Follow this and additional works at:

Transcription

1 The University of Toledo The University of Toledo Digital Repository Theses and Dissertations 2015 Comparison of radiation treatment plans for breast cancer between 3D conformal in prone and supine positions in contrast to VMAT and IMRT supine positions Ana Isabel Bejarano Buele University of Toledo Follow this and additional works at: Recommended Citation Bejarano Buele, Ana Isabel, "Comparison of radiation treatment plans for breast cancer between 3D conformal in prone and supine positions in contrast to VMAT and IMRT supine positions" (2015). Theses and Dissertations This Thesis is brought to you for free and open access by The University of Toledo Digital Repository. It has been accepted for inclusion in Theses and Dissertations by an authorized administrator of The University of Toledo Digital Repository. For more information, please see the repository's About page.

2 A Thesis entitled Comparison of Radiation Treatment Plans for Breast Cancer between 3D Conformal in Prone and Supine Positions in Contrast to VMAT and IMRT Supine Positions by Ana Isabel Bejarano Buele Submitted to the Graduate Faculty as partial fulfillment of the requirements for the Master of Science Degree in Medical Physics E. Ishmael Parsai, Ph.D., Committee Chair Krishna Reddy, M.D., Ph.D., Committee Member Diana Shvydka, Ph.D., Committee Member Dr. Patricia R. Komuniecki, Dean College of Graduate Studies The University of Toledo December, 2015

3 Copyright 2015, Ana Isabel Bejarano Buele This document is copyrighted material. Under copyright law, no parts of this document may be reproduced without the expressed permission of the author ii

4 An Abstract of Comparison of Radiation Treatment Plans for Breast Cancer between 3D Conformal in Prone and Supine Positions in Contrast to VMAT and IMRT Supine Positions by Ana Isabel Bejarano Buele Submitted to the Graduate Faculty as partial fulfillment of the requirements for the Master of Science Degree in Medical Physics The University of Toledo December, 2015 The treatment regimen for breast cancer patients typically involves Whole Breast Irradiation (WBI). The coverage and extent of the radiation treatment is dictated by location of tumor mass, breast tissue distribution, involvement of lymph nodes, and other factors. The current standard treatment approach used at our institution is a 3D tangential beam geometry, which involves two fields irradiating the breast, or a four field beam arrangement covering the whole breast and involved nodes, while decreasing the dose to organs as risk (OARs) such as the lung and heart. The coverage of these targets can be difficult to achieve in patients with unfavorable thoracic geometries, especially in those cases in which the planning target volume (PTV) is extended to the chest wall. It is a well-known fact that exposure of the heart to ionizing radiation has been proved to increase the subsequent rate of ischemic heart disease. In these cases, inverse planned treatments have become a proven alternative to the 3D approach. The goal of this research project is to evaluate the factors that affect our current techniques as well as to adapt the development of inverse modulated techniques for our iii

5 clinic, in which breast cancer patients are one of the largest populations treated. For this purpose, a dosimetric comparison along with the evaluation of immobilization devices was necessary. Radiation treatment plans were designed and dosimetrically compared for 5 patients in both, supine and prone positions. For 8 patients, VMAT and IMRT plans were created and evaluated in the supine position. Skin flash incorporation for inverse modulated plans required measurement of the surface dose as well as an evaluation of breast volume changes during a treatment course. It was found that prone 3D conformal plans as well as the VMAT and IMRT plans are generally superior in sparing OARs to supine plans with comparable PTV coverage. Prone setup leads to larger shifts in breast volume as well as in positioning due to the difference in target geometry and nature of the immobilization device. IMRT and VMAT plans offer sparing of OARs from high dose regions with an increase of irradiated volume in the low dose regions. Skin flash incorporation was found to be accurate with the use of virtual bolus in the TPS for inverse modulated plans. Various factors influencing dose delivery in breast cancer radiation treatments were examined and quantified. Practical recommendations developed in the course of this project can improve our current techniques and provide alternatives to treat unique and challenging clinical cases. iv

6 This work is dedicated to: My family. Without you, none of this would have been possible. You have been my support and my inspiration to follow this field. Words are not enough to express my gratitude to each one of you. My fiancé, Zaid. You supported me through this journey and patiently listened to my many ideas and problems. Chandra: You helped me find peace and beauty in the soul of little birds. Thank you for being my mom far away from home. My friend, Lizyamel. You made me experience the true meaning of this career. v

7 Acknowledgements I would like to thank my advisor, Dr. E. Parsai, for his guidance and help in this project, as well as Dr. Shydka and Dr. Pearson, who taught me the elemental concepts of Medical Physics on which my clinical training was built on. I would also like to express my gratitude to Dr. Nicholas Sperling, from whom I have learned different aspects of Medical Physics. These skills were critical during the completion of this research project. Thanks to Sean Tanny and Greg Warrell, for their support and help designing the measurements needed for this research project as well as for their help with technical aspects of OSLDs. Thanks to my friends, classmates and colleagues for their support and friendship through these years. I would like to thank the Radiation Oncologists, Dr. Reddy and Dr. Chen, for challenging me and helping me develop my skills as a Medical Physicist in the clinical setting. I would also like to thank the team of therapists at Dana Cancer Center, who taught me practical aspects of Radiation Therapy and let me observe their everyday work. Thanks to Michelle Giovanoli, for helping me acquire statistics about the cancer center. vi

8 Table of Contents Abstract Dedication Acknowledgements Table of Contents List of Tables List of Figures List of Abbreviations List of Symbols iii v vi vii x xiii xiv xv 1. Chapter 1: Background 1.1 Breast Cancer Statistics Treatments Available for Breast Cancer Surgery Systemic Treatments Radiation Therapy Local Epidemiology Chapter 2: Literature Survey 2.1 Irradiation Modalities Supine and Prone 3D Technique Considerations Dosimetric Comparison Studies Inverse Planning Techniques: IMRT and VMAT Technique Considerations 11 vii

9 Surface Dose Measurements Target Volume Correction Dosimetric Comparison Studies Treatment Reproducibility Breast Volume Variations Treatment Immobilization Methods and Materials 3.1 Dosimetric Comparison Planning Techniques D Conformal WBI VMAT/IMRT WBI Dosimetric Criteria for Analysis Inverse Planning Skin Flash Measurement of Surface Dose Evaluation of Inter-fractional Breast Volume Changes Evaluation of Immobilization Devices Anthropomorphic Phantom Setup Evaluation Patient Setup Evaluation Chapter 4: Results and Analysis 4.1 Dosimetric Comparison Supine and Prone 3D Conformal Plan Comparison Supine 3D Conformal, IMRT, and VMAT Plan Comparisons Inverse Planning Skin Flash.45 viii

10 4.2.1 Surface Dose Delivery Measurements Evaluation of Inter-fractional Breast Volume Variations Evaluation of Immobilization Devices Anthropomorphic Phantom Setup Patient Setup Evaluation Conclusions..54 References. 57 Appendices ix

11 List of Tables Table 1.1 Percent of US Women who Develop Breast Cancer over 10-, 20-, and 30- Year Intervals According to the Current Age, Table 1.2 Treatments for Patients Diagnosed at UTMC with Breast Cancer 4 Table 1.3 Diagnosis per Patient Admitted to the DCC Radiation Oncology Department.5 Table 3.1 Summary of Inverse Planning Optimization Parameters.25 Table 4.1 PTV dose distribution comparison for 3D conformal plans for the supine and prone positions.35 Table 4.2 Contralateral lung dose distribution comparison for 3D conformal plans for the supine and prone positions...36 Table 4.3 Ipsilateral lung dose distribution comparison for 3D conformal plans for the supine and prone positions...36 Table 4.4 Heart dose distribution comparison for 3D conformal plans for the supine and prone positions.37 Table 4.5 Contralateral breast dose distribution comparison for 3D conformal plans for the supine and prone positions.37 Table 4.6 PTV dose distribution comparison for 3D conformal, VMAT and IMRT plans in the supine position...39 Table 4.7 Contralateral lung dose distribution comparison for 3D conformal, VMAT and IMRT plans in the supine position..40 Table 4.8 Ipsilateral lung dose distribution comparison for 3D conformal, VMAT and IMRT plans in the supine position.40 x

12 Table 4.9 Heart dose distribution comparison for 3D conformal, VMAT and IMRT plans in the supine position..41 Table 4.10 Contralateral breast dose distribution comparison for 3D conformal, VMAT and IMRT plans in the supine position..42 Table 4.11 Summary of Dosimetric Comparison between Supine and Prone 3D Conformal Plans.43 Table 4.12 Summary of Dosimetric Comparison between Supine IMRT and VMAT (without bolus) Plans..44 Table 4.13 Summary of Total Average Deviation of Measured Values from Planned Values using MOSFETs and OSLDs for Each of the Plans Irradiated..45 Table 4.14 Summary of Average Deviation per Site of Measured Values from Planned Values using MOSFETs and OSLDs for Each of the Plans Irradiated...45 Table 4.15 OSLD Results from Irradiation at 0 and at 90 degrees. 46 Table 4.16 Summary of Analyzed Patient Imaging in the Supine Position 47 Table 4.17 Supine Breast (Base) Thickness Results...47 Table 4.18 Supine Breast (Middle) Thickness Results 48 Table 4.19 Supine Breast Length Results 48 Table 4.20 Supine Breast Angle Results..49 Table 4.21 Supine Breast Volume Results...49 Table 4.22 Volume Correlation Results in the Supine Position...50 Table 4.23 Summary of Average Rotation for the Supine and Prone Setup Using an Anthropomorphic Phantom..51 xi

