- Email: [email protected]
Dosimetric evaluation of automatic and manual plans for early nasopharyngeal carcinoma to radiotherapy
ARTICLE IN PRESS
JID: MDO
[mUS5Gb;August 21, 2019;19:0]
Medical Dosimetry xxx (xxxx) xxx
Medical Dosimetry journal homepage: www.meddos.org
Dosimetric evaluation of automatic and manual plans for early nasopharyngeal carcinoma to radiotherapy Quanbin Zhang, Yingying Peng, Xianlu Song, Hui Yu, Linjing Wang, Shuxu Zhang∗ Affiliated Cancer Hospital & Institute of Guangzhou Medical University, Guangzhou 510095, China
a r t i c l e
i n f o
Article history: Received 6 April 2019 Accepted 30 May 2019 Available online xxx Keywords: Automatic plans Manual plans Dose Nasopharyngeal carcinoma Radiotherapy
a b s t r a c t To investigate dosimetric differences and plan qualities between manual plans and automatic plans for nasopharyngeal carcinoma (NPC) in early stage, and provide better options to maximize the benefits. Sixteen cases diagnosed with early NPC were retrospectively investigated. Conventional step and shoot IMRT with 7-fields and full arc volumetric-modulated arc therapy (VMAT) with double arcs were manually generated by experienced planners and automatically generated by Auto-Planning module in Pinnacle3 respectively, such as IMRT manual-planning (mIMRT), IMRT auto-planning (aIMRT), VMAT manual-planning (mVMAT), and VMAT auto-planning (aVMAT) for each patient. Target coverage, organs at risk sparing, monitor units, and planning times were compared and evaluated. All parameters of plans are able to fulfill International Commission on Radiation Units and Measurements repor (ICRU) 83 recommendations. Automatic plans are comparable or superior to manual plans without time-consuming planning process. The CI and HI for PTVs are better in aVMAT when compared with aIMRT and mVMAT, but those are similar between aIMRT and mVMAT. Automatic plans not only have superior dose homogeneity and conformity in PTVs, but also have better sparing for spinal cord or slightly reduce the doses received by other OARs, while the VMAT plans have better sparing for brain stem, especially the aVMAT plans. However, Dmax, V30, and V40 of brain stem are similar between aIMRT and mVMAT without significant difference. The monitor units and planning time for treatment plans have been significantly decreased through automatic planning technique. The automatic VMAT plan has greater clinical advantages and should be recommended to a better option for treating NPC in early stage, while automatic IMRT would be preferentially considered instead of manual VMAT. © 2019 Published by Elsevier Inc. on behalf of American Association of Medical Dosimetrists.
Introduction Nasopharyngeal carcinoma (NPC) is one of the most common malignancies in southern China and Southeast Asia. Due to its anatomic characteristics, biological properties, and radiosensitivity, radiotherapy has become the main treatment modality for nonmetastatic NPC.1,2 Radiotherapy for NPC is highly complex because the targets are surrounded by inhomogeneous tissue and organs at risk (OARs). In order to achieve high treatment planning quality in clinic, the labor-intensive and time-consuming for planners are necessary, resulting appropriate target doses and balanced sparing of the various OARs.3,4 Because all the constraints may need to be remodified and new help structures need to be generated to take into account the cold and hot dose spots, never-
All authors participated in patient treatment and were involved in the preparation of the manuscript. All authors reviewed and approved the final manuscript. ∗ Reprint requests to Shuxu Zhang, Affiliated Cancer Hospital & Institute of Guangzhou Medical University, Guangzhou 510095, China. E-mail address: [email protected] (S. Zhang).
theless, multiple iterations are normally required and even worse is that each iteration needs to repeat the above steps. However, the planner experience and time pressure have a larger impact on quality of these plans, which is not practical for busy clinical applications. If an automated algorithm is able to consistently and efficiently generate a plan with a quality that is similar or not inferior to that manually generated by experienced planners, the impacts might be minimized.5 Therefore, Auto-Planning has been proposed to improve plan quality and efficiency. Auto-Planning module, integrated in Pinnacle3 Treatment planning system (TPS), is a volume-driven automatic planning platform.6-8 Auto-Planning module automatically adds and adjusts individual optimization goals, constraints, and weights, leading to planning efficiency and a standardization of the plan quality at a high level or even surpassing the plans with manual optimization. As a result, total time to generate a treatment plan will be reduced, as well as the variation of planner intervention will be effectively reduced due to independent on the experience of planner. Several studies have showed that automatic plan could achieve similar or superior plan quality in PTVs coverage and a
https://doi.org/10.1016/j.meddos.2019.05.005 0958-3947/© 2019 Published by Elsevier Inc. on behalf of American Association of Medical Dosimetrists.
JID: MDO 2
ARTICLE IN PRESS
[mUS5Gb;August 21, 2019;19:0]
Q. Zhang, Y. Peng and X. Song et al. / Medical Dosimetry xxx (xxxx) xxx
significant dose reduction for OARs when compared with manual plan.9-11 It could be indicated that the technique parameters from Auto-Planning module is biased toward OARs sparing relative to manual plan. In contrast, the automatic plans with high clinical quality could be considered as a plan reference and starting point to ensure a certain minimum quality for manual plan.7 According to reproducibility and broad compliance with clinical requirements in Auto-Planning module, it allows less experienced planner to generate treatment plans in the acceptable times, although experienced planners can win over these results for specific plans. If needed, the automatic plans can be adjusted and further optimized by experienced planners because manual interaction is still possible, though not necessary. Despite the differences between intensity-modulated radiotherapy (IMRT) and volumetric-modulated arc therapy (VMAT) treatment plans in NPC at both planning and clinical level have been investigated,12,13 and positive results from automatic plans for various sites of cancers have been studied, such as head and neck cancer,14,15 esophageal cancer,16 lung cancer,17 pelvis as prostate, rectal, and cervical cancers,18 which are mainly focused on sites with simple anatomical structures and less OARs. However, the complicated NPC, where multiple PTV dose levels are defined close to numerous OARs, needs more concern to use automatic technique to improve the plan quality and OARs sparing. Some studies have investigated the automatic technique for NPC,19 but the comparison between automatic IMRT and automatic VMAT, especially the difference between automatic IMRT and manual VMAT, has not yet to be reported in nasopharyngeal carcinoma. Because the time-consuming in planning and the costs for patients have obvious differences between 2 techniques in radiotherapy; additionally, the performance requirement of linac and loss of linac during treatment are also different between IMRT and VMAT due to the continuously modulation of multi-leaf collimator positions, dose rate, and gantry speed simultaneously in the latter. Therefore, it is necessary to perform a comparison of the 2 planning techniques, which may provide better options for treating NPC, especially in early stage. As a result, it would improve planning efficiency with comparable plan quality according to current conditions in radiation therapy. This study is aimed to investigate the dosimetric differences and provide information about plan qualities between manual plans and automatic plans, leading to a more comprehensive evaluation of the treatment techniques for nasopharyngeal carcinoma in early stage. The Auto-Planning module integrated in Pinnacle3 is used to generate automatic plans, and then automatic plans, such as full arc VMAT with double arcs, are compared with manual plans, which are generated by experienced planners and clinically accepted by dosimetrist and oncologist, such as step and shoot IMRT (s-IMRT) with 7-fields. In clinical plans evaluation, the monitor units (MUs) and planning time for treatment plans were recorded and analyzed along with Dose Volume Histogram (DVH) parameters for all treatment plans, including target coverage and normal tissue sparing. Methods and Materials Patient characteristics Between January 2016 and December 2017, 16 early NPC patients with biopsyproven and nonmetastatic were randomly selected from 210 NPC patients. Histological examination confirmed all patients had World Health Organization type II or III disease. Approval for retrospective analysis of the patient data was obtained from the ethics committee of Affiliated Cancer Hospital and Institute of Guangzhou Medical University. All patients have signed written consent for the radiation treatment and were informed the technique for radiotherapy, the possible side effects, the risk, and so on. The pretreatment evaluations included complete physical examination, hematologic and biochemistry profiles, fibrotic endoscope examination of the nasopharynx, and MRI or contrast-enhanced computed tomography of the nasopharyngeal and cervical region to evaluate the primary tumor extent and regional
Table 1 The characteristics of patients with nasopharyngeal carcinoma (n = 16)
Sex Male Female T 1 2 3 4 N 0 1 2 3 M 0 1 Stages I II III IV
Number
%
10 6
62.5 37.5
5 11 0 0
31.3 68.7 0 0
3 8 5 0
18.7 50.0 32.3 0
16 0
100 0
9 7 0 0
56.3 43.7 0 0
lymph node involvement. Chest radiography, bone scintigraphy, and abdominal region ultrasonography were used to exclude distant metastases. All patients were staged according to the American Joint Committee on Cancer (Manual for Staging of Cancer, 7th edition) system. The specific characteristics of the patients are showed in Table 1. The median age is 48 years ranging from 30 to 71 years. Simulation and immobilization All patients in the supine position were immobilized by using a head neck and shoulder thermoplastic mask. CT was performed after administration of intravenous contrast medium; 3 mm slices were obtained from the head to 2 cm below the sternoclavicular joint. These CT images were transferred to the Pinnacle3 TPS (version 9.10, Philips Medical Systems, Fitchburg, WI). Target volume delineation and dose prescription The target volumes were defined in accordance with the International Commission on Radiation Units and Measurements reports (ICRU) 50 and 62. The gross tumor volume (GTV) was determined from the MRI, clinical information, and endoscopic findings. The clinical target volumes (CTV) were individually delineated on the basis of tumor invasion. The first CT V(CT V1) was defined as the GTVnx with a 5 to 10 mm margin, including high-risk regions of microscopic extension encompassing the whole nasopharynx. The second CTV was defined as CTV1 with a margin of 5 to 10 mm, including low-risk regions of microscopic extension. If it was in close proximity to critical structures, this margin could be reduced. The OARs included the brainstem, spinal cord, temporal lobe, optic nerves, optic chiasm, lens, eyes, parotid glands, mandible, temporo-mandibular joint, middle ear, and larynx. The prescribed dose was 70 Gy to the PTV of GTVnx, 60 Gy to the PTV of CTV1, 56 Gy to the PTV of second CTV, and 60 to 68 Gy to the PTV of GTV for involved cervical lymph nodes, and it was set to 32 fractions. Treatment was delivered once daily and no more than 5 fractions per week. Treatment planning The PTVs of each target were created by adding a 3 mm 3D margin to the delineated target volume to compensate for treatment set-up variability and internal organ motion. In order to protect the spinal cord, lens, and brain stem, the spinal cord and lens were added a 5 mm 3D margin respectively while the brain stem was added a 4 mm 3D margin. The plan goal for each plan was set to 95% of the PTVs volume to reach 100% of the prescription dose; furthermore, no more than 20% of the PTVs were to receive 110% of the prescription dose and the PTVs receiving less than 93% of the prescription dose did not exceed 1%. In addition, the dose received by each OAR was limited to the dose constraints according to the RTOG0615 protocol, as shown in Table 2. Dose optimization and calculation for s-IMRT and VMAT plans were performed by Pinnacle3 TPS, and the Auto-Planning module in Pinnacle3 TPS was used to create automatic plans for the studied cases. Both IMRT and VMAT were created for an Elekta Synergy linear accelerator with 6 MV photons. Additionally, NPC patients were equally randomly assigned to 2 physicists who have 5 years of experience in planning s-IMRT and VMAT to generate plans, such as: (1) IMRT manual-planning (mIMRT), (2) IMRT auto-planning (aIMRT), (3) VMAT manual-planning (mVMAT) and (4) VMAT auto-planning (aVMAT).
