- Email: [email protected]
Clinical experience transitioning from IMRT to VMAT for head and neck cancer
Medical Dosimetry 38 (2013) 171–175
Medical Dosimetry journal homepage: www.meddos.org
Clinical experience transitioning from IMRT to VMAT for head and neck cancer Matthew T. Studenski, Ph.D.,* Voichita Bar-Ad, M.D.,* Joshua Siglin, M.D.,* David Cognetti, M.D.,y Joseph Curry, M.D.,y Madalina Tuluc, M.D.,z and Amy S. Harrison, M.S.* *Department of Radiation Oncology, yDepartment of Otolaryngology, and zDepartment of Pathology, Jefferson Medical College of Thomas Jefferson University, Philadelphia, PA
A R T I C L E I N F O
A B S T R A C T
Article history: Received 31 May 2012 Accepted 31 October 2012
To quantify clinical differences for volumetric modulated arc therapy (VMAT) versus intensity modulated radiation therapy (IMRT) in terms of dosimetric endpoints and planning and delivery time, twenty head and neck cancer patients have been considered for VMAT using Nucletron Oncentra MasterPlan delivered via an Elekta linear accelerator. Differences in planning time between IMRT and VMAT were estimated accounting for both optimization and calculation. The average delivery time per patient was obtained retrospectively using the record and verify software. For the dosimetric comparison, all contoured organs at risk (OARs) and planning target volumes (PTVs) were evaluated. Of the 20 cases considered, 14 had VMAT plans approved. Six VMAT plans were rejected due to unacceptable dose to OARs. In terms of optimization time, there was minimal difference between the two modalities. The dose calculation time was significantly longer for VMAT, 4 minutes per 358 degree arc versus 2 minutes for an entire IMRT plan. The overall delivery time was reduced by 9.2 ⫾ 3.9 minutes for VMAT (51.4 ⫾ 15.6%). For the dosimetric comparison of the 14 clinically acceptable plans, there was almost no statistical difference between the VMAT and IMRT. There was also a reduction in monitor units of approximately 32% from IMRT to VMAT with both modalities demonstrating comparable quality assurance results. VMAT provides comparable coverage of target volumes while sparing OARs for the majority of head and neck cases. In cases where high dose modulation was required for OARs, a clinically acceptable plan was only achievable with IMRT. Due to the long calculation times, VMAT plans can cause delays during planning but marked improvements in delivery time reduce patient treatment times and the risk of intra-fraction motion. & 2013 American Association of Medical Dosimetrists.
Keywords: VMAT IMRT Dosimetry Head and neck cancer
Introduction Modern radiation therapy has evolved into an extremely complex process with new imaging modalities, delivery systems, and patient immobilization devices. These technological advances have made it possible to reduce the dose to normal tissue structures and consequently minimize the risk of toxicity and morbidity, while allowing for dose escalation to the tumor volumes, potentially leading to improved locoregional control. The most common method of head and neck cancer radiotherapy treatment is intensity modulated radiation therapy (IMRT). A standard IMRT plan requires multiple fixed-angle radiation beams
Reprint requests to: Matthew T. Studenski, Ph.D. 111 S. 11th Street Room, G-321 Gibbon Bldg., Philadelphia, PA 19107. E-mail: [email protected]
with many treatment segments, leading to lengthy treatment delivery times. Extended time on the table reduces comfort and increases intrafraction motion of the tumor or organs at risk (OARs). Recently, volumetric modulated arc therapy (VMAT) has gained popularity due to improved delivery efficiency over IMRT, as this modality introduces extra degrees of freedom in the optimization process in the form of dose rate and gantry rotation speed modulation.1,2 Recent studies have examined dosimetric differences between IMRT and VMAT for head and neck cancers, with the consensus being that VMAT produces a similar or superior plan to IMRT with a more efficient delivery.3-11 However, little literature exists regarding how the introduction of head and neck VMAT in a clinic affects the workflow. Here, the first year of VMAT implementation at our institution is discussed in terms of difficulties in achieving an acceptable plan for some cases, how much time was required for both treatment planning and delivery, and the overall quality of the plans.
0958-3947/$ – see front matter Copyright Ó 2013 American Association of Medical Dosimetrists http://dx.doi.org/10.1016/j.meddos.2012.10.009
172
M.T. Studenski et al. / Medical Dosimetry 38 (2013) 171–175
Methods and Materials Patient selection In the first year of clinical implementation, 20 patients with head and neck cancer were considered for VMAT. These patients represented a variety of disease sites: squamous cell carcinoma of the tonsil (2), base of tongue (7), oral cavity (2), paranasal sinus (2), nasopharynx (1) and maxillary sinus (1), tracheal carcinoma (1), diffuse B-cell lymphoma (1), left ear squamous cell carcinoma (1), left ear skin melanoma (1), and skin basal cell carcinoma with orbital involvement (1). All patients were simulated supine using a large Aquaplast mask (Civco Medical Soutions, Kalona, IA) holding their chin in a neutral position on a GE LightSpeed 16-slice CT scanner (GE Medical Systems, Milwaukee, WI). Treatment planning and delivery As per departmental protocol for implementing a new technology, these patients had a standard IMRT (Elekta XIO v4.62, Eletka AB, Stockholm, Sweden) plan generated in addition to the VMAT (Nucletron Oncentra MasterPlan v4.1) plan. If the VMAT plan was incomplete at treatment start, the IMRT plan was initiated and the patient was later converted to VMAT. The initial IMRT plans were calculated in XIO using the superposition algorithm and a step-and-shoot technique. The VMAT plans were calculated using MasterPlan’s collapsed cone algorithm with GPU acceleration. VMAT control points were set at every 41. All plans were normalized so that at least 95% of each planning target volume (PTV) was covered by the prescription dose. This meant that some PTVs had coverage that was greater than 95%. All of the plans were delivered using an Elekta Synergy-S linear accelerator (Elekta Ltd., Crawley, UK) with a MLCi (80 noninterdigitating 1-cm leaves at isocenter) using 6 MV photons. The linac was equipped with Elekta’s PreciseBeam VMAT control system, allowing for a maximum dose rate of about 450 monitor unit(MU)/min and a maximum leaf speed of 1.5 cm/s. IMRT planning started with 5 coplanar beams, and additional beams were added until a satisfactory plan was achieved (up to 9 beams). Noncoplanar beam were allowed. VMAT planning started with 2 3581 arcs and the collimator angles set to 51 and 951 to reduce the tongue and groove effect. For a more efficient delivery, 1 arc rotated in the clockwise direction and the other counterclockwise. For simple, unilateral cases, 1 arc was removed, leaving a single 3581 arc, and for complex cases, a third, noncoplanar arc was added to help achieve the plan constraints, covering approximately 401. IMRT and VMAT comparison Planning time To quantify the effect of implementing VMAT on the clinical workflow, the required planning time was estimated accounting for both optimization and calculation time using both planning systems. Although it was difficult to quantify the planning time exactly due to planners having multiple cases at once, we tracked the number of iterations required for each plan by sequentially numbering the plans.
