- Email: [email protected]
A Prospective 4π Radiation Therapy Clinical Study in Recurrent High-Grade Glioma Patients
International Journal of
Radiation Oncology biology
physics
www.redjournal.org
Physics Contribution
A Prospective 4p Radiation Therapy Clinical Study in Recurrent High-Grade Glioma Patients Victoria Y. Yu, PhD, Angelia Landers, MS, Kaley Woods, MS, Dan Nguyen, PhD, Minsong Cao, PhD, Dongsu Du, PhD, Robert K. Chin, MD, PhD, Ke Sheng, PhD, and Tania B. Kaprealian, MD Department of Radiation Oncology, David Geffen School of Medicine, University of California, Los Angeles, Los Angeles, California Received Jun 23, 2017, and in revised form Dec 27, 2017. Accepted for publication Jan 12, 2018.
Summary 4p radiation therapy, a novel delivery technique utilizing inverse optimization for automatic non-coplanar beam orientation selection and fluence optimization, was shown capable of achieving dosimetry superior to conventional methods. This study summarizes the first prospective clinical trial of 4p radiation therapy in recurrent high-grade glioma patients. Substantial sparing to organs at risk was achieved compared with stateof-the-art VMAT plans, treatments were well tolerated, and intrafractional motion was <1.5 mm for all patients, motivating further clinical utilization of the technique.
Purpose: To evaluate the feasibility, safety, dosimetric benefits, delivery efficiency, and patient comfort in the clinical implementation of 4p radiation therapy. Methods and Materials: Eleven patients with recurrent high-grade glioma were recruited for the trial. 4 p plans integrating beam orientation and fluence-map optimization were created using an in-house column-generation algorithm. The collision-free beam solution space throughout the 4p steradian was determined using a computer-aided-design model of the Varian TrueBeam system and a human subject. Twenty beams were optimized for each case and imported into Eclipse for intensity modulated radiation therapy planning. Beam orientations with neighboring couch kicks were merged for increased delivery efficiency, generating plans with an average of 16 beam orientations. Volumetric modulated arc therapy (VMAT) plans with 3-4 arcs were also generated for each case, and the plan achieving superior dosimetric quality was selected for treatment. Patient comfort was surveyed after every fraction. Multiple 2-dimensional X-ray images were obtained to measure intrafractional motion. Results: Of 11 patients, 9 were treated with 4 p . Mean and maximum organ at risk doses were equal or significantly lower (P < .05) with 4p than with VMAT. Particularly substantial dose reduction of 2.92 Gy in the average accumulated brainstem maximum dose enabled treatments that would otherwise not satisfy safe dose constraints with VMAT. One patient was not treated because neither plan met the dosimetric criteria. The other was treated with VMAT owing to comparable dosimetry resulting from a planning target volume located in a separate co-plane superior to organs at risk. Treatments were well tolerated, with an average patient comfort score of 8.6/10. Intrafractional motion was <1.5 mm for all delivered fractions, and the average delivery time was 34.1 minutes.
Reprint requests to: Tania B. Kaprealian, MD, Department of Radiation Oncology, University of California, Los Angeles, 200 Medical Plaza,
Int J Radiation Oncol Biol Phys, Vol. 101, No. 1, pp. 144e151, 2018 0360-3016/$ - see front matter Ó 2018 Elsevier Inc. All rights reserved. https://doi.org/10.1016/j.ijrobp.2018.01.048
B265, Los Angeles, CA 90024. Tel: (310) 825-9775; E-mail: tkaprealian@ mednet.ucla.edu Conflict of interest: none.
Prospective 4p radiation therapy clinical study
Volume 101 Number 1 2018
145
Conclusions: The feasibility, safety, dosimetric benefits, delivery efficiency, and patient comfort of 4 p radiation therapy have been clinically demonstrated with a prospective clinical trial. The results elucidate the potential and challenges of wider clinical implementations. Ó 2018 Elsevier Inc. All rights reserved.
Introduction
Methods and Materials
Current state-of-the-art medical digital linear accelerators are capable of delivering radiation therapy with complex dynamic trajectories that involve orchestrated movements of the gantry, couch, and multileaf collimator (MLC), in conjunction with dose rate modulations. These new capabilities have motivated active research to explore the potential in further improving radiation therapy plan quality through optimization of beam orientations and dynamic trajectories, in which substantial dosimetric improvements have been shown (1-6). The optimal utilization of the non-coplanar beam solution space has been shown to better shape the spatial dose distribution. 4p radiation therapy is a novel planning and delivery technique integrating static intensity modulated radiation therapy (IMRT) beam orientation and fluence map optimization (7). Its dosimetry has been favorably compared with volumetric modulated arc therapy (VMAT) or IMRT with manually selected beam orientations in terms of the neighboring organs at risk (OARs) sparing for disease sites, including the brain (8), head and neck (9), lung (10), liver (11), and prostate (12, 13). The integral dose of highly noncoplanar radiation therapy has been shown to be comparable to or lower than that of clinical coplanar plans in both phantom simulations and patient case evaluations (14), but the patient volume receiving low-dose radiation may increase and should therefore be evaluated for curative or pediatric cases in which such dose volumes could be of clinical significance. The dosimetric benefit and feasibility of 4p radiation therapy were demonstrated through retrospective dosimetric evaluations and automated delivery in Varian Developer Mode (Varian Medical Systems, Palo Alto, CA) but have not yet been clinically implemented, partly owing to the lack of US Food and Drug Administration (FDA) cleared product. On the other hand, clinical implementations of 4p static beam radiation therapy, despite its complex coordinated couch and gantry movements, can be delivered in the clinical mode. This characteristic is distinctly different from several other experimental delivery methods using dynamic couch movements during beam-on (3, 5, 6, 15). In this study we demonstrate the first clinical implementation of 4p radiation therapy via a prospective clinical trial to test its feasibility, safety, dosimetric benefits, intrafractional motion, and delivery efficiency, and patient comfort on recurrent high-grade glioma patients.
Patient characteristics and treatment plan generation Under an institutional review boardeapproved phase 1 trial protocol, locally recurrent glioma was chosen as the first disease site for clinical translation of 4p radiation therapy, owing to the demanding dosimetric constraints to avoid over-irradiating previously treated normal tissue volumes (16-18). Specifically, these patients received 59.4 or 60 Gy from the original treatment before receiving another 25 Gy in 5 fractions or 30 Gy in 10 fractions to the recurrent tumor target volume that partially or completely overlapped with the original target volume. Because of this overlap, it is extremely challenging to meet the dose limits on critical organs, including the brainstem and optic apparatus. Additionally, the brain tumor location allows for access to more non-coplanar angles with greater clearance between the couch and gantry, making this patient cohort particularly well suited for the first prospective clinical trial of 4p radiation therapy. The recruitment inclusion criteria included a histologic diagnosis of primary and recurrent high-grade glioma, Karnofsky performance status >70, and age of 18 years or older. Eleven patients consented to participate in the clinical trial from December 2014 to January 2017. Patient and treatment characteristics of both the prior radiation therapy for the primary disease and the delivered plan during trial participation are shown in Table 1. Computed tomography (CT) simulation was performed with patients immobilized using a thermoplastic radiosurgical mask (Brainlab, Munich, Germany) in the supine position. The gross tumor volume was defined as the enhanced gross disease volume on the T1-weighted magnetic resonance image. The planning target volume (PTV) was then contoured by the physicians (TBK, RKC) as the gross tumor volume plus a 2- to 5-mm isotropic expansion, depending on each individual case. All contours were transferred from the magnetic resonance images to the CT domain through rigid image registration performed in MIM 6.6.5 (MIM Software, Cleveland, OH). With the CT image and the established PTV and OARs contours, a previously reported in-house 4p radiation therapy treatment planning optimization (7-9, 13) was used to select an optimal set of 20 beam angles for each patient. The number of beams for the optimization was empirically established as a balance between plan practicality and dose optimality. The number of beams for a
146
International Journal of Radiation Oncology Biology Physics
Yu et al.