13 Table 4.24 Summary of Pt. Axial Rotation and 3D Displacement Vector for the Supine Patient Setup 52 Table 4.25 Summary of Pt. Axial Rotation and 3D Displacement Vector for the Prone Patient Setup 52 Table 4.26 Summary of the Comparison Between the Supine and Prone Positions Setup.52 Table 4.27 Volume Correlation Results in the Prone Position 53 xii

14 List of Figures Figure 1.1 Diagnosis per Patient Admitted to DCC Radiation Oncology Department.6 Figure 2.1 Supine tangent setup (left) vs. prone tangent setup (right) for the same patient. 8 Figure 3.1. Optimization contours for Inverse Modulated Plans: PTV (yellow), PTV ring (orange), PTV optimization (purple), contralateral breast (green), sternum (red).24 Figure 3.2 General irradiation setup: a. Dosimeter placement on the medial, tip and lateral aspects of the breast. b. Superflab bolus setup. c. wet towels bolus setup.29 Figure 3.3 PTV coverage. 50 Gy (green), 32.5 Gy (blue), 25 Gy (purple): a. 1 cm of 1g/cm 3 bolus plan. b. 1 cm of 0.65 g/cm 3 bolus plan. c. 1cm of 0.65 g/cm 3 bolus 2 PTVs plan 29 Figure 3.4 Sample of breast measurements in a MV image..31 Figure 3.5 Positioning of the anthropomorphic phantom in the prone breast board.32 Figure 3.6 Identification of measurement slice in the anthropomorphic phantom and measurement of phantom rotation.32 Figure 4.1 Axial slices comparing the dose distribution for two 3-D conformal treatment plans for patient 4P. a. Supine 3D conformal dose distribution with a hotspot of 5400 cgy. b. Prone 3D conformal dose distribution with a hotspot of 5400 cgy.38 Figure 4.2. DVH comparison of 3D conformal dose distributions in the supine (dashed line) and prone positions (solid line) for patient 4P..38 Figure 4.3. CBCTs illustrating bolus conformity to breast tissue a. Wet towel bolus. b. superflab bolus..46 xiii

15 List of Abbreviations 3D-CRT...Three Dimensional Conformal Radiotherapy BCS Breast Conserving Surgery CBCT... Cone Beam Computed Tomography CT.. Computed Tomography DRR.. Digitally Reconstructed Radiograph DVH...Dose Volume Histogram IDL.Isodose Line IMRT...Intensity Modulated Radiation Therapy MLC... Multi-Leaf Collimator MV Mega Voltage OAR... Organ at Risk OBI.... On-Board Imager. PTV Planning Target Volume ROI Region of Interest RTOG...Radiation Therapy Oncology Group TPS Treatment Planning System VMAT Volumetric Modulated Arc Therapy WBI.. Whole Breast Irradiation xiv

16 List of Symbols cc...cubic Centimeter cm...centimeter cm 3...Cubic Centimeter g/cm 3...Gram per Cubic Centimeter mm...millimeter xv

17 Chapter 1 1. Background 1.1. Breast Cancer Statistics Breast cancer is the second most common cancer diagnosed in women around the world, having 1.7 million new cases diagnosed in 2012 according to the World Cancer Research Fund International. In the United States, the estimates for breast cancer fall second to lung cancer for women. The latest statistics obtained in the year 2012 reveal that 224,147 women and 2,125 men in the United States were diagnosed with breast cancer. Out of this group of patients, 41,150 women and 405 men died in the same year due to their cancer diagnosis (CDC Breast Cancer Statistics). Throughout the years, the use of a multidisciplinary approach surgery, radiotherapy, chemotherapy- has significantly reduced the mortality of breast cancer patients. Table 1.1 Percent of US Women who Develop Breast Cancer over 10-, 20-, and 30- Year Intervals According to the Current Age, Source: National Cancer Institute Statistics It has been found that the chances of getting breast cancer increase with age, as it is summarized on Table 1.1. According to the National Cancer Institute, 0.44% of the women who are now 30 years old can develop cancer sometime in the next 10 years. This estimate increases to 4.07% if the time lapse increases to 30 years after the current age. 1

18 As the diagnosed population ages, new factors have to be taken into consideration for successful interventions. Nowadays, cancer survival treatments do not only focus on stopping or slowing down the biological disease, but also on improving the quality of life of those patients. A standard breast cancer treatment approach often includes lumpectomy, systemic therapy (chemotherapy or hormone treatment) and radiation treatment. The specificity to the tumor cells of these treatments decreases the harm done to the healthy tissues, thus maintaining or improving the quality of life of the patients Treatments Available for Breast Cancer The American Cancer Society classifies the treatments available in three categories: surgery, systemic therapy and radiation treatment. The optimal treatment depends on the biological characteristics of the cancer as well as the staging at the time of diagnosis. Adjuvant treatments are not independent of each other. Here lies the importance of highly targeted treatments for the correct type of cancer. Highly targeted treatments do not only offer the patient the chance to have a better quality of life, but it also decrease the side effects when two or more therapies are combined. A review of these treatment categories is presented in this discussion. This research project focuses on the nature of the radiation treatment and irradiation dose distribution rather than other types of treatments, which are out of the scope of the field of Medical Physics Surgery. One of the most common treatment approaches after breast cancer diagnosis is surgery. After confirming the initial stage of the disease, surgery is used to remove the gross part of the cancer tissue as well as sentinel lymph nodes for pathological evaluation (re-staging). The surgical interventions range from performing breast-conserving surgery to radical mastectomy. Following these procedures, it is 2

19 possible to perform a reconstructive breast surgery for cosmetic purposes. Depending on the age at the time of diagnosis, it is estimated that 20-40% of women who have had mastectomy have breast reconstruction with implants or tissue from another part of the body, or a combination of the two (Breast Cancer Facts and Figures). Breast-conserving surgery (BCS) covers partial mastectomy, quadrantectomy and lumpectomy. The aim of these surgeries is to remove cancerous tissue with a margin of normal tissue, which is tested for cancer cells. On the other hand, simple or total mastectomy is the removal of the entire breast. Radical mastectomy includes the removal of the chest wall muscle and modified radical mastectomy includes the removal of the axillar lymph nodes for pathological testing and re-staging. Both BCS and mastectomy can remove the regional lymph nodes to pathologically determine if the disease has spread beyond the breast and help choose an appropriate course of treatment Systemic Treatments. Systemic therapy refers to treatments that affect all parts of the body through distribution of the pharmaceutical in the bloodstream after injection or ingestion. Chemotherapy, hormone therapy and targeted therapy are all systemic treatments even though they work through different mechanisms. When these therapies are administered to patients before surgery (called neoadjuvant therapy), their goal is to shrink the tumor and make the surgical procedure less extensive. In fact, neoadjuvant therapy has been found to be as effective as systemic therapy given after surgery in terms of survival, disease progression and distant recurrence (Breast Cancer Facts and Figures). The purpose of systemic therapy given after surgery (called adjuvant therapy) is to kill undetected tumor cells left from surgery. Patients with metastatic breast cancer are good candidates for systemic therapy due to the spread of the disease. 3

20 1.2.3 Radiation Therapy. Radiation therapy uses high-energy external beam treatments and/or internal brachytherapy sources for the treatment of breast cancer. Radiation therapy can be used to destroy cancer cells in the breast, chest wall, supraclavicular and axillar region. More than 20 years of follow-up data confirm that after breast conservative surgery, there is a higher risk of local recurrence. Women who choose breast conservative surgery and radiation treatment have the same long-term survival as if they had chosen mastectomy (Fisher, B., et al) Local Epidemiology Dana Cancer Center is located in Toledo, in the state of Ohio. It is a part of the University of Toledo Medical Center and serves the Toledo population as well as part of northwest Ohio. This center provides the following services: Radiation Oncology, Infusion Center, Radiology, clinics for surgical oncology and other cancer related specialties, Laboratory, Survivor shop, and a Center for Health/Successful Living. Getting a patient admitted to Dana Cancer Center does not mean necessarily that he or she will use all services. Table 1.2 Treatments for Patients Diagnosed at UTMC with Breast Cancer Year Types of Treatment Patients Diagnosed at UTMC with Breast Cancer Radiation Chemotherapy Radiation and Not treated Only Only chemotherapy at UTMC Total Number of Cases Source: UTMC Tumor Bank Statistics Table 1.2 refers to patients who have been diagnosed with breast cancer at UTMC and excludes all other diagnosis-related consults. In 2013, 101 patients were diagnosed with breast cancer at UTMC: 30 patients were treated with radiation only, 15 patients were treated with chemotherapy only, and 21 patients were treated with both radiation 4

21 and chemotherapy treatments. In the same year, 35 patients were diagnosed at UTMC but they did not receive their treatment in the same institution. On the other hand, in the year 2014, 81 patients were diagnosed with breast cancer at UTMC: 24 patients were treated with radiation only, 15 were treated with chemotherapy only and 21 patients were treated with a combination of radiation and chemotherapy treatments. In the same year, 26 diagnosed patients did not have their treatment at UTMC. Table 1.3 Diagnosis per Patient Admitted to the DCC Radiation Oncology Department Prostate Breast Brain Bladder Lung Cancer Diagnosis Bone Pancreas Year Total (j-o) Source: Dana Cancer Center Aria Radiation Oncology Patient Information System Table 1.3 includes all Radiation Oncology consults registered in the Aria Radiation Oncology Patient Information System. These patients have not necessarily been diagnosed at UTMC, thus the difference with Table 1.2 for breast cancer statistics. In the year 2013, breast cancer was the diagnosis of 23% of all patients admitted to the DCC Radiation Oncology Department, followed by lung cancer with 13%. In the year 2014, the cases for breast cancer increased and were accounted for 21% of the diagnosis of all patients admitted, followed by lung cancer (17%), and brain cancer (13%). Finally, since January to October 2015, breast cancer was the diagnosis of 19% of all patients admitted. Even though the percentages fluctuate year to year, the number of total patients admitted increases as well as the number of patients with other types of cancers. Patients 5 Esophagus Skin Lymphoma Rectum Cervix Liver Other

