Optimal selection of optimization bolus thickness in planning of VMAT breast radiotherapy treatments

ARTICLE IN PRESS

JID: MDO

[mUS5Gb;October 30, 2018;13:30]

Medical Dosimetry xxx (xxxx) xxx

Contents lists available at ScienceDirect

Medical Dosimetry journal homepage: www.elsevier.com/locate/meddos

Optimal selection of optimization bolus thickness in planning of VMAT breast radiotherapy treatments Maija Rossi, PhD a,b,∗, Eeva Boman, PhD a,b, Mika Kapanen, PhD a,b a b

Department of Medical Physics, Medical Imaging Centre, Tampere University Hospital, 33521 Tampere, Finland Department of Oncology, Tampere University Hospital, 33521 Tampere, Finland

a r t i c l e

i n f o

Article history: Received 1 June 2018 Revised 28 August 2018 Accepted 2 October 2018 Available online xxx Keywords: VMAT Breast cancer Skin flash Optimization bolus

a b s t r a c t The aim of this study was to find an optimal optimization skin flash thickness in volumetric modulated arc radiotherapy of the breast in consideration of soft tissue deformations during the treatment course. Ten breast radiotherapy patients with axillary lymph node inclusion were retrospectively planned with volumetric modulated arc radiotherapy technique. The plans were optimized with the planning target volume (PTV) extending outside the skin contour by 0, 5, 7, and 10 mm; and with optimization boluses of 3 or 5 mm on the extended PTV. The final dose was calculated without the bolus. The plans were compared in terms of PTV homogeneity and conformity, and dose minima and maxima. The doses to organs at risk were also evaluated. The doses were recalculated in real patient geometries based on cone beam computed tomography (CBCT) images captured 3 to 6 times during each patient’s treatment course. The optimization to the PTV without the PTV extension resulted in the best CTV coverage in the original plans (V95% = 98.0% ± 1.2%). However, when these plans were studied in real CBCT-based patient geometries, the CTV V95% was compromised (94.6% ± 8.3%). In addition, for the surface (4 mm slap inside the PTV 4 mm below the body contour) dose V95% was reduced from the planned 74.7% ± 7.5% to the recalculated 65.5% ± 11.5%. Optimization with an 8-mm bolus to a PTV with 5-mm extension was the most robust choice to ensure the CTV and surface dose coverage (recalculated V95% was 95.2% ± 6.4% and 74.6% ± 8.4%, respectively). In cases with the largest observed deformations, even a 10-mm PTV extension did not suffice to cover the target. Optimization with a 5-mm PTV extension and an 8-mm optimization bolus improved the surface dose and slightly improved the CTV dose when compared to no extension plans. For deformations over 1 cm, no benefit was seen in PTV extensions and replanning is recommended. Frequent tangential and CBCT imaging should be used during treatment course to detect potential large anatomical changes. © 2018 American Association of Medical Dosimetrists. Published by Elsevier Inc. All rights reserved.

Introduction Adjuvant radiotherapy (RT) is recommended for breast cancer patients undergoing breast-conserving surgery or node-positive (N+) mastectomy.1 The use of volumetric modulated arc therapy (VMAT) technique for breast cancer RT has become as standard practice replacing at least partly the traditional half blocked field treatments in many clinics.2-9 The use of VMAT technique has raised slight concern on the accuracy of treatment delivery due to inter- and intrafractional motion during treatment that is highly modulated compared to

∗ Reprint requests to Maija Rossi, Medical Imaging Centre, Department of Physics, Tampere University Hospital, 33521 Tampere, Finland. E-mail addresses: maija.rossi@pshp.fi (M. Rossi), eeva.boman@pshp.fi (E. Boman), mika.kapanen@pshp.fi (M. Kapanen).

static fields with large skin flash.10-12 Furthermore, the perpendicular beam entrance directions in the VMAT technique may decrease the skin dose compared to tangential beams.8-9 This is because the amount of scattered dose decreases at the surface due to lower irradiation contribution from tangential beam entrance directions. There are no recommendations for sufficient skin dose in breast RT, but a surface bolus (i.e., water equivalent build up material) is recommended for mastectomy patients above the surgical scar.13 As the skin dose of VMAT treatments is already slightly lower than that of FinF treatments, it should not be allowed to decrease considerably from the planned dose if the anatomy changes— especially for mastectomy patients for whom the chest wall is the most frequent site of local treatment failure.13 The dose distribution may be also affected by tissue deformations, such as swelling or postural changes during the treatment course.14,15 This may result in dose maxima and minima on the

https://doi.org/10.1016/j.meddos.2018.10.001 0958-3947/© 2018 American Association of Medical Dosimetrists. Published by Elsevier Inc. All rights reserved.