Table 1
Patient and treatment plan characteristics, along with average delivery time breakdown for each patient Average delivery time breakdown (min)
Pt no. Histology Age (y) Sex 1 2 3 4 5 6 7 8 9
GBM GBM GBM GBM GBM GBM GBM EPD GBM
54 64 60 54 52 48 39 80 59
F F F M M M M F M
Prior dose (Gy)
PTV volume (cm3)
59.4 60 60 60 59.4 60 46 54 60
13.99 110.14 0.95 1.84 124.51 18.17 47.15 2.57 3.86
Dosing scheme 5 3 5 5 3 6 3 5 5
Gy Gy Gy Gy Gy Gy Gy Gy Gy
5 10 5 5 10 5 10 5 5
Total Dose (Gy)
Machine
25 30 25 25 30 30 30 25 25
TrueBeam TrueBeam Novalis Tx Novalis Tx TrueBeam TrueBeam TrueBeam TrueBeam Novalis Tx
Recruited but not treated with 4p 10 GBM 11 GBM
56 66
M M
59.4 10.94 3 Gy 10 40 þ 20 SIB 141.09 3 Gy 10
Motion No. axes No. couch Initial Beam fields kicks setup on Imaging mvnt Total 18 19 15 13 20 15 16 17 15
12 14 11 11 15 9 10 10 11
9.4 7.7 3.0 2.4 5.3 9.1 10.3 12.6 3.0
2.1 3.5 1.9 1.8 2.4 2.0 2.1 3.5 3.1
1.4 1.5 2.6 2.2 2.8 3.4 2.0 2.4 2.8
27.1 32.1 25.5 21.0 23.5* 12.9* 16.6* 15.9* 23.1
40.1 44.8 33.1 27.5 34.0 27.3 31.1 34.4 31.9
Reason 30 30
Neither plans met dose constraints Comparable VMAT plan
Abbreviations: EPD Z ependymoma; F Z female; GBM Z glioblastoma multiforme; M Z male; PTV Z planning target volume; SIB Z simultaneous integrated boost. Initial setup: average time taken for initial setup, imaging, and physician approval; Imaging: total intrafractional imaging time in between beam angles; Motion axes mvnt: couch and gantry movement time between beams; Total time: total delivery time. * Cases delivered with remote couch kick capabilities.
good balance could vary between patients, but to standardize our process, we currently choose 20. These beams were selected from a candidate pool containing 1162 evenly distributed non-coplanar beams with approximately 6 separation between adjacent beams. Guided by a detailed and accurate computer-aided model of the Varian TrueBeam (Varian Medical Systems) and a 3-dimensional surface image of a volunteer placed on its couch, beams causing collision or being blocked by the couch pedestal were excluded from optimization (19). 4p treatment optimization was performed on the basis of precomputed beamlets. The dose calculation was performed using collapsed cone convolution/superposition and Monte Carlo calculated 6-MV poly-energetic X-ray kernels (7). The dose calculation resolution was 2.5 2.5 2.5 mm3 for 5 5 mm2. A greedy column generation algorithm with user-defined upper dose limits and structure priority weightings was used to iteratively select beam orientations and perform fluence optimization with all selected beam orientations until the desired number of beam angles was reached. The objective function of the optimization problem is based on a linear approximation of an equivalent uniform dose, defined in detail in previous publications (7, 20, 21). To generate a clinically deliverable plan using an FDA-approved treatment planning system and delivery technique, the 20 beam orientations selected by 4p were imported into Eclipse v13.6.23 (Varian Medical Systems) for IMRT reoptimization. To avoid excessively long delivery time and reduce the number of couch kicks, beams orientations with less than 20 separation in both the couch and gantry angles are merged to take the average position.
Beams with couch angles within 6 were merged to use a single couch position while maintaining their own gantry angles if they are separated by more than 20 . Merging was not performed if the merging of beam orientations resulted in clinically significant sacrifice in PTV coverage or OAR sparing. The resultant number of fields and couch kicks for each patient is summarized in Table 1. Volumetric modulated arc therapy plans with 3-4 full and partial coplanar or non-coplanar arcs were also generated by an experience dosimetrist for comparison. Guided by the comparison, the VMAT plans were reoptimized until no further improvement was achievable. The plan resulting in lower mean and maximum dose to surrounding OARs and increased PTV coverage was selected for treatment.
Pretreatment quality assurance Dosimetric quality assurance measurements were made using OCTAVIUS 729 (PTW, Freiburg, Germany) for cases with PTV diameter larger than 3 cm. OCTAVIUS 729 consists of 729 plane-parallel vented ionization chambers with 1-cm detector spacing on a 2-dimensional plane. The active volume of each ionization chamber is 0.125 cm3. Field-by-field measurements were performed at the corresponding gantry angles, with the couch angle maintained at the neutral position for all beam angles. Resultant distributions were analyzed in VeriSoft (PTW, Freiburg, Germany). For cases with PTV smaller than 3 cm, the relative dose distribution and absolute point dose was verified using a GafChromic QuiCk phantom, GafChromic film EBT3 (Ashland Advanced Materials, Bridgewater, NJ), and PTW
Volume 101 Number 1 2018
Pinpoint 3D TN31016 ion chamber with active volume of 0.016 cm3 (PTW) placed at the center of the high-dose volume. Film results were analyzed with FilmQA Pro software (Ashland Advanced Materials). A g index with 3%/3 mm was used in all dosimetric analyses (22). In addition to performing patient-specific dosimetric quality assurance, a dry run with patient-specific radiosurgical mask and couch translational positions aligned to the treatment position was performed. A delivery sequence was generated on the basis of the treatment beams and imaging nodes by sorting the couch angles to minimize the couch rotation. Imaging nodes (ExacTrac or on-board imager [OBI]) were placed between predetermined pairs of treatment beams. A generalized beam map containing the noncolliding beam solution space, such as the one shown in Figure 1a, was used to guide safe beam navigation, especially for beam angle transitions for which the couch and gantry need to be moved in a specific order to prevent collision (19). The sequenced beams, navigation route, and intrafractional imaging nodes were provided as detailed step-by-step written instructions for the therapists to strictly follow.
Treatment delivery and patient survey All patients were immobilized with radiosurgical masks during treatment. Treatments were delivered using a Varian TrueBeam or a Novalis Tx system (Brainlab, Munich, Germany). Cone-beam CT was performed to guide the initial setup. Intrafractional motion was monitored by acquiring 3- to 4-kV image pairs with either ExacTrac (Brainlab, Munich, Germany) or the Varian TrueBeam OBI during each fraction. Three to four imaging points guaranteed verification imaging for every 3 to 4 couch kicks. The OBI requires the couch at neutral position, which creates an additional imaging node in the delivery sequence. ExacTrac imaging can be performed at non-coplanar couch positions, thus eliminating the need for the extra imaging nodes. The typical treatment delivery workflow is demonstrated in Figure 1c. For treatments performed on the TrueBeam, remote couch rotation was used for increased efficiency. Partial treatments were handled as any other IMRT plan by continuing delivery of beam segments that were not yet delivered with the remaining monitor units. A patient survey was collected at the end of daily treatment. Overall treatment tolerability, dizziness, nausea, and pain were scored by the patients using a numerical scale of 0 to 10. Treatment tolerability of 10 indicates high tolerability, whereas the same score for the other 3 categories indicates the worst symptoms.
