Radiation therapy treatment of the prone breast using EZFluence: A case study

Kyle Garafolo, BAS, R.T.(R)(T), Stephanie Hufnal, BS, R.T.(R), Keenan Fuller, BS, R.T.(R),
Nishele Lenards, PhD, CMD, R.T.(R)(T), FAAMD, Ashley Hunzeker, MS, CMD, Matt Tobler,
RT(T), CMD, Alyssa Olson, MS, R.T.(T), CMD
Medical Dosimetry Program at the University of Wisconsin-La Crosse, WI

ABSTRACT
The aim of this study was to determine if the use of the EZFluence software script would
improve prone breast treatment planning time, organs at risk (OAR) sparing, and overall dose
uniformity to the target when compared with the traditional field-in-field (FiF) technique. Ten
patients were planned retrospectively to include the whole affected breast to a dose of 42.56 Gy
using both non-divergent medial and lateral 3-dimensional conformal radiation therapy (3D
CRT) tangential beams produced by both manual FiF techniques and EZFluence software. The
range of the Planning Target Volume (PTV) size for our included patients was 547-1090 cc; the
mean PTV size was 860 cc. All plans were generated with the goal of achieving at least 95% of
the whole breast PTV_Breast_Eval and the PTV_Lumpectomy_Eval structures receiving at least
95% of the prescription dose. Evaluation criteria included overall treatment planning time, the
amount of the PTV receiving ≥ 105% of the dose (V105), as well as dose to the OAR and nearby
critical structures. Plans using the EZFluence script demonstrated a 62.2% reduction in treatment
planning time and an 82.5% reduction in V105. Nearby OAR demonstrated a small increase in
dose with the EZFluence plans. The mean heart dose increased 2.8%, the mean lung dose
increased 2.7%, the mean spinal cord dose increased 1.2%, and the mean liver dose increased
2.6%. The V95% to the PTV_Breast_Eval demonstrated a 0.4% increase with EZFluence.
Overall, the plans utilizing the EZFluence script achieved similar coverage to the target
structures while greatly reducing treatment planning time and V105. However, the plans using
EZFluence did demonstrate a small increase in dose to the surrounding OAR.

