- Email: [email protected]
Skin dose estimation using virtual structures for Contura Multi-Lumen Balloon breast brachytherapy
Brachytherapy 17 (2018) 956e965
Breast/Soft Tissue
Skin dose estimation using virtual structures for Contura Multi-Lumen Balloon breast brachytherapy YuHuei Jessica Huang*, Fan-Chi Frances Su, David K. Gaffney, Kristine E. Kokeny, Hui Zhao, Prema Rassiah-Szegedi, Bill J. Salter, Matthew M. Poppe Department of Radiation Oncology, University of Utah, Salt Lake City, UT
ABSTRACT
PURPOSE: To propose a workflow that uses ultrasound (US)-measured skineballoon distances and virtual structure creations in the treatment planning system to evaluate the maximum skin dose for patients treated with Contura Multi-Lumen Balloon applicators. METHODS AND MATERIALS: Twenty-three patients were analyzed in this study. CT and US were used to investigate the interfractional skineballoon distance variations. Virtual structures were created on the planning CT to predict the maximum skin doses. Fitted curves and its equation can be obtained from the skineballoon distance vs. maximum skin dose plot using virtual structure information. The fidelity of US-measured skin distance and the skin dose prediction using virtual structures were assessed. RESULTS: The differences between CT- and US-measured skineballoon distances values had an average of 0.5 1.1 mm (95% confidence interval [CI] 5 1.0 to 0.1 mm). Using virtual structure created on CT, the average difference between the predicted and the actual dose overlay maximum skin dose was 1.7% (95% CI 5 3.0 to 0.4%). Furthermore, when applying the US-measured skin distance values in the virtual structure trendline equation, the differences between predicted and actual maximum skin dose had an average of 0.7 6.4% (95% CI 5 2.3% to 3.7%). CONCLUSIONS: It is possible to use US to observe interfraction skineballoon distance variation to replace CT acquisition. With the proposed workflow, based on the creation of virtual structures defined on the planning CT- and US-measured skineballoon distances, the maximum skin doses can be reasonably estimated. Ó 2018 American Brachytherapy Society. Published by Elsevier Inc. All rights reserved.
Keywords:
Accelerated partial breast irradiation; Contura; Brachytherapy; Skin dose
Introduction Accelerated partial-breast irradiation (APBI) is a type of radiation therapy that treats only the lumpectomy site with 1e2 cm margin in a hypofractionated regimen compared with whole-breast irradiation in standard fractionation. The efficacy of APBI has been proven by multiple trials with more than 10 years of followup results (1, 2, 3, 4).
Received 3 July 2018; received in revised form 12 August 2018; accepted 15 August 2018. Financial disclosure: The authors report no proprietary or commercial interest in any product mentioned or concept discussed in this article. * Corresponding author. Department of Radiation Oncology, University of Utah Huntsman Cancer Institute, 1950 Circle of HopeRoom 1570, Salt Lake City, UT 84112. Tel.: þ1-801-587-4759; fax: þ1-801581-2995. E-mail address: [email protected] (Y.J. Huang).
The treatment of APBI using brachytherapy can be carried out using either multicatheter interstitial needles, balloonbased applicators, or strut-based applicators. One of the most important dosimetric criteria during planning for APBI using brachytherapy is the maximum skin dose, as it is related to late toxicity and cosmetic results. For balloon-based applicators, the maximum skin dose is highly dependent on the minimum skineballoon distance. The Contura Multi-Lumen Balloon (MLB) is a single-entry balloon-based device that was developed because of its potential to improve skin dose compared with the original single-entry single-catheter balloon applicator, especially for patients with small skineballoon distances (5, 6). Previous studies had demonstrated even with small skin spacing, Contura MLB yields favorable cosmetic and toxicity outcomes. Akhtari (5) concluded that if the maximum skin dose for Contura MLB is kept under
1538-4721/$ - see front matter Ó 2018 American Brachytherapy Society. Published by Elsevier Inc. All rights reserved. https://doi.org/10.1016/j.brachy.2018.08.010
Y.J. Huang et al. / Brachytherapy 17 (2018) 956e965
125% of prescribed dose, patients with small skin spacing can be safely treated with good cosmetic outcomes; however, it is important to realize that skineballoon distance may change during the course of treatment. Because of this potential for change, it is important for the clinicians to monitor the skineballoon distances during the treatment course to ensure the maximum skin dose criteria are still within tolerance. In our institution, from the time the Contura MLB is implanted until the end of the treatment, the device is kept inside the surgical cavity for at least 8 days in the standard 34 Gy, 10-fraction, twice daily treatment regimen. Like other studies had observed, treatment variations, such as the size and shape of the balloon, seroma volume, catheter length, and skineballoon distance, etc, were found during the treatment course. A thorough analysis of day-to-day treatment variations had been published by Kuo (7). As was discussed in Kuo’s study, we also found the skine balloon distance as one of the main contributions of interfractional dose and geometry variation among our patients. To estimate the skineballoon distance variation, ultrasound (US) or CT can be used. The advantages of US include its ease of use and no added imaging dose from ionizing radiation. However, to most accurately account for the skine balloon distance change, CT still remains as the gold standard. CT has the advantage of higher resolution image quality and the ability to navigate in three-dimensional views to find the global minimal skineballoon distance. When a new CT is taken before the treatment fraction, the dose estimation of the maximum skin dose can be obtained from overlaying the treatment plan to the new CT. However, CT acquisition before each fraction means extra imaging doses to the patients. As the suitable patient population for APBI technique are the patients with early-stage breast cancer, higher imaging dose might also contribute to secondary malignancies, which should be avoided based on ALARA principle. One purpose of this study is to evaluate the potential of using US to replace CT images to measure skineballoon distance before each fraction. The other purpose of this study is to present a workflow to predict the maximum skin dose using virtual structures created on the planning CT. Based on the pretreatment US skin distance measurements and the virtual structure predictions, it is possible for the clinicians to determine the appropriateness of the maximum skin dose without adding additional CT imaging doses to the patients.
Methods and materials A total of 23 patients were included in this study between September 2015 and August 2017. Two prescription doses were used, 34 Gy in 10 fractions (BID) for 7 patients, and 22.5 Gy in three fractions (BID) for 16 patients who were enrolled in a prospective Phase II clinical trial, TRIUMPH-T (8). Typically, the implant of the Contura
957
MLB occurs 2 to 4 weeks after surgery. On the day of the implant, after the device placement, an immobilization device was made for each patient to ensure the reproducibility of the patient position during imaging and treatment. A postimplant CT (Day 0) was acquired on the day of implant and used for treatment planning. In this study, we define Day 0 as the implant/planning day, Day 1 the first treatment day, Day 2 the second treatment day, and so on. Depending on the prescription, the dosimetric goals used for planning in this study are listed in Table 1. CT acquisition schedule Our internal experience had shown that the soft tissue change, caused by the surgical cavity surrounding tissue settling down after the implanting procedure, seroma, hematoma, and/or air volume change, happens most dramatically on the day after the implant (Day 1). Therefore, we obtained Day 1 CTs for all patients before the first fraction of treatment delivery to verify the stability of the implant geometry. Replan using the Day 1 CT was used if drastic change of the implant geometry and/or patient anatomy was observed. With concern of geometry change over a weekend break, another CT was taken after the weekend before the morning treatment delivery. The typical CT acquisition schedules for both prescription regimens were listed in Tables 2 and 3. CT skineballoon distance measurement Oncentra Brachy (Elekta Brachytherapy Solutions, Veenendaal, the Netherlands) treatment planning system was used in this study. The minimum skineballoon distance can be determined by creating several margin structures from the balloon contour and observe when one of the margin structures starts intersecting with the body contour. The users can either scroll through each axial CT slices or Table 1 Treatment plan dosimetric criteria for Contura MLB brachytherapy Region of 34 Gy in 10 fractions a interest (ROI) Goal Acceptable PTV_ V95 $ 95% EVALc
Skin Rib a
V90 $ 90%
22.5 Gy in 3 fractions Goal
b
Acceptable
V90 $ 90%
V95 $ 90% with more air/fluidd V150 ! 30 cc !50 cc V150 ! 40 cc N/A V200 ! 15 cc !17 cc V200 ! 10 cc !10 cc Dmax # 100% Dmax # 145% Dmax # 100% #120% Dmax # 125% No limit Dmax # 100% #120%
Criteria based on Contura registry (9). Criteria based on Triumph-T protocol (8). c PTV_EVAL is created using 1 cm expansion from the balloon, excluding the balloon, chest wall, and 5 mm from the skin surface. d No adjustment needed if air-fluid volume is less than 5% of PTV_EVAL. If the displaced volume is between 5% and 10%, the constraint must be at least 95% of the PTV_EVAL receiving 90% of the prescribed dose. b
958
Y.J. Huang et al. / Brachytherapy 17 (2018) 956e965
Table 2 An example of typical treatment and imaging schedule for a patient with 34 Gy prescription Imaging modality
Day 0 (Tuesday)
Day 1 (Wednesday)
Day 2 (Thursday)
Day 3 (Friday)
Day 4 (Monday)
Day 5 (Tuesday)
Fraction No. CT US
— O O
1&2 O (AM) O (AM & PM)
3&4
5&6
9 & 10
O (AM & PM)
O (AM & PM)
7&8 O (AM) O (AM & PM)
O (AM & PM)
US 5 ultrasound. In this example, the patient was implanted on a Tuesday. Three CT data sets were taken, Day 0, Day 1, and Day 4 (after the weekend).