JID: MDO 2

ARTICLE IN PRESS

[mUS5Gb;October 30, 2018;13:30]

M. Rossi et al. / Medical Dosimetry 000 (2018) 1–8

skin or in deeper structures, depending on the arc geometry and the size and location of the deformation. During radiotherapy, the proportion of patients having edema has been reported to rise from 8% to 20%.16 Volume changes were more than 8% in half of the patients in a recent study from the beginning to the middle of the radiotherapy course.14 Smaller changes with median of 2 mm have been found based on tangential images; however having range between −5 and 27 mm, requiring VMAT replanning for 5/24 (20%) patients.15 Another study found 7% to 17% changes in tangential images in half of patients during the radiotherapy course.17 To minimize the dosimetric effect of tissue deformations, an automatic skin flash tool can be used in Monaco treatment planning system (Elekta AB, Stockholm, Sweden),6 or by using an optimization bolus in Varian treatment planning system and extending the target volume outside the skin contour (External Beam Planning, Varian Medical Systems, Palo Alto).5,7 , 15 The size of the optimization bolus has varied mainly between 0.5 and 1 cm.5,7 Thicker bolus is assumed to allow for more swelling. The increasing bolus thickness however increases the dose maxima in the final dose distribution which is calculated without the bolus. The optimal bolus thickness has not been studied earlier in terms of plan quality and actual tissue deformations. The aim of this study is to find an optimal bolus thickness and planning target volume (PTV) extension outside the skin that allows for deformations as large as possible without considerable loss of plan quality, based on actual skin surface changes detected with the cone beam computed tomography (CBCT) scans recorded during the patient’s treatment course. Materials and Methods This retrospective study included 10 breast cancer patients with axillary lymph node invasion who had been treated earlier with VMAT and at least 3 CBCT scans were available. The median age was 69 years, with range from 49 to 78 years. The patients included 5 right-sided (3 mastectomies, 2 conserving surgeries, 3 deep inspiration breath hold [DIBH], and 2 free breathing (FB) treatments) and 5 left-sided (4 mastectomies, 1 conserving surgery, 2 DIBH, and 3 FB treatments) cases. For all patients the axillary lymph nodes were included in the PTV. Treatment planning The patients were imaged with computed tomography (CT) using 3-mm slice thickness (Philips Brilliance Big Bore, Philips Medical Systems, Netherlands or Toshiba Aquilion LB, Toshiba Medical System, Japan). The plans were generated for Varian’s linear accelerator Clinac iX (Varian Medical systems). The machine was equipped with Millennium 120 multileaf collimator. For the optimization, photon optimizer algorithm v13.6.23 was used. To ensure consistency, all optimization parameters were kept constant for a given patient and all optimizations were undertaken by a single physicist. The Analytic Anisotropic Algorithm v13.6.23 was used for dose calculation using the grid size of 0.25 cm. The VMAT planning included 4 partial arcs as described by Rossi et al.9 Two lateral and 1 medial tangential arcs were used with collimator angles and jaw restrictions that best avoided the heart and lung. A longer frontal partial arc with gantry angles from 300° to 60° was aimed at the supraclavicular lymph nodes. The lower collimator jaw was limited to exclude the heart from the treatment field, and this arc was designed to cover the axillary lymph node volumes without going tangentially through the humeral head.9 The same arc design and same optimization criteria were used for a given patient in 6 comparing plans. The plans were thus oth-