Introduction
Breast cancer has remained the most commonly diagnosed cancer in women worldwide,
and the fifth leading cause of death from cancer overall.1 Historically, several methods have been
utilized for treating breast cancer including surgical resection, chemotherapy, and radiation
therapy or a combination of approaches. There have been a variety of different techniques and
technologies that have advanced treatment planning as well as treatment delivery, specifically
within the area of radiation therapy. One such technique that has been proven to be effective is
radiation therapy treatment delivery to the breast with the patient in the prone position.
Since its inception, several independent studies have recognized the benefits of whole
breast irradiation in the prone position in treating breast cancer.2,3 Benefits to treating breast
cancer in the prone position include, but are not limited to, decreasing acute toxicity, risks of
radiation induced lung cancer, and cardiac toxicity.2 The cosmetic benefits to prone treatments
have proven to be very effective as well. Prone treatment allows the breast to hang away from
the patient’s chest, which helps to alleviate the creation of skin folds thereby stopping the bolus
effect seen in supine treatments. In fact, Olson4 reported that only 4.5% of breast patients treated
prone had Grade 3 skin reactions and a 22% to 36% higher cosmetic outcomes when compared
to patients treated supine. Research done by Haffy,3 has also suggested additional benefits to
prone breast radiotherapy, such as improved dose coverage, better homogeneity, less areas of
maximum dose within the treatment volume, lower heart and ipsilateral lung dose, lower
contralateral breast dose, and reduced skin reactions.
Another potentially useful practice for treating breast cancer patients came in the form of
a software script known as EZFluence by RadFormation. The EZFluence software is a script that
is used with Varian Eclipse treatment planning system (TPS) in order to generate optimal fluence
patterns. EZFluence uses the patient anatomy and plan parameters from Eclipse and
automatically generates plans utilizing electronic compensators or field-in-field (FiF) techniques
for any treatment site (Figure 2). EZFluence quickly generates multiple plan options with various
maximum dose limits, giving the user the ability to choose their preferred plan based on target
coverage, dose to organs at risk (OAR), and maximum dose (Figure 3). The preferred plan is
then exported back into Eclipse where more edits or calculations can be performed. Only the FiF
component of EZFluence was used for this study, as this technique was the preference of the
clinic physicians.
While setup position for breast radiotherapy is one factor that affects dose to the target
and surrounding OAR, the ability to reduce dose inhomogeneities is another vital aspect to
minimizing patient toxicities.4 As a means to help achieve these aforementioned benefits of the
prone position, multiple planning techniques have been utilized over the years, such as physical
wedges, enhanced dynamic wedges, manual FiF, electronic compensators, and intensity
modulated radiation therapy (IMRT). As another way to aid in breast treatment planning, one
new software has been developed, known as EZFluence, which can be used to create an optimal
fluence pattern in order to generate a homogenous dose distribution for breast irradiation
treatments.5
Even with the use of better positional methods and planning techniques, medical
dosimetrists have continued to face several struggles when creating treatment plans. One struggle
that medical dosimetrists have faced is the speed and consistency in which plans are generated,
which has been a problem for even the most experienced medical dosimetrists.6 It can take up to
several hours to accurately plan and calculate complicated 3-dimensional conformal radiation
therapy (3D CRT) breast plans.6 Therefore, improving treatment planning time may lead to
increased productivity.
Another struggle that medical dosimetrists have faced when planning 3D CRT prone
breast radiotherapy treatments has been producing fields with an even dose distribution while
maintaining a low amount of breast tissue receiving greater than plus 5% of the prescription
amount. Dosimetric inhomogeneity in breast treatment planning, specifically in breast volumes
treated to ≥ 105% of the prescribed dose (V105), has been a proven predictor of long-term breast
pain.7 Based on their research, Keenan et al8 found that in cases when the volume of the breast
receiving V105 ≥ 30cc with conventional fractionation have a strong indicator for acute skin
toxicity in general breast planning. Another study by Hymas et al,9 proved that by reducing the
V105 to ≤ 10% of the overall breast volume would achieve excellent or good cosmesis following
radiotherapy to the breast, as well as an acceptable level of toxicity. Due to the increased risk of
skin toxicity, maximizing dose homogeneity and reducing inhomogeneities involving the region
of V105 have become important planning criterion to be evaluated when generating treatments.
Sparing of OAR has also been a longtime concern with any treatment planning technique,
with prone breast radiotherapy being no exception. In prone breast cases, the OAR of interest
typically include organs such as the heart, ipsilateral lung, spinal cord, and liver. The most
drastic differences using the prone technique in breast cancer patient treatment planning can be
seen in the dose amount to the heart and ipsilateral lung.10 Venkatesan et al10 showed that treating
breasts in the prone position would lower both the volume and the dose exposure to the lungs,
though there was no real statistical difference in mean heart dose on left sided breast cancer
treatments. Even with positional improvements to minimize the dose to critical structures and
OAR, attaining a homogenous plan with adequate dose to the tumor volume has continued to be
a challenge.
The utilization of both FiF techniques and software scripts, such as EZFluence, have
helped medical dosimetrists achieve optimal treatment plans. Field-in-field techniques have been
recognized as an effective means of controlling these problems. To date, however, there has been
limited data demonstrating the benefits of EZFluence compared to manual FiF techniques for
supine and prone breast patients.5 Yoder et al5 demonstrated that the EZFluence software script
greatly improved treatment planning time, reduced the volume of planning target volume (PTV)
receiving V105, and achieved similar dose constraints to the OAR for supine breast cancer
patients.
The purpose of this study was to provide information on the use of the EZFluence
software script to determine if it would improve prone breast treatment planning time, OAR
sparing, and overall dose uniformity to the target. Based on positional reasoning behind breast
treatment planning in the prone position, as well as dose uniformity and OAR constraints, the
metrics for this study were formed to demonstrate if these measures could be improved by using
the EZFluence software script over conventional FiF techniques. The main variables that the
researchers examined in this study were the overall treatment planning time, the amount of V105
within the plan, and the ability to meet dose constraint criteria placed on the OAR (Table 1).

