WHOLE-BRAIN RADIOTHERAPY WITH SIMULTANEOUS INTEGRATED BOOST TO MULTIPLE BRAIN METASTASES USING VOLUMETRIC MODULATED ARC THERAPY

doi:10.1016/j.ijrobp.2009.03.029 Int. J. Radiation Oncology Biol. Phys., Vol. 75, No. 1, pp. 253 259, 2009 Copyright Ó 2009 Elsevier Inc. Printed in the USA. All rights reserved 0360-3016/09/$ see front matter PHYSICS CONTRIBUTION WHOLE-BRAIN RADIOTHERAPY WITH SIMULTANEOUS INTEGRATED BOOST TO MULTIPLE BRAIN METASTASES USING VOLUMETRIC MODULATED ARC THERAPY FRANK J. LAGERWAARD, M.D., PH.D., ELLES A. P. VAN DER HOORN, WILKO F. A. R. VERBAKEL, PH.D., CORNELIS J. A. HAASBEEK, M.D., BEN J. SLOTMAN, M.D., PH.D., AND SURESH SENAN, M.R.C.P., F.R.C.R., PH.D. Department of Radiation Oncology, VU University Medical Center, Amsterdam, The Netherlands Purpose: Volumetric modulated arc therapy (RapidArc [RA]; Varian Medical Systems, Palo Alto, CA) allows for the generation of intensity-modulated dose distributions by use of a single gantry rotation. We used RA to plan and deliver whole-brain radiotherapy (WBRT) with a simultaneous integrated boost in patients with multiple brain metastases. Methods and Materials: Composite RA plans were generated for 8 patients, consisting of WBRT (20 Gy in 5 fractions) with an integrated boost, also 20 Gy in 5 fractions, to the brain metastases, and clinically delivered in 3 patients. Summated gross tumor volumes were 1.0 to 37.5 cm 3. RA plans were measured in a solid water phantom by use of Gafchromic films (International Specialty Products, Wayne, NJ). Results: Composite RA plans could be generated within 1 hour. Two arcs were needed to deliver the mean of 1,600 monitor units with a mean beam-on time of 180 seconds. RA plans showed excellent coverage of planning target volume for WBRT and planning target volume for the boost, with mean volumes receiving at least 95% of the prescribed dose of 100% and 99.8%, respectively. The mean conformity index was 1.36. Composite plans showed much steeper dose gradients outside the brain metastases than plans with a conventional summation of WBRT and radiosurgery. Comparison of calculated and measured doses showed a mean gamma for double-arc plans of 0.30, and the area with a gamma larger than 1 was 2%. In-room times for clinical RA sessions were approximately 20 minutes for each patient. Conclusions: RA treatment planning and delivery of integrated plans of WBRT and boosts to multiple brain metastases is a rapid and accurate technique that has a higher conformity index than conventional summation of WBRT and radiosurgery boost. Ó 2009 Elsevier Inc. Brain metastases, Volumetric modulated arc therapy, Radiosurgery. INTRODUCTION The combination of whole-brain radiotherapy (WBRT) and a radiosurgery boost has been shown to improve treatment results compared with WBRT alone in selected patients with brain metastases (1, 2). The prospective randomized Radiation Therapy Oncology Group study 9508 reported a survival benefit for the combined WBRT and radiosurgery approach for patients with a single brain metastasis but also a significant improvement in intracranial disease control, performance status, and steroid use for patients with multiple brain metastases (1). At our center, patients with multiple brain metastases have been treated with a combination of WBRT and linear accelerator based frameless radiosurgery, performed as separate procedures with a 1- to 2-week interval between them. Although the delivery of frameless radiosurgery is more patient friendly and faster than traditional frame-based radiosurgery techniques, it remains a time-consuming treatment for both patients and departments. RapidArc (RA) (Varian Medical Systems, Palo Alto, CA) is a volumetric modulated arc technique that allows for highly conformal intensity-modulated three-dimensional dose distributions to be delivered with a single 358 rotation of the gantry of the linear accelerator. The planning algorithm uses progressive sampling optimization by simultaneously changing the shape of the treatment aperture, dose rate, and rotation speed of the gantry (3). The combination of accurate patient setup by use of kilovoltage cone beam computed tomography (CT) and RA treatment planning and delivery constitutes an alternative to conventional stereotactic radiotherapy. It also allows for the generation and delivery of complex radiotherapy plans such as integrated delivery of WBRT with a fractionated stereotactic boost. Reprint requests to: Frank J. Lagerwaard, M.D., Ph.D., Department of Radiation Oncology, VU University Medical Center, de Boelelaan 1117, 1081 HV Amsterdam, The Netherlands. Tel: (+31) 20 4440414; Fax: (+31) 20 4440410; E-mail: fj.lagerwaard@vumc.nl 253 Conflict of interest: The VU University Medical Center has a research collaboration with Varian Medical Systems (Palo Alto, CA). Received Sept 4, 2008, and in revised form March 15, 2009. Accepted for publication March 19, 2009.