Results Of the 11 recruited patients, 9 were treated with 4p radiation therapy. An example of selected 4p beam orientations is shown in Figure 1b. One patient was not treated because
Prospective 4p radiation therapy clinical study
147
neither VMAT nor 4p met the dosimetric criteria for safe treatment, owing to the high and unsafe cumulative brainstem dose. It is worth noting that even for this untreated case, 4p achieved maximum and mean dose reductions of 8.9 Gy and 3.5 Gy compared with the patient’s VMAT plan. The other patient was treated with VMAT owing to comparable VMAT and 4p plans that resulted from the tumor location being relatively farther away and in a separate coplane from critical organs. The VMAT plan was therefore chosen to reduce treatment time. Substantial OAR sparing was demonstrated with 4p compared with VMAT for all recruited patients (n Z 9), as demonstrated in Table 2. Statistically significant improvements were found for the mean and maximum dose to the brainstem, chiasm, right eye, lens, and left optic nerve. Particularly remarkable average brainstem maximum dose and mean dose reductions of 2.92 Gy and 0.94 Gy in the accumulated plans for all 9 treated patients were observed. This substantial dose reduction enabled treatments for 4 patients in which the safe accumulated nontumor brainstem dose constraint of Dmax 62 Gy used at our institution could not be satisfied with VMAT. The 50% prescription isodose distributions of an initial VMAT plan, 4p plan, and the final reoptimized VMAT plan guided by the optimized 4p plan are shown in Figure 2a, where the dose spillage into the brainstem was markedly reduced with 4p. An example doseevolume histogram is shown in Figure 2b, in which global OAR dose reduction is apparent, particularly for the brainstem and chiasm. The spread of mean and maximum dose of the generated 4p plans compared with VMAT relative to plan prescription dose is shown in Fig. 3a and 3b. The differences in R50 and PTV homogeneity index (D5/D95) between the 4p and VMAT plans were not statistically significant. The cumulative OAR dose distribution from the original and recurrent treatment using 4p or VMAT is shown in Table 2. The average treatment time was 34.1 minutes (range, 19.9-64.5 minutes). The fastest treatments were achieved for cases delivered on the TrueBeam with remote couch kicks and ExacTrac intrafractional imaging. Factors that impacted delivery efficiency include the need to home the couch for OBI and the therapist entering the treatment room to perform manual rotation of the couch. The longest treatment resulted from an unforeseen machine issue that delayed the treatment by 12 minutes. A detailed breakdown of average time spent on initial setup, beam on, intrafractional imaging, and machine motion axes movements between beams for each patient is shown in Table 1. For all patients and fractions, with the exception of a single imaging point, intrafractional motion was shown to be less than 1 mm, in which cases the treatment continued without applying shifts. The single instance of 1.5-mm shift was corrected by applying a couch shift. A typical intrafractional imaging result is shown in Figure 1d. According to the patient survey, all treatment fractions were well tolerated with no adverse events. The survey scores (average standard deviation) or treatment duration
148
International Journal of Radiation Oncology Biology Physics
Yu et al.
a
Gantry vs. Couch 180
b
210 240 270
Gantry (IEC)
300 330 0 30 60 90 120 150 180 90 75 60 45 30 15 0 345 330 315 300 285 270 Couch (IEC) infeasible
extended STD
isocentric beam solution space
c
Beam on
Beam on kV image pairs kV imaging Initial setup for intrafractional motion correction CBCT d
kV 1
CBCT
kV 2
kV 3
Beam on kV imaging
kV 4
Final Beam
Fig. 1. (a) Beam solution space deliverable with isocentric setup with more than 5-cm clearance (blue hollow circles). Angles requiring extended source to target distance (STD) to avoid collision (black filled squares). Infeasible beams regardless of STD extensions (red crosses). (b) Example of selected 4p beam orientations (c) Schematic of treatment delivery workflow. (d) Intrafractional kV imaging results acquired with the TrueBeam on-board imager. Abbreviation: CBCT Z cone-beam computed tomography; IEC Z International Electrotechnical Commission. (A color version of this figure is available at www.redjournal.org.)
Prospective 4p radiation therapy clinical study
Volume 101 Number 1 2018 Table 2
149
Average OAR dose statistics comparison (n Z 9) Cochlea
Parameter 4p Mean Maximum VMAT Mean Maximum 4p þ PreRT Mean Maximum VMAT þ PreRT Mean Maximum
Eyes
Lens
Optic nerve
Brainstem
Chiasm
L
R
L
R
L
R
L
R
Brain
2.47* 8.44*
2.28* 4.12*
2.43 3.76
0.95 1.26*
1.07 2.23
0.47* 1.02*
0.72* 0.99
0.34* 0.47*
2.09* 2.86*
1.25 1.92
3.84 18.38
3.41 12.11
4.04 5.65
3.49 4.39
1.69 2.14
1.49 2.53
1.24 2.09
1.12 1.41
1.04 1.32
2.95 4.02
1.96 2.76
3.85 18.40
36.10* 61.07*
31.48* 42.21*
36.71 42.64
23.95 29.74
6.63 13.32
8.87 16.39*
4.07* 5.18
5.55* 7.33*
18.29* 27.37*
19.16 27.24
35.85 76.53
37.04 63.99
33.27 43.65
37.75 43.30
24.64 30.47
7.05 13.65
9.64 17.32
4.48 5.55
6.25 8.15
19.13 28.36
19.86 28.06
35.86 76.52
Abbreviations: L Z left; OAR Z organ at risk; R Z right; VMAT Z volumetric modulated arc therapy. Upper half of table is a comparison of 4p and VMAT plans generated during the clinical trial (average OAR dose statistics [Gy]). Lower half of table is a comparison of cumulative dose of previous plan (PreRT) and trial plans (average OAR dose statistics of cumulative plan [Gy]). * P < .05 from Wilcoxon signed-rank test.
tolerability, nausea, dizziness, and pain were 8.625 2.64, 0 0.78, 0.66 1.85, and 1.77 2.03, respectively. Six out of nine patients gave full treatment tolerability scores throughout the entire treatment course. Of the 3 patients who reported less than perfect tolerability, 2 reported pain
a
VMAT initial
Brainstem
b
and discomfort related to the time under the tight radiosurgical mask, but the pain seemed to decrease with the progress of treatment. One patient reported headache and dizziness that seemed to be related to the disease itself rather than radiation therapy.
VMAT final
Brainstem
Brainstem
4 π (solid), Clinical (dotted)
1
O-Brn O-Eye-Rt O-Eye-Lt O-Lens-Rt O-Lens-Lt O-Bstm O-Opnv-Lt O-Chsm PTV
0.8 Fractional volume
4π π
0.6
0.4
0.2
0
0
5
10
15 20 Dose (Gy)
25
30
Fig. 2. (a) Fifty percent prescription isodose distribution. Distance from planning target volume to isodose edge in the brainstem direction: 1.67 cm (volumetric modulated arc therapy [VMAT] initial), 0.94 cm (volumetric modulated arc therapy final), 0.29 cm (4p). (b) Doseevolume histogram illustrating the sparing power of 4p. Global organ at risk sparing, particularly for the brainstem and chiasm, in addition to lower planning target volume (PTV) maximum dose and increased homogeneity, can be visually observed.
150
International Journal of Radiation Oncology Biology Physics
Yu et al.
Relative Dose to Prescription (%)
a
110
Maximum Dose Comparison 4π VMAT
100 90 80 70 60 50 40 30 20 10 0 Bstm Chiasm CcleL
b
EyeL
EyeR
LensL LensR
OptL
OptR
Cord
Brain
Mean Dose Comparison
60
Relative Dose to Prescription (%)
CcleR
4π VMAT
55 50 45 40 35 30 25 20 15 10 5 0 Bstm Chiasm CcleL
CcleR
EyeL
EyeR
LensL LensR
OptL
OptR
Cord
Brain
Fig. 3. 4p versus volumetric modulated arc therapy (VMAT) dosimetric comparison, dose relative to the prescription dose of each plan. (a) Maximum dose. (b) Mean dose.
