Practical Radiation Oncology (2015) xx, xxx–xxx
www.practicalradonc.org
Original Report
Dosimetric impact of setup accuracy for an electron breast boost technique Scott Davidson PhD a,⁎, Steven Kirsner MS b , Bryan Mason MS b , Kelly Kisling MS b , Renee D. Barrett BS c , Anthony Bonetati BS c , Matthew T. Ballo MD d a
Department of Radiation Oncology, The University of Texas Medical Branch, Galveston, Texas Department of Radiation Physics, The University of Texas M. D. Anderson Cancer Center, Houston, Texas c Department of Radiation Oncology, The University of Texas M. D. Anderson Cancer Center, Houston, Texas d Department of Radiation Oncology, The University of Tennessee Health Science Center, Memphis, Tennessee b
Received 4 December 2014; revised 29 January 2015; accepted 21 February 2015
Abstract Purpose: To determine the setup error on an electron breast boost technique using daily cone beam computed tomography (CBCT). Patient and setup attributes were studied as contributing factors to the accuracy. Methods and materials: Reproducibility of a modified lateral decubitus position breast boost setup was verified for 33 patients using CBCT. Three-dimensional matching was performed between the CBCT and the initial planning CT for each boost fraction by matching the tumor bed and/or surgical clips. The dosimetric impact of the daily positioning error was achieved by rerunning the initial treatment plans incorporating the recorded shifts to study the dose differences. Breast compression, decubitus angle, tumor bed location and volume, and cup size were studied for their contribution to setup error. Results: The range of setup errors was: 1.5 cm anterior to 9 mm posterior, 1.3 cm superior to 2.3 cm inferior, and 3.2 cm medial to 2.4 cm lateral. Seven patients had setup errors that were ≥ 2-cm margin placed on the tumor bed and scar. Four of those 7 patients had unacceptable coverage as defined by the volume of the tumor bed plus scar that is covered by the 90% isodose line (V90) compared with the original plan. All other patients had no discernible difference in the coverage (V90). The use of compression, tumor bed location, or volumes N 20 mL showed no effect on coverage. Conclusions: In general, this study supported that a 2-cm margin was adequate (29 of 33 patients) when patients are treated under typical conditions. Care should be taken when high electron energies are selected because the coverage at depth is more difficult to maintain in the clinical environment. © 2015 American Society for Radiation Oncology. Published by Elsevier Inc. All rights reserved.
Introduction
Conflicts of interest: None. ⁎ Corresponding author. Scott Davidson, 301 University Blvd, Galveston, TX 77555. E-mail address: [email protected] (S. Davidson).
Postoperative radiation therapy for patients with breast cancer is traditionally delivered via tangential whole breast photon beams followed by an electron boost directed to the surgical cavity. The dimensions of the surgical cavity, defined either by surgical clips or surgical seroma, evolve
http://dx.doi.org/10.1016/j.prro.2015.02.011 1879-8500/© 2015 American Society for Radiation Oncology. Published by Elsevier Inc. All rights reserved.
2
S. Davidson et al
over a period of months following a lumpectomy. 1,2 These changes can continue through the 3- to 5-week course of tangential beam radiation and is inversely correlated with the length of time from lumpectomy. For this reason, many patients require replanning using computed tomography (CT) scanning for the electron boost. At our institution, we routinely use a modified lateral decubitus position for the breast boost in patients in which body habitus and tumor location precludes the use of the supine breast position. 3 This technique allows for freedom in patient rotation to optimize the patient anatomy for the boost delivery. In an effort to further reduce the electron energy (and thus the skin dose) and to improve the target volume coverage in larger patients, an in-house designed compression device is used. 4 This technique has also allowed a decreased distance between the skin and the target volume depth, improved electron coverage of the tumor bed, and reduced skin dose. Although much effort had been made to develop the modified lateral decubitus position technique and to use a compression device, the efficacy in terms of dose coverage had not been studied. In this study, cone beam CT (CBCT) was used to evaluate setup reproducibility and its impact on the dosimetry of the electron breast boost. In addition, any correlation between the dosimetric coverage and patient/tumor bed characteristics was investigated.