22 diagnosed with breast cancer conform the majority of all patients admitted to the Radiation Oncology department at the UTMC Dana Cancer Center for 2013 and 2014, and falls to second place in Figure 1.1. Diagnosis per Patient Admitted to DCC Radiation Oncology Department Source: Dana Cancer Center Aria Radiation Oncology Patient Information System The current treatments for breast cancer at Dana Cancer Center include the use of pharmacological agents, surgery and radiation therapy. Except for very unique chest geometries, most patients undergo 3D conformal radiation therapy in the supine position for breast cancer. Patients with pendulous breasts undergo treatment in the prone position. The rest of patients are treated with either IMRT or VMAT in the supine position. The advancements in early diagnosis and treatment modalities have resulted in a large population of long-term survivors. Evaluating the current treatment techniques for breast cancer enables health care providers to provide better disease control and care for these patients. 6

23 Chapter 2 Literature Survey 2.1 Irradiation Modalities There are four techniques for Whole Breast Irradiation (WBI) used at our institution: supine 3-D conformal, prone 3-D conformal, supine IMRT and supine VMAT. The supine 3-D conformal technique is predominant over the others and has become a standard practice due to its practical and reproducible setup. The other three techniques are used in special cases such as irradiating a target in a challenging thoracic position (IMRT or VMAT) or for patients with pendulous breasts (prone technique). In order to evaluate how these techniques perform through the treatment course, it is important to understand their nature as well as to compare them dosimetrically Supine and Prone 3D Technique Considerations. The main types of positioning for WBI treatments in Radiation Oncology cover two patient positions: supine and prone. The positioning for each case depends on institutional protocols as well as breast and chest geometry of the patient. The main advantage provided by the prone positioning WBI is the dislocation of the breast tissue for women with large or pendulous breasts. This decreases the skin folds accountable for the bolus effect observed in regions of skin-to-skin contact. Large breasts pose a dosimetric challenge in terms of dose inhomogeneity and hot spots in the treatment plan. In fact, increased radiation related toxicity and worse cosmetic outcome has been found in individuals with large breasts and/or increased BMI (Moody, et al. 1994). The position of the lesion or tumor bed is also important. Prone breast positioning 7

24 can separate the breast tissue from the chest wall, allowing for better treatment geometry. Coverage of the chest wall in the prone position is not as important as in the supine position: This coverage exists in the supine position just to cover the whole breast tissue as seen in figure 2.1. Figure 2.1 Supine tangent setup (left) vs. prone tangent setup (right) for the same patient. The tangents cover the chest wall and heart in the supine position while the dislocation of the breast in the prone position makes it possible to treat the breast without irradiating the heart and lungs. Source: Formenti, S. et al. Seminars in Radiation Oncology, 2004 Breathing patterns also prove dosimetrically favorable in the prone setup as the chest wall has limited range of expansion due to the nature of the position. It has been found that prone setup can reduce intra-fractional respiratory motion of the chest wall with a mean interfraction setup variability of less than 0.1 cm (Mitchell et. al., 2010). While supine positioning for WBI provides ease of patient positioning, it certainly can make the target concave, enwrapping itself around lungs and heart. Supine WBI positioning has limitations such as the lateral dislocation of the breast, formation of inframammary folds and the possible unintended presence of lung and heart tissue within the tangent fields when attempting to cover all the target breast tissue. Even though positioning does not seem to be difficult to hold for the majority of patients, lateral 8

25 dislocation of the breast can be unpredictable and difficult to reproduce in patients with large breasts Dosimetric Comparison Studies. Krengli et. al., conducted a prospective study comparing prone and supine WBI for patients with pendulous breasts. 29 patients with pendulous breasts and infra-mammary folds when supine were selected for prone WBI treatment while 14 patients were excluded because of insufficient gantry diameter of the CT scanner or low compliance to prone setup position. Besides dosimetric parameters, late effects were assessed during follow-up visits, which ran every 4 months after 6 months after radiotherapy for the first 2 years, and 6 months afterwards. Statistical analysis of the results to prove significant differences between the two setups was performed with the Student s t-test considering a p-value< 0.05 as statistically significant. The treatment reproducibility for the prone setup was evaluated using portal imaging checks, which showed inter-fraction differences of 2.0 mm, 1.8 mm and 2.5 mm in the lateral, longitudinal and vertical axes. The authors found that CTV and PTV coverage was significantly better in the supine position than in the prone position taking into consideration the irradiation to OARs and target shape. Lung V5, V10, and V20 were significantly lower in the prone position with p-values <0.05. The heart V5, V10, V20 and LAD mean and maximum doses were lower in 17 patients who were treated for left breast cancer. No significant difference was observed between the patients treated in prone and supine position for either acute or late toxicity. Similarly, Formenti et. al., studied the dosimetric differences between supine and prone positioning for WBI in 400 patients (200 patients had right breast cancer and 200 patients had left breast cancer). 314 patients had invasive breast cancer, including 47 9

26 individuals who had involved sentinel or axillary lymph nodes. The patients were simulated in both positions: supine and prone. A 3-D conformal treatment was planned for both datasets, placing the fields on a plane connecting the midline to the anterior extent of the latissimus dorsi muscle, achieving comparable coverage of the breast in both positions. For all patients, there was a reduction of the in-field lung volumes in the prone position compared with the supine setup: an 82.2% reduction was observed for right breast cancer and a 91.1% reduction for left breast cancer. In patients with left breast cancer, the prone position had an 85.7% reduction of the in-field heart volumes, except for 15% of patients for whom were found of having less in-field heart volume in the supine position. In this study, prone positioning for WBI had reduced the amount of irradiated lung in all patients, and the amount of heart volume irradiated was reduced in 85% of patients with left breast cancer. Mulliez, et. al., takes into account not only the different positioning with WBI 3- D conformal technique, but also makes a dosimetric comparison with tangential and multi-beam IMRT. For this study, 3 plans were created for 18 patients using their simulation CT dataset in the supine and prone positions: multi-beam IMRT (6 fields), tangential beam IMRT (2 fields), and wedged tangent fields (2 fields). The IMRT plans experienced the same benefits of wedged tangent fields between the prone and supine positions: The prone IMRT had a better dose homogeneity than prone wedged tangential fields and both supine IMRTs. The multi-beam IMRT lowered both lung and heart dose in the supine position (p<0.05), although the lowest ipsilateral lung doses were found in the prone position (p<0.001). Doses to the contralateral breast were similar regardless of positioning or technique. In the supine position, multi beam IMRT was the treatment of 10

27 choice in this study, and the prone IMRT was superior to any supine treatment for patients with right breast cancer. Prone IMRT obtained better conformity indices, target dose and sparing of the organs-at-risk Inverse Planning Techniques: IMRT and VMAT Technique Considerations. The Volumetric Modulated Arc Therapy (VMAT) technique consists of arc beams optimized for dose modulation. One of the main problems of VMAT for WBI is the potential lack of skin flash. Strategies to incorporate flash will be reviewed but they can be summarized to a single technique: adding a virtual bolus by overriding air to the density of water in the tissue around the breast. The most important aspects of setting up a VMAT for WBI are: machine limitations and target size, coverage range of arcs and number of control points Surface Dose Measurements. The use of inverse planning limits the amount of skin flash in whole breast irradiation versus standard 3-D conformal planning. The skin flash region accommodates breast volume changes due to setup uncertainties and intra- and inter- fractional changes due to respiratory movement, inflammation, edema and others. Strategies to incorporate flash into inverse-planned treatments found in literature typically involve overriding air to the density of water or tissue by the use of a virtual bolus. Since the treatment plan is optimized with virtual bolus at first, overwriting its density to air can introduce uncertainties to the superficial dose distribution, potentially degrading the coverage at the skin-bolus interface. It is important to investigate the accuracy of various commonly used bolus materials to incorporate flash in inverse-modulated WBI plans while minimizing the perturbation near the skin. 11

28 Zankar et. al., tested the intensity extension and their impact on entrance dose for breast radiotherapy in IMRT plans using TLDs and the Eclipse TPS Skin Flash tool and virtual bolus technique. This study was done on a hypothetical target volume resembling the shape of a breast attached to a PTW cylindrical phantom. Treatment plans were generated with and without the skin flash tool and the values were compared. The surface dose readings for IMRT plans without skin flash were low, having a value of 10% of the dose of those plans with dose extension. These measurements were done with 3 mm backscatter material on the dosimeters, hence the surface measurements were done at 3mm from surface. The group also demonstrated that there was no significant difference in entrance dose due to different methods of skin flash implementation (virtual bolus or IMRT skin flash tool). The lack of skin flash implementation did show a potential for under dosage in the surface region, especially when small positional errors or patient movement is involved. Berg et. al., made measurements on surface dosimetry for VMAT plans in different anatomic places using MOSFETs. The team used three MOSFET detectors placed adjacent to each other at the point of interest on the patient s skin to improve measurement statistics due to the non-uniformity of the VMAT beam. A 3.0 x 3.0 cm mini-bolus of D max thickness for the given x-ray beam was placed over the detectors during treatment delivery and during planning. All the results fell within 10 percent of the treatment planning system predictions. Ito, et. al., made skin surface measurements for the verification of skin dose in postmastectomy helical tomotherapy with Thermoluminescent Dosimeters (TLDs) in 14 patients. A potential issue with chest wall irradiation treatment plans is the accuracy of TPS dose calculations at shallow depths. A physical bolus was used, covering the TLDs 12