Delivery time In terms of the difference in delivery time between IMRT and VMAT, temporal data were obtained retrospectively using record and verify software (Elekta MOSAIQ, Elekta AB, Stockholm, Sweden), which accounted for any machine errors and patient repositioning for noncoplanar beams. The average time that the patient was on the treatment table was calculated from all treatment fractions. Twelve patients were initially treated with IMRT and subsequently switched to VMAT during the treatment course. Dosimetry As with other dosimetric studies,3-11 the quality of the IMRT plans was compared to the VMAT plans. To eliminate any differences between the algorithms in XIO and MasterPlan, the dose in the IMRT plan was recalculated using the collapsed cone algorithm in MasterPlan. All contoured OARs and PTVs were evaluated (maximum dose, mean dose, and PTV D95%). As clinical dose prescriptions varied per disease site, results for all cases were rescaled to a dose of 70 Gy in 2 Gy fractions for consistency in the comparison. Statistical significance was established with a two-tailed paired t-test (p o 0.05). The total number of MUs delivered was also recorded. Quality assurance Patient-specific quality assurance (QA) results were obtained for both IMRT and VMAT plans. IMRT QA was carried out using the MapCheck diode array (Sun Nuclear, Melbourne FL) placed on the table at isocenter with the gantry at the nominal position. For VMAT QA, the Delta4 diode array (Scandidos AB, Uppsala, Sweden) was used with the arc delivered as during treatment except for removal of any couch kicks. For the first 10 VMAT cases, an ion chamber point measurement was also obtained for the VMAT plans.
Results Of the 20 cases considered, 14 had VMAT plans approved. Pertinent results can be seen in Table 1. Six VMAT plans were rejected due to unacceptable dose to critical structures. The beam arrangements used for the IMRT and VMAT plans in each of these cases and some relevant dosimetric endpoints can be found in Table 2. Table 2 shows that for some dosimetric endpoints, the VMAT plan was superior to the IMRT plan but was still rejected. Acceptable plans were the most difficult to achieve for primary tumors of the nasopharynx and sinuses. Of the 6 rejected plans, 4 fell into these locations. In 1 maxillary sinus case (patient 20), adding a constraint to reduce the lens dose degraded the PTV coverage to an unacceptable level. A rejection reason for 1 paranasal sinus case (patient 16) was that the chiasm dose exceeded the tolerance range, and the reason for the rejection of
Table 1 Summary of relevant results for the 20 cases analyzed Patient
1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20
Site
Lt. tonsil squamous cell carcinoma Lt. tongue squamous cell carcinoma BOT squamous cell carcinoma Lt. ear melanoma (5 fxs) Nasopharyngal cancer Floor of mouth squamous cell carcinoma Rt. Tonsil squamous cell carcinoma Lt. BOT squamous cell carcinoma BOT squamous cell carcinoma Paranasal sinus squamous cell carcinoma Tracheal carcinoma Lt. ear squamous cell carcinoma B call lymphoma Lt. forehead/orbit basal cell carcinoma BOT squamous cell carcinoma Rt. Paranasal papillary squamous cell carcinoma BOT squamous cell carcinoma BOT squamous cell carcinoma Oral cavity squamous cell carcinoma Maxillary sinus squamous cell carcinoma
BOT ¼ base of tongue, Lt ¼ left, Rt ¼ right.
VMAT accepted
Y Y Y Y N Y N Y Y N Y Y Y Y N N Y Y Y N
Reason for VMAT rejection
Larynx maximum dose Low dose volume to plexus & mandible
Separated PTV
Global maximum dose Optic chaism maximum dose
Maximum lens dose
Treatment time (min) IMRT
VMAT
23.6 23.0 18.9 34.3 24.9 15.8 22.3 19.8 16.6 25.3 11.6 12.3 23.0 11.9 31.8 23.3 19.7 18.1 16.8 33.1
10.7 11.3 4.0 25.8 9.9 9.8 13.7 6.5 4.0 8.6 5.4 8.7 9.3 9.7 -
M.T. Studenski et al. / Medical Dosimetry 38 (2013) 171–175
173
Table 2 Plan details for the patients with rejected VMAT plans Beam arrangement
Dose (Gy)
Relevant dosimetric
Patient IMRT
VMAT
5
9 beams (aso)
2 full arcs þ 1801 coplanar
7
7 beams (aio)
2 full þ 501 noncoplanar
10
9 beams (aso & aio)
-
15
8 beams (aio)
2 full arcs
16
8 beams (aso & aio)
1 full arc þ 2 1651 noncoplanar
20
9 beams (aso)
2 full arcs þ 1001 noncoplanar
the other paranasal sinus case (patient 10) was a large separation between PTVs resulting in the multileaf collimator (MLC) leaves not covering the multiple targets without interdigitating. The nasopharyngal case (patient 5) was rejected because the larynx dose was above the constraints. The other 2 cases that were rejected were a base of tongue squamous cell carcinoma (patient 15) where the global maximum fell outside of the PTV and a tonsil squamous cell carcinoma (patient 7) where the dose to the brachial plexus was unacceptably high. In an attempt to move the global maximum dose, weighting and virtual structures (rings and tuners) were used unsuccessfully.
Results
IMRT
VMAT
Larynx mean Brainstem mean Oral cavity max Lens max Optic chiasm max Brachial plexus mean Brainstem mean Larynx max Oral cavity mean Oral cavity max Esophagus mean Separated PTV Global max outside of PTV Parotid mean Oral cavity mean Mandible max Optic chiasm max Brainstem max Parotid mean Optic chiasm max Optic nerve max Esophagus mean Larynx mean
35.5 29.6 66.6 5.7 47.8 54.7 47.3 56.9 33.1 58.6 30.6
44.7 45.3 57.7 18.6 54.4 60.7 37.6 60.7 25.5 64.6 34.9 no VMAT plan
68.1 47.7 44.9 66.2 38.7 54.4 15.7 34.4 49.9 55.5 39.8
75.5 33.6 37.5 75.5 54.6 43.9 11.2 38.6 74.9 72.4 68.1
Treatment delivery time The treatment time was reduced by 9.2 ⫾ 3.9 minutes for VMAT over IMRT, an average reduction of 51.4 ⫾ 15.6%. The maximum time reduction was 15 minutes (78.8%) and the minimum was 2.9 minutes (17.5%). Table 1 shows the differences in average treatment time for patients treated with both IMRT and VMAT plans. This is a significant change that reduces the time the patient is in the immobilization mask on the table. There was also a significant reduction in MUs delivered, 289.3 ⫾ 179.9 MUs (32.9 ⫾ 14.3%) for the VMAT plan, when compared to the IMRT plans. The maximum reduction was 541 MU (48.9%) and the minimum was 54 MU (7.4%).
