- Open Access
The impact of dose calculation algorithms on partial and whole breast radiation treatment plans
- Parminder S Basran†1, 2Email author,
- Sergei Zavgorodni†1, 2,
- Tanya Berrang†3, 4,
- Ivo A Olivotto†3, 4 and
- Wayne Beckham†1, 2
© Basran et al; licensee BioMed Central Ltd. 2010
- Received: 27 August 2010
- Accepted: 16 December 2010
- Published: 16 December 2010
This paper compares the calculated dose to target and normal tissues when using pencil beam (PBC), superposition/convolution (AAA) and Monte Carlo (MC) algorithms for whole breast (WBI) and accelerated partial breast irradiation (APBI) treatment plans.
Plans for 10 patients who met all dosimetry constraints on a prospective APBI protocol when using PBC calculations were recomputed with AAA and MC, keeping the monitor units and beam angles fixed. Similar calculations were performed for WBI plans on the same patients. Doses to target and normal tissue volumes were tested for significance using the paired Student's t-test.
For WBI plans the average dose to target volumes when using PBC calculations was not significantly different than AAA calculations, the average PBC dose to the ipsilateral breast was 10.5% higher than the AAA calculations and the average MC dose to the ipsilateral breast was 11.8% lower than the PBC calculations. For ABPI plans there were no differences in dose to the planning target volume, ipsilateral breast, heart, ipsilateral lung, or contra-lateral lung. Although not significant, the maximum PBC dose to the contra-lateral breast was 1.9% higher than AAA and the PBC dose to the clinical target volume was 2.1% higher than AAA. When WBI technique is switched to APBI, there was significant reduction in dose to the ipsilateral breast when using PBC, a significant reduction in dose to the ipsilateral lung when using AAA, and a significant reduction in dose to the ipsilateral breast and lung and contra-lateral lung when using MC.
There is very good agreement between PBC, AAA and MC for all target and most normal tissues when treating with APBI and WBI and most of the differences in doses to target and normal tissues are not clinically significant. However, a commonly used dosimetry constraint, as recommended by the ASTRO consensus document for APBI, that no point in the contra-lateral breast volume should receive >3% of the prescribed dose needs to be relaxed to >5%.
- Monte Carlo
- Planning Target Volume
- Contralateral Breast
- Accelerate Partial Breast Irradiation
- Ipsilateral Breast
For early stage breast cancer, whole breast irradiation (WBI) is used extensively to minimize the risk of ipsilateral breast cancer recurrence after breast conserving surgery. Over the last decade, there has been increased interest in the use of accelerated partial breast irradiation (APBI) as opposed to WBI . The use of APBI offers fewer fractions and lower dose to uninvolved regions of the breast. A number of clinical trials comparing WBI with various methods of APBI treatments are ongoing , however mature randomized data on the efficacy and toxicity of APBI compared to standard WBI will not be available for a number of years.
Publications supporting the dosimetric advantages of using APBI as an alternative to WBI have mainly focused on intra-cavitary brachytherapy, interstitial brachytherapy or intra-operative radiation therapy [3–6]. A common method of delivering APBI in ongoing randomized trials is linac-based, 3-dimensional conformal external beam radiation therapy (3DCRT) employing the same widely-available technology, staff, and treatment planning systems as WBI .
Given the potential importance of linear accelerator based delivery of APBI, the influence of dose calculation algorithms on trial eligibility and interpretation of risks to normal tissues is relevant. The impact of scatter corrections with WBI techniques comparing pencil beam convolution (PBC), the analytic anisotropic algorithm (AAA), and Monte Carlo (MC) calculations has been previously described , with several articles discussing the benefits of using AAA over PBC [9, 10]. However, there are no studies that examine the accuracy of the dose to target and normal tissues for 3DCRT APBI techniques. The accuracy of the calculated dose in regions well outside the irradiated volume is particularly important when trying to ascertain the risk of secondary cancer or normal tissue toxicity . Obtaining a better understanding of the potential increase, or decrease, in dose to target and normal tissues could facilitate a better understanding of the risks associated with APBI treatment strategies. This is a report of the consequences of changing dose calculation algorithms on doses to target volumes and important normal tissues during whole breast and partial breast irradiation.