Methods and materials A total of 33 patients ranging from 33 to 77 years old were entered into an institutional protocol. Consenting postsegmental mastectomy patients diagnosed with invasive carcinoma of any type or ductal carcinoma in situ were selected. Patients having any contraindication for external beam radiation therapy, such as systemic lupus, scleroderma, or previous ipsilateral breast irradiation, or who had undergone modified radical or total mastectomy were excluded from the study. No other selection criteria were included. Patients were simulated for their electron breast boost using a GE LightSpeed scanner (GE Healthcare, Little Chalfont, UK). Patients were immobilized using a Vac-Lok bag (Civco, Orange City, IA) in the lateral decubitus position so that the surface of the breast in the location of the scar and tumor bed was en face. Care was taken to ensure that the Vac-Lok bag was formed around the patient to maintain a reproducible position. Alignment marks were placed on the patient from the simulation room lasers. In 25 of the 33 patients studied, an in-house custom compression device was used to reduce the depth of the distal tumor bed relative to the skin, thus allowing for a lower electron energy level. The daily reproducibility of the compressed device was indexed to the breast by skin markings as well as index markings made to the Vac-Lok
Practical Radiation Oncology: Month 2015
bag. The CT images were sent to the Pinnacle treatment planning system (Philips, Amsterdam, the Netherlands) for planning of the breast boost. In the treatment planning process, the tumor bed, surgical scar, and surgical clips were contoured and then combined to create the clinical tumor volume. A 2-cm margin was added to the combined structure to create the planning tumor volume extension. A custom electron cutout was then fabricated to define the beam portal. The margin was used to account for positioning variability, breathing motion, breast shape changes, penumbra, and microscopic disease. The electron energy was chosen to achieve tumor bed coverage at a minimum of 90% of the total dose (V90). Typical fractionation schemes were 5 to 7, with 1 patient receiving only 2 fractions. All fractions were prescribed to deliver either 200 or 250 cGy. For treatment, each patient was set to the marks established at simulation in her modified lateral decubitus position. A CBCT using the Varian linear accelerator’s on-board imager (Varian Medical Systems, Palo Alto, CA) was performed for each fraction for every patient to quantify the setup accuracy and its dosimetric impact. At the treatment console, the radiation oncologist then performed a 3-dimensional–3-dimensional (3D-3D) match between the planning CT data set and the CBCT data set while the patient was on the treatment table. The radiation oncologist used soft-tissue registration and relied on seroma in the tumor bed and/or surgical clips as landmarks on which to match the contoured anatomy. The table shifts in the lateral, longitudinal, and vertical directions were recorded for subsequent data analysis. The table shifts were transformed into the patient’s coordinate system by a shift of the isocenter for the dosimetric analysis. For the 33 patients entered into the study, the total number of fractions, and thus the total number of table shift recordings, was 167. A descriptive analysis was performed on all table shifts. The average and range of the shifts in all 3 directions for each patient were determined. The radial displacement, defined as the vector that resulted from the combined linear table translations in the lateral, longitudinal, and vertical directions within a single fraction, was calculated and expressed as an average, range, and maximum for each patient. A dosimetric analysis was performed using the treatment planning system. A composite treatment plan was constructed using the shifts obtained from the 3D-3D matching. A separate electron beam was created for each fraction and the inverse of the obtained shifts was applied to each isocenter for each beam. The plan was then recalculated to achieve a composite plan that represented the setup inaccuracies from the 3D-3D matching. The dose received by 100%, 95%, and 90% of the breast volume, V100, V95, and V90 were recorded for both the original plan and the recalculated composite plans. In this report, only the V90 was reported because this was the minimum coverage deemed acceptable.
Practical Radiation Oncology: Month 2015
Additionally, several factors were identified as having a potential effect on the setup error. The use of breast compression, the lateral decubitus angle position, tumor bed location, tumor bed volume, and breast cup size were recorded and analyzed for dependencies on the outcome of the setup accuracy study. The lateral decubitus angle was defined based on the position of the sternum and vertebrae relative the position of the sternum and vertebrae when in the supine position. Tumor bed location was divided into quadrants (ie, upper-outer, upper-inner, lower-inner, and lower-outer) and depth. The electron energy selected for treatment was used as a surrogate for depth.