Planning time Dosimetry In terms of the effect on the clinical workflow, the optimization time was similar for both IMRT and VMAT. Each round of optimization took between 5 and 10 minutes depending on the complexity of the case. The authors also found that regardless of being IMRT or VMAT, the number of iterations required to achieve an acceptable plan was similar, that is, the complexity of the case, not the modality used, determined how many iterations needed to be run. For simple cases, an acceptable plan was frequently found in the first or second iteration of optimization. For complex cases, up to 10 iterations were run prior to the development of an acceptable plan. The dose calculation time was significantly longer for VMAT: 4 minutes per 3581 arc vs 2 minutes for an entire IMRT plan. For complex VMAT plans with 3 arcs, the calculation time was about 10 minutes. Additionally, the workflow in MasterPlan used a second optimization and dose calculation to achieve an acceptable plan, bringing the time per iteration up to 40 minutes for some complex cases (10-minute optimization plus 10-minute calculation, twice). The second optimization and calculation was carried out taking into account the more accurate collapsed cone dose calculation to compensate for areas that did not meet plan constraints, as the optimization used a less accurate but faster pencil beam algorithm. Overall, the planning time per iteration was about 3 times longer for VMAT when compared to IMRT.
For the dosimetric comparison between the VMAT and IMRT plans, several endpoints were analyzed for each OAR from the clinically acceptable plans, with all the cases rescaled to a prescription dose of 70 Gy for consistency. Thus, the values do not represent the actual dose delivered or the constraints applied to each OAR. A statistically significant difference was only found for the parotid gland where the maximum dose was 55.1 Gy vs 49.8 Gy and the mean dose was 29.3 Gy vs 26.6 Gy for IMRT and VMAT, respectively. For all other OARs, Figures 1 and 2 show the average maximum and mean dose between IMRT and VMAT for all patients. These results show that for some dosimetric endpoints, IMRT provides superior results, whereas for others, VMAT provides improved dosimetry. Quality assurance Using the gamma criteria with a dose difference of 3% and a distance-to-agreement of 3 mm, the results of the patient-specific QA were similar for IMRT and VMAT. The average pass rate for the IMRT plans was 92.0% and 96.1% for VMAT (85% is the departmental threshold to pass). This slight increase was statistically insignificant, and we suspect that it is related to the total number of measurement points, which is higher for the Delta4 with 2
174
M.T. Studenski et al. / Medical Dosimetry 38 (2013) 171–175
Fig. 1. Average maximum dose to selected OARs for IMRT and VMAT for all patients. Error bars represent 1 standard deviation.
diode arrays rather than 1. For the ion chamber point measurement, the plan was delivered to the solid water phantom in a manner attempting to keep the chamber in a high-dose, lowgradient region. This was possible for 7 of the 10 cases. For these 7 cases, the average chamber deviation from expected was 0.75%. For the other 3 cases, the average deviation was 4.3% with an overall average of 1.9%. A 3% deviation was the tolerance for these measurements, but for the 3 falling outside this range, the chamber was in a high-gradient dose region increasing the uncertainty in the expected dose. Although outside of the tolerance, these results were considered acceptable accounting for the increased uncertainty in the expected dose and the passing Delta4 measurement.
Discussion Various studies have assessed the benefits of VMAT when compared to IMRT in terms of dosimetric parameters, total MUs,
and delivery time, but they have all failed to address the impact that implementing a new procedure can have clinically.3-11 All conclude that VMAT reduces the total MUs and treatment time, while maintaining similar dosimetric endpoints in terms of PTV coverage and OAR sparing.3-11 The current study is unique in that it focuses on the clinical aspect of switching from IMRT to VMAT for treating head and neck cancer. For the 14 patients who were treated with VMAT plans, the results of the present study are very similar to data published previously. The dosimetric comparison showed minimal difference between VMAT and IMRT plans in terms of PTV coverage and OAR sparing. There was a significant reduction in MUs and in the time that the patient was on the treatment table. The reduction in MUs is one of the benefits of VMAT as this can potentially reduce the risk of secondary malignancies.12-14 The reduction in time on the treatment table is also important in terms of comfort as the patient is not in the immobilization mask for a long period of time, there is a reduced probability of intrafraction motion, and there is increased patient throughput.
Fig. 2. Average mean dose to selected OARs for IMRT and VMAT for all patients. Error bars represent 1 standard deviation.
