Treating patients with Dynamic Wave Arc: First clinical experience

Radiotherapy and Oncology xxx (2017) xxx–xxx

Contents lists available at ScienceDirect

Radiotherapy and Oncology journal homepage: www.thegreenjournal.com

Original article

Treating patients with Dynamic Wave Arc: First clinical experience Manuela Burghelea a,c,e,⇑, Dirk Verellen b, Jennifer Dhont a, Cecilia Hung c, Thierry Gevaert a, Robbe Van den Begin a, Christine Collen a, Kenneth Poels d, Koen Tournel a, Marlies Boussaer a, Cyril Jaudet a, Truus Reynders a, Viorica Simon e, Mark de Ridder a a Universitair Ziekenhuis Brussel, Vrije Universiteit Brussel, Department of Radiotherapy; b GZA Ziekenhuizen, Sint Augustinus – Iridium Kankernetwerk Antwerpen, VUB, Faculty of Medicine and Pharmacy, Brussels, Belgium; c Brainlab AG, Feldkirchen, Germany; d Department of Radiation Oncology, University Hospitals Leuven, Belgium; and e Babes Bolyai University, Faculty of Physics, Cluj-Napoca, Romania

a r t i c l e

i n f o

Article history: Received 5 October 2016 Received in revised form 2 January 2017 Accepted 3 January 2017 Available online xxxx Keywords: Dynamic Wave Arc Clinical implementation Noncoplanar rotational IMRT

a b s t r a c t Background and purpose: Dynamic Wave Arc (DWA) is a system-specific noncoplanar arc technique that combines synchronized gantry-ring rotation with D-MLC optimization. This paper presents the clinical workflow, quality assurance program, and reports the geometric and dosimetric results of the first patient cohort treated with DWA. Methods and materials: The RayStation TPS was clinically integrated on the Vero SBRT platform for DWA treatments. The first 15 patients treated with DWA represent a broad range of treatment sites: breast boost, prostate, lung SBRT and bone metastases, which allowed us to explore the potentials and assess the limitations of the current DWA site-specific template solution. For the DWA verification a variety of QA equipment was used, from 3D diode array to an anthropomorphic end-to-end phantom. The geometric accuracy of each arc was verified with an independent orthogonal fluoroscopy method. Results: The average beam-on delivery time was 3 min, ranging from 1.22 min to 8.82 min. All patient QAs passed our institutional clinical criteria of gamma index. For both EBT3 film and Delta4 measurements, DWA planned versus delivered dose distributions presented an average agreement above 97%. An overall mean gantry-ring geometric deviation of 0.03° ± 0.46° and 0.18° ± 0.26° was obtained, respectively. Conclusion: For the first time, DWA has been translated into the clinic and used to treat various treatment sides. DWA has been successfully added to the noncoplanar rotational IMRT techniques arsenal, allowing additional flexibility in dose shaping while preserving dosimetrically robust delivery. Ó 2017 Elsevier B.V. All rights reserved. Radiotherapy and Oncology xxx (2017) xxx–xxx

Noncoplanar trajectories offer the potential to reduce normal tissue complication probability, a hypothesis that is supported by several reports in literature [1–3]. However, the translation from research to clinical practice is less straightforward. While modern LINACs can deliver noncoplanar VMAT, the clinical implementation has been limited to Dynamic Wave Arc and a pilot clinical trial that studies the feasibility 4p radiotherapy [4]. Dynamic Wave Arc (DWA) is a system-specific noncoplanar arc technique that combines synchronized gantry-ring rotation with dynamic multileaf (D-MLC) optimization. The theoretical concept was first introduced in 2013 [5], and highlighted the potential benefits of a new technique that allows both the radiation head unit and the ring-shaped gantry to rotate simultaneously during arc irradiation. However, it was not until 2015 that the novel tech-

⇑ Corresponding author at: Department of Radiotherapy, UZ Brussel, Laarbeeklaan 101, B-1090 Brussels, Belgium. E-mail address: [email protected] (M. Burghelea).

