Advertisement

Fully automated planning and delivery of hippocampal-sparing whole brain irradiation

Published:September 01, 2021DOI:https://doi.org/10.1016/j.meddos.2021.06.004

      Abstract

      The goal of this study is to fully automate the treatment planning and delivery process of hippocampal-sparing whole brain irradiation (HS-WBRT) by combining a RapidPlan (RP) knowledge-based planning model and HyperArc (HA) technology. Additionally, this study compares the dosimetric performance of RapidPlan-HyperArc (RP-HA) treatment plans with RP plans and volumetric modulated arc therapy (VMAT) plans.
      Ten patients previously treated with HS-WBRT using conventional VMAT were re-planned using RP-HA technique and RP model for HS-WBRT. Treatment plans were generated for 30Gy in 3Gy fractions using 6MV photon beam on a TrueBeam linear accelerator (Varian Medical Systems, Palo Alto, CA) equipped with high definition multileaf collimator (HDMLC). Target coverage, homogeneity index (HI), Paddick Conformity index (CI), dose to organs-at-risk (OARs) provided by the 3 planning modalities were compared, and a paired t-test was performed. Total number of monitor units (MU), effective planning time and beam-on-time time were reported and evaluated for each plan.
      RP-HA plans achieved on average a 4% increase in D98% of PTV, a 26% improvement in HI, a 2.3% increase in CI, when compared to RP plans. Furthermore, RP-HA plans provided on average 11% decrease in D100% of hippocampi when compared to VMAT plans. All RP-HA plans were generated in less than 30 minutes while RP plans took 40 minutes and VMAT plans required on average 9 hours to complete. Regarding beam-on-time time, it was estimated that RP-HA plans take on average 5 minutes to deliver while RP and VMAT plans require 6.5 and 10 minutes, respectively.
      RP-HA method provides fully automated planning and delivery for HS-WBRT. The auto-generated plans together with automated treatment delivery allow standardization of plan quality, increased efficiency and ultimately improved patient care.

      Keywords

      Introduction

      It is estimated that up to 40% of cancer patients will develop brain metastasis
      • Khuntia D
      • Brown P
      • Li J
      Whole-brain radiotherapy in the management of brain metastasis.
      during the course of their disease. Various treatment options such as surgical resection, stereotactic radiosurgery (SRS) and whole brain radiotherapy (WBRT) are available for these patients and treatment recommendations are based on patient's age and performance status as well as size, location and number of brain metastases
      • Aoyama H
      • Tago M
      • Kato N
      • et al.
      Neurocognitive function of patients with brain metastasis who received either whole brain radiotherapy plus stereotactic radiosurgery or radiosurgery alone.
      ,
      • Monaco 3rd EA
      • Faraji AH
      • Berkowitz O
      • et al.
      Leukoencephalopathy after whole-brain radiation therapy plus radiosurgery versus radiosurgery alone for metastatic lung cancer.
      . While WBRT is considered the standard treatment for multiple brain metastasis it is associated with significant neurocognitive deficit and decline in patient reported quality of life (QOL)
      • Monaco 3rd EA
      • Faraji AH
      • Berkowitz O
      • et al.
      Leukoencephalopathy after whole-brain radiation therapy plus radiosurgery versus radiosurgery alone for metastatic lung cancer.
      • Monje M
      • Mizumatsu S
      • Fike J
      • et al.
      Irradiation induces neural precursor-cell dysfunction.
      • Roman DD
      • Sperduto PW.
      Neuropsychological effects of cranial radiation: Current knowledge and future directions.
      .
      Over the last several years hippocampal sparing whole brain radiation therapy (HS-WBRT) has gained support
      • Gondi V
      • Tome WA
      • Mehta MP
      • et al.
      Why avoid the hippocampus? A comprehensive review.
      as a method to mitigate neurocognitive toxicity due to radiation. The Radiation Therapy Oncology Group (RTOG) 0933
      • Gondi V
      • Pugh SL
      • Tome WA
      • et al.
      Preservation of memory with conformal avoidance of the hippocampal neural stem-cell compartment during whole-brain radiotherapy for brain metastases (RTOG 0933): a phase II multi-institutional trial.
      Phase II multi-institutional trial showed preservation of memory and improved QOL for HS-WBRT compared to a historical series of patients receiving standard WBRT. Additionally, results of the NRG Oncology CC001 Phase III
      • Brown PD
      • Gondi V
      • Pugh SL
      • et al.
      Hippocampal avoidance during whole-brain radiotherapy plus memantine for patients with brain metastases: phase III trial NRG oncology CC001.
      randomized trial, investigating WBRT and memantine with and without hippocampal avoidance found significantly higher risk of neurocognitive failure for the WBRT and memantine without hippocampal avoidance arm.
      Sparing the hippocampus while providing acceptable dose coverage and homogeneity to the remaining brain parenchyma poses significant challenges with respect to treatment planning. Several planning techniques and delivery methods including helical tomotherapy and linear accelerator-based intensity modulated radiation therapy (IMRT)
      • Gondi V
      • Tolakanahalli R
      • Mehta MP
      • et al.
      Hippocampal-sparing whole-brain radiotherapy: a "how-to" technique using helical tomotherapy and linear accelerator-based intensity-modulated radiotherapy.
      have been reported in the literature. More recently, volumetric arc therapy (VMAT)
      • Rong Y
      • Evans J
      • Xu-Welliver M
      • et al.
      Dosimetric evaluation of intensity-modulated radiotherapy, volumetric modulated arc therapy, and helical tomotherapy for hippocampal-avoidance whole brain radiotherapy.
      • Lagerwaard FJ
      • Van der Hoorn EA
      • Verbakel WF
      • et al.
      Whole-brain radiotherapy with simultaneous integrated boost to multiple brain metastases using volumetric modulated arc therapy.
      • Shen J
      • Bender E
      • Yaparpalvi R
      • et al.
      An efficient Volumetric Arc Therapy treatment planning approach for hippocampal-avoidance whole-brain radiation therapy (HA-WBRT).
      showed superior performance compared to IMRT in minimizing the dose to the hippocampus and reducing the overall delivery time. While both IMRT and VMAT modalities are capable of delivering HS-WBRT, they require extensive planning times and experienced planners capable of addressing the challenging trade-off between target coverage and organs at risk (OARs) sparing.
      Recently, a commercial knowledge base planning (KBP) tool - RapidPlan (RP) engine
      DVH Estimation Algorithm for RapidPlan
      Eclipse Photon and Electron Algorithms Reference Guide.
      (Varian Medical Systems, Palo Alto, CA) – has been made available that uses machine learning methods to build models based on a library of high-quality plans. The models are used to estimate dose-volumes histograms (DVHs) for OARs on new patients. The DVH estimations and the automatically generated optimization parameters and priorities are then directed to the optimizer to generate deliverable plans
      • Fogliata A
      • Cozzi L
      • Reggiori G
      • et al.
      RapidPlan knowledge based planning: iterative learning process and model ability to steer planning strategies.
      . RP models have been developed for multiple anatomical sites including HS-WBRT