254 I. J. Radiation Oncology d Biology d Physics Volume 75, Number 1, 2009 In this study we investigated whether RA plans consisting of integrated WBRT and boost doses to multiple brain metastases would be an appropriate alternative to our traditional technique. Plans were compared and evaluated with respect to dosimetry and treatment delivery time. The actual clinical treatments, including quality assurance (QA) measurements for this an integrated approach with RA delivery, are described. METHODS AND MATERIALS For 8 patients with brain metastases, two treatment plans were generated: (1) an integrated RA plan consisting of WBRT and a concomitant boost to the brain metastases and (2) a conventional approach of sequential WBRT followed by a stereotactic boost by use of multiple non-coplanar conformal arcs, which was customary at our center. After the evaluation of RA treatment plans and dosimetric verification in the first 5 patients, the integrated RA plans were actually delivered in the last 3 patients (patients F H). Patient H had undergone neurosurgical resection for the largest of two brain metastases and was treated with postoperative WBRT and a simultaneous integrated stereotactic boost on the remaining brain metastasis. None of the other patients underwent surgical resection. Target definition and treatment planning Patients were positioned supine in a custom-made mask (CIVCO Medical Solutions [formerly SINMED], Reeuwijk, The Netherlands), and planning CT scans without intravenous contrast were obtained with a 2.5-mm slice thickness. Contrast-enhanced T1 sequences of a coregistered diagnostic magnetic resonance imaging scan (slice thickness, 2 mm; enhanced with gadolinium with a three-dimensional distortion-correction protocol) were used for target contouring. The summated gross tumor volumes (GTVs) (i.e., the summated volume of all GTVs per patient) ranged from 1.0 to 37.5 cm 3 (Table 1). The planning target volume (PTV) for the boost (PTV boost ) was derived by adding a 2-mm margin to the GTVs to correct for possible residual positional inaccuracies by use of an online cone beam CT setup protocol. The PTV for WBRT (PTV WBRT ) was derived from autosegmentation of the brain plus the addition of a 2-mm symmetric margin. For all patients, a composite RA treatment plan (version 8.2.22) was generated, consisting of WBRT (20 Gy in 5 fractions) with a simultaneous integrated stereotactic boost, also 20 Gy in 5 fractions, to the PTV boost. The cumulative dose received by the center of the brain metastases was consequently 40 Gy in 5 fractions. Both WBRT and boost doses were prescribed at 100%, according to International Commission on Radiation Units & Measurements criteria. Treatment plans were generated with 6-MV photons, by use of multileaf collimation with a leaf width of 5 mm (Varian 120 MLC; Varian Medical Systems) and a collimator rotation of 45. All final dose calculations were performed with the Eclipse system, version 8.6.3 (Varian Medical Systems, Palo Alto, CA), by use of the anisotropic analytical algorithm calculation model, with a calculation grid of 2.5 mm, and with tissue heterogeneity correction. The maximum dose rate for treatment delivery was 600 monitor units (MU) per minute, and only a maximum of 999 MU can be delivered in a single arc with the present version of RA (version 8.2.22). The minimal accepted doses to the PTV WBRT and the PTV boost were 95% of the prescribed fraction dose of 4 Gy and 8 Gy, respectively. In the conventional stereotactic radiation treatment used at our center, a dose of 21 Gy is prescribed to the 80% Table 1. Patient characteristics. Patient No. of metastases