|Ahead of print publication
Robustly optimized hybrid intensity-modulated proton therapy for craniospinal irradiation
Manthala Padannayil Noufal1, Shamurailatpam Dayananda Sharma1, Ganapathy Krishnan1, Mayur Sawant1, Utpal Gaikwad2, Rakesh Jalali2
1 Department of Medical Physics, Apollo Proton Cancer Centre, Chennai, Tamil Nadu, India
2 Department of Radiation Oncology, Apollo Proton Cancer Centre, Chennai, Tamil Nadu, India
|Date of Submission||09-Jan-2020|
|Date of Decision||30-Dec-2020|
|Date of Acceptance||23-Feb-2021|
|Date of Web Publication||18-Aug-2021|
Shamurailatpam Dayananda Sharma,
Apollo Proton Cancer Centre, Dr Vikram Sarabai Instronic, Estate 7th Street, Dr. Vasi Estate, Phase II, Tharamani, Chennai - 600 096, Tamil Nadu
Source of Support: None, Conflict of Interest: None
Aim: The aim of the study was to investigate the hybrid robust optimization planning approach in intensity-modulated proton therapy (IMPT) of craniospinal irradiation (CSI).
Subjects and Methods: Five IMPT-based adult CSI plans in supine position were created using Raystation treatment planning system (TPS) modelled for Proteus plus proton therapy system. A hybrid planning strategy was implemented, where clinical target volume was robustly optimized (RB) for set up uncertainties and planning target volume was optimized for target coverage using minimax algorithm in the TPS. Beam angle selection, optimization, and dose calculation approach were carefully performed to ensure optimum organ at risk (OAR) sparing, even with potential setup and range errors. The complementary dose gradients in junctions were generated using spot assignment and RB technique. Dosimetric outcomes in both nominal plan and the 16 error scenarios (±3 mm setup and 3.5% range) were analyzed using standard dose volume histogram.
Results: This planning approach resulted in a homogeneous dose distribution in the target volume of CSI, including the junction regions, by explicitly reducing number of robust optimization scenarios. The proposed technique was also able to achieve excellent coverage to cribriform plate with lower lens doses and minimal dose to other OARs. Target and OAR doses in the nominal plans as well as in the worst case scenarios with setup and range errors were able to meet the predefined clinical goal.
Conclusions: This proposed planning technique is efficient, robust against the uncertainties. It could be adopted in other proton therapy centers.
Keywords: Craniospinal irradiation, intensity modulated proton therapy, proton therapy, robust optimization
|How to cite this URL:|
Noufal MP, Sharma SD, Krishnan G, Sawant M, Gaikwad U, Jalali R. Robustly optimized hybrid intensity-modulated proton therapy for craniospinal irradiation. J Can Res Ther [Epub ahead of print] [cited 2021 Dec 5]. Available from: https://www.cancerjournal.net/preprintarticle.asp?id=324038
| > Introduction|| |
Medulloblastoma is the most common malignant neoplasm in central nervous system, constituting approximately 20% of all pediatric brain tumors. The current standard of care for medulloblastoma consists of maximal safe resection, followed by craniospinal irradiation (CSI), involved field or posterior fossa radiation boost, and chemotherapy., The tremendous biological/molecular and technological advancements in the past two decade have significantly improved the 5-year survival rate to >80% for average risk and >50% for high-risk medulloblastoma., Consequently, there is a growing concern on long-term radiogenic side effects which includes neurocognitive decline, cataract formation, endocrine dysfunction, hypothyroidism, hearing impairment, cardiomyopathy, growth retardation, impaired fertility, and second malignancies. Several radiotherapy techniques have been implemented for CSI and continuously refined by employing photon, electron, and proton to reduce the dose to the surrounding organs at risk (OAR).,, Proton radiotherapy has demonstrated distinct dosimetric superiority over the most advanced photon radiotherapy techniques due to its Bragg peak characteristics. Intensity-modulated proton therapy (IMPT) using proton pencil beam scanning (PBS) technique allows further improvement in target dose conformity as compared to double scattering technique without the need of patient-specific beam modifying devices, which are the sources of unwanted neutron dose to patient.,, In IMPT treatment planning, proton energy, spot positions, spot spacing, and monitor unit/spot from each field were simultaneously optimized to deliver a homogeneous dose to target.
