Skip to main content
NCBI home page
Journal of Radiosurgery and SBRT logoLink to Journal of Radiosurgery and SBRT
. 2016;4(3):235–243.

Dosimetric comparison of intracranial metastasis treatment using two radiosurgery systems: TrueBeam STx with VMAT and Gamma Knife Model 4C

Vance Keeling 1, Ozer Algan 1, Salahuddin Ahmad 1, Sabbir Hossain 1,✉
PMCID: PMC5658806  PMID: 29296448

Abstract

We compared treatment plan quality based on target coverage and normal brain tissue sparing for two intracranial stereotactic radiosurgery systems: TrueBeam STx using VMAT and Gamma Knife (GK). Ten patients with 24 tumors (seven with 1-2 and three with 4-6 ranging from 0.1 to 20.2 cc), previously treated with GK Model 4C (prescription doses ranging from 14-23 Gy), were re-planned for VMAT using Eclipse treatment planning system. Various photon beam energies and MLC leaf widths with and without jaw tracking were studied to achieve optimal plans. Plan qualities were assessed by target coverages using Paddick Conformity Index (PCI), normal-brain-tissue integral dose (Gy-cc) and sparing. In all cases critical structure dose criteria were met. The average PCI was 0.76±0.21 for VMAT and 0.46±0.20 for GK plans (p≤0.001), respectively. On average 81% reduction of 12 Gy normal-brain-tissue volumes was achieved by VMAT. The average integral dose ratio of GK to VMAT plans was 1.50±0.61 (p=0.006). VMAT was capable of producing higher quality treatment plans in terms of target coverage and normal brain tissue sparing than GK while using optimal beam geometries and optimization techniques.

Keywords: RapidArc, VMAT, Gamma Knife, brain metastases, beam optimization

1. Introduction

The use of stereotactic radiosurgery or stereotactic radiotherapy (SRS/SRT) has been a common modality in the treatment of multiple intracranial metastasis for many years [1-4]. Numerous platforms such as Gamma Knife (GK) (Elekta, Atlanta, Georgia), Cyberknife (Accuray, Sunnyvale, California), Truebeam STx, and Novalis Tx (Varian, Palo Alto, California) are clinically used for treatments. GK is usually considered to be the gold standard of treatment in terms of sparing normal healthy tissue. However, GK has certain disadvantages including a need for invasive immobilization, difficulty in treating certain locations in the brain, and longer treatment times than LINAC based SRS systems. The implementation of Flattening Filter Free (FFF) beams [5,6] on LINACs has allowed increased dose rates, reduced scatter, leakage, and out of field doses [7-9]. Several studies comparing LINAC based systems to GK have been performed. In one study, the GK Perfexion was shown to have superior conformity and significantly lower normal brain tissue doses compared to CyberKnife and Novalis for multiple metastasis treatment [10]. Sio et. al. [11] concluded that CyberKnife achieved superior conformity, however, GK Model 4C (GK-4C) plans had sharper peripheral dose falloff in majority of cases. With the implementation of VMAT technology such as RapidArc (Varian, Palo Alto, California) [12,13] much faster treatment times can be achieved with higher conformity and less peripheral dose compared to other LINAC based delivery systems. Jaw tracking technology [14,15] implemented on the TrueBeam STx leads to further reduction in normal brain tissue doses.

The goal of this study is the dosimetric comparison of GK-4C and the optimal VMAT (RapidArc) plans based on normal brain tissue dose, conformity, integral dose, and dose falloff. A recent study comparing single isocenter VMAT and GK-4C systems [16] showed that VMAT could achieve equivalent plan quality based on conformity and normal brain tissue dose while treating multiple metastases. We evaluated plan quality of RapidArc plans for energies (6MV, 6MV-FFF, 8, 10, and 10MV-FFF), MLC widths (2.5mm vs 5mm), and Jaw Tracking vs No Jaw Tracking during dosimetric comparisons.