      Magliari A, Magliari V, Foster R. Hippocampal Sparing Whole Brain (HCSWB) Model Description. Varian Medical Systems. https://oncopeer.com.

      , stereotactic radiosurgery (SRS)
      • Thomas E
      • Phillips H
      • Popple R
      Development of a knowledge based model (RapidPlan) for brain metastasis stereotactic radiosurgery and validation with automated non-coplanar treatment planning (HyperArc).
      , lung
      • Snyder KC
      • Kim J
      • Reding A
      • et al.
      Development and evaluation of a clinical model for lung cancer patients using stereotactic body radiotherapy (SBRT) within a knowledge-based algorithm for treatment planning.
      , prostate
      • Fogliata A
      • Belosi F
      • Clivio A
      • et al.
      On the pre-clinical validation of a commercial model-based optimization engine: application to volumetric modulated arc therapy for patients with lung or prostate cancer.
      , head and neck
      • Tol JP
      • Delaney AR
      • Dahele M
      • et al.
      Evaluation of a knowledge-based planning solution for head and neck cancer.
      , liver
      • Fogliata A
      • Wang P
      • Belosi F
      • et al.
      Assessment of a model based optimization engine for volumetric modulated arc therapy for patients with advanced hepatocellular cancer.
      , spine
      • Foy JJ
      • Marsh R
      • Ten Haken RK
      • et al.
      An analysis of knowledge-based planning for stereotactic body radiation therapy of the spine.
      etc. and their performance has been evaluated extensively in the literature. Those studies have confirmed that RP models were able to generate plans that are clinically acceptable, often times in a single optimization. Moreover, RP allows less experienced planners to efficiently generate high quality plans improving consistency and efficiency in the treatment planning process.
      The RP model for intracranial SRS
      • Thomas E
      • Phillips H
      • Popple R
      Development of a knowledge based model (RapidPlan) for brain metastasis stereotactic radiosurgery and validation with automated non-coplanar treatment planning (HyperArc).
      was created and validated with an automated non-coplanar treatment planning technique called HyperArcTM (Varian Medical Systems, Palo Alto, CA). HyperArc is a novel VMAT solution for SRS treatments that reduces the treatment planning time compared to conventional VMAT and provides automated treatment delivery on the linac. Recent work has found HyperArc capable of achieving steeper dose gradient for targets compared to conventional VMAT and lower dose spread into the normal brain while minimizing inter-planner variability and treatment time
      • Ohira S
      • Ueda Y
      • Akino Y
      • et al.
      HyperArc VMAT planning for single and multiple brain metastases stereotactic radiosurgery: a new treatment planning approach.
      .
      Our study aims to investigate the dosimetric benefit of combining a HS-WBRT RP model and HyperArc technology (RP-HA) as a fully automated treatment planning and delivery process for HS-WBRT. Additionally, the study will compare the dosimetric performance of RP-HA treatment plans with RP plans and conventional VMAT plans. While extensive literature has been dedicated to VMAT techniques for HS-WBRT treatments to our knowledge this is the first investigation regarding effectiveness of RP-HA as a fully automated treatment modality.

      Methods

      Patient selection

      Ten patients previously treated with HS-WBRT using conventional VMAT that were either enrolled on NRG CC001, or followed NRG CC001 protocol planning requirements were selected for this study. All 10 patients were re-planned using RP-HA technique and RP model. The RP-HA plans were compared to the RP and the VMAT plans. This retrospective study was approved by the Institutional Review Board of Loyola University Medical Center.