Summated GTV boost (cm 3 ) A 4 3.1 B 2 37.5 C 3 9.8 D 2 8.0 E 5 1.0 F* 3 3.3 G* 3 25.8 H* 1 1.5 Abbreviation: GTV boost = summated gross tumor volume for boost. * Patients in whom RapidArc treatment was clinically delivered. PTV encompassing isodose, corresponding to a biologically effective dose of 65 Gy 10, calculated using an alpha/beta ratio of 10 for tumor tissue. The minimum dose of 95% of 8 Gy per fraction to the PTV boost volume with RA corresponds to a biologically effective dose of 67 Gy 10. There was no maximum dose limit for the brain metastasis, although typically, this was confined to 110% of the prescribed dose. Because of the steep dose gradients generated around the PTV boost, a maximum dose for the PTV WBRT is difficult to define. Conformity indices, which were calculated from the ratio between the total volume receiving more than 95% of the prescribed boost dose and the volume of the PTV boost, were compared for both techniques. Doses to critical organs, such as the lens, were evaluated by use of dose volume histograms. QA of RA plans All calculated RA plans were delivered on a Varian Trilogy linear accelerator (Varian Medical Systems) and measured in a 23-cm cube solid water phantom in three coronal or sagittal planes with 2 cm of separation by use of Gafchromic EBT films (International Specialty Products, Wayne, NJ) (Fig. 1). In each plane double films were used to reduce the statistical uncertainty per film, which is about 1.8% for double films (4). All film measurements were compared with calculated dose distributions of the respective RA plans on the solid water phantom. Comparisons were performed by means of a gamma evaluation (5, 6) with dose and distance criteria of 3% of the WBRT dose and 2 mm, respectively. Areas that do not meet these criteria will have a gamma larger than 1. Patient setup procedure To ensure the accuracy of delivery of the integrated treatment plans, daily online setup by use of a combination of a lateral kilovoltage image and cone beam CT scans was performed for the 3 patients treated with RA. The lateral kilovoltage image was used for detecting a pitch larger than 0.8. Each part of the PTV WBRT and PTV boost is within 10 cm of the isocenter. A maximum pitch or roll of 0.8 leads to a maximum positioning error of 1.4 mm; a combination of the two rotations would lead to a maximum error of 2 mm (i.e., the clinical target volume PTV margin used). All rotations exceeding 0.8 were corrected for by reapplying the mask and repeating the cone beam CT to ensure that the correct position was obtained. All detected shifts in patient position were corrected. RESULTS In this study we performed a dosimetric comparison of integrated RA plans and a conventional summation of WBRT and

RA for WBRT and boost to multiple metastases d F. J. LAGERWAARD et al. 255 Fig. 1. Quality assurance measurements of RapidArc plans reconstructed in a cube solid water phantom in coronal or sagittal planes using Gafchromic EBT films in patient G. stereotactic boosts in 8 patients with brain metastases. After completion of the planning study and QA measurements, we performed clinical delivery of the integrated plans in 3 patients. Dosimetric results are provided in Tables 2 and 3. More than 999 MU is needed to deliver the fraction doses of 4 Gy (PTV WBRT ) and 8 Gy (PTV boost ), and consequently, two arcs are needed to deliver the dose. Although it would have been possible to deliver two identical arcs, we have opted for the solution of a second arc that uses the results of the first arc as a starting point for further optimization. For this second (compensatory) arc, the collimator is rotated