M.T. Studenski et al. / Medical Dosimetry 38 (2013) 171–175
A key finding of the current study is the number of cases (6 out of 20) for which a clinically acceptable VMAT plan was not achieved. One explanation for this is that a learning curve exists for VMAT planning. With more experience, some of these cases might have been accepted. This same reasoning applies to the physician making the clinical decision. As the physicians became more familiar with the VMAT plans, they accepted cases that they might have rejected at the beginning. With experience, the physicians became more comfortable with the dose distributions and were willing to push their constraints to have a faster treatment delivery. When we first implemented VMAT, the constraints were absolute. That being said, a recurring issue throughout the first year with the VMAT planning was the inability to spare small structures such as the lens, optic chiasm, or larynx and also to control the hot spots. Only 1 other study has reported similar results with inferior PTV coverage between the orbits and reduced lens sparing for VMAT.11 We suspect that 1 reason for this observation stems from the fact that with IMRT, ideal gantry and collimator angles can be selected for each beam to optimize OAR sparing. Many segments can be delivered at these ideal gantry angles, whereas with VMAT, each arc is limited to a single collimator angle and the gantry can only slow down, not stop, to deliver a high dose rate at the optimal angles. This hypothesis is supported by observing VMAT plans where a higher dose is delivered at beam angles similar to those selected for the IMRT plans. Directly related to this is the ability of IMRT to apply fluence painting manual optimizations. Because the IMRT beam is fixed, by painting down or reducing the fluence directly over an OAR or hot spot, the segmentation algorithm blocks this area more effectively then VMAT where the gantry and MLC leaves are constantly moving. Fluence painting is not possible in VMAT, and increasing the weight to a small structure usually resulted in an unacceptable reduction in PTV coverage. Adding noncoplanar arcs can assist in VMAT planning as it provides additional collimator and gantry angles that allow for more effective optimization. Unlike the full 3581 arcs, the noncoplanar arcs should be designed with a couch kick and arc length to maximize PTV coverage and OAR sparing just as IMRT beams are selected for the same reasons. Additionally, plan quality could have been affected by the 1cm MLC leaves, although IMRT was able to achieve the constraints on the same machine and other studies also used MLCs with 1-cm leaves and were able to achieve acceptable plans.4,5,8 Another explanation could be deficiencies in the planning system’s segmentation algorithm or the linear accelerator’s MLC control mechanisms that cannot achieve the high degree of modulation required. A final explanation could be the higher integral dose delivered by the VMAT plans, increasing the strain on the optimizer to bring down the maximum dose for these small OARs. Although VMAT can significantly improve clinical workflow at the treatment machine, it can create problems that suspend the treatment planning process. The planning process is about 3 times longer for VMAT, which is significant for complex cases. For some patients, the treatment delivery might be postponed if a backup plan is not made. This is an especially relevant issue in patients initiating concurrent chemotherapy. The final component of the clinical implementation of VMAT is the analysis of patient-specific QA and how it affects the clinical workflow. Our QA passing results corroborate previous reports.3,4,8,9 In terms of the effect on workflow, VMAT plans must be recalculated on a phantom, which will take longer than the IMRT procedure. However, a portion of this time is regained as the delivery will be faster for VMAT.
175
It is also important to understand what the QA results mean before treating patients clinically. At our institution, the QA is analyzed using gamma analysis with absolute, rather than relative, dose. Depending on the method used, the type of phantom is important as a planar array of diodes or ion chambers can have significant angular dependence, making absolute dose evaluation difficult with an arc delivery.15,16 These arrays can be supplemented with another device such as an ion chamber or film for an improved measure of absolute dose, although this will result in additional QA time. A phantom with a nonplanar array can greatly improve QA efficiency. Whether treating with IMRT or VMAT, patient-specific QA is vital to safe and effective patient care.
Conclusions VMAT provides comparable coverage of target volumes and sparing of the majority of OARs for most head and neck cases. In certain cases where high dose modulation is required around small OARs, a clinically acceptable plan may only be achievable with IMRT. As VMAT requires a longer calculation time, difficult plans can greatly reduce clinical efficiency and should be chosen carefully. The greatest benefit of VMAT is a rapid delivery allowing improved patient comfort, reduced intrafraction motion, and increased patient throughput. References 1. Bedford, J.L. Treatment planning for volumetric modulated arc therapy. Med. Phys. 36:5128–38; 2009. 2. Teoh, M.; Clark, C.H.; Wook, K.; et al. Volumetric modulated arc therapy: A review of current literature and clinical use in practice. Br. J. Radiol. 84:967–96; 2011. 3. Alvarez-Moret, J.; Pohl, F.; Koelbl, O.; et al. Evaluation of volumetric modulated arc therapy (VMAT) with Oncentra MasterPlans for the treatment of head and neck cancer. Radiat. Oncol. 5:110; 2010. 4. Bertelsen, A.; Hansen, C.R.; Johansen, J.; et al. Single arc volumetric modulated arc therapy of head and neck cancer. Radiother. Oncol. 95:142–8; 2010. 5. Clemente, S.; Wu, B.; Sanguineti, G.; et al. SmartArc-Based volumetric modulated arc therapy for otopharyngeal cancer: A dosimetric comparison with both intensity-modulated radiation therapy and helical tomotherapy. Int. J. Radiat. Oncology. Biol. Phys. 80:1248–55; 2011. 6. Johnston, M.; Clifford, S.; Bromley, R.; et al. Volumetric-modulated arc therapy in head and neck radiotherapy: A planning comparison using simultaneous integrated boost for nasopharynx and oropharynx carcinoma. Clin. Oncol. 23:503–11; 2011. 7. Wiezorek, T.; Brachwitz, T.; Georg, D.; et al. Rotational IMRT techniques compared to fixed gantry IMRT and Tomotherapy: Multi-institutional planning study for head-and-neck cases. Radiat. Oncol. 6:20; 2011. 8. Rao, M.; Yang, W.; Chen, F.; et al. Comparison of Elekta VMAT with helical tomotherapy and fixed field IMRT: Plan quality, delivery efficiency and accuracy. Med. Phys. 37:1350–9; 2010. 9. Verbakel W.F., Cuijpers J.P., Hoffmans D., et al. Volumetric intensitymodulated arc therapy vs. conventional IMRT in head-and-neck cancer: A comparative planning and dosimetric study. Int. J. Radiat. Oncol. Biol. Phys. 2009;74:252–259. 10. Vanetti, E.; Clivio, A.; Nicolini, G.; et al. Volumetric modulated arc radiotherapy for carcinomas of the oro-pharynx, hypo-pharynx and larynx: A treatment planning comparison with fixed field IMRT. Radiother. Oncol. 92:111–7; 2009. 11. Guckenberger, M.; Richter, A.; Krieger, T.; et al. Is a single arc sufficient in volumetric modulated arc therapy (VMAT) for complex-shaped target volumes? Radiother. Oncol. 93:259–65; 2009. 12. Hall, E.J.; Wuu, C.S. Radiation-induced second cancers: The impact of 3D-CRT and IMRT. Int. J. Radiat. Oncol. Biol. Phys. 56:83–8; 2003. 13. Verellen, D.; Vanhavere, F. Risk assessment of radiation induced malignancies based on whole-body equivalent dose estimates for IMRT treatment in the head and neck region. Radiother. Oncol. 53:199–203; 1999. 14. Ruben, J.D.; Davis, S.; Evans, C.; et al. The effect of intensity-modulated radiotherapy on radiation-induced second malignancies. Int. J. Radiat. Oncol. Biol. Phys. 70:1530–6; 2008. 15. Jursinic, P. Angular dependence of dose sensitivity of surface diodes. Med. Phys. 36:2165–71; 2009. 16. O’Daniel, J.; Das, S.; Wu, Q.J.; et al. Volumetric-modulated arc therapy: Effective and efficient end-to-end patient-specific quality assurance. Int. J. Radiat. Oncol. Biol. Phys. 82:1567–74; 2012.