Results Figure 1 shows the average and range of daily table shifts in all 3 directions for each of the 33 patients. The patients have been identified numerically to organize the data. Four patients (patients 1, 5, 22, and 27) had the largest reduction (≥ 10%) in dose coverage of V90 relative to the planned coverage. In addition, 3 other patients (patients 15, 25, and 28) had setup errors that were ≥ 2-cm margin placed on the tumor bed and scar; however, there was no appreciable change in dosimetric coverage. In the anteroposterior direction, the shifts ranged from 1.5 cm anteriorly to 0.9 cm posteriorly. The medial-lateral direction had the most setup variation. The shifts ranged
Dosimetric impact on breast boost setup
3
from 3.2 cm medial to 2.4 cm lateral. In the craniocaudal direction, the shifts ranged from 2.3 cm inferior to 1.3 cm superior. Figure 2 shows the magnitude, or radial displacement, of the shift data. Four of the 5 patients with the maximum radial displacement, which are the same 4 patients identified in Fig 1, had the appreciable loss in dose coverage (reduction in V90 of more than 10%). The dosimetric impact is presented in Fig 3. These data show the percent change in the dose coverage of V90 between the approved plan and the recalculated composite plan. The data included all 33 patients with and without compression. Four (patients 1, 5, 22, and 27) of the 33 patients studied show decreased coverage (V90) of at least 10% from the approved plan. The composite plans for patients 19, 20, and 23 had 5% to 8% increases in the V90 coverage compared with the approved plan. In 3 of the 25 patients who received breast compression, the dose coverage was suboptimal. The majority of patients treated (76%) had their tumor bed located in the upper-outer quadrant. No correlation was made between dose coverage and tumor bed location. In this study, the electron energy selected for each patient was used as a surrogate for depth. Most patients received either 9- (27%) or 12-MeV (52%) electrons for their boost. Of these 79%, 3 of the 4 patients with poor coverage were treated with these energies. The 20-MeV electrons were used to treat the fourth patient (patient 1) with poor coverage. The 4 patients with compromised coverage had small tumor bed volumes (less than 20 mL). Eight patients had tumor bed
Figure 1 Average and range of table shifts in the vertical (anteroposterior), longitudinal (superior-inferior), and lateral (medial-lateral) direction for 33 patients. The symbol x indicates those patients (patients 1, 5, 22, and 27) in which the V90 degraded by at least 10% based on the clinical setup.
4
S. Davidson et al
Practical Radiation Oncology: Month 2015
Figure 2 Average radial displacement and range of table shifts and maximum radial displacement table shifts for each patient during fractionated treatment. The data have been sorted by increasing radial displacement. The patient data that had at least a 10% reduction in dose coverage at V90 compared with the planned coverage are shown with the symbol x for the average radial displacement and with the symbol Δ for the maximum radial displacement shifts. Patient identification numbers are shown adjacent to the average and maximum shift data points for a select number of patients to relate to the data in the other figures.
volumes greater than 20 mL and as large as 140 mL with no degradation in dose coverage. The volume data for 2 patients were not available. The 4 patients with at least a 10% reduction in dose coverage resulting from setup variations had a breast cup size between C and DD.
Discussion Randomized studies have demonstrated that breast-conserving therapy consisting of partial mastectomy followed by whole breast radiation therapy provide equivalent disease-free
Figure 3 Dosimetric impact of the clinical setup. The data are expressed in terms of the percent change from the approved plan to the prospective plan as determined by the 3-dimensional–3-dimensional match from the daily cone beam computed tomography. Both compression and non-compression data is shown. Patient identification is along the x-axis. V90, volume of the tumor bed plus scar covered by the 90% isodose line.
Practical Radiation Oncology: Month 2015
and overall survival rates compared with mastectomy for early-stage (stages I and II) breast cancer. 5,6 In general, there is a two-thirds relative reduction in the rate of local failure with the addition of radiation therapy. Other randomized studies have demonstrated that a breast boost results in improved local control when compared with standard doses of radiation therapy. 7-9 The relative reduction in the rate of local failure is 20% to 50%. The clinical technique for designing the breast boost in which the radiation oncologist draws an electron field around a surgical scar is inadequate because the scar does not always correlate with the location of the underlying surgical excision site. 10 Subsequent studies have indicated that CT is the technique of choice for target delineation, whereas ultrasound frequently underestimates the true size of the cavity. 11,12 CT planning allows for improved target coverage while sparing uninvolved breast tissue. Research has shown that the traditional supine breast position for the electron breast boost is associated with an 8- to 11-mm setup uncertainty; this can easily be accounted for using an adequate margin around the clinical tumor volume. 13 At our institution, we have traditionally used a 2-cm margin around the clinical tumor volume to create the planning tumor volume. In this study, the dose coverage in 4 of the 33 patients was decreased by at least 10% because of setup error. One of these patients had the smallest volume tumor bed and received 20-MeV electrons and had extremely poor coverage, yet the table shifts were within the 2-cm margin used to create the planning tumor volume. Although the radial displacement was large and consistent with the group of patients that had poor coverage, it was determined that even a small change in a table shift would have resulted in poor coverage because of the higher energy electrons used to treat the small volume. Higher energy electron beams are more restrictive laterally at the 90% isodose level, and this condition is worse with decreasing field size. 14 As a result, there was not adequate margin at the tumor edges to allow for setup changes; therefore, this patient’s treatment did not resemble the typical electron breast boost. In retrospect, a different technique should have been used, such as reduced tangents for the boost. The table shifts in the lateral and craniocaudal directions exhibited the greatest amount of shift. Others have reported setup errors that were less than what has been reported here and with no correlation between setup error and the directional shift. 13,15 Topolnjak et al 13 reported using CBCT to register to the rib and sternum, whereas we registered to surgical clips and seroma. Fraser et al 15 reported the use of ultrasound for registration in a small number of patients. In both of those studies, the patients were positioned supine without compression and not in the modified lateral decubitus position. This technique may result in larger setup uncertainty; however, a 2-cm margin is still adequate for most patients. It was believed that the anteroposterior direction had the least amount of setup variation because of the use of compression. As expected, the patients that experienced the greatest amount of setup error
Dosimetric impact on breast boost setup
5
based on the CBCT 3D-3D matching tended to have worse coverage as determined by V90. However, in some patients, there were daily shifts beyond 2 cm (Fig 1), yet there was no dosimetric impact. This was due, in part, to the finite energy selection at which coverage at depth was more forgiving to setup error and to fractionation, whereas only 1 fraction was subject to large setup error such that the dose coverage remained adequate.