2. Methods and materials

Ten patients with 24 brain metastasis (seven with 1-2 and three with 4-6 lesions) previously treated with GK Model 4C were selected. All patients underwent treatment planning MRI (1.5T General Electric Excite MRI, USA) utilizing T1, T2, T1+C, and thin slice SPGR+C sequences. The gross tumor volume (GTV) was used as the planning target volume. No uniform expansion was used around the GTV. The complexity of the plans varied with some plans having the tumor close to critical structures such as optic chiasm and brainstem and other plans being much simpler with tumor volumes far from critical structures. The average tumor volumes and their standard deviation were 3.5±5.3cc ranging from 0.10 to 20.20cc. GK plans were generated using GammaPlan version 10. Prescription dose ranged from 14 to 22 Gy with a majority being prescribed to the 50% IDL. Almost all of the GK plans utilized more than one isocenter, in an attempt to conform the isodose distribution to the geometry of the tumors. The plans were optimized to achieve at least 99% target coverage, while trying to maintain the highest selectivity. Larger isocenters were used initially to achieve tumor coverage, and regions that were not covered were filled-in using smaller isocenters. Beam on time and gradient index were monitored during the planning process, reviewed in the final plan evaluation, and further optimization was performed as indicated. Depending on the proximity of the target volume to critical structures, plugs were used to shape the isodose distribution and further limit the dose to critical structures. For VMAT plans, a single isocenter was used for seven patients and two iscocenters were used for three. The isocenter was generated using Arc Geometry Tool in Eclipse (Varian, Palo Alto, California). For multiple target plans, Boolean function was used to combine targets into one structure to determine isocenter in Arc Geometry Tool. Plans were generated using 6MV-FFF beams (2 to 10 arcs) at a dose rate of 1400 MU/minute for the Varian TrueBeam STx machine utilizing high definition MLCs, 2.5mm in width in the inner 8cm of the leaf carriage. Plans were calculated on a 1mm grid size using AAA algorithm. All targets were prescribed to cover at least 99% of each individual target volume.

Plans with more than 2 arcs were accomplished with half rotation arcs using non-coplanar beams with couch angles of either 15°, 30°, 45°, 345°, 330°, or 315°. The beam angle selections were based on target depths. Areas that penetrated more than 10 cm of tissues were avoided either by using partial arcs or creating avoidance sectors in optimization.

Plans were optimized using jaw tracking [14] to prevent dose to the normal brain due to leakage and transmission through the MLC leaves utilizing Normal Tissue Objective (NTO) to achieve optimal dose falloff and to prevent streaking of the dose. In plans with multiple metastases the goal in optimization was to achieve similar target coverage. Setting no upper limit objective for the PTV allowed more freedom in optimization producing better normal brain tissue sparing and dose fall-off around the target and led to maximum dose as high as 150% of the prescription dose located within the PTV. If needed, a maximum dose objective (systematically lowered and priorities raised) was placed on the optic chiasm and brainstem until the dosimetric objectives were met. The PTV subtracted from the brain (BrainsubPTV) objective was placed at the inflection point of the curve which was usually around 200 – 800cGy depending upon the prescription. The priority of BrainsubPTV was determined by slowly increasing it until target coverage was compromised. Optimization was always paused in the first multi-resolution steps until the optimization curves reached their minimum values. Evaluations and comparisons of plans were accomplished through Paddick Conformity Index (PCI) [17], integral dose (Gy-cc), number of Monitor Units (MUs), normal-brain-tissue in absolute volume (cc) for the following criteria (V4, V8, V12, and V16 Gy, where Vx Gy indicates the amount of volume in cc receiving x Gy dose), and maximum and mean dose for brainstem and optic chiasm.

The Paddick conformity index was defined as follows [18]:

PCI = TVPV2 / (TV × PIV),

where PIV is the prescription isodose volume, TVPV is the target volume within the prescribed isodose surface, and TV is the target volume. The Radiation Therapy Oncology Group (RTOG) conformity index (CI) defined as,

RTOG CI = PIV/TV.

A p-value ≥0.05 indicated statistically insignificant differences at the 95% confidence interval. The integral dose of the brain was determined from the DVH curve using Trapezoidal technique [18].

The 6MV-FFF RapidArc plans were then re-optimized and calculated with same objectives and beam geometry using 6MV, 8MV, 10MV, and 10MV-FFF beams and compared with 6MV-FFF reference beams. Plans were calculated on a Varian Trilogy with 5mm leaf width and compared to 2.5mm plans. Plans were also performed with Jaw Tracking turned off in optimization and compared to plans with Jaw Tracking.