Treatment planning of CSI poses many challenges due to large irregularly shaped target volume, requiring multiple isocenters and field junctions. Above all, IMPT is susceptible to set up, range, and interfield uncertainties., While the setup and inter-field uncertainties can lead to creation of hot or cold spots in the junction region of target volume, the range uncertainty may produce underdose of target and overdose of OARs, if not taken into account during the treatment planning. Previous investigators have reported different planning approaches to mitigate these uncertainties.,,, The reported techniques differ largely in the cranial field geometry, optimization techniques, and managing the dose distribution in the junction region. An optimum cranial field arrangement should provide minimum dose to neurological OARs, specially the lens, without compromising the dose to cranial target, especially the cribriform plate. Studies show that under dose to cribriform plate to reduce the dose to lens is the primary reason for treatment failure., Second, a robust CSI treatment plan should produce a highly conformal uniform dose distribution to the entire clinical target volume (CTV) and remain unchanged even in case of possible setup, range, and inter-field uncertainty. The commonly adopted planning target volume (PTV) based optimization with complimentary dose gradient in the junction region does not guarantee robust match lines. Whereas, in the recently introduced robust optimization technique, possible setup error and range uncertainty can be included either to the plan or to each individual fields during the optimization. CTV-based robust optimization method is being widely adopted in proton therapy in the recent times. However, creating and optimizing for 81 scenarios for both setup and range uncertainty of a CSI plan with three isocenters independently leads to substantial increase in computation time, even with the graphic processing unit-based latest hardware configuration. In addition, Monte-Carlo (MC) algorithm required for better mitigation of range uncertainty and accurate dose computation demand more computational time and hence practically unrealistic in routine clinical environment. In this study, we investigate a hybrid CSI planning technique, wherein PTV was optimized for target coverage and CTV was robustly optimized for the setup uncertainties, only in superior-inferior (S-I) direction of the junction to achieve a robust dose distribution. Furthermore, the cranial beam geometry was optimized for least possible dose to neurological structure without compromising target coverage. This study describes in details the hybrid IMPT techniques developed and implemented in our center for CSI.
| > Subjects and Methods|| |
Patient simulation and contouring
Seventeen pediatric and adult patients of CNS (central nervous system) tumors were treated for CSI employing image-guided IMPT between January 2019 and May 2020 at our center. Among these, five adult/adolescent (median age: 20 years, age range: 18–36 years), patients were selected for this dosimetric study. The details of the patients are summarized in [Table 1]. All patients were immobilized in supine position on a special base of skull base plate inserts with customized moldcare headrest (Qfix Systems, Avondale, PA, USA), head-and-neck thermoplastic mask and body vaclock cradle as shown in [Figure 1]. Validation tests were carried out to ensure that the modeling of these devices is perfect in the treatment planning system (TPS) before the clinical implementation of the CSI. The planning computed tomography (CT) data were acquired from vertex to the coccyx with 2 mm slice thickness on 85 cm bore Toshiba Aquilion LB (Toshiba Medical Systems, Japan) multislice CT scanner. The delineation of gross target volume, CTV, and OAR were carried out in Raystation (Version 7 Raysearch Laboratories, Stockholm) TPS following standard guidelines., Briefly, the brain CTV comprises entire brain, cranial nerves, and meninges. The spine CTV consists of the spinal canal including cerebrospinal fluid extension to the spinal ganglia. CTVs brain and spine were summed together to define CTV-CSI. An isotropic margin of 3 mm on CTV brain and 5 mm on CTV spine were grown to create corresponding PTVs. Critical structures such as lens, eyes, optical structures, kidneys, cochleas, lungs, thyroid, esophagus, heart, liver, and genital structure were delineated.