      Patient simulation

      Computed tomography (CT) scans were performed on a Somatom scanner (Siemens Medical Solutions Malvern, PA) following the NRG CC001 protocol criteria. Patients were immobilized in the supine position with the EncompassTM SRS Immobilization system (QFix, Avondale, PA) and CT scans were acquired with a slice thickness of 1.5mm.
      T1-weighted contrast-enhanced magnetic resonance images (MRI) were acquired for all patients with a slice thickness no greater than 1.5mm. The MRI images were registered to the planning CT and used for target and OARs delineation. The clinical target volume (CTV) was defined as the whole-brain parenchyma to the foreman magnum. The hippocampus was contoured following the NRG CC001 protocol guidelines and the hippocampus avoidance region was generated by expanding the bilateral hippocampi volumetrically by 5mm. The PTV consisted of the CTV excluding the hippocampus avoidance region. Other normal structures delineated included eyes, lenses, optic nerves, chiasm and brainstem.

      Treatment planning

      Treatment plans were generated for 30Gy in 3Gy fractions using 6MV photon beam on a TrueBeam linear accelerator (Varian Medical Systems, Palo Alto, CA) equipped with high definition multileaf collimator (HDMLC). All plans were generated in Eclipse TPS using the anisotropic analytic algorithm (AAA, Varian Eclipse TPS, version 15.5) with a 2mm dose calculation grid size. Conventional VMAT plans were generated by 4 different experienced planners following the NRG CC001 guidelines. An average of 5 non-coplanar arcs (range 4-6 arcs) were used and multiple optimizations were required to achieve dosimetrically acceptable plans according to NRG CC001 protocol.

      RapidPlan (RP) model

      RP-VMAT treatment plans were generated using the publicly available RP model for HS-WBRT

      Magliari A, Magliari V, Foster R. Hippocampal Sparing Whole Brain (HCSWB) Model Description. Varian Medical Systems. https://oncopeer.com.

      . The model was created following RTOG 0933 guidelines and was trained and validated using 20 clinical cases. The model utilizes a 4 arc VMAT technique, 2 full coplanar arcs and 2 vertex arcs, with clock-wise and counter clock-wise gantry rotation and mirrored collimator angles of 3150 and 450. The coplanar arcs have 359.8 degrees of arc rotation while the vertex arcs have 184.9 degrees, extending 50 past the 00 gantry position.
      For optimization purposes, the HS-WBRT model requires the user to generate an optimization target (PTVopti), defined as the whole brain minus the hippocampus expanded volumetrically by 8mm. In other words, the PTVopti is the protocol PTV with an additional 3mm margin around the hippocampus. Additionally, the model uses a MU objective with a minimum and maximum MU of 1000 and 2000 respectively. After generating the PTVopti structure and inserting the 4 VMAT arcs the HS-WBRT model is applied and the optimization is performed using the automatically generated objectives and priorities. All RP plans were created using a single optimization.

      HyperArc (HA) technique

      HyperArc is a single isocenter VMAT technique for SRS treatments that provides automated placement of the isocenter and non-coplanar arcs along with optimized collimator rotation. The arc setup geometry requires 4 non-coplanar arcs with 1 full coplanar arc with couch rotation of 00, 2 half arcs with couch rotation of +/- 450 and 1 half vertex arc with couch rotation of + or – 900. HA requires the EncompassTM SRS immobilization system to be used as the patient support system which allows clearance prediction between the patient and the linear accelerator. This approach permits a preliminary collision check and a visual dry run prior to treatment delivery. Additionally, HA enables automated treatment delivery, without the need to enter the room and manually move the couch between the treatment fields.

      RapidPlan-Hyperarc (RP-HA) planning

      Fig. 1 below illustrates the workflow for RP-HA planning. First, HA plans were generated using the automated placement of the isocenter and non-coplanar arcs along with the optimized collimator rotation. After VMAT arcs have been inserted using HA, the RP model was applied for plan optimization. The optimization objectives based on DVH estimates automatically generated by the RP model were used for planning. For RP-HA planning, the protocol PTV (whole-brain parenchyma excluding hippocampal avoidance region) was used during optimization, not the PTVopti required by the RP model. This way, the automated features provided by HA (isocenter placement, non-coplanar arcs) along with auto-generated optimization objectives of RP allow full automation of the treatment plan without the need of generating pseudostructures for optimization. All RP-HA plans were created by a single iteration of plan optimization.
      Fig. 1
      Fig. 1Diagram of planning and treatment workflow for HS-WBRT using RP-HA.

      Dosimetric and plan quality analysis

      Dose metrics extracted from RTOG 0933 protocol compliance criteria were used in this analysis. All treatment plans were normalized such that 30Gy covered 95% of the PTV volume. Target coverage was assessed by the minimum dose covering 98% of the PTV volume (D98%). The dose received by the hottest 2% of the PTV (D2%) was recorded along with the median dose of the PTV (Dmedian).To quantify dose homogeneity within the target, a homogeneity index (HI)
      International Commission on Radiation Units and Measurements
      Prescribing, recording and reporting photon-beam intensity0modulated radiation therapy (IMRT):.
      was calculated for each plan as follows:
      HI=D2%D98%Dmedian.
      (1)