Fig. 1. Contours of the original CT V (orange), PT V+0 (red), PTV−5 (yellow), PTV+5 (green), PTV+7 (blue), and PTV+10 (light blue) are shown in A and B. The cranial ending of PTVb/c is shown in pink in B. The contours of PTV+0, PTV+5, PTV+7, PT V+10, and PT Vb/c are posteriorly similar to PTV−5. In (C), the surface slabs PT V0-4 (magenta), PT V4-8 (blue), and PT V8-12 (green) are shown inside the red PTV+0, and marked with arrows. (Color version of figure is available online.)

erwise similar but created with 6 different PTV extension and optimization bolus combinations. The original PTV was drawn with a 5-mm margin to the clinical target volume (CTV), and cropped to the skin contour (PTV+0). With the purpose of creating a better skin flash, the PTV was then extended 5, 7, and 10 mm to the air (PT V+5, PT V+7, and PT V+10, respectively). The PT V+0, PT V+5, and PTV+7 were optimized with optimization boluses of 5, 10, and 12 mm, respectively, thus the bolus extending 5 mm further than the target volume. In addition, the PTV+5, PTV+7, and PTV+10 were optimized with optimization boluses of 8, 10, and 13 mm, respectively, thus reaching 3 mm further than the planning PTV. The CT V and PT Vs are presented in Fig. 1. The plans were named as Plan_X_Y, with X referring to the extension of the PTV outside the skin contour, and Y referring to the thickness of the optimization bolus in mm.

JID: MDO

ARTICLE IN PRESS M. Rossi et al. / Medical Dosimetry 000 (2018) 1–8

[mUS5Gb;October 30, 2018;13:30] 3

The optimization was computed with the optimization bolus, with 1 intermediate dose calculation during the optimization. The final dose calculation without the bolus resulted typically in higher dose maxima in the final plan when compared to the plan which was calculated with the optimization bolus. These maxima may appear in different locations than they would if the bolus was included. For these reasons a copy of each plan was calculated without the bolus, and a volume of the maximum dose (D1%) was segmented. The volume and the location of the D1% depended on patient anatomy, PTV extension, and bolus thickness. The optimization was run again on the original plan with the optimization bolus, now restricting the dose in D1% to reduce the dose maxima in the final plan. The final plan was calculated without the bolus. Dose was normalized to 100% mean dose in PTV−5, which was the original PTV cropped 5 mm from the body contour (Fig. 1A and B). The mastectomy patients had a 3-mm bolus on the scar area during their treatment course, but for simplicity this was not included in the calculations. Plan analysis The plans were evaluated for the dose homogeneity (HI) and Paddick dose conformity (CI). The CI was calculated as: CI =

V95(PTV )(cc ) V95(PTV )(cc ) , V(PTV )(cc ) V95(cc )

(1)

where V95(PTV)(cc) and V95(cc) were the PTV and whole-body volumes, respectively, that received at least 95% of the prescribed dose, and V(PTV)(cc) is the PTV volume. Higher values indicate better conformity. The HI was calculated as: HI =

D2%(PTV ) − D98%(PTV ) , D50(PTV )

(2)

where DX%(PTV)(Gy) was the dose to X% of the PTV volume, and D50(PTV)(Gy) is the PTV median dose. Lower HI values indicate the better homogeneity. The PTV minima and maxima were analyzed using V90% (%), V95% (%), V105% (%), and V108% (%). The surface doses were evaluated with surface structures (slaps) inside the PTV: PTV0-4mm, PT V4-8mm, and PT V8-12mm, in which the numbers represented the outer and inner borders of those slaps by the distances from the skin (Fig. 1C). The PTV at the breast or chest wall (PTVb/c) was analyzed separately, excluding the axillary lymph node volumes from the original PTV. The PTVb/c ended cranially to the CT slice where the PTV no longer touched the skin, and was cropped 5 mm inside from the skin contour for analysis (Fig. 1B). Doses to organs at risk were analyzed for the heart, lungs, humeral head, and the contralateral breast. Due to the small number of patients, the statistical analyses were calculated with Wilcoxon signed ranks test, and the limit of p < 0.05 was used for statistical significance. The combinations of PTV+7 with 10-mm bolus, PTV+5 mm with 8-mm bolus and PTV with 5-mm bolus were selected for further analysis using real patient data on tissue deformations. Each CBCT was registered to the original CT with automatic registration emphasizing the region of the PTV. No dose calculations were performed on the CBCT images. Instead, to take into account the tissue deformation seen in CBCT images, the body contour of the original CT was modified accordingly. This was done by creating new outside- and inside-structures to the original CT. The structures were created using Boolean operators on the body contours of the CBCT and original CT. The inside structure consisted of swollen tissue or tissue moved to the location from elsewhere, and was assigned the Hounsfield unit of −100 as in fat. The outside structure consisted of volumes of tissue shrinkage or where tissue had moved to another location, and was assigned the Hounsfield unit of −10 0 0 as in air. The original body contour was modified to include the inside structure and to exclude the outside structure. The