Conclusions It was not clear that any of the factors studied (ie, compression, decubitus angle, tumor bed size, location, depth, breast size), either individually or combined had an effect on the setup accuracy as determined by the V90 dose coverage. In general, this study showed that the 2-cm margin was adequate in most of the patients (29/33) treated under typical conditions, and that care should be taken when high electron energies are used because the coverage at depth is more difficult to maintain in the clinical environment.
References 1. Sharma R, Spierer M, Mutyala S, et al. Change in seroma volume during whole-breast radiation therapy. Int J Radiat Oncol Biol Phys. 2009;75:89-93. 2. Hepel JT, Evans SB, Hiatt JR, et al. Planning the breast boost: Comparison of three techniques and evolution of tumor bed during treatment. Int J Radiat Oncol Biol Phys. 2009;74:458-463. 3. Ludwig MS, McNeese MD, Buchholz TA, et al. The lateral decubitus breast boost: Description, rationale, and efficacy. Int J Radiat Oncol Biol Phys. 2010;76:100-103. 4. Schinkel CG, Strom EA, Johnson JL, et al. Use of a compression device to improve radiation dose coverage of the tumor bed for treatment of breast cancer. Int J Radiat Oncol Biol Phys. 2008;72:S187. 5. Fisher B, Anderson S, Bryant J, et al. Twenty-year follow-up of a randomized trial comparing total mastectomy, lumpectomy, and lumpectomy plus irradiation for the treatment of invasive breast cancer. N Engl J Med. 2002;347:1233-1241. 6. Veronesi U, Cascinelli N, Mariani L, et al. Twenty-year follow-up of a randomized study comparing breast-conserving surgery with radical mastectomy for early breast cancer. N Engl J Med. 2002;347:1227-1232. 7. Romestaing P, Lehingue Y, Carrie C, et al. Role of a 10-Gy boost in the conservative treatment of early breast cancer: Results of a randomized clinical trial in Lyon, France. J Clin Oncol. 1997;15:963-968. 8. Bartelink H, Horiot JC, Poortmans P, et al. Recurrence rates after treatment of breast cancer with standard radiotherapy with or without additional radiation. N Engl J Med. 2001;345:1378-1387. 9. Polgár C, Fodor J, Orosz Z, et al. Electron and high-dose-rate brachytherapy boost in the conservative treatment of stage I-II breast cancer first results of the randomized Budapest boost trial. Strahlenther Onkol. 2002;178:615-623. 10. Regine WF, Ayyangar KM, Komarnicky LT, et al. Computer-CT planning of the electron boost in definitive breast irradiation. Int J Radiat Oncol Biol Phys. 1991;20:121-125. 11. Benda RK, Yasuda G, Sethi A, et al. Breast boost: Are we missing the target? Cancer. 2003;97:905-909. 12. Rabinovitch R, Finlayson C, Pan Z, et al. Radiographic evaluation of surgical clips is better than ultrasound for defining the lumpectomy
6
S. Davidson et al
cavity in breast boost treatment planning: A prospective clinical study. Int J Radiat Oncol Biol Phys. 2000;47:313-317. 13. Topolnjak R, van Vliet-Vroegindeweij C, Sonke JJ, et al. Breastconserving therapy: Radiotherapy margins for breast tumor bed boost. Int J Radiat Oncol Biol Phys. 2008;72:941-948.