We retrospectively examined plans for 10 consecutive patients enrolled in a prospective APBI trial who met all the dosimetry constraints of the protocol when using simplified pencil beam calculations . All plans were initially calculated with a pencil beam convolution (PBC) algorithm with Batho inhomogeneity corrections using the Eclipse Treatment Planning System (Version 8.617, Varian Medical Systems, Palo Alto, USA) . Plans were then recomputed (keeping the monitor units, beam weights and angles fixed) within Eclipse using AAA. All calculations were performed on 2.5 mm dose grid.
The WBI prescription was 42.5 Gy in 16 fractions, normalized to a point mid-plane in the breast tissue and to be delivered through a segmented MLC delivery with 6 MV photon beams.
The APBI prescription was 38.5 Gy in 10 fractions normalized to a point within the target volume. The planning guidelines for APBI patients follow those articulated in the American Society of Therapeutic Radiology and Oncology (ASTRO) consensus document .
Monte Carlo Verification
WBI and APBI treatment plans were recomputed with the Vancouver Island Monte Carlo (VIMC) system [15, 16]. The system provides a platform for Monte Carlo verification of the treatment plans generated by a TPS and exported in DICOM format.
The main "calculation engines" within the system are BEAMnrc for modelling particle fluence and DOSXYZnrc for modelling the dose deposition within the patient . The beam model for Varian 21EX treatment machine was used in this study. The model utilises a two-stage approach in calculating the dose where in the "first stage" all non-variable linac components are modelled and the particle fluence is stored in the phase space file. Then, in the "second stage" the phase space file is used in subsequent calculations as a radiation source for transporting the fluence through the patient phantom. Standard energy cut-off values were AP = PCUT = 0.01 MeV and AE = ECUT = 0.700 MeV, where AP and AE are the low energy thresholds for the production of secondary bremsstrahlung photons and knock-on electrons and PCUT and ECUT are the global cut-off energies for photon and electron transport used during electron and photon transport. In addition, "azimuthal particle redistribution" has been used to substantially reduce phase space latent variance [18, 19]. The model has been tuned and verified (except the build-up and penumbra regions) demonstrating dose agreement with the measured open field dose profiles within 1% for the field sizes within the range of 4 × 4 to 40 × 40 cm2. This excluded build-up and penumbra regions where the dose differences were higher, as expected, but still agreed to within 2% or within a 2 mm distance. Modelling of IMRT and RapidArc, as well as fixed-aperture fields' delivery has been performed with the dynamic multi-leaf collimators (dMLC) model by Siebers et al. and verified in our previous publications [11, 20, 21].
As most of the treatment fields used in the current study utilise the Varian implementation of collimator-controlled wedging, or enhanced dynamic wedges (EDWs), it is important that the dose from such fields is calculated correctly. Radiation transport through the moving jaw of EDWs is modelled in VIMC system using the method developed by Verhaegen and Liu . Each particle is transported through the dynamic jaw with its position sampled from a probability density function that describes jaw motion. Then, the particle is transported through the physical jaw in its sampled position. This method naturally models the radiation transmitted through the dynamic jaw towards the patient as well as radiation backscattered from the jaw into the linac monitor chamber. The latter is essential for correct absolute dose calculation implemented in the VIMC linac model . Verhaegen and Liu demonstrated excellent agreement of this EDW model with measured data. Our implementation of this model has been verified against the EDW commissioning measurements collected in our department. The measurements were done using Scanditronix Wellhofer CA24 ionization chamber array with IC-10 ionization chambers that have effective volume of 0.13 cm3. Examples of this verification for Monte Carlo as well as PBC and AAA calculations that include 10 × 10 and 20 × 20 cm2 fields with 60° wedge are shown in the Results section.
MC simulations of the treatment plans presented in this study were performed on 2.5 mm dose grid with less than 1% statistical uncertainty at the DEV.
Target and normal tissue dosimetric definitions and the average volumes for 10 patients in this study.
Target & Normal Tissue
Average Volume [cm3]
Planning Target Volume (PTV)
Relative volume covered by 95% of the prescription dose
Dose Evaluation Volume (DEV)
Relative volume covered by 95% of the prescription dose
Ipsilateral Breast (IPS-BR)
Relative volume covered by 95% of the prescription dose
Ipsilateral Lung (IPS-LUNG)
Relative volume receiving 10% of the prescription dose
Percent of prescription dose delivered to 10% of the volume
Contra-lateral lung (CON-LUNG)
Percent of prescription dose delivered to 5% of the volume
Contra-lateral breast (CON-BR)
Maximum point dose as a percent of the prescription dose
Verification of MC, AAA and PBC dose calculations for EDW fields
Dose Calculation Algorithm Effects on Whole Breast Irradiation
Mean, standard deviation and ranges of volumetric coverage and percent dose delivered to selected target and normal tissues as defined in Table 1 for three dose calculation algorithms during whole breast tangent radiation therapy.