|Table 1: List of patients with diagnosis, prescription dose in Gy RBE, planning target volume, treatment length and beam angles|
Click here to view
|Figure 1: Photograph of setup of a representative patient with skull base plate inserts, customized moldCare headrest, head-and-neck thermoplastic mask, and body vacLock cradle|
Click here to view
Hybrid robust optimization intensity-modulated proton therapy treatment planning
All treatment plans were generated in Raystation TPS, modelled for Proteus plus (Ion Beam Application, Belgium) proton therapy system. Proteus plus is an isochronous cyclotron-based three-room (two gantry and one fix beam line) image-guided proton therapy system equipped with dedicated PBS nozzle. It can modulate proton energy from 70.2 to 226.2 MeV (range 4.1–32 g/cm2) with corresponding spot size of 6.7–3 mm sigma in air and at isocenter. For a head first supine patient position, our system can magnetically position the spot anywhere in a 30 × 40 cm2 field size which corresponds to the lateral and longitudinal axis of the patient, respectively. The range of proton beam can be reduced further by 7.5 g/cm2 to treat tumor superficial to 4.1 g/cm2 using an add-on Lexan (density = 1.25 g/cm2) range shifter (RS) having physical thickness of 6.0 cm. Raystation TPS supports both pencil beam (PB) and MC dose calculation algorithms and the same have been validated for our beam model during the clinical commissioning.
The robustly optimized IMPT treatment planning of CSI demand creation of spot assignment structures (SAS) to enable the optimizer to create spot intensity map (SIM) from each field and complementary dose gradient in the junction region with robust match lines. First, spinal PTV was split into upper and lower spine. Three SAS were created from PTV brain and PTV spine as shown in [Figure 2], forming two junctions in the regions of brain and upper spine (J1) and upper and lower spine (J2). The length of the overlapping regions in each junction was maintained at 8 cm based on the finding by Lin et al. Every plan was created with three isocenters having same lateral and vertical co-ordinates [Figure 2]. The beam geometry and prescribed dose of every patient are summarized in [Table 1]. The technique and beam geometry, two noncoplanar posterior oblique fields (F1 and F2) for the brain, and a part of upper spine and one posterior field each for upper (F3) and lower spine (F4) used in this study were arrived after an extensive comparative dosimetric study conducted in three test patient datasets before clinical implementation of IMPT for CSI. Robust optimization was carried out on CTV-CSI by incorporating setup error of ± 5 mm only in S-I direction to each independent beam. This setting alone creates 81 perturbed scenarios. Raystation adopts minimax optimization formulation of stochastic programming, which enables accounting for uncertainties in the probability distributions of the errors., The minimax optimization aims at minimizing the effects of setup and range errors in the worst-case scenarios. Robust optimization also creates a complementary dose gradient in the junction regions (J1 and J2). As the robust optimization was not carried out in lateral and anterio-posterio direction, SIM was also optimized to PTV-CSI for adequate and robust dose delivery to conformal radiotherapy-CSI. Positioning of unwanted proton spots during the optimization was restricted using control structures. RS was used in all the four fields and air gap between the face of RS and patient's most extended surface was kept at a practically achievable minimum value. The plans were optimized until the clinical goal of covering at least 98% of each PTVs by 98% of the prescribed dose and 98% of each CTVs receiving at least 99% of the prescribed dose while maintaining the dose to other OARs within the clinically acceptable limit. Final doses were computed with MC algorithm using 10,000–15,000 ions/spot for 1% uncertainty.
|Figure 2: Sagittal view of a tall patient with three spot assignment structures (brain, upper [UP] spine, and lower [LW] spine) along with the two junctions (J1 between brain and upper spine and J2 between upper and lower spine)|
Click here to view
Robust plan evaluation
The dosimetry outcome of the robustly optimized hybrid plans was evaluated using standard dose volume indices derived from the cumulative dose volume histogram (DVH). Target volume coverage was analyzed as Vx%, which is the volume of the CTVs and PTVs receiving at least x% of the prescribed dose (V98%, V95%) and high dose volume as V107%. Dose homogeneity index (DHI) defined as a ratio between the doses to 95% (D95%) and 5% (D5%) of the target volume was used to evaluate dose homogeneity within each CTVs and PTVs. Further, the conformation of therapeutic dose volume to the target volume was estimated using the conformity index (CI) as defined by Paddick. The D1%, Dmean, and Vx% for OARs were also analyzed (V80%, V50%, V30%, and V10%) for every patient.