Practical Radiation Oncology: Month 2015 14. Kahn FM, Gibbons JP. The Physics of Radiation Therapy. 4th ed. Philadelphia: Lippincott Williams & Wilkins; 2010. 15. Fraser DJ, Wong P, Sultanem K, et al. Dosimetric evolution of the breast electron boost target using 3D ultrasound imaging. Radiother Oncol. 2010;96:185-191.
www.practicalradonc.org
Original Report
Dosimetric impact of setup accuracy for an electron breast boost technique Scott Davidson PhD a,⁎, Steven Kirsner MS b , Bryan Mason MS b , Kelly Kisling MS b , Renee D. Barrett BS c , Anthony Bonetati BS c , Matthew T. Ballo MD d a
Department of Radiation Oncology, The University of Texas Medical Branch, Galveston, Texas Department of Radiation Physics, The University of Texas M. D. Anderson Cancer Center, Houston, Texas c Department of Radiation Oncology, The University of Texas M. D. Anderson Cancer Center, Houston, Texas d Department of Radiation Oncology, The University of Tennessee Health Science Center, Memphis, Tennessee b
Received 4 December 2014; revised 29 January 2015; accepted 21 February 2015
Abstract Purpose: To determine the setup error on an electron breast boost technique using daily cone beam computed tomography (CBCT). Patient and setup attributes were studied as contributing factors to the accuracy. Methods and materials: Reproducibility of a modified lateral decubitus position breast boost setup was verified for 33 patients using CBCT. Three-dimensional matching was performed between the CBCT and the initial planning CT for each boost fraction by matching the tumor bed and/or surgical clips. The dosimetric impact of the daily positioning error was achieved by rerunning the initial treatment plans incorporating the recorded shifts to study the dose differences. Breast compression, decubitus angle, tumor bed location and volume, and cup size were studied for their contribution to setup error. Results: The range of setup errors was: 1.5 cm anterior to 9 mm posterior, 1.3 cm superior to 2.3 cm inferior, and 3.2 cm medial to 2.4 cm lateral. Seven patients had setup errors that were ≥ 2-cm margin placed on the tumor bed and scar. Four of those 7 patients had unacceptable coverage as defined by the volume of the tumor bed plus scar that is covered by the 90% isodose line (V90) compared with the original plan. All other patients had no discernible difference in the coverage (V90). The use of compression, tumor bed location, or volumes N 20 mL showed no effect on coverage. Conclusions: In general, this study supported that a 2-cm margin was adequate (29 of 33 patients) when patients are treated under typical conditions. Care should be taken when high electron energies are selected because the coverage at depth is more difficult to maintain in the clinical environment. © 2015 American Society for Radiation Oncology. Published by Elsevier Inc. All rights reserved.
Introduction
Conflicts of interest: None. ⁎ Corresponding author. Scott Davidson, 301 University Blvd, Galveston, TX 77555. E-mail address: [email protected] (S. Davidson).
Postoperative radiation therapy for patients with breast cancer is traditionally delivered via tangential whole breast photon beams followed by an electron boost directed to the surgical cavity. The dimensions of the surgical cavity, defined either by surgical clips or surgical seroma, evolve
http://dx.doi.org/10.1016/j.prro.2015.02.011 1879-8500/© 2015 American Society for Radiation Oncology. Published by Elsevier Inc. All rights reserved.
2
S. Davidson et al
over a period of months following a lumpectomy. 1,2 These changes can continue through the 3- to 5-week course of tangential beam radiation and is inversely correlated with the length of time from lumpectomy. For this reason, many patients require replanning using computed tomography (CT) scanning for the electron boost. At our institution, we routinely use a modified lateral decubitus position for the breast boost in patients in which body habitus and tumor location precludes the use of the supine breast position. 3 This technique allows for freedom in patient rotation to optimize the patient anatomy for the boost delivery. In an effort to further reduce the electron energy (and thus the skin dose) and to improve the target volume coverage in larger patients, an in-house designed compression device is used. 4 This technique has also allowed a decreased distance between the skin and the target volume depth, improved electron coverage of the tumor bed, and reduced skin dose. Although much effort had been made to develop the modified lateral decubitus position technique and to use a compression device, the efficacy in terms of dose coverage had not been studied. In this study, cone beam CT (CBCT) was used to evaluate setup reproducibility and its impact on the dosimetry of the electron breast boost. In addition, any correlation between the dosimetric coverage and patient/tumor bed characteristics was investigated.
