|Year : 2019 | Volume
| Issue : 8 | Page : 33-38
Comparison of biological-based and dose volume-based intensity-modulated radiotherapy plans generated using the same treatment planning system
K Senthilkumar1, KJ Maria Das2
1 Department of Medical Physics, Karnataka Cancer Therapy and Research Institute, Hubli, Karnataka; Research and Development Centre, Bharathiar University, Coimbatore, Tamil Nadu, India
2 Research and Development Centre, Bharathiar University, Coimbatore, Tamil Nadu; Department of Radiotherapy, Sanjay Gandhi Postgraduate Institute of Medical Sciences, Lucknow, Uttar Pradesh, India
|Date of Web Publication||22-Mar-2019|
Dr. K J Maria Das
Department of Radiotherapy, Sanjay Gandhi Postgraduate Institute of Medical Sciences, Lucknow - 226 014, Uttar Pradesh
Source of Support: None, Conflict of Interest: None
Purpose: Nowadays, most of the radiotherapy (RT) treatment planning systems (TPSs) uses dose or dose-volume (DV)-based cost functions for Intensity modulated radiation therapy (IMRT) fluence optimization. Recently, some of the TPSs incorporated biological-based cost function for IMRT optimization. Most of the previous studies compared IMRT plans optimized using biological-based and DV-based cost functions in two different TPSs. Hence, the purpose of the study is to compare equivalent uniform dose (EUD)-based and DV-based IMRT plans generated using the same TPS.
Materials and Methods: Twenty patients with prostate cancer were retrospectively selected for this study. For each patient, two IMRT plans were generated using EUD-based cost function (EUD_TP) and DV-based cost (DV_Treatment Plan (TP)), respectively. The generated IMRT plans were evaluated using both physical and biological dose evaluation indices.
Results: Biological-based plans ended up with a highly inhomogeneous target dose when compared to DV-based plans. For serial organs, Dnear-max or D2%(Gy) of EUD-based plans showed significant difference with DV-based plans (P = 0.003). For both rectum and bladder, there was a significant difference in mean dose and D30%(Gy) dose between EUD-based plans and DV-based plans.
Conclusion: In this study, we decoupled the influence of optimization parameters from the potential use of EUD-based cost functions on plan quality by generating both plans in the same TPS.
Keywords: Biological-based optimization, dose volume-based optimization, equivalent uniform dose, intensity-modulated radiotherapy
|How to cite this article:|
Senthilkumar K, Maria Das K J. Comparison of biological-based and dose volume-based intensity-modulated radiotherapy plans generated using the same treatment planning system. J Can Res Ther 2019;15, Suppl S1:33-8
|How to cite this URL:|
Senthilkumar K, Maria Das K J. Comparison of biological-based and dose volume-based intensity-modulated radiotherapy plans generated using the same treatment planning system. J Can Res Ther [serial online] 2019 [cited 2020 Jul 7];15:33-8. Available from: http://www.cancerjournal.net/text.asp?2019/15/8/33/243505
| > Introduction|| |
The major goal of radiation therapy (RT) would be to deliver high dose to target while minimizing dose to normal tissues. Using intensity-modulated RT (IMRT), treatment plan could be optimized to produce desired dose distribution inside the tumor with reduced normal tissue dose. Dose- or dose-volume (DV)-based physical cost functions or objectives are commonly used for IMRT fluence optimization in clinical practice., As the DV-based cost function does not reflect the dose-response nature of both tumor and normal structure, need for the cost functions which simulate the dose-response nature of the tumor and organ at risk (OAR) was in great demand. IMRT optimization using biological-based cost functions such as tumor control probability (TCP), normal tissue complication probability (NTCP), and equivalent uniform dose (EUD) have been recently developed. IMRT optimization using EUD-based cost functions were extensively analyzed by Wu et al. To explore the real potential of biological-based cost functions, some investigators have already compared biological-based IMRT plans with DV-based IMRT Plans.,,,, Most of the above studies were limited because they used two different treatment planning system (TPS) for comparison of biological-based IMRT plans with DV-based IMRT Plans. For instance, Monaco TPS (CMS Inc., St. Louis, MO) was used for generating biological-based treatment plans whereas XIO TPS (CMS Inc., St. Louis, MO, USA) was used for generating DV-based treatment plans. The inherent parameters which may influence final dose distribution between the TPSs are IMRT fluence optimization algorithm, IMRT dose calculation algorithm, and IMRT beam segmentation methods., For example, Monaco TPS IMRT fluence optimization is based on constraint optimization algorithm in which cost function specified to target would be considered as objective and other cost function specified to OARs would be treated as constraints. In contrast to Monaco TPS, XIO TPS uses conjugate gradient optimization algorithm in which target and OAR cost functions are contributing to the overall plan score (objective score) in accordance with user-specified weights., Further, Monaco TPS uses X-ray volume Monte Carlo algorithm for final dose calculation whereas XIO TPS uses superposition algorithm.,, Because of two different TPS were used for optimization and dose computation, it is still not clear if the claimed advantage of EUD-based cost function over DV-based cost function on plan quality is really due to EUD-based cost function or due to changes in the above said influencing factors between the TPS. Hence, this study aims to decouple these factors to understand if EUD-based cost functions are really making a positive impact on plan quality.