The robustness of all plans was evaluated using a robust evaluation module and an in-house python script to check the robsutness between the iso-center shifts in Raystation TPS. It enables us to create all possible error scenarios of both setup and range uncertainties and accordingly allows to study the perturbed dose distribution of the nominal plans. For each plan, 16 error scenarios were created by introducing setup uncertainties of ±3 mm along the three translational axes, namely, the anterior-posterior (A-P), S-I, and right-left directions and range uncertainties of ±3.5%. Furthermore, the setup error between the iso-centers were checked using an in-house python script by shifting the iso-centers independently by ±3 mm. This resulted in IMPT plans with perturbed dose distributions derived from the nominal IMPT plans. The worst decrease in target coverage represented by least values of V95% for CTVs and worst increase (WI) in maximum and mean dose to OARs represented by highest increase in D1% and Dmean were also evaluated from the 16 perturbed plans.
| > Results|| |
Nominal dosimetric parameters of the intensity-modulated proton therapy plans
The dosimetric outcome of the proposed robustly optimized hybrid IMPT plan of a representative patient is shown in [Figure 3]a. Whereas, [Figure 3]b, [Figure 3]c, [Figure 3]d represents the dose gradients from the complimentary fields in the junction (J1 and J2) regions. The robust optimization technique produced a homogeneous dose distribution throughout the CSI target volume including the junction regions, where the dose gradient from the complementary beams is matched at around 50% isodose [Figure 3]e and [Figure 3]f. [Table 2] shows the mean (±standard deviation [SD]) values of V98%, V95%, V107%, Dmean, D1%, DHI, and CI of CTVs and PTVs of the brain and spine separately and together from the nominal plans of the five patients enrolled for this study. The same dosimetric parameters were also reported separately for the cribriform plate. In all patients, V98% of PTVs fulfill the predefined clinical goal of 98%, with overall mean (SD) of 99.33% (±0.39%) for brain and 99.75% (±0.31%) for spinal PTVs. The CTV coverage (V98%) was also well within the clinical goal of 99% with overall mean (SD) of 99.75% (±0.26%) for brain and 99.99% (±0.03%) for spine, respectively. The proposed technique was also able to achieve excellent coverage to cribriform plate with mean (SD) value of V98% at 99.58% (±0.79). The high dose volume within the target, represented by V107% was negligible with a mean (SD) of 0.12% (±0.26%) and 0.06% (±0.04%) in PTV brain and spine, respectively. Besides, this technique resulted in homogeneous and conformal dose distribution. The mean (SD) of DHI and CI was 0.98 (±0) and 0.90 (±0.02) for PTV brain and 0.97 (±0.02) and 0.70 (±0.04) for PTV spine, respectively.
|Figure 3: Isodose distribution of the nominal robustly optimized hybrid intensity-modulated proton therapy plan in one of the representative cases showing (a) whole dose distributions (b) junction brain, (c) upper spine junction, (d) lower spine, (e) junction dose distributions and gradient formation along the junction between brain and upper spine (J1), and (f) junction between upper and lower spine (J2)|
Click here to view
|Table 2: Dose-volume indices of target in the two different treatment techniques|
Click here to view
The dose volume indices of OARs resulted from the five nominal plans are depicted in [Table 3]. All data reported as percentage of prescription dose are the mean (±SD) of the five treatment plans. The OAR doses were within the clinically acceptable limit. None of the patients receive any dose to their genitalia. The Dmean was observed to be very less in most of the extracranial OARs, especially heart and liver (<1%), followed by esophagus (<4.5%), thyroid (<5%), kidneys (<5.5%), and lungs (8.5%). The high-dose volume represented by V80% to all extracranial OARs was <1%, while the intermediate (V50% and V30%) dose volumes were <6% and 9%, respectively. The low-dose volume (V10%) was 13%–18%, except in heart and liver (<2%). The mean of D1% and Dmean was <20% and <12% for right lens, whereas it was <14% and <10% for left lens. All dose volume indices were high for eyes and optic nerves as it was part of the target volume.