29 with 1 cm in the PTV during irradiation and plans were optimized with the bolus in the TPS. The mean difference and standard error of the mean difference between the measurement and calculation for scar measurements was 1.8%±0.2%. The mean difference between measured and calculated TLD doses was statistically significant at two standard deviations of the mean, but was not clinically significant (<5%) and the mean of the measured TLD doses agreed with the calculated doses within 5% Target Volume Correction. Breathing motion or breast volume changes in the positive margin (outside the patient) is a factor that can easily be taken care of in 3D conformal planning by opening the field and creating a skin flash area. The position of the MLCs on an inverse planned treatment conforms to the target, making it difficult to integrate this area. As with any other radiation treatment delivery, immobilization devices help to reduce inter- and intra-fraction motion of the target during irradiation. The current use of IGRT plays an important role in helping minimize the effect of inter-fraction motion by analyzing patient positioning daily and performing online correction of set offs. Approaches such as deep inspiration breath-hold techniques during VMAT irradiation are suggested in literature (Tsai, et al.)(michalski, et al.) in order to avoid significant variations in dosage to the PTV as well as to reduce the dose to delivered to the heart volume and lungs. Gating techniques have also been suggested (Nicolini, et al.), but their implementation is not time effective in most institutions. Nicolini et al., has been able to explore both possibilities: delivering respiratorygated VMAT and accounting for respiratory motion during the treatment planning. The respiratory gating approach used for the Real-time Position Management (RPM) respiratory gating system was from Varian in a group of six patients and it was tested on 13

30 an Octavius phantom (PTW, Freiburg). The gated delivery was compared with the nongated planed dose and the dosimetric impact was a mean Delta MU <0.02% for all gating conditions. The gated VMAT did not affect the quality of dose delivery but it did interrupt a single arc 20 to 50 times during delivery, which can be time consuming and difficult to use if the patient does not have regular breathing patterns. On their other approach, Nicolini et al., expanded the contour of the breast area to the outside of the CT image and made its density equivalent to soft tissue to extend the dose fluence outside of the body contour to account for changes in size and position of the target and other tissues due to respiratory movement or edema. VMAT plans were optimized for six patients for whom three datasets were used: an original planning CT and a planning CT with an artificial expansion of 10 mm and 5 mm outside of the body in the breast region which was assigned a HU number equivalent to that of soft tissue. It was concluded that the target coverage could be compromised if no action is taken towards correcting for edema or respiratory motion in modulated radiation treatments. If no steps are taken in case of important changes in morphology, for example 10 mm, the target dose could be affected with 4% average reduction, while a 5 mm discrepancy could lead to a 5% reduction in v95% rising to 25% in the case of 10 mm. This study only included whole breast irradiation treatment plans and did not include radiation treatments in which the internal mammary or supra-clavicular nodes were involved. In order to further analyze the importance of having an optimization PTV structure that accounts for volume changes, Michalski, et al., analyzed eighteen articles about the inter- and intra-fraction motion during radiation therapy to the whole breast in the supine position, ranging from the years 1991 to The motion was measured for 14

31 each image set acquired for each patient based on five parameters: central lung distance, central irradiated width, central beam edge-to-skin distance, cranio-caudal distance and/or central breast distance. From the seven groups that reported the intrafraction motion of the target in breast cancer patients never exceeded 5 mm. The intrafraction motion is naturally smaller than interfraction motion, since interfraction motion includes changes in the tissue such as edema or tumor shrinkage. Michalski, et al. recommends a clinical PTV margin of 5 mm for all whole breast radiotherapy Dosimetric Comparison Studies. In 2010, Popescu et al., conducted a study comparing 3-D conventional, IMRT and VMAT techniques on 5 previously treated patients. Her team demonstrated the possibility to produce high quality plans using VMAT for large treatment volumes of an average of 945 cc by using two coplanar arcs of 190 with a 2 cm overlap jaws. Her group concluded that this technique successfully covered the large PTV regardless of the short MLC travel distance of Varian Machines (15 cm). The VMAT plans had a total of 386 MLC segments, giving enough control points to achieve an optimal dose distribution as well as to spare OARs, regardless of the long optimization time (10 hours). Popescu et al. concluded that the PTV homogeneity was similar across all techniques being compared. The average heart volumes receiving >30 Gy for VMAT, IMRT, and 3-D conformal were 2.6% ± 0.7%, 3.5% ± 0.8%, and 16.4% ± 4.3%, respectively, and the average ipsilateral lung volumes receiving >20 Gy were 16.9% ± 1.1%, 17.3% ± 0.9%, and 37.3% ± 7.2%, respectively. The healthy tissue volume percentages receiving 5 Gy were significantly larger with VMAT (33.1% ± 2.1%) and IMRT (45.3% ±3.1%) than with 3-D conformal (19.4% ± 3.7%). VMAT decreased the number of monitor units by 30% and the treatment time by 55% compared 15

32 with IMRT. In more advanced approaches, Popescu and her team have also compared traditional 3D conformal techniques with static couch VMAT and simultaneous couch rotation VMAT in 20 patients. The VMAT plans consisted of a single 200 to 240 degree arc, while the VMAT with couch rotation consisted of two tangential couch arcs with the same isocenter with simultaneous non-coplanar gantry rotation directions. Even though the VMAT with dynamic couch rotation yielded superior results to static couch VMAT in OARs sparing, the mean dose to the ipsilateral breast increased by up to 15% compared to 3D conformal technique. The dynamic couch VMAT resulted in superior target coverage when compared to 3D-CRT and VMAT being the V95%: 98.2% vs. 97.1% respectively. This type of treatment planning requires a modification of the VMAT algorithm to link the couch and gantry rotation. Tsai, et al., studied the feasibility of using multiple partial arcs for VMAT plans in early stage left-sided breast cancer patients. For this study, the VMAT technique was planned for ten patients with left breast cancer and with a PTV average volume of cc. The plans had six partial arcs with a 50-degree gantry rotation, with the two closest arcs to the chest wall with their collimator rotated to decrease dose to OARs. The optimization was done with the Anisotropic Analytic Algorithm (AAA) with heterogeneity correction in Eclipse 8.6. This arc distribution and optimization achieved sparing of the organs at risk while reaching 95% of the dose coverage to 95% of the PTV volume, with a maximum dose of ±1% of 110% of the prescription dose. The mean dose, V5, V10, V25, and V30 of the heart were 7.61 ± 1.38 Gy, 59.73% ±15.87%, 24.39%± 6.82%, 2.52%±1.11%, and 1.57% ± 0.71%, respectively. The mean dose, V5, V10, V20 of the ipsilateral lung were 8.22±0.57, 40.46±3.81, 23.32±2.07, and 12.71±

33 respectively. Other researches such as Marlies, et al., have also put further consideration to the beam arrangement by comparing a plan with one 230-degree arc with another plan with two small tangential arc segments of 50 degrees. This group compared these plans for ten patients with positive lymphonode left-sided breast cancer and concluded that the VMAT technique with the smaller tangential arc segments enabled better OAR sparing and better dose distribution in the target; The heart received a mean dose of 4.4 ± 0.8 Gy using single arc VMAT and 3.3 ± 1.0 Gy using the second VMAT technique, the contralateral breast received 1.5 ± 0.3 Gy and 0.9 ± 0.3 Gy, respectively Treatment Reproducibility Breast Volume Variations. Jin, et al., conducted a randomized study to treat patients who had undergone breast-conserving surgery with different types of WBI. The volume of the PTV in these patients ranged from cc to cc, with an average of 360.8±149.1 cc, which correspond to small breast volumes. These volumes are significantly less than those reported by Popescu (945 cc), Tsai (562.1 cc) and other authors. Even though the optimization process was run with the same constraints for all plans, the VMAT plan did not achieve the same degree of dose homogeneity in the PTV as the other planning techniques including IMRT and 3-D conformal. It was found that the VMAT technique had the greatest scatter dose to the contralateral breast. On the other hand, Dumane, et al., tested the VMAT technique on the case of one patient with a modified radical mastectomy of the right breast with flap reconstruction. In this case, other techniques such as IMRT and 3D were also planned. While the VMAT plan was comparable to the 3D CRT plan in PTV coverage and dose homogeneity, VMAT was the only plan to lower the dose to the heart and lungs in comparison to the other plans. In 17

34 cases in which the mastectomy has left little to none breast tissue, techniques such as 3D CRT and tangential IMRT have been proposed, with the modulated planning sparing the OAR on either breast. The size of the primary breast volume does not only affect the dose distribution of modulated techniques, but it can also be the cause of scatter to the opposite breast. A study done by Surgent et al., demonstrates the relationship between main breast size and scatter dose to the opposite breast. This study consisted of sixty-five patients (thirty six patients had left breast cancer) who were treated with WBI IMRT. The dose on the contralateral breast was measured with thermoluminescent dosimeters placed four centimeters from the center of the medial border of the primary breast irradiation field. The volume of the primary breast for this study ranged from to cc, having a positive correlation with the irradiated volume through a correlation test value of 0.552, which they found to be statistically significant. All other OARs, such as the ipsilateral lung or the heart dose did not have such correlation Treatment Immobilization. Having a radiation treatment plan with a good dose distribution does not guarantee a successful treatment course delivery if the immobilization setup is not reproducible. Having an effective immobilization system can make the positioning variations minimal and improve the total outcome of the radiotherapy treatment. There are specific immobilization devices for specific cancer sites as well as techniques. For WBI, the most common immobilization devices are butterfly boards and Vaclok bags in the supine position and prone breast boards. The study of the effectiveness of an immobilization device has been studied using two-dimensional (2D) and three-dimensional (3D) information. Two-dimensional 18

