically, a margin of 1-2 mm from the GTV was suggested in frameless VMAT-based SRS for multiple brain metastases to ensure target coverage due to spatial uncertainties.[21,25,26] No planning margin was used in this work in order to compare the results between GK and VMAT with the same volumes. Our clinical experience for multiple brain metastases SRS is mainly from GK. To retrospectively compare VMAT-based SRS plans with GK plans, there are limitations as in any retrospective studies.
Specific beam arrangements used in the present study differed from those used in previous studies due to the fact that arc configurations varied among studies. It is reported that a plan with increasing number of arcs can typically generate better plan quality than 1- and 2-arc plans.[6,27] When optimizing multiple targets using a single isocenter technique, we found that increasing the number of arcs by optimizing one target at a time (i.e., each target was associated with a set of arcs) can alleviate dose spillage in the normal tissue among the targets. In some cases, we reduced the arc lengths and the number of arcs that were associated with each target, depending on the location of the target. For example, when a two-isocenter technique was implemented in the treatment plan, the number of arcs on the contralateral side of the target was reduced to avoid the exposure to the targets on the opposite side. Individual planner skills represent a minor limitation in the present study.
Previous studies have shown that the use of HyperArc could improve the optimization efficiency and provide better dosimetric performance compared to manual VMAT.[13,28,29] We found that plans planned using HyperArc can indeed achieve a lower CI and comparable NT V12Gy to GK plans for targets ≤0.4 cc, which is similar to the result presented by Vergalasova et al.[13] However, the purpose of this study was not to directly compare the results between SmartArc and HyperArc SRS plans. It is beyond the scope of this paper due to the discrepancies between beam arrangements for the VMAT plans between SmartArc and HyperArc. Rather, it is to compare SmartArc and HyperArc plan results individually to GK plans for this study. Numerous variables may be introduced for VMAT-based SRS planning. Different treatment planning algorithms used in different treatment planning systems may produce different results. More dosimetric studies on comparisons among various treatment planning systems as well as planning techniques are required to evaluate plans delivered with VMAT technique regarding the size, number and location of the targets when treating multiple brain metastases.
CONCLUSIONS
VMAT may be considered a viable alternative treatment method to GK for SRS of multiple brain metastases when target volumes are greater than 0.4 cc. Future studies are desired to explore the limitations of VMAT when different treatment planning systems are used.
ACKNOWLEDGMENTS
Funding
This research is supported by a research grant from Advanced Oncotherapy.
Authors’ declaration of potential conflicts of interest
Dr. Neyman reports personal fees from Elekta AB, outside the submitted work. Dr. Chao reports personal fees from Varian Medical Systems, outside the submitted work. Dr. Suh reports personal fees from Philips, other from Neutron Therapeutics, personal fees from Novocure, outside the submitted work. Dr. Xia reports grants from AVO, during the conduct of the study; grants from Philips, outside the submitted work. Other authors have nothing to disclose.
Author contributions
Conception and design: Chieh-Wen Liu, Ping Xia
Data collection: Tianjun Ma, Gennady Neyman
Data analysis and interpretation: Chieh-Wen Liu, Saeed Ahmed
Manuscript writing: Chieh-Wen Liu, Saeed Ahmed, Tara Gray, Young-Bin Cho, Ping Xia
Final approval of manuscript: Samuel Chao, John Suh, Ping Xia
REFERENCES
- 1.Tsao M, Xu W, Sahgal A. A meta-analysis evaluating stereotactic radiosurgery, whole-brain radiotherapy, or both for patients presenting with a limited number of brain metastases. Cancer. 2012;118(9):2486-2493. [DOI] [PubMed] [Google Scholar]