|Table 3: Mean (standard deviation) of maximum dose, mean dose and the volumes of each organ at risk receiving 80% (V80%), 50% (V50%), 30% (V30%) and 10% (V10%) of the prescription dose, in two different treatment techniques|
Click here to view
[Figure 4] shows the 16 perturbed and nominal DVHs of CTV CSI [Figure 4]a and OARs [Figure 4]b of a representative patient. The narrow DVH band demonstrates the robustness of the treatment plan in terms of target coverage and dose to OARs even in the presence of 16 setup error scenarios. The widest shoulder and elongated tail in the DVH band of CTV represent the worst case scenario in terms of reduction in CTV coverage and increase in high dose. The corresponding dose distribution from nominal [Figure 5]a and worst cases scenario [Figure 5]b plans and their differences [Figure 5]c are shown in [Figure 5]. Even the worst case scenario (high and low) produces a maximum dose difference of approximately 2%–3% mostly around the junction regions and anterior to the base of skull. In all patients, the worst decrease in V95% of PTVs, CTVs, and cribriform plate from its corresponding nominal values were within 2% as shown in [Figure 6]a. The mean (SD) of V95% from all worst case scenario plans of the 6 patients was 98.38% (±0.38%) for cribriform plate, 99.64% (±0.37%) for CTV brain and 99.75% (±0.36%) for CTV spine, respectively. The corresponding values for PTVs brain and spine were 98.80% (±0.30) and 98.06% (±0.59%), respectively. [Figure 6]b and [Figure 6]c represent the distribution of WI in D1% and Dmean of OARs from the corresponding nominal plans. The increase in D1% and Dmean was noticeable in most of the OARs, except the genital organs which remain at zero in all error scenarios. The mean (SD) increase in the D1% from the nominal plans was 10.4% (±2.7%), 21.3% (±4.1%), 14.9% (±5.1%), 3.4% (±3.8%), 7% (±3.4%), 10.5% (±4.2%), and 13.3% (±0.7%) for lens, eye, esophagus, thyroid, heart, lung, liver, and kidney, respectively. The increase in Dmean was observed to be higher in lens 16.6% (±9.2%) and eye 18.4% (±2.6%), while it was lesser (<5%) in esophagus, thyroid, heart, lung, liver, and kidney.
|Figure 4: Dose volume histogram band resulted from the perturbed dose distributions of the nominal plan in 16 scenarios with 3 mm setup error and 3.5% range uncertainties in (a) clinical target volume of brain and (b) organ at risk|
Click here to view
|Figure 5: Isodose distributions of (a) nominal plan (b) corresponding worst case scenario of perturbed dose with 3 mm setup error and 3.5% range uncertainties and (c) dose difference between a and b|
Click here to view
|Figure 6: Box plot of (a) difference between worst decrease in V95% and corresponding nominal (NM) value in planning target volume, clinical target volume, and cribriform plate and (b and c) difference between worst increase in D1% and Dmean from its corresponding nominal values in OARs|
Click here to view
| > Discussion|| |
In this study, we describe a robustly optimized hybrid IMPT technique of CSI routinely used in our institute. Our technique is different from the techniques reported elsewhere,,,,, with regard to optimization approach and beam geometry to the brain target. The two posterior non-coplanar beams used in our study were chosen as a fine balance between minimum number of beams, less dose to lens, and improved target coverage, especially in cribriform plate. The overall mean (SD) of D1% of right (5.63 Gy RBE [1.04 Gy RBE]) and left (4.14 Gy RBE [1.01 Gy RBE]) lens doses observed in our study was comparable to the best results reported in the literature.,, In the study by Farace et al., right and left lens doses were least at 5.8 Gy RBE and 6.0 Gy RBE from three-beam (bilateral and posterior) arrangement when compared to 15.7 and 15.7 Gy RBE from two oblique-posterior beams and 17.6 and 17.5 Gy RBE from two opposed-lateral beams. Stoker et al. reported a mean maximum dose of 11.16 Gy RBE (1.3 Gy RBE) and 10.5 Gy RBE (2 Gy RBE) to lens for pediatric and adult patients planned with two anterior oblique. The two noncoplanar posterior oblique beam used in our study not only reduced treatment time as compared to three-field technique but also provide the flexibility in treating cranial and spinal field junction, and hence the possibility of treating short patients with only two isocenters. Adequate coverage of cribriform plate without exceeding the tolerance doses to neighboring OARs, especially lens, were challenging and any sub-optimal coverage of this region may lead to treatment relapse., These studies reported recurrence in cribriform plate which was undertreated with commonly employed radiotherapy techniques., We have ensured that cribriform plate was covered adequately by the prescription dose in our patients without increasing the dose to lenses and eyes. The mean (SD) value of V98% in cribriform plate of 99.9% (±0.04%) achieved in our study was comparable to the minimum dose of 35.42 ± 9.8 Gy RBE (98.4%) reported by Lin et al. In all these studies,,, PB algorithm was used, whereas we used MC due to the limitation of PB algorithm in handling inhomogeneous medium and inaccuracy of dose distribution in proximal depth when the RS was used. The dose to all other OARs was comparable with the previously published data from other centers.,,,