Case Description
Patient Selection
Ten female individuals with breast cancer that had intact, pendulous breasts (post-
lumpectomy) without nodal involvement were selected for this study. Only patients that were set
up in the prone position, per the attending radiation oncologist’s request, were included. The
range of the PTV size for the included patients was 547-1090 cc, with the mean PTV size being
860 cc. Planning was done retrospectively to include the whole affected breast to a dose of 42.56
Gy using both non-divergent medial and lateral 3D CRT tangential beams produced by both
manual FiF techniques and EZFluence software. Additional dose from any boost prescribed after
the initial 42.56 Gy was not included in this study.
 All 10 patients received a CT simulation with a General Electronic (GE) Lightspeed 16
multi-slice CT scanner. All patients were setup in the prone position on a CDR Systems prone
breast board with their arms up prior to treatment planning. The radiation oncologist placed
radiopaque wire stickers at the time of CT simulation on the patient’s midsagittal plane and
approximately 2.0 cm beyond the palpable breast tissue in the superior, inferior, and lateral
aspects. These wires were used to determine the approximate breast tissue borders, location of
the lumpectomy scar, and assist in defining beam angles, collimator angles, and field sizes
during treatment planning. Alignment reference markers were used for positioning and
subsequent markings were placed for future set-up as part of the CT simulation (Figure
1). The isocenter location was chosen by the attending radiation oncologist at the time of CT
simulation. Isocenter was located approximately at the midpoint of the breast tissue at the
approximate center of the breast volume (roughly at the midpoint of the breast tissue external
separation and in proximity of the chest wall).

Target Delineation
All target delineation was performed in Varian Medical Systems Eclipse TPS, version
15.6. Targets were delineated for treatment planning by the attending radiation oncologist (Table
2).11,12 The general OAR for prone breast treatment planning were contoured as well, which
included the bilateral lungs, heart, liver, and spinal cord. Contouring of the OAR was performed
per the Radiation Therapy Oncology Group (RTOG) contouring atlas recommendations.12

Treatment Planning
The CT scans of the 10 patients in this study were imported from CT simulation to the
Eclipse TPS. Each patient was prescribed a radiation dose of 42.56 Gy to the intact whole breast
delivered in 16 fractions at 2.66 Gy per fraction. To ensure adequate anterior coverage to the
breast tissue, a 2.0 cm margin beyond the breast tissue was included anteriorly in air. Two fields
were tangentially aligned to the PTV to create non-divergent opposed fields. Two different 3D
CRT plans were created for each patient, one utilizing the manual FiF technique and the other
using EZFluence software. All plans were constructed using opposed non-divergent tangential
field arrangements with 6 MV photon energy beams.
 For the manual FiF treatment plans, the initial fields included open multileaf collimator
(MLC) leaves for the medial and lateral tangent fields. A reference point was placed within the
field, and the plan was normalized so that upon calculation, 100% of the prescription dose was
delivered to that point. The 107% dose cloud was turned on to verify that the reference point was
not placed within the 107% dose cloud. The field weightings were adjusted slightly to achieve a
uniform dose distribution to the PTV_Lumpectomy_4256. At the dosimetrist’s discretion, the
collimator was rotated 90 degrees in order to allow for optimal blocking with the MLC leaves.
The dosimetrist manually created the FiF every 2% until the 104% isodose line was reached,
which created between 2 to 4 subfields per gantry angle. The FiF weighting was still set to 0% at
this point. Starting with the first FiF, the maximum dose was blocked at every 2% interval
utilizing the MLC leaves. The reference point was not blocked by any of the MLC leaves. The
process of blocking the maximum dose regions continued until the 104% isodose line, while also
removing as much as the 105% isodose line throughout the breast volume but keeping
appropriate coverage to the targets. Each subfield was weighted 2% of the total weight. This was
done by removing the dose contribution to the original field and distributing it to the respective
subfields.
The EZFluence plans were created using the software accessed through the Eclipse TPS.
The EZFluence plans utilized the same structure sets and the primary open tangent fields as in
the manual FiF plans. There was no plan normalization enabled and the maximum number of
field segments was limited to 8 for each tangential field. The minimum number of monitor units
(MU) for each tangential field was 4. EZFluence creates a PTV_EVAL_EZ structure on the
patient to calculate the initial fluences. The target was then formed using the EZFluence
guidelines (Table 3).
All plans were generated with the goal of achieving at least 95% of the whole breast
PTV_Breast_Eval and the PTV_Lumpectomy_Eval structures receiving at least 95% of the
prescription dose. The initial dose was calculated using a grid size of 0.25 cm with the analytical
anisotropic algorithm (AAA) from Eclipse TPS. In order to equally compare both the manual FiF
and EZFluence plans, the EZFluence plan was normalized so that 100% of the prescription dose
covered the same volume of the PTV_Breast_Eval (V100) as in the manual FiF plan.
Normalization was performed on a patient-by-patient basis, which was then analyzed across the
entire 10 patients in this study. Dose volume histogram parameters were evaluated and the
planning measures of the 2 different techniques were compared. The values assessed were the
amount of the PTV receiving ≥ 95% of the prescribed dose, the volume of the breast receiving
105% prescription dose, and the dose metrics to the ipsilateral lung, heart, liver, and spinal cord.