3. Results

3.1. Paddick & RTOG Conformity Indices

In 23 out of 24 lesions RapidArc had a higher PCI than GK as shown in Figure 1(a). The average PCI was 0.76±0.21 for 6MV-FFF RapidArc and 0.46±0.20 for GK plans, where the differences were statistically significant (p ≤0.001). RTOG conformity indices were higher in 22 out of 24 lesions in GK than RapidArc shown in Figure 1(b). The average RTOG CI was 2.56±1.47 for 6MV-FFF RapidArc and 1.49±0.78 for GK plans (p ≤0.003).

Figure 1(a).

Figure 1(a)

Open in a new tab

PCI for Gamma Knife 4C and RapidArc plans for all 24 tumors.

Figure 1(b).

Figure 1(b)

Open in a new tab

RTOG Conformity Index for Gamma Knife 4C and RapidArc plans for all 24 tumors.

3.2. Normal Brain Tissue Dose

Normal Brain Tissue doses for all measured criteria were lower for RapidArc plans in all patients (except patient #4) shown in Figure 2. The normal-brain-tissue volumes in GK plans were approximately 28%, 31%, and 81% higher than RapidArc plans for V4, V8, and V12 Gy, respectively. The ratio GK/RapidArc for normal brain was large for V16 Gy, but for absolute volume the differences were minimal.

Figure 2.

Figure 2

Open in a new tab

Normal brain tissue volume (cc) receiving 4, 8, 12, and 16 Gy for each patient for the Gamma Knife 4C and TrueBeam STx RapidArc plans.

3.3. Integral Dose

The integral dose (Gy-cc) for normal brain was higher (9 of 10) GK plans to RapidArc plans (Figure 3) and the differences were statistically significant (p=0.01). On average the integral dose was 50% higher for GK plans with standard deviation of 0.61. For one patient integral dose was 7% lower for GK.

Figure 3.

Figure 3

Open in a new tab

Comparison of Integral Dose (Gy-cc) for Gamma Knife and RapidArc plans.

3.4. Energy Dependence and Jaw Tracking

The average PCI±SD was 0.76±0.21 and 0.76±0.22 for 6MV-FFF and 10MV-FFF, respectively. Statistically significant differences in normal-brain-tissue for V4, V8, and V12 Gy were observed (Table 1) in all cases for different energies (p-values <0.05). A 12% volume reduction for V4 Gy could be achieved using 6MV-FFF compared to 10MV-FFF arcs. Brain tissue sparing from best to worst shown in Table 1 occurred in this order 6MV-FFF, 6MV-FFF no-JT, 10MV-FFF, 8MV, and 10MV. Table 1 also shows that the integral dose increased with increasing energy and decreased when FFF arcs were used. The integral dose for 10MV-FFF plans were on average 4% higher than 6MV-FFF plans.

Table 1.

Ratio of brain tissue for different energies and Jaw Tracking/No Jaw Tracking for V4 (4), V8 (8), V12 (12), V16 (16), and V20 (20) Gy, and average ratio of integral doses of all 10 patients for various beam energies.

Ratio of Brain Tissue
  4Gy 8Gy 12Gy 16Gy 20Gy
No Jaw Tracking/Jaw Tracking
Average 1.04 1.03 1.03 1.00 1.13
Stdev 0.03 0.03 0.04 0.07 0.30
p-value 0.04 0.19 0.22 0.35 0.15
8x/6xFFF
Average 1.12 1.12 1.14 1.04 1.10
Stdev 0.07 0.07 0.04 0.15 0.32
p-value 0.01 0.00 0.00 0.11 0.17
10x/FFF/6xFFF
Average 1.12 1.12 1.13 1.02 0.98
Stdev 0.08 0.09 0.06 0.17 0.32
p-value 0.02 0.01 0.00 0.14 0.23
10xFFF/6xFFF
Average 1.19 1.18 1.20 1.07 1.07
Stdev 0.10 0.10 0.04 0.19 0.21
p-value 0.02 0.00 0.00 0.08 0.14
Ration of Integral Dose
  8x/6xFFF 10x/FFF/6xFFF 10xFFF/6xFFF
Average 1.05 1.04 1.09
Stdev 0.04 0.05 0.06
p-value 0.21 0.33 0.04
Min 0.94 0.92 0.96
Max 1.09 1.08 1.19
Open in a new tab