ARTICLE IN PRESS
JID: MDO
Q. Zhang, Y. Peng and X. Song et al. / Medical Dosimetry xxx (xxxx) xxx Table 2 OARs optimization goals in treatment planning OARs
Index
Optimization goal
Priority£
Compromise§
Brain stem Brain stem Brain stem + 2 mm PRV Brain stem + 2 mm PRV Spinal cord Spinal cord + 5 mm PRV Spinal cord + 5 mm PRV Lens+5 mm PRV Optic chiasm Optic nerves Temporal lobes Middle ears Eyes Parotids Parotids TM joint Mandible Trachea Larynx
Max Dose Max DVH Max Dose
54 Gy V50¶ < 4% 60 Gy
High High Medium
No No No
Max DVH
V60# < 1%
High
No
Max Dose Max Dose
40 Gy 45 Gy
High Medium
No No
Max DVH
V45& < 1%
High
No
Max Dose Max Dose Max Dose Max Dose Max Dose Max Dose Mean Dose Max DVH Max Dose Max Dose Max Dose Mean Dose
8 Gy 54 Gy 50 Gy 60 Gy 60 Gy 25 Gy 28 Gy V30∗ ≤ 50% 65 Gy 65 Gy 40 Gy 40 Gy
High High High Low Medium Medium Medium Medium Medium Medium Low Low
No Yes Yes Yes Yes No Yes Yes Yes Yes Yes Yes
Note: The 2 mm and 5 mm represent the 3D margins of OARs respectively, and PRV is planning risk volume. In addition, ∗ , &, ¶, and # is defined as the percentage volume received >30, 45, 50 and 60 Gy respectively. Additionally, £ and § represent the parameters for automatic planning in Auto-Planning module.
[mUS5Gb;August 21, 2019;19:0] 3
Based on many years of clinical experience and combined with the characteristics of the target area, T category of NPC and the principle of sparing OARs, a standard coplanar 7-fields gantry arrangement for IMRT was designed in all cases and delivered in the step and shoot mode, and gantry angles were 0°, 50°, 100°, 150°, 210°, 260°, and 310° respectively. Furthermore, total number of subfields was ≤ 100 and the minimum segment area was set to 10 cm2 and the minimum segment monitor unit was set to 10 MU. On the other hand, VMAT plans used double coplanar full arcs with a 0° collimator angle. One arc was set-up in a clockwise direction from 182° to 178°; conversely, the second arc was performed in a counterclockwise direction from 178° to 182°, and gantry spacing was 4° in VMAT plans. It was noted that the final results of the Auto-Planning module were obtained without more than 3 postoptimizations or slight manual intervention. Meanwhile, the number of iterations for dose calculation in planning process does not exceed 100. All treatment plans were approved by a clinical physicist and a radiation oncologist for all cases.
Plans evaluation According to ICRU 83, quantitative evaluation for the PTVs and OARs was performed using a standard DVH. The conformity index (CI),20 used as a measure of target volume dose distribution conformity, was defined as CI = VT,ref /VT × VT,ref /Vref , where VT,ref was target volume covered by reference isodose, VT was target volume, and Vref was volume of the reference isodose. If the CI is closer to 1, the dose conformity is better. The homogeneity index (HI),21 used as a measure of the evenness of dose distribution, was defined as HI = (D2 to D98) / D50, where D2, D98, and D50 were the doses covering 2%, 98%, and 50% of the PTV, respectively. HI = 0 is the ideal value. Analysis of the OARs included the maximum dose, mean dose and a set of appropriate define (Vx) and define (Dy) values. In addition, the planning times and MUs for treatment plans were recorded and evaluated respectively. The planning time was defined as the effective working time, which started when the PTVs and OARs were defined by the clinicians and finished when the plans were optimized and approved by dosimetrist and oncologist.
Fig. 1. The dose distributions for a representative NPC patient (stage T2N0M0) in 4 types of plans. T stands for transverse, S stands for sagittal, and C stands for coronal; while 1, 2, 3, and 4 represent mIMRT, aIMRT, mVMAT, and aVMAT respectively. (Color version of figure is available online.)
JID: MDO 4
ARTICLE IN PRESS
[mUS5Gb;August 21, 2019;19:0]
Q. Zhang, Y. Peng and X. Song et al. / Medical Dosimetry xxx (xxxx) xxx
Fig. 2. The dosimetric parameters of PTV70 and PTV56 in 4 types of plans, A for PTV70 and B for PTV56 .
Statistical analysis
Monitor unit and planning time
Statistical analysis was performed using SPSS 19.0 (SPSS Inc., Chicago, IL). All datas were expressed as mean ± standard deviation. The data for the different plans were evaluated using Wilcoxon’s signed rank test. The difference was considered significant when P< 0.05.
As shown in Fig. 5, the MUs and planning time for all plans were different. Compared with manual plans, automatic plans had made an average MUs reduction of 7.95% (p <0.05) and an average planning time reduction of 22.27% (p <0.05). In terms of MUs, the average MUs from VMAT were significantly less than those from s-IMRT, while the MUs from aVMAT were similar or evenly less to those from mVMAT; however, compared to mIMRT, the MUs from aIMRT were significantly reduced to some extent. In addition, the average planning times for VMAT were significantly more than those for s-IMRT, and it was an average increase of about 97.11% (p <0.05). Furthermore, the planning times for aVMAT were significantly less than those for mVMAT. The same trend was observed between mIMRT and aIMRT (all p < 0.05).
Results The dose distribution of the PTVs The mean volumes for PTV70 and PTV56 were 52.47 ± 26.60 cm3 and 690.57 ± 130.76 cm3 respectively. All of the parameters obtained from 4 types of plans for early NPC were complied with the ICRU 83 recommendation for target coverage. The dose distributions for a representative NPC patient (stage T2N0M0) were shown in Fig. 1, including 4 types of plans. The dose distributions of the PTVs from VMAT were similar or a certain extent better to s-IMRT. The parameters from DVH of the PTVs in 4 types of plans were shown in Fig. 2. The automatic plans not only had slightly or obviously less maximum dose, especially in aVMAT plans, but also produced more conformal and homogeneous dose distributions in PTVs, as shown in Fig. 3. The CI and HI of automatic plans for PT V70 and PT V56 were 0.802 ± 0.049, 0.037 ± 0.010, and 0.825 ± 0.050, 0.261 ± 0.014 respectively, which were greatly improved when compared with those of manual plans, such as 0.774 ± 0.039, 0.046 ± 0.007 and 0.808 ± 0.047, 0.275 ± 0.011 respectively. It reflected that a more flat dose distribution in target area. In addition, the CI and HI for PTV70 in aVMAT were significantly better than those in aIMRT (p = 0.016 and 0.002 respectively); when regarding PTV56 in aIMRT, the CI was similar and the HI was inferior to those from aVMAT. However, CI and HI both in PTV70 and PTV56 were similar between aIMRT and mVMAT without significant difference (p >0.05). It revealed that aVMAT showed superior dose homogeneity and conformity in PTVs. In 4 types of plans, the qualities of plans for PTVs from mIMRT were the last of relative ranking, while the first of relative ranking was for aVMAT, followed by mVMAT equal to aIMRT.
The dose distribution of OARs Table 3 and Table 4 described the parameters of OARs from the patients. According to dose objectives for OARs, all plans were similar between patients. But all plans did not meet the mean dose objective for parotids, only partial V30 of parotids met the planning objective. Compared with manual plans, the automatic plans had decreased dose to some OARs, including spinal cord, brain stem, eyes, optic nerves, and optic chiasm. In addition, the low dose burden values achieved from VMAT plans were significantly lower than those from s-IMRT plans. Automatic plans had better sparing for spinal cord, while the VMAT had better sparing for brain stem. When compared with all plans, the mVMAT plans resulted in significantly poorer sparing for spinal cord (all p <0.05). Regarding dose received by brain stem, Dmax, V30, and V40 were significantly lower in aVMAT than aIMRT and mVMAT, but these parameters were similar between aIMRT and mVMAT without significant difference, as seen in Fig. 4. Additionally, it could note that the irradiation dose coming from mIMRT for brain stem was relatively highest in 4 types of plans. When considering other normal tissue sparring, it showed no significant difference (p >0.05).