nique was supported by a commercially available treatment planning system (TPS) and the Vero machine was mechanically capable of combined gantry-ring rotations. Sato et al. [6] assessed the accuracy of the machine motion and dosimetric delivery for various gantry-ring pathways and velocities. Gantry-ring differences were reported to be within 0.2°, while the accumulated MU deviations remained within ±0.21%. Independently, Burghelea et al. [7] reported gantry-ring deviations below 0.6°. Another pre-clinical study showed additional flexibility in dose shaping and significantly faster treatment times compared to the other Vero clinical solutions [8] and therefore will not be covered in this report. The clinical motivation behind DWA implementation is the additional degrees-of-freedom in dose shaping to obtain a more patient-individualized treatment plan, and ultimately a better outcome for the patient, i.e. higher tumor control with fewer side effects. In this article, we report on the first clinical application of noncoplanar VMAT on a cohort of 15 patients. This study includes a detailed description of the clinical workflow, focusing on quality

http://dx.doi.org/10.1016/j.radonc.2017.01.006 0167-8140/Ó 2017 Elsevier B.V. All rights reserved.

Please cite this article in press as: Burghelea M et al. Treating patients with Dynamic Wave Arc: First clinical experience. Radiother Oncol (2017), http://dx. doi.org/10.1016/j.radonc.2017.01.006

2

Treating patients with DWA

assurance and dosimetric results. Additionally, it is written with the aim of facilitating introduction of DWA into other clinical sites. To our knowledge, no clinical use of this technology has been published at this moment. Material and methods Clinical implementation of DWA To support DWA deliveries, the Vero system required additional software and hardware upgrades: an MLC secondary feedback unit upgrade allowing faster dynamic MLC leaf movement of up to 4 cm/s at isocenter level, and a machine controller offering gantry-ring synchronous rotations. The R&V system (MHITM2000 v3.5.8) and ExacTrac Vero (SW v3.5.3) are now able to automatically import and recognize DWA plans. The RayStation v5.02 (RaySearch Laboratories, Sweden) TPS was clinically integrated with the Vero SBRT system for DWA. The current planning solution used at UZ Brussels aimed to deliver fixed anatomical site-specific trajectories that can be selected from a Wave Arc template list before the optimization process. Each trajectory is defined by 5 to 8 Manipulation Points (MP), gantry-ring positions where the direction of the ring rotation can change. Hence, DWA is an extended form of VMAT, delivered with continuously varying ring positions at constant dose rate [9]. The main difference in the optimization modules of VMAT and DWA is during the angular spacing, where the DWA optimization algorithm does not consider the gantry spacing, but only the Euclidian norm of the ring and gantry angle. The final dose was calculated on a 2  2  2 mm3 resolution dose grid using collapse cone dose engine v3.2. A video of the D-MLC modulation following a predefined DWA trajectory for a representative lung SBRT case is provided in Supplementary Movie M1. The standard patient positioning protocol on the Vero system, based on IR markers and CBCT, remained unchanged for the DWA technique. As a general rule in the workflow for noncoplanar treatments, the patients with peripheral indications are offcentered on the couch in order to have minor lateral couch displacement. As such it avoids potential collision with the ring. During every first fraction of a DWA treatment, a collision verification step is added into the workflow where the wave trajectory is simulated without beam-on, using the control pendant inside the treatment room. Patient population and DWA planning The first 15 patients treated with DWA on the Vero system represent a broad range of treatment sites: breast boost, prostate, lung SBRT and bone metastases. Table 1 provides further information for each patient case including the corresponding DWA plan information. The site variability allowed us to explore the DWA potentials and assess its limitations. The planning objectives for target volumes were individualized for each patient, while OAR doses were kept as low as achievable, which originated from recommendations by RTOG clinical trials (RTOG 0813, 0618) and EORTC boost versus no boost trial [10]. In Fig. 1 are presents the most common DWA trajectories of different treatment sites. For breast boost, a single DWA was enough to fulfill the required dose distribution. This trajectory has two ‘‘tangential” sub-arcs at the beginning/end of the path, while the central part is shaped to follow the thoracic wall to best avoid the heart and lung. For long lesions, based on the tumor position (i.e. left, right, central, peripheral), the gantry-ring collision free zone is different and this can be observed in Fig. 1b where the higher ring rotation is on the lesion side. Full and half DWA trajectories were combined to achieve the lung SBRT planning

constraints but also to allow second CBCT verification between arcs. The prescription dose ranged from 7.5 Gy to 17 Gy per fraction, based on the tumor localization and proximity to OARs. For prostate, two templates whose ring movements are in ‘‘W” shapes of 10° and 15° rotations to sharper conform the dose gradient around the target, were generally used (Table 1), whereas the dual-DWA modality was used to accomplish the horseshoe dose shape around the PTV to spare the rectum. The multiple lesion case was planned using the multiple beam set module, with one DWA per lesion and background dose dependency between sets during optimization. In the case presented in Fig. 1d, the specific curve was chosen to avoid the kidneys as much as possible.