3.5. MLC Leaf Width Dependence

In all cases, HD MLC plans performed better in sparing normal-brain-tissue, achieving a higher PCI with lower integral dose. The average PCI for all 24 targets was 0.75±0.23 and 0.70±0.23 (p≤0.0015) for HD-MLC and Millennium-MLC plans, respectively. The average ratio of normal-brain-doses for Millennium-MLC to HD-MLC plans was 1.30±0.16, 1.27±0.15, and 1.31±0.18 for the V4, V8, and V12, respectively. Figure 4 shows the volume (cc) for each patient for the 4, 8, and 12 Gy isodose lines. The differences in normal brain dose for all criteria were statistically significant (p≤0.05). Additionally, Millennium-MLC plans had a 16% higher integral dose than HD-MLC plans (Figure 5).

Figure 4.

Figure 4

Open in a new tab

Normal brain tissue volume (cc) receiving 4, 8, and 12 Gy for 2.5 and 5 mm leaf widths.

Figure 5.

Figure 5

Open in a new tab

Integral dose for each patient for 2.5 and 5 mm leaf widths.

4. Discussion

4.1. Paddick Conformity Index

The PCI was much higher for the RapidArc plans due to the ability of HD-MLCs to conform tumor volumes. GK-4C isn’t as proficient to conform irregular shaped targets due to its fixed circular shots. The conformity of the GK can be improved by filling tumor volume with many 4-mm circular shots at the cost of treatment time increase.

4.2. Normal Brain Tissue Dose (NBTD)

One of the main advantages of GK is field geometry used for treatment. Because of GK-4C’s ability to deliver 201 beams hemi-spherically around the head, normal brain tissue doses can be lowered. However, the geometry of GK can nearly be achieved on a LINAC by delivering multiple coplanar and non-coplanar arcs. Another key to lowering NBTD is proper beam selection. Reduction in NBTD was achieved in planning by using partial arcs and/or avoidance sectors to not deliver beam in areas where depth of beam penetration was high. In cases where there were multiple metastases spread throughout the brain we found it beneficial to group the metastases and use two isocenters.

The advantages of using a LINAC to spare normal-brain-tissue are due to its beam energy and penumbra size. The 6MV-FFF beam is more penetrative compared to Co-60 beam. For treatment planning 6MV-FFF beams were selected over 6MV because of higher dose rates, lower out of field doses [8,9,19], and smaller penumbra [19-21]. A disadvantage of LINAC MLC over GK-4C is due to interleaf and intra-leaf leakage through the MLC’s which was minimized using Jaw Tracking technique.

Optimal techniques (optimizing NTO and assigning higher priority to normal brain tissue structure over PTV), proper geometry, partial arcs and/or avoidance sectors are critical in reducing NBTD. For one of the 10 patients, GK spared NBT much better. This patient had a small circular tumor centered in the brain and under this condition GK performs better due to its superiority in beam geometry compared to LINAC.

This study showed statistically significant differences in normal brain tissue sparing for all criteria, in 9 out of 10 cases in terms of V4, V8, V12, and V16 Gy in favor of LINAC over GK. Another advantage to this type of planning is using more than one isocenter. Normal brain tissue can be spared more if groups of tumors can be treated separately. Furthermore, selection of 6MV-FFF beams was found to produce better results than 10MV-FFF beams in terms of normal-brain-tissue sparing and integral dose. This was likely caused by the fact that 6MV-FFF beams possessed a narrower penumbra than the 10MV-FFF beams. Leaf Transmission and interleaf leakage are also lower for 6MV-FFF beams than 10MV FFF beams.