Discussion Dose comparison of PTVs between automatic and manual plans In general, planning and optimization of treatment plans for NPC are complex process, where hard constraints have to be fulfilled. The Auto-Planning module for generating automatic plans would create a lot of special ROIs to ensure a high quality of dose distribution for PTVs and OARs sparing.22 However, if such special ROIs are manually created in manual plans, it is very difficult or even impossible to achieve, especially depending on the efforts and experiences of manual planners. In this sense, quantitatively better treatment plans could be generated by using the Auto-Planning module. For example, several researches have shown that it is feasible and similar or superior to previous manual plans because of maintaining the target coverage and better sparing OARs.9,10,23 As expected, compared to manual plans, automatic plans have shown significantly superior conformity and homogeneity. The CI and HI of aVMAT are more superior to those of aIMRT and mVMAT, especially in cold and hot dose spots and OARs sparing, while the CI and HI in PTVs are similar between aIMRT and mVMAT. Many researchers have found that VMAT provides similar or better results in dose coverage of the PTVs in NPC when compared with IMRT, especially in early T category.2 From our study cases, automatic IMRT is comparable to manual VMAT in dose homogeneity and conformity for PTVs. It means that automatic IMRT may provide an option to replace the manual VMAT, though manual VMAT may offer dosimetric advantages in head and neck cancer. In this way, it does not need to blindly adjust the dose coverage of the PTVs by manual VMAT technique, which has higher performance requirement of linac. As a result, it not only can reduce the loss of linac, but also can decrease the costs of patients due to the cost
ARTICLE IN PRESS
JID: MDO
[mUS5Gb;August 21, 2019;19:0]
Q. Zhang, Y. Peng and X. Song et al. / Medical Dosimetry xxx (xxxx) xxx
5
Fig. 3. Statistical comparison of CI and HI are for PT V70 and PT V56 respectively in 4 types of plans, A for PTV70 and B for PTV56 .
Table 3 The objective and pass rate for various OARs in 4 types of plans OARs
Brain stem Brain stem Brain stem + 2 mm PRV Spinal cord Spinal cord + 5 mm PRV Lens+5 mm PRV Optic chiasm Optic nerves Temporal lobes Middle ears Eyes Parotids Parotids TM joint Mandible Larynx
Objective
Max V50¶ D1# Max D1# Max Max Max D2& D2& D2& Mean V30∗ Max D1# Mean
Pass rate (%)
≤ < < ≤ ≤ ≤ ≤ ≤ ≤ ≤ ≤ ≤ ≤ < < <
54 Gy 4% 60 Gy 45 Gy 50 Gy 12 Gy 54 Gy 50 Gy 60 Gy 60 Gy 30 Gy 26 Gy 50% 70 Gy 75 Gy 45 Gy
mIMRT
aIMRT
mVMAT
aVMAT
56.25 10 0.0 0 10 0.0 0 10 0.0 0 10 0.0 0 10 0.0 0 93.75 81.25 56.25 10 0.0 0 10 0.0 0 0.00 62.50 10 0.0 0 10 0.0 0 75.00
62.50 10 0.0 0 10 0.0 0 10 0.0 0 10 0.0 0 10 0.0 0 93.75 87.50 56.25 10 0.0 0 10 0.0 0 0.00 68.75 10 0.0 0 10 0.0 0 81.25
75.00 10 0.0 0 10 0.0 0 10 0.0 0 10 0.0 0 10 0.0 0 10 0.0 0 75.00 68.75 10 0.0 0 10 0.0 0 0.00 62.50 10 0.0 0 10 0.0 0 81.25
81.25 10 0.0 0 10 0.0 0 10 0.0 0 10 0.0 0 10 0.0 0 10 0.0 0 81.25 62.50 10 0.0 0 10 0.0 0 0.00 62.50 10 0.0 0 10 0.0 0 87.50
Manual vs automatic
0.429 N/A N/A N/A N/A N/A 0.192 0.072 0.719 N/A N/A N/A 0.599 N/A N/A 0.488
p aIMRT vs mVMAT 0.509 N/A N/A N/A N/A N/A 0.843 0.429 0.227 N/A N/A N/A 0.669 N/A N/A 0.537
aIMRT vs aVMAT 0.154 N/A N/A N/A N/A N/A 0.345 0.950 0.339 N/A N/A N/A 0.645 N/A N/A 0.916
Note: The 2 mm and 5 mm represent the 3D margins of OARs respectively, and PRV is planning risk volume. In addition, ¶ is defined as the percentage volume received >50 Gy; # is defined as the dose received by 1% of the volume; & is defined as the dose received by 2% of the volume; ∗ is defined as the percentage volume received >30 Gy.
ARTICLE IN PRESS
JID: MDO 6
[mUS5Gb;August 21, 2019;19:0]
Q. Zhang, Y. Peng and X. Song et al. / Medical Dosimetry xxx (xxxx) xxx
Table 4 The comparison of doses received by various OARs in 4 types of plans OARs
Brain stem Brain stem Brain stem + 2 mm PRV Spinal cord Spinal cord + 5 mm PRV Lens + 5 mm PRV Optic chiasm Optic nerves Temporal lobes Middle ears Eyes Parotids Parotids TM joint Mandible Larynx
Index
Max (Gy) V50¶ (%) D1# (Gy) Max (Gy) D1# (Gy) Max (Gy) Max (Gy) Max (Gy) D2& (Gy) D2& (Gy) D2& (Gy) Mean (Gy) V30∗ (%) Max (Gy) D1# (Gy) Mean (Gy)
Mean dose ± SD mIMRT
aIMRT
mVMAT
aVMAT
54.29 ± 2.24 1.83 ± 1.15 57.69 ± 1.18 37.65 ± 2.04 41.79 ± 1.89 6.91 ± 0.88 43.58 ± 7.71 47.79 ± 3.62 59.55 ± 4.33 51.12± 2.01 14.93 ± 3.13 34.41 ± 2.60 47.60 ± 5.31 62.54 ± 4.59 62.70 ± 2.83 43.39 ± 2.45
53.64 ± 2.26 0.99 ± 0.95 57.74 ± 1.28 35.13 ± 1.07 40.96 ± 1.93 6.26 ± 0.87 40.65 ± 8.14 45.69 ± 4.41 59.15 ± 3.53 51.71 ± 2.61 12.98 ± 2.88 34.75 ± 2.36 47.01 ± 6.04 61.14 ± 5.93 61.78 ± 1.91 42.50 ±2.34
52.76 ± 2.23 0.46 ± 0.63 57.23 ± 1.54 38.25 ± 0.88 42.32 ± 2.14 5.81 ± 0.71 39.13 ± 7.99 47.20 ± 4.30 57.45 ± 3.78 51.71 ± 2.98 11.88 ± 3.92 34.43 ± 1.49 47.77 ± 7.90 61.82 ± 4.86 61.43 ±4.69 42.99 ± 2.08
51.96 ± 1.96 0.41 ± 0.42 57.12 ± 2.28 35.20 ± 1.61 40.93 ± 1.06 5.47 ± 1.38 37.25 ± 6.72 45.92 ± 4.40 58.06 ± 4.10 50.31 ± 4.82 11.24 ± 3.79 34.29 ± 0.92 49.11 ±6.82 61.38 ± 3.71 60.77 ± 4.97 42.43 ± 2.20
Manual vs automatic
0.239 0.272 0.946 0.0 0 0 0.028 0.164 0.222 0.107 0.919 0.831 0.514 0.955 0.817 0.649 0.305 0.200
p aIMRT vs mVMAT 0.246 0.284 0.316 0.0 0 0 0.095 0.429 0.598 0.336 0.200 0.990 0.399 0.854 0.765 0.870 0.760 0.533
aIMRT vs aVMAT 0.028 0.203 0.346 0.677 0.961 0.198 0.207 0.884 0.429 0.991 0.235 0.653 0.367 0.953 0.461 0.935
Note: The 2 mm and 5 mm represent the 3D margins of OARs respectively, and PRV is planning risk volume. In addition, ¶ is defined as the percentage volume received >50 Gy; # is defined as the dose received by 1% of the volume; & is defined as the dose received by 2% of the volume; ∗ is defined as the percentage volume received >30 Gy.
of VMAT is more expensive than IMRT. More importantly, it would reduce the dependence on planner experience to minimize the impacts in treatment plans, and improve the planning efficiency. On the contrary, if conditions permit, automatic VMAT is a better option for recommendation. Dose comparison of OARs between automatic and manual plans
Fig. 4. Comparison of dosimetric quality for brain stem in 4 types of plans.
Due to it does not know how much lower dose is better sparing an OAR for a particular patient, a dose guideline only provides a guidance in planning process and cannot ensure an optimal dose distribution in a treatment plan. During the planning process, individual optimization performed by the experienced planner has a very large solution space, which leaves with much uncertainty to ensure an optimal treatment plan, especially sparing OARs. If the optimization is performed by automatic algorithm, it would decrease uncertainty to increase the probability for the achievable optimal plan through the reduction of variation in planner intervention. In the study cases, the average doses for OARs from automatic plans are similar or less to those from manual plans.
Fig. 5. Comparison of monitor units and planning times in 4 types of plans.