use the 3D display to find the minimum skineballoon distance (Fig. 1). The minimum skineballoon distance from all CT data sets for all cases were determined and recorded regardless of the time point when the CT was taken. US skineballoon distance measurement A subset of patients had data points with both CT and US skin distance measurements on various acquisition days for comparison purposes. Unlike CT images, where 3D global minimum skine balloon distance can be found, a US measurement can only approximate the minimal skineballoon distance. During the Day 0 US acquisition, the US skineballoon minimum distance baseline was determined and recorded. Several skin marks were also placed to show the position of the US probe when the skin distance baseline was determined (Fig. 2). On subsequent days, the US probe was positioned according to the skin marks to find the US Day X skine balloon distance value. The changes in US-measured skin distance between Day X and Day 0 were used to compare with the skineballoon distance changes found on CT on the corresponding days. Maximum skin dose estimation (DskinMax and DSkinMax_dmm)
Results
In the treatment planning system, virtual structures were created on the CT to estimate the maximum skin dose. As shown in Fig. 3a, the Contura balloon was contoured in cyan and the patient external body contour in yellow. The magenta structure was created by adding a large margin (2.5 cm in the example shown) to the balloon and subtracting the body contour. The amount of margin needed varies based on the patient’s anatomy; it should be large enough so that it extends outside of the body contour. This virtual structure, SkinMax (in magenta), would provide us with the maximum Table 3 The treatment and imaging schedule for a patient with 22.5 Gy prescription Imaging modality
Day 0 (planning day)
Day 1
Fraction No. CT US
1 (PM) O O
2&3 O (AM) O (AM & PM)
US 5 ultrasound. CT data sets were taken on Day 0 and Day 1.
skin point dose (DSkinMax) that was calculated according the treatment plan design. The maximum skin dose on the planning CT (Day 0) was obtained for all patients. The original treatment plans were applied to all Day X CTs to investigate the dosimetric effects due to changes of the minimum skine balloon distances. A total of 23 Day 0 plans and 34 Day X plans were created in this study. In Day 0 plans, virtual structures, SkinMax_dmm (d 5 1, 2.7), were created to estimate the changes of maximum skin point doses when skineballoon distances became smaller. Small outer margins in mm steps were added to this SkinMax structure. As shown in Fig. 3b, SkinMax_3 mm (light green) was used to simulate when the skineballoon distance decreased 3 mm. The maximum doses to the SkinMax_dmm structures were calculated based on the treatment plan. These structures would provide us information about the predicted maximum skin dose when the skineballoon distance decreases ‘‘d’’ mm. When a decreasing skineballoon distance, ‘‘d’’, was observed on the US image, we can find the predicted maximum skin dose from the corresponding SkinMax_dmm structure, DSkinMax_dmm. This process gives us the opportunity to estimate whether the maximum skin dose should be of concern without new CT taken.
Minimum skineballoon distance changes on CT (dCT) A total of 57 CT sets, including Day 0 and Day X data, were acquired for 23 patients. The minimum skineballoon distances (skin bridge) were measured on each CT, with an average of 11.4 6.2 mm, ranging between 2.5 mm and 32.5 mm; median value 11.5 mm. Because Day 0 CT minimum skineballoon distances were used as the baseline, 34 skineballoon distance delta values were obtained. The results of the delta values of the minimum skineballoon distance found between CT Day X and CT Day 0 were listed in Table 4. The delta values for minimum skineballoon distance ranged between 6.0 mm and 2.0 mm, with an average of 0.0 1.6 mm (median: 0.0 mm). While we observed one patient with a decrease of 6.0 mm of skine balloon distance, the rest of the skineballoon distance changes were all less than 2.0 mm, in either positive (increasing skin bridge) or negative (decreasing skin bridge) directions. Furthermore, about one-third of the data
Y.J. Huang et al. / Brachytherapy 17 (2018) 956e965
959
Fig. 1. Determination of minimum skineballoon distance on the 2D axial (left) and 3D (right) displays. Four structure contours are shown here: balloon (cyan), 5 mm margin structure (white), 6-mm margin structure (yellow), and body (blue). In this example, both 5-mm and 6-mm margin structures are shown, but only 6-mm structure intersected with the body contour. Therefore, the minimum skineballoon distance was 6 mm for this case. (For interpretation of the references to color in this figure legend, the reader is referred to the Web version of this article.)
points resulted in no change in skineballoon distance. The resulting maximum skin doses (DSkinMax), either on Day 0 or Day X, were plotted against the corresponding minimum skineballoon distance values, as shown in Fig. 4. Minimum skineballoon distance change comparisons: dCT vs. dUS For a subset of patients, there were corresponding US skineballoon distance measurements on the day when the CTs were taken. The changes on skineballoon distance measurement between Day X and Day 0 on CT and US were compared (Table 5). The CT-measured skineballoon distance change (dCT) ranged between 2.0 mm and 2.0 mm. The average was 0.4 1.3 mm, with a median
value of 1.0 mm. On the skineballoon distance changes measured by the US (dUS), an average of 0.0 1.6 mm was found (range between 4.4 mm and 4.2 mm, median: 0.1 mm). Among the skineballoon distance change found on CT, 12 of them had positive values (increasing skin bridge), four stayed the same, and five with decreasing skin bridge. The US results showed nine cases with positive values, one stayed the same, and 11 with decreasing values. The US found the same direction in change, either increasing or decreasing skin bridge, with CT in 13 cases. The average of the differences between dCT and dUS was 0.5 1.1 mm, ranging from 2.4 mm to 2.2 mm, with a median value of 0.5 mm (95% confidence interval [CI] 5 1.0 to 0.1 mm). The values of skineballoon distance changes on CT and US were plotted in Fig. 5. Ideally in Fig. 5, when the CT measure delta is 1.0 mm, the US will also measure 1.0 mm; and if CT is 1.0 mm, US will also correspond with 1.0 mm. In this plot, a slight trend was found (R2 5 0.49, p ! 0.05). Furthermore, it was noted that in almost all of the cases when US did not agree with CT in skin change directions, US overestimated the shrinking of the skin, which resulted in overestimating the increasing of skin maximum dose. From two cases (No. 12 & 15), the US observed increased skin bridge when CT found no change in skineballoon distance. This subset analysis gave us some insight on the possibilities to use US to monitor the changes in skineballoon distance throughout the course of treatment without extra imaging dose from CT. Maximum skin dose prediction using virtual structures
Fig. 2. Skin marks (red arrows) made on Day 0 indicating the US probe position to find the US skineballoon distance. US 5 ultrasound. (For interpretation of the references to color in this figure legend, the reader is referred to the Web version of this article.)