Discussion Recent progress in non-coplanar treatment planning and collision space modeling unveiled the feasibility of additional significant radiation dosimetry improvement from the state-of-the-art IMRT and VMAT methods (1-3, 5-7, 23). However, clinical translation of many such experimental delivery methods, particularly ones with dynamic couch movements during beam-on, remains prohibitive owing to the lack of FDAapproved procedures and products. In contrast, the current implementation of 4 p radiation therapy utilizing static IMRT beams can be performed completely within the FDA-approved domain. This characteristic
allowed us to conduct the first prospective 4 p phase 1 clinical trial, which is reported in this article. In this study the clinical utility and substantial dosimetric benefit of 4p was demonstrated and validated. The global inverse optimization approach that mathematically incorporates the entire noncolliding non-coplanar beam solution space and fluence map optimization resulted in dosimetry that was consistently superior to that of stateof-the-art VMAT plans. In the context of reirradiating patients who have previously received high radiation doses to critical normal organs, such as the brainstem, and then experienced in-field recurrences, 4p helped create safer treatment plans without compromising the PTV coverage. For these well-immobilized patients using rigid
Volume 101 Number 1 2018
radiosurgical masks, secondary head motion induced by the couch rotation was shown to be negligible (24). According to the collected patient survey, the treatments were well tolerated. Motivated by this phase 1 trial, 3 challenging spine SBRT retreatment cases have also been treated with 4p, and we continue to implement the method on cases that demand superior dosimetry. Compared with clinically utilized 3 to 4 non-coplanar arc plans, our inhouse optimizer allowed for optimal selection of beam orientations. Similar implementation that incorporates couch kicks into the VMAT optimization problem may have the potential to produce similar results. One major drawback of this technique is the prolonged treatment time. However, with the aid of pre-established couch and gantry motion sequencing from the Computeraided design-model generated beam solution space map and dry run, treatment delivery efficiency was improved with safe remote couch movements. Using the currently nonclinical Developer Mode, we tested the feasibility and efficiency of fully automated delivery. The automated delivery time for 20-beam 4p treatments to the brain was 10 minutes (19). Therefore, there is a potential to deliver fully automated 4p treatment in current clinical treatment time slots. Another potential challenge less relevant to the current clinical trial is internal organ motion that may be worsened with the couch rotation and that is certainly more difficult to monitor with couch rotation. Periodically homing the couch for imaging is a solution but can have obvious downsides, such as lengthening the treatment. This is an ongoing research topic for many body sites that tend to move during the treatment, even with state-of-the-art immobilization.
Conclusion The feasibility, safety, dosimetric benefits, delivery efficiency, and patient comfort of 4p radiation therapy have been clinically demonstrated with a prospective clinical trial. Treatments were well tolerated despite prolonged treatment time, which can be substantially reduced with automation. These results pave the way for 4p implementation in more clinically challenging cases.
References 1. Breedveld S, Storchi PR, Voet PW, et al. iCycle: Integrated, multicriterial beam angle, and profile optimization for generation of coplanar and noncoplanar IMRT plans. Med Phys 2012;39:951-963. 2. Fahimian B, Yu V, Horst K, et al. Trajectory modulated prone breast irradiation: A LINAC-based technique combining intensity modulated delivery and motion of the couch. Radiother Oncol 2013;109:475-481. 3. Yang Y, Zhang P, Happersett L, et al. Choreographing couch and collimator in volumetric modulated arc therapy. Int J Radiat Oncol Biol Phys 2011;80:1238-1247.
Prospective 4p radiation therapy clinical study
151
4. Ma Y, Popple R, Suh TS, et al. Beam’s-eye-view dosimetricseguided inverse planning for aperture-modulated arc therapy. Int J Radiat Oncol Biol Phys 2009;75:1587-1595. 5. MacDonald RL, Thomas CG. Dynamic trajectory-based couch motion for improvement of radiation therapy trajectories in cranial SRT. Med Phys 2015;42:2317-2325. 6. Smyth G, Bamber JC, Evans PM, et al. Trajectory optimization for dynamic couch rotation during volumetric modulated arc radiotherapy. Phys Med Biol 2013;58:8163-8177. 7. Dong P, Lee P, Ruan D, et al. 4pi non-coplanar liver SBRT: A novel delivery technique. Int J Radiat Oncol Biol Phys 2013;85: 1360-1366. 8. Nguyen D, Rwigema JCM, Victoria YY, et al. Feasibility of extreme dose escalation for glioblastoma multiforme using 4p radiotherapy. Radiat Oncol 2014;9:239. 9. Rwigema JCM, Nguyen D, Heron DE, et al. 4p noncoplanar stereotactic body radiation therapy for head-and-neck cancer: Potential to improve tumor control and late toxicity. Int J Radiat Oncol Biol Phys 2015;91:401-409. 10. 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;86:407-413. 11. Woods K, Nguyen D, Tran A, et al. Viability of non-coplanar VMAT for liver SBRT as compared to coplanar VMAT and beam orientation optimized 4p IMRT. Adv Radiat Oncol 2016;1:67-75. 12. Tran A, Zhang J, Woods K, et al. Treatment planning comparison of IMPT, VMAT and 4p radiotherapy for prostate cases. Radiat Oncol 2017;12:10. 13. Dong P, Nguyen D, Ruan D, et al. Feasibility of prostate robotic radiation therapy on conventional C-arm linacs. Pract Radiat Oncol 2014;4:254-260. 14. Nguyen D, Dong P, Long T, et al. Integral dose investigation of noncoplanar treatment beam geometries in radiotherapy. Med Phys 2014; 41:011905. 15. Liang J, Atwood T, von Eyben R, et al. Trajectory modulated arc therapy: A fully dynamic delivery with synchronized couch and gantry motion significantly improves dosimetric indices correlated with poor cosmesis in accelerated partial breast irradiation. Int J Radiat Oncol Biol Phys 2015;92:1148-1156. 16. Adamson C, Kanu OO, Mehta AI, et al. Glioblastoma multiforme: A review of where we have been and where we are going. Expert Opin Investig Drugs 2009;18:1061-1083. 17. Chan JL, Lee SW, Fraass BA, et al. Survival and failure patterns of high-grade gliomas after three-dimensional conformal radiotherapy. J Clin Oncol 2002;20:1635-1642. 18. Wallner KE, Galicich JH, Krol G, et al. Patterns of failure following treatment for glioblastoma multiforme and anaplastic astrocytoma. Int J Radiat Oncol Biol Phys 1989;16:1405-1409. 19. Yu VY, Tran A, Nguyen D, et al. The development and verification of a highly accurate collision prediction model for automated noncoplanar plan delivery. Med Phys 2015;42:6457. 20. Romeijn HE, Ahuja RK, Dempsey JF, et al. A column generation approach to radiation therapy treatment planning using aperture modulation. SIAM J Optim 2005;15:838-862. 21. Nguyen D, Thomas D, Cao M, et al. Computerized triplet beam orientation optimization for MRI-guided Co-60 radiotherapy. Med Phys 2016;43:5667-5675. 22. Low DA, Harms WB, Mutic S, et al. A technique for the quantitative evaluation of dose distributions. Med Phys 1998;25:656-661. 23. Burghelea M, Verellen D, Poels K, et al. Initial characterization, dosimetric benchmark and performance validation of dynamic wave arc. Radiat Oncol 2016;11:63. 24. Cervin˜o LI, Pawlicki T, Lawson JD, et al. Frame-less and mask-less cranial stereotactic radiosurgery: A feasibility study. Phys Med Biol 2010;55:1863-1873.