JID: MDO
ARTICLE IN PRESS
[mUS5Gb;August 21, 2019;19:0]
Q. Zhang, Y. Peng and X. Song et al. / Medical Dosimetry xxx (xxxx) xxx
Although the target area of NPC is close to the spinal cord, especially brain stem, the results have demonstrated that automatic plan could protect the spinal cord better than manual plan. Additionally, aIMRT is comparable to aVMAT in spinal cord sparing. Compared to s-IMRT, VMAT plans have advantages for brain stem sparing, as shown in Fig. 4. It illustrates that aVMAT plans are confined to a narrower region in doses distribution of brain stem when compared with mVMAT plans, but the dose received by brain stem does not have significant difference between aIMRT and mVMAT. It reveals that aIMRT is a certain extent better than mVMAT in OARs sparing. The doses of other OARs in automatic plan, used to evaluate the quality of treatment plan, are slight lower or even similar to those in manual plan. It can be suggested that planning balance between various optimization criteria is clinically favored during the automatic planning process. Due to parotids are partial overlap with the PTVs, it is more difficult to judge parotids doses whether meeting dose constraints.24 If more consideration is need for better sparing of OARs, the automatic planning technique is more needed to be carried out. MUs and planning times of automatic and manual plans In the study cases, aVMAT plans have fewer MUs, followed by mVMAT plans, where the MUs are much less than those in aIMRT plans. The fewer MUs can lead to significantly improve the efficiency of plan implementation to reduce the treatment time, which could benefit the patients to reduce discomfort and intrafraction variation during the treatment. In particular, it is worth to note that secondary cancer risk may be potentially reduced during radiotherapy due to the benefits from fewer MUs.25 Using standard settings in Auto-Planning module, the automatic planning times have decreased about 26% and 20% respectively when compared with manual planning. Because the Auto-Planning module could potentially make it possible for planners to spend less time on the more tedious steps in planning process during optimization. In addition, this is not just a matter of shortening planning time to improve planning efficiency by using Auto-Planning module. First, the Auto-Planning module has an advantage to leave planner free for other tasks of plans due to reducing the need for multiple plan reviews and runs in the background. Second, this makes it possible for planners to spend more time to focus on planning goals with difficult dose. On the other hand, the planning times spent on aIMRT plans are significantly shorter when compared with aVMAT plans. Because 180 control points need to be considered to calculate the dose of VMAT plans with double arcs. It is equivalent to calculate 180 irradiation fields, which are significantly more than those of IMRT plans. Conclusions Auto-Planning module could represent an effective way to improve a planning process with consistent plan qualities for NPC in early stage. Standardization of the plan quality is consistently achieved from automatic plans, which are similar or better to those from manual plans. But the effective times required for planning and optimization and MUs are significantly decreased in automatic process. Furthermore, when compared automatic plans with manual plans, the dose to most of OARs could be similar or decreased to some extent, especially in brain stem and spinal cord. It means that the automatic plans provide comparable or superior plans when compared to manual plans. After a comprehensive analysis, the plan qualities of mIMRT are the last of relative ranking, while the first of relative ranking is for aVMAT, followed by mVMAT that equal to aIMRT. The automatic VMAT plan has greater clinical advantages and should be recommended to a better option for treating NPC in early stage, but automatic IMRT would be
7
preferentially considered instead of manual VMAT. However, owing to small sample size and retrospective study, a large sample of prospective studies is needed to further confirm the above results before extensive implementation for NPC. Conflict of Interest The authors declare no conflict of interest. Ethics Approval and Consent to Participate All patients included into this study have given their approval to use their data for scientific research. All personal information to identify patients was removed from the image data and analyzed retrospectively. Acknowledgments This work was supported financially by Guangzhou Medical Key Discipline Construction Project (2017 to 2019): Cancer Therapeutics and Experimental Oncology Project, Technology Project of Guangzhou Medical and Health Science (Grant number: 20171A010314; 20181A011095). References 1. Kam, M.M.; Chau, R.M.; Suen, J.; et al. Intensity-modulated radiotherapy in nasopharyngeal carcinoma: Dosimetric advantage over conventional plans and feasibility of dose escalation. Int J Radiat Oncol Biol Phys 56:145–57; 2003. 2. Sun, Y.; Guo, R.; Yin, W.J.; et al. Which t category of nasopharyngeal carcinoma may benefit most from volumetric modulated arc therapy compared with step and shoot intensity modulated radiation therapy. PLoS ONE 8:e75304; 2013. 3. Nelms, B.E.; Robinson, G.; Markham, J.; et al. Variation in external beam treatment plan quality: an inter-institutional study of planners and planning systems. Pract Radiat Oncol 2:296–305; 2012. 4. Batumalai, V.; Jameson, M.G.; Forstner, D.F.; et al. How important is dosimetrist experience for intensity modulated radiation therapy? A comparative analysis of a head and neck case. Pract Radiat Oncol 3:e99–e106; 2013. 5. Hansen, C.R.; Nielsen, M.; Bertelsen, A.S.; et al. Automatic treatment planning facilitates fast generation of high-quality treatment plans for esophageal cancer. Acta Oncol 56:1–6; 2017. 6. Wang, S.; Zheng, D.; Zhang, C.; et al. Automatic planning on hippocampal avoidance whole-brain radiotherapy. Med Dosim 42:63–8; 2017. 7. Hazell, I.; Bzdusek, K.; Kumar, P.; et al. Automatic planning of head and neck treatment plans. J Appl Clin Med Phys 17:272–82; 2016. 8. Chen, H.; Wang, H.; Gu, H.; et al. Study for reducing lung dose of upper thoracic esophageal cancer radiotherapy by auto-planning: Volumetric-modulated arc therapy vs intensity-modulated radiation therapy. Med Dosim 43:243–50; 2018. 9. Krayenbuehl, J.; Norton, I.; Studer, G.; et al. Evaluation of an automated knowledge based treatment planning system for head and neck. Radiat Oncol 10:226; 2015. 10. Hansen, C.R.; Bertelsen, A.; Hazell, I.; et al. Automatic treatment planning improves the clinical quality of head and neck cancer treatment plans. Clin Translat Radiat Oncol 119:S396–7; 2016. 11. Li, K.; Chang, X.; Wang, J.; et al. SU-F-T-358: is auto-planning useful for volumetric-modulated arc therapy planning in rectal cancer radiotherapy? Med Phys 43:3545; 2016. 12. Lee, T.F.; Chao, P.J.; Ting, H.M.; et al. Comparative analysis of smartarc-based dual arc volumetric-modulated arc radiotherapy (VMAT) versus intensity-modulated radiotherapy (IMRT) for nasopharyngeal carcinoma. J Appl Clin Med Phys 12:158–74; 2011. 13. Ning, Z.H.; Mu, J.M.; Jin, J.X.; et al. Single arc volumetric-modulated arc therapy is sufficient for nasopharyngeal carcinoma: A dosimetric comparison with dual arc VMAT and dynamic MLC and step-and-shoot intensity-modulated radiotherapy. Radiat Oncol 8:237; 2013. 14. Tol, J.P.; Delaney, A.R.; Dahele, M.; et al. Evaluation of a knowledge-based planning solution for head and neck cancer. Int J Radiat Oncol Biol Phys 91:612–20; 2015. 15. Doornaert, P.; Verbakel, W.F.; Bieker, M.; et al. RapidArc planning and delivery in patients with locally advanced head-and-neck cancer undergoing chemoradiotherapy. Int J Radiat Oncol Biol Phys 79:429–35; 2011. 16. Hansen, C.R.; Nielsen, M.; Bertelsen, A.S.; et al. Automatic treatment planning facilitates fast generation of high-quality treatment plans for esophageal cancer. Acta Oncol 56:1495–500; 2017. 17. Fogliata, A.; Belosi, F.; Clivio, A.; et al. On the pre-clinical validation of a commercial model-based optimisation engine: application to volumetric modulated arc therapy for patients with lung or prostate cancer. Radiother Oncol 113(3):385–91; 2014.
JID: MDO 8
ARTICLE IN PRESS
[mUS5Gb;August 21, 2019;19:0]
Q. Zhang, Y. Peng and X. Song et al. / Medical Dosimetry xxx (xxxx) xxx
18. Hussein, M.; South, C.P.; Barry, M.A.; et al. Clinical validation and benchmarking of knowledge-based IMRT and VMAT treatment planning in pelvic anatomy. Radiother Oncol 120:473–9; 2016. 19. Wang, J.; Chen, Z.; Li, W.; et al. A new strategy for volumetric-modulated arc therapy planning using autoplanning based multicriteria optimization for nasopharyngeal carcinoma. Radiat Oncol 13:94; 2018. 20. Paddick, I. A simple scoring ratio to index the conformity of radiosurgical treatment plans. Technical note. J Neurosurg 93(Suppl 3):219–22; 20 0 0. 21. Kataria, T.; Sharma, K.; Subramani, V.; et al. Homogeneity Index: an objective tool for assessment of conformal radiation treatments. J Med Phys 37:207–13; 2012.
22. Nawa, K.; Haga, A.; Nomoto, A.; et al. Evaluation of a commercial automatic treatment planning system for prostate cancers. Med Dosim 42:203–9; 2017. 23. Wu, B.; Kusters, M.; Kunzebusch, M.; et al. Cross-institutional knowledge-based planning (KBP) implementation and its performance comparison to auto-planning engine (APE). Radiother Oncol 123:57–62; 2017. 24. Johnston, M.; Clifford, S.; Bromley, R.; et al. Volumetric-modulated arc therapy in head and neck radiotherapy: a planning comparison using simultaneous integrated boost for nasopharynx and oropharynx carcinoma. Clin Oncol 23:503–11; 2011. 25. Hall, E.J. Intensity-modulated radiation therapy, protons, and the risk of second cancers. Int J Radiat Oncol Biol Phys 65:1–7; 2006.