Medical Dosimetry journal homepage: www.meddos.org
Clinical experience transitioning from IMRT to VMAT for head and neck cancer Matthew T. Studenski, Ph.D.,* Voichita Bar-Ad, M.D.,* Joshua Siglin, M.D.,* David Cognetti, M.D.,y Joseph Curry, M.D.,y Madalina Tuluc, M.D.,z and Amy S. Harrison, M.S.* *Department of Radiation Oncology, yDepartment of Otolaryngology, and zDepartment of Pathology, Jefferson Medical College of Thomas Jefferson University, Philadelphia, PA
A R T I C L E I N F O
A B S T R A C T
Article history: Received 31 May 2012 Accepted 31 October 2012
To quantify clinical differences for volumetric modulated arc therapy (VMAT) versus intensity modulated radiation therapy (IMRT) in terms of dosimetric endpoints and planning and delivery time, twenty head and neck cancer patients have been considered for VMAT using Nucletron Oncentra MasterPlan delivered via an Elekta linear accelerator. Differences in planning time between IMRT and VMAT were estimated accounting for both optimization and calculation. The average delivery time per patient was obtained retrospectively using the record and verify software. For the dosimetric comparison, all contoured organs at risk (OARs) and planning target volumes (PTVs) were evaluated. Of the 20 cases considered, 14 had VMAT plans approved. Six VMAT plans were rejected due to unacceptable dose to OARs. In terms of optimization time, there was minimal difference between the two modalities. The dose calculation time was significantly longer for VMAT, 4 minutes per 358 degree arc versus 2 minutes for an entire IMRT plan. The overall delivery time was reduced by 9.2 ⫾ 3.9 minutes for VMAT (51.4 ⫾ 15.6%). For the dosimetric comparison of the 14 clinically acceptable plans, there was almost no statistical difference between the VMAT and IMRT. There was also a reduction in monitor units of approximately 32% from IMRT to VMAT with both modalities demonstrating comparable quality assurance results. VMAT provides comparable coverage of target volumes while sparing OARs for the majority of head and neck cases. In cases where high dose modulation was required for OARs, a clinically acceptable plan was only achievable with IMRT. Due to the long calculation times, VMAT plans can cause delays during planning but marked improvements in delivery time reduce patient treatment times and the risk of intra-fraction motion. & 2013 American Association of Medical Dosimetrists.
Keywords: VMAT IMRT Dosimetry Head and neck cancer
Introduction Modern radiation therapy has evolved into an extremely complex process with new imaging modalities, delivery systems, and patient immobilization devices. These technological advances have made it possible to reduce the dose to normal tissue structures and consequently minimize the risk of toxicity and morbidity, while allowing for dose escalation to the tumor volumes, potentially leading to improved locoregional control. The most common method of head and neck cancer radiotherapy treatment is intensity modulated radiation therapy (IMRT). A standard IMRT plan requires multiple fixed-angle radiation beams
Reprint requests to: Matthew T. Studenski, Ph.D. 111 S. 11th Street Room, G-321 Gibbon Bldg., Philadelphia, PA 19107. E-mail: [email protected]
with many treatment segments, leading to lengthy treatment delivery times. Extended time on the table reduces comfort and increases intrafraction motion of the tumor or organs at risk (OARs). Recently, volumetric modulated arc therapy (VMAT) has gained popularity due to improved delivery efficiency over IMRT, as this modality introduces extra degrees of freedom in the optimization process in the form of dose rate and gantry rotation speed modulation.1,2 Recent studies have examined dosimetric differences between IMRT and VMAT for head and neck cancers, with the consensus being that VMAT produces a similar or superior plan to IMRT with a more efficient delivery.3-11 However, little literature exists regarding how the introduction of head and neck VMAT in a clinic affects the workflow. Here, the first year of VMAT implementation at our institution is discussed in terms of difficulties in achieving an acceptable plan for some cases, how much time was required for both treatment planning and delivery, and the overall quality of the plans.
0958-3947/$ – see front matter Copyright Ó 2013 American Association of Medical Dosimetrists http://dx.doi.org/10.1016/j.meddos.2012.10.009
172
M.T. Studenski et al. / Medical Dosimetry 38 (2013) 171–175
Methods and Materials Patient selection In the first year of clinical implementation, 20 patients with head and neck cancer were considered for VMAT. These patients represented a variety of disease sites: squamous cell carcinoma of the tonsil (2), base of tongue (7), oral cavity (2), paranasal sinus (2), nasopharynx (1) and maxillary sinus (1), tracheal carcinoma (1), diffuse B-cell lymphoma (1), left ear squamous cell carcinoma (1), left ear skin melanoma (1), and skin basal cell carcinoma with orbital involvement (1). All patients were simulated supine using a large Aquaplast mask (Civco Medical Soutions, Kalona, IA) holding their chin in a neutral position on a GE LightSpeed 16-slice CT scanner (GE Medical Systems, Milwaukee, WI). Treatment planning and delivery As per departmental protocol for implementing a new technology, these patients had a standard IMRT (Elekta XIO v4.62, Eletka AB, Stockholm, Sweden) plan generated in addition to the VMAT (Nucletron Oncentra MasterPlan v4.1) plan. If the VMAT plan was incomplete at treatment start, the IMRT plan was initiated and the patient was later converted to VMAT. The initial IMRT plans were calculated in XIO using the superposition algorithm and a step-and-shoot technique. The VMAT plans were calculated using MasterPlan’s collapsed cone algorithm with GPU acceleration. VMAT control points were set at every 41. All plans were normalized so that at least 95% of each planning target volume (PTV) was covered by the prescription dose. This meant that some PTVs had coverage that was greater than 95%. All of the plans were delivered using an Elekta Synergy-S linear accelerator (Elekta Ltd., Crawley, UK) with a MLCi (80 noninterdigitating 1-cm leaves at isocenter) using 6 MV photons. The linac was equipped with Elekta’s PreciseBeam VMAT control system, allowing for a maximum dose rate of about 450 monitor unit(MU)/min and a maximum leaf speed of 1.5 cm/s. IMRT planning started with 5 coplanar beams, and additional beams were added until a satisfactory plan was achieved (up to 9 beams). Noncoplanar beam were allowed. VMAT planning started with 2 3581 arcs and the collimator angles set to 51 and 951 to reduce the tongue and groove effect. For a more efficient delivery, 1 arc rotated in the clockwise direction and the other counterclockwise. For simple, unilateral cases, 1 arc was removed, leaving a single 3581 arc, and for complex cases, a third, noncoplanar arc was added to help achieve the plan constraints, covering approximately 401. IMRT and VMAT comparison Planning time To quantify the effect of implementing VMAT on the clinical workflow, the required planning time was estimated accounting for both optimization and calculation time using both planning systems. Although it was difficult to quantify the planning time exactly due to planners having multiple cases at once, we tracked the number of iterations required for each plan by sequentially numbering the plans.