      Smaller HI values indicate better dose homogeneity within the target volume and a value of 0 corresponding to a completely homogeneous dose within the target.
      The conformity index was calculated according to the formula introduced by Paddick
      • Paddick I.
      A simple scoring ratio to index the conformity of radiosurgical treatment plans. Technical note.
      :
      CI=TVPIV2TV×VRI
      (2)


      where TV is the target volume, TVPIV is the target volume covered by the prescription isodose and VRI is the total volume covered by the prescription isodose. Higher values of CI indicate better dose conformity to the target volume, with a value of 1 corresponding to a reference isodose covering the exact target volume without irradiating any healthy tissue.
      Two metrics were used to assess the hippocampus structure sparing: dose to 100% of hippocampus (D100%) and the maximum dose (Dmax). Maximum dose for eyes, lenses, optic nerves and optic chiasm were also recorded. The NRG CC001 compliance criteria for target volume and normal structures are shown below in Table 1. The listed parameters correspond to ideal cases or "Per Protocol" category and the ones in parentheses belong to the "Variation Acceptable" category. Treatment plans falling outside of “Per Protocol” or “Variation Acceptable” categories are considered suboptimal and require further treatment planning optimization.
      Table 1NRG CC001 compliance criteria for target and normal structures
      OrganDose constraints
      PTVD2% ≤ 37.5Gy (D2% < 40Gy is allowed)

      D98% ≥ 25Gy (D98% > 22.5Gy is allowed)

      V30Gy ≥ 95% (V30Gy > 90% is allowed)

      HippocampiD100% ≤ 9Gy (D100% < 10Gy is allowed)

      Dmax ≤ 16Gy (Dmax < 17Gy is allowed)

      Optic Nerves and Optic ChiasmDmax ≤ 30Gy (Dmax < 37.5Gy is allowed)
      To evaluate the dose falloff between the brain and the hippocampus, three concentric ring structures were generated around the hippocampus. A 5mm 3D margin was applied to the hippocampi to create the first ring (R1). The second (R2) and the third ring (R3) were generated by applying the 5mm outer margin to R1 and R2 respectively. Ring 1 (R1) represents the buffer zone between the hippocampi and the PTV with its internal boundary abutting the hippocampi surface and external boundary abutting the PTV surface. While brain tissue encompassed by R1 structure will exhibit an unavoidable dose gradient from the hippocampal surface to the prescription isodose line, the R2 and R3 ring structures are part of the PTV and therefore expected to exhibit improved dose homogeneity compared to R1. Dose heterogeneity was assessed in each ring by computing HI as defined by Equation 1.
      The dose fall-off in the R1 structures was quantified by the Rx% and defined as follows:
      Rx%=VX%presVPTV
      (3)


      where VX%pres is the volume of brain in R1 receiving at least X% of the target prescription dose. Assuming the same target coverage, a lower Rx% value corresponds to a steeper dose fall-off between the 100% and X% dose levels. In this study we evaluated R50% and R60% to characterize the dose fall-off. The mean dose to each of the three rings along with the homogeneity index were reported. Additionally, the aforementioned Rx% values for R1 and the percentage of volume receiving 30Gy in R2 (V30Gy R2) and in R3 (V30Gy R3) were also included. .
      Along with the dosimetric parameters mentioned above the total number of monitor units (MU) and planning time were reported and evaluated for each plan. Additionally, the time needed to deliver a single fraction was estimated for all three modalities. Delivery time excluded patient setup and daily imaging procedures and was recorded from beam-on to beam-off only.
      Statistical comparison between the three modalities using a paired t-test was performed to determine if there was any significant difference between the examined parameters. A p-value < 0.05 was considered statistically significant.

      Results

      Mean and standard deviation (mean±SD) of D98%, D2%, HI and CI for the PTV are reported in Table 2 for the three planning techniques along with the corresponding p values of the paired t-test. The table also includes the RTOG 0933 protocol dose compliance criteria. Comparing the dose covering 98% of the PTV volume (D98%), RP-HA plans achieved an average of 28.6Gy which was higher compared to RP plans (27.2Gy, p<0.01) and conventional VMAT (27.8Gy, p=0.159). In the evaluation of PTV hotspots, D2%, all three modalities provided a hotspot significantly lower than the protocol recommended value of 37.5Gy.
      Table 2Dosimetric parameters for PTV, expressed as mean ± SD for each planning techniques and related p values. ** Statistically significant with p value threshold of 0.05
      PTV- Dosimetric ParametersProtocol CriteriaRP-HARPVMATp value
      RP-HA vs. RPRP-HA vs. VMAT


      D98%(Gy)


      ≥ 25Gy


      28.6 ± 0.5


      27.2 ± 0.9


      27.8 ± 1.9


      <0.01**


      0.159


      D2%(Gy)


      ≤ 37.5Gy


      32.9 ± 0.2


      32.8± 0.3


      32.3 ± 0.9


      0.822


      0.036**


      HI


      N/A


      0.14 ± 0.02


      0.19 ± 0.03


      0.15 ± 0.08


      <0.01**


      0.788


      CI


      N/A


      0.88 ± 0.01


      0.86 ± 0.01


      0.85 ± 0.07


      <0.01**


      0.229
      RP-HA plans offered statistically significant improvement for target dose homogeneity and conformity when compared to RP plans.
      Mean and standard deviations (mean±SD) of OARs for the 10 patients are reported in Table 3 along with the p values among the three planning modalities. In general, RP-HA plans achieved statistically significant decrease in D100% of hippocampus when compared to VMAT plans. Both RP-HA and RP plans accomplished significantly lower dose compared to the 9Gy allowed by the protocol. The average Dmax to optic nerves and optic chiasm was similar in all three modalities. For the doses to the eyes and lenses, structures not included in the protocol dosimetric compliance criteria, RP-HA plans performed similar to RP plans and provided statistically significant improvement when compared to VMAT plans. Of note, the standard deviations of dosimetric parameters for VMAT are much larger than RP-HA and RP as shown in Table 2 and 3.
      Table 3Dosimetric parameters for OARs, expressed as mean ± SD for each planning techniques and related p values. ** Statistically significant with p value threshold of 0.05
      StructureDosimetric Parameter (Protocol criteria)RP-HARPVMATp value
      RP-HA vs. RPRP-HA vs. VMAT
      Hippocampus