JID: MDO
[mUS5Gb;August 21, 2019;19:0]
Medical Dosimetry xxx (xxxx) xxx
Medical Dosimetry journal homepage: www.meddos.org
Dosimetric evaluation of automatic and manual plans for early nasopharyngeal carcinoma to radiotherapy Quanbin Zhang, Yingying Peng, Xianlu Song, Hui Yu, Linjing Wang, Shuxu Zhang∗ Affiliated Cancer Hospital & Institute of Guangzhou Medical University, Guangzhou 510095, China
a r t i c l e
i n f o
Article history: Received 6 April 2019 Accepted 30 May 2019 Available online xxx Keywords: Automatic plans Manual plans Dose Nasopharyngeal carcinoma Radiotherapy
a b s t r a c t To investigate dosimetric differences and plan qualities between manual plans and automatic plans for nasopharyngeal carcinoma (NPC) in early stage, and provide better options to maximize the benefits. Sixteen cases diagnosed with early NPC were retrospectively investigated. Conventional step and shoot IMRT with 7-fields and full arc volumetric-modulated arc therapy (VMAT) with double arcs were manually generated by experienced planners and automatically generated by Auto-Planning module in Pinnacle3 respectively, such as IMRT manual-planning (mIMRT), IMRT auto-planning (aIMRT), VMAT manual-planning (mVMAT), and VMAT auto-planning (aVMAT) for each patient. Target coverage, organs at risk sparing, monitor units, and planning times were compared and evaluated. All parameters of plans are able to fulfill International Commission on Radiation Units and Measurements repor (ICRU) 83 recommendations. Automatic plans are comparable or superior to manual plans without time-consuming planning process. The CI and HI for PTVs are better in aVMAT when compared with aIMRT and mVMAT, but those are similar between aIMRT and mVMAT. Automatic plans not only have superior dose homogeneity and conformity in PTVs, but also have better sparing for spinal cord or slightly reduce the doses received by other OARs, while the VMAT plans have better sparing for brain stem, especially the aVMAT plans. However, Dmax, V30, and V40 of brain stem are similar between aIMRT and mVMAT without significant difference. The monitor units and planning time for treatment plans have been significantly decreased through automatic planning technique. The automatic VMAT plan has greater clinical advantages and should be recommended to a better option for treating NPC in early stage, while automatic IMRT would be preferentially considered instead of manual VMAT. © 2019 Published by Elsevier Inc. on behalf of American Association of Medical Dosimetrists.
Introduction Nasopharyngeal carcinoma (NPC) is one of the most common malignancies in southern China and Southeast Asia. Due to its anatomic characteristics, biological properties, and radiosensitivity, radiotherapy has become the main treatment modality for nonmetastatic NPC.1,2 Radiotherapy for NPC is highly complex because the targets are surrounded by inhomogeneous tissue and organs at risk (OARs). In order to achieve high treatment planning quality in clinic, the labor-intensive and time-consuming for planners are necessary, resulting appropriate target doses and balanced sparing of the various OARs.3,4 Because all the constraints may need to be remodified and new help structures need to be generated to take into account the cold and hot dose spots, never-
All authors participated in patient treatment and were involved in the preparation of the manuscript. All authors reviewed and approved the final manuscript. ∗ Reprint requests to Shuxu Zhang, Affiliated Cancer Hospital & Institute of Guangzhou Medical University, Guangzhou 510095, China. E-mail address: [email protected] (S. Zhang).
theless, multiple iterations are normally required and even worse is that each iteration needs to repeat the above steps. However, the planner experience and time pressure have a larger impact on quality of these plans, which is not practical for busy clinical applications. If an automated algorithm is able to consistently and efficiently generate a plan with a quality that is similar or not inferior to that manually generated by experienced planners, the impacts might be minimized.5 Therefore, Auto-Planning has been proposed to improve plan quality and efficiency. Auto-Planning module, integrated in Pinnacle3 Treatment planning system (TPS), is a volume-driven automatic planning platform.6-8 Auto-Planning module automatically adds and adjusts individual optimization goals, constraints, and weights, leading to planning efficiency and a standardization of the plan quality at a high level or even surpassing the plans with manual optimization. As a result, total time to generate a treatment plan will be reduced, as well as the variation of planner intervention will be effectively reduced due to independent on the experience of planner. Several studies have showed that automatic plan could achieve similar or superior plan quality in PTVs coverage and a
https://doi.org/10.1016/j.meddos.2019.05.005 0958-3947/© 2019 Published by Elsevier Inc. on behalf of American Association of Medical Dosimetrists.
JID: MDO 2
ARTICLE IN PRESS
[mUS5Gb;August 21, 2019;19:0]
Q. Zhang, Y. Peng and X. Song et al. / Medical Dosimetry xxx (xxxx) xxx
significant dose reduction for OARs when compared with manual plan.9-11 It could be indicated that the technique parameters from Auto-Planning module is biased toward OARs sparing relative to manual plan. In contrast, the automatic plans with high clinical quality could be considered as a plan reference and starting point to ensure a certain minimum quality for manual plan.7 According to reproducibility and broad compliance with clinical requirements in Auto-Planning module, it allows less experienced planner to generate treatment plans in the acceptable times, although experienced planners can win over these results for specific plans. If needed, the automatic plans can be adjusted and further optimized by experienced planners because manual interaction is still possible, though not necessary. Despite the differences between intensity-modulated radiotherapy (IMRT) and volumetric-modulated arc therapy (VMAT) treatment plans in NPC at both planning and clinical level have been investigated,12,13 and positive results from automatic plans for various sites of cancers have been studied, such as head and neck cancer,14,15 esophageal cancer,16 lung cancer,17 pelvis as prostate, rectal, and cervical cancers,18 which are mainly focused on sites with simple anatomical structures and less OARs. However, the complicated NPC, where multiple PTV dose levels are defined close to numerous OARs, needs more concern to use automatic technique to improve the plan quality and OARs sparing. Some studies have investigated the automatic technique for NPC,19 but the comparison between automatic IMRT and automatic VMAT, especially the difference between automatic IMRT and manual VMAT, has not yet to be reported in nasopharyngeal carcinoma. Because the time-consuming in planning and the costs for patients have obvious differences between 2 techniques in radiotherapy; additionally, the performance requirement of linac and loss of linac during treatment are also different between IMRT and VMAT due to the continuously modulation of multi-leaf collimator positions, dose rate, and gantry speed simultaneously in the latter. Therefore, it is necessary to perform a comparison of the 2 planning techniques, which may provide better options for treating NPC, especially in early stage. As a result, it would improve planning efficiency with comparable plan quality according to current conditions in radiation therapy. This study is aimed to investigate the dosimetric differences and provide information about plan qualities between manual plans and automatic plans, leading to a more comprehensive evaluation of the treatment techniques for nasopharyngeal carcinoma in early stage. The Auto-Planning module integrated in Pinnacle3 is used to generate automatic plans, and then automatic plans, such as full arc VMAT with double arcs, are compared with manual plans, which are generated by experienced planners and clinically accepted by dosimetrist and oncologist, such as step and shoot IMRT (s-IMRT) with 7-fields. In clinical plans evaluation, the monitor units (MUs) and planning time for treatment plans were recorded and analyzed along with Dose Volume Histogram (DVH) parameters for all treatment plans, including target coverage and normal tissue sparing. Methods and Materials Patient characteristics Between January 2016 and December 2017, 16 early NPC patients with biopsyproven and nonmetastatic were randomly selected from 210 NPC patients. Histological examination confirmed all patients had World Health Organization type II or III disease. Approval for retrospective analysis of the patient data was obtained from the ethics committee of Affiliated Cancer Hospital and Institute of Guangzhou Medical University. All patients have signed written consent for the radiation treatment and were informed the technique for radiotherapy, the possible side effects, the risk, and so on. The pretreatment evaluations included complete physical examination, hematologic and biochemistry profiles, fibrotic endoscope examination of the nasopharynx, and MRI or contrast-enhanced computed tomography of the nasopharyngeal and cervical region to evaluate the primary tumor extent and regional
Table 1 The characteristics of patients with nasopharyngeal carcinoma (n = 16)
Sex Male Female T 1 2 3 4 N 0 1 2 3 M 0 1 Stages I II III IV
Number
%
10 6
62.5 37.5
5 11 0 0
31.3 68.7 0 0
3 8 5 0
18.7 50.0 32.3 0
16 0
100 0
9 7 0 0
56.3 43.7 0 0
lymph node involvement. Chest radiography, bone scintigraphy, and abdominal region ultrasonography were used to exclude distant metastases. All patients were staged according to the American Joint Committee on Cancer (Manual for Staging of Cancer, 7th edition) system. The specific characteristics of the patients are showed in Table 1. The median age is 48 years ranging from 30 to 71 years. Simulation and immobilization All patients in the supine position were immobilized by using a head neck and shoulder thermoplastic mask. CT was performed after administration of intravenous contrast medium; 3 mm slices were obtained from the head to 2 cm below the sternoclavicular joint. These CT images were transferred to the Pinnacle3 TPS (version 9.10, Philips Medical Systems, Fitchburg, WI). Target volume delineation and dose prescription The target volumes were defined in accordance with the International Commission on Radiation Units and Measurements reports (ICRU) 50 and 62. The gross tumor volume (GTV) was determined from the MRI, clinical information, and endoscopic findings. The clinical target volumes (CTV) were individually delineated on the basis of tumor invasion. The first CT V(CT V1) was defined as the GTVnx with a 5 to 10 mm margin, including high-risk regions of microscopic extension encompassing the whole nasopharynx. The second CTV was defined as CTV1 with a margin of 5 to 10 mm, including low-risk regions of microscopic extension. If it was in close proximity to critical structures, this margin could be reduced. The OARs included the brainstem, spinal cord, temporal lobe, optic nerves, optic chiasm, lens, eyes, parotid glands, mandible, temporo-mandibular joint, middle ear, and larynx. The prescribed dose was 70 Gy to the PTV of GTVnx, 60 Gy to the PTV of CTV1, 56 Gy to the PTV of second CTV, and 60 to 68 Gy to the PTV of GTV for involved cervical lymph nodes, and it was set to 32 fractions. Treatment was delivered once daily and no more than 5 fractions per week. Treatment planning The PTVs of each target were created by adding a 3 mm 3D margin to the delineated target volume to compensate for treatment set-up variability and internal organ motion. In order to protect the spinal cord, lens, and brain stem, the spinal cord and lens were added a 5 mm 3D margin respectively while the brain stem was added a 4 mm 3D margin. The plan goal for each plan was set to 95% of the PTVs volume to reach 100% of the prescription dose; furthermore, no more than 20% of the PTVs were to receive 110% of the prescription dose and the PTVs receiving less than 93% of the prescription dose did not exceed 1%. In addition, the dose received by each OAR was limited to the dose constraints according to the RTOG0615 protocol, as shown in Table 2. Dose optimization and calculation for s-IMRT and VMAT plans were performed by Pinnacle3 TPS, and the Auto-Planning module in Pinnacle3 TPS was used to create automatic plans for the studied cases. Both IMRT and VMAT were created for an Elekta Synergy linear accelerator with 6 MV photons. Additionally, NPC patients were equally randomly assigned to 2 physicists who have 5 years of experience in planning s-IMRT and VMAT to generate plans, such as: (1) IMRT manual-planning (mIMRT), (2) IMRT auto-planning (aIMRT), (3) VMAT manual-planning (mVMAT) and (4) VMAT auto-planning (aVMAT).