Methods and materials A total of 33 patients ranging from 33 to 77 years old were entered into an institutional protocol. Consenting postsegmental mastectomy patients diagnosed with invasive carcinoma of any type or ductal carcinoma in situ were selected. Patients having any contraindication for external beam radiation therapy, such as systemic lupus, scleroderma, or previous ipsilateral breast irradiation, or who had undergone modified radical or total mastectomy were excluded from the study. No other selection criteria were included. Patients were simulated for their electron breast boost using a GE LightSpeed scanner (GE Healthcare, Little Chalfont, UK). Patients were immobilized using a Vac-Lok bag (Civco, Orange City, IA) in the lateral decubitus position so that the surface of the breast in the location of the scar and tumor bed was en face. Care was taken to ensure that the Vac-Lok bag was formed around the patient to maintain a reproducible position. Alignment marks were placed on the patient from the simulation room lasers. In 25 of the 33 patients studied, an in-house custom compression device was used to reduce the depth of the distal tumor bed relative to the skin, thus allowing for a lower electron energy level. The daily reproducibility of the compressed device was indexed to the breast by skin markings as well as index markings made to the Vac-Lok
Practical Radiation Oncology: Month 2015
bag. The CT images were sent to the Pinnacle treatment planning system (Philips, Amsterdam, the Netherlands) for planning of the breast boost. In the treatment planning process, the tumor bed, surgical scar, and surgical clips were contoured and then combined to create the clinical tumor volume. A 2-cm margin was added to the combined structure to create the planning tumor volume extension. A custom electron cutout was then fabricated to define the beam portal. The margin was used to account for positioning variability, breathing motion, breast shape changes, penumbra, and microscopic disease. The electron energy was chosen to achieve tumor bed coverage at a minimum of 90% of the total dose (V90). Typical fractionation schemes were 5 to 7, with 1 patient receiving only 2 fractions. All fractions were prescribed to deliver either 200 or 250 cGy. For treatment, each patient was set to the marks established at simulation in her modified lateral decubitus position. A CBCT using the Varian linear accelerator’s on-board imager (Varian Medical Systems, Palo Alto, CA) was performed for each fraction for every patient to quantify the setup accuracy and its dosimetric impact. At the treatment console, the radiation oncologist then performed a 3-dimensional–3-dimensional (3D-3D) match between the planning CT data set and the CBCT data set while the patient was on the treatment table. The radiation oncologist used soft-tissue registration and relied on seroma in the tumor bed and/or surgical clips as landmarks on which to match the contoured anatomy. The table shifts in the lateral, longitudinal, and vertical directions were recorded for subsequent data analysis. The table shifts were transformed into the patient’s coordinate system by a shift of the isocenter for the dosimetric analysis. For the 33 patients entered into the study, the total number of fractions, and thus the total number of table shift recordings, was 167. A descriptive analysis was performed on all table shifts. The average and range of the shifts in all 3 directions for each patient were determined. The radial displacement, defined as the vector that resulted from the combined linear table translations in the lateral, longitudinal, and vertical directions within a single fraction, was calculated and expressed as an average, range, and maximum for each patient. A dosimetric analysis was performed using the treatment planning system. A composite treatment plan was constructed using the shifts obtained from the 3D-3D matching. A separate electron beam was created for each fraction and the inverse of the obtained shifts was applied to each isocenter for each beam. The plan was then recalculated to achieve a composite plan that represented the setup inaccuracies from the 3D-3D matching. The dose received by 100%, 95%, and 90% of the breast volume, V100, V95, and V90 were recorded for both the original plan and the recalculated composite plans. In this report, only the V90 was reported because this was the minimum coverage deemed acceptable.
Practical Radiation Oncology: Month 2015
Additionally, several factors were identified as having a potential effect on the setup error. The use of breast compression, the lateral decubitus angle position, tumor bed location, tumor bed volume, and breast cup size were recorded and analyzed for dependencies on the outcome of the setup accuracy study. The lateral decubitus angle was defined based on the position of the sternum and vertebrae relative the position of the sternum and vertebrae when in the supine position. Tumor bed location was divided into quadrants (ie, upper-outer, upper-inner, lower-inner, and lower-outer) and depth. The electron energy selected for treatment was used as a surrogate for depth.