from 45 (first arc) to 40 (second arc), to ensure that possible tongue and groove underdosage did not add up along the same lines. The use of this two-arc approach also has the advantage of noninterrupted irradiation without rotation of the gantry back to its starting point. The principle of the compensatory second arc is illustrated in Fig. 2. The relative cold spots and hotspots that can be seen in the dose distribution of the baseline plan for the first arc (left panel) are compensated for in the second arc (middle panel), and the resulting composite plan of both arcs shows improved homogeneity (right panel). All evaluated RA plans were derived from the summation of the first arc and the compensatory second arc. An example of an integrated RA plan for WBRT with simultaneous integrated boost to multiple metastases is shown in Fig. 3. Dosimetric analysis of the composite RA plans showed excellent coverage of both PTV WBRT and PTV boost, with mean volumes receiving at least 95% of the prescribed dose of 100% and 99.8%, respectively (Table 2). The maximum point dose, which was in all cases located within the PTV boost, had a mean value of 108.9% (range, 106.2 110.4%). Table 2. Dosimetric results in patients A through H obtained with RA planning V 95 WBRT (%) Boost (%) D max (%) Patient A 99.9 99.9 107.3 B 100 99.8 110.4 C 99.9 99.9 109.0 D 99.9 99.9 106.2 E 99.9 100 109.3 F 100 99.7 108.2 G 100 99.3 110.4 H 100 99.8 110.4 Mean SD 100 0.1 99.8 0.2 108.9 1.6 Abbreviations: RA = RapidArc; V 95 = volume receiving at least 95% of prescribed dose; D max = maximum dose.

256 I. J. Radiation Oncology d Biology d Physics Volume 75, Number 1, 2009 Table 3. Dosimetric comparison between integrated RA plans and a conventional summation of WBRT and stereotactic boosts Conventional summation RA V25 (%) V30 (%) V35 (%) Conformity index V25 (%) V30 (%) V35 (%) Conformity index Patient A 12.9 4.1 2.1 2.4 4.3 2.1 1.1 1.3 B 21.5 8.5 5.2 1.5 9.3 5.9 4.0 1.2 C 7.1 2.7 1.5 1.6 3.5 2 1.2 1.2 D 8.2 2.8 1.6 2.0 3.3 1.9 1.1 1.2 E 3.7 1.0 0.4 3.4 1.9 0.8 0.3 2.0 F 3.4 1.3 0.7 2.2 2 0.9 0.5 1.2 G 18 7.9 4.9 2.3 6.6 4.2 2.7 1.1 H 0.9 0.4 0.3 1.8 0.6 0.3 0.2 1.1 Mean SD 9.5 7.4 3.5 3.1 2.1 1.9 2.1 0.6 3.9 2.8 2.3 1.9 1.4 1.3 1.3 0.3 Abbreviations: RA = RapidArc; WBRT = whole-brain radiotherapy; V25 = percent of normal brain receiving total dose of 25 Gy; V30 = percent of normal brain receiving total dose of 30 Gy; V35 = percent of normal brain receiving total dose of 35 Gy. One of the most striking, though obvious, advantages of generating integrated WBRT and boost plans is illustrated in Fig. 4. The right panel shows a summation of a conventional WBRT plan with a standard radiosurgery boost derived from five non-coplanar arcs with dynamic conformal arcs. In the left panel the comparative RA plan shows much steeper dose gradients outside the PTV boost, resulting from the modulation of the WBRT dose within the area of the boost dose gradient. This is reflected by a significantly better conformity index for RA plans of 1.3 0.3, which contrasted to a conformity index of 2.1 0.6 for the conventional summation (p < 0.001, t test) (Table 3). The volume of normal brain receiving doses of between 25 and 35 Gy was also smaller with RA planning. A maximum dose (D max ) constraint of 5 Gy for both lenses was routinely used in RA planning, which resulted in a mean D max of 9.5 Gy with RA. With the conventional summation of WBRT plus radiosurgery, the D max was 5 Gy. The mean number of monitor units and the mean beamon time needed to deliver both arcs were 1,600 MU (range, 1,404 1,790 MU) and 180 seconds (range, 160 210 seconds), respectively. In