ARTICLE IN PRESS
JID: MDO
Q. Zhang, Y. Peng and X. Song et al. / Medical Dosimetry xxx (xxxx) xxx Table 2 OARs optimization goals in treatment planning OARs
Index
Optimization goal
Priority£
Compromise§
Brain stem Brain stem Brain stem + 2 mm PRV Brain stem + 2 mm PRV Spinal cord Spinal cord + 5 mm PRV Spinal cord + 5 mm PRV Lens+5 mm PRV Optic chiasm Optic nerves Temporal lobes Middle ears Eyes Parotids Parotids TM joint Mandible Trachea Larynx
Max Dose Max DVH Max Dose
54 Gy V50¶ < 4% 60 Gy
High High Medium
No No No
Max DVH
V60# < 1%
High
No
Max Dose Max Dose
40 Gy 45 Gy
High Medium
No No
Max DVH
V45& < 1%
High
No
Max Dose Max Dose Max Dose Max Dose Max Dose Max Dose Mean Dose Max DVH Max Dose Max Dose Max Dose Mean Dose
8 Gy 54 Gy 50 Gy 60 Gy 60 Gy 25 Gy 28 Gy V30∗ ≤ 50% 65 Gy 65 Gy 40 Gy 40 Gy
High High High Low Medium Medium Medium Medium Medium Medium Low Low
No Yes Yes Yes Yes No Yes Yes Yes Yes Yes Yes
Note: The 2 mm and 5 mm represent the 3D margins of OARs respectively, and PRV is planning risk volume. In addition, ∗ , &, ¶, and # is defined as the percentage volume received >30, 45, 50 and 60 Gy respectively. Additionally, £ and § represent the parameters for automatic planning in Auto-Planning module.
[mUS5Gb;August 21, 2019;19:0] 3
Based on many years of clinical experience and combined with the characteristics of the target area, T category of NPC and the principle of sparing OARs, a standard coplanar 7-fields gantry arrangement for IMRT was designed in all cases and delivered in the step and shoot mode, and gantry angles were 0°, 50°, 100°, 150°, 210°, 260°, and 310° respectively. Furthermore, total number of subfields was ≤ 100 and the minimum segment area was set to 10 cm2 and the minimum segment monitor unit was set to 10 MU. On the other hand, VMAT plans used double coplanar full arcs with a 0° collimator angle. One arc was set-up in a clockwise direction from 182° to 178°; conversely, the second arc was performed in a counterclockwise direction from 178° to 182°, and gantry spacing was 4° in VMAT plans. It was noted that the final results of the Auto-Planning module were obtained without more than 3 postoptimizations or slight manual intervention. Meanwhile, the number of iterations for dose calculation in planning process does not exceed 100. All treatment plans were approved by a clinical physicist and a radiation oncologist for all cases.
Plans evaluation According to ICRU 83, quantitative evaluation for the PTVs and OARs was performed using a standard DVH. The conformity index (CI),20 used as a measure of target volume dose distribution conformity, was defined as CI = VT,ref /VT × VT,ref /Vref , where VT,ref was target volume covered by reference isodose, VT was target volume, and Vref was volume of the reference isodose. If the CI is closer to 1, the dose conformity is better. The homogeneity index (HI),21 used as a measure of the evenness of dose distribution, was defined as HI = (D2 to D98) / D50, where D2, D98, and D50 were the doses covering 2%, 98%, and 50% of the PTV, respectively. HI = 0 is the ideal value. Analysis of the OARs included the maximum dose, mean dose and a set of appropriate define (Vx) and define (Dy) values. In addition, the planning times and MUs for treatment plans were recorded and evaluated respectively. The planning time was defined as the effective working time, which started when the PTVs and OARs were defined by the clinicians and finished when the plans were optimized and approved by dosimetrist and oncologist.
Fig. 1. The dose distributions for a representative NPC patient (stage T2N0M0) in 4 types of plans. T stands for transverse, S stands for sagittal, and C stands for coronal; while 1, 2, 3, and 4 represent mIMRT, aIMRT, mVMAT, and aVMAT respectively. (Color version of figure is available online.)
JID: MDO 4
ARTICLE IN PRESS
[mUS5Gb;August 21, 2019;19:0]
Q. Zhang, Y. Peng and X. Song et al. / Medical Dosimetry xxx (xxxx) xxx
Fig. 2. The dosimetric parameters of PTV70 and PTV56 in 4 types of plans, A for PTV70 and B for PTV56 .
Statistical analysis
Monitor unit and planning time
Statistical analysis was performed using SPSS 19.0 (SPSS Inc., Chicago, IL). All datas were expressed as mean ± standard deviation. The data for the different plans were evaluated using Wilcoxon’s signed rank test. The difference was considered significant when P< 0.05.
As shown in Fig. 5, the MUs and planning time for all plans were different. Compared with manual plans, automatic plans had made an average MUs reduction of 7.95% (p <0.05) and an average planning time reduction of 22.27% (p <0.05). In terms of MUs, the average MUs from VMAT were significantly less than those from s-IMRT, while the MUs from aVMAT were similar or evenly less to those from mVMAT; however, compared to mIMRT, the MUs from aIMRT were significantly reduced to some extent. In addition, the average planning times for VMAT were significantly more than those for s-IMRT, and it was an average increase of about 97.11% (p <0.05). Furthermore, the planning times for aVMAT were significantly less than those for mVMAT. The same trend was observed between mIMRT and aIMRT (all p < 0.05).
Results The dose distribution of the PTVs The mean volumes for PTV70 and PTV56 were 52.47 ± 26.60 cm3 and 690.57 ± 130.76 cm3 respectively. All of the parameters obtained from 4 types of plans for early NPC were complied with the ICRU 83 recommendation for target coverage. The dose distributions for a representative NPC patient (stage T2N0M0) were shown in Fig. 1, including 4 types of plans. The dose distributions of the PTVs from VMAT were similar or a certain extent better to s-IMRT. The parameters from DVH of the PTVs in 4 types of plans were shown in Fig. 2. The automatic plans not only had slightly or obviously less maximum dose, especially in aVMAT plans, but also produced more conformal and homogeneous dose distributions in PTVs, as shown in Fig. 3. The CI and HI of automatic plans for PT V70 and PT V56 were 0.802 ± 0.049, 0.037 ± 0.010, and 0.825 ± 0.050, 0.261 ± 0.014 respectively, which were greatly improved when compared with those of manual plans, such as 0.774 ± 0.039, 0.046 ± 0.007 and 0.808 ± 0.047, 0.275 ± 0.011 respectively. It reflected that a more flat dose distribution in target area. In addition, the CI and HI for PTV70 in aVMAT were significantly better than those in aIMRT (p = 0.016 and 0.002 respectively); when regarding PTV56 in aIMRT, the CI was similar and the HI was inferior to those from aVMAT. However, CI and HI both in PTV70 and PTV56 were similar between aIMRT and mVMAT without significant difference (p >0.05). It revealed that aVMAT showed superior dose homogeneity and conformity in PTVs. In 4 types of plans, the qualities of plans for PTVs from mIMRT were the last of relative ranking, while the first of relative ranking was for aVMAT, followed by mVMAT equal to aIMRT.
The dose distribution of OARs Table 3 and Table 4 described the parameters of OARs from the patients. According to dose objectives for OARs, all plans were similar between patients. But all plans did not meet the mean dose objective for parotids, only partial V30 of parotids met the planning objective. Compared with manual plans, the automatic plans had decreased dose to some OARs, including spinal cord, brain stem, eyes, optic nerves, and optic chiasm. In addition, the low dose burden values achieved from VMAT plans were significantly lower than those from s-IMRT plans. Automatic plans had better sparing for spinal cord, while the VMAT had better sparing for brain stem. When compared with all plans, the mVMAT plans resulted in significantly poorer sparing for spinal cord (all p <0.05). Regarding dose received by brain stem, Dmax, V30, and V40 were significantly lower in aVMAT than aIMRT and mVMAT, but these parameters were similar between aIMRT and mVMAT without significant difference, as seen in Fig. 4. Additionally, it could note that the irradiation dose coming from mIMRT for brain stem was relatively highest in 4 types of plans. When considering other normal tissue sparring, it showed no significant difference (p >0.05).