The dosimetric goals for the maximum skin dose were met for all patients. The average maximum skin dose for Day 0 (DSkinMax, Day0) was 84.2%, ranging between 38.5% and 113.2%, with a median value of 85.8%. The original treatment plan (Day 0) for each patient was applied to all of the Day X CTs. It was found the maximum skin
960
Y.J. Huang et al. / Brachytherapy 17 (2018) 956e965
Fig. 3. (a) Three structure contours shown on an axial CT image: balloon (cyan), external body (yellow), and SkinMax (magenta). (b) A 3-mm margin was added to the SkinMax structure to create SkinMmax_3 mm (light green) structures. The yellow dotted contour is used to show a simulated skineballoon distance decrease of 3 mm. (c) and (d) are simplified drawings to show how the SkinMax and SkinMax_dmm were created. (For interpretation of the references to color in this figure legend, the reader is referred to the Web version of this article.)
dose on Day X (DSkinMax, Day X) had an average of 88.6%; range: 34.0e127.0%; median: 85.9%. For each patient on their Day 0 CT, seven virtual structures were created, SkinMax_dmm (d 5 1, 2.7), where d represented the assumed values of the skine balloon distance delta. These virtual structures were used to predict the maximum skin dose (DSkinMax_dmm) when the minimum skineballoon distance decreased in d mm. One patient’s predicted maximum skin dose based on the virtual structures and the actual maximum skin dose on Day X CT were shown in Fig. 6. In this example, the original maximum skin dose was 58.0%. During the treatment course, two Day X CTs were taken. The resulting dCT on Day X CTs were 2.0 mm and 6.0 mm, with DSkinMax, Day X values of 62.1% and 77.3%, respectively. The overlay of the predicted and the actual maximum skin doses showed
good agreement in this example. Using the trendline equation obtained from the plot, one can apply the d value and find the predicted maximum skin dose. In the example in Fig. 6, the predicated DSkinMax_dmm(CT) values by the equation using dCT 5 0, 2, and 6 mm were 57.7%, 64.8%, and 81.7%, which were different from the actual DSkinMax, Day 0 or Day X values by 0.3%, 2.7%, and 4.7%, respectively. The predicted maximum skin doses based on each patient’s trendline equation were calculated and compared with the actual DSkinMax, Day X values. The average difference between the predicted value and the actual maximum skin dose was 1.7%, with a range of 21.8% to 4.9%, and a median value of 0.6% (95% CI 5 3.0 to 0.4%, 99% CI 5 3.4% to 0.0%).
Table 4 Statistics of minimum skineballoon distance delta found in this study
For the cases with both Day X CT and corresponding US-measured skineballoon distance delta value available, the predicated DSkinMax_dmm(US) values (using dUS) were used to compared with DSkinMax, Day X values, as listed in Table 6. The resulting differences between actual Day X CT found maximum skin doses and the predicted values using dUS and trendline equation established by Day 0 virtual structures plots had a range of 11.7% to 14.8%, with an average of 0.7 6.4% and median value of 0.9% (95% CI 5 2.3 to 3.7%, 99% CI 5 3.4 to 4.7%).
Minimum skineballoon distance delta (dCT) between Day X and Day 0 CT
Frequency
d ! 5.0 mm 5.0 mm # d ! 2.0 mm 2.0 mm # d ! 0.0 mm d 5 0.0 mm 0.0 mm ! d # 2.0 mm Total
1 0 9 10 14 34
(2.9%) (0.0%) (26.5%) (29.4%) (41.2%) (100.0%)
Note that the minus sign meant the decrease of the skin bridge.
Maximum skin dose prediction using US found skin bridge change (dUS)
Y.J. Huang et al. / Brachytherapy 17 (2018) 956e965
961
Fig. 4. Data from all patients showing maximum skin dose on the original (Day 0) and the subsequent CT (Day X) based on the minimum skineballoon distance. An exponential trendline was also plotted (R2 0.9235).
Discussion Even though skin toxicity is one of the major concerns in APBI brachytherapy, no consensus is available to measure day-to-day skineballoon distance variations. This study is Table 5 Skineballoon distance delta between Day X and Day 0 results measured by CT and US
No.
CT skin distance delta (dCT, mm)
US skin distance delta (dUS, mm)
Difference between CT and US (dUSdCT) (mm)
1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 Average
0.0 2.0 1.5 1.0 1.0 2.0 1.0 1.0 1.0 2.0 0.0 0.0 1.5 1.0 0.0 2.0 1.0 1.0 2.0 2.0 2.0 0.4 1.3
0.5 4.4 0.6 0.6 0.1 1.4 0.4 0.3 0.2 4.2 0.5 1.4 0.4 0.5 0.3 1.5 0.3 0.1 1.9 0.0 1.5 0.0 1.6
0.5 2.4 2.1 0.4 0.9 0.6 1.4 1.3 0.8 2.2 0.5 1.4 1.1 0.5 0.3 0.5 1.3 0.9 0.1 2.0 0.5 0.5 1.1
US 5 ultrasound.
US and CT skin change in the same direction U U U U U U U U U U U U U
looking into using US along with the creation of virtual structures on Day 0 CT to predict maximum skin dose variation. Interfraction variation of APBI using Contura had been reported previously (7, 10, 11, 12). Kuo’s study had included detail analysis of 7 patients with many aspects of changes found during the course of MLB-based brachytherapy. The ranges of observed minimum skineballoon distances were similar to our results. It was found both in Kuo’s study and ours that the smaller initial skineballoon distances resulted in higher dosimetric impacts and tended to differ more than the fitted curve values (Fig. 4). Kuo concluded a mean decrease of 2 mm skineballoon distance could result in a 20e40% dose increase if the initial skine balloon distance is less than 12 mm. Their findings also demonstrated the importance of skineballoon distance monitoring during the treatment course to prevent the maximum skin dose exceeding the tolerance values. In our clinical experience, we observed more skineballoon distance changes on bigger, more pendulous patients. In fact, the case with 6.0 mm skin bridge change from our study was from one patient with larger breast and more loose fat tissue surrounding the balloon. We saw a big part of her breast tissue shifted back to its natural position on Day 1, which changed the geometric relationship between the balloon and the skin surface. Based on our findings, we recommend the clinics to at least take another CT before the first fraction or treatment in case there is drastic soft tissue motion that could change the dosimetric parameters, especially if the skineballoon distance is small or if the maximum skin dose is very close to the tolerable values in the original plan.
962
Y.J. Huang et al. / Brachytherapy 17 (2018) 956e965
Fig. 5. CT and US found skineballoon distance change between Day X and Day 0 plot. A positive delta value meant an increasing of skineballoon distance, which would result in lower maximum skin dose, and negative distance delta caused higher maximum skin dose. US 5 ultrasound.
When daily CT is not available, the utilization of virtual structures had been investigated previously for dose predictions based on geometric changes. In Bhatt’s study series
(10, 12), they looked into potential effect of seroma using virtual structures on CT to calculate dosimetric coverage of the target volume. This study demonstrated the possibilities to
Fig. 6. Maximum skin dose value predictions using virtual structures and Day 0 or X plan calculation comparisons for one patient. The predicated DSkinMax_dmm values were obtained from doses to SkinMax_dmm virtual structures, d 5 1, 2.7. The maximum skin dose on Day 0 and Day X CT, DSkinMax, Day 0 and DSkinMax, Day X, were also shown.
Y.J. Huang et al. / Brachytherapy 17 (2018) 956e965
963
Table 6 The comparison results between the actual maximum skin doses calculated on Day X CT vs. using US skin delta value measurement and virtual structure trendline equation method to predict maximum skin dose No.
DSkinMax,
1 2 3a 4a 5a 6a 7a 8a 9a 10a 11a 12a 13a 14a 15a 16a 17a 18 19 20 21 Average standard deviation
66.8% 77.0% 74.6% 80.2% 84.0% 92.8% 67.4% 86.2% 84.4% 61.5% 110.6% 34.0% 79.1% 93.2% 62.6% 74.4% 115.8% 102.2% 111.7% 86.0% 127.0% 84.4 20.9%
Day X
US skin distance delta (dUS, in mm)
DSkinMax_dmm(US)
Difference (DSkinMax_dmm(US)DSkinMax,
0.5 4.4 0.6 0.6 0.1 1.4 0.4 0.3 0.2 4.2 1.0 1.4 0.4 0.5 0.3 1.5 0.3 0.1 1.9 0.0 1.5 0.0 1.6
69.4% 86.3% 89.4% 77.5% 88.8% 91.4% 74.3% 87.1% 80.2% 50.9% 101.3% 36.2% 82.5% 93.2% 63.0% 77.6% 114.6% 99.2% 115.6% 94.1% 115.2% 85.0 19.4%
2.6% 9.3% 14.8% L2.6% 4.8% L1.4% 6.8% 0.9% L4.2% L10.7% L9.2% 2.2% 3.3% 0.0% 0.4% 3.2% L1.2% L3.0% 3.9% 8.1% L11.7% 0.7 6.4%
Day X)
The maximum deviation between the two methods (14.8%) was from an overestimation of the maximum skin predicted values. Minus values (in bold) demonstrated the cases when the virtual structure prediction method using US delta underestimated the maximum skin doses. Note that all of the maximum skin doses (actual or predicted) were still within the dose goal tolerance as listed in Table 1. a Patients treated with 22.5 Gy in 3-fraction regimen.