There are certain limitations to this type of analysis, as it is difficult to compare a treatment plan that is generated under time and patient constraints, as were the Gamma Knife 4C plans, to those that are generated under less constraints for the purposes of research. Although, part of the differences seen in our analysis could be explained by a greater amount of time being spent on LINAC plans, this difference is also present in the clinical work flow for patient treatment planning. All of our GK 4C patients were treated in a single fraction, and required treatment imaging, planning and treatment delivery to be completed in a single day. On the LINAC side, there is generally several days from the completion of treatment planning simulation to the start of therapy. Some of the differences seen in selectivity and PCI also relates to the location of the tumor. On the GK 4C model there are certain location, especially in the region of the posterior-lateral base of skull, that were difficult to achieve, necessitating the use of a larger collimator to cover the tumor. The upgrades made to the GK Perfexion system, including a larger treatable volume, composite shots with sector blocking, and improved optimization alghorithms will likely result in higher PCI and selectivity when compared to the 4C system. The elimination of helmets and plugs allows for increased automation with increased conformity utilizing smaller shots or sectors without increasing the overall treatment time [22, 23].

The clinical significance of our findings depends on the size and location of the treated metastases, its proximity to dose limiting structures as well as prior treatments the patients have undergone. Studies have shown a correlation between the volume of NBT receiving 10 or 12 Gy and toxicity [3,4,24]. An effort was made to limit the size of the 10 Gy isodose line, and small compromises to target volume coverage may be necessary to limit critical doses to normal structures.

5. Conclusions

For the tumors evaluated in this study, significantly better dose conformity with reduced volume of normal-brain-tissue and integral dose was achieved with HD MLC plans compared to Millennium MLC plans. Normal-brain-tissue and integral dose improved using the lower energy and FFF beams, however plan conformity showed to be independent of energy. Jaw Tracking was shown to be able to effectively lower normal brain tissue dose. It was found that VMAT was superior to GK-4C in terms of conformity, normal brain tissue sparing, and integral dose. LINAC based single/double isocenter VMAT is a viable option for the treatment of single/multiple brain metastasis while delivering fast treatments without invasive immobilization.

Footnotes

Authors’ disclosure of potential conflicts of interest

The authors have no conflicts of interest to report.

Author contributions

Conception and design: Sabbir Hossain

Data analysis and interpretation: Vance Keeling, Ozer Algan, Salahuddin Ahmad, Sabbir Hossain

Manuscript writing: Vance Keeling, Ozer Algan, Salahuddin Ahmad, Sabbir Hossain

Final approval of manuscript: Vance Keeling, Ozer Algan, Salahuddin Ahmad, Sabbir Hossain