Radiation Oncology biology
physics
www.redjournal.org
Physics Contribution
A Prospective 4p Radiation Therapy Clinical Study in Recurrent High-Grade Glioma Patients Victoria Y. Yu, PhD, Angelia Landers, MS, Kaley Woods, MS, Dan Nguyen, PhD, Minsong Cao, PhD, Dongsu Du, PhD, Robert K. Chin, MD, PhD, Ke Sheng, PhD, and Tania B. Kaprealian, MD Department of Radiation Oncology, David Geffen School of Medicine, University of California, Los Angeles, Los Angeles, California Received Jun 23, 2017, and in revised form Dec 27, 2017. Accepted for publication Jan 12, 2018.
Summary 4p radiation therapy, a novel delivery technique utilizing inverse optimization for automatic non-coplanar beam orientation selection and fluence optimization, was shown capable of achieving dosimetry superior to conventional methods. This study summarizes the first prospective clinical trial of 4p radiation therapy in recurrent high-grade glioma patients. Substantial sparing to organs at risk was achieved compared with stateof-the-art VMAT plans, treatments were well tolerated, and intrafractional motion was <1.5 mm for all patients, motivating further clinical utilization of the technique.
Purpose: To evaluate the feasibility, safety, dosimetric benefits, delivery efficiency, and patient comfort in the clinical implementation of 4p radiation therapy. Methods and Materials: Eleven patients with recurrent high-grade glioma were recruited for the trial. 4 p plans integrating beam orientation and fluence-map optimization were created using an in-house column-generation algorithm. The collision-free beam solution space throughout the 4p steradian was determined using a computer-aided-design model of the Varian TrueBeam system and a human subject. Twenty beams were optimized for each case and imported into Eclipse for intensity modulated radiation therapy planning. Beam orientations with neighboring couch kicks were merged for increased delivery efficiency, generating plans with an average of 16 beam orientations. Volumetric modulated arc therapy (VMAT) plans with 3-4 arcs were also generated for each case, and the plan achieving superior dosimetric quality was selected for treatment. Patient comfort was surveyed after every fraction. Multiple 2-dimensional X-ray images were obtained to measure intrafractional motion. Results: Of 11 patients, 9 were treated with 4 p . Mean and maximum organ at risk doses were equal or significantly lower (P < .05) with 4p than with VMAT. Particularly substantial dose reduction of 2.92 Gy in the average accumulated brainstem maximum dose enabled treatments that would otherwise not satisfy safe dose constraints with VMAT. One patient was not treated because neither plan met the dosimetric criteria. The other was treated with VMAT owing to comparable dosimetry resulting from a planning target volume located in a separate co-plane superior to organs at risk. Treatments were well tolerated, with an average patient comfort score of 8.6/10. Intrafractional motion was <1.5 mm for all delivered fractions, and the average delivery time was 34.1 minutes.
Reprint requests to: Tania B. Kaprealian, MD, Department of Radiation Oncology, University of California, Los Angeles, 200 Medical Plaza,
Int J Radiation Oncol Biol Phys, Vol. 101, No. 1, pp. 144e151, 2018 0360-3016/$ - see front matter Ó 2018 Elsevier Inc. All rights reserved. https://doi.org/10.1016/j.ijrobp.2018.01.048
B265, Los Angeles, CA 90024. Tel: (310) 825-9775; E-mail: tkaprealian@ mednet.ucla.edu Conflict of interest: none.
Prospective 4p radiation therapy clinical study
Volume 101 Number 1 2018
145
Conclusions: The feasibility, safety, dosimetric benefits, delivery efficiency, and patient comfort of 4 p radiation therapy have been clinically demonstrated with a prospective clinical trial. The results elucidate the potential and challenges of wider clinical implementations. Ó 2018 Elsevier Inc. All rights reserved.
Introduction
Methods and Materials
Current state-of-the-art medical digital linear accelerators are capable of delivering radiation therapy with complex dynamic trajectories that involve orchestrated movements of the gantry, couch, and multileaf collimator (MLC), in conjunction with dose rate modulations. These new capabilities have motivated active research to explore the potential in further improving radiation therapy plan quality through optimization of beam orientations and dynamic trajectories, in which substantial dosimetric improvements have been shown (1-6). The optimal utilization of the non-coplanar beam solution space has been shown to better shape the spatial dose distribution. 4p radiation therapy is a novel planning and delivery technique integrating static intensity modulated radiation therapy (IMRT) beam orientation and fluence map optimization (7). Its dosimetry has been favorably compared with volumetric modulated arc therapy (VMAT) or IMRT with manually selected beam orientations in terms of the neighboring organs at risk (OARs) sparing for disease sites, including the brain (8), head and neck (9), lung (10), liver (11), and prostate (12, 13). The integral dose of highly noncoplanar radiation therapy has been shown to be comparable to or lower than that of clinical coplanar plans in both phantom simulations and patient case evaluations (14), but the patient volume receiving low-dose radiation may increase and should therefore be evaluated for curative or pediatric cases in which such dose volumes could be of clinical significance. The dosimetric benefit and feasibility of 4p radiation therapy were demonstrated through retrospective dosimetric evaluations and automated delivery in Varian Developer Mode (Varian Medical Systems, Palo Alto, CA) but have not yet been clinically implemented, partly owing to the lack of US Food and Drug Administration (FDA) cleared product. On the other hand, clinical implementations of 4p static beam radiation therapy, despite its complex coordinated couch and gantry movements, can be delivered in the clinical mode. This characteristic is distinctly different from several other experimental delivery methods using dynamic couch movements during beam-on (3, 5, 6, 15). In this study we demonstrate the first clinical implementation of 4p radiation therapy via a prospective clinical trial to test its feasibility, safety, dosimetric benefits, intrafractional motion, and delivery efficiency, and patient comfort on recurrent high-grade glioma patients.
Patient characteristics and treatment plan generation Under an institutional review boardeapproved phase 1 trial protocol, locally recurrent glioma was chosen as the first disease site for clinical translation of 4p radiation therapy, owing to the demanding dosimetric constraints to avoid over-irradiating previously treated normal tissue volumes (16-18). Specifically, these patients received 59.4 or 60 Gy from the original treatment before receiving another 25 Gy in 5 fractions or 30 Gy in 10 fractions to the recurrent tumor target volume that partially or completely overlapped with the original target volume. Because of this overlap, it is extremely challenging to meet the dose limits on critical organs, including the brainstem and optic apparatus. Additionally, the brain tumor location allows for access to more non-coplanar angles with greater clearance between the couch and gantry, making this patient cohort particularly well suited for the first prospective clinical trial of 4p radiation therapy. The recruitment inclusion criteria included a histologic diagnosis of primary and recurrent high-grade glioma, Karnofsky performance status >70, and age of 18 years or older. Eleven patients consented to participate in the clinical trial from December 2014 to January 2017. Patient and treatment characteristics of both the prior radiation therapy for the primary disease and the delivered plan during trial participation are shown in Table 1. Computed tomography (CT) simulation was performed with patients immobilized using a thermoplastic radiosurgical mask (Brainlab, Munich, Germany) in the supine position. The gross tumor volume was defined as the enhanced gross disease volume on the T1-weighted magnetic resonance image. The planning target volume (PTV) was then contoured by the physicians (TBK, RKC) as the gross tumor volume plus a 2- to 5-mm isotropic expansion, depending on each individual case. All contours were transferred from the magnetic resonance images to the CT domain through rigid image registration performed in MIM 6.6.5 (MIM Software, Cleveland, OH). With the CT image and the established PTV and OARs contours, a previously reported in-house 4p radiation therapy treatment planning optimization (7-9, 13) was used to select an optimal set of 20 beam angles for each patient. The number of beams for the optimization was empirically established as a balance between plan practicality and dose optimality. The number of beams for a
146
International Journal of Radiation Oncology Biology Physics
Yu et al.