      D100% (≤9Gy)


      7.9 ± 0.2


      7.8 ± 0.3


      8.8 ± 0.2


      0.080


      <0.001**
      Dmax (≤16Gy)12.9 ± 0.712.3 ± 0.513.9 ± 1.30.040 **0.196


      Optic Nerves


      Dmax(<30Gy)


      28.7 ± 0.2


      28.5 ± 0.2


      27.5 ± 2.7


      0.078


      0.083


      Optic Chiasm


      Dmax(<30Gy)


      28.9 ± 0.1


      28.9 ± 0.3


      28.2 ± 2.6


      0.307


      0.424


      Eyes


      Dmax(Gy) (N/A)


      11.4 ± 0.7


      11.3 ± 1.2


      17.9 ± 3.8


      0.675


      <0.001**


      Lenses


      Dmax(Gy) (N/A)


      4.8 ± 0.5


      5.0 ± 0.8


      8.9 ± 3.2


      0.174


      <0.001**
      Table 4 summarizes the average mean dose and the corresponding SD along with the HI, Rx% values and V30Gy R2 and V30Gy R3 for the ring structures surrounding the hippocampi. In the first 5mm away from the hippocampi (R1) the RP-HA average mean dose was significantly higher compared to RP plans (15.8Gy vs 14.7Gy, p<0.01). While the same trend was observed for R2, starting at 1cm away from the hippocampi (R3) the mean doses were comparable between the two planning modalities. Regarding HI, RP-HA plans exhibit the largest dose heterogeneity in R1 compared to RP plans and VMAT plans. In the first shell of the PTV (R2), RP-HA plans offer improved dose homogeneity and significantly superior volume of brain PTV receiving prescription dose (V30Gy R2 %) when compared to RP plans. Moreover, V30Gy R2 % and Dmean are comparable when compared to VMAT plans despite the significant sparing of the hippocampi offered by RP-HA plans.
      Table 4Mean±SD of Dmean, HI, Rx and V30Gy R2(R3) values for the ring structures surrounding the hippocampi for each planning techniques and related p values. ** Statistically significant with p value threshold of 0.05
      Ring StructureDosimetric ParameterRP-HARPVMATp value
      RP-HA vs. RPRP-HA vs. VMAT


      R1


      Dmean(Gy)


      15.8 ± 0.6


      14.7 ± 0.5


      18.4 ± 1.9


      <0.001**


      0.015**
      HI1.05 ± 0.080.91 ± 0.70.78 ± 0.08<0.001**<0.001**


      R2

      R50%

      R60%

      Dmean (Gy)
      0.011±0.002

      0.006±0.002

      28.1 ± 0.7
      0.009±0.002

      0.005±0.001

      26.3±0.5
      0.015±0.003

      0.010±0.004

      28.8±1.6
      0.002**

      <0.001**

      <0.001**
      0.003**

      0.012**

      0.250
      HI

      V30Gy R2 (%)
      0.45 ± 0.03

      30.9 ± 9.4
      0.50 ± 0.03

      12.4 ± 4.2
      0.32 ± 0.09

      38.6 ± 16.7
      <0.001**

      <0.001**
      <0.001**

      0.289


      R3


      Dmean (Gy)


      31.4 ± 0.3


      31.2 ± 0.2


      31.1 ± 0.3


      0.121


      0.247
      HI

      V30Gy R3 (%)
      0.17 ± 0.03

      85.5 ± 4.9

      0.15 ± 0.03

      85.9 ± 5.5

      0.14 ± 0.06

      87.2 ± 6.8
      <0.001**

      0.705
      <0.001**

      0.381
      Fig. 2 shows the dose distribution at the level of the hippocampi for the three planning modalities for one representative patient. The coronal and sagittal views display doses ranging from 9Gy to 35Gy. The RP dose distribution shows larger hotspot regions compared to RP-HA and VMAT dose distribution. This translates into an inferior HI for the RP plans as shown by the mean values of HI in Table 2. For the patient presented in Fig. 2, a total of 50cc of the PTV received a dose of 33Gy for the RP plan, while RP-HA and VMAT plans resulted in 33Gy covering 27cc and 0.02cc volumes respectively. A total of 0.02cc of the PTV received a dose of 35Gy for both RP-HA and RP plan whereas the VMAT plan had no brain volume receiving 35Gy. Fig. 2 also illustrates superior sparing of the hippocampi in RP-HA and RP compared to VMAT plans. This can be seen by the dark blue dose cloud covering the hippocampi (magenta contour) in the VMAT plan and being absent from both RP-HA and RP plans.
      Fig. 2
      Fig. 2Dose color wash distribution in coronal (top) and sagittal (bottom) view at the level of the hippocami for (A) RP-HA, (B) RP and (C) VMAT. (Color version of figure ia avilable online).
      Fig. 3 compares the DVH for the PTV and various normal organs for the same patient presented in Fig. 2. The DVH supports the dosimetric analysis presented in Tables 1 and 2. While RA-HA and RP plans resulted in larger hotspots in the brain (consistent with D2%) both planning techniques significantly reduced the dose to hippocampi and hippocampi avoidance region compared to VMAT. Additionally, both RP-HA and RP led to lower doses to the lenses and the globes.
      Fig. 3
      Fig. 3Dose volume histogram (DVH) for the PTV and OARs for the patient illustrated in . Solid line = VMAT, dashed line = RP, dotted line = RP-HA.
      The average of total number of MU was similar between the RP-HA plans (1001±3) and RP plans (980±36) with a p value of 0.058, while the conventional VMAT plans used significantly lower MUs (886±140, P=0.036) than the RP-HA plans. All RP-HA plans were generated in less than 30 minutes, while the effective working time was kept to less than 4 minutes. By contrast, RP plan took on average 40 min to generate while the VMAT plans required a significantly longer time to complete, on the order of 9 hours.
      Regarding treatment delivery time, ARIA recorded on average 10 minutes delivery time for conventional VMAT plans. Although the RP-HA and RP plans presented in this study were not clinically delivered to the patients, it was estimated that treatment delivery for RP-HA plans will take on average 5 minutes while RP plans will require 6.5 minutes to be delivered. Since the total number of MUs is very consistent between patients and the treatment geometry is identical for RP-HA plans, the 5 minute delivery time was extracted from patients treated with RP-HA method but not included in this retrospective planning analysis. It was estimated that RP plans will take an additional 1.5 minute compared to RP-HA plans due to therapist walking into the treatment room and manually moving the couch.