References

  • 1.Schell MC, Bova FJ, Larson DA, et al. Stereotactic Radiosurgery. AAPM Report 54 1995. [Google Scholar]
  • 2.Lightstone AW, Benedict SH, Bova FJ, et al. Intracranial stereotactic positioning systems: Report of the American Association of Physicists in Medicine Radiation Therapy Committee Task Group No. 68. Medical Physics. 2005; 32 (7): 2380-98. [DOI] [PubMed] [Google Scholar]
  • 3.Chin LS, Ma L, DiBiase S. Radiation necrosis following gamma knife surgery: a case-controlled comparison of treatment parameters and long-term clinical follow up. Journal of Neurosurgery. 2001; 94(6): 899–904. [DOI] [PubMed] [Google Scholar]
  • 4.Korytko T, Radivoyevitch T, Colussi V, et al. 12 Gy gamma knife radiosurgical volume is a predictor for radiation necrosis in non-AVM intracranial tumors. Int. J. Radiat. Oncol. Biol. Phys. 2006; 64(2): 419–24. [DOI] [PubMed] [Google Scholar]
  • 5.Kragl G, af Wetterstedt S, Knäusl B, et al. Dosimetric characteristics of 6 and 10 MV unflattened photon beams. Radiotherapy and Oncology. 2009; 93(1): 141–6. [DOI] [PubMed] [Google Scholar]
  • 6.Hrbacek J, Lang S, Klöck S. Commissioning of Photon Beams of a Flattening Filter-Free Linear Accelerator and the Accuracy of Beam Modeling Using an Anisotropic Analytical Algorithm. Int. J. Radiat. Oncol. Biol. Phys. 2011; 80(4): 1228–37. [DOI] [PubMed] [Google Scholar]
  • 7.Jason C. The characterization of unflattened photon beams from a 6 MV linear accelerator. Physics in Medicine and Biology. 2008; 53(7): 1933. [DOI] [PubMed] [Google Scholar]
  • 8.Kragl G, Baier F, Lutz S, et al. Flattening filter free beams in SBRT and IMRT: Dosimetric assessment of peripheral doses. Zeitschrift für Medizinische Physik. 2011; 21(2): 91–101. [DOI] [PubMed] [Google Scholar]
  • 9.Titt U, Vassiliev ON, Pönisch F, Dong L, Liu H, Mohan R. A flattening filter free photon treatment concept evaluation with Monte Carlo. Medical Physics. 2006; 33(6): 1595–602. [DOI] [PubMed] [Google Scholar]
  • 10.Lijun Ma, Petti Paula, Wang Brian, Descovich Martina, Chuang Cynthia, Barani Igor J., et al. Apparatus dependence of normal brain tissue dose in stereotactic radiosurgery for multiple brain metastases. Journal of Neurosurgery. 2011; 114(6): 1580–4. [DOI] [PubMed] [Google Scholar]
  • 11.Sio TT, Jang S, Lee S-W, Curran B, Pyakuryal AP, Sternick ES. Comparing Gamma Knife and CyberKnife in patients with brain metastases. JACMP. 2014; 15(1): 4095. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 12.Otto K. Volumetric modulated arc therapy: IMRT in a single gantry arc. Medical Physics. 2008; 35(1): 310–7. [DOI] [PubMed] [Google Scholar]
  • 13.Kang J, Ford EC, Smith K, et al. A method for optimizing LINAC treatment geometry for volumetric modulated arc therapy of multiple brain metastases. Medical Physics. 2010; 37(8): 4146–54. [DOI] [PubMed] [Google Scholar]
  • 14.Kim J-i, Park JM, Park S-Y, et al. Assessment of potential jaw-tracking advantage using control point sequences of VMAT planning. JACMP. 2014; 15(2):160-168. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 15.Joy S, Starkschall G, Kry S, et al. Dosimetric effects of jaw tracking in step-and-shoot intensity-modulated radiation therapy. JACMP. 2012; 13(2). [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 16.Thomas EM, Popple RA, Wu X, et al. Comparison of Plan Quality and Delivery Time Between Volumetric Arc Therapy (RapidArc) and Gamma Knife Radiosurgery for Multiple Cranial Metastases. Neurosurgery. 2014; 75(4): 409–18. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 17.Paddick Ian. A simple scoring ratio to index the conformity of radiosurgical treatment plans. Journal of Neurosurgery. 2000; 93(supplement 3): 219-22. [DOI] [PubMed] [Google Scholar]
  • 18. http://www.mathwords.com/t/trapezoid_rule.htm. Trapezoid Rule [updated 7/28/2014].
  • 19.Georg D, Knöös T, McClean B. Current status and future perspective of flattening filter free photon beamsa). Medical Physics. 2011; 38(3): 1280–93. [DOI] [PubMed] [Google Scholar]
  • 20.Yarahmadi M, Allahverdi M, Nedaie HA, et al. Improvement of the penumbra for small radiosurgical fields using flattening filter free low megavoltage beams. Zeitschrift für Medizinische Physik. 2013; 23(4): 291–9. [DOI] [PubMed] [Google Scholar]
  • 21.Pönisch F, Titt U, Vassiliev ON, et al. Properties of unflattened photon beams shaped by a multileaf collimator. Medical Physics. 2006; 33(6): 1738–46. [DOI] [PubMed] [Google Scholar]
  • 22.Ma L, Nichol A, Hossain S, et al. Variable dose interplay effects across radiosurgical apparatus in treating multiple brain metastases, International Journal of Computer Assisted Radiology and Surgery, 2014, 014-1001-4. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 23.Kendall E, Algan O, Ahmad S, Chen Y, Hossain S. Treatment plan quality comparison of Gamma Knife 4C and Perfexion for brain metastatic tumors and meningioma, 18th International LekSell Gamma Knife Society Meeting, Amsterdam, 2016 May 15-19.
  • 24.Lawrence YR, Li XA, el Naqa I, et al. Radiation dose-volume effects in the brain. Int J Radiat Oncol Biol Phys. 2010; 76(3 suppl): s20-s27. [DOI] [PMC free article] [PubMed] [Google Scholar]

Articles from Journal of Radiosurgery and SBRT are provided here courtesy of Old City Publishing

RESOURCES