- Email: [email protected]
Intrapatient study comparing 3D printed bolus versus standard vinyl gel sheet bolus for postmastectomy chest wall radiation therapy
Intra-patient study comparing 3D printed bolus versus standard vinyl gel sheet bolus for postmastectomy chest wall radiation therapy James L. Robar, Kathryn Moran, James Allan, James Clancey, Tami Joseph, Krista Chytyk-Praznik, R. Lee MacDonald, John Lincoln, Parisa Sadeghi, Robert Rutledge PII: DOI: Reference:
S1879-8500(17)30384-3 doi: 10.1016/j.prro.2017.12.008 PRRO 861
To appear in:
Practical Radiation Oncology
Received date: Revised date: Accepted date:
14 September 2017 11 December 2017 20 December 2017
Please cite this article as: Robar James L., Moran Kathryn, Allan James, Clancey James, Joseph Tami, Chytyk-Praznik Krista, MacDonald R. Lee, Lincoln John, Sadeghi Parisa, Rutledge Robert, Intra-patient study comparing 3D printed bolus versus standard vinyl gel sheet bolus for postmastectomy chest wall radiation therapy, Practical Radiation Oncology (2017), doi: 10.1016/j.prro.2017.12.008
This is a PDF file of an unedited manuscript that has been accepted for publication. As a service to our customers we are providing this early version of the manuscript. The manuscript will undergo copyediting, typesetting, and review of the resulting proof before it is published in its final form. Please note that during the production process errors may be discovered which could affect the content, and all legal disclaimers that apply to the journal pertain.
ACCEPTED MANUSCRIPT Intra-patient study comparing 3D printed bolus versus standard vinyl gel sheet bolus for postmastectomy chest wall radiation therapy
RI P
T
James L. Robar1,2,3,*, Kathryn Moran3, James Allan3, James Clancey3, Tami Joseph3, Krista Chytyk-Praznik1,2,3, R. Lee MacDonald2, John Lincoln2, Parisa Sadeghi2, Robert Rutledge1,3 1
Department of Radiation Oncology, Dalhousie University, Halifax, Canada Department of Physics and Atmospheric Science, Dalhousie University, Halifax, Canada 3 Nova Scotia Health Authority, Halifax Canada * Corresponding author
NU
SC
2
MA
Running title: 3D printed bolus for chest wall radiotherapy
AC
CE
PT
ED
Conflict of interest (COI) statements: JR provides technical direction to 3D Bolus, Inc. but does not receive income. KM has provided part-time consultant services to 3D Bolus, Inc. JC, TJ, KCP, RLM, JL, PS and RR report no COI.
1
ACCEPTED MANUSCRIPT Abstract Purpose: This patient study evaluated the use of 3D printed bolus for chest wall radiotherapy, with
T
comparison to standard sheet bolus with regard to accuracy of fit, surface dose measured in vivo and
RI P
efficiency of patient setup. By alternating bolus type over the course of therapy, each patient served as her own control. Methods and materials: For sixteen patients undergoing chest wall radiotherapy,
SC
custom 5.0 mm thick bolus was designed based on the treatment planning CT and 3D printed using
NU
polylactic acid (PLA). Cone beam CT was used to image and quantify the accuracy of fit of the two bolus types with regard to air gaps between the bolus and skin. As a QA measure for the 3D printed bolus,
MA
Optically Stimulated Luminescent Dosimetry provided in vivo comparison of surface dose at seven points on the chest wall. Durations of patient setup and image guidance were recorded and compared.
ED
Results: In 13 of 16 patients the bolus was printed without user intervention and the median print time was 12.6 hours. the accuracy of fit of the bolus to the chest wall was improved significantly relative to
PT
standard sheet bolus with the frequency of air gaps 5 mm or greater reduced from 30% to 13%
CE
(p<0.0003), and maximum air gap dimension diminished from 0.5 +/- 0.3 mm to 0.3 +/- 0.3 mm, on average. Surface dose was within 3% for both standard sheet and 3D printed bolus. On average, the use
AC
of 3D printed bolus reduced the setup time from 104 to 76 seconds. Conclusions: This study demonstrates 3D printed bolus in post-mastectomy radiation therapy improves fit of the bolus and reduces patient setup time marginally compared with standard vinyl gel sheet bolus. The time savings on patient setup must be weighed against the considerable time for the 3D printing process.
Introduction Radiation therapy following mastectomy significantly improves survival in patients with high risk breast cancer (1-3). In treatment planning and delivery, a primary goal is the administration of a sufficient and uniform dose to a target volume that extends to and includes the skin (4). For either megavoltage
2
ACCEPTED MANUSCRIPT photon or electron beams, this typically necessitates the placement of a bolus on the patient surface, and commonly the bolus consists of a vinyl gel (e.g. Superflab) sheet that is uniform in thickness, or a
T
manually fabricated thermoplastic material, for example (5). Accurate fit of the bolus to the chest wall is
RI P
important because inadvertent air gaps between the bolus and the skin can reduce the skin dose in the affected areas. For example, Butson et al (6) showed that at the basal cell layer, for a 6 MV beam,
SC
underdosing can occur by up to 4% and 10% for 4 mm and 10 mm air gaps below the bolus, respectively,
NU
depending on field size and angle of incidence. Khan et al (7) demonstrated that, for a 6 MV beam, reduction of surface dose becomes significant for air gaps greater than 5 mm below Superflab bolus.
MA
Customizing the fit of the bolus to the patient may have implications with regard to complication rate: for example Anderson et al (8) compared complication rates among 85 patients who had received post-
ED
mastectomy breast reconstruction. Patients who had received a custom wax/thermoplastic mesh bolus
CE
sheet bolus (24%, p=0.05).
PT
showed a significantly lower incidence of complications (9%) versus those who had used traditional
There is growing interest in the application of 3D printing to the radiation therapy process (9-21), and a
AC
natural application of this developing technology is the generation of treatment accessories based on CT image data. While a majority of studies to date focus on phantom experiments or characterize the dosimetric properties of printable media (19, 21-24), few studies have demonstrated the utility of 3D printed bolus in application to actual patient treatment. For instance, in a series with 11 patients receiving treatment for non-melanoma skin cancer, Canters et al (9) created bolus by 3D printing a shell used as a mold which is filled with silicone rubber. This study concluded that the fabricated devices satisfied dosimetric objectives and that the 3D printing process was efficient. Zhao et al (20) described case studies using 3D printed modulated electron bolus for treatment of mycosis fungoides, photon bolus in the treatment of the nasal septum and surface applicators for HDR brachytherapy. For electron
3
ACCEPTED MANUSCRIPT therapy of the canthus for 11 patients, Lukowiak et al(25) showed that 3D printing produces bolus that improved correspondence between the bolus designed in the planning system and that placed on the
RI P
T
patient compared to wax bolus, and this facilitated improved dosimetric accuracy.
It is unknown whether 3D print detailed accessories will provide improved accuracy over accessories
SC
that are manually fabricated or generic, when applied to diverse patients in multiple treatment sites.
NU
This work presents the findings from a study of sixteen patients, where 3D printed chest wall bolus was compared directly to status quo sheet bolus (Superflab) in the same individuals with three main goals: i)
MA
to determine whether 3D printed bolus provides a more accurate fit to the patient surface based on CBCT imaging, ii) to compare the two bolus types with regard to in vivo dosimetry of dose to surface of
CE
Patient selection
PT
Methods and Materials
ED
skin, and iii) to provide comparative data regarding the efficiency of patient setup at the treatment unit.
Twenty women receiving post-mastectomy chest wall radiation therapy were selected for the study
AC
after consenting to the Research Ethics Board-approved protocol at our institution. Selection criteria included i) anticipated use of tangential fields in treatment planning and ii) requirement for bolus on alternate treatment fractions. Four patients became ineligible or withdrew from the study for various reasons. One patient exhibited excessive respiratory motion (over 15 mm amplitude) causing motion artefacts in acquired CBCT, one patient had sustained a shoulder injury (unrelated to RT treatment), and two patients exhibited apparent anatomical change between simulation and treatment delivery. No patient developed skin toxicity requiring discontinuation from the study. Of the sixteen eligible patients 12 and 4 patients were treated for right and left sided disease, respectively. The median age of the 16
4
ACCEPTED MANUSCRIPT female breast cancer patients is 61 years (range 38-83) and the median Body Mass Index is 26.1 (range
T
18.7-34.9).