35 information can be obtained through MV portal images as well as kv radiographs with the disadvantages that one cannot assess rotational variations and that image quality might pose a difficulty measuring other parameters. The integration of three-dimensional (3D) imaging not only facilitates the setup positioning during treatment, but also can provide additional information in a similar way to a simulation CT dataset. Therefore, 3D images provide more information than 2D images and it can be said that they are superior tools to assess the reproducibility of immobilization devices. Inverse planned radiotherapy delivers highly conformal dose distribution to spare critical organs and normal tissues while achieving optimal target coverage. Image-guidance during treatment delivery as well as a reduction in setup uncertainty become more important to guarantee correct treatment delivery. In a study by Cheng, et al., daily treatment set-up MVCT data from 212 patients was evaluated in terms of reproducibility and set-up corrections using an Accuray helical tomotherapy unit. Within the sample population, 41 patients were evaluated for chest radiation treatment accounting for a total of 862 treatment fractions. Cheng assessed the reproducibility by looking at the lateral, longitudinal and vertical locations of a point in each MVCT as a measure of translational deviation and roll locations for each patient. The systematic error was calculated as the average error over all the treatment fractions and random error was calculated as the average magnitude of errors that were expected to be distributed as a Gaussian function about a mean. Other parameters included a 3-D vector resultant from the displacement of the treatment position from the reference position calculated with the formula 19

36 2.1 In this study, the prone breast group used the CIVCO Medical Solutions prone board while the supine breast group used a Vaclok and a headrest. Cheng and his team discovered that for the breast cases, the systematic error in the long position, and the random errors for the lateral, long and vertical positions were greater for the prone breast group than the supine breast group by 2.2, 3.8, 2.0 and 1.7 mm respectively. The mean 3D vectors of the prone breast group and the supine breast group were 13.9 and 7.2 mm respectively. For the prone breast, shifts more than 6 mm were more frequent (91.2%) than for the supine breast group (66.2%). No breast volumes were obtained for either group, but the authors did note that the treatment region is more mobile in patients with large breasts. For the rest of the study, the overall set-up accuracy was acceptable with mean set-up variations being less than 10 mm. Another setup uncertainty study done by Oh, et al., assessed the setup uncertainties for various tumor sites with CBCT for more than 2200 VMAT treatments. Among those patients whose setup was evaluated, 19 thorax pathologies were treated with a total of 313 fractions. The distribution of setup corrections in all directions and frequencies of 3D vector lengths were analyzed. The setup errors calculated in this study were those defined by Remeijer et al, as the overall mean error (M), the systematic error (Σ) and the random error (σ): 20

37 2.2 F p is the number of measured fractions for each patient p, f is the number of treatment fractions, x pf is a measurement of a setup error, m p is the average patient setup error and N is the total number of measured fractions. The 3D vector length of translational shift of more than 1 degree was the second highest for the thorax (first was the brain) with 19.5%. The overall mean error in the lateral, longitudinal, vertical and rotational directions was the highest for the thorax with 0.5, -2.0, 1.0, and -0.1 respectively. For systematic and random errors, the thorax also has the highest values of all other anatomical sites being evaluated such as the brain, and head and neck. The authors note that not only bony structures were taken as reference on daily CBCT but changes in target, tumor location and shrinkage also affected the setup reproducibility. 21

38 Chapter 3 3. Methods and Materials 3.1 Dosimetric Comparison Planning Techniques D Conformal WBI. Once the simulation takes place and the contours are drawn on the planning CT, they are transferred to the planning system (Pinnacle 9.8). On the planning system, the simulation isocenter is found by looking at the marks (crosshairs and lateral marks) and a point of interest is created at this location. The simulation isocenter is usually set at the patient s surface in our clinic with an SSD of 100 cm for the ease of localization. On the same CT slice as the isocenter point, a new isocenter for the beams is placed so that the tangent beams have the same SSD. The gantry angles are chosen so that the medial and the lateral field wires have an overlap with the central axis of the field. The superior and inferior borders of the fields are set to match their respective wires and the lateral borders are set at least 2 cm beyond the skin edge. The jaws are brought to block half of the field and the collimator angle is set to even the amount of chest wall coverage on the superior and inferior borders of the field. Once the beams are set up, the radiation oncologist delineates the first block and adjusts the flash region. After these blocks are drawn and approved by the radiation oncologist, control points are added to reduce the hotspots using the Field in Field technique. The dose distribution is normalized to a calculation point. The maximum acceptable hotspot is % but lesser hotspots are usually accomplished in smaller 22

39 patients. For this study, the irradiation of the whole breast only was considered. Irradiation of axillary or supraclavicular nodes was not taken into consideration VMAT / IMRT WBI. For the general setup of the VMAT WBI technique, a few optimization contours are drawn as follows: PTV: In case that this contour was non-existent, it was drawn from the physician approved prescription isodose line in the 3-D conformal plan. This contour was fixed to avoid irregularities as well as to cut structures that did not belong to the anatomy of the whole breast. 2 mm were subtracted from the surface to account for skin sparing. For this study, the irradiation of the whole breast only was considered. Irradiation of axillary or supraclavicular nodes was not taken into consideration. PTV ring: It was a structure with 1 cm thickness, away from the PTV by 1.5 cm. It was the result of the 1.5 and 2.5 cm expansions of the PTV, avoiding the exterior of the exterior contour. GI Avoid: GI stands for the term gastrointestinal. This contour has the same structure as the PTV ring but it only covers what is considered gastrointestinal anatomy. This contour helped control the 50% isodose line spill into the liver, bowel and stomach when applicable. PTV Optimization: This contour was the result of expanding the PTV contour anteriorly by 1 cm (including the bolus). This structure was used for optimization and it accounted for any movement or volume variation of the PTV during treatment delivery (respiratory motion) or interfractional target variation. Bolus: A bolus structure was drawn as an external 1 cm contour over the target breast to force the treatment planning system to allow a region of skin flash in the case of 23

40 the inverse modulated plan. The density of this structure was overridden to wet towel density of 0.65 g/cc. Organs at risk: The contoured OARs were: the lungs, the heart, the contralateral breast, and sternum. Figure 3.1. Optimization contours for Inverse Modulated Plans: PTV (yellow), PTV ring (orange), PTV optimization (purple), contralateral breast (green), sternum (red). The beam arrangement consisted of two coplanar dynamic arcs with no collimator rotation. The fields were long enough to cover the PTV in the superior and inferior aspect of the patient: the width of each field covered 14 cm, with a jaw overlap of 2 cm between the two fields. The first optimization was done with the SmartArc algorithm, having 135 maximum iterations and a value of 35 for the convolution dose iteration setting. The SmartArc algorithm was configured to not have the allow jaw motion setting so the jaws could stay at their original position. The first optimization was run along the following parameter template below. Since prescriptions might differ from case to case, a description of what was done is described. 24

41 Table 3.1 Summary of Inverse Planning Optimization Parameters Region of Parameter Type Target (cgy) Volume Weight Interest PTV Opt Min Dose Prescription N/A 25 PTV Opt Max Dose 105% of the Prescription N/A 1 PTV Opt Uniform Dose Prescription +50 cgy N/A 75 PTV Opt Min DVH Prescription PTV Ring Max Dose Prescription N/A 1 Avoid GI Max Dose Prescription N/A 1 Heart Max DVH Check individual plan 1 Heart Max Dose Less than 50% IDL N/A 1 Left Lung Max DVH Check individual plan 1 Left Lung Max Dose Less than 50% IDL N/A 1 The minimum dose objective applies to the region of interest. This objective is met when the minimum dose is greater than or equal to the target dose and it is always set at 100%. The maximum dose also applies to the whole region of interest and it is met when the maximum dose is less than or equal to the target dose, and it is set to 0% volume. The uniform dose objective applies to the whole region of interest (PTV Opt) and helps the dose distribution to become closer to its target prescription. The minimum and maximum dose DVH set particular areas of the DVH to a certain goal. In the case this optimization, the 3D conformal plan was below the Timmerman s radiobiological constraints for fractionated radiation treatments. After the first optimization, further optimizations with 75 maximum iterations and 15 convolution dose iterations were run, adjusting the parameters and adding new objectives according to the initial dose distribution. The dose distribution was set up to be proportional to the number of monitor units with 95% of the PTV volume being covered by the prescription. The IMRT optimization for each plan was run on the same list of optimization parameters from the original VMAT plan. Six to seven beams were used to achieve a 25

42 total of control points per IMRT trial. Minor modifications were done to the optimization process as it was needed Dosimetric Criteria for Analysis The quantification of dose coverage or sparing of the OARs was given by the following indexes: The hotspot is the highest isodose line within the PTV contained by 3% of total PTV volume. The RTOG Conformity Index is a measure of how well the prescription volume conforms or fits the target. It is found by dividing the prescription isodose line volume by the PTV volume. The volume of the PTV was obtained from the compute ROI volume option in Pinnacle TPS, while the prescription isodose line volume was obtained through the tabular view in the DVH plot in the TPS. CI = Prescription isodose line volume PTV volume 3.1 The Paddick conformity index is found by squaring the overlap volume between the PTV and prescription isodose line and dividing this quantity by the PTV volume multiplied by the prescription isodose volume. This index is different from the RTOG CI in the aspect that it takes into account the overlap from the PTV and the prescription isodose line. For cases in which the PTV and prescription isodose line do share the same or approximate volume, this index helps analyze whether this dose is inside or outside the PTV. Paddick CI = Overlap Volume2 (PTV vol)(pres. vol)