A commonly used approach to create the dose gradient in junction region is the volumetric gradient dose optimization., In another method, numerous optimization structures can be created which force the optimizer dose to incrementally step down within the match lines to create the gradient. However, these approaches require considerable time to delineate the optimization structures in order to optimize the dose in the junction region. Farace et al. proposed a method of using ancillary beam to create the gradient, but it requires an external script, implemented through scripting module in the TPS. In our study, we have adopted hybrid optimization approach where PTV was used for the target coverage, and the robustness was applied only in SI direction not only to create the dose gradient in the junction but also to produce a robust match line. The advantage of our method is that the junction dose gradient was created automatically based on the overlapping volumes of the SAS structures, thus reducing the optimization time and also without any external script. At the time of this study, RayStation TPS version 7 supports introduction of setup errors either in each field or to the whole plan. As the CSI is always planned and treated with multiple isocenters, we have introduced setup error and range uncertainty in each field separately. Ideally, we must have introduced setup error in all three translational axes and both the direction along with range uncertainty. In such condition, the robust optimizer will have to optimize for 7203 scenarios, which will increase the computation time enormously. Moreover, we used MC algorithm with tight uncertainty of 1% in majority of clinical sites including CSI. This would demand unrealistic treatment planning time in routine clinical practice. Therefore, full CTV-based robust optimization approach is impractical citing the computation time required to get a clinically acceptable plan and hence lead us to explore an alternative option, wherein we get a robust plan yet in lesser time. We have used robust optimization only in the S-I direction to avoid the dosimetric variation due to inter-field shift and produce robust match line. Moreover, as the proton range variation is predominant along the direction of the beam, we decided not to use range uncertainty during robust optimization. This optimization setting has reduced the optimization scenarios to just 81 and in turn the planning time, which is still considered large as compared to other clinical plans with single isocenter. This plan can potentially result a robust plan in S-I direction and in junction region, but does not guarantee the same in lateral and A-P direction. Therefore, we hybrid method hybrid method, wherein PTVs were optimized for adequate CTV coverage even in worst case scenarios. This hybrid method does not lead to any significant increase in planning time. The final treatment plan of every patient resulted in satisfactory dosimetry outcomes and passed the clinical goals, even in worst case scenarios.
| > Conclusions|| |
We presented an IMPT plan for CSI, using CTV-based robust optimization along the length of the target and PTV-based optimization for uniform target coverage. This planning technique is efficient and robust against the setup and range uncertainties. It could be adopted in other proton therapy centers with PBS proton technology.
Financial support and sponsorship
Conflicts of interest
There are no conflicts of interest.
| > References|| |
Central Brain Tumor Registry of the United States (CBTRUS). Statistical Report 2005–2006. Primary Brain Tumors in the United States: 1998–2002. Hinsdale, IL: CBTRUS; 2005.
Packer RJ, Sutton LN, Goldwein JW, Perilongo G, Bunin G, Ryan J, et al
. Improved survival with the use of adjuvant chemotherapy in the treatment of medulloblastoma. J Neurosurg 1991;74:433-40.
Packer RJ, Gajjar A, Vezina G, Rorke-Adams L, Burger PC, Robertson PL, et al
. Phase III study of craniospinal radiation therapy followed by adjuvant chemotherapy for newly diagnosed average-risk medulloblastoma. J Clin Oncol 2006;24:4202-8.
Gajjar A, Hernan R, Kocak M, Fuller C, Lee Y, McKinnon PJ, et al
. Clinical, histopathologic, and molecular markers of prognosis: Toward a new disease risk stratification system for medulloblastoma. J Clin Oncol 2004;22:984-93.
Packer RJ, Goldwein J, Nicholson HS, Vezina LG, Allen JC, Ris MD, et al
. Treatment of children with medulloblastomas with reduced-dose craniospinal radiation therapyandadju- vant chemotherapy: A Children's Cancer Group study. J Clin Oncol 1999;17:2127-36.
Chojnacka M, Skowrońska-Gardas A, Morawska-Kaczyńska M, Zygmuntowicz-Piętka A, Pędziwiatr K, Semaniak A. Craniospinal radiotherapy in children: Electron- or photon-based technique of spinal irradiation. Rep Pract Oncol Radiother 2010;15:21-4.