Table 1
Patient and treatment plan characteristics, along with average delivery time breakdown for each patient Average delivery time breakdown (min)
Pt no. Histology Age (y) Sex 1 2 3 4 5 6 7 8 9
GBM GBM GBM GBM GBM GBM GBM EPD GBM
54 64 60 54 52 48 39 80 59
F F F M M M M F M
Prior dose (Gy)
PTV volume (cm3)
59.4 60 60 60 59.4 60 46 54 60
13.99 110.14 0.95 1.84 124.51 18.17 47.15 2.57 3.86
Dosing scheme 5 3 5 5 3 6 3 5 5
Gy Gy Gy Gy Gy Gy Gy Gy Gy
5 10 5 5 10 5 10 5 5
Total Dose (Gy)
Machine
25 30 25 25 30 30 30 25 25
TrueBeam TrueBeam Novalis Tx Novalis Tx TrueBeam TrueBeam TrueBeam TrueBeam Novalis Tx
Recruited but not treated with 4p 10 GBM 11 GBM
56 66
M M
59.4 10.94 3 Gy 10 40 þ 20 SIB 141.09 3 Gy 10
Motion No. axes No. couch Initial Beam fields kicks setup on Imaging mvnt Total 18 19 15 13 20 15 16 17 15
12 14 11 11 15 9 10 10 11
9.4 7.7 3.0 2.4 5.3 9.1 10.3 12.6 3.0
2.1 3.5 1.9 1.8 2.4 2.0 2.1 3.5 3.1
1.4 1.5 2.6 2.2 2.8 3.4 2.0 2.4 2.8
27.1 32.1 25.5 21.0 23.5* 12.9* 16.6* 15.9* 23.1
40.1 44.8 33.1 27.5 34.0 27.3 31.1 34.4 31.9
Reason 30 30
Neither plans met dose constraints Comparable VMAT plan
Abbreviations: EPD Z ependymoma; F Z female; GBM Z glioblastoma multiforme; M Z male; PTV Z planning target volume; SIB Z simultaneous integrated boost. Initial setup: average time taken for initial setup, imaging, and physician approval; Imaging: total intrafractional imaging time in between beam angles; Motion axes mvnt: couch and gantry movement time between beams; Total time: total delivery time. * Cases delivered with remote couch kick capabilities.
good balance could vary between patients, but to standardize our process, we currently choose 20. These beams were selected from a candidate pool containing 1162 evenly distributed non-coplanar beams with approximately 6 separation between adjacent beams. Guided by a detailed and accurate computer-aided model of the Varian TrueBeam (Varian Medical Systems) and a 3-dimensional surface image of a volunteer placed on its couch, beams causing collision or being blocked by the couch pedestal were excluded from optimization (19). 4p treatment optimization was performed on the basis of precomputed beamlets. The dose calculation was performed using collapsed cone convolution/superposition and Monte Carlo calculated 6-MV poly-energetic X-ray kernels (7). The dose calculation resolution was 2.5 2.5 2.5 mm3 for 5 5 mm2. A greedy column generation algorithm with user-defined upper dose limits and structure priority weightings was used to iteratively select beam orientations and perform fluence optimization with all selected beam orientations until the desired number of beam angles was reached. The objective function of the optimization problem is based on a linear approximation of an equivalent uniform dose, defined in detail in previous publications (7, 20, 21). To generate a clinically deliverable plan using an FDA-approved treatment planning system and delivery technique, the 20 beam orientations selected by 4p were imported into Eclipse v13.6.23 (Varian Medical Systems) for IMRT reoptimization. To avoid excessively long delivery time and reduce the number of couch kicks, beams orientations with less than 20 separation in both the couch and gantry angles are merged to take the average position.
Beams with couch angles within 6 were merged to use a single couch position while maintaining their own gantry angles if they are separated by more than 20 . Merging was not performed if the merging of beam orientations resulted in clinically significant sacrifice in PTV coverage or OAR sparing. The resultant number of fields and couch kicks for each patient is summarized in Table 1. Volumetric modulated arc therapy plans with 3-4 full and partial coplanar or non-coplanar arcs were also generated by an experience dosimetrist for comparison. Guided by the comparison, the VMAT plans were reoptimized until no further improvement was achievable. The plan resulting in lower mean and maximum dose to surrounding OARs and increased PTV coverage was selected for treatment.
Pretreatment quality assurance Dosimetric quality assurance measurements were made using OCTAVIUS 729 (PTW, Freiburg, Germany) for cases with PTV diameter larger than 3 cm. OCTAVIUS 729 consists of 729 plane-parallel vented ionization chambers with 1-cm detector spacing on a 2-dimensional plane. The active volume of each ionization chamber is 0.125 cm3. Field-by-field measurements were performed at the corresponding gantry angles, with the couch angle maintained at the neutral position for all beam angles. Resultant distributions were analyzed in VeriSoft (PTW, Freiburg, Germany). For cases with PTV smaller than 3 cm, the relative dose distribution and absolute point dose was verified using a GafChromic QuiCk phantom, GafChromic film EBT3 (Ashland Advanced Materials, Bridgewater, NJ), and PTW
Volume 101 Number 1 2018
Pinpoint 3D TN31016 ion chamber with active volume of 0.016 cm3 (PTW) placed at the center of the high-dose volume. Film results were analyzed with FilmQA Pro software (Ashland Advanced Materials). A g index with 3%/3 mm was used in all dosimetric analyses (22). In addition to performing patient-specific dosimetric quality assurance, a dry run with patient-specific radiosurgical mask and couch translational positions aligned to the treatment position was performed. A delivery sequence was generated on the basis of the treatment beams and imaging nodes by sorting the couch angles to minimize the couch rotation. Imaging nodes (ExacTrac or on-board imager [OBI]) were placed between predetermined pairs of treatment beams. A generalized beam map containing the noncolliding beam solution space, such as the one shown in Figure 1a, was used to guide safe beam navigation, especially for beam angle transitions for which the couch and gantry need to be moved in a specific order to prevent collision (19). The sequenced beams, navigation route, and intrafractional imaging nodes were provided as detailed step-by-step written instructions for the therapists to strictly follow.
Treatment delivery and patient survey All patients were immobilized with radiosurgical masks during treatment. Treatments were delivered using a Varian TrueBeam or a Novalis Tx system (Brainlab, Munich, Germany). Cone-beam CT was performed to guide the initial setup. Intrafractional motion was monitored by acquiring 3- to 4-kV image pairs with either ExacTrac (Brainlab, Munich, Germany) or the Varian TrueBeam OBI during each fraction. Three to four imaging points guaranteed verification imaging for every 3 to 4 couch kicks. The OBI requires the couch at neutral position, which creates an additional imaging node in the delivery sequence. ExacTrac imaging can be performed at non-coplanar couch positions, thus eliminating the need for the extra imaging nodes. The typical treatment delivery workflow is demonstrated in Figure 1c. For treatments performed on the TrueBeam, remote couch rotation was used for increased efficiency. Partial treatments were handled as any other IMRT plan by continuing delivery of beam segments that were not yet delivered with the remaining monitor units. A patient survey was collected at the end of daily treatment. Overall treatment tolerability, dizziness, nausea, and pain were scored by the patients using a numerical scale of 0 to 10. Treatment tolerability of 10 indicates high tolerability, whereas the same score for the other 3 categories indicates the worst symptoms.