RI P
Treatment planning, patient setup and bolus sequencing
Twelve patients received 42.4 Gy in 16 fractions, two patients received 45 Gy in 25 fractions, one patient
SC
received 42.5 Gy in 16 fractions and one patient received 50 Gy in 25 fractions. Patients
NU
were positioned as per departmental standards with both arms fixed above the head using an inclined breast board (Civco, Orange City, USA). Two anterior and two ipsilateral tattoos were used for initial
MA
positioning. Standard tangential fields with MLC fluence compensation were defined in the treatment planning system (Eclipse 11, Varian Medical Systems, Palo Alto, USA). Eleven patients were planned
ED
with 6 MV beams only and five patients were planned with a combination of 6 and 10 MV beams. Bolus was prescribed for alternate treatment fractions throughout the course of radiation therapy resulting in
PT
8 and 12 bolus fractions for the 16 and 25 fraction schedules, respectively. On fractions with bolus, the
CE
application of standard 5.0 mm vinyl gel sheet bolus (Superflab) and 5.0 mm 3D printed bolus was alternated. A block randomization process was used to determine which type bolus would be applied
AC
first for each patient and to guarantee an equal number of patients would be start with each type of bolus. On setup, planned shifts from the user origin were implemented, the bolus was placed, and then the CBCT image guidance was performed to finalize patient positioning.
3D printed bolus design, fabrication and quality assurance Bolus of 5.0 mm thickness was defined in the treatment planning system and exported as a Polygon File Format (PLY) object. This structure was then manipulated in 3D design software (Meshmixer, AutoDesk) to apply smoothing and to crop at the inferior edge to subsequent adhesion to the 3D printer build plate. An example bolus is shown in Figure 1A. All boluses were 3D printed using Fused Deposition
5
ACCEPTED MANUSCRIPT Modeling (FDM) of polylactic acid (PLA) filament. This material offers the advantages (20) of being nontoxic, easily cleaned during the treatment course and, compared to some other materials, faster during
T
3D printing. Previous reports (19, 21) have shown the electron density relative to water to be
RI P
approximately 1.1 to 1.2 when printed as a solid object (100% infill) with electron density(19) of 3.8 x 1023 electrons/cm3. 3D printing was performed using a Taz5 3D printer (Aleph Objects, Colorado, USA),
SC
which allows a sufficiently large print volume compatible to most chest wall boluses (maximum 290 mm
NU
x 275 mm x 250 mm). Individual boluses were variable in dimension according to patient anatomy, with the largest bolus printed having greatest dimension of approximately 29 cm. The time required to print
MA
each individual bolus was recorded, and at the conclusion of the study, the median, mean and range of durations for printing were calculated. Quality assurance (QA) was performed on the 3D printed bolus to
ED
assess spatial fidelity and material uniformity by CT imaging of the bolus in absence of the patient.
PT
Assessment of conformity of bolus to patient surface
CE
3D printed bolus was compared directly to the standard 5.0 mm thick vinyl gel sheet bolus (SuperFlab, CIVCO, Orange City, USA) on alternate bolus fractions. For the standard sheet bolus, usual protocol was
AC
followed by the radiation therapists to place the bolus as uniformly on the chest wall to minimize the extent of air gaps, e.g., tape and rolled linens were applied to assist in conforming the bolus to concavities on the patient surface as deemed necessary. Once the bolus was judged to be properly in place, CBCT was performed for image guidance of patient position. CBCT image data were subsequently analyzed to assess fit to the patient. All CBCT image datasets were analyzed post-treatment in the treatment planning system by auto-contouring all air cavities between the bolus and the skin surface using a CT tool with consistent limits between -1000 and -800 HU. Any air cavities outside of the tangential treatment fields were discarded since they would not have an effect on dose delivery. The
6
ACCEPTED MANUSCRIPT resultant air cavity structure was then reviewed to quantify the total air cavity volume and maximum air
T
cavity dimension along a normal to the skin surface.
RI P
In vivo dosimetry
At the time of treatment planning, seven positions on the patient surface were selected for placement
SC
of OSLDs (NanoDot, Landauer, USA). Three positions were selected on the mid-plane of the cranial-
NU
caudal extent of the tangent fields, one near the apex of the chest wall and two near the medial and lateral aspects of the field. Two OSLD positions were selected approximately 3 cm inferior to the cranial
MA
field border, and two more were approximately 3 cm superior to the caudal field border, as illustrated in Figure 1B. For the 3D printed bolus, pockets were created in the skin-side of the bolus for insertion of
ED
OSLDs, in order to prevent the introduction of inadvertent air gaps between the rigid PLA material and the skin. To achieve this, OSLD locations were then translated from the planning system to the mesh-
PT
handling software (Meshmixer, Autodesk). In cross-section, OSLD pockets were designed to be just
CE
larger than the dimensions of the OSLD (4 mm x 5 mm) as shown in Figure 1C. The depth of the pockets was 2 mm to place the OSLD flush with the skin-side surface, and the remainder of the bolus thickness
AC
(3.0 mm) was filled with printed PLA material. For fractions with sheet bolus, OSLDs were placed at the same locations on the chest wall. Finally, the planned dose values were recorded at these locations in order to allow determination of the ratio of measured to planned dose for both bolus types. We note that this in vivo dosimetry measurement was performed primarily as a QA measure using the new 3D printed bolus, and not to investigate the effect of possible air cavities, given that OSLDs were placed in a regular arrangement and not necessarily at locations of compromised fit of the bolus.
Duration of patient setup and image guidance To assess possible differences in efficiency of setting up patients for treatment, it was of interest to
7
ACCEPTED MANUSCRIPT perform a timing comparison between standard sheet and 3D printed bolus. For both bolus types, stopwatch times were recorded to reflect the total amount of time required for initial bolus placement.
T
The timed period began after the patient had been visually aligned with tattoos, and the therapists
RI P
initiated the bolus positioning, and ended at the point where therapists were satisfied with bolus placement and ready to proceed to image guidance. Separately, the time required for CBCT image
NU
SC
guidance was recorded.
Results
MA
3D printed bolus fabrication
All but three boluses were printed without user intervention—in these exceptional cases, the common
ED
cause of interruption was loss of adhesion to the print bed. The median and mean durations for 3D printing were 10.8 and 12.6 5.4 hours, respectively. The range was 7.2 to 28.0 hours, with four of the
PT
16 boluses for eligible patients requiring 15 hours or longer. The average mass of PLA used was 421 g
CE
resulting in a materials cost of approximately US$10 per bolus. No spatial or compositional irregularities
AC
were detected for any of the boluses as a result of the QA process.
Conformity of bolus to patient surface Figure 2 shows a frequency histogram of the CBCT-measured maximum air gap observed, including all patients and all fractions. These distributions exhibit significantly different medians (p<0.0003, Wilcoxon rank test). For the sheet bolus, approximately 30% percent of all fractions involved an air gap of over 5 mm; this is reduced to 13% with the use of 3D printed bolus. For the sheet bolus, among all observations the mean air gap was 0.5 +/- 0.3 cm for the sheet bolus; this is reduced to 0.3 +/- 0.3 cm for the 3D printed bolus. One outlying air gap of 2.2 cm was observed on a single fraction for the 3D
8
ACCEPTED MANUSCRIPT bolus; this resulted from gradual anatomical change over the treatment course, as described further in
T
the Discussion section.
RI P
In addition, the median total air gap volume differed significantly (p<0.00003, Wilcoxon rank test), with the 3D printed bolus resulting in lower values. Whereas the sheet bolus exhibits a broad distribution of
SC
air volume, the 3D printed bolus is characterized by a peak just above zero, with a rapid decline to single
NU
observations above 2 cc. Over all observations, the mean total air gap volumes were 5.4 +/- 9 cc and 4.0 +/ 11 cc for sheet bolus and 3D printed bolus, respectively. Outliers of total volume are possible for
MA
either bolus type, however with reference to Figure 2, for the 3D printed bolus this total volume is comprised of multiple, small air cavities (mostly < 5 mm in dimension) whereas for the sheet bolus larger
ED
individual air gaps occurred. Qualitative review showed that these larger air gaps occurred at regions of surface complexity, and in particular, locations featuring large convexities or concavities. Examples are
CE
In vivo dosimetry
PT
shown in Figure 3.