predict skin dose changes using virtual structures. Our results had shown the predicted doses using virtual structures agreed well with the actual maximum skin dose, DSkinMax, Day X (99% CI 5 3.4% to 0.0%). It was noted, however, that when the minimum skin bridge contained a continuous strip of tissue with similar thickness to the minimum skineballoon distance, and the minimum skineballoon distance was less than 7 mm, the virtual structure prediction method had resulted in the most deviation from actual DSkinMax, Day X values. Therefore, cautions should be taken when considering applying the virtual structures to estimate the skin dose variations for patients with the anatomical characteristics as described previously. In our study, we were able to establish fitted curve with its equation to predict the maximum skin dose based on the US-found skineballoon distance delta values (dUS) with an average difference of 0.7% 6.4% (95% CI 5 2.3% to 3.7%). As shown in Table 6, three cases underestimated the maximum skin doses with more than 5% differences when compared with actual DSkinMax, Day X values. Note that only one of the case’s maximum skin doses was of concern based on the dose goal of 100% (Table 1), but would still be within minimum acceptable range of #120%, as it was a 3-fraction regimen case (case No. 17). Our data had shown potentials of using US-measured skineballoon distances and CT virtual structures to provide the clinicians with additional information to evaluate the maximum skin dose without additional CT imaging acquisition.
Based on our findings, the following workflow is proposed. On Day 0 CT, after the treatment plan is approved, virtual structures, SkinMax_dmm, are created to establish a scatterplot curve and equation, as shown in Fig. 7. This Figure was plotted using data from one of the studied patients, who was prescribed with 34 Gy/10 fx regimen. The goal of the maximum skin dose is #100%, with an acceptable value of #145%. In this example, the original planned DSkinMax was 97.7%, and to exceed skin dose of more than 145%, the delta value of the minimum skineballoon distance needs to be more than 5.0 mm (~5.2 mm). Based on this plot and the projected maximum d value, we can have confidence that the maximum skin dose will be within acceptable ranges as long as the US measure dUS is less than 5.2 mm. If a more conservative measure is taken, with 15% margin (99% CI up to 5%, plus an arbitrary 10% error bar based on our institutional experience), as long as the change of the minimum skineballoon distance found on the patient is less than 3.8 mm, the maximum skin dose would be well within tolerance. This proposed workflow offers quick and simple mechanism to provide the clinicians with confidences about the quality of the treatment without adding additional imaging doses to the patients. As with other imaging modalities used in the Radiation Oncology department, the quality of the information provided by the image is highly dependent on the experience levels of the users. It is even more critical when using US to measure an unstable (breast tissue) curved surface (Contura balloon).
964
Y.J. Huang et al. / Brachytherapy 17 (2018) 956e965
Fig. 7. The maximum skin dose prediction plot according to skineballoon distance delta values. In the example, the original maximum skin dose was 97.7%; the maximum skin dose would be more than 100% (dose goal) when the delta value was about 0.4 mm (green arrow). However, to exceed the minimum acceptable value of 145%, the minimum skineballoon distance delta would need to be more than 5.2 mm (red arrow). The more conservative values (in orange) were also shown if study uncertainties (15% margin) were also included for consideration. (For interpretation of the references to color in this figure legend, the reader is referred to the Web version of this article.)
It is also important to ensure no excessive pressure is applied to the probe during US measurement, as too much pressure could introduce distorted images and affect our skin distance results. In our practice, we have instructed our staff to use enough but the lightest pressure that allows for good image quality for skin bridge determination. Furthermore, instead of searching for the global minimum skineballoon distance, the users should look for a stabilized position with a clear view of the balloon and the skin bridge for best reproducibility to find the minimum skin distance baseline. The ideal position of the US probe relative to the balloon varies based on the patient’s anatomy; and could either be parallel or perpendicular to the balloon applicator axis. We had observed more interuser differences on US-measured skineballoon distances in our first couple of treated patients. As a group, the process of obtaining US information should be well established so that the results of the US measurements are reproducible. It is important for the clinicians to understand only with good quality US measurements can this proposed workflow provide useful information to predict the maximum skin dose. With the proposed workflow, other interfraction variations, such as applicator rotation, can still occur. The clinicians should understand the possibilities that some of these variations during MLB treatment course might not be as detectable when CT is eliminated. Therefore, when concerns arise during pretreatment checks, additional CT for verification or even new plan should still be considered to ensure minimum dose variation in treatment delivery. In this study, maximum skin point dose was investigated instead of the more robust measures using D0.2cc for a 2 mm thick skin layer or D1 cc as concluded by Hilts (13). We had
intentionally chosen to keep point dose for investigation as we were not looking for clinical cosmetic representation of the skin dose value. Furthermore, the point dose matrix used in the study was a more conservative mechanism, which was reasonable for the purpose of the study.
Conclusions In conclusion, it is possible to use US to observe interfraction skineballoon distance variation to replace CT acquisition. With the proposed workflow, based on the creation of virtual structures defined on the planning CT- and US-measured skineballoon distances, the skin maximum doses can be reasonably estimated. The clinicians can determine when a new CT acquisition and a new plan are needed if the skineballoon distance delta exceeded a predefined tolerance value. This can result in elimination of extra imaging dose from CT and facilitate treatment efficiency for patients treated with Contura MLB brachytherapy.
References [1] Hepel JT, Arthur D, Shaitelman S, et al. American Brachytherapy Society consensus report for accelerated partial breast irradiation using interstitial multicatheter brachytherapy. Brachytherapy 2017;16: 919e928. [2] Shah C, Vicini F, Shaitelman SF, et al. The American Brachytherapy Society consensus statement for accelerated partial-breast irradiation. Brachytherapy 2018;17:154e170.
Y.J. Huang et al. / Brachytherapy 17 (2018) 956e965 [3] Njeh CF, Saunders MW, Langton CM. Accelerated partial breast irradiation (APBI): A review of available techniques. Radiat Oncol 2010;5:90. [4] Cuttino LW, Arthur DW, Vicini F, et al. Long-term results from the contura multilumen balloon breast brachytherapy catheter phase 4 registry trial. Int J Radiat Oncol Biol Phys 2014;90:1025e1029. [5] Akhtari M, Abboud M, Szeja S, et al. Clinical outcomes, toxicity, and cosmesis in breast cancer patients with close skin spacing treated with accelerated partial breast irradiation (APBI) using multi-lumen/catheter applicators. J Contemp Brachytherapy 2016; 8:497e504. [6] Lee K, Quillo A, Dillon D, et al. Acute toxicity and early cosmetic outcome in patients treated with multilumen balloon brachytherapy with skin spacing # 7.0 millimeters. J Contemp Brachytherapy 2012;4:8e13. [7] Kuo HC, Mehta KJ, Hong L, et al. Day to day treatment variations of accelerated partial breast brachytherapy using a multi-lumen balloon. J Contemp Brachytherapy 2014;6:68e75.
965
[8] Accelerated partial breast radiation therapy using high-dose rate brachytherapy in treating patients with early stage breast cancer after surgery - full text view - ClinicalTrials.gov. Available at: https:// clinicaltrials.gov/ct2/show/NCT02526498. Accessed May 7, 2018. [9] Arthur DW, Vicini FA, Todor DA, et al. Contura multi-lumen balloon breast brachytherapy catheter: Comparative dosimetric findings of a phase 4 trial. Int J Radiat Oncol Biol Phys 2013;86:264e269. [10] Bhatt A, Sowards K, Bhatt G, et al. Impact of interfraction seroma collection on breast brachytherapy dosimetry e a mathematical model. J Contemp Brachytherapy 2012;4:101e105. [11] Bhatt AD, Crew JB, Bhatt G, et al. Interfraction accumulation of seroma during accelerated partial breast irradiation: Preliminary results of a prospective study. Brachytherapy 2012;11:374e379. [12] Bhatt AD, Barry PN, Sowards KT, et al. Negative effect of seroma on breast balloon brachytherapy dosimetry. Pract Radiat Oncol 2014;4: e1ee5. [13] Hilts M, Halperin H, Morton D, et al. Skin dose in breast brachytherapy: Defining a robust metric. Brachytherapy 2015;14:970e978.