Delivery time In terms of the difference in delivery time between IMRT and VMAT, temporal data were obtained retrospectively using record and verify software (Elekta MOSAIQ, Elekta AB, Stockholm, Sweden), which accounted for any machine errors and patient repositioning for noncoplanar beams. The average time that the patient was on the treatment table was calculated from all treatment fractions. Twelve patients were initially treated with IMRT and subsequently switched to VMAT during the treatment course. Dosimetry As with other dosimetric studies,3-11 the quality of the IMRT plans was compared to the VMAT plans. To eliminate any differences between the algorithms in XIO and MasterPlan, the dose in the IMRT plan was recalculated using the collapsed cone algorithm in MasterPlan. All contoured OARs and PTVs were evaluated (maximum dose, mean dose, and PTV D95%). As clinical dose prescriptions varied per disease site, results for all cases were rescaled to a dose of 70 Gy in 2 Gy fractions for consistency in the comparison. Statistical significance was established with a two-tailed paired t-test (p o 0.05). The total number of MUs delivered was also recorded. Quality assurance Patient-specific quality assurance (QA) results were obtained for both IMRT and VMAT plans. IMRT QA was carried out using the MapCheck diode array (Sun Nuclear, Melbourne FL) placed on the table at isocenter with the gantry at the nominal position. For VMAT QA, the Delta4 diode array (Scandidos AB, Uppsala, Sweden) was used with the arc delivered as during treatment except for removal of any couch kicks. For the first 10 VMAT cases, an ion chamber point measurement was also obtained for the VMAT plans.
Results Of the 20 cases considered, 14 had VMAT plans approved. Pertinent results can be seen in Table 1. Six VMAT plans were rejected due to unacceptable dose to critical structures. The beam arrangements used for the IMRT and VMAT plans in each of these cases and some relevant dosimetric endpoints can be found in Table 2. Table 2 shows that for some dosimetric endpoints, the VMAT plan was superior to the IMRT plan but was still rejected. Acceptable plans were the most difficult to achieve for primary tumors of the nasopharynx and sinuses. Of the 6 rejected plans, 4 fell into these locations. In 1 maxillary sinus case (patient 20), adding a constraint to reduce the lens dose degraded the PTV coverage to an unacceptable level. A rejection reason for 1 paranasal sinus case (patient 16) was that the chiasm dose exceeded the tolerance range, and the reason for the rejection of
Table 1 Summary of relevant results for the 20 cases analyzed Patient
1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20
Site
Lt. tonsil squamous cell carcinoma Lt. tongue squamous cell carcinoma BOT squamous cell carcinoma Lt. ear melanoma (5 fxs) Nasopharyngal cancer Floor of mouth squamous cell carcinoma Rt. Tonsil squamous cell carcinoma Lt. BOT squamous cell carcinoma BOT squamous cell carcinoma Paranasal sinus squamous cell carcinoma Tracheal carcinoma Lt. ear squamous cell carcinoma B call lymphoma Lt. forehead/orbit basal cell carcinoma BOT squamous cell carcinoma Rt. Paranasal papillary squamous cell carcinoma BOT squamous cell carcinoma BOT squamous cell carcinoma Oral cavity squamous cell carcinoma Maxillary sinus squamous cell carcinoma
BOT ¼ base of tongue, Lt ¼ left, Rt ¼ right.
VMAT accepted
Y Y Y Y N Y N Y Y N Y Y Y Y N N Y Y Y N
Reason for VMAT rejection
Larynx maximum dose Low dose volume to plexus & mandible
Separated PTV
Global maximum dose Optic chaism maximum dose
Maximum lens dose
Treatment time (min) IMRT
VMAT
23.6 23.0 18.9 34.3 24.9 15.8 22.3 19.8 16.6 25.3 11.6 12.3 23.0 11.9 31.8 23.3 19.7 18.1 16.8 33.1
10.7 11.3 4.0 25.8 9.9 9.8 13.7 6.5 4.0 8.6 5.4 8.7 9.3 9.7 -
M.T. Studenski et al. / Medical Dosimetry 38 (2013) 171–175
173
Table 2 Plan details for the patients with rejected VMAT plans Beam arrangement
Dose (Gy)
Relevant dosimetric
Patient IMRT
VMAT
5
9 beams (aso)
2 full arcs þ 1801 coplanar
7
7 beams (aio)
2 full þ 501 noncoplanar
10
9 beams (aso & aio)
-
15
8 beams (aio)
2 full arcs
16
8 beams (aso & aio)
1 full arc þ 2 1651 noncoplanar
20
9 beams (aso)
2 full arcs þ 1001 noncoplanar
the other paranasal sinus case (patient 10) was a large separation between PTVs resulting in the multileaf collimator (MLC) leaves not covering the multiple targets without interdigitating. The nasopharyngal case (patient 5) was rejected because the larynx dose was above the constraints. The other 2 cases that were rejected were a base of tongue squamous cell carcinoma (patient 15) where the global maximum fell outside of the PTV and a tonsil squamous cell carcinoma (patient 7) where the dose to the brachial plexus was unacceptably high. In an attempt to move the global maximum dose, weighting and virtual structures (rings and tuners) were used unsuccessfully.
Results
IMRT
VMAT
Larynx mean Brainstem mean Oral cavity max Lens max Optic chiasm max Brachial plexus mean Brainstem mean Larynx max Oral cavity mean Oral cavity max Esophagus mean Separated PTV Global max outside of PTV Parotid mean Oral cavity mean Mandible max Optic chiasm max Brainstem max Parotid mean Optic chiasm max Optic nerve max Esophagus mean Larynx mean
35.5 29.6 66.6 5.7 47.8 54.7 47.3 56.9 33.1 58.6 30.6
44.7 45.3 57.7 18.6 54.4 60.7 37.6 60.7 25.5 64.6 34.9 no VMAT plan
68.1 47.7 44.9 66.2 38.7 54.4 15.7 34.4 49.9 55.5 39.8
75.5 33.6 37.5 75.5 54.6 43.9 11.2 38.6 74.9 72.4 68.1
Treatment delivery time The treatment time was reduced by 9.2 ⫾ 3.9 minutes for VMAT over IMRT, an average reduction of 51.4 ⫾ 15.6%. The maximum time reduction was 15 minutes (78.8%) and the minimum was 2.9 minutes (17.5%). Table 1 shows the differences in average treatment time for patients treated with both IMRT and VMAT plans. This is a significant change that reduces the time the patient is in the immobilization mask on the table. There was also a significant reduction in MUs delivered, 289.3 ⫾ 179.9 MUs (32.9 ⫾ 14.3%) for the VMAT plan, when compared to the IMRT plans. The maximum reduction was 541 MU (48.9%) and the minimum was 54 MU (7.4%).