AC
Figure 4 shows frequency histograms of all OSLD measurements made with sheet bolus and 3D printed bolus, for all patients in the study. The bimodal distribution reflects the different fractionations schemes used. On average, the ratio of dose for 3D printed bolus to sheet bolus was 0.98 +/- 0.05. OSLD-measured surface dose data were also analyzed for each patient independently by calculating the mean dose over the treatment course, for corresponding OSLD locations, and comparing between the two bolus types. On average, the ratio of the mean dose with 3D printed bolus to sheet bolus is 0.97 +/0.04. The deviation of these dose ratios from unity is explained by a difference in the geometry of the OSLD detector below the bolus; for the sheet bolus, the OSLD is located below 5 mm of Superflab, and for the 3D printed bolus, the detector is located within a pocket below 3 mm of PLA. It was also of
9
ACCEPTED MANUSCRIPT interest to determine whether there would be a difference in variability in measured dose, for 3D printed bolus compared to sheet bolus. The mean coefficients of variation were 0.029 and 0.028 for 3D
T
printed and sheet bolus, respectively, and no significant difference was observed (p=0.95, Wilcoxon rank
RI P
test) over the patient sample. Finally, the ratios of OSLD-measured to planned dose (Dmeasured/Dplanned) at each of the seven locations on the chest wall are shown in the box plots of Figure 5. The mean absolute
SC
deviation of this ratio from unity was higher for sheet bolus (0.04) and compared to 3D printed bolus
NU
(0.02) however differences between bolus type did not reach statistical significance (p=0.10, Wilcoxon
MA
rank test).
Duration of patient setup and image guidance
ED
Among all patients, the ratio of setup time (sheet bolus to 3D printed bolus) was reduced significantly (p<0.000005, Wilcoxon rank test), by a factor of 1.42 on average. On average, the mean time reduced
PT
from 104 to 76 seconds. With regard to individual patients, 14 of 16 patients showed reduced setup
CE
time with the use of 3D printed bolus. On the other hand, the image guidance duration was not
Discussion
AC
significantly different between bolus types.
This unique study examined 16 patients breast cancer patients, using each as her own control to assess 3D printed chest wall bolus compared to standard vinyl gel sheet bolus. A decrease of the frequency of air gaps 5 mm or greater from 30% with sheet bolus to 13% for 3D printed bolus was observed. The dosimetric consequence of these larger air gaps on the surface of the chest wall will depend on multiple parameters including energy, field size, beam modifiers and angle of incidence. A key determinant of dose error is field size, whereby a larger dosimetric error result as the field size is reduced. Referring to the study by Khan et al(7), for a 5x5 cm2 field, air gaps 5 mm or greater between Superflab and skin
10
ACCEPTED MANUSCRIPT produced significant dosimetric error, but this became insignificant for 10x10 cm2 fields. Thus, the error may become more important with increasing adoption of highly conformal techniques for chest wall
T
radiotherapy such as IMRT or VMAT (26, 27) where the overall dose is delivered using a series of smaller
RI P
MLC defined apertures.
SC
While 3D printed bolus produced a better fit on average, Figure 2 shows a single outlier where an air gap
NU
of over 2 cm was observed. In this case the patient exhibited gradual but persistent anatomical change between the first and last fraction such that bolus did not fit by the end of the final fraction (Figure 6).
MA
Anatomical change, as opposed to mis-positioning of the bolus, was confirmed for this case by coregistering and comparing sequential CBCTs over the treatment course. The research protocol did not
ED
permit redesign of the 3D printed bolus mid-treatment course. However, in practice, 3D printing allows for the unique concept of “adaptive bolus”, whereby CBCT imaging may be monitored after each
CE
using these image data.
PT
fraction, and the appearance of a significant air gap can trigger generation of a new 3D printed bolus
AC
Since this is the first study using 3D printed chest wall bolus, in vivo dosimetry was deemed important as a QA measure. Compared to the sheet bolus, 3D printed bolus yielded dose measurements that were lower by approximately 2 to 3% on average. This can be explained by the fact that for sheet bolus, the OSLD NanoDot was located below 5 mm Superflab material, and for the 3D printed bolus, below 3 mm PLA. According to the data of Burleson et al(19), if the physical depth were equal at 5 mm, these two materials would have approximately the same Tissue Maximum Ratio (TMR) value. With the reduced thickness of buildup material for the 3D printed bolus, the measured dose should be reduced, but the disparity is complicated by the variable obliquity of fields which will determine the actual difference in ray-line depth to the OSLD. We emphasize that the purpose of the in vivo dosimetry was to verify
11
ACCEPTED MANUSCRIPT approximate equivalence of the bolus types, not to quantify the relationship between air gap dimension and dose error, as this has been done elsewhere (6,7). Spatial coincidence between OSLD position and
T
air gaps below the bolus were infrequent, and thus the improvement in fit for the 3D printed bolus
RI P
should not be reflected in the OSLD measurements. While differences were seen between measured and planned doses were observed (Figure 5), the magnitude of these differences are comparable to
SC
previous work using nanoDot OSLDs on the chest wall (28). On average the Dmeasured/Dplanned ratio was
NU
higher than unity, and this is consistent with the finding that at depths in the vicinity of 5 mm, Eclipse AAA dose calculations are systematically low for beams oblique to the surface by approximately 2-4%
MA
compared to Monte Carlo calculations(29). The OSLD methodology applied here shows that 3D printed bolus offers the option of in-printing pockets for dosimeters, which addresses two issues: first, it avoids
ED
inadvertently creating an air gap by introducing a dosimeter between the bolus material and the skin.
CE
tattoo marks on the patient.
PT
Second, it facilitates standardization of the locations of measurements, e.g., without making multiple
With regard to surface dose, although not explored in this study, 3D printing of photon bolus allows
AC
customization of thickness, rather than selecting from a range of generic thicknesses of sheet bolus. Thus, dose build-up may be customized by the practitioner, and in concept could be variable by location on the patient. For example, where a tangential field arrangement delivers dose medial or lateral to the PTV, the bolus could be thinned, or the printed density (infill) could be selectively reduced. This could be useful where the radiation oncologist is concerned with severe skin toxicity during chest wall radiotherapy (30), or where an alternating bolus approach is to be replaced with a customized single thickness of bolus for all fractions (31), for example.
12
ACCEPTED MANUSCRIPT Grading of skin reaction was not an end-point of this study because for each patient, bolus type was alternated, and thus observations would reflect a combination of bolus type. However, among all 16
T
patients assessed weekly on treatment by their Radiation Oncologist, the maximum skin reaction as
RI P
graded on the NCI CTCAE V4.03 was as follows: One patient grade 3 (moist desquamation), two patients grade 2 (marked erythema), and 13 patients with mild or no dermatitis. This distribution of skin reaction
NU
technique chest wall radiotherapy in our department.
SC
for women undergoing chest wall radiotherapy was consistent with that in women receiving standard
MA
There are some drawbacks to the 3D printing approach, notably the time required to fabricate the bolus (over 12 hours, on average), and minimizing user-intervention of this step (i.e. to allow printing
ED
overnight) requires a high-quality 3D printer with low failure-rate. In addition, time is required for QA on the fabricated bolus. This QA may include physical measurement of dimensions with comparison to
PT
the design, as well as CT imaging (or CBCT imaging prior to treatment delivery at first fraction) to verify
CE
homogeneity of the bolus medium. Finally, as illustrated above, for this study a very rigid printable medium (PLA) was used. Although personalized to the surface of the patient, this material does not flex
AC
with minor patient deformation or anatomical change; therefore, further research will include investigation of more flexible materials.
Conclusion This study demonstrates the practical use of 3D printed bolus for post-mastectomy radiation therapy. Among a sample of 16 patients each treated at least four times with 3D bolus, the accuracy of fit of the 3D bolus to the chest wall was improved relative to standard sheet (Superflab) bolus, with the frequency of air gaps 5 mm or greater reduced from 30% to 13%. Surface dose, as measured using in vivo dosimetry, was within 3% for sheet and 3D printed bolus and there was no significant difference
13
ACCEPTED MANUSCRIPT between sheet and 3D printed bolus with regard to agreement with the treatment planning system. The setup time was reduced only marginally with 3D printed bolus (104 to 76 seconds), however this time
T
saving must be weighed against the considerable time for fabrication of the bolus (median 10.8 hours,
RI P
mean 12.6 5.4 hours) and related quality assurance, although the printing process was found to be
SC
largely automated.
NU
References
MA
1. Overgaard M, Hansen PS, Overgaard J, et al. Postoperative radiotherapy in high-risk premenopausal women with breast cancer who receive adjuvant chemotherapy. Danish Breast Cancer Cooperative
ED
Group 82b Trial. N. Engl. J. Med. 1997;337:949–955.