To eliminate the influence of the above-said parameters in the final dose distribution, both biological-based IMRT plans and DV-based IMRT plans should be performed in the same TPS. New Monaco TPS version 5.0 allows generation of IMRT treatment plan using purely biological-based cost function and vice versa. Although there was an extensive experience for using Monaco TPS for EUD-based planning study, those studies were limited because most of the studies evaluated their treatment plans using physical dose evaluation parameters such as conformity, heterogeneity, mean dose, and dose volume.,, Hence, the generated treatment plans were evaluated using both physical and biological dose evaluation parameters such as TCP and NTCP.
| > Materials and Methods|| |
Twenty patients with prostate cancer were retrospectively selected for this study. All these patients were already been treated using Xio generated IMRT plans. DV-based cost functions were used to generate the Xio IMRT plans. The acquired computed tomography (CT) and magnetic resonance imaging (MRI) images were imported into the Monaco TPS and fused using fusion mode. Gross tumor volume (GTV) was delineated on MRI image and transformed to CT data set for both the cases. GTV on CT data were manually edited based on physical examination. CTV containing the microscopic disease extension around the GTV was delineated manually by considering tumor spread. For prostate case, single high-risk volume planning target volume (PTV)_72 Gy was created by expanding CTV. The contours for all OARs such as rectum, bladder, and femoral heads were delineated on the CT data set.
For each patient, two IMRT plans were generated in Monaco TPS (1) IMRT plan using EUD-based cost functions called EUD_TP and (2) IMRT plan using DV-based cost functions called DV_TP. [Table 1] shows the details of the cost function used for target and OARs for both the plans. The generated treatment plans were prescribed a dose of 72 Gy in 36 fractions to PTV_72 Gy volume. All IMRT plans were created using 6MV photon beam and modulated with multileaf collimator (80 leaves) from the linear accelerator (Synergy platform, Elekta Medical Systems). Treatment plan calculations are carried out with tissue heterogeneities. To make both the treatment plans almost comparable, optimization was stopped when there was degradation in specified target coverage while trying to reduce the OAR dose further. The treatment plan acceptance criteria were to deliver 95% of the prescription dose to 95% of the volume of PTV. The single experienced planner was assigned to generate all the plans to avoid any variation of IMRT plan quality due to operator skill and experience.
|Table 1: Details of cost function used for target and organs at risks in Monaco treatment planning system|
Click here to view
Biological optimization in Monaco treatment planning system
Monaco TPS is the first commercial IMRT TPS to incorporate biological-based optimization features. Apart from several DV-based cost functions, Monaco TPS offers three biological-based cost functions named Poisson statistics cell kill model for the target, serial complication model, and parallel complication model for OAR. The biological cost functions incorporated into Monaco TPS was developed by Alber and Reemtsen. Detailing the full mathematics of their work is beyond the scope of this paper. For each cost function, a three-dimensional dose distribution is reduced to a single index called isoeffect. On the other hand, clinical goals specified by the user referred as isoconstraint. After each optimization step, calculated isoeffects values for all cost functions are compared with user-specified isoconstraints. Optimization process will end when there is no possibility to further improve the target dose distribution without affecting the OAR dose.
Physical dose evaluation indices
For high-risk tumor volume Dmean(Gy), Dnear-min or D98%(Gy), Dnear-max or D2%(Gy), and V107%values were evaluated for both cases. For OARs Dmean(Gy), D30%(Gy), Dnear-max or D2%(Gy), V30 Gy (%), and D5%(Gy) values were evaluated for both the cases. Target conformity index (CI) and homogeneity index (HI) were calculated as per ICRU 83 report. HI was expressed in terms of the ratio (D2%-D98%/D50%, where D50%(Gy) is the minimum dose represents 50% of the target volume.