Fig. 2. (A) The CT-based body contour (green) is modified to correspond to the CBCT-based contour (magenta). The HU values are replaced with -100 HU as in fat or -10 0 0 HU as in air, where the body contour has changed. PTV−5 (yellow) is cropped 5 mm from the modified body contour. (B) The dose distribution of Plan_5_8 and (C) the recalculated dose distribution in the modified structure set. (Color version of figure is available online.)

original plans were recalculated to this structure set using the original monitor units. All dose calculations were performed on the skin-contour-modified CT images instead of CBCT images. The contours are shown in Fig. 2A. Results Optimization plans The HI and CI are shown in Fig. 3A. A continuous decrease in plan quality was found with increasing PTV extension and bolus thickness. Both HI and CI were best in Plan_0_5, and worst in Plan_10_13. In Plan_5_10 and 7_10 they were closely equal (p > 0.06), and slightly better than in Plan_7_12 (p < 0.04) and Plan_10_13 (p < 0.01). In PTV dose minima (V95% and V90%, Fig. 3B) and maxima (V105% and V108%, Fig. 3C), the same trend continued, with Plan_0_5 showing best dose coverage and smallest

JID: MDO 4

ARTICLE IN PRESS

[mUS5Gb;October 30, 2018;13:30]

M. Rossi et al. / Medical Dosimetry 000 (2018) 1–8

Fig. 3. Planned dose distribution presented with homogeneity index (HI) and conformity index (CI) (A), PTV minima (B) and maxima (C) using different optimization methods X_Y, where X is the extension of the PTV to the air and Y is the bolus thickness in mm.

dose maxima, and Plan_10_13 the least dose coverage and highest dose maxima. The minima and maxima in PT V−5, PT Vb/c, and CTV were in favor of 3-mm bolus over 5-mm bolus above the optimization structure with optimization to both PTV+5 and PTV+7 for which 2 different bolus thicknesses were used. The final dose calculation and normalization to PTV−5 mean without the optimization bolus decreased the organs at risk (OAR) doses slightly as the removal of the optimization bolus decreased the monitor units needed. However, this dose decrease was similar among the plans and did not depend on the bolus thickness. Although statistically significant differences for the lungs were found between the plans generated with different thicknesses, the magnitude of the differences was negligible. Recalculated plans The tissue deformations in the CBCT data showed that the maximum change was at median 5 mm, with a range from 2 to 33 mm. The largest deformation was in all cases directed outside the body, as in swelling. Maximum regional tissue loss was 11 mm. The tissue losses seemed to be related to tissue deformations

as in Fig. 2, not to total volume changes (i.e., caused by weight loss). Instead, tissue gains were related to deformation, swelling, or seroma. Three plans (Plan_0_5, Plan_5_8, and Plan_7_10) were recalculated in the CBCT-based geometries for each patient. The dosimetric parameters and comparisons to original plan values are presented in Fig. 4. The changes in patient geometries from the planning CT to the treatment caused slightly decreased homogeneity and conformity. The recalculations showed larger PTV dose minima than planned. Improvements were gained as decreasing dose maxima. The actual dose coverage in PTV−5 mm decreased with increasing optimization structure and bolus thickness in the recalculated dose distributions as predicted by the original plans. However, in PTVb/c the actual coverages were closely similar despite slight differences between the original plans (Fig. 4B), and for the actual coverage of CTV, Plan_5_8 was slightly better than Plan_0_5 (pV95 = 0.218, pV90 = 0.004), unlike in the original plans. The decrease in dose maxima from plan to recalculated plan was similar for all 3 bolus thicknesses. The doses near the surface were the best in Plan_5_8 for both the original plans and the recalculated dose distributions (Fig. 5).