Results Of the 11 recruited patients, 9 were treated with 4p radiation therapy. An example of selected 4p beam orientations is shown in Figure 1b. One patient was not treated because
Prospective 4p radiation therapy clinical study
147
neither VMAT nor 4p met the dosimetric criteria for safe treatment, owing to the high and unsafe cumulative brainstem dose. It is worth noting that even for this untreated case, 4p achieved maximum and mean dose reductions of 8.9 Gy and 3.5 Gy compared with the patient’s VMAT plan. The other patient was treated with VMAT owing to comparable VMAT and 4p plans that resulted from the tumor location being relatively farther away and in a separate coplane from critical organs. The VMAT plan was therefore chosen to reduce treatment time. Substantial OAR sparing was demonstrated with 4p compared with VMAT for all recruited patients (n Z 9), as demonstrated in Table 2. Statistically significant improvements were found for the mean and maximum dose to the brainstem, chiasm, right eye, lens, and left optic nerve. Particularly remarkable average brainstem maximum dose and mean dose reductions of 2.92 Gy and 0.94 Gy in the accumulated plans for all 9 treated patients were observed. This substantial dose reduction enabled treatments for 4 patients in which the safe accumulated nontumor brainstem dose constraint of Dmax 62 Gy used at our institution could not be satisfied with VMAT. The 50% prescription isodose distributions of an initial VMAT plan, 4p plan, and the final reoptimized VMAT plan guided by the optimized 4p plan are shown in Figure 2a, where the dose spillage into the brainstem was markedly reduced with 4p. An example doseevolume histogram is shown in Figure 2b, in which global OAR dose reduction is apparent, particularly for the brainstem and chiasm. The spread of mean and maximum dose of the generated 4p plans compared with VMAT relative to plan prescription dose is shown in Fig. 3a and 3b. The differences in R50 and PTV homogeneity index (D5/D95) between the 4p and VMAT plans were not statistically significant. The cumulative OAR dose distribution from the original and recurrent treatment using 4p or VMAT is shown in Table 2. The average treatment time was 34.1 minutes (range, 19.9-64.5 minutes). The fastest treatments were achieved for cases delivered on the TrueBeam with remote couch kicks and ExacTrac intrafractional imaging. Factors that impacted delivery efficiency include the need to home the couch for OBI and the therapist entering the treatment room to perform manual rotation of the couch. The longest treatment resulted from an unforeseen machine issue that delayed the treatment by 12 minutes. A detailed breakdown of average time spent on initial setup, beam on, intrafractional imaging, and machine motion axes movements between beams for each patient is shown in Table 1. For all patients and fractions, with the exception of a single imaging point, intrafractional motion was shown to be less than 1 mm, in which cases the treatment continued without applying shifts. The single instance of 1.5-mm shift was corrected by applying a couch shift. A typical intrafractional imaging result is shown in Figure 1d. According to the patient survey, all treatment fractions were well tolerated with no adverse events. The survey scores (average standard deviation) or treatment duration
148
International Journal of Radiation Oncology Biology Physics
Yu et al.
a
Gantry vs. Couch 180
b
210 240 270
Gantry (IEC)
300 330 0 30 60 90 120 150 180 90 75 60 45 30 15 0 345 330 315 300 285 270 Couch (IEC) infeasible
extended STD
isocentric beam solution space
c
Beam on
Beam on kV image pairs kV imaging Initial setup for intrafractional motion correction CBCT d
kV 1
CBCT
kV 2
kV 3
Beam on kV imaging
kV 4
Final Beam
Fig. 1. (a) Beam solution space deliverable with isocentric setup with more than 5-cm clearance (blue hollow circles). Angles requiring extended source to target distance (STD) to avoid collision (black filled squares). Infeasible beams regardless of STD extensions (red crosses). (b) Example of selected 4p beam orientations (c) Schematic of treatment delivery workflow. (d) Intrafractional kV imaging results acquired with the TrueBeam on-board imager. Abbreviation: CBCT Z cone-beam computed tomography; IEC Z International Electrotechnical Commission. (A color version of this figure is available at www.redjournal.org.)
Prospective 4p radiation therapy clinical study
Volume 101 Number 1 2018 Table 2
149
Average OAR dose statistics comparison (n Z 9) Cochlea
Parameter 4p Mean Maximum VMAT Mean Maximum 4p þ PreRT Mean Maximum VMAT þ PreRT Mean Maximum
Eyes
Lens
Optic nerve
Brainstem
Chiasm
L
R
L
R
L
R
L
R
Brain
2.47* 8.44*
2.28* 4.12*
2.43 3.76
0.95 1.26*
1.07 2.23
0.47* 1.02*
0.72* 0.99
0.34* 0.47*
2.09* 2.86*
1.25 1.92
3.84 18.38
3.41 12.11
4.04 5.65
3.49 4.39
1.69 2.14
1.49 2.53
1.24 2.09
1.12 1.41
1.04 1.32
2.95 4.02
1.96 2.76
3.85 18.40
36.10* 61.07*
31.48* 42.21*
36.71 42.64
23.95 29.74
6.63 13.32
8.87 16.39*
4.07* 5.18
5.55* 7.33*
18.29* 27.37*
19.16 27.24
35.85 76.53
37.04 63.99
33.27 43.65
37.75 43.30
24.64 30.47
7.05 13.65
9.64 17.32
4.48 5.55
6.25 8.15
19.13 28.36
19.86 28.06
35.86 76.52
Abbreviations: L Z left; OAR Z organ at risk; R Z right; VMAT Z volumetric modulated arc therapy. Upper half of table is a comparison of 4p and VMAT plans generated during the clinical trial (average OAR dose statistics [Gy]). Lower half of table is a comparison of cumulative dose of previous plan (PreRT) and trial plans (average OAR dose statistics of cumulative plan [Gy]). * P < .05 from Wilcoxon signed-rank test.
tolerability, nausea, dizziness, and pain were 8.625 2.64, 0 0.78, 0.66 1.85, and 1.77 2.03, respectively. Six out of nine patients gave full treatment tolerability scores throughout the entire treatment course. Of the 3 patients who reported less than perfect tolerability, 2 reported pain
a
VMAT initial
Brainstem
b
and discomfort related to the time under the tight radiosurgical mask, but the pain seemed to decrease with the progress of treatment. One patient reported headache and dizziness that seemed to be related to the disease itself rather than radiation therapy.
VMAT final
Brainstem
Brainstem
4 π (solid), Clinical (dotted)
1
O-Brn O-Eye-Rt O-Eye-Lt O-Lens-Rt O-Lens-Lt O-Bstm O-Opnv-Lt O-Chsm PTV
0.8 Fractional volume
4π π
0.6
0.4
0.2
0
0
5
10
15 20 Dose (Gy)
25
30
Fig. 2. (a) Fifty percent prescription isodose distribution. Distance from planning target volume to isodose edge in the brainstem direction: 1.67 cm (volumetric modulated arc therapy [VMAT] initial), 0.94 cm (volumetric modulated arc therapy final), 0.29 cm (4p). (b) Doseevolume histogram illustrating the sparing power of 4p. Global organ at risk sparing, particularly for the brainstem and chiasm, in addition to lower planning target volume (PTV) maximum dose and increased homogeneity, can be visually observed.
150
International Journal of Radiation Oncology Biology Physics
Yu et al.
Relative Dose to Prescription (%)
a
110
Maximum Dose Comparison 4π VMAT
100 90 80 70 60 50 40 30 20 10 0 Bstm Chiasm CcleL
b
EyeL
EyeR
LensL LensR
OptL
OptR
Cord
Brain
Mean Dose Comparison
60
Relative Dose to Prescription (%)
CcleR
4π VMAT
55 50 45 40 35 30 25 20 15 10 5 0 Bstm Chiasm CcleL
CcleR
EyeL
EyeR
LensL LensR
OptL
OptR
Cord
Brain
Fig. 3. 4p versus volumetric modulated arc therapy (VMAT) dosimetric comparison, dose relative to the prescription dose of each plan. (a) Maximum dose. (b) Mean dose.