Discussion Dose comparison of PTVs between automatic and manual plans In general, planning and optimization of treatment plans for NPC are complex process, where hard constraints have to be fulfilled. The Auto-Planning module for generating automatic plans would create a lot of special ROIs to ensure a high quality of dose distribution for PTVs and OARs sparing.22 However, if such special ROIs are manually created in manual plans, it is very difficult or even impossible to achieve, especially depending on the efforts and experiences of manual planners. In this sense, quantitatively better treatment plans could be generated by using the Auto-Planning module. For example, several researches have shown that it is feasible and similar or superior to previous manual plans because of maintaining the target coverage and better sparing OARs.9,10,23 As expected, compared to manual plans, automatic plans have shown significantly superior conformity and homogeneity. The CI and HI of aVMAT are more superior to those of aIMRT and mVMAT, especially in cold and hot dose spots and OARs sparing, while the CI and HI in PTVs are similar between aIMRT and mVMAT. Many researchers have found that VMAT provides similar or better results in dose coverage of the PTVs in NPC when compared with IMRT, especially in early T category.2 From our study cases, automatic IMRT is comparable to manual VMAT in dose homogeneity and conformity for PTVs. It means that automatic IMRT may provide an option to replace the manual VMAT, though manual VMAT may offer dosimetric advantages in head and neck cancer. In this way, it does not need to blindly adjust the dose coverage of the PTVs by manual VMAT technique, which has higher performance requirement of linac. As a result, it not only can reduce the loss of linac, but also can decrease the costs of patients due to the cost
ARTICLE IN PRESS
JID: MDO
[mUS5Gb;August 21, 2019;19:0]
Q. Zhang, Y. Peng and X. Song et al. / Medical Dosimetry xxx (xxxx) xxx
5
Fig. 3. Statistical comparison of CI and HI are for PT V70 and PT V56 respectively in 4 types of plans, A for PTV70 and B for PTV56 .
Table 3 The objective and pass rate for various OARs in 4 types of plans OARs
Brain stem Brain stem Brain stem + 2 mm PRV Spinal cord Spinal cord + 5 mm PRV Lens+5 mm PRV Optic chiasm Optic nerves Temporal lobes Middle ears Eyes Parotids Parotids TM joint Mandible Larynx
Objective
Max V50¶ D1# Max D1# Max Max Max D2& D2& D2& Mean V30∗ Max D1# Mean
Pass rate (%)
≤ < < ≤ ≤ ≤ ≤ ≤ ≤ ≤ ≤ ≤ ≤ < < <
54 Gy 4% 60 Gy 45 Gy 50 Gy 12 Gy 54 Gy 50 Gy 60 Gy 60 Gy 30 Gy 26 Gy 50% 70 Gy 75 Gy 45 Gy
mIMRT
aIMRT
mVMAT
aVMAT
56.25 10 0.0 0 10 0.0 0 10 0.0 0 10 0.0 0 10 0.0 0 93.75 81.25 56.25 10 0.0 0 10 0.0 0 0.00 62.50 10 0.0 0 10 0.0 0 75.00
62.50 10 0.0 0 10 0.0 0 10 0.0 0 10 0.0 0 10 0.0 0 93.75 87.50 56.25 10 0.0 0 10 0.0 0 0.00 68.75 10 0.0 0 10 0.0 0 81.25
75.00 10 0.0 0 10 0.0 0 10 0.0 0 10 0.0 0 10 0.0 0 10 0.0 0 75.00 68.75 10 0.0 0 10 0.0 0 0.00 62.50 10 0.0 0 10 0.0 0 81.25
81.25 10 0.0 0 10 0.0 0 10 0.0 0 10 0.0 0 10 0.0 0 10 0.0 0 81.25 62.50 10 0.0 0 10 0.0 0 0.00 62.50 10 0.0 0 10 0.0 0 87.50
Manual vs automatic
0.429 N/A N/A N/A N/A N/A 0.192 0.072 0.719 N/A N/A N/A 0.599 N/A N/A 0.488
p aIMRT vs mVMAT 0.509 N/A N/A N/A N/A N/A 0.843 0.429 0.227 N/A N/A N/A 0.669 N/A N/A 0.537
aIMRT vs aVMAT 0.154 N/A N/A N/A N/A N/A 0.345 0.950 0.339 N/A N/A N/A 0.645 N/A N/A 0.916
Note: The 2 mm and 5 mm represent the 3D margins of OARs respectively, and PRV is planning risk volume. In addition, ¶ is defined as the percentage volume received >50 Gy; # is defined as the dose received by 1% of the volume; & is defined as the dose received by 2% of the volume; ∗ is defined as the percentage volume received >30 Gy.
ARTICLE IN PRESS
JID: MDO 6
[mUS5Gb;August 21, 2019;19:0]
Q. Zhang, Y. Peng and X. Song et al. / Medical Dosimetry xxx (xxxx) xxx
Table 4 The comparison of doses received by various OARs in 4 types of plans OARs
Brain stem Brain stem Brain stem + 2 mm PRV Spinal cord Spinal cord + 5 mm PRV Lens + 5 mm PRV Optic chiasm Optic nerves Temporal lobes Middle ears Eyes Parotids Parotids TM joint Mandible Larynx
Index
Max (Gy) V50¶ (%) D1# (Gy) Max (Gy) D1# (Gy) Max (Gy) Max (Gy) Max (Gy) D2& (Gy) D2& (Gy) D2& (Gy) Mean (Gy) V30∗ (%) Max (Gy) D1# (Gy) Mean (Gy)
Mean dose ± SD mIMRT
aIMRT
mVMAT
aVMAT
54.29 ± 2.24 1.83 ± 1.15 57.69 ± 1.18 37.65 ± 2.04 41.79 ± 1.89 6.91 ± 0.88 43.58 ± 7.71 47.79 ± 3.62 59.55 ± 4.33 51.12± 2.01 14.93 ± 3.13 34.41 ± 2.60 47.60 ± 5.31 62.54 ± 4.59 62.70 ± 2.83 43.39 ± 2.45
53.64 ± 2.26 0.99 ± 0.95 57.74 ± 1.28 35.13 ± 1.07 40.96 ± 1.93 6.26 ± 0.87 40.65 ± 8.14 45.69 ± 4.41 59.15 ± 3.53 51.71 ± 2.61 12.98 ± 2.88 34.75 ± 2.36 47.01 ± 6.04 61.14 ± 5.93 61.78 ± 1.91 42.50 ±2.34
52.76 ± 2.23 0.46 ± 0.63 57.23 ± 1.54 38.25 ± 0.88 42.32 ± 2.14 5.81 ± 0.71 39.13 ± 7.99 47.20 ± 4.30 57.45 ± 3.78 51.71 ± 2.98 11.88 ± 3.92 34.43 ± 1.49 47.77 ± 7.90 61.82 ± 4.86 61.43 ±4.69 42.99 ± 2.08
51.96 ± 1.96 0.41 ± 0.42 57.12 ± 2.28 35.20 ± 1.61 40.93 ± 1.06 5.47 ± 1.38 37.25 ± 6.72 45.92 ± 4.40 58.06 ± 4.10 50.31 ± 4.82 11.24 ± 3.79 34.29 ± 0.92 49.11 ±6.82 61.38 ± 3.71 60.77 ± 4.97 42.43 ± 2.20
Manual vs automatic
0.239 0.272 0.946 0.0 0 0 0.028 0.164 0.222 0.107 0.919 0.831 0.514 0.955 0.817 0.649 0.305 0.200
p aIMRT vs mVMAT 0.246 0.284 0.316 0.0 0 0 0.095 0.429 0.598 0.336 0.200 0.990 0.399 0.854 0.765 0.870 0.760 0.533
aIMRT vs aVMAT 0.028 0.203 0.346 0.677 0.961 0.198 0.207 0.884 0.429 0.991 0.235 0.653 0.367 0.953 0.461 0.935
Note: The 2 mm and 5 mm represent the 3D margins of OARs respectively, and PRV is planning risk volume. In addition, ¶ is defined as the percentage volume received >50 Gy; # is defined as the dose received by 1% of the volume; & is defined as the dose received by 2% of the volume; ∗ is defined as the percentage volume received >30 Gy.
of VMAT is more expensive than IMRT. More importantly, it would reduce the dependence on planner experience to minimize the impacts in treatment plans, and improve the planning efficiency. On the contrary, if conditions permit, automatic VMAT is a better option for recommendation. Dose comparison of OARs between automatic and manual plans
Fig. 4. Comparison of dosimetric quality for brain stem in 4 types of plans.
Due to it does not know how much lower dose is better sparing an OAR for a particular patient, a dose guideline only provides a guidance in planning process and cannot ensure an optimal dose distribution in a treatment plan. During the planning process, individual optimization performed by the experienced planner has a very large solution space, which leaves with much uncertainty to ensure an optimal treatment plan, especially sparing OARs. If the optimization is performed by automatic algorithm, it would decrease uncertainty to increase the probability for the achievable optimal plan through the reduction of variation in planner intervention. In the study cases, the average doses for OARs from automatic plans are similar or less to those from manual plans.
Fig. 5. Comparison of monitor units and planning times in 4 types of plans.