JID: MDO

ARTICLE IN PRESS M. Rossi et al. / Medical Dosimetry 000 (2018) 1–8

[mUS5Gb;October 30, 2018;13:30] 5

Fig. 4. The PTV coverage in actual patient geometry based on CBCT imaging. Homogeneity index (HI) and conformity index (CI) (A), PTV minima (B) and maxima (D). In (C) the PTV minima are presented without the four outliers with deformations larger than 10 mm. For comparison, the corresponding planned doses (same as in Fig. 3) are shown.

JID: MDO 6

ARTICLE IN PRESS

[mUS5Gb;October 30, 2018;13:30]

M. Rossi et al. / Medical Dosimetry 000 (2018) 1–8

Fig. 5. Doses near the surface for different optimization thicknesses in the original plans and recalculated plans.

Fig. 6. CBCT-based deformation as a minor displacement medially (A) and as major seroma (B). In C,D the planned dose profiles are shown for Plan_0_5, Plan_5_8, Plan_10_7, and for the seroma case Plan_13_10. The dotted lines show the recalculations, where in the case of minor deformation the planned dose is nicely shifted for all bolus thicknesses (C); but with the major deformation the dose level changes considerably, depending on the bolus thickness (D).

Figure 4C presents the dose parameters V95 and V90 when the large (>1cm) soft tissue deformations are excluded from the analyses. In this case, the changes in dose distribution were small between the plan and CBCT-based recalculation. Figure 2B and C shows a dose distribution in one CT-slice in the original and with CBCT-based recalculation in the typical case of minor soft tissue deformation. The planned dose was shifted to another location in case of minor tissue deformations and filled the PTV−5 in the new location, independent of the optimization structure and bolus thickness in use (Fig. 6A, C).

The effect of deformation was calculated as changes in HI, CI, and PTV dose minima and maxima from plan to recalculated plan (subtraction) for each bolus thickness. The amount of tissue swelling in cm (maximum difference seen in the original body contour to the CBCT body contour) correlated with these changes (Table 1). Larger tissue swelling led to increasing volumes of dose minima, but the effect on V90% was smaller for Plan_5_8 than for Plan_0_5 or Plan_7_10. Also the dose maxima were decreased with larger swelling but this correlation was mainly not significant.

ARTICLE IN PRESS

JID: MDO

[mUS5Gb;October 30, 2018;13:30]

M. Rossi et al. / Medical Dosimetry 000 (2018) 1–8 Table 1 Correlation coefficients between tissue swelling and dose parameter changes from plan to recalculated plan using the CBCT images (Plan_X_Y – recalculated Plan_X_Y)

HI(PTV − 5 mm) HI(PTVb/c) HI(CTV) CI(PTV − 5 mm) V95(PTV − 5 mm) V95(PTVb/c) V95(CTV) V90(PTV − 5 mm) V90(PTVb/c) V90(CTV) V105(PTV − 5 mm) V105(PTVb/c) V105(CTV) V108(PTV − 5 mm) V108(PTVb/c) V108(CTV) ∗ ∗∗

Plan_0_5

Plan_5_8

Plan_7_10

0.584∗∗ 0.535∗∗ 0.596∗∗ −0.529∗∗ −0.549∗∗ −0.566∗∗ −0.550∗∗ −0.575∗∗ −0.549∗∗ −0.603∗∗ −0.220 −0.190 −0.186 0.195 0.254 0.451∗∗

0.605∗∗ 0.563∗∗ 0.638∗∗ −0.244 −0.459∗∗ −0.487∗∗ −0.490∗∗ −0.353∗ −0.338∗ −0.298 0.002 0.002 0.039 0.069 0.137 0.180

0.387∗ 0.395∗∗ 0.190 −0.403∗∗ −0.557∗∗ −0.567∗∗ −0.357∗ −0.599∗∗ −0.510∗∗ −0.325∗ −0.304 −0.299 −0.301 −0.353∗ −0.347∗ −0.338∗

p < 0.05. p < 0.01.