      Discussion

      In this study, we evaluated the ability of RP-HA to produce high quality HS-WBRT treatment plans. In general, all RP-HA plans achieved the RTOG guidelines with some dosimetric parameters performing substantially better than the protocol recommended values. Additionally, RP-HA outperformed RP and conventional VMAT plans in terms of HI and CI, treatment deliverability and efficiency. Specifically, RP-HA achieved on average a 26% improvement in HI, a 2.3% increase in CI, a 4% increase in D98% of PTV when compared to RP plans.
      Although avoiding the hippocampi during WBRT is important to achieve neurocognitive benefits, it carries the likelihood of underdosing lesions near the hippocampi and ultimately resulting in an increase in local recurrence
      • Harth S
      • Abo-Madyan Y
      • Zheng L
      • et al.
      Estimation of intracranial failure risk following hippocampal-sparing whole brain radiotherapy.
      . While the dose gradient near the hippocampi must be sufficiently sharp in order to spare the hippocampi the benefit of decreased dose to the hippocampi should be weighed against the cost of lower dose to potential brain metastasis close to the hippocampi. Given the central location of the hippocampi and the surrounding PTV, the brain encompassed by R2 structure is probably more at risk for being undertreated. Therefore, it is important to evaluate not only the dose falloff around the hippocampi (R1) but also the volume of brain receiving prescription dose in the areas close to hippocampi, as in the R2 and R3 shells. Despite the steeper dose fall-off in R1 offered by RP plans, RP-HA improved the volume receiving prescription dose by a factor of 2.5 and offered a 7% increase in the average mean dose in R2 without compromising hippocampal sparing. Compared to VMAT, RP-HA plans offered the same prescription dose coverage in R2 and R3 while achieving on average a 9% decrease in hippocampi D100% dose.
      RP-HA plans were superior to RP and VMAT for both planning time and beam-on-time. Although both RP-HA and RP plans were generated using a single iteration of plan optimization, RP-HA planning required less active working time, since it allowed automated non-coplanar arc placement and eliminated the need for generating planning pseudostructures. By comparison, the VMAT plans, although generated by experienced dosimetrists, required multiple optimizations that translated into significantly longer effective working time, as well as a longer time expected for treatment delivery.
      The patient-to-patient dosimetric variability was minimal for RP-HA and RP planning modalities but larger for conventional VMAT plans, as shown by the standard deviation of the analyzed dosimetric parameters in Tables 1 and 2. Small variability in plan quality for RP-HA and RP plans is consistent with recent studies that highlight the role of automated treatment planning systems (ATPS) to improve planning quality and efficiency independent of the planner experience, therefore allowing standardization of treatment planning
      • Krayenbuehl J
      • Matino M
      • Guckenberger M
      • et al.
      Improved plan quality with automated radiotherapy planning for whole brain with hippocampus sparing: a comparison to the RTOG 0933 trial.
      • Wang S
      • Zheng D
      • Zhang C
      • et al.
      Automatic planning on hippocampal avoidance whole-brain radiotherapy.
      • Zieminski S.
      • Khandekar M.
      • Wang Y.
      Assessment of multi-criteria optimization (MCO) for volumetric modulated arc therapy (VMAT) in hippocampal avoidance whole brain radiation therapy (HA-WBRT).
      . With respect to treatment time, the automated treatment delivery and the preliminary collision check, which allows the avoidance of dry runs prior to the treatment provide a significant decrease in delivery time and potentially reduce the risk of intrafraction motion and improve patient comfort. Our RP-HA method allows automated treatment delivery with virtual collision check in Eclipse TPS while meeting all the RTOG dose parameter criteria.
      The study performed by Krayenbuehl et al.
      • Krayenbuehl J
      • Matino M
      • Guckenberger M
      • et al.
      Improved plan quality with automated radiotherapy planning for whole brain with hippocampus sparing: a comparison to the RTOG 0933 trial.
      examined the feasibility of Automated Planning (AP) included in Pinnacle (Philips Radiation Oncology Systems, Fitchburg, WI) for HS-WBRT using 4 non-coplanar VMAT technique. While they reported hippocampi's D100% of 8.1 Gy (7.8-8.5 range) and hippocampi's Dmax of 14.1 Gy (12.0-15.3 range), the HI was 0.24 (0.21-0.26 range) and the V30Gy was only 92% (plans normalized to 92% of the PTV receiving 30Gy). Both HI and V30Gy were substantially inferior compared to our RP-HA plans. The authors emphasized the minimum effective working time for plan generation and although treatment delivery time is not reported one can assume it remains longer compared to the automated treatment delivery offered by RP-HA modality.
      Wang et al.
      • Wang S
      • Zheng D
      • Zhang C
      • et al.
      Automatic planning on hippocampal avoidance whole-brain radiotherapy.
      also looked at AP for HS WBRT using only two coplanar arcs. They reported the hippocampi's D100% and Dmax above the protocol recommended values with a range of 8.3-10 Gy and 14.6-16.9 Gy respectively and the HI in the range of 0.23-0.33. While the use of two coplanar arcs for plan optimization reduces the beam-on time, the reported dosimetric parameters are inferior to our study. The work by Zieminski et al.
      • Zieminski S.
      • Khandekar M.
      • Wang Y.
      Assessment of multi-criteria optimization (MCO) for volumetric modulated arc therapy (VMAT) in hippocampal avoidance whole brain radiation therapy (HA-WBRT).
      compared the dosimetric performance of VMAT using standard optimization and multi-criteria optimization (MCO) for HS-WBRT with simultaneous integrated boost (SIB) to 37.5 Gy for specific metastatic lesions. For the MCO-VMAT plans they reported 7.8±0.5 Gy for hippocampi's D100% while the hippocampi Dmax and V30Gy of PTV were above the protocol recommendations, with values of 15.9±1.5 Gy and 92.7±1.8% respectively. The authors highlight the MCO operational flexibility and planning efficiency and conclude that MCO-VMAT is superior to standard optimization for HS-WBRT.
      In summary, we demonstrated that the RP-HA method for HS-WBRT reduces treatment planning and delivery time by automating both processes. Furthermore, this method consistently generates plans meeting all the RTOG criteria for clinical cases, which indicates that it creates higher quality plans compared to other planning techniques (RP and VMAT) and previous studies of HS-WBRT.