PT
2. Ragaz J, Jackson SM, Le N, et al. Adjuvant radiotherapy and chemotherapy in node-positive
CE
premenopausal women with breast cancer. N. Engl. J. Med. 1997;337:956–962.
3. Overgaard M, Nielsen HM, Overgaard J. Is the benefit of postmastectomy irradiation limited to
AC
patients with four or more positive nodes, as recommended in international consensus reports? A subgroup analysis of the DBCG 82 b&c randomized trials. Radiother Oncol. 2007;82:247–253.
4. White J, Tai A, Arthur D, et al. Radiation Therapy Oncology Group (RTOG) Breast cancer atlas for radiation therapy planning: consensus definitions. RTOG; 2009.
5. Hsu S-H, Roberson PL, Chen Y, et al. Assessment of skin dose for breast chest wall radiotherapy as a function of bolus material. Phys Med Biol. 2008;53:2593–2606.
6. Butson MJ, Cheung T, Yu P, et al. Effects on skin dose from unwanted air gaps under bolus in photon beam radiotherapy. Radiation Measurements. 2000;32:201–204.
14
ACCEPTED MANUSCRIPT 7. Khan Y, Villarreal-Barajas JE, Udowicz M, et al. Clinical and Dosimetric Implications of Air Gaps between Bolus and Skin Surface during Radiation Therapy. JCT. 2013;04:1251–1255.
RI P
T
8. Anderson PR, Hanlon AL, McNeeley SW, et al. Low complication rates are achievable after postmastectomy breast reconstruction and radiation therapy. International Journal of Radiation
SC
Oncology*Biology*Physics. 2004;59:1080–1087.
NU
9. Canters RA, Lips IM, Wendling M, et al. Clinical implementation of 3D printing in the construction of patient specific bolus for electron beam radiotherapy for non-melanoma skin cancer. Radiother Oncol.
MA
2016;121:148–153.
10. Dancewicz OL, Sylvander SR, Markwell TS, et al. Radiological properties of 3D printed materials in
ED
kilovoltage and megavoltage photon beams. Phys Med. 2017;38:111–118.
PT
11. Fujimoto K, Shiinoki T, Yuasa Y, et al. Efficacy of patient-specific bolus created using three-
CE
dimensional printing technique in photon radiotherapy. Phys Med. 2017;38:1–9.
AC
12. Hongtao Z, Xuemin D, Huimin Y, et al. Dosimetry study of three-dimensional print template-guided precision (125)I seed implantation. J Cancer Res Ther. 2016;12:C159–C165.
13. Ji Z, Jiang Y, Guo F, et al. Dosimetry verification of radioactive seed implantation for malignant tumors assisted by 3D printing individual templates and CT guidance. Appl Radiat Isot. 2017;124:68–74.
14. Kim M-J, Lee S-R, Lee M-Y, et al. Characterization of 3D printing techniques: Toward patient specific quality assurance spine-shaped phantom for stereotactic body radiation therapy. Chung CK, ed. PLoS ONE. 2017;12:e0176227.
15. Lee M-Y, Han B, Jenkins C, et al. A depth-sensing technique on 3D-printed compensator for total body irradiation patient measurement and treatment planning. Med Phys. 2016;43:6137–6144.
15
ACCEPTED MANUSCRIPT 16. Lindsay C, Kumlin J, Martinez DM, et al. Design and application of 3D-printed stepless beam modulators in proton therapy. Phys Med Biol. 2016;61:N276–90.
RI P
T
17. Park JW, Oh SA, Yea JW, et al. Fabrication of malleable three-dimensional-printed customized bolus using three-dimensional scanner. Sherman JH, ed. PLoS ONE. 2017;12:e0177562.
SC
18. Ricotti R, Ciardo D, Pansini F, et al. Dosimetric characterization of 3D printed bolus at different infill
NU
percentage for external photon beam radiotherapy. Phys Med. 2017;39:25–32.
19. Burleson S, Baker J, Hsia AT, et al. Use of 3D printers to create a patient-specific 3D bolus for external
MA
beam therapy. J Appl Clin Med Phys. 2015;16:166–178.
ED
20. Zhao Y, Moran K, Yewondwossen M, et al. Clinical applications of 3-dimensional printing in radiation
PT
therapy. Med Dosim. 2017;42:150–155.
CE
21. **** (redacted for blinding during review)
22. Park S-Y, Choi CH, Park JM, et al. A Patient-Specific Polylactic Acid Bolus Made by a 3D Printer for
AC
Breast Cancer Radiation Therapy. Deutsch E, ed. PLoS ONE. 2016;11:e0168063.
23. Lindsay C, Kumlin J, Jirasek A, et al. 3D printed plastics for beam modulation in proton therapy. Phys Med Biol. 2015;60:N231–40.
24. Ju SG, Kim MK, Hong C-S, et al. New technique for developing a proton range compensator with use of a 3-dimensional printer. Int. J. Radiat. Oncol. Biol. Phys. 2014;88:453–458.
25. Łukowiak M, Jezierska K, Boehlke M, et al. Utilization of a 3D printer to fabricate boluses used for electron therapy of skin lesions of the eye canthi. J Appl Clin Med Phys. 2017;18:76–81.
26. Zhang Q, Yu XL, Hu WG, et al. Dosimetric comparison for volumetric modulated arc therapy and
16
ACCEPTED MANUSCRIPT intensity-modulated radiotherapy on the left-sided chest wall and internal mammary nodes irradiation in treating post-mastectomy breast cancer. Radiol Oncol. 2015;49:91–98.
RI P
T
27. Munshi A, Sarkar B, Anbazhagan S, et al. Short tangential arcs in VMAT based breast and chest wall radiotherapy lead to conformity of the breast dose with lesser cardiac and lung doses: a prospective
SC
study of breast conservation and mastectomy patients. Australas Phys Eng Sci Med. 2017;23:433–8.
NU
28. Yusof, F. H. et al. On the Use of Optically Stimulated Luminescent Dosimeter for Surface Dose Measurement during Radiotherapy. PLoS One. 2105;10: e0128544.
MA
29. Panettieri, V., Barsoum, P., Westermark, M., Brualla, L. & Lax, I. AAA and PBC calculation accuracy in the surface build-up region in tangential beam treatments. Phantom and breast case study with the
ED
Monte Carlo code PENELOPE. Radiother Oncol 2009;93:94–101.
PT
30. Pignol J-P, Vu TTT, Mitera G, et al. Prospective evaluation of severe skin toxicity and pain during
CE
postmastectomy radiation therapy. Int. J. Radiat. Oncol. Biol. Phys. 2015;91:157–164.
AC
31. Ordonez-Sanz C, Bowles S, Hirst A, et al. A single plan solution to chest wall radiotherapy with bolus? BJR. 2014;87:20140035–9.
Figure Captions Figure 1 (A) Example of a 3D printed bolus, (B) locations of seven OSLDs (labeled A to G) relative to patient and tangential fields, and (C) embedded OSLD contained by in-printed pocket, where OSLD is flush with the skin-side of the bolus surface. Figure 2 Frequency histogram of maximum dimension of airgap between bolus and patient surface as measured on CBCT, comparing sheet bolus with 3D printed bolus.
17
ACCEPTED MANUSCRIPT Figure 3 Examples of differences in accuracy of fit between Superflab sheet bolus (left) and 3D printed bolus (right) for patients with convexity (Patient 4, top row) or concavity (Patient 13, bottom row).
T
Figure 4 Frequency histogram of OSLD-measured dose values for sheet bolus and 3D printed bolus,
RI P
including all measurements, for all OSLDs and patients.
Figure 5 Ratio of measured over planned dose at each of the seven OSLD locations, averaged over all
NU
OSLDs (indicated by A to F) are depicted in Figure 1.
SC
fractions and patients, comparing sheet bolus (top) with 3D printed bolus (bottom). The locations of the
Figure 6 Images explaining the outlying result in Figure 2. for 3D printed bolus. The patient exhibited
AC
CE
PT
ED
fraction (B) causing a loss of accuracy of fit.