Results
Data for both the manual FiF technique plans and EZFluence generated plans were
compiled for all 10 patients. To ensure an accurate comparison between planning techniques, the
EZFluence plans were normalized using the aforementioned methods. Dose distributions and
planning metrics were evaluated based on DVH analysis. The generalized results of the two
different planning techniques indicated relatively similar outcomes in terms of OAR sparing and
target coverage (Table 4). The volume of breast receiving ≥ V105 and total planning time was
noticeably improved using the EZFluence planning technique.
Treatment planning constraints for the heart, ipsilateral lung, liver, and spinal cord were
all achieved. However, dose to these OAR was greater in nearly every instance with the
EZFluence method. For left sided breast cancer patients, mean dose to the heart was 3.4% greater
with the EZFluence plans. For right sided breast cancer patients, this difference was only 1.7%
greater. The mean ipsilateral lung dose for the EZFluence plans was 2.7% greater than the
manual FiF plans. There was a 1.2% increase in mean dose to the spinal cord as well as a 2.7%
increase in mean liver dose with the EZFluence plans.
The volume of the PTV_Lumpectomy_Eval as well as the PTV_Breast_Eval receiving ≥
95% of the prescription dose were similar. Coverage to the PTV_Lumpectomy_Eval was 0.2%
less with the EZFluence plans; coverage to the PTV_Breast_Eval showed a 0.4% increase with
the EZFluence plans. While coverage to the target structures was similar, the plans generated
with EZFluence produced less V105 as compared to the manual FiF method. This reduction in
V105 also contributed to a lower global maximum dose within the breast tissue. The normalized
EZFluence plans averaged over a 1% reduction in global maximum dose as compared the
manual FiF technique. The treatment planning time was vastly improved with the EZFluence
plans, which resulted in an overall 62.2% decrease over manual FiF technique (Table 4).
Conclusion
Based on the results of this case study, initiating EZFluence script, as opposed to using a
manual FiF technique, produced a marked improvement in the overall time needed to plan
treatments for breast patients in the prone position. Treatment planning time averaged 40.7
minutes when planning with a manual FiF technique and 15.4 minutes when using EZFluence,
which was a 62.2% decrease. Much of the time saved was a result of the ability of the software
to quickly generate the necessary FiF MLC arrangements while reducing the need to remove
remaining areas of high dose. When comparing the percentage of V105 within the plans, there
was also a significant decrease when using EZFluence over the manual FiF technique. Utilizing
the EZFluence script also proved effective at controlling dose to OAR with only a minimal
increase to segmented structures, while still maintaining adequate coverage to the
PTV_Breast_Eval structure. The minimal increase to the OAR was not of great clinical
significance due to the inherent ability of prone patient positioning to reduce dose to nearby
OAR as compared to supine breast positioning.2,3 The findings of this study demonstrated the
value of EZFluence in the treatment planning of prone breast cancer patients. The treatment
plans generated with EZFluence helped to produce clinically acceptable treatment plans in a
fraction of the time, therefore greatly improving the productivity of the medical dosimetrist.
The main limitation to this study was the small sample size, yet the results for each of the