      Conclusions

      HA technology combined with RP model provides fully automated planning and delivery for HS-WBRT. The auto-generated plans together with automated treatment delivery increase planning consistency and efficiency. PP-HA ultimately improves patient care by delivering a quality complex treatment safely in a shorter period of time.

      Conflict of Interest

      No conflict of interest needs to be reported

      References

        • Khuntia D
        • Brown P
        • Li J
        Whole-brain radiotherapy in the management of brain metastasis.
        J Clin Oncol. 2006; 24: 1295-1304https://doi.org/10.1200/JCO.2005.04.6185
        • Aoyama H
        • Tago M
        • Kato N
        • et al.
        Neurocognitive function of patients with brain metastasis who received either whole brain radiotherapy plus stereotactic radiosurgery or radiosurgery alone.
        Int J Radiat Oncol Biol Phys. 2007; 68: 1388-1395https://doi.org/10.1016/j.ijrobp.2007.03.048
        • Monaco 3rd EA
        • Faraji AH
        • Berkowitz O
        • et al.
        Leukoencephalopathy after whole-brain radiation therapy plus radiosurgery versus radiosurgery alone for metastatic lung cancer.
        Cancer. 2013; 119: 226-232https://doi.org/10.1002/cncr.27504
        • Monje M
        • Mizumatsu S
        • Fike J
        • et al.
        Irradiation induces neural precursor-cell dysfunction.
        Nat Med. 2002; 8: 955-962https://doi.org/10.1038/nm749
        • Roman DD
        • Sperduto PW.
        Neuropsychological effects of cranial radiation: Current knowledge and future directions.
        J Radiat Oncol Biol Phys. 1995; 31: 983-998https://doi.org/10.1016/0360-3016(94)00550-8
        • Gondi V
        • Tome WA
        • Mehta MP
        • et al.
        Why avoid the hippocampus? A comprehensive review.
        Radiother Oncol. 2010; 97: 370-376https://doi.org/10.1016/j.radonc.2010.09.013
        • Gondi V
        • Pugh SL
        • Tome WA
        • et al.
        Preservation of memory with conformal avoidance of the hippocampal neural stem-cell compartment during whole-brain radiotherapy for brain metastases (RTOG 0933): a phase II multi-institutional trial.
        J Clin Oncol. 2014; 32: 3810-3816https://doi.org/10.1200/JCO.2014.57.2909
        • Brown PD
        • Gondi V
        • Pugh SL
        • et al.
        Hippocampal avoidance during whole-brain radiotherapy plus memantine for patients with brain metastases: phase III trial NRG oncology CC001.
        J Clin Oncol. 2020; 38: 1019-1029https://doi.org/10.1200/JCO.19.02767
        • Gondi V
        • Tolakanahalli R
        • Mehta MP
        • et al.
        Hippocampal-sparing whole-brain radiotherapy: a "how-to" technique using helical tomotherapy and linear accelerator-based intensity-modulated radiotherapy.
        Int J Radiat Oncol Biol Phys. 2010; 78: 1244-1252https://doi.org/10.1016/j.ijrobp.2010.01.039
        • Rong Y
        • Evans J
        • Xu-Welliver M
        • et al.
        Dosimetric evaluation of intensity-modulated radiotherapy, volumetric modulated arc therapy, and helical tomotherapy for hippocampal-avoidance whole brain radiotherapy.
        PLoS ONE. 2015; 10e0126222https://doi.org/10.1371/journal.pone.0126222
        • Lagerwaard FJ
        • Van der Hoorn EA
        • Verbakel WF
        • et al.
        Whole-brain radiotherapy with simultaneous integrated boost to multiple brain metastases using volumetric modulated arc therapy.
        Int J Radiat Oncol Biol Phys. 2009; 75: 253-259https://doi.org/10.1016/j.ijrobp.2009.03.029
        • Shen J
        • Bender E
        • Yaparpalvi R
        • et al.
        An efficient Volumetric Arc Therapy treatment planning approach for hippocampal-avoidance whole-brain radiation therapy (HA-WBRT).
        Med Dos. 2015; 40: 205-209https://doi.org/10.1016/j.meddos.2014.11.007
        • DVH Estimation Algorithm for RapidPlan
        Eclipse Photon and Electron Algorithms Reference Guide.
        Varian Medical Systems, Palo Alto, CADec 2014
        • Fogliata A
        • Cozzi L
        • Reggiori G
        • et al.
        RapidPlan knowledge based planning: iterative learning process and model ability to steer planning strategies.
        Radiat Oncol. 2019; 14: 187https://doi.org/10.1186/s13014-019-1403-0
      1. Magliari A, Magliari V, Foster R. Hippocampal Sparing Whole Brain (HCSWB) Model Description. Varian Medical Systems. https://oncopeer.com.