Planning time Dosimetry In terms of the effect on the clinical workflow, the optimization time was similar for both IMRT and VMAT. Each round of optimization took between 5 and 10 minutes depending on the complexity of the case. The authors also found that regardless of being IMRT or VMAT, the number of iterations required to achieve an acceptable plan was similar, that is, the complexity of the case, not the modality used, determined how many iterations needed to be run. For simple cases, an acceptable plan was frequently found in the first or second iteration of optimization. For complex cases, up to 10 iterations were run prior to the development of an acceptable plan. The dose calculation time was significantly longer for VMAT: 4 minutes per 3581 arc vs 2 minutes for an entire IMRT plan. For complex VMAT plans with 3 arcs, the calculation time was about 10 minutes. Additionally, the workflow in MasterPlan used a second optimization and dose calculation to achieve an acceptable plan, bringing the time per iteration up to 40 minutes for some complex cases (10-minute optimization plus 10-minute calculation, twice). The second optimization and calculation was carried out taking into account the more accurate collapsed cone dose calculation to compensate for areas that did not meet plan constraints, as the optimization used a less accurate but faster pencil beam algorithm. Overall, the planning time per iteration was about 3 times longer for VMAT when compared to IMRT.
For the dosimetric comparison between the VMAT and IMRT plans, several endpoints were analyzed for each OAR from the clinically acceptable plans, with all the cases rescaled to a prescription dose of 70 Gy for consistency. Thus, the values do not represent the actual dose delivered or the constraints applied to each OAR. A statistically significant difference was only found for the parotid gland where the maximum dose was 55.1 Gy vs 49.8 Gy and the mean dose was 29.3 Gy vs 26.6 Gy for IMRT and VMAT, respectively. For all other OARs, Figures 1 and 2 show the average maximum and mean dose between IMRT and VMAT for all patients. These results show that for some dosimetric endpoints, IMRT provides superior results, whereas for others, VMAT provides improved dosimetry. Quality assurance Using the gamma criteria with a dose difference of 3% and a distance-to-agreement of 3 mm, the results of the patient-specific QA were similar for IMRT and VMAT. The average pass rate for the IMRT plans was 92.0% and 96.1% for VMAT (85% is the departmental threshold to pass). This slight increase was statistically insignificant, and we suspect that it is related to the total number of measurement points, which is higher for the Delta4 with 2
174
M.T. Studenski et al. / Medical Dosimetry 38 (2013) 171–175
Fig. 1. Average maximum dose to selected OARs for IMRT and VMAT for all patients. Error bars represent 1 standard deviation.
diode arrays rather than 1. For the ion chamber point measurement, the plan was delivered to the solid water phantom in a manner attempting to keep the chamber in a high-dose, lowgradient region. This was possible for 7 of the 10 cases. For these 7 cases, the average chamber deviation from expected was 0.75%. For the other 3 cases, the average deviation was 4.3% with an overall average of 1.9%. A 3% deviation was the tolerance for these measurements, but for the 3 falling outside this range, the chamber was in a high-gradient dose region increasing the uncertainty in the expected dose. Although outside of the tolerance, these results were considered acceptable accounting for the increased uncertainty in the expected dose and the passing Delta4 measurement.
Discussion Various studies have assessed the benefits of VMAT when compared to IMRT in terms of dosimetric parameters, total MUs,
and delivery time, but they have all failed to address the impact that implementing a new procedure can have clinically.3-11 All conclude that VMAT reduces the total MUs and treatment time, while maintaining similar dosimetric endpoints in terms of PTV coverage and OAR sparing.3-11 The current study is unique in that it focuses on the clinical aspect of switching from IMRT to VMAT for treating head and neck cancer. For the 14 patients who were treated with VMAT plans, the results of the present study are very similar to data published previously. The dosimetric comparison showed minimal difference between VMAT and IMRT plans in terms of PTV coverage and OAR sparing. There was a significant reduction in MUs and in the time that the patient was on the treatment table. The reduction in MUs is one of the benefits of VMAT as this can potentially reduce the risk of secondary malignancies.12-14 The reduction in time on the treatment table is also important in terms of comfort as the patient is not in the immobilization mask for a long period of time, there is a reduced probability of intrafraction motion, and there is increased patient throughput.
Fig. 2. Average mean dose to selected OARs for IMRT and VMAT for all patients. Error bars represent 1 standard deviation.