Breast/Soft Tissue
Skin dose estimation using virtual structures for Contura Multi-Lumen Balloon breast brachytherapy YuHuei Jessica Huang*, Fan-Chi Frances Su, David K. Gaffney, Kristine E. Kokeny, Hui Zhao, Prema Rassiah-Szegedi, Bill J. Salter, Matthew M. Poppe Department of Radiation Oncology, University of Utah, Salt Lake City, UT
ABSTRACT
PURPOSE: To propose a workflow that uses ultrasound (US)-measured skineballoon distances and virtual structure creations in the treatment planning system to evaluate the maximum skin dose for patients treated with Contura Multi-Lumen Balloon applicators. METHODS AND MATERIALS: Twenty-three patients were analyzed in this study. CT and US were used to investigate the interfractional skineballoon distance variations. Virtual structures were created on the planning CT to predict the maximum skin doses. Fitted curves and its equation can be obtained from the skineballoon distance vs. maximum skin dose plot using virtual structure information. The fidelity of US-measured skin distance and the skin dose prediction using virtual structures were assessed. RESULTS: The differences between CT- and US-measured skineballoon distances values had an average of 0.5 1.1 mm (95% confidence interval [CI] 5 1.0 to 0.1 mm). Using virtual structure created on CT, the average difference between the predicted and the actual dose overlay maximum skin dose was 1.7% (95% CI 5 3.0 to 0.4%). Furthermore, when applying the US-measured skin distance values in the virtual structure trendline equation, the differences between predicted and actual maximum skin dose had an average of 0.7 6.4% (95% CI 5 2.3% to 3.7%). CONCLUSIONS: It is possible to use US to observe interfraction skineballoon distance variation to replace CT acquisition. With the proposed workflow, based on the creation of virtual structures defined on the planning CT- and US-measured skineballoon distances, the maximum skin doses can be reasonably estimated. Ó 2018 American Brachytherapy Society. Published by Elsevier Inc. All rights reserved.
Keywords:
Accelerated partial breast irradiation; Contura; Brachytherapy; Skin dose
Introduction Accelerated partial-breast irradiation (APBI) is a type of radiation therapy that treats only the lumpectomy site with 1e2 cm margin in a hypofractionated regimen compared with whole-breast irradiation in standard fractionation. The efficacy of APBI has been proven by multiple trials with more than 10 years of followup results (1, 2, 3, 4).
Received 3 July 2018; received in revised form 12 August 2018; accepted 15 August 2018. Financial disclosure: The authors report no proprietary or commercial interest in any product mentioned or concept discussed in this article. * Corresponding author. Department of Radiation Oncology, University of Utah Huntsman Cancer Institute, 1950 Circle of HopeRoom 1570, Salt Lake City, UT 84112. Tel.: þ1-801-587-4759; fax: þ1-801581-2995. E-mail address: [email protected] (Y.J. Huang).
The treatment of APBI using brachytherapy can be carried out using either multicatheter interstitial needles, balloonbased applicators, or strut-based applicators. One of the most important dosimetric criteria during planning for APBI using brachytherapy is the maximum skin dose, as it is related to late toxicity and cosmetic results. For balloon-based applicators, the maximum skin dose is highly dependent on the minimum skineballoon distance. The Contura Multi-Lumen Balloon (MLB) is a single-entry balloon-based device that was developed because of its potential to improve skin dose compared with the original single-entry single-catheter balloon applicator, especially for patients with small skineballoon distances (5, 6). Previous studies had demonstrated even with small skin spacing, Contura MLB yields favorable cosmetic and toxicity outcomes. Akhtari (5) concluded that if the maximum skin dose for Contura MLB is kept under
1538-4721/$ - see front matter Ó 2018 American Brachytherapy Society. Published by Elsevier Inc. All rights reserved. https://doi.org/10.1016/j.brachy.2018.08.010
Y.J. Huang et al. / Brachytherapy 17 (2018) 956e965
125% of prescribed dose, patients with small skin spacing can be safely treated with good cosmetic outcomes; however, it is important to realize that skineballoon distance may change during the course of treatment. Because of this potential for change, it is important for the clinicians to monitor the skineballoon distances during the treatment course to ensure the maximum skin dose criteria are still within tolerance. In our institution, from the time the Contura MLB is implanted until the end of the treatment, the device is kept inside the surgical cavity for at least 8 days in the standard 34 Gy, 10-fraction, twice daily treatment regimen. Like other studies had observed, treatment variations, such as the size and shape of the balloon, seroma volume, catheter length, and skineballoon distance, etc, were found during the treatment course. A thorough analysis of day-to-day treatment variations had been published by Kuo (7). As was discussed in Kuo’s study, we also found the skine balloon distance as one of the main contributions of interfractional dose and geometry variation among our patients. To estimate the skineballoon distance variation, ultrasound (US) or CT can be used. The advantages of US include its ease of use and no added imaging dose from ionizing radiation. However, to most accurately account for the skine balloon distance change, CT still remains as the gold standard. CT has the advantage of higher resolution image quality and the ability to navigate in three-dimensional views to find the global minimal skineballoon distance. When a new CT is taken before the treatment fraction, the dose estimation of the maximum skin dose can be obtained from overlaying the treatment plan to the new CT. However, CT acquisition before each fraction means extra imaging doses to the patients. As the suitable patient population for APBI technique are the patients with early-stage breast cancer, higher imaging dose might also contribute to secondary malignancies, which should be avoided based on ALARA principle. One purpose of this study is to evaluate the potential of using US to replace CT images to measure skineballoon distance before each fraction. The other purpose of this study is to present a workflow to predict the maximum skin dose using virtual structures created on the planning CT. Based on the pretreatment US skin distance measurements and the virtual structure predictions, it is possible for the clinicians to determine the appropriateness of the maximum skin dose without adding additional CT imaging doses to the patients.
Methods and materials A total of 23 patients were included in this study between September 2015 and August 2017. Two prescription doses were used, 34 Gy in 10 fractions (BID) for 7 patients, and 22.5 Gy in three fractions (BID) for 16 patients who were enrolled in a prospective Phase II clinical trial, TRIUMPH-T (8). Typically, the implant of the Contura
957
MLB occurs 2 to 4 weeks after surgery. On the day of the implant, after the device placement, an immobilization device was made for each patient to ensure the reproducibility of the patient position during imaging and treatment. A postimplant CT (Day 0) was acquired on the day of implant and used for treatment planning. In this study, we define Day 0 as the implant/planning day, Day 1 the first treatment day, Day 2 the second treatment day, and so on. Depending on the prescription, the dosimetric goals used for planning in this study are listed in Table 1. CT acquisition schedule Our internal experience had shown that the soft tissue change, caused by the surgical cavity surrounding tissue settling down after the implanting procedure, seroma, hematoma, and/or air volume change, happens most dramatically on the day after the implant (Day 1). Therefore, we obtained Day 1 CTs for all patients before the first fraction of treatment delivery to verify the stability of the implant geometry. Replan using the Day 1 CT was used if drastic change of the implant geometry and/or patient anatomy was observed. With concern of geometry change over a weekend break, another CT was taken after the weekend before the morning treatment delivery. The typical CT acquisition schedules for both prescription regimens were listed in Tables 2 and 3. CT skineballoon distance measurement Oncentra Brachy (Elekta Brachytherapy Solutions, Veenendaal, the Netherlands) treatment planning system was used in this study. The minimum skineballoon distance can be determined by creating several margin structures from the balloon contour and observe when one of the margin structures starts intersecting with the body contour. The users can either scroll through each axial CT slices or Table 1 Treatment plan dosimetric criteria for Contura MLB brachytherapy Region of 34 Gy in 10 fractions a interest (ROI) Goal Acceptable PTV_ V95 $ 95% EVALc
Skin Rib a
V90 $ 90%
22.5 Gy in 3 fractions Goal
b
Acceptable
V90 $ 90%
V95 $ 90% with more air/fluidd V150 ! 30 cc !50 cc V150 ! 40 cc N/A V200 ! 15 cc !17 cc V200 ! 10 cc !10 cc Dmax # 100% Dmax # 145% Dmax # 100% #120% Dmax # 125% No limit Dmax # 100% #120%
Criteria based on Contura registry (9). Criteria based on Triumph-T protocol (8). c PTV_EVAL is created using 1 cm expansion from the balloon, excluding the balloon, chest wall, and 5 mm from the skin surface. d No adjustment needed if air-fluid volume is less than 5% of PTV_EVAL. If the displaced volume is between 5% and 10%, the constraint must be at least 95% of the PTV_EVAL receiving 90% of the prescribed dose. b
958
Y.J. Huang et al. / Brachytherapy 17 (2018) 956e965
Table 2 An example of typical treatment and imaging schedule for a patient with 34 Gy prescription Imaging modality
Day 0 (Tuesday)
Day 1 (Wednesday)
Day 2 (Thursday)
Day 3 (Friday)
Day 4 (Monday)
Day 5 (Tuesday)
Fraction No. CT US
— O O
1&2 O (AM) O (AM & PM)
3&4
5&6
9 & 10
O (AM & PM)
O (AM & PM)
7&8 O (AM) O (AM & PM)
O (AM & PM)
US 5 ultrasound. In this example, the patient was implanted on a Tuesday. Three CT data sets were taken, Day 0, Day 1, and Day 4 (after the weekend).