the 3 patients actually treated by use of the daily setup with cone beam CT scans and RA delivery, the total time needed for patients to enter and leave the treatment room after delivery has been limited to approximately 20 minutes with growing experience with online patient setup. QA measurements The measured dose distributions generally agreed well with the calculated distributions. QA film measurements of single-arc plans showed maximum differences between calculated and measured doses of up to 7.5%. However, the use of summated plans with two separate arcs averaged out these differences. The mean gamma, averaged for all measured planes for the 8 patients, for the single arc and double arc were 0.50 and 0.30, respectively. The area with a gamma larger than 1 (3% of WBRT dose, 2 mm), also averaged for all measurements, was 6% for the single arc comparisons and only 2% for the double-arc comparisons. Figure 5 shows a typical example of a comparison in the sagittal plane of a composite plan consisting of two arcs. DISCUSSION Despite the increased speed with the use of noninvasive immobilization devices, conventional radiosurgery remains a time-consuming technique. Treatment delivery times on Fig. 2. RapidArc dose distributions showing baseline plan (left) and a compensatory second plan (middle), leading to improved homogeneity with an absence of hotspots and cold spots in the summated two-arc plan (right).

RA for WBRT and boost to multiple metastases d F. J. LAGERWAARD et al. 257 Fig. 3. Composite RapidArc plan with whole-brain radiotherapy and integrated boost to multiple metastases for patient C. our Novalis linear accelerator (BrainLAB, Feldkirchen, Germany), by use of the frameless method, varies from approximately 30 minutes for a single metastasis to well over 1 hour for three metastases. Our planning analysis and early clinical data indicate that integrated WBRT and fractionated stereotactic boost are feasible in a very short mean beam-on time of 180 seconds by use of RA. Although the use of two arcs is inevitable as a result of the maximum of 1,000 MU per arc, this does not prolong the treatment delivery time, because the gantry does not need to be rotated back to its starting point. The total in-room time span, which includes patient setup by cone beam CT scans and treatment delivery, has been approximately 20 minutes in patients who have actually been treated with the RA plans. The high delivery speed may not only increase patient tolerance of radiosurgery procedures and the efficiency of radiation oncology departments but can also decrease the risk of intrafractional positional shifts of the patient within the fixation device. Although intrafractional three-dimensional positional shifts were generally small, with a mean SD of 0.3 mm, extremes of up to 1.5 mm were observed in our population treated with the frameless mask system (7). As a result of modulation of the WBRT dose in the area of the dose gradient of the boost doses, integrated plans have much steeper dose gradients than comparable plans with conventionally summated WBRT and radiosurgery. This leads to a large increase in conformity of the high-dose region, which has been correlated to reduced normal tissue Fig. 4. Comparison between RapidArc (RA) composite plan (left) and standard summation of whole-brain radiotherapy (WBRT) and radiosurgery boost (right) for patient B. The dose color wash shows only doses above 23 Gy (115% of WBRT dose) to demonstrate the steeper dose gradient with RA plans. The dose volume histograms of the brain in both cases are shown (middle). The steepest dose volume histogram (yellow) is for the RA plan.