43 The RTOG Homogeneity Index is the ratio between the difference of maximum and minimum dose and the prescription dose. However, Dmax and Dmin are not representative of true maximum or minimum dose values within the PTV. These values are sensitive to dose-calculation parameters such as the grid size and grid placement. This is the reason to use a modified homogeneity index HI including the hotspot as maximum dose. HI = Hotspot Dmin Prescription Dose 3.3 The HI indicates the ratio between the maximum and minimum dose in the PTV. A low value indicates a more homogeneous dose distribution within the PTV. The 50% IDL in Organs-at-Risk percent is the ratio between the 50% isodose line volume in the organs at risk (lungs, heart and contralateral breast) and the 50% isodose line volume in the patient minus the PTV volume. It is a measure of fall-off dose from the PTV target. 50% IDL in OARs % = 50% isodose line volume in OARS 50% isodose line volume outside PTV 3.4 An average of percent isodose lines and percent volumes was configured to obtain the summary results due to the different prescriptions for some of the patients. 3.2 Inverse Planning Skin Flash Measurement of Surface Dose. An anthropomorphic phantom with a left breast attachment was simulated using a Phillips Gemini TF PET-CT scanner with the standard thorax simulation protocol at the clinic. The isocenter was marked with CT markers during the simulation process. A DICOM viewer and image processing software 27

44 (MIM) was used for contouring OARs and the TPS (Pinnacle 9.8, Phillips) was used to construct the optimization contours. Three VMAT plans were created with different clinically available boluses: 1 cm of 1 g/cm 3 bolus (Plan 1), 1 cm of 0.65 g/cm 3 bolus (Plan 2) and 1 cm of g/cm 3 bolus with 2 dose levels accounting for the difference between bolus and tissue density (Plan 3). For Plan 3, the two levels consisted on the in-body PTV and the bolus only PTV. Plan 1 was optimized to use the Superflab bolus material, while Plan 2 and Plan 3 were optimized to use wet towels as boluses. The treatment plan consisted of two coplanar beams covering 190 degrees over the left breast. The density of the bolus structure was overridden to its respective value prior to the optimization process. The plans were optimized using the SmartArc algorithm and a dose matrix resolution of 0.3 cm. After optimization, the treatment plans were irradiated on an ArcCheck phantom for quality assurance purposes. All the plans had a passing rate equal or superior to 95% with a gamma analysis of 3% dose difference and 3mm distance to agreement (DTA). A Varian Truebeam accelerator was used to deliver the plans to the phantom. Each delivery was preceded by a CBCT using the accelerator s OBI for setup and dosimeter localization processes. The dosimeters were positioned under the bolus and prior to the CBCT acquisition time. The same positioning was followed for both types of dosimeters: two dosimeters on the medial, lateral and tip aspects of the phantom s breast. After irradiation, the exact position and dose of the dosimeters was obtained from transferring the dose distribution to the CBCTs and looking at the position of the dosimeters. The OSLD dosimeters were expected to have a total uncertainty of ±5% for screened dosimeters according to the manufacturer (Landauer, 2012), but according to 28

45 the latest calibration and evaluation of the equipment at our institution, the random error of our dosimeters was ±3%. a. b. c. Figure 3.2. General irradiation setup: a. Dosimeter placement on the medial, tip and lateral aspects of the breast. b. Superflab bolus setup. c. wet towels bolus setup. Figure 3.3. PTV coverage. 50 Gy (green), 32.5 Gy (blue), 25 Gy (purple): a. 1 cm of 1g/cm 3 bolus plan. b. 1 cm of 0.65 g/cm 3 bolus plan. c. 1cm of 0.65 g/cm 3 bolus 2 PTVs plan Evaluation of Inter-fractional Breast Volume Changes. The setup image analysis using images of previously treated patients with a WBI treatment was done in 29

46 two parts because of the nature of the datasets: MV analysis (2-dimensional image) and CBCT analysis (3-dimensional image). The MV image analysis was performed in the Offline Review viewer in the Aria Information System. The DRRs from planning as well as each MV image obtained during the treatment course was measured in the following aspects: thickness of the breast at the base, thickness of the breast at middle length, length of the breast from the chest wall to the tip, angle of the breast with respect to the vertical. The thickness of the breast at the base was measured at the visible separation of the breast from the chest wall skin at the respective portal angle. The thickness of the breast at the middle length was measured halfway between the base and the tip of the breast. The length of the breast from the chest wall was measured by adjusting the contrast of the image to visualize the chest wall and the tip of the breast, dividing the breast equally in the superior and inferior aspects. The angle of the breast was measured between the line defining the length of the breast and the horizontal. Positive angles rotate towards the gantry and negative angles rotate towards the couch. The CBCT image analysis takes into consideration the same measurements as the MV image analysis, performed in the middle sagittal slice of the dataset. Breast volume variation evaluation was done by drawing a breast volume in all datasets: a base contour containing the chest wall was drawn and subtracted from the total breast contour. The base contour helped delimit the area of the breast volume while excluding the chest wall. The data was compared to the values measured on the simulation CT dataset. 30

47 Figure 3.4. Sample of breast measurements in a MV image For the parameter evaluation among all patients, the average values were obtained as a weighted average, taking into consideration the number of MV images from each tangent angle. The standard deviation was taken as a quadrature sum of the standard deviation for each type of tangent angle image. Correlation values were obtained with the data from all images using Microsoft Excel. 3.3 Evaluation of Immobilization Devices Anthropomorphic Phantom Setup Evaluation. The setup uncertainty of immobilization devices was evaluated for the Bionix Prone Breast System and the VacLok bag setup for supine positioning by placing an anthropomorphic phantom on the immobilization device and obtaining ten CT datasets for each device setup. The phantom was marked and aligned in the sagittal, axial and coronal LAP laser system lasers in each of the devices. CT markers were placed on the CT table to ensure device position constancy between scans. Additional CT markers were placed on the chin support area as well as in the abdomen cushion of the Bionix Prone Breast System to help with phantom alignment. No CT markers were placed on the VacLok system after 31

48 creating a vacuum inside the bag. The phantom was repositioned after each scan to simulate daily setup. The simulation CT scans were acquired with a Phillips Gemini TF Big Bore PET/CT scanner, using the thorax simulation protocol that is used at our clinic. Figure 3.5. Positioning of the anthropomorphic phantom in the prone breast board Image processing and positioning measurements were performed in MIM. For each one of the CT datasets, the same slice containing the central bolt of the phantom breast was identified. The contrast was adjusted to the bone window. Measurements in the superior, medial and inferior aspects of the axial cut of the sternum were used to measure the angle of this bony structure with respect to the vertical. The first dataset was considered as the baseline while the following datasets were considered as the daily setup images of a normal treatment. Figure 3.6. Identification of measurement slice in the anthropomorphic phantom and measurement of phantom rotation 32

49 The butterfly/breast board was not taken into consideration for this evaluation since the support it offers depends on the positioning of the arms of the patient and the phantom lacks these anatomical structures. The sternum was used for these measurements as a fixed reference structure. The spinal process of the vertebral body was not used because it was not always in correct alignment with the vertebra inside the anthropomorphic phantom. Averages, standard deviation and T-test values were obtained from the information obtained from the measurements, using Microsoft Excel. For the T-test, the setup was arranged to run a one-tailed test, taking into consideration unequal variances Patient Setup Evaluation. For every treatment fraction, the patients were immobilized and positioned with the help of the in-room setup lasers, according to tattoos made on their simulation day. They underwent imaging using the OBI and EPID from a Varian TrueBeam Linear Accelerator during normal and boost treatment fractions. MV images were taken twice a week from any of the treatment gantry angles before each fraction. In cases in which a CBCT was needed, it was acquired before every irradiation. After performing an automatic image match with the help of the Varian accelerator software, the resulting images were analyzed by both, therapists and physicians, for an accurate anatomic registration around the chest wall and breast tissue. The couch shifts were then authorized and performed. After each shift, additional imaging was required to confirm good anatomical placement. Monthly quality assurance of the OBI alignment was within ±1 mm in this institution. The images were reviewed using the Offline Image Review action in the Aria Patient Information System. The couch shifts from initial positioning to treatment 33

50 position were seen with the function show trends. The couch adjustments included three translational directions: lateral, longitudinal, and vertical. These values were recorded for all available treatments in the supine and prone positions and a 3D vector representing the overall shift was obtained with the formula: 3D Vector = lateral 2 + longitudinal 2 + vertical The patient axial rotation was measured from the CT simulation and registrations with the treatment CBCTs by picking the same thoracic vertebra in all datasets and measuring the angle of variations with respect to the vertical axis using the MIM software after exporting all CBCT images from the Aria Patient Information System to the MIM server. The data analysis included a treatment course average 3D vector, an average patient axial rotation, the calculation of standard deviations and T-tests to confirm the difference between the shifts done in the supine and prone positions taking into account all the data collected for all the available patients. For the T-test, the setup was arranged to run a one-tailed test, taking into consideration unequal variances. 34