Cases with large deformations Two patients had developed large seroma of more than 2 cm between the planning CT and the beginning of treatment. Another 2 patients had consistent 1-cm deformations due to large breasts. For these patients also Plan_10_13 was calculated and compared to the CBCT with the largest deformation. The dose coverages V95% and V90% were recorded for PTVb/c and CTV (Table 2). One case with seroma is also shown in Fig. 6B, D. Treatment planning with Plan_10_13 would not have reached proper CTV coverage for neither the seroma cases nor the postural changes any better than Plan_0_5 or Plan_5_8. Discussion The use of an optimization bolus has been reported earlier for the creation of the skin flash for VMAT radiotherapy of the breast,5,7 , 14,15 but the optimal thickness of the skin flash has not been studied. In this study, the PTV coverage was shown to decrease and the dose maxima to increase with thicker optimization bolus and PTV extension. In addition, the CBCT analyses showed that the largest skin flash with the thickest optimization bolus and PTV extension did not automatically result in the best skin coverage for the deformations seen among the studied patients. The optimization strategies of Nicolini et al.5 and Boman et 7 al. were similar to Plan_5_10 and Plan_0_5, respectively, using optimization bolus that extended 5 mm outside the optimization structure. In this study, the best dose distribution was obtained with Plan_0_5 in terms of dose conformity and homogeneity. A continuous decrease in plan quality was seen with increasing bolus thickness and PTV extension, with unacceptable maxima in Plan_10_13. The constancy in optimization parameters was neces-

7

sary for an unbiased study; however for patient treatments the dose maxima of Plan_5_8 and Plan_7_10 may be reduced with further optimization. The Plan_5_10 and Plan_7_12 were generally more suboptimal than the Plan_5_8 and Plan_7_10, respectively. An optimization bolus that extends only 3 mm outside the optimization structure seems therefore preferable. Swelling or other daily deformations in breast during radiotherapy has been reported in 20% to 50% of patients,15-17 with the size of 4 to 27 mm in 25% of patients15 or 7% to 17% in 50% of patients17 in tangential images. With the addition of intrafractional motion, adequate skin flash in breast RT is needed; especially if the patient is treated in free breathing.18 In the recalculations on actual CBCT-based patient geometries, PTV coverage was slightly lower using Plan_5_8 than Plan_0_5, but CTV coverage was similar in case of minor deformations; and slightly better when including the outliers. Plan_0_5 and Plan_5_8 had better CTV dose coverage than Plan_7_10. Thus, increasing the allowance for deformations from Plan_5_8 to Plan_7_10 did not increase the robustness, as the dose coverages (V90% and V95%) were similar or lower in Plan_7_10 than in Plan_5_8. Our median of largest tissue deformation of 5 mm was in line with previous studies. Plan_7_10 was therefore too much exaggerating the possible tissue swelling and the lower original plan quality than in Plan_0_5 and Plan_5_8 was realized in the treatment. The recalculated PTV dose minima of Plan_5_8 correlated the least with the amount of tissue swelling of the 3 plans, thus providing the best robustness. Without the need for unbiased study, in clinical practice Plan_5_8 may be improved close to that of Plan_0_5 with further reoptimization, thus combining good dose coverage and low sensitivity to tissue changes. Furthermore, it is known that breast cancer recurrence for postmastectomy RT is most often located at the chest wall,13 and the dose to the skin slabs was best in Plan_5_8. The 4 cases with large tissue deformations showed that Plan_0_5 was the least sensitive to postural changes, but the most sensitive to swelling. The tangentially emphasized beam directions allowed for postural changes in the tangential direction; but with the lack of PTV extension, not for swelling. Neither Plan_7_10 nor Plan_10_13 provided proper CTV coverage suggesting a need for replanning in cases of extensive swelling even with the largest PTV extension. Instead, those plans had the disadvantage of large maxima when compared to Plan_0_5 and Plan_5_8. Therefore, the VMAT plans are always sensitive to considerable tissue swelling or deformation even with large skin flash optimization. Unlike in conventional 3D planning, large tissue deformations seem to require replanning in VMAT. The monitoring of tissue deformations with tangential images and CBCT is important with VMAT RT of the breast or chest wall. The choice between Plan_0_5 and Plan_5_8 depends on the frequency of imaging and the resources available for replanning. The thicker skin flash may also prepare for daily variations, positioning errors and intrafractional movements.15,18 The need for replanning depends on patient anatomy, and should be considered individually. Based on the pa-