MA
significant anatomical change such that the final fraction (A) differed significantly from that in the first
18
SC
RI P
T
ACCEPTED MANUSCRIPT
AC
CE
PT
ED
MA
NU
Fig. 1
19
ED
MA
NU
SC
RI P
T
ACCEPTED MANUSCRIPT
AC
CE
PT
Fig. 2
20
MA
NU
SC
RI P
T
ACCEPTED MANUSCRIPT
AC
CE
PT
ED
Fig. 3
21
NU
SC
RI P
T
ACCEPTED MANUSCRIPT
AC
CE
PT
ED
MA
Fig. 4
22
AC
Fig. 5
CE
PT
ED
MA
NU
SC
RI P
T
ACCEPTED MANUSCRIPT
23
MA
NU
SC
RI P
T
ACCEPTED MANUSCRIPT
AC
CE
PT
ED
Fig.6
24
S1879-8500(17)30384-3 doi: 10.1016/j.prro.2017.12.008 PRRO 861
To appear in:
Practical Radiation Oncology
Received date: Revised date: Accepted date:
14 September 2017 11 December 2017 20 December 2017
Please cite this article as: Robar James L., Moran Kathryn, Allan James, Clancey James, Joseph Tami, Chytyk-Praznik Krista, MacDonald R. Lee, Lincoln John, Sadeghi Parisa, Rutledge Robert, Intra-patient study comparing 3D printed bolus versus standard vinyl gel sheet bolus for postmastectomy chest wall radiation therapy, Practical Radiation Oncology (2017), doi: 10.1016/j.prro.2017.12.008
This is a PDF file of an unedited manuscript that has been accepted for publication. As a service to our customers we are providing this early version of the manuscript. The manuscript will undergo copyediting, typesetting, and review of the resulting proof before it is published in its final form. Please note that during the production process errors may be discovered which could affect the content, and all legal disclaimers that apply to the journal pertain.
ACCEPTED MANUSCRIPT Intra-patient study comparing 3D printed bolus versus standard vinyl gel sheet bolus for postmastectomy chest wall radiation therapy
RI P
T
James L. Robar1,2,3,*, Kathryn Moran3, James Allan3, James Clancey3, Tami Joseph3, Krista Chytyk-Praznik1,2,3, R. Lee MacDonald2, John Lincoln2, Parisa Sadeghi2, Robert Rutledge1,3 1
Department of Radiation Oncology, Dalhousie University, Halifax, Canada Department of Physics and Atmospheric Science, Dalhousie University, Halifax, Canada 3 Nova Scotia Health Authority, Halifax Canada * Corresponding author
NU
SC
2
MA
Running title: 3D printed bolus for chest wall radiotherapy
AC
CE
PT
ED
Conflict of interest (COI) statements: JR provides technical direction to 3D Bolus, Inc. but does not receive income. KM has provided part-time consultant services to 3D Bolus, Inc. JC, TJ, KCP, RLM, JL, PS and RR report no COI.
1
ACCEPTED MANUSCRIPT Abstract Purpose: This patient study evaluated the use of 3D printed bolus for chest wall radiotherapy, with
T
comparison to standard sheet bolus with regard to accuracy of fit, surface dose measured in vivo and
RI P
efficiency of patient setup. By alternating bolus type over the course of therapy, each patient served as her own control. Methods and materials: For sixteen patients undergoing chest wall radiotherapy,
SC
custom 5.0 mm thick bolus was designed based on the treatment planning CT and 3D printed using
NU
polylactic acid (PLA). Cone beam CT was used to image and quantify the accuracy of fit of the two bolus types with regard to air gaps between the bolus and skin. As a QA measure for the 3D printed bolus,
MA
Optically Stimulated Luminescent Dosimetry provided in vivo comparison of surface dose at seven points on the chest wall. Durations of patient setup and image guidance were recorded and compared.
ED
Results: In 13 of 16 patients the bolus was printed without user intervention and the median print time was 12.6 hours. the accuracy of fit of the bolus to the chest wall was improved significantly relative to
PT
standard sheet bolus with the frequency of air gaps 5 mm or greater reduced from 30% to 13%
CE
(p<0.0003), and maximum air gap dimension diminished from 0.5 +/- 0.3 mm to 0.3 +/- 0.3 mm, on average. Surface dose was within 3% for both standard sheet and 3D printed bolus. On average, the use
AC
of 3D printed bolus reduced the setup time from 104 to 76 seconds. Conclusions: This study demonstrates 3D printed bolus in post-mastectomy radiation therapy improves fit of the bolus and reduces patient setup time marginally compared with standard vinyl gel sheet bolus. The time savings on patient setup must be weighed against the considerable time for the 3D printing process.
Introduction Radiation therapy following mastectomy significantly improves survival in patients with high risk breast cancer (1-3). In treatment planning and delivery, a primary goal is the administration of a sufficient and uniform dose to a target volume that extends to and includes the skin (4). For either megavoltage
2
ACCEPTED MANUSCRIPT photon or electron beams, this typically necessitates the placement of a bolus on the patient surface, and commonly the bolus consists of a vinyl gel (e.g. Superflab) sheet that is uniform in thickness, or a
T
manually fabricated thermoplastic material, for example (5). Accurate fit of the bolus to the chest wall is
RI P
important because inadvertent air gaps between the bolus and the skin can reduce the skin dose in the affected areas. For example, Butson et al (6) showed that at the basal cell layer, for a 6 MV beam,
SC
underdosing can occur by up to 4% and 10% for 4 mm and 10 mm air gaps below the bolus, respectively,
NU
depending on field size and angle of incidence. Khan et al (7) demonstrated that, for a 6 MV beam, reduction of surface dose becomes significant for air gaps greater than 5 mm below Superflab bolus.
MA
Customizing the fit of the bolus to the patient may have implications with regard to complication rate: for example Anderson et al (8) compared complication rates among 85 patients who had received post-
ED
mastectomy breast reconstruction. Patients who had received a custom wax/thermoplastic mesh bolus
CE
sheet bolus (24%, p=0.05).
PT
showed a significantly lower incidence of complications (9%) versus those who had used traditional
There is growing interest in the application of 3D printing to the radiation therapy process (9-21), and a
AC
natural application of this developing technology is the generation of treatment accessories based on CT image data. While a majority of studies to date focus on phantom experiments or characterize the dosimetric properties of printable media (19, 21-24), few studies have demonstrated the utility of 3D printed bolus in application to actual patient treatment. For instance, in a series with 11 patients receiving treatment for non-melanoma skin cancer, Canters et al (9) created bolus by 3D printing a shell used as a mold which is filled with silicone rubber. This study concluded that the fabricated devices satisfied dosimetric objectives and that the 3D printing process was efficient. Zhao et al (20) described case studies using 3D printed modulated electron bolus for treatment of mycosis fungoides, photon bolus in the treatment of the nasal septum and surface applicators for HDR brachytherapy. For electron
3
ACCEPTED MANUSCRIPT therapy of the canthus for 11 patients, Lukowiak et al(25) showed that 3D printing produces bolus that improved correspondence between the bolus designed in the planning system and that placed on the
RI P
T
patient compared to wax bolus, and this facilitated improved dosimetric accuracy.
It is unknown whether 3D print detailed accessories will provide improved accuracy over accessories
SC
that are manually fabricated or generic, when applied to diverse patients in multiple treatment sites.
NU
This work presents the findings from a study of sixteen patients, where 3D printed chest wall bolus was compared directly to status quo sheet bolus (Superflab) in the same individuals with three main goals: i)
MA
to determine whether 3D printed bolus provides a more accurate fit to the patient surface based on CBCT imaging, ii) to compare the two bolus types with regard to in vivo dosimetry of dose to surface of
CE
Patient selection
PT
Methods and Materials
ED
skin, and iii) to provide comparative data regarding the efficiency of patient setup at the treatment unit.
Twenty women receiving post-mastectomy chest wall radiation therapy were selected for the study
AC
after consenting to the Research Ethics Board-approved protocol at our institution. Selection criteria included i) anticipated use of tangential fields in treatment planning and ii) requirement for bolus on alternate treatment fractions. Four patients became ineligible or withdrew from the study for various reasons. One patient exhibited excessive respiratory motion (over 15 mm amplitude) causing motion artefacts in acquired CBCT, one patient had sustained a shoulder injury (unrelated to RT treatment), and two patients exhibited apparent anatomical change between simulation and treatment delivery. No patient developed skin toxicity requiring discontinuation from the study. Of the sixteen eligible patients 12 and 4 patients were treated for right and left sided disease, respectively. The median age of the 16
4
ACCEPTED MANUSCRIPT female breast cancer patients is 61 years (range 38-83) and the median Body Mass Index is 26.1 (range
T
18.7-34.9).