Pre-treatment quality assurance (QA) As part of our institution’s pre-treatment QA procedure, the relative dose distribution and absolute point dose of each clinical plan were verified on GafChromic EBT3 film (Ashland ISP Advanced Materials, USA) [11] and with an A1SL ion chamber (0.053ccm, Standard Imaging, USA), respectively. Both were simultaneously inserted in the middle of the ‘‘Cheese” phantom (Gammex RMI, WI). To ensure the safety and efficiency of the SBRT treatments, additional measures were implemented for these particular cases using the E2E SBRT anthropomorphic thorax phantom (CIRS, USA), allowing high-resolution dose distribution measurements to the target and OARs in a single delivery. Further information regarding this verification step is presented in Supplementary File S3. The volumetric absolute dose distributions were measured with a biplanar diode array (Delta4, ScandiDos AB, Sweden) that takes into consideration the corresponding angular correction factors during dose reconstruction [12]. Using an independent verification method based on treatment log files and on-board orthogonal fluoroscopy previously reported [7], the geometric accuracy and stability of synchronized gantry-ring rotations during DWA were evaluated. The kV images were acquired during the QA process, while the log files were registered for the first three treatment fractions of each patient. Results The planned dose statistics are summarized per indication in Supplementary Table S2. Without exception, all OAR doses were kept below our institutional standards. The average beam-on delivery time was 3 min, ranging from 1.22 min to 8.82 min (Table 1). The total treatment time for breast boost and prostate cases was kept below 7 min, which includes the following steps: patient positioning (0.8 min), DWA collision verification (1 min) and CBCT acquisition plus verification (2.5 min). The longest treatment time was registered for the lung SBRT cases due to the high prescription dose and secondary CBCT verification between DWAs. The average ɣ (3%, 3 mm) passing rate for film measurements was 97.0 ± 1.6% (range from 93.3% to 98.8%), while the Delta4 measurements presented an average ɣ (2%, 2 mm) of 97.7 ± 1.4% (range from 95.3% to 99.6%) respectively. The average isocentric point ratio was 99.9 ± 1.2% (range from 98.0% to 102.8%). The results of the lung SBRT verifications with the CIRS phantom are presented in Supplementary Table S3. An average local ɣ of 97.0 ± 1, 0% and 96.6 ± 0.5% was obtained during the coronal and sagittal film analysis, whereas the average isocentric point ratio was 100.0 ± 1.4%. The measurements in the low dose gradient presented a local ɣ (4%, 4 mm) of 0.7 ± 0.2 and 0.7 ± 0.1 for the spine insert and the low dose lung insert respectively. Fig. 2 presents the mean gantry and ring angular relative deviation per DWA over all investigated cases obtained through the geometric accuracy analysis of detected positions from kV images.

Please cite this article in press as: Burghelea M et al. Treating patients with Dynamic Wave Arc: First clinical experience. Radiother Oncol (2017), http://dx. doi.org/10.1016/j.radonc.2017.01.006

M. Burghelea et al. / Radiotherapy and Oncology xxx (2017) xxx–xxx

3

Fig. 1. DWA trajectories for (a) left breast boost, (b) right lung, (c) prostate and (d) bone metastases. The diagrams present the 2D gantry-ring rotational positions with highlighted turning points (square red markers), while the images present the DWA curve in 3D view.