Table 2 V95% and V90% coverage for patients with large swelling or deformation in CBCT-based dose recalculations Plan

V95%

V90%

0_5

5_8

7_10

10_13

0_5

5_8

7_10

10_13

PTVb/c

Seroma #1 (2.3 cm) Seroma #2 (3.3 cm) Deformation #1 (1.1 cm) Deformation #2 (1.1 cm)

87.6% 59.5% 91.8% 96.2%

88.2% 68.1% 85.3% 93.6%

87.1% 71.5% 83.7% 92.9%

86.3% 70.8% 77.6% 94.4%

97.9% 85.3% 98.0% 99.6%

98.9% 91.5% 98.2% 98.8%

98.1% 92.8% 96.8% 98.3%

98.2% 93.2% 96.4% 98.8%

CTV

Seroma #1 (2.3 cm) Seroma #2 (3.3 cm) Deformation #1 (1.1 cm) Deformation #2 (1.1 cm)

91.9% 69.6% 95.4% 98.2%

92.3% 77.8% 95.6% 97.8%

92.5% 79.9% 93.8% 97.3%

91.5% 77.8% 92.7% 97.0%

99.0% 90.7% 99.2% 99.9%

99.8% 96.7% 99.5% 99.9%

99.6% 97.4% 99.0% 99.9%

99.8% 98.4% 98.8% 99.9%

JID: MDO 8

ARTICLE IN PRESS

[mUS5Gb;October 30, 2018;13:30]

M. Rossi et al. / Medical Dosimetry 000 (2018) 1–8

tient population in this study, at least deformations larger than 10 mm should be evaluated and potentially replanned. However, also smaller deformations may decrease the PTV coverage depending on both the plan and patient anatomy. At our clinic we currently treat with Plan_5_8 and monitor deformations in tangential images daily (DIBH patients) or at least weekly (FB patients). Confirmation with CBCT imaging is made at the beginning of treatment course, and further CBCT imaging is advised if tangential images show deformations larger than 7 mm. The dose coverage of PTV−5 decreased with increasing optimization thickness, whereas the dose to the PTVb/c remained relatively similar between the different bolus thicknesses. During dose optimization, the optimization bolus was placed only on the chest wall. Therefore when the final dose was calculated without the bolus, the dose was normalized downwards because less radiation was sufficient for the smaller volume. The dose normalization however decreases the dose to the axillary lymph node areas simultaneously, with more significant dose decrease with larger optimization bolus. Therefore one might consider drawing the axillar lymph node area separately and optimizing this to a slightly larger dose (around 100.5% for Plan_5_8), thus obtaining a 100% mean dose in the final dose calculation. Alternatively, the optimization bolus may be extended cranially to the lymph node area. Conclusions The optimization to PTV structure extended by 0 or 5 mm outside the skin contour with 5- or 8-mm optimization bolus was demonstrated to be the best option, from which the latter values gave better surface dose. Plan_5_8 showed the most optimal CTV coverage, acceptable dose maxima, and least sensitivity to tissue deformations and swelling. The choice of the skin flash depends on resources available for potential replanning. Larger bolus and PTV extension are not recommended. However, tissue deformations should be monitored frequently with tangential imaging or CBCT to detect potential large deformations exceeding optimization bolus thickness. Funding This study was financially supported by Seppo Nieminen Research Fund (grant number 150634). Conflicts of Interest No conflicts of interest.