Dose Calculation Algorithm Effects on Accelerated Partial Breast Irradiation
Mean, standard deviation and ranges of volumetric coverage and percent dose delivered to selected target and normal tissues as defined in Table 1 for three dose calculation algorithms during partial breast radiation therapy.
Accelerated Partial Breast versus Whole Breast Irradiation
Differences in percentage of volumetric coverage and percent dose delivered to selected target and normal tissues as defined in Table 1 when WBI plans are replanned with ABPI.
Dose PBC [%]
Dose AAA [%]
Dose MC [%]
This study demonstrates very good agreement between the AAA and PBC algorithms when planning either WBI or ABPI. This suggests that there are no major concerns associated with target and normal tissue coverage if switching from PBC to AAA for WBI or ABPI. Given that AAA provides a significant improvement over the PBC plus Batho-heterogeneity corrections in lung tissue, our clinical practice has migrated from PBC to AAA along with dose calculations for the APBI clinical trial.
For APBI plans, the dose to target and normal tissue volumes varied with the dose calculation algorithm. This result is in agreement with work that explored the impact of PBC, AAA, and MC algorithms in non-clinical scenarios . The volumes of the DEV and PTV receiving 95% of the prescription dose from PBC plans were higher or equal to the plans recomputed with AAA and MC. This is predictable because the lung-tissue interface is poorly calculated with PBC. If APBI plans are switched from PBC to AAA calculations, the dose to the PTV and DEV requires re-evaluation. Based on our results, a plan generated using AAA compared to PBC calculations would deliver approximately 2% more dose within the DEV. This may not have any measurable effect on tumour control but could influence the risk of late breast fibrosis because during APBI the dose per fraction is already high. This may be a particular risk if the DEV or PTV is large. The doses (and volumes receiving a specific dose) to normal structures will also correspondingly increase. Apart from the contra-lateral breast, the treatment plan can be re-configured to ensure that normal tissue constraints are maintained. This is not difficult to achieve since the doses to normal tissues are relatively independent of the calculation algorithm, with the important exception of the contra-lateral breast.
There may be a small but important difference in the contra-lateral breast dose when comparing APBI plans computed with PBC, AAA and MC algorithms. The dose to the contra-lateral breast was 2-3% higher with AAA as compared to MC. Despite the fact that dose calculation algorithms are not generally validated for dose points far away from the treatment volume and that this metric is sensitive and unstable, existing accelerated partial breast clinical trials use a maximum point dose as a constraint to the contra-lateral breast. The selection of this constraint stems from a desire to have simple planning objectives and constraints for dosimetrists. The ASTRO consensus document states that no point in the contra-lateral breast volume should receive > 3% of the prescribed dose. This work suggests that switching from the PBC to the AAA treatment planning algorithm could affect the apparent eligibility of patients for accelerated partial breast treatment. Out of ten patients in the current study, two would have failed the ASTRO contralateral breast dosimetry guideline when calculated using the PBC or MC algorithm. However, delivering an identical amount of MUs and using the same beam angles and weightings but calculated with AAA, seven patients would have not met the contra-lateral breast constraint. If reproduced across the population of patients considered for APBI, this could represent a significant reduction in eligibility. An examination of the DVH data for APBI plans suggests that relaxing the contra-lateral breast maximum dose constraint from 3% to 5% would retain eligibility for APBI without any real increase in the risk of radiation exposure or second breast cancer that is considered acceptable using existing PBC planning algorithms.
The APBI technique often employs wedges to achieve tumor coverage, hence the accuracy of the dose calculation to the contra-lateral breast can be largely affected by the algorithm's ability to correctly calculate the in-field and penumbra dose for the EDW fields. The AAA algorithm uses a semi-analytic model to account for leakage radiation, jaw and multi-leaf transmission for open and wedged fields and can over-estimate the dose in penumbra by 1-2% when compared with MC . In our centre, in-field open and wedged field agreement between measurement and calculations was better than 2% for AAA, and better than 1.5% for MC. This leads us to hypothesise that the dose differences in the contra-lateral breast are mostly due to head scatter and leakage modelling within AAA . These contributions are modelled as extra-focal and electron contamination parameters within the treatment planning system, which are optimized in the beam fitting procedure. In the fitting procedure, these extra-focal parameters cannot be distinguished from other parameters in the beam tuning, leading to excellent agreement in the open field and penumbra, but not necessarily far from the open beam.