use the 3D display to find the minimum skineballoon distance (Fig. 1). The minimum skineballoon distance from all CT data sets for all cases were determined and recorded regardless of the time point when the CT was taken. US skineballoon distance measurement A subset of patients had data points with both CT and US skin distance measurements on various acquisition days for comparison purposes. Unlike CT images, where 3D global minimum skine balloon distance can be found, a US measurement can only approximate the minimal skineballoon distance. During the Day 0 US acquisition, the US skineballoon minimum distance baseline was determined and recorded. Several skin marks were also placed to show the position of the US probe when the skin distance baseline was determined (Fig. 2). On subsequent days, the US probe was positioned according to the skin marks to find the US Day X skine balloon distance value. The changes in US-measured skin distance between Day X and Day 0 were used to compare with the skineballoon distance changes found on CT on the corresponding days. Maximum skin dose estimation (DskinMax and DSkinMax_dmm)
Results
In the treatment planning system, virtual structures were created on the CT to estimate the maximum skin dose. As shown in Fig. 3a, the Contura balloon was contoured in cyan and the patient external body contour in yellow. The magenta structure was created by adding a large margin (2.5 cm in the example shown) to the balloon and subtracting the body contour. The amount of margin needed varies based on the patient’s anatomy; it should be large enough so that it extends outside of the body contour. This virtual structure, SkinMax (in magenta), would provide us with the maximum Table 3 The treatment and imaging schedule for a patient with 22.5 Gy prescription Imaging modality
Day 0 (planning day)
Day 1
Fraction No. CT US
1 (PM) O O
2&3 O (AM) O (AM & PM)
US 5 ultrasound. CT data sets were taken on Day 0 and Day 1.
skin point dose (DSkinMax) that was calculated according the treatment plan design. The maximum skin dose on the planning CT (Day 0) was obtained for all patients. The original treatment plans were applied to all Day X CTs to investigate the dosimetric effects due to changes of the minimum skine balloon distances. A total of 23 Day 0 plans and 34 Day X plans were created in this study. In Day 0 plans, virtual structures, SkinMax_dmm (d 5 1, 2.7), were created to estimate the changes of maximum skin point doses when skineballoon distances became smaller. Small outer margins in mm steps were added to this SkinMax structure. As shown in Fig. 3b, SkinMax_3 mm (light green) was used to simulate when the skineballoon distance decreased 3 mm. The maximum doses to the SkinMax_dmm structures were calculated based on the treatment plan. These structures would provide us information about the predicted maximum skin dose when the skineballoon distance decreases ‘‘d’’ mm. When a decreasing skineballoon distance, ‘‘d’’, was observed on the US image, we can find the predicted maximum skin dose from the corresponding SkinMax_dmm structure, DSkinMax_dmm. This process gives us the opportunity to estimate whether the maximum skin dose should be of concern without new CT taken.
Minimum skineballoon distance changes on CT (dCT) A total of 57 CT sets, including Day 0 and Day X data, were acquired for 23 patients. The minimum skineballoon distances (skin bridge) were measured on each CT, with an average of 11.4 6.2 mm, ranging between 2.5 mm and 32.5 mm; median value 11.5 mm. Because Day 0 CT minimum skineballoon distances were used as the baseline, 34 skineballoon distance delta values were obtained. The results of the delta values of the minimum skineballoon distance found between CT Day X and CT Day 0 were listed in Table 4. The delta values for minimum skineballoon distance ranged between 6.0 mm and 2.0 mm, with an average of 0.0 1.6 mm (median: 0.0 mm). While we observed one patient with a decrease of 6.0 mm of skine balloon distance, the rest of the skineballoon distance changes were all less than 2.0 mm, in either positive (increasing skin bridge) or negative (decreasing skin bridge) directions. Furthermore, about one-third of the data
Y.J. Huang et al. / Brachytherapy 17 (2018) 956e965
959
Fig. 1. Determination of minimum skineballoon distance on the 2D axial (left) and 3D (right) displays. Four structure contours are shown here: balloon (cyan), 5 mm margin structure (white), 6-mm margin structure (yellow), and body (blue). In this example, both 5-mm and 6-mm margin structures are shown, but only 6-mm structure intersected with the body contour. Therefore, the minimum skineballoon distance was 6 mm for this case. (For interpretation of the references to color in this figure legend, the reader is referred to the Web version of this article.)
points resulted in no change in skineballoon distance. The resulting maximum skin doses (DSkinMax), either on Day 0 or Day X, were plotted against the corresponding minimum skineballoon distance values, as shown in Fig. 4. Minimum skineballoon distance change comparisons: dCT vs. dUS For a subset of patients, there were corresponding US skineballoon distance measurements on the day when the CTs were taken. The changes on skineballoon distance measurement between Day X and Day 0 on CT and US were compared (Table 5). The CT-measured skineballoon distance change (dCT) ranged between 2.0 mm and 2.0 mm. The average was 0.4 1.3 mm, with a median
value of 1.0 mm. On the skineballoon distance changes measured by the US (dUS), an average of 0.0 1.6 mm was found (range between 4.4 mm and 4.2 mm, median: 0.1 mm). Among the skineballoon distance change found on CT, 12 of them had positive values (increasing skin bridge), four stayed the same, and five with decreasing skin bridge. The US results showed nine cases with positive values, one stayed the same, and 11 with decreasing values. The US found the same direction in change, either increasing or decreasing skin bridge, with CT in 13 cases. The average of the differences between dCT and dUS was 0.5 1.1 mm, ranging from 2.4 mm to 2.2 mm, with a median value of 0.5 mm (95% confidence interval [CI] 5 1.0 to 0.1 mm). The values of skineballoon distance changes on CT and US were plotted in Fig. 5. Ideally in Fig. 5, when the CT measure delta is 1.0 mm, the US will also measure 1.0 mm; and if CT is 1.0 mm, US will also correspond with 1.0 mm. In this plot, a slight trend was found (R2 5 0.49, p ! 0.05). Furthermore, it was noted that in almost all of the cases when US did not agree with CT in skin change directions, US overestimated the shrinking of the skin, which resulted in overestimating the increasing of skin maximum dose. From two cases (No. 12 & 15), the US observed increased skin bridge when CT found no change in skineballoon distance. This subset analysis gave us some insight on the possibilities to use US to monitor the changes in skineballoon distance throughout the course of treatment without extra imaging dose from CT. Maximum skin dose prediction using virtual structures
Fig. 2. Skin marks (red arrows) made on Day 0 indicating the US probe position to find the US skineballoon distance. US 5 ultrasound. (For interpretation of the references to color in this figure legend, the reader is referred to the Web version of this article.)