References 1. EBCTCG Effect of radiotherapy after mastectomy and axillary surgery on 10-year recurrence and 20-year breast cancer mortality: Meta-analysis of individual patient data for 8135 women in 22 randomised trials. Lancet 383:2127–35; 2014. 2. Tournel, K.; Verellen, D.; Duchateau, M.; et al. An assessment of the use of skin flashes in helical tomotherapy using phantom and in-vivo dosimetry. Radiother Oncol 84:34–9; 2007. 3. Johansen, S.; Cozzi, L.; Olsen, D.R. A planning comparison of dose patterns in organs at risk and predicted risk for radiation induced malignancy in the contralateral breast following radiation therapy of primary breast using conventional, IMRT and volumetric modulated arc treatment techniques. Acta Oncol 48:495–503; 2009. 4. Popescu, C.C.; Olivotto, I.A.; Beckham, W.A.; et al. Volumetric modulated arc therapy improves dosimetry and reduces treatment time compared to conventional intensity-modulated radiotherapy for locoregional radiotherapy of left-sided breast cancer and internal mammary nodes. Int J Radiat Oncol Biol Phys 76:287–95; 2010. 5. Nicolini, G.; Fogliata, A.; Clivio, A.; Vanetti, E.; Cozzi, L. Planning strategies in volumetric modulated arc therapy for breast. Med Phys 38:4025–31; 2011. 6. Virén, T.; Heikkilä, J.; Myllyoja, K.; Koskela, K.; Lahtinen, T.; Seppälä, J. Tangential volumetric modulated arc therapy technique for left-sided breast cancer radiotherapy. Radiat Oncol 10:79; 2015. 7. Boman, E.; Rossi, M.; Haltamo, M.; Skyttä, T.; Kapanen, M. A new split arc VMAT technique for lymph node positive breast cancer. Phys Med; 2016. doi:10.1016/j. ejmp.2016.10.012. 8. Fogliata, A.; Seppälä, J.; Reggiori, G.; et al. Dosimetric trade-offs in breast treatment with VMAT technique. Br J Radiol 90:20160701; 2017. 9. Rossi, M.; Boman, E.; Kapanen, M. Contralateral tissue sparing in lymph nodepositive breast cancer radiotherapy with VMAT technique. Med Dosim; 2018. doi:10.1016/j.meddos.2018.03.005. 10. Mancosu, P.; Nicollini, G.; De Rose, F.; et al. EP-1637: Critical appraisal of deep inspiration breath hold CBCT for left breast using VMAT. Radiother Oncol 123:S887–8; 2017. 11. Michalski, A.; Atyeo, J.; Cox, J.; Rinks, M. Inter- and intra-fraction motion during radiation therapy to the whole breast in the supine position: a systematic review. J Med Imaging Radiat Oncol 56:499–509; 2012. 12. Kusters, M.; Dankers, F.; Monshouwer, R. EP-1639: Does intrafraction motion between vmDIBHs during breast cancer treatment impact on delivered dose? Radiother Oncol 123:S888–9; 2017. 13. Chao, K.; Perez, C.; Brady, L.. Radiation Oncology Management Decisions. 3rd Edition. Philadelphia: Wolters Kluwer-Lippincott Williams & Wilkins; 2011. 14. Tyran, M.; Tallet, A.; Resbeut, M.; et al. Safety and benefit of using a virtual bolus during treatment planning for breast cancer treated with arc therapy. J Appl Clin Med Phys; 2018. doi:10.1002/acm2.12398. 15. Rossi, M.; Boman, E.; Skyttä, T.; Haltamo, M.; Laaksomaa, M.; Kapanen, M. Dosimetric effects of anatomical deformations and positioning errors in VMAT breast radiotherapy. J Appl Clin Med Phys; 2018. doi:10.1002/acm2.12409. 16. Back, M.; Guerrieri, M.; Wratten, C.; Steigler, A. Impact of radiation therapy on acute toxicity in breast conservation therapy for early breast cancer. Clin Oncol (R Coll Radiol) 16:12–16; 2004. 17. Jain, P.; Marchant, T.; Green, M.; et al. Inter-fraction motion and dosimetric consequences during breast intensity-modulated radiotherapy (IMRT). Radiother Oncol 90:93–8; 2009. 18. George, R.; Keall, P.J.; Kini, V.R.; et al. Quantifying the effect of intrafraction motion during breast IMRT planning and dose delivery. Med Phys 30:552–62; 2003.