- 2.Soliman H, Das S, Larson DA, Sahgal A. Stereotactic radiosurgery (SRS) in the modern management of patients with brain metastases. Oncotarget. 2016;7(11):12318-12330. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 3.Mayo CS, Ding L, Addesa A, Kadish S, Fitzgerald TJ, Moser R. Initial experience with volumetric IMRT (RapidArc) for intracranial stereotactic radiosurgery. Int J Radiat Oncol Biol Phys. 2010;78(5):1457-1466. [DOI] [PubMed] [Google Scholar]
- 4.Wolff HA, Wagner DM, Christiansen H, Hess CF, Vorwerk H. Single fraction radiosurgery using Rapid Arc for treatment of intracranial targets. Radiat Oncol (London, England). 2010;5:77. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 5.Clark GM, Popple RA, Prendergast BM, Spencer SA, Thomas EM, Stewart JG, Guthrie BL, Markert JM, Fiveash JB. Plan quality and treatment planning technique for single isocenter cranial radiosurgery with volumetric modulated arc therapy. Pract Radiat Oncol. 2012;2(4):306-313. [DOI] [PubMed] [Google Scholar]
- 6.Thomas EM, Popple RA, Wu X, Clark GM, Markert JM, Guthrie BL, Yuan Y, Dobelbower MC, Spencer SA, Fiveash JB. Comparison of plan quality and delivery time between volumetric arc therapy (RapidArc) and Gamma Knife radiosurgery for multiple cranial metastases. Neurosurgery. 2014;75(4):409-417; discussion 417-408. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 7.Kim H, Potrebko P, Rivera A, Liu H, Eldredge-Hindy HB, Gunn V, Werner-Wasik M, Andrews DW, Evans JJ, Farrell CJ, Judy K, Shi W. Tumor volume threshold for achieving improved conformity in VMAT and Gamma Knife stereotactic radiosurgery for vestibular schwannoma. Radiother Oncol. 2015;115(2):229-234. [DOI] [PubMed] [Google Scholar]
- 8.Liu H, Andrews DW, Evans JJ, Werner-Wasik M, Yu Y, Dicker AP, Shi W. Plan quality and treatment efficiency for radiosurgery to multiple brain metastases: Non-coplanar RapidArc vs. Gamma Knife. Frontiers Oncol. 2016;6:26. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 9.Balik S, Chao S, Neyman G. Gamma Knife and volumetric modulated arc therapy stereotactic radiosurgery plan quality and OAR sparing comparison for pituitary adenomas and vestibular schwannomas. J Radiosurg SBRT. 2018;5(3):237-247. [PMC free article] [PubMed] [Google Scholar]
- 10.Zhang I, Antone J, Li J, Saha S, Riegel AC, Vijeh L, Lauritano J, Marrero M, Salas S, Schulder M, Zinkin H, Goenka A, Knisely J. Hippocampal-sparing and target volume coverage in treating 3 to 10 brain metastases: A comparison of Gamma Knife, single-isocenter VMAT, CyberKnife, and TomoTherapy stereotactic radiosurgery. Pract Radiat Oncol. 2017;7(3):183-189. [DOI] [PubMed] [Google Scholar]
- 11.Potrebko PS, Keller A, All S, Sejpal S, Pepe J, Saigal K, Kandula S, Sensakovic WF, Shridhar R, Poleszczuk J, Biagioli M. GammaKnife versus VMAT radiosurgery plan quality for many brain metastases. J Appl Clin Med Phys. 2018;19(6):159-165. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 12.Huss M, Barsoum P, Dodoo E, Sinclair G, Toma-Dasu I. Fractionated SRT using VMAT and Gamma Knife for brain metastases and gliomas--a planning study. J Appl Clin Med Phys. 2015;16(6):3-16. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 13.Vergalasova I, Liu H, Alonso-Basanta M, Dong L, Li J, Nie K, Shi W, Teo BK, Yu Y, Yue NJ, Zou W, Li T. Multi-Institutional Dosimetric Evaluation of Modern Day Stereotactic Radiosurgery (SRS) Treatment Options for Multiple Brain Metastases. Frontiers Oncol. 2019;9:483. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 14.Shaw E, Kline R, Gillin M, Souhami L, Hirschfeld A, Dinapoli R, Martin L. Radiation therapy oncology group: Radiosurgery quality assurance guidelines. Int J Radiat Oncol Biol Phys. 1993;27(5):1231-1239. [DOI] [PubMed] [Google Scholar]
- 15.Shaw E, Scott C, Souhami L, Dinapoli R, Bahary J-P, Kline R, Wharam M, Schultz C, Davey P, Loeffler J, Del Rowe J, Marks L, Fisher B, Shin K. Radiosurgery for the treatment of previously irradiated recurrent primary brain tumors and brain metastases: Initial report of radiation therapy oncology group protocol 90-05.. Int J Radiat Oncol Biol Phys. 1996;34(3):647-654. [DOI] [PubMed] [Google Scholar]