Biological dose evaluation indices
EUD based TCP and NTCP models were proposed by Niemierko. The generalized EUD (gEUD) was calculated using the following formula,
In this equation, vi is the ith partial volume receiving dose Di and “a” is tissue-specific parameters which is negative for target and positive for normal tissues. The vi and Di were obtained from dose-volume histogram and the value of “a” was obtained from Wu et al.
TCP is defined as the probability that no clonogenic cells survive after the treatment. TCP can be calculated from the following equation,
Where TCD50 is the tumor dose to control 50% of the tumor when irradiated homogenously and γ50 is the slope of the dose-response curve which is specific to the tumor or normal tissue. TCD50 was obtained from Rana and Cheng and γ50 from Park et al.
NTCP is based on the linear quadratic model, and it is function of the delivered dose and irradiated normal tissue volume.
In this equation, TD50 is the tolerance dose of bladder and rectum for a 50% complication rate at a specific time interval. TD50 values were obtained from the model of Emami et al. The radiobiological parameters used to calculate EUD, TCP, and NTCP were given in [Table 2]. Microsoft Excel application was used for numerical calculations. Calculated TCP and NTCP values were shown in [Table 3].
|Table 2: Radiobiological parameters used to calculate equivalent uniform dose, tumor control probability and normal tissue complication probability|
Click here to view
|Table 3: Summary of evaluated biological and physical dosimetric values target and organs at risks in prostate case|
Click here to view
To determine the statistical significance, two-tailed paired t-tests were performed with P < 0.05 considered to be statistically significant. All calculations were performed using the online statistical packages software called VassarStats (Vassar College Poughkeepsie, NY, USA).
| > Results|| |
Both plans were clinically acceptable in terms of target coverage and OAR sparing. More than 98% of the PTVs received at least 95% of the prescribed dose. From [Table 3], it was shown that EUD_TP-based plans showed slightly better conformity, but an increased in-homogeneity when compared to DV_TP plans. It is around 4% of the PTV volume received a dose ≥107% of the prescribed dose while it is 1.7% for DV_TP plans. [Figure 1]a shows the cumulative DV histograms (DVH) curve of high-risk volume PTV_72 Gy for a typical head and neck patient.
|Figure 1: Comparison of cumulative dose-volume histograms between equivalent uniform dose-TP and dose-volume-TP for (a) planning target volume-72 Gy (b) rectum (c) bladder and (d) left femur and (e) right femur of typical prostate case|
Click here to view
As shown in [Table 3], the results of OARs of all prostate cases. For rectum, there was a significant difference in the evaluated mean dose and D30%(Gy) dose between EUD_TP plans and DV_TP plans. The calculated NTCP value was 3.3% for EU_TP whereas it was 5.7% for DV_TP plans and was statistically significant. Also for bladder, there was a significant difference in the evaluated mean dose and D30%(Gy) dose between EUD_TP plans and DV_TP plans. The calculated NTCP values were 1.2% for EUD_TP plans while it was 2.7% for DV_TP plans and was statistically significant. For both the femoral heads D5%(Gy) dose shows the significant difference between the two plans. [Figure 1]b,[Figure 1]c,[Figure 1]d,[Figure 1]e shows the cumulative DVH curves of all OARs evaluated for a typical prostate cancer patient.
Further, [Figure 2] shows the calculated correlation coefficient values between Monaco and Xio plan evaluation indices.
|Figure 2: Calculated correlation coefficients between plans evaluation indices derived from Monaco dose-volume-based plans and Xio-based plans respectively|
Click here to view
| > Discussion|| |
For each patient, two IMRT plans were generated using EUD- and DV-based cost functions in Monaco TPS. The generated plans were compared using physical dose evaluation indices such as Dmean(Gy), Dnear-min or D98%(Gy), Dnear-max or D2%(Gy), CI, HI and biological dose evaluation indices such as TCP and NTCP. Both the optimized plans results in clinically acceptable dose level in terms of target coverage and OAR sparing. From [Table 3], it was noticed that EUD-based plans result in slightly better conformity than DV-based plans for both the cases. Dmean(Gy) and Dnear-min or D98%(Gy) in EUD-based plans were higher than DV based but not statistically significant. Although we used a DV-based cost functions to control the in-homogeneity inside the target in EUD-based plans, it was noticed from HI and Dnear-max or D2%(Gy) values that EUD-based plans always ended up with a slightly increased inhomogeneous target dose when compared with DV-based plans. This is because EUD-based cost function is insensitive to hot spots inside the target and always tries to nullify the cold spots to maximize the target cell kill. As a consequence, it leads to better conformity and increased heterogeneity inside the tumor. On the other hand, it is also possible to achieve similar plans by carefully chosen DV-based objective comparable to EUD-based plans. However, again it is cumbersome and time-consuming.