51 Chapter 4 4. Results and Analysis 4.1 Dosimetric Comparison Supine and Prone 3D Conformal Plan Comparison. The comparison for 3D conformal plans between supine and prone positions was done in 5 patients who had two CT simulation datasets: one in the supine position using a Vaclock device or a butterfly board, and one in the prone position using our institution s prone breast board (Bionix). Dose distribution on the CT image as well as DVH for each of the individual plans can be found in Appendix A. Table 4.1 PTV dose distribution comparison for 3D conformal plans for the supine and prone positions Min (%IDL) Mean (%IDL) Max (%IDL) Hotspot (%IDL) PTV Standard CI Paddick's CI Fall Off Dose in OARs (%) Plan HI Supine 3D Prone 3D The average PTV coverage for all five patients is summarized and presented in table 4.1. This analysis indicates that the supine and prone 3D conformal plans have comparable PTV coverage, with similar hotspots. However, the minimum isodose line in the prone is higher by 14% increment. Other aspects, such as the conformity index are comparable in both planning techniques. The homogeneity index for the supine position is higher and this could be attributed to the lower minimum isodose line. The fall-off dose in OARs (%) is lower in the prone distribution by 1.23%. 35

52 Table 4.2 Contralateral lung dose distribution comparison for 3D conformal plans for the supine and prone positions Min (%IDL) Mean (% IDL) Max (%IDL) Contralateral Lung V25 (%vol) V20 (%vol) V5 (%vol) V5% of prescription (cc) Plan Supine 3D Prone 3D For the contralateral lung, both plans have comparable values. Small dose increments between supine and prone for maximum and mean IDL are not considered to be dosimetrically significant. Table 4.3 Ipsilateral lung dose distribution comparison for 3D conformal plans for the supine and prone positions Min %IDL Mean %IDL Max %IDL Vol prescription (% Vol) Ipsilateral Lung V25 (%vol) V20 (%vol) V10 (%vol) V5 (%vol) V30% of Prescription Plan Supine 3D Prone 3D For the ipsilateral lung, the prone plans deliver an overall less dose than the supine plans. Even though the maximum isodose line covering the ipsilateral lung is 3% higher in the prone plans than in the supine plans, dose drops off significantly when looking at other dose coverage parameters (V25, V20, V10, V5 and V30%). The percentage of volume that receives prescription is lower in the prone position (0.22% of the ipsilateral lung volume) than in the supine position (1.05% of the ipsilateral lung volume). Other parameter such as V25, V20, V10, V5 and V30% of prescription are 50% lower in the prone position than in the supine position. 36

53 Table 4.4 Heart dose distribution comparison for 3D conformal plans for the supine and prone positions Min (cgy) Mean (%IDL) Max (%IDL) Vprescription (%vol) Heart V25 (%vol) V20 (%vol) V10 (%vol) V5 (%vol) V2 (%vol) Plan Supin e 3D Prone 3D Table 4.4 summarizes the dosimetric parameters for the heart. Parameters such as V25, V20, V10, V5 and V2 are overall lower in the prone plan than in the supine plan. Table 4.5 Contralateral breast dose distribution comparison for 3D conformal plans for the supine and prone positions Contralateral Breast Plan Min (%IDL) Mean (% IDL) Max (%IDL) Supine 3D Prone 3D The contralateral breast does not receive considerable mean dose in the supine or prone plan. The only remarkable finding for the contralateral breast dosimetry is that this OAR was found to be covered with an isodose line of 37.16% in the prone plans, which is considerably higher than the values obtained in the supine plan. This can be a product of unique beam positioning for individual plans taken into account in the average results. Figure 4.1 and 4.2 contain the dose distribution and DVH for the dataset belonging to patient 4P. In the supine-prone 3D conformal plan comparison, the dose distribution for this dataset was considerably different from the other plans. While all other plans yielded low dose values for OARs in the prone position, the opposite trend was reported for this case (Figure 4.2). The main difference in this CT dataset over all other datasets is the shape of the chest wall. The anterior portion of the chest wall is wider than its posterior 37

54 counterpart. This box appearance (figure 4.1) affects the dose distribution coming from tangent beams. Achieving similar PTV coverage in the supine plan would have delivered full dose to the contralateral breast. Figure 4.1. Axial slices comparing the dose distribution for two 3-D conformal treatment plans for patient 4P. a. Supine 3D conformal dose distribution with a hotspot of 5400 cgy. b. Prone 3D conformal dose distribution with a hotspot of 5400 cgy. Figure 4.2. DVH comparison of 3D conformal dose distributions in the supine (dashed line) and prone positions (solid line) for patient 4P. 38

55 4.1.2 Supine 3D Conformal, IMRT and VMAT Plan Comparisons. This comparison was done in 8 patients, 5 of which were simulated in both supine and prone positions. Dose distribution on the CT image as well as DVH for each individual plan can be found in Appendix B. Table 4.6 PTV dose distribution comparison for 3D conformal, VMAT and IMRT plans in the supine position Min (%IDL) Mean (%IDL) Max (%IDL) Hotspot (%IDL) PTV Standard CI Paddick's CI Fall Off Dose in OARs (%) Plan HI Supine VMAT w/ Bolus Supine VMAT w/o Bolus Supine IMRT w/ Bolus Supine IMRT w/o Bolus Supine 3D The VMAT plans differ in hotspot when the bolus is removed: An increase of 4 percent in average for the hotspot of all the plans is seen, as well as a decrease of about 2% in the fall off dose in OARs %. The hotspot for IMRT plans is about the same for plans with and without bolus (1% difference, without bolus being larger). The lowest fall off dose percentage is given in the supine 3D conformal plan. VMAT plans have the lowest hotspot while 3D conformal has the best conformity index (Paddick s). IMRT without bolus had the best homogeneity index. 39

56 Table 4.7 Contralateral lung dose distribution comparison for 3D conformal, VMAT and IMRT plans in the supine position Min (%IDL) Mean (%IDL) Max (%IDL) Contralateral Lung V25 (%vol) V20 (%vol) V5 (%vol) V5% of prescription (cc) Plan Supine VMAT w/ Bolus Supine VMAT w/o Bolus Supine IMRT w/ Bolus Supine IMRT w/o Bolus Supine 3D For the contralateral lung, the 3D conformal plan gives lower doses overall than the VMAT and IMRT plans. This could be related to the geometry of the beams in the inverse planned treatments. The effect of removing the bolus can be seen for both inverse modulated plans with an increase of 2% in the V5% of the prescription as well as on the V5. Major changes can be seen at V5% changes by removing the bolus: The removal of the bolus results in an increase of 10%. Table 4.8 Ipsilateral lung dose distribution comparison for 3D conformal, VMAT and IMRT plans in the supine position Min (%IDL) Mean (%IDL) Max (%IDL) Ipsilateral Lung Vprescription (% Vol) V25 (%vol) V20 (%vol) V10 (%vol) V5 (%vol) V30% of Prescription Plan Supine VMAT w/ Bolus Supine VMAT w/o Bolus Supine IMRT w/ Bolus Supine IMRT w/o Bolus Supine 3D

57 For the ipsilateral lung, the maximum isodose line is the lowest for the 3D conformal plan. However, the Vprescription is the highest for the 3D conformal plan as well. The maximum isodose line is taken at 0.03 cc of the OAR volume. Between the inverse modulated plans, VMAT offers the lowest % volume irradiated at V25, V20, V10, V5 and V30% of prescription. The inverse modulated plans offer less volume irradiated at high isodose line values, with a tradeoff of more irradiated volume at low isodose line values. Table 4.9 Heart dose distribution comparison for 3D conformal, VMAT and IMRT plans in the supine position Min (%IDL) Mean (%IDL) Max (%IDL) V prescription (%vol) Heart Plan Supine VMAT w/ Bolus Supine VMAT w/o Bolus Supine IMRT w/ Bolus Supine IMRT w/o Bolus Supine 3D The V25 and V20 are comparable for all plans. V10, V5 and V2 are considerably higher for inverse modulated plans than for the 3D conformal plan. Between the two inverse planned treatments, the IMRT gives a lower dose to the heart overall. V25 (%vol) V20 (%vol) V10 (%vol) V5 (%vol) V2 (%vol) 41

58 Table 4.10 Contralateral breast dose distribution comparison for 3D conformal, VMAT and IMRT plans in the supine position Contralateral Breast Min Mean Max Plan (%IDL) (% IDL) (%IDL) Supine VMAT w/ Bolus Supine VMAT w/o Bolus Supine IMRT w/ Bolus Supine IMRT w/o Bolus Supine 3D The contralateral breast receives more dose in the inverse modulated plans (mean dose). The maximum isodose is considerably higher for the inverse modulated plans than for the 3D conformal plan as well. The following tables summarize the comparison between supine and prone 3D conformal plans, and the comparison between supine VMAT and IMRT (without bolus) plans. 42

59 Table 4.11 Summary of Dosimetric Comparison between Supine and Prone 3D Conformal Plans ROI Dosimetric Criteria Supine 3D Prone 3D PTV Min (%IDL) Mean (% IDL) Max (%IDL) Hospot (%IDL) Standard CI Paddick's CI HI Fall Off Dose in OARs (%) Contralateral Lung Ipsilateral Lung Min (%IDL) 0 0 Mean (% IDL) 0 1 Max (%IDL) 3 2 V25( %vol) V20 ( %vol) V5 ( %vol) V5% of prescription (cc) Min %IDL 0 0 Mean % IDL 11 7 Max %IDL V prescription (%vol) V25 (%vol) V20 ( %vol) V5 ( %vol) V30% of Prescription (%vol) Heart Min (%IDL) Mean (% IDL) 4 2 Max (%IDL) V prescription (%vol) V25 (%vol) V20 ( %vol) V10 ( %vol) V5 ( %vol) V2 ( %vol) Contralateral Breast Min (%IDL) Mean (% IDL) Max (%IDL)