The dosimetric goals for the maximum skin dose were met for all patients. The average maximum skin dose for Day 0 (DSkinMax, Day0) was 84.2%, ranging between 38.5% and 113.2%, with a median value of 85.8%. The original treatment plan (Day 0) for each patient was applied to all of the Day X CTs. It was found the maximum skin
960
Y.J. Huang et al. / Brachytherapy 17 (2018) 956e965
Fig. 3. (a) Three structure contours shown on an axial CT image: balloon (cyan), external body (yellow), and SkinMax (magenta). (b) A 3-mm margin was added to the SkinMax structure to create SkinMmax_3 mm (light green) structures. The yellow dotted contour is used to show a simulated skineballoon distance decrease of 3 mm. (c) and (d) are simplified drawings to show how the SkinMax and SkinMax_dmm were created. (For interpretation of the references to color in this figure legend, the reader is referred to the Web version of this article.)
dose on Day X (DSkinMax, Day X) had an average of 88.6%; range: 34.0e127.0%; median: 85.9%. For each patient on their Day 0 CT, seven virtual structures were created, SkinMax_dmm (d 5 1, 2.7), where d represented the assumed values of the skine balloon distance delta. These virtual structures were used to predict the maximum skin dose (DSkinMax_dmm) when the minimum skineballoon distance decreased in d mm. One patient’s predicted maximum skin dose based on the virtual structures and the actual maximum skin dose on Day X CT were shown in Fig. 6. In this example, the original maximum skin dose was 58.0%. During the treatment course, two Day X CTs were taken. The resulting dCT on Day X CTs were 2.0 mm and 6.0 mm, with DSkinMax, Day X values of 62.1% and 77.3%, respectively. The overlay of the predicted and the actual maximum skin doses showed
good agreement in this example. Using the trendline equation obtained from the plot, one can apply the d value and find the predicted maximum skin dose. In the example in Fig. 6, the predicated DSkinMax_dmm(CT) values by the equation using dCT 5 0, 2, and 6 mm were 57.7%, 64.8%, and 81.7%, which were different from the actual DSkinMax, Day 0 or Day X values by 0.3%, 2.7%, and 4.7%, respectively. The predicted maximum skin doses based on each patient’s trendline equation were calculated and compared with the actual DSkinMax, Day X values. The average difference between the predicted value and the actual maximum skin dose was 1.7%, with a range of 21.8% to 4.9%, and a median value of 0.6% (95% CI 5 3.0 to 0.4%, 99% CI 5 3.4% to 0.0%).
Table 4 Statistics of minimum skineballoon distance delta found in this study
For the cases with both Day X CT and corresponding US-measured skineballoon distance delta value available, the predicated DSkinMax_dmm(US) values (using dUS) were used to compared with DSkinMax, Day X values, as listed in Table 6. The resulting differences between actual Day X CT found maximum skin doses and the predicted values using dUS and trendline equation established by Day 0 virtual structures plots had a range of 11.7% to 14.8%, with an average of 0.7 6.4% and median value of 0.9% (95% CI 5 2.3 to 3.7%, 99% CI 5 3.4 to 4.7%).
Minimum skineballoon distance delta (dCT) between Day X and Day 0 CT
Frequency
d ! 5.0 mm 5.0 mm # d ! 2.0 mm 2.0 mm # d ! 0.0 mm d 5 0.0 mm 0.0 mm ! d # 2.0 mm Total
1 0 9 10 14 34
(2.9%) (0.0%) (26.5%) (29.4%) (41.2%) (100.0%)
Note that the minus sign meant the decrease of the skin bridge.
Maximum skin dose prediction using US found skin bridge change (dUS)
Y.J. Huang et al. / Brachytherapy 17 (2018) 956e965
961
Fig. 4. Data from all patients showing maximum skin dose on the original (Day 0) and the subsequent CT (Day X) based on the minimum skineballoon distance. An exponential trendline was also plotted (R2 0.9235).
Discussion Even though skin toxicity is one of the major concerns in APBI brachytherapy, no consensus is available to measure day-to-day skineballoon distance variations. This study is Table 5 Skineballoon distance delta between Day X and Day 0 results measured by CT and US
No.
CT skin distance delta (dCT, mm)
US skin distance delta (dUS, mm)
Difference between CT and US (dUSdCT) (mm)
1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 Average
0.0 2.0 1.5 1.0 1.0 2.0 1.0 1.0 1.0 2.0 0.0 0.0 1.5 1.0 0.0 2.0 1.0 1.0 2.0 2.0 2.0 0.4 1.3
0.5 4.4 0.6 0.6 0.1 1.4 0.4 0.3 0.2 4.2 0.5 1.4 0.4 0.5 0.3 1.5 0.3 0.1 1.9 0.0 1.5 0.0 1.6
0.5 2.4 2.1 0.4 0.9 0.6 1.4 1.3 0.8 2.2 0.5 1.4 1.1 0.5 0.3 0.5 1.3 0.9 0.1 2.0 0.5 0.5 1.1
US 5 ultrasound.
US and CT skin change in the same direction U U U U U U U U U U U U U
looking into using US along with the creation of virtual structures on Day 0 CT to predict maximum skin dose variation. Interfraction variation of APBI using Contura had been reported previously (7, 10, 11, 12). Kuo’s study had included detail analysis of 7 patients with many aspects of changes found during the course of MLB-based brachytherapy. The ranges of observed minimum skineballoon distances were similar to our results. It was found both in Kuo’s study and ours that the smaller initial skineballoon distances resulted in higher dosimetric impacts and tended to differ more than the fitted curve values (Fig. 4). Kuo concluded a mean decrease of 2 mm skineballoon distance could result in a 20e40% dose increase if the initial skine balloon distance is less than 12 mm. Their findings also demonstrated the importance of skineballoon distance monitoring during the treatment course to prevent the maximum skin dose exceeding the tolerance values. In our clinical experience, we observed more skineballoon distance changes on bigger, more pendulous patients. In fact, the case with 6.0 mm skin bridge change from our study was from one patient with larger breast and more loose fat tissue surrounding the balloon. We saw a big part of her breast tissue shifted back to its natural position on Day 1, which changed the geometric relationship between the balloon and the skin surface. Based on our findings, we recommend the clinics to at least take another CT before the first fraction or treatment in case there is drastic soft tissue motion that could change the dosimetric parameters, especially if the skineballoon distance is small or if the maximum skin dose is very close to the tolerable values in the original plan.
962
Y.J. Huang et al. / Brachytherapy 17 (2018) 956e965
Fig. 5. CT and US found skineballoon distance change between Day X and Day 0 plot. A positive delta value meant an increasing of skineballoon distance, which would result in lower maximum skin dose, and negative distance delta caused higher maximum skin dose. US 5 ultrasound.
When daily CT is not available, the utilization of virtual structures had been investigated previously for dose predictions based on geometric changes. In Bhatt’s study series
(10, 12), they looked into potential effect of seroma using virtual structures on CT to calculate dosimetric coverage of the target volume. This study demonstrated the possibilities to
Fig. 6. Maximum skin dose value predictions using virtual structures and Day 0 or X plan calculation comparisons for one patient. The predicated DSkinMax_dmm values were obtained from doses to SkinMax_dmm virtual structures, d 5 1, 2.7. The maximum skin dose on Day 0 and Day X CT, DSkinMax, Day 0 and DSkinMax, Day X, were also shown.
Y.J. Huang et al. / Brachytherapy 17 (2018) 956e965
963
Table 6 The comparison results between the actual maximum skin doses calculated on Day X CT vs. using US skin delta value measurement and virtual structure trendline equation method to predict maximum skin dose No.
DSkinMax,
1 2 3a 4a 5a 6a 7a 8a 9a 10a 11a 12a 13a 14a 15a 16a 17a 18 19 20 21 Average standard deviation
66.8% 77.0% 74.6% 80.2% 84.0% 92.8% 67.4% 86.2% 84.4% 61.5% 110.6% 34.0% 79.1% 93.2% 62.6% 74.4% 115.8% 102.2% 111.7% 86.0% 127.0% 84.4 20.9%
Day X
US skin distance delta (dUS, in mm)
DSkinMax_dmm(US)
Difference (DSkinMax_dmm(US)DSkinMax,
0.5 4.4 0.6 0.6 0.1 1.4 0.4 0.3 0.2 4.2 1.0 1.4 0.4 0.5 0.3 1.5 0.3 0.1 1.9 0.0 1.5 0.0 1.6
69.4% 86.3% 89.4% 77.5% 88.8% 91.4% 74.3% 87.1% 80.2% 50.9% 101.3% 36.2% 82.5% 93.2% 63.0% 77.6% 114.6% 99.2% 115.6% 94.1% 115.2% 85.0 19.4%
2.6% 9.3% 14.8% L2.6% 4.8% L1.4% 6.8% 0.9% L4.2% L10.7% L9.2% 2.2% 3.3% 0.0% 0.4% 3.2% L1.2% L3.0% 3.9% 8.1% L11.7% 0.7 6.4%
Day X)
The maximum deviation between the two methods (14.8%) was from an overestimation of the maximum skin predicted values. Minus values (in bold) demonstrated the cases when the virtual structure prediction method using US delta underestimated the maximum skin doses. Note that all of the maximum skin doses (actual or predicted) were still within the dose goal tolerance as listed in Table 1. a Patients treated with 22.5 Gy in 3-fraction regimen.