        • Thomas E
        • Phillips H
        • Popple R
        Development of a knowledge based model (RapidPlan) for brain metastasis stereotactic radiosurgery and validation with automated non-coplanar treatment planning (HyperArc).
        Int J Radiat Oncol Biol Phys. 2017; 99: E727-E728https://doi.org/10.1016/j.ijrobp.2017.06.2353
        • Snyder KC
        • Kim J
        • Reding A
        • et al.
        Development and evaluation of a clinical model for lung cancer patients using stereotactic body radiotherapy (SBRT) within a knowledge-based algorithm for treatment planning.
        J Appl Clin Med Phys. 2016; 17: 263-275https://doi.org/10.1120/jacmp.v17i6.6429
        • Fogliata A
        • Belosi F
        • Clivio A
        • et al.
        On the pre-clinical validation of a commercial model-based optimization engine: application to volumetric modulated arc therapy for patients with lung or prostate cancer.
        Radiother Oncol. 2014; 113: 385-391https://doi.org/10.1016/j.radonc.2014.11.009
        • Tol JP
        • Delaney AR
        • Dahele M
        • et al.
        Evaluation of a knowledge-based planning solution for head and neck cancer.
        Int J Radiat Oncol Biol Phys. 2015; 91: 612-620https://doi.org/10.1016/j.ijrobp.2014.11.014
        • Fogliata A
        • Wang P
        • Belosi F
        • et al.
        Assessment of a model based optimization engine for volumetric modulated arc therapy for patients with advanced hepatocellular cancer.
        Radiat Oncol. 2014; 9: 236-242https://doi.org/10.1186/s13014-014-0236-0
        • Foy JJ
        • Marsh R
        • Ten Haken RK
        • et al.
        An analysis of knowledge-based planning for stereotactic body radiation therapy of the spine.
        Pract Radiat Oncol. 2017; 7: e355-e360https://doi.org/10.1016/j.prro.2017.02.007
        • Ohira S
        • Ueda Y
        • Akino Y
        • et al.
        HyperArc VMAT planning for single and multiple brain metastases stereotactic radiosurgery: a new treatment planning approach.
        Radiat Oncol. 2018; 13https://doi.org/10.1186/s13014-017-0948-z
        • International Commission on Radiation Units and Measurements
        Prescribing, recording and reporting photon-beam intensity0modulated radiation therapy (IMRT):.
        ICRU Report. 2010; 83https://doi.org/10.1093/jicru/ndq002
        • Paddick I.
        A simple scoring ratio to index the conformity of radiosurgical treatment plans. Technical note.
        J. Neurosurg. 2000; 93: 219-222https://doi.org/10.3171/jns.2000.93.supplement_3.0219
        • Harth S
        • Abo-Madyan Y
        • Zheng L
        • et al.
        Estimation of intracranial failure risk following hippocampal-sparing whole brain radiotherapy.
        Radiother Oncol. 2013; 109: 152-158https://doi.org/10.1016/j.radonc.2013.09.009
        • Krayenbuehl J
        • Matino M
        • Guckenberger M
        • et al.
        Improved plan quality with automated radiotherapy planning for whole brain with hippocampus sparing: a comparison to the RTOG 0933 trial.
        Rad, Oncol. 2017; 12: 161-168https://doi.org/10.1186/s13014-017-0896-7
        • Wang S
        • Zheng D
        • Zhang C
        • et al.
        Automatic planning on hippocampal avoidance whole-brain radiotherapy.
        Med Dos. 2017; 42: 63-68https://doi.org/10.1016/j.meddos.2016.12.002
        • Zieminski S.
        • Khandekar M.
        • Wang Y.
        Assessment of multi-criteria optimization (MCO) for volumetric modulated arc therapy (VMAT) in hippocampal avoidance whole brain radiation therapy (HA-WBRT).
        J Appl Clin Med Phys. 2018; 2: 184-190https://doi.org/10.1002/acm2.12277