Discussion Recent progress in non-coplanar treatment planning and collision space modeling unveiled the feasibility of additional significant radiation dosimetry improvement from the state-of-the-art IMRT and VMAT methods (1-3, 5-7, 23). However, clinical translation of many such experimental delivery methods, particularly ones with dynamic couch movements during beam-on, remains prohibitive owing to the lack of FDAapproved procedures and products. In contrast, the current implementation of 4 p radiation therapy utilizing static IMRT beams can be performed completely within the FDA-approved domain. This characteristic
allowed us to conduct the first prospective 4 p phase 1 clinical trial, which is reported in this article. In this study the clinical utility and substantial dosimetric benefit of 4p was demonstrated and validated. The global inverse optimization approach that mathematically incorporates the entire noncolliding non-coplanar beam solution space and fluence map optimization resulted in dosimetry that was consistently superior to that of stateof-the-art VMAT plans. In the context of reirradiating patients who have previously received high radiation doses to critical normal organs, such as the brainstem, and then experienced in-field recurrences, 4p helped create safer treatment plans without compromising the PTV coverage. For these well-immobilized patients using rigid
Volume 101 Number 1 2018
radiosurgical masks, secondary head motion induced by the couch rotation was shown to be negligible (24). According to the collected patient survey, the treatments were well tolerated. Motivated by this phase 1 trial, 3 challenging spine SBRT retreatment cases have also been treated with 4p, and we continue to implement the method on cases that demand superior dosimetry. Compared with clinically utilized 3 to 4 non-coplanar arc plans, our inhouse optimizer allowed for optimal selection of beam orientations. Similar implementation that incorporates couch kicks into the VMAT optimization problem may have the potential to produce similar results. One major drawback of this technique is the prolonged treatment time. However, with the aid of pre-established couch and gantry motion sequencing from the Computeraided design-model generated beam solution space map and dry run, treatment delivery efficiency was improved with safe remote couch movements. Using the currently nonclinical Developer Mode, we tested the feasibility and efficiency of fully automated delivery. The automated delivery time for 20-beam 4p treatments to the brain was 10 minutes (19). Therefore, there is a potential to deliver fully automated 4p treatment in current clinical treatment time slots. Another potential challenge less relevant to the current clinical trial is internal organ motion that may be worsened with the couch rotation and that is certainly more difficult to monitor with couch rotation. Periodically homing the couch for imaging is a solution but can have obvious downsides, such as lengthening the treatment. This is an ongoing research topic for many body sites that tend to move during the treatment, even with state-of-the-art immobilization.
Conclusion The feasibility, safety, dosimetric benefits, delivery efficiency, and patient comfort of 4p radiation therapy have been clinically demonstrated with a prospective clinical trial. Treatments were well tolerated despite prolonged treatment time, which can be substantially reduced with automation. These results pave the way for 4p implementation in more clinically challenging cases.
References 1. Breedveld S, Storchi PR, Voet PW, et al. iCycle: Integrated, multicriterial beam angle, and profile optimization for generation of coplanar and noncoplanar IMRT plans. Med Phys 2012;39:951-963. 2. Fahimian B, Yu V, Horst K, et al. Trajectory modulated prone breast irradiation: A LINAC-based technique combining intensity modulated delivery and motion of the couch. Radiother Oncol 2013;109:475-481. 3. Yang Y, Zhang P, Happersett L, et al. Choreographing couch and collimator in volumetric modulated arc therapy. Int J Radiat Oncol Biol Phys 2011;80:1238-1247.
Prospective 4p radiation therapy clinical study
151
4. Ma Y, Popple R, Suh TS, et al. Beam’s-eye-view dosimetricseguided inverse planning for aperture-modulated arc therapy. Int J Radiat Oncol Biol Phys 2009;75:1587-1595. 5. MacDonald RL, Thomas CG. Dynamic trajectory-based couch motion for improvement of radiation therapy trajectories in cranial SRT. Med Phys 2015;42:2317-2325. 6. Smyth G, Bamber JC, Evans PM, et al. Trajectory optimization for dynamic couch rotation during volumetric modulated arc radiotherapy. Phys Med Biol 2013;58:8163-8177. 7. Dong P, Lee P, Ruan D, et al. 4pi non-coplanar liver SBRT: A novel delivery technique. Int J Radiat Oncol Biol Phys 2013;85: 1360-1366. 8. Nguyen D, Rwigema JCM, Victoria YY, et al. Feasibility of extreme dose escalation for glioblastoma multiforme using 4p radiotherapy. Radiat Oncol 2014;9:239. 9. Rwigema JCM, Nguyen D, Heron DE, et al. 4p noncoplanar stereotactic body radiation therapy for head-and-neck cancer: Potential to improve tumor control and late toxicity. Int J Radiat Oncol Biol Phys 2015;91:401-409. 10. 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;86:407-413. 11. Woods K, Nguyen D, Tran A, et al. Viability of non-coplanar VMAT for liver SBRT as compared to coplanar VMAT and beam orientation optimized 4p IMRT. Adv Radiat Oncol 2016;1:67-75. 12. Tran A, Zhang J, Woods K, et al. Treatment planning comparison of IMPT, VMAT and 4p radiotherapy for prostate cases. Radiat Oncol 2017;12:10. 13. Dong P, Nguyen D, Ruan D, et al. Feasibility of prostate robotic radiation therapy on conventional C-arm linacs. Pract Radiat Oncol 2014;4:254-260. 14. Nguyen D, Dong P, Long T, et al. Integral dose investigation of noncoplanar treatment beam geometries in radiotherapy. Med Phys 2014; 41:011905. 15. Liang J, Atwood T, von Eyben R, et al. Trajectory modulated arc therapy: A fully dynamic delivery with synchronized couch and gantry motion significantly improves dosimetric indices correlated with poor cosmesis in accelerated partial breast irradiation. Int J Radiat Oncol Biol Phys 2015;92:1148-1156. 16. Adamson C, Kanu OO, Mehta AI, et al. Glioblastoma multiforme: A review of where we have been and where we are going. Expert Opin Investig Drugs 2009;18:1061-1083. 17. Chan JL, Lee SW, Fraass BA, et al. Survival and failure patterns of high-grade gliomas after three-dimensional conformal radiotherapy. J Clin Oncol 2002;20:1635-1642. 18. Wallner KE, Galicich JH, Krol G, et al. Patterns of failure following treatment for glioblastoma multiforme and anaplastic astrocytoma. Int J Radiat Oncol Biol Phys 1989;16:1405-1409. 19. Yu VY, Tran A, Nguyen D, et al. The development and verification of a highly accurate collision prediction model for automated noncoplanar plan delivery. Med Phys 2015;42:6457. 20. Romeijn HE, Ahuja RK, Dempsey JF, et al. A column generation approach to radiation therapy treatment planning using aperture modulation. SIAM J Optim 2005;15:838-862. 21. Nguyen D, Thomas D, Cao M, et al. Computerized triplet beam orientation optimization for MRI-guided Co-60 radiotherapy. Med Phys 2016;43:5667-5675. 22. Low DA, Harms WB, Mutic S, et al. A technique for the quantitative evaluation of dose distributions. Med Phys 1998;25:656-661. 23. Burghelea M, Verellen D, Poels K, et al. Initial characterization, dosimetric benchmark and performance validation of dynamic wave arc. Radiat Oncol 2016;11:63. 24. Cervin˜o LI, Pawlicki T, Lawson JD, et al. Frame-less and mask-less cranial stereotactic radiosurgery: A feasibility study. Phys Med Biol 2010;55:1863-1873.