M.T. Studenski et al. / Medical Dosimetry 38 (2013) 171–175
A key finding of the current study is the number of cases (6 out of 20) for which a clinically acceptable VMAT plan was not achieved. One explanation for this is that a learning curve exists for VMAT planning. With more experience, some of these cases might have been accepted. This same reasoning applies to the physician making the clinical decision. As the physicians became more familiar with the VMAT plans, they accepted cases that they might have rejected at the beginning. With experience, the physicians became more comfortable with the dose distributions and were willing to push their constraints to have a faster treatment delivery. When we first implemented VMAT, the constraints were absolute. That being said, a recurring issue throughout the first year with the VMAT planning was the inability to spare small structures such as the lens, optic chiasm, or larynx and also to control the hot spots. Only 1 other study has reported similar results with inferior PTV coverage between the orbits and reduced lens sparing for VMAT.11 We suspect that 1 reason for this observation stems from the fact that with IMRT, ideal gantry and collimator angles can be selected for each beam to optimize OAR sparing. Many segments can be delivered at these ideal gantry angles, whereas with VMAT, each arc is limited to a single collimator angle and the gantry can only slow down, not stop, to deliver a high dose rate at the optimal angles. This hypothesis is supported by observing VMAT plans where a higher dose is delivered at beam angles similar to those selected for the IMRT plans. Directly related to this is the ability of IMRT to apply fluence painting manual optimizations. Because the IMRT beam is fixed, by painting down or reducing the fluence directly over an OAR or hot spot, the segmentation algorithm blocks this area more effectively then VMAT where the gantry and MLC leaves are constantly moving. Fluence painting is not possible in VMAT, and increasing the weight to a small structure usually resulted in an unacceptable reduction in PTV coverage. Adding noncoplanar arcs can assist in VMAT planning as it provides additional collimator and gantry angles that allow for more effective optimization. Unlike the full 3581 arcs, the noncoplanar arcs should be designed with a couch kick and arc length to maximize PTV coverage and OAR sparing just as IMRT beams are selected for the same reasons. Additionally, plan quality could have been affected by the 1cm MLC leaves, although IMRT was able to achieve the constraints on the same machine and other studies also used MLCs with 1-cm leaves and were able to achieve acceptable plans.4,5,8 Another explanation could be deficiencies in the planning system’s segmentation algorithm or the linear accelerator’s MLC control mechanisms that cannot achieve the high degree of modulation required. A final explanation could be the higher integral dose delivered by the VMAT plans, increasing the strain on the optimizer to bring down the maximum dose for these small OARs. Although VMAT can significantly improve clinical workflow at the treatment machine, it can create problems that suspend the treatment planning process. The planning process is about 3 times longer for VMAT, which is significant for complex cases. For some patients, the treatment delivery might be postponed if a backup plan is not made. This is an especially relevant issue in patients initiating concurrent chemotherapy. The final component of the clinical implementation of VMAT is the analysis of patient-specific QA and how it affects the clinical workflow. Our QA passing results corroborate previous reports.3,4,8,9 In terms of the effect on workflow, VMAT plans must be recalculated on a phantom, which will take longer than the IMRT procedure. However, a portion of this time is regained as the delivery will be faster for VMAT.
175
It is also important to understand what the QA results mean before treating patients clinically. At our institution, the QA is analyzed using gamma analysis with absolute, rather than relative, dose. Depending on the method used, the type of phantom is important as a planar array of diodes or ion chambers can have significant angular dependence, making absolute dose evaluation difficult with an arc delivery.15,16 These arrays can be supplemented with another device such as an ion chamber or film for an improved measure of absolute dose, although this will result in additional QA time. A phantom with a nonplanar array can greatly improve QA efficiency. Whether treating with IMRT or VMAT, patient-specific QA is vital to safe and effective patient care.
Conclusions VMAT provides comparable coverage of target volumes and sparing of the majority of OARs for most head and neck cases. In certain cases where high dose modulation is required around small OARs, a clinically acceptable plan may only be achievable with IMRT. As VMAT requires a longer calculation time, difficult plans can greatly reduce clinical efficiency and should be chosen carefully. The greatest benefit of VMAT is a rapid delivery allowing improved patient comfort, reduced intrafraction motion, and increased patient throughput. References 1. Bedford, J.L. Treatment planning for volumetric modulated arc therapy. Med. Phys. 36:5128–38; 2009. 2. Teoh, M.; Clark, C.H.; Wook, K.; et al. Volumetric modulated arc therapy: A review of current literature and clinical use in practice. Br. J. Radiol. 84:967–96; 2011. 3. Alvarez-Moret, J.; Pohl, F.; Koelbl, O.; et al. Evaluation of volumetric modulated arc therapy (VMAT) with Oncentra MasterPlans for the treatment of head and neck cancer. Radiat. Oncol. 5:110; 2010. 4. Bertelsen, A.; Hansen, C.R.; Johansen, J.; et al. Single arc volumetric modulated arc therapy of head and neck cancer. Radiother. Oncol. 95:142–8; 2010. 5. Clemente, S.; Wu, B.; Sanguineti, G.; et al. SmartArc-Based volumetric modulated arc therapy for otopharyngeal cancer: A dosimetric comparison with both intensity-modulated radiation therapy and helical tomotherapy. Int. J. Radiat. Oncology. Biol. Phys. 80:1248–55; 2011. 6. Johnston, M.; Clifford, S.; Bromley, R.; et al. Volumetric-modulated arc therapy in head and neck radiotherapy: A planning comparison using simultaneous integrated boost for nasopharynx and oropharynx carcinoma. Clin. Oncol. 23:503–11; 2011. 7. Wiezorek, T.; Brachwitz, T.; Georg, D.; et al. Rotational IMRT techniques compared to fixed gantry IMRT and Tomotherapy: Multi-institutional planning study for head-and-neck cases. Radiat. Oncol. 6:20; 2011. 8. Rao, M.; Yang, W.; Chen, F.; et al. Comparison of Elekta VMAT with helical tomotherapy and fixed field IMRT: Plan quality, delivery efficiency and accuracy. Med. Phys. 37:1350–9; 2010. 9. Verbakel W.F., Cuijpers J.P., Hoffmans D., et al. Volumetric intensitymodulated arc therapy vs. conventional IMRT in head-and-neck cancer: A comparative planning and dosimetric study. Int. J. Radiat. Oncol. Biol. Phys. 2009;74:252–259. 10. Vanetti, E.; Clivio, A.; Nicolini, G.; et al. Volumetric modulated arc radiotherapy for carcinomas of the oro-pharynx, hypo-pharynx and larynx: A treatment planning comparison with fixed field IMRT. Radiother. Oncol. 92:111–7; 2009. 11. Guckenberger, M.; Richter, A.; Krieger, T.; et al. Is a single arc sufficient in volumetric modulated arc therapy (VMAT) for complex-shaped target volumes? Radiother. Oncol. 93:259–65; 2009. 12. Hall, E.J.; Wuu, C.S. Radiation-induced second cancers: The impact of 3D-CRT and IMRT. Int. J. Radiat. Oncol. Biol. Phys. 56:83–8; 2003. 13. Verellen, D.; Vanhavere, F. Risk assessment of radiation induced malignancies based on whole-body equivalent dose estimates for IMRT treatment in the head and neck region. Radiother. Oncol. 53:199–203; 1999. 14. Ruben, J.D.; Davis, S.; Evans, C.; et al. The effect of intensity-modulated radiotherapy on radiation-induced second malignancies. Int. J. Radiat. Oncol. Biol. Phys. 70:1530–6; 2008. 15. Jursinic, P. Angular dependence of dose sensitivity of surface diodes. Med. Phys. 36:2165–71; 2009. 16. O’Daniel, J.; Das, S.; Wu, Q.J.; et al. Volumetric-modulated arc therapy: Effective and efficient end-to-end patient-specific quality assurance. Int. J. Radiat. Oncol. Biol. Phys. 82:1567–74; 2012.