predict skin dose changes using virtual structures. Our results had shown the predicted doses using virtual structures agreed well with the actual maximum skin dose, DSkinMax, Day X (99% CI 5 3.4% to 0.0%). It was noted, however, that when the minimum skin bridge contained a continuous strip of tissue with similar thickness to the minimum skineballoon distance, and the minimum skineballoon distance was less than 7 mm, the virtual structure prediction method had resulted in the most deviation from actual DSkinMax, Day X values. Therefore, cautions should be taken when considering applying the virtual structures to estimate the skin dose variations for patients with the anatomical characteristics as described previously. In our study, we were able to establish fitted curve with its equation to predict the maximum skin dose based on the US-found skineballoon distance delta values (dUS) with an average difference of 0.7% 6.4% (95% CI 5 2.3% to 3.7%). As shown in Table 6, three cases underestimated the maximum skin doses with more than 5% differences when compared with actual DSkinMax, Day X values. Note that only one of the case’s maximum skin doses was of concern based on the dose goal of 100% (Table 1), but would still be within minimum acceptable range of #120%, as it was a 3-fraction regimen case (case No. 17). Our data had shown potentials of using US-measured skineballoon distances and CT virtual structures to provide the clinicians with additional information to evaluate the maximum skin dose without additional CT imaging acquisition.
Based on our findings, the following workflow is proposed. On Day 0 CT, after the treatment plan is approved, virtual structures, SkinMax_dmm, are created to establish a scatterplot curve and equation, as shown in Fig. 7. This Figure was plotted using data from one of the studied patients, who was prescribed with 34 Gy/10 fx regimen. The goal of the maximum skin dose is #100%, with an acceptable value of #145%. In this example, the original planned DSkinMax was 97.7%, and to exceed skin dose of more than 145%, the delta value of the minimum skineballoon distance needs to be more than 5.0 mm (~5.2 mm). Based on this plot and the projected maximum d value, we can have confidence that the maximum skin dose will be within acceptable ranges as long as the US measure dUS is less than 5.2 mm. If a more conservative measure is taken, with 15% margin (99% CI up to 5%, plus an arbitrary 10% error bar based on our institutional experience), as long as the change of the minimum skineballoon distance found on the patient is less than 3.8 mm, the maximum skin dose would be well within tolerance. This proposed workflow offers quick and simple mechanism to provide the clinicians with confidences about the quality of the treatment without adding additional imaging doses to the patients. As with other imaging modalities used in the Radiation Oncology department, the quality of the information provided by the image is highly dependent on the experience levels of the users. It is even more critical when using US to measure an unstable (breast tissue) curved surface (Contura balloon).
964
Y.J. Huang et al. / Brachytherapy 17 (2018) 956e965
Fig. 7. The maximum skin dose prediction plot according to skineballoon distance delta values. In the example, the original maximum skin dose was 97.7%; the maximum skin dose would be more than 100% (dose goal) when the delta value was about 0.4 mm (green arrow). However, to exceed the minimum acceptable value of 145%, the minimum skineballoon distance delta would need to be more than 5.2 mm (red arrow). The more conservative values (in orange) were also shown if study uncertainties (15% margin) were also included for consideration. (For interpretation of the references to color in this figure legend, the reader is referred to the Web version of this article.)
It is also important to ensure no excessive pressure is applied to the probe during US measurement, as too much pressure could introduce distorted images and affect our skin distance results. In our practice, we have instructed our staff to use enough but the lightest pressure that allows for good image quality for skin bridge determination. Furthermore, instead of searching for the global minimum skineballoon distance, the users should look for a stabilized position with a clear view of the balloon and the skin bridge for best reproducibility to find the minimum skin distance baseline. The ideal position of the US probe relative to the balloon varies based on the patient’s anatomy; and could either be parallel or perpendicular to the balloon applicator axis. We had observed more interuser differences on US-measured skineballoon distances in our first couple of treated patients. As a group, the process of obtaining US information should be well established so that the results of the US measurements are reproducible. It is important for the clinicians to understand only with good quality US measurements can this proposed workflow provide useful information to predict the maximum skin dose. With the proposed workflow, other interfraction variations, such as applicator rotation, can still occur. The clinicians should understand the possibilities that some of these variations during MLB treatment course might not be as detectable when CT is eliminated. Therefore, when concerns arise during pretreatment checks, additional CT for verification or even new plan should still be considered to ensure minimum dose variation in treatment delivery. In this study, maximum skin point dose was investigated instead of the more robust measures using D0.2cc for a 2 mm thick skin layer or D1 cc as concluded by Hilts (13). We had
intentionally chosen to keep point dose for investigation as we were not looking for clinical cosmetic representation of the skin dose value. Furthermore, the point dose matrix used in the study was a more conservative mechanism, which was reasonable for the purpose of the study.
Conclusions In conclusion, it is possible to use US to observe interfraction skineballoon distance variation to replace CT acquisition. With the proposed workflow, based on the creation of virtual structures defined on the planning CT- and US-measured skineballoon distances, the skin maximum doses can be reasonably estimated. The clinicians can determine when a new CT acquisition and a new plan are needed if the skineballoon distance delta exceeded a predefined tolerance value. This can result in elimination of extra imaging dose from CT and facilitate treatment efficiency for patients treated with Contura MLB brachytherapy.
References [1] Hepel JT, Arthur D, Shaitelman S, et al. American Brachytherapy Society consensus report for accelerated partial breast irradiation using interstitial multicatheter brachytherapy. Brachytherapy 2017;16: 919e928. [2] Shah C, Vicini F, Shaitelman SF, et al. The American Brachytherapy Society consensus statement for accelerated partial-breast irradiation. Brachytherapy 2018;17:154e170.
Y.J. Huang et al. / Brachytherapy 17 (2018) 956e965 [3] Njeh CF, Saunders MW, Langton CM. Accelerated partial breast irradiation (APBI): A review of available techniques. Radiat Oncol 2010;5:90. [4] Cuttino LW, Arthur DW, Vicini F, et al. Long-term results from the contura multilumen balloon breast brachytherapy catheter phase 4 registry trial. Int J Radiat Oncol Biol Phys 2014;90:1025e1029. [5] Akhtari M, Abboud M, Szeja S, et al. Clinical outcomes, toxicity, and cosmesis in breast cancer patients with close skin spacing treated with accelerated partial breast irradiation (APBI) using multi-lumen/catheter applicators. J Contemp Brachytherapy 2016; 8:497e504. [6] Lee K, Quillo A, Dillon D, et al. Acute toxicity and early cosmetic outcome in patients treated with multilumen balloon brachytherapy with skin spacing # 7.0 millimeters. J Contemp Brachytherapy 2012;4:8e13. [7] Kuo HC, Mehta KJ, Hong L, et al. Day to day treatment variations of accelerated partial breast brachytherapy using a multi-lumen balloon. J Contemp Brachytherapy 2014;6:68e75.
965
[8] Accelerated partial breast radiation therapy using high-dose rate brachytherapy in treating patients with early stage breast cancer after surgery - full text view - ClinicalTrials.gov. Available at: https:// clinicaltrials.gov/ct2/show/NCT02526498. Accessed May 7, 2018. [9] Arthur DW, Vicini FA, Todor DA, et al. Contura multi-lumen balloon breast brachytherapy catheter: Comparative dosimetric findings of a phase 4 trial. Int J Radiat Oncol Biol Phys 2013;86:264e269. [10] Bhatt A, Sowards K, Bhatt G, et al. Impact of interfraction seroma collection on breast brachytherapy dosimetry e a mathematical model. J Contemp Brachytherapy 2012;4:101e105. [11] Bhatt AD, Crew JB, Bhatt G, et al. Interfraction accumulation of seroma during accelerated partial breast irradiation: Preliminary results of a prospective study. Brachytherapy 2012;11:374e379. [12] Bhatt AD, Barry PN, Sowards KT, et al. Negative effect of seroma on breast balloon brachytherapy dosimetry. Pract Radiat Oncol 2014;4: e1ee5. [13] Hilts M, Halperin H, Morton D, et al. Skin dose in breast brachytherapy: Defining a robust metric. Brachytherapy 2015;14:970e978.