Results Figure 1 shows the average and range of daily table shifts in all 3 directions for each of the 33 patients. The patients have been identified numerically to organize the data. Four patients (patients 1, 5, 22, and 27) had the largest reduction (≥ 10%) in dose coverage of V90 relative to the planned coverage. In addition, 3 other patients (patients 15, 25, and 28) had setup errors that were ≥ 2-cm margin placed on the tumor bed and scar; however, there was no appreciable change in dosimetric coverage. In the anteroposterior direction, the shifts ranged from 1.5 cm anteriorly to 0.9 cm posteriorly. The medial-lateral direction had the most setup variation. The shifts ranged
Dosimetric impact on breast boost setup
3
from 3.2 cm medial to 2.4 cm lateral. In the craniocaudal direction, the shifts ranged from 2.3 cm inferior to 1.3 cm superior. Figure 2 shows the magnitude, or radial displacement, of the shift data. Four of the 5 patients with the maximum radial displacement, which are the same 4 patients identified in Fig 1, had the appreciable loss in dose coverage (reduction in V90 of more than 10%). The dosimetric impact is presented in Fig 3. These data show the percent change in the dose coverage of V90 between the approved plan and the recalculated composite plan. The data included all 33 patients with and without compression. Four (patients 1, 5, 22, and 27) of the 33 patients studied show decreased coverage (V90) of at least 10% from the approved plan. The composite plans for patients 19, 20, and 23 had 5% to 8% increases in the V90 coverage compared with the approved plan. In 3 of the 25 patients who received breast compression, the dose coverage was suboptimal. The majority of patients treated (76%) had their tumor bed located in the upper-outer quadrant. No correlation was made between dose coverage and tumor bed location. In this study, the electron energy selected for each patient was used as a surrogate for depth. Most patients received either 9- (27%) or 12-MeV (52%) electrons for their boost. Of these 79%, 3 of the 4 patients with poor coverage were treated with these energies. The 20-MeV electrons were used to treat the fourth patient (patient 1) with poor coverage. The 4 patients with compromised coverage had small tumor bed volumes (less than 20 mL). Eight patients had tumor bed
Figure 1 Average and range of table shifts in the vertical (anteroposterior), longitudinal (superior-inferior), and lateral (medial-lateral) direction for 33 patients. The symbol x indicates those patients (patients 1, 5, 22, and 27) in which the V90 degraded by at least 10% based on the clinical setup.
4
S. Davidson et al
Practical Radiation Oncology: Month 2015
Figure 2 Average radial displacement and range of table shifts and maximum radial displacement table shifts for each patient during fractionated treatment. The data have been sorted by increasing radial displacement. The patient data that had at least a 10% reduction in dose coverage at V90 compared with the planned coverage are shown with the symbol x for the average radial displacement and with the symbol Δ for the maximum radial displacement shifts. Patient identification numbers are shown adjacent to the average and maximum shift data points for a select number of patients to relate to the data in the other figures.
volumes greater than 20 mL and as large as 140 mL with no degradation in dose coverage. The volume data for 2 patients were not available. The 4 patients with at least a 10% reduction in dose coverage resulting from setup variations had a breast cup size between C and DD.
Discussion Randomized studies have demonstrated that breast-conserving therapy consisting of partial mastectomy followed by whole breast radiation therapy provide equivalent disease-free
Figure 3 Dosimetric impact of the clinical setup. The data are expressed in terms of the percent change from the approved plan to the prospective plan as determined by the 3-dimensional–3-dimensional match from the daily cone beam computed tomography. Both compression and non-compression data is shown. Patient identification is along the x-axis. V90, volume of the tumor bed plus scar covered by the 90% isodose line.
Practical Radiation Oncology: Month 2015
and overall survival rates compared with mastectomy for early-stage (stages I and II) breast cancer. 5,6 In general, there is a two-thirds relative reduction in the rate of local failure with the addition of radiation therapy. Other randomized studies have demonstrated that a breast boost results in improved local control when compared with standard doses of radiation therapy. 7-9 The relative reduction in the rate of local failure is 20% to 50%. The clinical technique for designing the breast boost in which the radiation oncologist draws an electron field around a surgical scar is inadequate because the scar does not always correlate with the location of the underlying surgical excision site. 10 Subsequent studies have indicated that CT is the technique of choice for target delineation, whereas ultrasound frequently underestimates the true size of the cavity. 11,12 CT planning allows for improved target coverage while sparing uninvolved breast tissue. Research has shown that the traditional supine breast position for the electron breast boost is associated with an 8- to 11-mm setup uncertainty; this can easily be accounted for using an adequate margin around the clinical tumor volume. 13 At our institution, we have traditionally used a 2-cm margin around the clinical tumor volume to create the planning tumor volume. In this study, the dose coverage in 4 of the 33 patients was decreased by at least 10% because of setup error. One of these patients had the smallest volume tumor bed and received 20-MeV electrons and had extremely poor coverage, yet the table shifts were within the 2-cm margin used to create the planning tumor volume. Although the radial displacement was large and consistent with the group of patients that had poor coverage, it was determined that even a small change in a table shift would have resulted in poor coverage because of the higher energy electrons used to treat the small volume. Higher energy electron beams are more restrictive laterally at the 90% isodose level, and this condition is worse with decreasing field size. 14 As a result, there was not adequate margin at the tumor edges to allow for setup changes; therefore, this patient’s treatment did not resemble the typical electron breast boost. In retrospect, a different technique should have been used, such as reduced tangents for the boost. The table shifts in the lateral and craniocaudal directions exhibited the greatest amount of shift. Others have reported setup errors that were less than what has been reported here and with no correlation between setup error and the directional shift. 13,15 Topolnjak et al 13 reported using CBCT to register to the rib and sternum, whereas we registered to surgical clips and seroma. Fraser et al 15 reported the use of ultrasound for registration in a small number of patients. In both of those studies, the patients were positioned supine without compression and not in the modified lateral decubitus position. This technique may result in larger setup uncertainty; however, a 2-cm margin is still adequate for most patients. It was believed that the anteroposterior direction had the least amount of setup variation because of the use of compression. As expected, the patients that experienced the greatest amount of setup error
Dosimetric impact on breast boost setup
5
based on the CBCT 3D-3D matching tended to have worse coverage as determined by V90. However, in some patients, there were daily shifts beyond 2 cm (Fig 1), yet there was no dosimetric impact. This was due, in part, to the finite energy selection at which coverage at depth was more forgiving to setup error and to fractionation, whereas only 1 fraction was subject to large setup error such that the dose coverage remained adequate.