In terms of OAR sparing, both the plans produced clinically acceptable dose distributions. From physical dose evaluation indices, it was evident that EUD-based plans achieve better OAR sparing than DV-based plans with most of the differences were statistically significant. Even calculated biological dose evaluation indices substantiate the improved OARs dose in EUD-based plans. Hence, in general, EUD-based plans offers improved OAR sparing when compared to DV-based plans. This is because EUD-based cost functions are degenerative in nature and has the tendency to find the optimal solution. This results in improved OAR sparing using EUD optimized plans than DV plans.
Semenenko VA et al. have compared EUD-based IMRT plans with DV-based IMRT plans using Monaco TPS and XIO TPS, respectively. Furthermore, they compared DV-based plans with semi-biological-based plans because the Poisson cell kill model was mandatory cost function for the target, it was not possible to create IMRT plans using entirely EUD-based cost functions in Monaco TPS at that time. Similarly, Semenenko VA et al. also made the same kind of comparison using Monaco TPS and Pinnacle3 (P3IMRT, Version 8.0 h, Philips Medical Systems, Milpitas, CA) for biological plans and XIO TPS and Tomotherapy (Version 3.1.2, Tomotherapy Inc., Madison, WI) for DV-plans. Using Pinnacle3 TPS, they made a comparison between EUD-based plans and DV-based plans. However, methodology was not clear about what cost functions were used for unspecified tissue in EUD-based plans. Another study was performed using Pinnacle3 system, in which initial optimization process was started using DV-based cost function and then gEUD (generalized EUD)-based objective was added to assist the optimization process. Finally, Kan MW et al. used Eclipse TPS (Varian Medical Systems, Palo Alto, CA) to perform the same comparison and they used DV-based cost function in addition to the EUD-based cost function for OARs. Hence, most of the reported publications on this topic were limited because those studies were compared IMRT plans optimized with biological-based and DV-based cost functions using two different TPS. IMRT optimization algorithm, IMRT dose calculation algorithm, and IMRT segmentation method are the inherent parameter which may affect the final dose distribution was not addressed by most of the above studies.
To eliminate the influence of the of the above-said parameters on final dose distributions, both biological-based IMRT plan and DV-based IMRT plan should be performed in the same TPS. Incidentally, new Monaco TPS version 5.0 allows generation of IMRT treatment plan using purely EUD-based cost function and vice versa. The major advantage in generating both types of plans in the same TPS is to decouple the effect of above said influencing parameters in the potential use of EUD-cost functions on plan quality. This will explore the real potential of EUD-based cost functions on plan quality. To eliminate these influence parameters in the final dose distribution, both EUD and DV-based IMRT plans were generated using single TPS and plans were compared.
Hence, it was noticed that EUD-based IMRT plans were better than DV-based IMRT plans generated using Monaco TPS in terms of OAR sparing. As the EUD-based plans generated using Monaco TPS produced better OAR sparing with statistically significant difference, the impact of above said influencing parameters on plan quality were limited. Therefore, we conclude that EUD-based cost function is purely responsible for the improved OAR sparing when compared to DV-based cost functions, not the above-mentioned influence parameters. Furthermore, this study showed the real potential of EUD-based cost functions when compared to DV-based cost function by eliminating the above-said influencing parameters.
| > Conclusion|| |
From our study, it can be concluded that the resultant improved EUD-based treatment is really due to EUD cost function-based IMRT optimization. By decoupling the influencing parameters from EUD-based cost functions optimization, it was understood that EUD-based optimization is really making some positive impact on plan quality. As a result, EUD-based IMRT plans results in improved OAR sparing. Furthermore, we substantiate the improved OAR sparing of EUD-based plans using biological-based dose evaluation indices such as TCP and NTCP.
Many thanks to Dr. Azhagurajan, Tohoku University and Mr. Vaitheesvaran, Philips Medical Systems India (P) Ltd for discussion on this topic.
Financial support and sponsorship
Conflicts of interest
There are no conflicts of interest.
| > References|| |
Wu Q, Mohan R. Algorithms and functionality of an intensity modulated radiotherapy optimization system. Med Phys 2000;27:701-11.