RI P
Treatment planning, patient setup and bolus sequencing
Twelve patients received 42.4 Gy in 16 fractions, two patients received 45 Gy in 25 fractions, one patient
SC
received 42.5 Gy in 16 fractions and one patient received 50 Gy in 25 fractions. Patients
NU
were positioned as per departmental standards with both arms fixed above the head using an inclined breast board (Civco, Orange City, USA). Two anterior and two ipsilateral tattoos were used for initial
MA
positioning. Standard tangential fields with MLC fluence compensation were defined in the treatment planning system (Eclipse 11, Varian Medical Systems, Palo Alto, USA). Eleven patients were planned
ED
with 6 MV beams only and five patients were planned with a combination of 6 and 10 MV beams. Bolus was prescribed for alternate treatment fractions throughout the course of radiation therapy resulting in
PT
8 and 12 bolus fractions for the 16 and 25 fraction schedules, respectively. On fractions with bolus, the
CE
application of standard 5.0 mm vinyl gel sheet bolus (Superflab) and 5.0 mm 3D printed bolus was alternated. A block randomization process was used to determine which type bolus would be applied
AC
first for each patient and to guarantee an equal number of patients would be start with each type of bolus. On setup, planned shifts from the user origin were implemented, the bolus was placed, and then the CBCT image guidance was performed to finalize patient positioning.
3D printed bolus design, fabrication and quality assurance Bolus of 5.0 mm thickness was defined in the treatment planning system and exported as a Polygon File Format (PLY) object. This structure was then manipulated in 3D design software (Meshmixer, AutoDesk) to apply smoothing and to crop at the inferior edge to subsequent adhesion to the 3D printer build plate. An example bolus is shown in Figure 1A. All boluses were 3D printed using Fused Deposition
5
ACCEPTED MANUSCRIPT Modeling (FDM) of polylactic acid (PLA) filament. This material offers the advantages (20) of being nontoxic, easily cleaned during the treatment course and, compared to some other materials, faster during
T
3D printing. Previous reports (19, 21) have shown the electron density relative to water to be
RI P
approximately 1.1 to 1.2 when printed as a solid object (100% infill) with electron density(19) of 3.8 x 1023 electrons/cm3. 3D printing was performed using a Taz5 3D printer (Aleph Objects, Colorado, USA),
SC
which allows a sufficiently large print volume compatible to most chest wall boluses (maximum 290 mm
NU
x 275 mm x 250 mm). Individual boluses were variable in dimension according to patient anatomy, with the largest bolus printed having greatest dimension of approximately 29 cm. The time required to print
MA
each individual bolus was recorded, and at the conclusion of the study, the median, mean and range of durations for printing were calculated. Quality assurance (QA) was performed on the 3D printed bolus to
ED
assess spatial fidelity and material uniformity by CT imaging of the bolus in absence of the patient.
PT
Assessment of conformity of bolus to patient surface
CE
3D printed bolus was compared directly to the standard 5.0 mm thick vinyl gel sheet bolus (SuperFlab, CIVCO, Orange City, USA) on alternate bolus fractions. For the standard sheet bolus, usual protocol was
AC
followed by the radiation therapists to place the bolus as uniformly on the chest wall to minimize the extent of air gaps, e.g., tape and rolled linens were applied to assist in conforming the bolus to concavities on the patient surface as deemed necessary. Once the bolus was judged to be properly in place, CBCT was performed for image guidance of patient position. CBCT image data were subsequently analyzed to assess fit to the patient. All CBCT image datasets were analyzed post-treatment in the treatment planning system by auto-contouring all air cavities between the bolus and the skin surface using a CT tool with consistent limits between -1000 and -800 HU. Any air cavities outside of the tangential treatment fields were discarded since they would not have an effect on dose delivery. The
6
ACCEPTED MANUSCRIPT resultant air cavity structure was then reviewed to quantify the total air cavity volume and maximum air
T
cavity dimension along a normal to the skin surface.
RI P
In vivo dosimetry
At the time of treatment planning, seven positions on the patient surface were selected for placement
SC
of OSLDs (NanoDot, Landauer, USA). Three positions were selected on the mid-plane of the cranial-
NU
caudal extent of the tangent fields, one near the apex of the chest wall and two near the medial and lateral aspects of the field. Two OSLD positions were selected approximately 3 cm inferior to the cranial
MA
field border, and two more were approximately 3 cm superior to the caudal field border, as illustrated in Figure 1B. For the 3D printed bolus, pockets were created in the skin-side of the bolus for insertion of
ED
OSLDs, in order to prevent the introduction of inadvertent air gaps between the rigid PLA material and the skin. To achieve this, OSLD locations were then translated from the planning system to the mesh-
PT
handling software (Meshmixer, Autodesk). In cross-section, OSLD pockets were designed to be just
CE
larger than the dimensions of the OSLD (4 mm x 5 mm) as shown in Figure 1C. The depth of the pockets was 2 mm to place the OSLD flush with the skin-side surface, and the remainder of the bolus thickness
AC
(3.0 mm) was filled with printed PLA material. For fractions with sheet bolus, OSLDs were placed at the same locations on the chest wall. Finally, the planned dose values were recorded at these locations in order to allow determination of the ratio of measured to planned dose for both bolus types. We note that this in vivo dosimetry measurement was performed primarily as a QA measure using the new 3D printed bolus, and not to investigate the effect of possible air cavities, given that OSLDs were placed in a regular arrangement and not necessarily at locations of compromised fit of the bolus.
Duration of patient setup and image guidance To assess possible differences in efficiency of setting up patients for treatment, it was of interest to
7
ACCEPTED MANUSCRIPT perform a timing comparison between standard sheet and 3D printed bolus. For both bolus types, stopwatch times were recorded to reflect the total amount of time required for initial bolus placement.
T
The timed period began after the patient had been visually aligned with tattoos, and the therapists
RI P
initiated the bolus positioning, and ended at the point where therapists were satisfied with bolus placement and ready to proceed to image guidance. Separately, the time required for CBCT image
NU
SC
guidance was recorded.
Results
MA
3D printed bolus fabrication
All but three boluses were printed without user intervention—in these exceptional cases, the common
ED
cause of interruption was loss of adhesion to the print bed. The median and mean durations for 3D printing were 10.8 and 12.6 5.4 hours, respectively. The range was 7.2 to 28.0 hours, with four of the
PT
16 boluses for eligible patients requiring 15 hours or longer. The average mass of PLA used was 421 g
CE
resulting in a materials cost of approximately US$10 per bolus. No spatial or compositional irregularities
AC
were detected for any of the boluses as a result of the QA process.
Conformity of bolus to patient surface Figure 2 shows a frequency histogram of the CBCT-measured maximum air gap observed, including all patients and all fractions. These distributions exhibit significantly different medians (p<0.0003, Wilcoxon rank test). For the sheet bolus, approximately 30% percent of all fractions involved an air gap of over 5 mm; this is reduced to 13% with the use of 3D printed bolus. For the sheet bolus, among all observations the mean air gap was 0.5 +/- 0.3 cm for the sheet bolus; this is reduced to 0.3 +/- 0.3 cm for the 3D printed bolus. One outlying air gap of 2.2 cm was observed on a single fraction for the 3D
8
ACCEPTED MANUSCRIPT bolus; this resulted from gradual anatomical change over the treatment course, as described further in
T
the Discussion section.
RI P
In addition, the median total air gap volume differed significantly (p<0.00003, Wilcoxon rank test), with the 3D printed bolus resulting in lower values. Whereas the sheet bolus exhibits a broad distribution of
SC
air volume, the 3D printed bolus is characterized by a peak just above zero, with a rapid decline to single
NU
observations above 2 cc. Over all observations, the mean total air gap volumes were 5.4 +/- 9 cc and 4.0 +/ 11 cc for sheet bolus and 3D printed bolus, respectively. Outliers of total volume are possible for
MA
either bolus type, however with reference to Figure 2, for the 3D printed bolus this total volume is comprised of multiple, small air cavities (mostly < 5 mm in dimension) whereas for the sheet bolus larger
ED
individual air gaps occurred. Qualitative review showed that these larger air gaps occurred at regions of surface complexity, and in particular, locations featuring large convexities or concavities. Examples are
CE
In vivo dosimetry
PT
shown in Figure 3.