10 patients were consistent across all the measured constraints that were applied. Future research
should include a greater number of patients with varying breast tissue volumes to further
evaluate the findings of this study. EZFluence has proven to be a viable choice when it comes to
the treatment planning of prone breast patients, however, additional research and evaluation
should be conducted to prove its usefulness in plans other than breast cases.
 References
1. Ferlay J, Soerjomataram I, Dikshit R, et al. Cancer incidence and mortality worldwide:
Sources, methods and major patterns in GLOBOCAN 2012. Int J Cancer.
2015;136(5):E359-E386. https://dx.doi.org/10.1002/ijc.29210
2. Boute B, De Neve W, Speleers B, et al. Potential benefits of crawl position for prone
radiation therapy in breast cancer. J Appl Clin Med Phys. 2017;18(4):200–205.
https://dx.doi.org/10.1002/acm2.12118
3. Haffty, BG. Supine or prone breast radiation: Upsides and downsides. Int J Radiat Oncol
Biol Phys. 2018;101(3):510-512. https://dx.doi.org/10.1016/j.ijrobp.2018.03.023
4. Olson KN. Improving treatment outcomes of breast radiation therapy: the prone position.
Radiat Therapist. 2014;23(1):21-26. https://web-b-ebscohost-
com.libweb.uwlax.edu/ehost/pdfviewer/pdfviewer?vid=7&sid=6c25b527-3254-4499-9769-
da885b9aa3d5%40sessionmgr102. Published Spring 2014. Accessed May 31, 2019.
5. Yoder T, Hsia AT, Xu Z, Stessin A, Ryu S. Usefulness of EZFluence software for
radiotherapy planning of breast cancer treatment. Med Dosim. 2019;S0958-3947(18):30137-
7. https://dx.doi.org/10.1016/j.meddos.2018.12.001
6. Bolan C. Expediting the treatment planning process. Appl Radiat Oncol. 2013;2(4):19-23.
http://cdn.agilitycms.com/applied-radiation-oncology/ARO_12-13_TechTrends.pdf.
Published Winter 2013. Accessed July 11, 2019.
7. Mak KS, Chen YH, Catalano PJ, et al. Dosimetric inhomogeneity predicts for long-term
breast pain after breast-conserving therapy. Int J Radiat Oncol Biol Phys. 2014;93(5):1087-
1095. https://dx.doi.org/10.1016/j.ijrobp.2014.05.021
8. Keenan LG, Lavan N, Dunne M, McArdle O. Modifiable risk factors for acute skin toxicity
in adjuvant breast radiotherapy: Dosimetric analysis and review of the literature. Med
Dosim. 2019;44(1):51-55. https://dx.doi.org/10.1016/j.meddos.2018.01.004
9. Hymas RV, Jawad MS, Mangona VS, et al. Dosimetric predictors of toxicity and cosmesis in
women treated with hypofractionated whole-breast irradiation. Int J Radiat Oncol Biol Phys.
2014;90(1):S270-S271. https://dx.doi.org/10.1016/j.ijrobp.2014.05.930
10. Venkatesan K, Deshpande S, Anand V, et al. Comparison of heart and lung doses in deep
inspiration breath hold radiation therapy and prone position radiation therapy for whole
 breast radiation therapy. Int J Radiat Oncol Biol Phys. 2018;102(3):489-490.
https://dx.doi.org/10.1016/j.ijrobp.2018.07.1393
11. Mamounas E, White J, Khan A, et al. NRG Oncology. A randomized phase III clinical trial
evaluating post-mastectomy chestwall and regional nodal XRT and post-lumpectomy
regional nodal XRT in patients with positive axillary nodes before neoadjuvant
chemotherapy who convert to pathologically negative axillary nodes after neoadjuvant
chemotherapy. J Clin Oncol. 2014;32(15).
https://dx.doi:10.1200/jco.2014.32.15_suppl.tps1141
12. White J, Tai A, Arthur D, et al. Breast cancer atlas for radiation therapy planning: Consensus
definitions. Radiation Therapy Oncology Group (RTOG). In: RTOG Breast Cancer
Contouring Atlas. Retrieved from:
https://www.rtog.org/LinkClick.aspx?fileticket=vzJFhPaBipE=.
Figures