258 I. J. Radiation Oncology d Biology d Physics Volume 75, Number 1, 2009 Fig. 5. Results of quality assurance film measurements in patient F, showing excellent agreement between calculated dose (red line) and measured dose (green line). Gamma analysis showed only very limited areas with a gamma greater than 1 (3% of whole-brain radiotherapy dose, 2 mm). complication probability values (8). The advantages of steeper dose gradients appear to be increased with the number of metastases and the size of metastases treated. Integrated plans of WBRT and simultaneous boosts for brain metastases have previously been described in planning studies for helical TomoTherapy (9, 10). RA treatment differs from TomoTherapy by the simultaneous irradiation of the entire target volume in contrast to the slice-by-slice delivery of the latter. This is reflected in the difference in the beamon time, which was reported to be on the order of 8 minutes for TomoTherapy (TomoTherapy Inc, Madison, WI) for this indication (9). QA film measurements of single RA plans in a solid water phantom were performed for all plans. Despite maximum differences between calculated and measured doses of up to 7.5% being observed in single-arc plans, the results were improved for plans using two summated arcs. The mean gamma values for the single- and double-arc plans were 0.50 and 0.30, respectively. Further improvements in the RA planning algorithm may help to decrease the inaccuracies observed with a single arc. RA treatment has now replaced successive WBRT with radiosurgery boost for patients with multiple brain metastases at our center. Verification of calculated RA plans by use of phantom measurements is performed before treatment in all cases, and dedicated online patient setup by use of kilovoltage cone beam CT is performed before each fraction. In the first 3 patients RA treatment was well tolerated under corticosteroid protection, although it is too early to assess the efficacy with respect to intracranial disease control. Future improvements in RA delivery can be expected shortly, when our Novalis Tx unit is commissioned by use of micro-multileaf collimation with very low leaf transmission (<1.2% compared with 1.6% for the Millennium 120 MLC), as well as the ability to correct patient rotations by use of a Robotics treatment couch (BrainLAB). A newer version of RA software will allow for non-coplanar arcs or selected arc ranges to be used, which may be desirable in some cases (e.g., for avoiding beam directions where radiation is transmitted through immobilization devices). In conclusion, RA treatment planning of integrated WBRT and simultaneous fractioned boost to multiple brain metastases results in highly conformal dose distributions. QA measurements showed high agreement with calculated dose distributions. Treatment delivery is feasible in a short beam-on time on the order of 3 minutes. The clinical benefit of this approach with respect to intracranial disease control will be investigated in a Phase II study.

RA for WBRT and boost to multiple metastases d F. J. LAGERWAARD et al. 259 REFERENCES 1. Andrews DW, Scott CB, Sperduto PW, et al. Whole brain radiation therapy with or without stereotactic radiosurgery boost for patients with one to three brain metastases: Phase III results of the RTOG 9508 randomised trial. Lancet 2004;363:1665 1672. 2. Kondziolka D, Patel A, Lunsford LD, et al. Stereotactic radiosurgery plus whole brain radiotherapy versus radiotherapy alone for patients with multiple brain metastases. Int J Radiat Oncol Biol Phys 1999;45:427 434. 3. Otto K. Volumetric modulated arc therapy: IMRT in a single gantry arc. Med Phys 2008;35:310 317. 4. van Battum LJ, Hoffmans D, Piersma H, et al. Accurate dosimetry with GafChromic EBT film of a 6 MV photon beam in water: What level is achievable? Med Phys 2008;35:704 716. 5. Stock M, Kroupa B, Georg D. Interpretation and evaluation of the gamma index and the gamma index angle for the verification of IMRT hybrid plans. Phys Med Biol 2005;50:399 411. 6. Depuydt T, Van Esch A, Huyskens DP. A quantitative evaluation of IMRT dose distributions: Refinement and clinical assessment of the gamma evaluation. Radiat Oncol 2002;62:309 319. 7. Verbakel WF, Cuijpers JP, Verduim AJ, et al. Accuracy of frameless stereotactic intracranial radiotherapy. Int J Radiat Oncol Biol Phys 2007;69:S701 S702. 8. Pasciuti K, Iaccarino G, Soriani A, et al. DVHs evaluation in brain metastases stereotactic radiotherapy treatment plans. Radiother Oncol 2008;87:110 115. 9. Bauman G, Yartsev S, Fisher B, et al. Simultaneous infield boost with helical tomotherapy for patients with 1 to 3 brain metastases. Am J Clin Oncol 2007;30:38 44. 10. Gutiérrez AN, Westerly DC, Tomé WA, et al. Whole brain radiotherapy with hippocampal avoidance and simultaneously integrated brain metastases boost: A planning study. Int J Radiat Oncol Biol Phys 2007;69:589 597.