AC
Figure 4 shows frequency histograms of all OSLD measurements made with sheet bolus and 3D printed bolus, for all patients in the study. The bimodal distribution reflects the different fractionations schemes used. On average, the ratio of dose for 3D printed bolus to sheet bolus was 0.98 +/- 0.05. OSLD-measured surface dose data were also analyzed for each patient independently by calculating the mean dose over the treatment course, for corresponding OSLD locations, and comparing between the two bolus types. On average, the ratio of the mean dose with 3D printed bolus to sheet bolus is 0.97 +/0.04. The deviation of these dose ratios from unity is explained by a difference in the geometry of the OSLD detector below the bolus; for the sheet bolus, the OSLD is located below 5 mm of Superflab, and for the 3D printed bolus, the detector is located within a pocket below 3 mm of PLA. It was also of
9
ACCEPTED MANUSCRIPT interest to determine whether there would be a difference in variability in measured dose, for 3D printed bolus compared to sheet bolus. The mean coefficients of variation were 0.029 and 0.028 for 3D
T
printed and sheet bolus, respectively, and no significant difference was observed (p=0.95, Wilcoxon rank
RI P
test) over the patient sample. Finally, the ratios of OSLD-measured to planned dose (Dmeasured/Dplanned) at each of the seven locations on the chest wall are shown in the box plots of Figure 5. The mean absolute
SC
deviation of this ratio from unity was higher for sheet bolus (0.04) and compared to 3D printed bolus
NU
(0.02) however differences between bolus type did not reach statistical significance (p=0.10, Wilcoxon
MA
rank test).
Duration of patient setup and image guidance
ED
Among all patients, the ratio of setup time (sheet bolus to 3D printed bolus) was reduced significantly (p<0.000005, Wilcoxon rank test), by a factor of 1.42 on average. On average, the mean time reduced
PT
from 104 to 76 seconds. With regard to individual patients, 14 of 16 patients showed reduced setup
CE
time with the use of 3D printed bolus. On the other hand, the image guidance duration was not
Discussion
AC
significantly different between bolus types.
This unique study examined 16 patients breast cancer patients, using each as her own control to assess 3D printed chest wall bolus compared to standard vinyl gel sheet bolus. A decrease of the frequency of air gaps 5 mm or greater from 30% with sheet bolus to 13% for 3D printed bolus was observed. The dosimetric consequence of these larger air gaps on the surface of the chest wall will depend on multiple parameters including energy, field size, beam modifiers and angle of incidence. A key determinant of dose error is field size, whereby a larger dosimetric error result as the field size is reduced. Referring to the study by Khan et al(7), for a 5x5 cm2 field, air gaps 5 mm or greater between Superflab and skin
10
ACCEPTED MANUSCRIPT produced significant dosimetric error, but this became insignificant for 10x10 cm2 fields. Thus, the error may become more important with increasing adoption of highly conformal techniques for chest wall
T
radiotherapy such as IMRT or VMAT (26, 27) where the overall dose is delivered using a series of smaller
RI P
MLC defined apertures.
SC
While 3D printed bolus produced a better fit on average, Figure 2 shows a single outlier where an air gap
NU
of over 2 cm was observed. In this case the patient exhibited gradual but persistent anatomical change between the first and last fraction such that bolus did not fit by the end of the final fraction (Figure 6).
MA
Anatomical change, as opposed to mis-positioning of the bolus, was confirmed for this case by coregistering and comparing sequential CBCTs over the treatment course. The research protocol did not
ED
permit redesign of the 3D printed bolus mid-treatment course. However, in practice, 3D printing allows for the unique concept of “adaptive bolus”, whereby CBCT imaging may be monitored after each
CE
using these image data.
PT
fraction, and the appearance of a significant air gap can trigger generation of a new 3D printed bolus
AC
Since this is the first study using 3D printed chest wall bolus, in vivo dosimetry was deemed important as a QA measure. Compared to the sheet bolus, 3D printed bolus yielded dose measurements that were lower by approximately 2 to 3% on average. This can be explained by the fact that for sheet bolus, the OSLD NanoDot was located below 5 mm Superflab material, and for the 3D printed bolus, below 3 mm PLA. According to the data of Burleson et al(19), if the physical depth were equal at 5 mm, these two materials would have approximately the same Tissue Maximum Ratio (TMR) value. With the reduced thickness of buildup material for the 3D printed bolus, the measured dose should be reduced, but the disparity is complicated by the variable obliquity of fields which will determine the actual difference in ray-line depth to the OSLD. We emphasize that the purpose of the in vivo dosimetry was to verify
11
ACCEPTED MANUSCRIPT approximate equivalence of the bolus types, not to quantify the relationship between air gap dimension and dose error, as this has been done elsewhere (6,7). Spatial coincidence between OSLD position and
T
air gaps below the bolus were infrequent, and thus the improvement in fit for the 3D printed bolus
RI P
should not be reflected in the OSLD measurements. While differences were seen between measured and planned doses were observed (Figure 5), the magnitude of these differences are comparable to
SC
previous work using nanoDot OSLDs on the chest wall (28). On average the Dmeasured/Dplanned ratio was
NU
higher than unity, and this is consistent with the finding that at depths in the vicinity of 5 mm, Eclipse AAA dose calculations are systematically low for beams oblique to the surface by approximately 2-4%
MA
compared to Monte Carlo calculations(29). The OSLD methodology applied here shows that 3D printed bolus offers the option of in-printing pockets for dosimeters, which addresses two issues: first, it avoids
ED
inadvertently creating an air gap by introducing a dosimeter between the bolus material and the skin.
CE
tattoo marks on the patient.
PT
Second, it facilitates standardization of the locations of measurements, e.g., without making multiple
With regard to surface dose, although not explored in this study, 3D printing of photon bolus allows
AC
customization of thickness, rather than selecting from a range of generic thicknesses of sheet bolus. Thus, dose build-up may be customized by the practitioner, and in concept could be variable by location on the patient. For example, where a tangential field arrangement delivers dose medial or lateral to the PTV, the bolus could be thinned, or the printed density (infill) could be selectively reduced. This could be useful where the radiation oncologist is concerned with severe skin toxicity during chest wall radiotherapy (30), or where an alternating bolus approach is to be replaced with a customized single thickness of bolus for all fractions (31), for example.
12
ACCEPTED MANUSCRIPT Grading of skin reaction was not an end-point of this study because for each patient, bolus type was alternated, and thus observations would reflect a combination of bolus type. However, among all 16
T
patients assessed weekly on treatment by their Radiation Oncologist, the maximum skin reaction as
RI P
graded on the NCI CTCAE V4.03 was as follows: One patient grade 3 (moist desquamation), two patients grade 2 (marked erythema), and 13 patients with mild or no dermatitis. This distribution of skin reaction
NU
technique chest wall radiotherapy in our department.
SC
for women undergoing chest wall radiotherapy was consistent with that in women receiving standard
MA
There are some drawbacks to the 3D printing approach, notably the time required to fabricate the bolus (over 12 hours, on average), and minimizing user-intervention of this step (i.e. to allow printing
ED
overnight) requires a high-quality 3D printer with low failure-rate. In addition, time is required for QA on the fabricated bolus. This QA may include physical measurement of dimensions with comparison to
PT
the design, as well as CT imaging (or CBCT imaging prior to treatment delivery at first fraction) to verify
CE
homogeneity of the bolus medium. Finally, as illustrated above, for this study a very rigid printable medium (PLA) was used. Although personalized to the surface of the patient, this material does not flex
AC
with minor patient deformation or anatomical change; therefore, further research will include investigation of more flexible materials.
Conclusion This study demonstrates the practical use of 3D printed bolus for post-mastectomy radiation therapy. Among a sample of 16 patients each treated at least four times with 3D bolus, the accuracy of fit of the 3D bolus to the chest wall was improved relative to standard sheet (Superflab) bolus, with the frequency of air gaps 5 mm or greater reduced from 30% to 13%. Surface dose, as measured using in vivo dosimetry, was within 3% for sheet and 3D printed bolus and there was no significant difference
13
ACCEPTED MANUSCRIPT between sheet and 3D printed bolus with regard to agreement with the treatment planning system. The setup time was reduced only marginally with 3D printed bolus (104 to 76 seconds), however this time
T
saving must be weighed against the considerable time for fabrication of the bolus (median 10.8 hours,
RI P
mean 12.6 5.4 hours) and related quality assurance, although the printing process was found to be
SC
largely automated.
NU
References
MA
1. Overgaard M, Hansen PS, Overgaard J, et al. Postoperative radiotherapy in high-risk premenopausal women with breast cancer who receive adjuvant chemotherapy. Danish Breast Cancer Cooperative
ED
Group 82b Trial. N. Engl. J. Med. 1997;337:949–955.
PT
2. Ragaz J, Jackson SM, Le N, et al. Adjuvant radiotherapy and chemotherapy in node-positive
CE
premenopausal women with breast cancer. N. Engl. J. Med. 1997;337:956–962.
3. Overgaard M, Nielsen HM, Overgaard J. Is the benefit of postmastectomy irradiation limited to
AC
patients with four or more positive nodes, as recommended in international consensus reports? A subgroup analysis of the DBCG 82 b&c randomized trials. Radiother Oncol. 2007;82:247–253.