Spirou SV, Chui CS. A gradient inverse planning algorithm with dose-volume constraints. Med Phys 1998;25:321-33.
Allen Li X, Alber M, Deasy JO, Jackson A, Ken Jee KW, Marks LB, et al.
The use and QA of biologically related models for treatment planning: Short report of the TG-166 of the therapy physics committee of the AAPM. Med Phys 2012;39:1386-409.
Wu Q, Mohan R, Niemierko A, Schmidt-Ullrich R. Optimization of intensity-modulated radiotherapy plans based on the equivalent uniform dose. Int J Radiat Oncol Biol Phys 2002;52:224-35.
Semenenko VA, Reitz B, Day E, Qi XS, Miften M, Li XA, et al.
Evaluation of a commercial biologically based IMRT treatment planning system. Med Phys 2008;35:5851-60.
Qi XS, Semenenko VA, Li XA. Improved critical structure sparing with biologically based IMRT optimization. Med Phys 2009;36:1790-9.
Anderson N, Lawford C, Khoo V, Rolfo M, Joon DL, Wada M, et al.
Improved normal tissue sparing in head and neck radiotherapy using biological cost function based-IMRT. Technol Cancer Res Treat 2011;10:575-83.
Lee TF, Ting HM, Chao PJ, Wang HY, Shieh CS, Horng MF, et al
. Dosimetric advantage of generalized equivalent uniform dose-based optimization on dose-volume objectives in intensity-modulated radiotherapy planning for bilateral breast cancer. Br J Radiol 2012;85:1499-506.
Kan MW, Leung LH, Yu PK. The use of biologically related model (Eclipse) for the intensity-modulated radiation therapy planning of nasopharyngeal carcinomas. PLoS One 2014;9:e112229.
Bertsekas DP. Exact penalty methods and lagrangian methods. In: Flaherty M, editor. Constrained Optimization and Lagrange Multiplier Methods. 2nd
ed. Belmont, Massachusetts: Athena Scientific; 1996. p. 180-297.
Nocedal J, Wright SJ. Conjugate gradient methods. In: Peter G, editor. Numerical Optimization. 3rd
ed. New York: Springer; 2000. p. 100-32.
Press WH, Teukolsky SA, Vetterling WT, Flannery BP. Numerical Recipes in C++: The Art of Scientific Computing. 2nd
ed. Cambridge: Cambridge University Press; 2002.
Fippel M. Fast Monte Carlo dose calculation for photon beams based on the VMC electron algorithm. Med Phys 1999;26:1466-75.
Mackie TR, el-Khatib E, Battista J, Scrimger J, Van Dyk J, Cunningham JR, et al.
Lung dose corrections for 6-and 15-MVxrays. Med Phys 1985;12:327-32.
Mohan R, Chui C, Lidofsky L. Differential pencil beam dose computation model for photons. Med Phys 1986;13:64-73.
Boyer AL, Zhu YP, Wang L, Francois P. Fast Fourier transform convolution calculations of x-ray isodose distributions in homogeneous media. Med Phys 1989;16:248-53.
Eisbruch A, Foote RL, O'Sullivan B, Beitler JJ, Vikram B. Intensity-modulated radiation therapy for head and neck cancer: Emphasis on the selection and delineation of the targets. Semin Radiat Oncol 2002;12:238-49.
Alber M, Reemtsen R. Intensity modulated radiation therapy planning by use of a barrier-penalty multiplier method. Optim Methods Softw 2007;22:391-411.
International Commission on Radiation Units and Measurements. Prescribing, Recording, and Reporting Photon-Beam Intensity Modulation Radiation Therapy (IMRT). ICRU Report 83, Oxford, UK; 2010.
Niemierko A. Reporting and analyzing dose distributions: A concept of equivalent uniform dose. Med Phys 1997;24:103-10.
Rana S, Cheng C. Radiobiological impact of planning techniques for prostate cancer in terms of tumor control probability and normal tissue complication probability. Ann Med Health Sci Res 2014;4:167-72.
] [Full text]
Park JY, Lee JW, Chung JB, Choi KS, Kim YL, Park BM, et al
. Radiological model-based bio-anatomical quality assurance in intensity-modulated radiation therapy for prostate cancer. J Radiat Res 2012;53:978-85.
Emami B, Lyman J, Brown A, Coia L, Goitein M, Munzenrider JE, et al.
Tolerance of normal tissue to therapeutic irradiation. Int J Radiat Oncol Biol Phys 1991;21:109-22.
[Figure 1], [Figure 2]
[Table 1], [Table 2], [Table 3]