There is very good agreement between PBC, AAA and MC for most tissues when treating with APBI. However, if calculation algorithms are switched from a simple pencil beam to a scatter-correction convolution/superposition algorithm, careful consideration should be given to tissues peripheral to the treated volume. In this study, it was found that a commonly used dosimetry constraint, as recommended by the ASTRO consensus document, that no point in the contra-lateral breast volume should receive >3% of the prescribed dose needs to be relaxed to >5%.
The authors would like to thank Michael Crane for his assistance with some of the planning of the patients in this study. The authors also greatly appreciate VIC Monte Carlo group and particularly Karl Bush for technical support of VIMC system used in this study.
- Smith BD, Arthur DW, Buchholz TA, Haffty BG, Hahn CA, Hardenbergh PH, Julian TB, Marks LB, Todor DA, Vicini FA, Whelan TJ, White J, Wo JY, Harris JR: Accelerated partial breast irradiation consensus statement from the American Society for Radiation Oncology (ASTRO). Int J Radiat Oncol Biol Phys 2009, 74: 987-1001. 10.1016/j.ijrobp.2009.02.031View ArticlePubMedGoogle Scholar
- RAPID: Randomized Trial of Accelerated Partial Breast Irradiation[http://clinicaltrials.gov/ct2/show/NCT00282035] --- Either ISSN or Journal title must be supplied.
- Kuerer HM, Julian TB, Strom EA, Lyerly HK, Giuliano AE, Mamounas EP, Vicini FA: Accelerated partial breast irradiation after conservative surgery for breast cancer. Ann Surg 2004, 239: 338-351. 10.1097/01.sla.0000114219.71899.13PubMed CentralView ArticlePubMedGoogle Scholar
- Scanderbeg D, Yashar C, White G, Rice R, Pawlicki T: Evaluation of three APBI techniques under NSABP B-39 guidelines. J Appl Clin Med Phys 2010, 11: 274-280.Google Scholar
- Stewart AJ, O'Farrell DA, Cormack RA, Hansen JL, Khan AJ, Mutyala S, Devlin PM: Dose volume histogram analysis of normal structures associated with accelerated partial breast irradiation delivered by high dose rate brachytherapy and comparison with whole breast external beam radiotherapy fields. Radiat Oncol 2008, 3: 39. 10.1186/1748-717X-3-39PubMed CentralView ArticlePubMedGoogle Scholar
- Bush DA, Slater JD, Garberoglio C, Yuh G, Hocko JM, Slater JM: A technique of partial breast irradiation utilizing proton beam radiotherapy: comparison with conformal x-ray therapy. Cancer J 2007, 13: 114-118. 10.1097/PPO.0b013e318046354bView ArticlePubMedGoogle Scholar
- Lomax AJ, Cella L, Weber D, Kurtz JM, Miralbell R: Potential role of intensity-modulated photons and protons in the treatment of the breast and regional nodes. Int J Radiat Oncol Biol Phys 2003, 55: 785-792. 10.1016/S0360-3016(02)04210-4View ArticlePubMedGoogle Scholar
- Baglan KL, Sharpe MB, Jaffray D, Frazier RC, Fayad J, Kestin LL, Remouchamps V, Martinez AA, Wong J, Vicini FA: Accelerated partial breast irradiation using 3 D conformal radiation therapy (3D-CRT). Int J Radiat Oncol Biol Phys 2003, 55: 302-311. 10.1016/S0360-3016(02)03811-7View ArticlePubMedGoogle Scholar
- Panettieri V, Barsoum P, Westermark M, Brualla L, Lax I: AAA and PBC calculation accuracy in the surface build-up region in tangential beam treatments. Phantom and breast case study with the Monte Carlo code PENELOPE. Radiother Oncol 2009, 93: 94-101. 10.1016/j.radonc.2009.05.010View ArticlePubMedGoogle Scholar
- Breitman K, Rathee S, Newcomb C, Murray B, Robinson D, Field C, Warkentin H, Connors S, Mackenzie M, Dunscombe P, Fallone G: Experimental validation of the Eclipse AAA algorithm. J Appl Clin Med Phys 2007, 8: 76-92.PubMedGoogle Scholar