JID: MDO
ARTICLE IN PRESS
[mUS5Gb;August 21, 2019;19:0]
Q. Zhang, Y. Peng and X. Song et al. / Medical Dosimetry xxx (xxxx) xxx
Although the target area of NPC is close to the spinal cord, especially brain stem, the results have demonstrated that automatic plan could protect the spinal cord better than manual plan. Additionally, aIMRT is comparable to aVMAT in spinal cord sparing. Compared to s-IMRT, VMAT plans have advantages for brain stem sparing, as shown in Fig. 4. It illustrates that aVMAT plans are confined to a narrower region in doses distribution of brain stem when compared with mVMAT plans, but the dose received by brain stem does not have significant difference between aIMRT and mVMAT. It reveals that aIMRT is a certain extent better than mVMAT in OARs sparing. The doses of other OARs in automatic plan, used to evaluate the quality of treatment plan, are slight lower or even similar to those in manual plan. It can be suggested that planning balance between various optimization criteria is clinically favored during the automatic planning process. Due to parotids are partial overlap with the PTVs, it is more difficult to judge parotids doses whether meeting dose constraints.24 If more consideration is need for better sparing of OARs, the automatic planning technique is more needed to be carried out. MUs and planning times of automatic and manual plans In the study cases, aVMAT plans have fewer MUs, followed by mVMAT plans, where the MUs are much less than those in aIMRT plans. The fewer MUs can lead to significantly improve the efficiency of plan implementation to reduce the treatment time, which could benefit the patients to reduce discomfort and intrafraction variation during the treatment. In particular, it is worth to note that secondary cancer risk may be potentially reduced during radiotherapy due to the benefits from fewer MUs.25 Using standard settings in Auto-Planning module, the automatic planning times have decreased about 26% and 20% respectively when compared with manual planning. Because the Auto-Planning module could potentially make it possible for planners to spend less time on the more tedious steps in planning process during optimization. In addition, this is not just a matter of shortening planning time to improve planning efficiency by using Auto-Planning module. First, the Auto-Planning module has an advantage to leave planner free for other tasks of plans due to reducing the need for multiple plan reviews and runs in the background. Second, this makes it possible for planners to spend more time to focus on planning goals with difficult dose. On the other hand, the planning times spent on aIMRT plans are significantly shorter when compared with aVMAT plans. Because 180 control points need to be considered to calculate the dose of VMAT plans with double arcs. It is equivalent to calculate 180 irradiation fields, which are significantly more than those of IMRT plans. Conclusions Auto-Planning module could represent an effective way to improve a planning process with consistent plan qualities for NPC in early stage. Standardization of the plan quality is consistently achieved from automatic plans, which are similar or better to those from manual plans. But the effective times required for planning and optimization and MUs are significantly decreased in automatic process. Furthermore, when compared automatic plans with manual plans, the dose to most of OARs could be similar or decreased to some extent, especially in brain stem and spinal cord. It means that the automatic plans provide comparable or superior plans when compared to manual plans. After a comprehensive analysis, the plan qualities of mIMRT are the last of relative ranking, while the first of relative ranking is for aVMAT, followed by mVMAT that equal to aIMRT. The automatic VMAT plan has greater clinical advantages and should be recommended to a better option for treating NPC in early stage, but automatic IMRT would be
7
preferentially considered instead of manual VMAT. However, owing to small sample size and retrospective study, a large sample of prospective studies is needed to further confirm the above results before extensive implementation for NPC. Conflict of Interest The authors declare no conflict of interest. Ethics Approval and Consent to Participate All patients included into this study have given their approval to use their data for scientific research. All personal information to identify patients was removed from the image data and analyzed retrospectively. Acknowledgments This work was supported financially by Guangzhou Medical Key Discipline Construction Project (2017 to 2019): Cancer Therapeutics and Experimental Oncology Project, Technology Project of Guangzhou Medical and Health Science (Grant number: 20171A010314; 20181A011095). References 1. Kam, M.M.; Chau, R.M.; Suen, J.; et al. Intensity-modulated radiotherapy in nasopharyngeal carcinoma: Dosimetric advantage over conventional plans and feasibility of dose escalation. Int J Radiat Oncol Biol Phys 56:145–57; 2003. 2. Sun, Y.; Guo, R.; Yin, W.J.; et al. Which t category of nasopharyngeal carcinoma may benefit most from volumetric modulated arc therapy compared with step and shoot intensity modulated radiation therapy. PLoS ONE 8:e75304; 2013. 3. Nelms, B.E.; Robinson, G.; Markham, J.; et al. Variation in external beam treatment plan quality: an inter-institutional study of planners and planning systems. Pract Radiat Oncol 2:296–305; 2012. 4. Batumalai, V.; Jameson, M.G.; Forstner, D.F.; et al. How important is dosimetrist experience for intensity modulated radiation therapy? A comparative analysis of a head and neck case. Pract Radiat Oncol 3:e99–e106; 2013. 5. Hansen, C.R.; Nielsen, M.; Bertelsen, A.S.; et al. Automatic treatment planning facilitates fast generation of high-quality treatment plans for esophageal cancer. Acta Oncol 56:1–6; 2017. 6. Wang, S.; Zheng, D.; Zhang, C.; et al. Automatic planning on hippocampal avoidance whole-brain radiotherapy. Med Dosim 42:63–8; 2017. 7. Hazell, I.; Bzdusek, K.; Kumar, P.; et al. Automatic planning of head and neck treatment plans. J Appl Clin Med Phys 17:272–82; 2016. 8. Chen, H.; Wang, H.; Gu, H.; et al. Study for reducing lung dose of upper thoracic esophageal cancer radiotherapy by auto-planning: Volumetric-modulated arc therapy vs intensity-modulated radiation therapy. Med Dosim 43:243–50; 2018. 9. Krayenbuehl, J.; Norton, I.; Studer, G.; et al. Evaluation of an automated knowledge based treatment planning system for head and neck. Radiat Oncol 10:226; 2015. 10. Hansen, C.R.; Bertelsen, A.; Hazell, I.; et al. Automatic treatment planning improves the clinical quality of head and neck cancer treatment plans. Clin Translat Radiat Oncol 119:S396–7; 2016. 11. Li, K.; Chang, X.; Wang, J.; et al. SU-F-T-358: is auto-planning useful for volumetric-modulated arc therapy planning in rectal cancer radiotherapy? Med Phys 43:3545; 2016. 12. Lee, T.F.; Chao, P.J.; Ting, H.M.; et al. Comparative analysis of smartarc-based dual arc volumetric-modulated arc radiotherapy (VMAT) versus intensity-modulated radiotherapy (IMRT) for nasopharyngeal carcinoma. J Appl Clin Med Phys 12:158–74; 2011. 13. Ning, Z.H.; Mu, J.M.; Jin, J.X.; et al. Single arc volumetric-modulated arc therapy is sufficient for nasopharyngeal carcinoma: A dosimetric comparison with dual arc VMAT and dynamic MLC and step-and-shoot intensity-modulated radiotherapy. Radiat Oncol 8:237; 2013. 14. Tol, J.P.; Delaney, A.R.; Dahele, M.; et al. Evaluation of a knowledge-based planning solution for head and neck cancer. Int J Radiat Oncol Biol Phys 91:612–20; 2015. 15. Doornaert, P.; Verbakel, W.F.; Bieker, M.; et al. RapidArc planning and delivery in patients with locally advanced head-and-neck cancer undergoing chemoradiotherapy. Int J Radiat Oncol Biol Phys 79:429–35; 2011. 16. Hansen, C.R.; Nielsen, M.; Bertelsen, A.S.; et al. Automatic treatment planning facilitates fast generation of high-quality treatment plans for esophageal cancer. Acta Oncol 56:1495–500; 2017. 17. Fogliata, A.; Belosi, F.; Clivio, A.; et al. On the pre-clinical validation of a commercial model-based optimisation engine: application to volumetric modulated arc therapy for patients with lung or prostate cancer. Radiother Oncol 113(3):385–91; 2014.
JID: MDO 8
ARTICLE IN PRESS
[mUS5Gb;August 21, 2019;19:0]
Q. Zhang, Y. Peng and X. Song et al. / Medical Dosimetry xxx (xxxx) xxx
18. Hussein, M.; South, C.P.; Barry, M.A.; et al. Clinical validation and benchmarking of knowledge-based IMRT and VMAT treatment planning in pelvic anatomy. Radiother Oncol 120:473–9; 2016. 19. Wang, J.; Chen, Z.; Li, W.; et al. A new strategy for volumetric-modulated arc therapy planning using autoplanning based multicriteria optimization for nasopharyngeal carcinoma. Radiat Oncol 13:94; 2018. 20. Paddick, I. A simple scoring ratio to index the conformity of radiosurgical treatment plans. Technical note. J Neurosurg 93(Suppl 3):219–22; 20 0 0. 21. Kataria, T.; Sharma, K.; Subramani, V.; et al. Homogeneity Index: an objective tool for assessment of conformal radiation treatments. J Med Phys 37:207–13; 2012.
22. Nawa, K.; Haga, A.; Nomoto, A.; et al. Evaluation of a commercial automatic treatment planning system for prostate cancers. Med Dosim 42:203–9; 2017. 23. Wu, B.; Kusters, M.; Kunzebusch, M.; et al. Cross-institutional knowledge-based planning (KBP) implementation and its performance comparison to auto-planning engine (APE). Radiother Oncol 123:57–62; 2017. 24. Johnston, M.; Clifford, S.; Bromley, R.; et al. Volumetric-modulated arc therapy in head and neck radiotherapy: a planning comparison using simultaneous integrated boost for nasopharynx and oropharynx carcinoma. Clin Oncol 23:503–11; 2011. 25. Hall, E.J. Intensity-modulated radiation therapy, protons, and the risk of second cancers. Int J Radiat Oncol Biol Phys 65:1–7; 2006.