- 16.Souhami L, Seiferheld W, Brachman D, Podgorsak EB, Werner-Wasik M, Lustig R, Schultz CJ, Sause W, Okunieff P, Buckner J, Zamorano L, Mehta MP, Curran WJ. Randomized comparison of stereotactic radiosurgery followed by conventional radiotherapy with carmustine to conventional radiotherapy with carmustine for patients with glioblastoma multiforme: Report of Radiation Therapy Oncology Group 93-05 protocol.. Int J Radiat Oncol Biol Phys. 2004;60(3):853-860. [DOI] [PubMed] [Google Scholar]
- 17.Flickinger JC, Kondziolka D, Lunsford LD, Kassam A, Phuong LK, Liscak R, Pollock B. Development of a model to predict permanent symptomatic postradiosurgery injury for arteriovenous malformation patients. Arteriovenous Malformation Radiosurgery Study Group. Int J Radiat Oncol Biol Phys. 2000;46(5):1143-1148. [DOI] [PubMed] [Google Scholar]
- 18.Blonigen BJ, Steinmetz RD, Levin L, Lamba MA, Warnick RE, Breneman JC. Irradiated volume as a predictor of brain radionecrosis after linear accelerator stereotactic radiosurgery. Int J Radiat Oncol Biol Phys. 2010;77(4):996-1001. [DOI] [PubMed] [Google Scholar]
- 19.Minniti G, Clarke E, Lanzetta G, Osti MF, Trasimeni G, Bozzao A, Romano A, Enrici RM. Stereotactic radiosurgery for brain metastases: analysis of outcome and risk of brain radionecrosis. Radiat Oncol (London, England). 2011;6:48. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 20.Guckenberger M, Roesch J, Baier K, Sweeney RA, Flentje M. Dosimetric consequences of translational and rotational errors in frame-less image-guided radiosurgery. Radiat Oncol (London, England). 2012;7:63. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 21.Winey B, Bussiere M. Geometric and dosimetric uncertainties in intracranial stereotatctic treatments for multiple nonisocentric lesions. J Appl Clin Med Phys. 2014;15(3):122-132. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 22.Roper J, Chanyavanich V, Betzel G, Switchenko J, Dhabaan A. Single-Isocenter Multiple-Target Stereotactic Radiosurgery: Risk of Compromised Coverage. Int J Radiat Oncol Biol Phys. 2015;93(3):540-546. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 23.Prentou G, Pappas EP, Logothetis A, Koutsouveli E, Pantelis E, Papagiannis P, Karaiskos P. Dosimetric impact of rotational errors on the quality of VMAT-SRS for multiple brain metastases: Comparison between single- and two-isocenter treatment planning techniques. J Appl Clin Med Phys. 2020;21(3):32-44. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 24.Dhabaan A, Elder E, Schreibmann E, Crocker I, Curran WJ, Oyesiku NM, Shu HK, Fox T. Dosimetric performance of the new high-definition multileaf collimator for intracranial stereotactic radiosurgery. J Appl Clin Med Phys. 2010;11(3):3040. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 25.Stanhope C, Chang Z, Wang Z, Yin FF, Kim G, Salama JK, Kirkpatrick J, Adamson J. Physics considerations for single-isocenter, volumetric modulated arc radiosurgery for treatment of multiple intracranial. Pract Radiat Oncol. 2016;6(3):207-213. [DOI] [PubMed] [Google Scholar]
- 26.Briscoe M, Voroney J-P, Ploquin N. Establishing a threshold for rotational patient setup errors in linear accelerator-based stereotactic radiosurgery. Biomed Phys Eng Express. 2016;2(4):045018. [Google Scholar]
- 27.Abisheva Z, Floyd SR, Salama JK, Kirkpatrick J, Yin FF, Moravan MJ, Giles W, Adamson J. The effect of MLC leaf width in single-isocenter multi-target radiosurgery with volumetric modulated arc therapy. J Radiosurg SBRT. 2019;6(2):131-138. [PMC free article] [PubMed] [Google Scholar]
- 28.Ohira S, Ueda Y, Akino Y, Hashimoto M, Masaoka A, Hirata T, Miyazaki M, Koizumi M, Teshima T. HyperArc VMAT planning for single and multiple brain metastases stereotactic radiosurgery: a new treatment planning approach. Radiat Oncol (London, England). 2018;13(1):13-13. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 29.Slosarek K, Bekman B, Wendykier J, Grządziel A, Fogliata A, Cozzi L. In silico assessment of the dosimetric quality of a novel, automated radiation treatment planning strategy for linac-based radiosurgery of multiple brain metastases and a comparison with robotic methods. Radiat Oncol (London, England). 2018;13(1):41-41. [DOI] [PMC free article] [PubMed] [Google Scholar]