Figure 1: Diagram of CT simulation setup devices, patient position, and setup markings
Figure 2: EZFluence isodose / color wash for FiF plan.
Figure 3: EZFluence DVH selection screen.
Tables
Structure Volume Receiving (Dose)

Ipsilateral Lung ≥ 16 Gy (V16) ≤ 10%

≥ 8 Gy (V8) ≤ 15%
≥ 4 Gy (V4) ≤ 25%

Liver Mean Dose (Dmean) ≤ 2.8 Gy

Spinal Cord Dmax ≤ 50 Gy (used for reporting purposes)

Heart Dmean ≤ 3.2 Gy and Dmax ≤ 16 Gy

Table 1: Dose constraints for OAR and critical structures.
 Target Structure
1. Cavity
2. PTV_Lumpectomy_4256
3. PTV_Lumpectomy_Eval
4. PTV_Breast
5. PTV_Breast_Eval
Table 2: Definition of target structure delineation.11,12
Determining Factors
Contour using all available clinical and radiographic
information including the excision cavity volume,
architectural distortion, lumpectomy scar, seroma and/or
extent of surgical clips
Cavity + 1.0 cm expansion
Since a substantial part of the Lumpectomy PTV often
extends outside the patient (especially for superficial cavities),
the Lumpectomy PTV is then copied to a Lumpectomy PTV
Eval which is edited. This Lumpectomy PTV Eval is limited
to exclude the part outside the ipsilateral breast and the first 5
mm of tissue under the skin (in order to remove most of the
build-up region for the DVH analysis) and excluding the
Lumpectomy PTV expansion beyond the posterior extent of
breast tissue (chestwall, pectoralis muscles, and lung) when
pertinent. The lumpectomy PTV should not cross midline
Considers reference clinical breast at time of CT. This
includes the apparent CT glandular breast tissue and
incorporates consensus definitions of anatomical borders.
The breast PTV includes the lumpectomy cavity. It is limited
posteriorly to the anterior surface of the pectoralis, serratous
anterior muscle excluding chestwall, boney thorax, and lung.
In general, the pectoralis and/or serratous anterior muscles are
excluded from the Breast PTV unless clinically warranted by
the patient’s pathology
The Breast PTV Eval is intended to exclude the portion of the
Breast PTV that extends outside the patient. The Breast PTV
is copied to a Breast PTV Eval which is edited. This Breast
PTV Eval is limited anteriorly to exclude the part outside the
patient and the first 5 mm of tissue under the skin (in order to
remove most of the build-up region for the DVH analysis)
 PTV_EVAL_EZ Target Guidelines
Cropped 0.5 cm from the defined body surface
Cropped 0.5 cm from the field edges
Cropped 0.5 cm from identified OAR structures (Lung, Heart, Spinal
Cord)
Table 3: EZFluence PTV structure guidelines.
 Heart
Heart
(Right
(Left Side Lung V95% Planning
Planning Side V105
Treatment (Mean - PTV_Breast_Eval Time
Technique Treatment (cc)
– mean cGy) (%) (Minutes)
– mean
cGy )
cGy)
Manual
35.52 67.96 59.45 9.99 95.55 40.7
FiF
EZFluence 36.14 70.28 61.07 1.75 95.97 15.4
% Change
1.7% 3.4 % 2.7% 82.5% 62.2%
for 0.4% Increase
Increase Increase Increase Decrease Decrease
EZFluence
Table 4: Comparison on mean OAR and target dose, V105, and planning time for manual FiF
and EZFluence techniques.