Lee CT, Bilton SD, Famiglietti RM, Riley BA, Mahajan A, Chang EL, et al
. Treatment planning with protons for pediatric retinoblastoma, medulloblastoma, and pelvic sarcoma: How do protons compare with other conformal techniques? Int J Radiat Oncol Biol Phys 2005;63:362-72.
Seravalli E, Bosman M, Lassen-Ramshad Y, Vestergaard A, Oldenburger F, Visser J, et al
. Dosimetric comparison of five different techniques for craniospinal irradiation across 15 European centers: Analysis on behalf of the SIOP-E-BTG (radiotherapy working group). Acta Oncol 2018;57:1240-9.
Stoker JB, Grant J, Zhu XR, Pidikiti R, Mahajan A, Grosshans DR. Intensity modulated proton therapy for craniospinal irradiation: Organ-at-risk exposure and a low-gradient junctioningtechnique. Int J Radiat Oncol Biol Phys 2014;90:637-44.
Farace P, Bizzocchi N, Righetto R, Fellin F, Fracchiolla F, Lorentini S, et al
. Supine craniospinal irradiation in pediatric patients by proton pencil beam scanning. Radiother Oncol 2017;123:112-8.
Lin H, Ding X, Kirk M, Liu H, Zhai H, Hill-Kayser CE, et al
. Supine craniospinal irradiation using a proton pencil beam scanning technique without match line changes for field junctions. Int J Radiat Oncol Biol Phys 2014;90:71-8.
Zaghian M, Cao W, Liu W, Kardar L, Randeniya S, Mohan R, et al
. Comparison of linear and nonlinear programming approaches for “worst case dose” and “minmax” robust optimization of intensity-modulated proton therapy dose distributions. J Appl Clin Med Phys 2017;18:15-25.
Fredriksson A, Forsgren A, Hårdemark B. Minimax optimization for handling range and setup uncertainties in proton therapy. Med Phys 2011;38:1672-84.
Weiss E, Krebeck M, Kohler B, Pradier O, Hess CF. Does the standardised helmet technique load to adequate coverage of the cribriform plate? An analysis of current practice with respect to the ICRU50 report. Int J Radiat Oncol Biol Phys 2001;49:1475-80.
Jereb B, Krishnaswami S, Reid A, Allen JC. Radiation for medulloblastoma adjusted to prevent recurrence to the cribriform plate region. Cancer 1984;54:602-4.
Liu W, Zhang X, Li Y, Mohan R. Robust optimization of intensity modulated proton therapy. Med Phys 2012;39:1079-91.
Paganetti H. Range uncertainties in proton therapy and the role of Monte Carlo simulations. Phys Med Biol 2012;57:R99-117.
Widesott L, Lorentini S, Fracchiolla F, Farace P, Schwarz M. Improvements in pencil beam scanning proton therapy dose calculation accuracy in brain tumor cases with a commercial Monte Carlo algorithm. Phys Med Biol 2018;63:145016.
Ajithkumar T, Horan G, Padovani L, Thorp N, Timmermann B, Alapetite C, et al
. SIOPE-Brain tumor group consensus guideline on craniospinal target volume delineation for high-precision radiotherapy. Radiother Oncol 2018;128:192-7.
Parker W, Filion E, Roberge D, Freeman CR. Intensity-modulated radiotherapy for craniospinal irradiation: Target volume considerations, dose constraints, and competing risks. Int J Radiat Oncol Biol Phys 2007;69:251-7.
Paddick I. A simple scoring ratio to index the conformity of radiosurgical treatment plans. Technical note. J Neurosurg 2000;93 Suppl 3:219-22.
Sharma DS, Gupta T, Jalali R, Master Z, Phurailatpam RD, Sarin R. High-precision radiotherapy for craniospinal irradiation: Evaluation of three-dimensional conformal radiotherapy, intensity-modulated radiation therapy and helical TomoTherapy. Br J Radiol 2009;82:1000-9.
Tasson A, Laack NN, Beltran C. Clinical implementation of robust optimization for craniospinal irradiation. Cancers (Basel) 2018;10:7.
Canamas LO, Crispín V, Gomà C. Robust proton CSI planning in proteusone. Phys Medica. 2018;52:137.
[Figure 1], [Figure 2], [Figure 3], [Figure 4], [Figure 5], [Figure 6]
[Table 1], [Table 2], [Table 3]