4. White J, Tai A, Arthur D, et al. Radiation Therapy Oncology Group (RTOG) Breast cancer atlas for radiation therapy planning: consensus definitions. RTOG; 2009.
5. Hsu S-H, Roberson PL, Chen Y, et al. Assessment of skin dose for breast chest wall radiotherapy as a function of bolus material. Phys Med Biol. 2008;53:2593–2606.
6. Butson MJ, Cheung T, Yu P, et al. Effects on skin dose from unwanted air gaps under bolus in photon beam radiotherapy. Radiation Measurements. 2000;32:201–204.
14
ACCEPTED MANUSCRIPT 7. Khan Y, Villarreal-Barajas JE, Udowicz M, et al. Clinical and Dosimetric Implications of Air Gaps between Bolus and Skin Surface during Radiation Therapy. JCT. 2013;04:1251–1255.
RI P
T
8. Anderson PR, Hanlon AL, McNeeley SW, et al. Low complication rates are achievable after postmastectomy breast reconstruction and radiation therapy. International Journal of Radiation
SC
Oncology*Biology*Physics. 2004;59:1080–1087.
NU
9. Canters RA, Lips IM, Wendling M, et al. Clinical implementation of 3D printing in the construction of patient specific bolus for electron beam radiotherapy for non-melanoma skin cancer. Radiother Oncol.
MA
2016;121:148–153.
10. Dancewicz OL, Sylvander SR, Markwell TS, et al. Radiological properties of 3D printed materials in
ED
kilovoltage and megavoltage photon beams. Phys Med. 2017;38:111–118.
PT
11. Fujimoto K, Shiinoki T, Yuasa Y, et al. Efficacy of patient-specific bolus created using three-
CE
dimensional printing technique in photon radiotherapy. Phys Med. 2017;38:1–9.
AC
12. Hongtao Z, Xuemin D, Huimin Y, et al. Dosimetry study of three-dimensional print template-guided precision (125)I seed implantation. J Cancer Res Ther. 2016;12:C159–C165.
13. Ji Z, Jiang Y, Guo F, et al. Dosimetry verification of radioactive seed implantation for malignant tumors assisted by 3D printing individual templates and CT guidance. Appl Radiat Isot. 2017;124:68–74.
14. Kim M-J, Lee S-R, Lee M-Y, et al. Characterization of 3D printing techniques: Toward patient specific quality assurance spine-shaped phantom for stereotactic body radiation therapy. Chung CK, ed. PLoS ONE. 2017;12:e0176227.
15. Lee M-Y, Han B, Jenkins C, et al. A depth-sensing technique on 3D-printed compensator for total body irradiation patient measurement and treatment planning. Med Phys. 2016;43:6137–6144.
15
ACCEPTED MANUSCRIPT 16. Lindsay C, Kumlin J, Martinez DM, et al. Design and application of 3D-printed stepless beam modulators in proton therapy. Phys Med Biol. 2016;61:N276–90.
RI P
T
17. Park JW, Oh SA, Yea JW, et al. Fabrication of malleable three-dimensional-printed customized bolus using three-dimensional scanner. Sherman JH, ed. PLoS ONE. 2017;12:e0177562.
SC
18. Ricotti R, Ciardo D, Pansini F, et al. Dosimetric characterization of 3D printed bolus at different infill
NU
percentage for external photon beam radiotherapy. Phys Med. 2017;39:25–32.
19. Burleson S, Baker J, Hsia AT, et al. Use of 3D printers to create a patient-specific 3D bolus for external
MA
beam therapy. J Appl Clin Med Phys. 2015;16:166–178.
ED
20. Zhao Y, Moran K, Yewondwossen M, et al. Clinical applications of 3-dimensional printing in radiation
PT
therapy. Med Dosim. 2017;42:150–155.
CE
21. **** (redacted for blinding during review)
22. Park S-Y, Choi CH, Park JM, et al. A Patient-Specific Polylactic Acid Bolus Made by a 3D Printer for
AC
Breast Cancer Radiation Therapy. Deutsch E, ed. PLoS ONE. 2016;11:e0168063.
23. Lindsay C, Kumlin J, Jirasek A, et al. 3D printed plastics for beam modulation in proton therapy. Phys Med Biol. 2015;60:N231–40.
24. Ju SG, Kim MK, Hong C-S, et al. New technique for developing a proton range compensator with use of a 3-dimensional printer. Int. J. Radiat. Oncol. Biol. Phys. 2014;88:453–458.
25. Łukowiak M, Jezierska K, Boehlke M, et al. Utilization of a 3D printer to fabricate boluses used for electron therapy of skin lesions of the eye canthi. J Appl Clin Med Phys. 2017;18:76–81.
26. Zhang Q, Yu XL, Hu WG, et al. Dosimetric comparison for volumetric modulated arc therapy and
16
ACCEPTED MANUSCRIPT intensity-modulated radiotherapy on the left-sided chest wall and internal mammary nodes irradiation in treating post-mastectomy breast cancer. Radiol Oncol. 2015;49:91–98.
RI P
T
27. Munshi A, Sarkar B, Anbazhagan S, et al. Short tangential arcs in VMAT based breast and chest wall radiotherapy lead to conformity of the breast dose with lesser cardiac and lung doses: a prospective
SC
study of breast conservation and mastectomy patients. Australas Phys Eng Sci Med. 2017;23:433–8.
NU
28. Yusof, F. H. et al. On the Use of Optically Stimulated Luminescent Dosimeter for Surface Dose Measurement during Radiotherapy. PLoS One. 2105;10: e0128544.
MA
29. Panettieri, V., Barsoum, P., Westermark, M., Brualla, L. & Lax, I. AAA and PBC calculation accuracy in the surface build-up region in tangential beam treatments. Phantom and breast case study with the
ED
Monte Carlo code PENELOPE. Radiother Oncol 2009;93:94–101.
PT
30. Pignol J-P, Vu TTT, Mitera G, et al. Prospective evaluation of severe skin toxicity and pain during
CE
postmastectomy radiation therapy. Int. J. Radiat. Oncol. Biol. Phys. 2015;91:157–164.
AC
31. Ordonez-Sanz C, Bowles S, Hirst A, et al. A single plan solution to chest wall radiotherapy with bolus? BJR. 2014;87:20140035–9.
Figure Captions Figure 1 (A) Example of a 3D printed bolus, (B) locations of seven OSLDs (labeled A to G) relative to patient and tangential fields, and (C) embedded OSLD contained by in-printed pocket, where OSLD is flush with the skin-side of the bolus surface. Figure 2 Frequency histogram of maximum dimension of airgap between bolus and patient surface as measured on CBCT, comparing sheet bolus with 3D printed bolus.
17
ACCEPTED MANUSCRIPT Figure 3 Examples of differences in accuracy of fit between Superflab sheet bolus (left) and 3D printed bolus (right) for patients with convexity (Patient 4, top row) or concavity (Patient 13, bottom row).
T
Figure 4 Frequency histogram of OSLD-measured dose values for sheet bolus and 3D printed bolus,
RI P
including all measurements, for all OSLDs and patients.
Figure 5 Ratio of measured over planned dose at each of the seven OSLD locations, averaged over all
NU
OSLDs (indicated by A to F) are depicted in Figure 1.
SC
fractions and patients, comparing sheet bolus (top) with 3D printed bolus (bottom). The locations of the
Figure 6 Images explaining the outlying result in Figure 2. for 3D printed bolus. The patient exhibited
AC
CE
PT
ED
fraction (B) causing a loss of accuracy of fit.
MA
significant anatomical change such that the final fraction (A) differed significantly from that in the first
18
SC
RI P
T
ACCEPTED MANUSCRIPT
AC
CE
PT
ED
MA
NU
Fig. 1
19
ED
MA
NU
SC
RI P
T
ACCEPTED MANUSCRIPT
AC
CE
PT
Fig. 2
20
MA
NU
SC
RI P
T
ACCEPTED MANUSCRIPT
AC
CE
PT
ED
Fig. 3
21
NU
SC
RI P
T
ACCEPTED MANUSCRIPT
AC
CE
PT
ED
MA
Fig. 4
22
AC
Fig. 5
CE
PT
ED
MA
NU
SC
RI P
T
ACCEPTED MANUSCRIPT
23
MA
NU
SC
RI P
T
ACCEPTED MANUSCRIPT
AC
CE
PT
ED
Fig.6
24