An overall mean gantry-ring deviation of -0.03° ± 0.46° and 0.18° ± 0.26° was obtained, with the largest SDs for the prostate cases. The gantry-ring angular deviation comparison with the values from the log files acquired during delivery is presented in Suppl. Fig. S2. DWA is verified step by step with an increment equal to the number of images acquired. The mean gantry-ring deviation was 0.04° ± 0.11° and 0.17° ± 0.16° with the largest SDs registered also for the prostate cases. The LogActed recorded for the first three fractions presented an average angular difference of 0.08° ± 0.05° for gantry and 0.04° ± 0.06° for ring, demonstrating a good daily reproducibility of DWA (data not shown). Discussion Defined by synchronized gantry-ring rotation combined with DMLC modulation, DWA has been successfully added to the noncoplanar rotational IMRT techniques arsenal allowing additional flexibility in dose shaping while preserving dosimetrically robust delivery. This paper reports on the clinical workflow, quality assur-

ance, and geometric and dosimetric results of the first patient treatments with DWA. Introduction of the novel technique into the daily workflow was straightforward, with no major changes. Although DWA involves complex movements in the LINAC system, its performance could not be interfered by the patient behavior or system operator. A real time DWA delivery in the treatment room is presented in Supplementary Movie M2. Another broadly reported technique yet to be clinically used is the noncoplanar VMAT by dynamic couch rotation, available with the Varian Developer Mode on TrueBeam LINACs (Varian Medical Systems, Inc., Palo Alto). Several groups have developed in-house software planning solutions [1,3,13], quality control procedures [14] or collision prediction models [15] for dynamic treatment delivery techniques involving couch motion. However, the transition into a standard clinical workflow is still pending due to lack of clinical release. The current DWA planning solution can be easily applied on mainstream machines, nevertheless extra considerations for the intra-fraction patient movement due to the moving couch [2] are recommended.

Please cite this article in press as: Burghelea M et al. Treating patients with Dynamic Wave Arc: First clinical experience. Radiother Oncol (2017), http://dx. doi.org/10.1016/j.radonc.2017.01.006

4

Treating patients with DWA

Table 1 Summary of patient diagnoses, dose prescription and DWA plan parameters. Patient Nr.

Indication

Fractionation

DWA template

Nr. of MPs

Nr. of CPs

Nr. kV images

Nr. MU

Delivery time (min)

1 2

Right breast boost Prostate

8  2 Gy 39  2 Gy

Lung LLL centrally

8  7,5 Gy

4

Prostate

39  2 Gy

5

Lung RML centrally

3  17 Gy

6 7

Left Breast boost Lung RUL peripherally

4  2,5 Gy 4  12 Gy

8

Lung RUL centrally

8  7,5 Gy

9

Lung RUL paravertebral

4  12 Gy

10 11 12

Left Breast boost Right Breast boost Prostate

4  2,5 Gy 8  2 Gy 39  2 Gy

13 14

Left breast boost Prostate

8  2 Gy 39  2 Gy

15

Bone metastases

6 9 9 5 5 6 6 7 5 6 6 9 6 6 6 5 6 6 9 9 6 9 9 5 5 5 5

59 89 89 90 45 89 89 90 45 51 51 89 89 51 89 45 51 51 89 89 51 89 89 45 45 90 90

195 146 160 298 132 129 130 523 571 185 335 556 324 288 625 161 175 195 148 149 188 152 150 155 228 213 200

505 239 220 969 423 168 233 1686 1840 435 1082 1823 1057 842 2062 522 427 521 194 200 472 208 282 487 740 697 662

1,60 2,37

3

B3 X1 X1 ccw L5 X5 X2 X2 ccw L1 X3 B2 B2 X1 X2 B2 X2 X3 B2 B1 X1 X1 ccw B2 X1 X1 ccw X5 B2 P2 E1

Right 8th rib Left 3th rib Vertebra L5 Right ilium

5  4 Gy 5  4 Gy 5  4 Gy 5  4 Gy

3,48 2,07 8,82 1,43 7,28 4,92 6,48 1,42 1,60 2,37 1,50 2,37 1,22 1,85 1,73 1,67

Abbreviations: LLL = left lower lobe; RML = right middle lobe, RUL = right upper lobe; MP = Manipulation Points i.e. gantry-ring angle positions where the direction of the ring rotation can change; CP = Control Points, i.e. gantry-ring angle positions spaced every 4° where the beam aperture shaped by the MLC can change; Nr. kV images = fluoroscopic images acquired during DWA QA delivery for the gantry-ring geometric verification; Delivery time = measured beam-on delivery time.