- Gagné IM, Zavgorodni S: Evaluation of the analytical anisotropic algorithm in an extreme water-lung interface phantom using Monte Carlo dose calculations. J Appl Clin Med Phys 2007, 8: 33-46.Google Scholar
- Stathakis S, Li J, Ma CCM: Monte Carlo determination of radiation-induced cancer risks for prostate patients undergoing intensity-modulated radiation therapy. J Appl Clin Med Phys 2007, 8: 2685. 10.1120/jacmp.v8i4.2685View ArticlePubMedGoogle Scholar
- Storchi PR, van Battum LJ, Woudstra E: Calculation of a pencil beam kernel from measured photon beam data. Phys Med Biol 1994, 44: 2917-2928. 10.1088/0031-9155/44/12/305View ArticleGoogle Scholar
- Mitchell T, Truong PT, Salter L, Graham C, Gaffney H, Beckham W, Olivotto IA: A low-dose ipsilateral lung restriction improves 3-D conformal planning for partial breast radiation therapy. Med Dosim, in press.Google Scholar
- Zavgorodni S, Bush K, Locke C, Beckham W: Vancouver Island Monte Carlo (VIMC) system for radiotherapy treatment planning dosimetry and research. Radiother Oncol 2007,84(Supplement 1):S49.Google Scholar
- Bush K, Townson R, Zavgorodni S: Monte Carlo simulation of RapidArc radiotherapy delivery. Phys Med Biol 2008, N359-7.Google Scholar
- Walters BRB, Kawrakow I, Rogers DWO: History by history statistical estimators in the BEAM code system. Med Phys 2002, 29: 2745-2752. 10.1118/1.1517611View ArticlePubMedGoogle Scholar
- Bush K, Zavgorodni SF, Beckham WA: Azimuthal particle redistribution for the reduction of latent phase-space variance in Monte Carlo simulations. Phys Med Biol 2007, 52: 4345-4360. 10.1088/0031-9155/52/14/021View ArticlePubMedGoogle Scholar
- Sempau J, Sánchez-Reyes A, Salvat F, ben Tahar HO, Jiang SB, Fernández-Varea JM: Monte Carlo simulation of electron beams from an accelerator head using PENELOPE. Phys Med Biol 2001, 46: 1163-1186. 10.1088/0031-9155/46/4/318View ArticlePubMedGoogle Scholar
- Siebers JV, Keall PJ, Kim JO, Mohan R: A method for photon beam Monte Carlo multileaf collimator particle transport. Phys Med Biol 2002, 47: 3225-3249. 10.1088/0031-9155/47/17/312View ArticlePubMedGoogle Scholar
- Stapleton S, Zavgorodni S, Popescu IA, Beckham WA: Implementation of random set-up errors in Monte Carlo calculated dynamic IMRT treatment plans. Phys Med Biol 2005, 50: 429-439. 10.1088/0031-9155/50/3/003View ArticlePubMedGoogle Scholar
- Verhaegen F, Liu HH: Incorporating dynamic collimator motion in Monte Carlo simulations: an application in modelling a dynamic wedge. Phys Med Biol 2001, 46: 287-296. 10.1088/0031-9155/46/2/302View ArticlePubMedGoogle Scholar
- Popescu IA, Shaw CP, Zavgorodni SF, Beckham WA: Absolute dose calculations for Monte Carlo simulations of radiotherapy beams. Phys Med Biol 2005, 50: 3375-3392. 10.1088/0031-9155/50/14/013View ArticlePubMedGoogle Scholar
- Wackerly Dennis, Mendenhall William, Scheaffer RichardL: Mathematical Statistics With Applications. 6th edition. Duxbury Press Toronto; 2008.Google Scholar
- Tillikainen L, Siljamäki S, Helminen H, Alakuijala J, Pyyry J: Determination of parameters for a multiple-source model of megavoltage photon beams using optimization methods. Phys Med Biol 2007, 52: 1441-1467. 10.1088/0031-9155/52/5/015View ArticlePubMedGoogle Scholar
This article is published under license to BioMed Central Ltd. This is an Open Access article distributed under the terms of the Creative Commons Attribution License (<url>http://creativecommons.org/licenses/by/2.0</url>), which permits unrestricted use, distribution, and reproduction in any medium, provided the original work is properly cited.