Conclusions It was not clear that any of the factors studied (ie, compression, decubitus angle, tumor bed size, location, depth, breast size), either individually or combined had an effect on the setup accuracy as determined by the V90 dose coverage. In general, this study showed that the 2-cm margin was adequate in most of the patients (29/33) treated under typical conditions, and that care should be taken when high electron energies are used because the coverage at depth is more difficult to maintain in the clinical environment.
References 1. Sharma R, Spierer M, Mutyala S, et al. Change in seroma volume during whole-breast radiation therapy. Int J Radiat Oncol Biol Phys. 2009;75:89-93. 2. Hepel JT, Evans SB, Hiatt JR, et al. Planning the breast boost: Comparison of three techniques and evolution of tumor bed during treatment. Int J Radiat Oncol Biol Phys. 2009;74:458-463. 3. Ludwig MS, McNeese MD, Buchholz TA, et al. The lateral decubitus breast boost: Description, rationale, and efficacy. Int J Radiat Oncol Biol Phys. 2010;76:100-103. 4. Schinkel CG, Strom EA, Johnson JL, et al. Use of a compression device to improve radiation dose coverage of the tumor bed for treatment of breast cancer. Int J Radiat Oncol Biol Phys. 2008;72:S187. 5. Fisher B, Anderson S, Bryant J, et al. Twenty-year follow-up of a randomized trial comparing total mastectomy, lumpectomy, and lumpectomy plus irradiation for the treatment of invasive breast cancer. N Engl J Med. 2002;347:1233-1241. 6. Veronesi U, Cascinelli N, Mariani L, et al. Twenty-year follow-up of a randomized study comparing breast-conserving surgery with radical mastectomy for early breast cancer. N Engl J Med. 2002;347:1227-1232. 7. Romestaing P, Lehingue Y, Carrie C, et al. Role of a 10-Gy boost in the conservative treatment of early breast cancer: Results of a randomized clinical trial in Lyon, France. J Clin Oncol. 1997;15:963-968. 8. Bartelink H, Horiot JC, Poortmans P, et al. Recurrence rates after treatment of breast cancer with standard radiotherapy with or without additional radiation. N Engl J Med. 2001;345:1378-1387. 9. Polgár C, Fodor J, Orosz Z, et al. Electron and high-dose-rate brachytherapy boost in the conservative treatment of stage I-II breast cancer first results of the randomized Budapest boost trial. Strahlenther Onkol. 2002;178:615-623. 10. Regine WF, Ayyangar KM, Komarnicky LT, et al. Computer-CT planning of the electron boost in definitive breast irradiation. Int J Radiat Oncol Biol Phys. 1991;20:121-125. 11. Benda RK, Yasuda G, Sethi A, et al. Breast boost: Are we missing the target? Cancer. 2003;97:905-909. 12. Rabinovitch R, Finlayson C, Pan Z, et al. Radiographic evaluation of surgical clips is better than ultrasound for defining the lumpectomy
6
S. Davidson et al
cavity in breast boost treatment planning: A prospective clinical study. Int J Radiat Oncol Biol Phys. 2000;47:313-317. 13. Topolnjak R, van Vliet-Vroegindeweij C, Sonke JJ, et al. Breastconserving therapy: Radiotherapy margins for breast tumor bed boost. Int J Radiat Oncol Biol Phys. 2008;72:941-948.
Practical Radiation Oncology: Month 2015 14. Kahn FM, Gibbons JP. The Physics of Radiation Therapy. 4th ed. Philadelphia: Lippincott Williams & Wilkins; 2010. 15. Fraser DJ, Wong P, Sultanem K, et al. Dosimetric evolution of the breast electron boost target using 3D ultrasound imaging. Radiother Oncol. 2010;96:185-191.