The DWA delivery time per treatment indication (Table 1) is lower in comparison with other noncoplanar IMRT approaches. Fahimian et al. [16] reported a delivery time for accelerated breast boost with trajectory modulated arc therapy of 4 min an 56 s. A 4p plan of 30 noncoplanar beams was estimated at approx. 10 min for a 12 Gy lung SBRT treatment [17]. The DWA geometric accuracy results are in good agreement with previous reports [6,7], indicating that DWA is a robust treatment technique across different Vero systems. The prostate cases have the fastest delivery time per arc (reflected by the lower fraction dose), therefore the lowest sampling number of kV images, which can be the reason for the larger SDs values. DWA plan versus delivered dose distributions presented an overall average agreement above 96% for all indications, using a wide variety of QA equipment from routine 3D diode array to an anthropomorphic end-to-end phantom. This milestone represents an important step forward in the Vero community, making DWA a clinical standard noncoplanar rotational IMRT. In order to draw conclusions with respect to the expected clinical gain of DWA, gathering experience from a larger patient group will be necessary. The DWA impact would depend on the target localization, shape and proximity to critical normal tissue. There are several avenues of DWA for further improvements. For instance, the current template based DWA implementation is acceptable as starting point to validate and get accustomed with the new technology; however higher flexibility would be required in clinical practice. Patient-specific, personalized trajectories would allow the planner to create more individualized plans and fully utilize the optimization module of the planning system in combination with adaptive noncoplanar curves. Moreover, for patients with low mobility or increased weight, it’s a requirement to be able to change the trajectory. A collision situation was experienced with one of the breast boost cases, where the DWA starting point collided with the patient’s ipsilateral elbow. The repositioning ‘‘to fit” in the DWA collision-free zone took 30 min. Since there were only two non-editable breast templates for each side avail-

able, the preventive clinical solution was to position the following patients with the ipsilateral arm next to the body. Based on our experience, pure template-based DWA represents a shortcoming for the technique and a DWA editing tool is mandatory for the efficiency of the workflow. Considering that real-time gimbal tracking proved to add substantial treatment quality gains compared to no motion compensation [18,19], DWA in combination with real-time tumor tracking (RTTT) could likely be another advancement in terms of both dosimetry and treatment comfort. Preliminary tests with DWARTTT have been performed to determine the interplay between MLC non-coplanar modulation behavior in synchrony with the gimbal movements.

Conclusion/clinical translational relevance The DWA implementation in the clinical workflow represents a crucial step in a long pathway from the bench to the bedside. Template-based DWA has been successfully implemented in a clinical setting and used to treat a variety of indications for the first time. Further plans include individualized patient DWA trajectories that could be of particular interest in comparison with alternative noncoplanar clinical treatment modalities by others investigators, for example by using automated couch rotation. Also further investigations are planned to explore the dosimetric and geometric accuracy of DWA-RTTT technical combination for realtime tumor motion management. In a short period of time, DWA has already become a standard treatment technique on the Vero system in our department.

Competing interests This collaborative work was supported by corporate funding from BrainLAB AG. MB and CH are financially supported by BrainLAB AG. The other authors have no conflict of interest.

Please cite this article in press as: Burghelea M et al. Treating patients with Dynamic Wave Arc: First clinical experience. Radiother Oncol (2017), http://dx. doi.org/10.1016/j.radonc.2017.01.006

M. Burghelea et al. / Radiotherapy and Oncology xxx (2017) xxx–xxx

5

Fig. 2. Statistical analysis of the Gantry (a) and Ring (b) geometrical accuracy per Wave Arc. The angular difference between each Control Point (CP) and the closest GantryRing detected position from the kV images (DetPositions) is represented with a box-and-whiskers plot: whiskers above and below the box = the locations of the minimum and maximum; upper edge of box = 75th percentile; the red segment inside the rectangle = the median; lower edge of the box = 25th percentile; the red plus markers represent outliers (3 range of variation) in the data series. Note: The Gantry-Ring positions detected from the fluoroscopy images (DetPositions) were benchmarked against the Control Points of the DWA RT plan (CPs) to verify if the system accurately passes through each commanded set of gantry and ring angles of the trajectory at that given moment.

Acknowledgments The authors express their gratitude to BrainLAB Vero team and RaySearch Laboratories for their assistance during this project and for supplying the hardware and software clinical upgrades. Appendix A. Supplementary data Supplementary data associated with this article can be found, in the online version, at http://dx.doi.org/10.1016/j.radonc.2017.01. 006. References [1] Wild E, Bangert M, Nill S, Oelfke U. Noncoplanar VMAT for nasopharyngeal tumors: plan quality versus treatment time. Med Phys 2015;42:2157–68. [2] Smyth G, Evans PM, Bamber JC, et al. Non-coplanar trajectories to improve organ at risk sparing in volumetric modulated arc therapy for primary brain tumours. Radiother Oncol 2016;16. 31206–3. [3] MacDonald RL, Thomas CG. Dynamic trajectory-based couch motion for improvement of radiation therapy trajectories in cranial SRT. Med Phys 2015;42:2317–25. [4] Jonsson Comprehensive Cancer Center, Palliative 4pi Radiotherapy in Treating Patients With Recurrent Glioblastoma Multiforme. 2015. ClinicalTrials.gov. Bethesda (MD): National Library of Medicine (US). Available from: https://clinicaltrials.gov/ct2/show/NCT02575027. [5] Mizowaki T, Takayama K, Nagano K, et al. Feasibility evaluation of a new irradiation technique: three-dimensional unicursal irradiation with the Vero4DRT (MHI-TM2000). J Radiat Res 2013;54:330–6. [6] Sato S, Miyabe Y, Takahashi K, et al. Commissioning and quality assurance of Dynamic WaveArc irradiation. J Appl Clin Med Phys 2015;16:73–86. [7] Burghelea M, Verellen D, Poels K, et al. Geometric verification of dynamic wave arc delivery with the vero system using orthogonal X-ray fluoroscopic imaging. Int J Radiat Oncol 2015;92:754–61.

[8] Burghelea M, Verellen D, Poels K. Initial characterization, dosimetric benchmark and performance validation of Dynamic Wave Arc. Radiat Oncol 2016;11:63. [9] Bzdusek K, Friberger H, Eriksson K, Hårdemark B, Robinson D, Kaus M. Development and evaluation of an efficient approach to volumetric arc therapy planning. Med Phys 2009;36:2328. [10] P. Poortmans, H. Bartelink, J. Horiot, et al. The influence of the boost technique on local control in breast conserving treatment in the EORTC ‘ boost versus no boost’ randomised trial, Radiother Oncol. 2004;72:25–33. [11] Casanova Borca V, Pasquino M, Russo G, et al. Dosimetric characterization and use of GAFCHROMIC EBT3 film for IMRT dose verification. J Appl Clin Med Phys 2013;14:4111. [12] Feygelman V, Stambaugh C, Opp D, Zhang G, Moros EG, Nelms BE. Crossvalidation of two commercial methods for volumetric high-resolution dose reconstruction on a phantom for non-coplanar VMAT beams. Radiother Oncol 2014;110:558–61. [13] Yu VY, Fahimian BP, Xing L, Hristov DH. Quality control procedures for dynamic treatment delivery techniques involving couch motion. Med Phys 2014;41:81712. [14] Yu VY, Tran A, Nguyen D, Cao M, Ruan D, Low DA, et al. The development and verification of a highly accurate collision prediction model for automated noncoplanar plan delivery. Med Phys 2015;42:6457. [15] Fahimian B, Yu V, Horst K, Xing L, Hristov D. Trajectory modulated prone breast irradiation: A LINAC-based technique combining intensity modulated delivery and motion of the couch. Radiother Oncol 2013;109:475–81. [16] Dong P, Lee P, Ruan D, et al. 4P Noncoplanar stereotactic body radiation therapy for centrally located or larger lung tumors. Int J Radiat Oncol Biol Phys 2013;86:407–13. [17] Papp D, Bortfeld T, Unkelbach J. A modular approach to intensity-modulated arc therapy optimization with noncoplanar trajectories. Phys Med Biol 2015;60:5179–98. [18] Depuydt T, Poels K, Verellen D, et al. Treating patients with real-time tumor tracking using the Vero gimbaled linac system: Implementation and first review. Radiother Oncol 2014;112:343–51. [19] Colvill E, Booth J, Nill S, et al. A dosimetric comparison of real-time adaptive and non-adaptive radiotherapy: A multi-institutional study encompassing robotic, gimbaled, multileaf collimator and couch tracking. Radiother Oncol 2016;119:159–65.

Please cite this article in press as: Burghelea M et al. Treating patients with Dynamic Wave Arc: First clinical experience. Radiother Oncol (2017), http://dx. doi.org/10.1016/j.radonc.2017.01.006