60 Table 4.12 Summary of Dosimetric Comparison between Supine IMRT and VMAT (without bolus) Plans ROI Dosimetric Criteria Supine IMRT w/o Bolus Supine VMAT w/o Bolus PTV Min (%IDL) Mean (% IDL) Max (%IDL) Hospot (%IDL) Standard CI Paddick's CI HI Fall Off Dose in OARs (%) Contralateral Lung Ipsilateral Lung Min (%IDL) Mean (% IDL) Max (%IDL) V25( %vol) 0 0 V20 ( %vol) 0 0 V5 ( %vol) V5% of prescription (cc) Min %IDL 1 2 Mean % IDL Max %IDL V prescription (%vol) 0 0 V25 (%vol) V20 ( %vol) V5 ( %vol) V30% of Prescription (%vol) Heart Min (%IDL) 1 3 Mean (% IDL) Max (%IDL) V prescription (%vol) 0 0 V25 (%vol) 2 1 V20 ( %vol) 4 2 V10 ( %vol) V5 ( %vol) V2 ( %vol) Contralateral Breast Min (%IDL) 0 0 Mean (% IDL) 8 9 Max (%IDL)

61 4.2 Inverse Planning Skin Flash Surface Dose Delivery Measurements. The average thickness of the wettowel bolus on delivery was 8.5 mm with a CBCT-measured density of 0.6 g/cm3. OSLD and MOSFET measurements demonstrated good agreement with predicted doses from the TPS. Table 4.13, summarizes the total average deviations of measured values: OSLD average deviations were: 5.63%, -2.02%, and -2.48% for plans 1, 2, and 3, respectively; MOSFET average deviations were 2.092%, % and 0.864% respectively for plans 1, 2 and 3. Table Summary of Total Average Deviation of Measured Values from Planned Values using MOSFETs and OSLDs for Each of the Plans Irradiated Total Average Plans Total Average Deviation with MOSFET (%) Deviation with OSLD (%) 1 cm, 1g/cc Bolus cm, 0.65 g/cc Bolus. 1 PTV cm, 0.65 g/cc Bolus. 2 PTV Table Summary of Average Deviation per Site of Measured Values from Planned Values using MOSFETs and OSLDs for Each of the Plans Irradiated Average Deviation per site (%) Lateral Medial Tip Plan OSLD MOSFET OSLD MOSFET OSLD MOSFET 1 cm, 1g/cc Bolus cm, 0.65 g/cc Bolus. 1 PTV cm, 0.65 g/cc Bolus. 2 PTVs Table 4.14 summarizes the average deviation per site. Dosimeters placed at the medial and lateral portions of the breast showed the largest average deviations. The maximum recorded deviation from planned values was -8.6% for a dosimeter to bolus 45

62 perpendicular distance of 2.51 cm with the superflab bolus. Our results are in agreement with Zankar, Ito and Berg et al ( 5%). Table 4.15 OSLD Results from Irradiation at 0 and at 90 degrees ANGLE (degrees) OSLD1 (cgy) OSLD 2 (cgy) OSLD 3 (cgy) AVERAGE (cgy) Difference -2.4% The results for the two boluses vary by the same magnitude for both types of dosimeters used. The numerical difference between the values obtained with MOSFET versus OSLDs can be attributed to the physical characteristics of both detectors. Sensitive volume size, and directional dependence plays an important role as well. Three pairs of OSLDs were irradiated with 200 MU, 6X, 10x10 fs, 100 SAD at 0 and 90 degree gantry angles. The difference between the two readings was 2.4%. The manufacturer of MOSFETs guarantees an excellent directional response. Previous to taking the measurements, all dosimeters were calibrated and the readings for each type of dosimeter were found to be constant through this research. Figure 4.3. CBCTs illustrating bolus conformity to breast tissue a. Wet towel bolus. b. superflab bolus. 46

63 Evaluation of Inter-fractional Breast Volume Variations. The values for interfractional breast volume variation play an important role on determining the virtual bolus thickness for the implementation of skin flash in inverse planning. Table 4.16 Summary of Analyzed Patient Imaging in the Supine Position Patient Immobilization Device Number of MV Images Analyzed Number of CBCT Images 1S Breast Board S Breast Board 9 4 3S Vaclok S Breast Board S Vaclok 0 25 Total A total of 102 images, including MV and CBCT images were analyzed for the following parameters: breast thickness at the base and in the middle of the breast, breast length, breast angle and volume. Only the CBCT images were taken into consideration for breast volume measurements. Table 4.17 Supine Breast (Base) Thickness Results Patient Daily Average % Change from Simulation Average Value (cm) Standard Deviation (cm) 1S S S S S In table 4.17, it can be seen that the thickness at the base of the breast for the supine patients has an average daily fluctuation from -9.3% to 4.97% across all patients. The standard deviation for all patients stands well below 1 cm. 47

64 Table 4.18 Supine Breast (Middle) Thickness Results Patient Daily Average % Change from Average Value Simulation (cm) Standard Deviation (cm) 1S S S S S The average daily change from simulation CT ranges from -0.44% to -9.33%. The standard deviation ranges from 0.24 to 1.31 cm (Table 4.18). It was observed that the daily average percent change from the simulation dataset represented a larger value than the standard deviation. Regardless of these fluctuations, the standard deviation values stayed below 1 cm for all criteria, except for patient 1S and 3S (thickness of the middle aspect of the breast). The breast might change slightly in volume from the time of simulation to the first treatment for reasons unrelated to the radiation treatment. Once treatment starts, the standard deviation reveals a more stable trend. Michalski et. al., concluded that patients had a volume variation of no more than 5 mm, this value corresponded to an intrafractionary approach depending mostly on the respiratory movement of the chest rather than an integration of all possible factors that could change the target volume. Table 4.19 Supine Breast Length Results Daily Average % Change from Simulation Patient Average Value (cm) 1S S S S S Standard Deviation (cm) 48

65 Table 4.19, summarizes the supine breast length measurements. The breast length for supine patients had a daily average percent change ranging from 3.36% to -7.73% from the values measured in the simulation CT. The standard deviation for al patients ranges from 0.11 to 0.73, well below 1 cm. Table 4.20 Supine Breast Angle Results Daily Average Change from Simulation (degrees) Patient Average Value (degrees) 1S S S S S Standard Deviation (degrees) The daily average change from simulation in degrees ranges from 0.64 to degrees and the standard deviation for the images taken during the treatment course ranges from 4.5 to 0.03 degrees. The breast angle in the supine position does not vary as much as the other parameters. Table Supine Breast Volume Results Patient Average % Change Average Value Standard from Simulation (cc) Deviation (cc) 1S S S S S Table 4.21 summarizes the supine breast volume measurement results. The breast volume changes throughout the treatment course a range from 5.75% to -3.87% from the simulation CT. The standard deviation for this criteria ranges from to cc. 49

66 Even though the standard deviation is numerically large to other standard deviations from other measurements, it represents a 2.27% to 7.58% of the breast volume. Table 4.22 Volume Correlation Results in the Supine Position Criteria Correlation Value P Value Breast Thickness (base) and Volume <0.001 Breast Thickness (middle) and Volume 0.35 <0.05 Breast Length and Volume 0.65 <0.001 Breast Angle and Volume <0.001 The volume is positively correlated in most cases, except with the breast thickness in the base aspect of the breast and the breast angle. The correlations have a P-value of <0.001 and <0.05, which denotes a significant finding. In the supine position, there was a positive correlation between the breast volume and the breast thickness (middle) and length of the breast were found with P-values <0.001 and <0.05 respectively. The thicker the breast is in the middle and the longer it is, the more volume it will have. This is an expected result since these changes are situated in the middle and throughout the volume of the breast. Negative correlation values for the breast thickness (base) and breast angle were found with P-values < This means that during the treatment course, when more tissue is found at the base of the breast, the treated breast volume decreases. In the same way, as the breast angle becomes more positive (towards the gantry), the breast volume decreases. These findings can be related to the breast/fat tissue distribution during the radiation course, and of course, the unnatural angle of the breast tissue going against its natural angle. 4.3 Evaluation of Immobilization Devices Anthropomorphic Phantom Setup. Measurements taken with the anthropomorphic phantom depend mostly on the nature of the immobilization device, 50

67 taking away uncertainty from skin sagging and anatomical day-to-day variations in morphology. Table 4.23 Summary of Average Rotation for the Supine and Prone Setup Using an Anthropomorphic Phantom Prone Rando w/ board Supine Rando w/ VacLok C T # Average Rotation with Respect to the Horizontal Axis (degrees) Variation (degrees) Average Rotation with Respect to the Horizontal Axis (degrees) Variation (degrees) AVERAGE ROTATION (degrees) AVERAGE ROTATION (degrees) Standard Deviation (degrees) 2.82 Standard Deviation (degrees) 0.43 T-test P-value The average rotation for the prone CT setup was degrees with a standard deviation of 2.82 degrees. The average rotation for the supine setup was degrees with a standard deviation of 0.43 degrees. The values obtained between the two datasets is significantly different, with a T-test P-value of <0.05. These results were obtained with the anthropomorphic phantom, which has no movable tissue such as skin and fat. These results correspond to the deviations produced by the nature of the immobilization devices: vaclok for the supine setup and the prone breast board for the prone setup Patient Setup Evaluation. The patient setup evaluation included measurements of the axial rotation and a 3D displacement vector between the original setup (using tattoos) to the treatment position, in the supine and prone position. 51

68 Table 4.24 Summary of Pt. Axial Rotation and 3D Displacement Vector for the Supine Patient Setup Patient Pt Axial Rotation 3D vector Average (degrees) St Dev Average (mm) St Dev (mm) 1S S S S S In the supine position, the standard deviation for the patient axial deviation ranged from 0.24 degrees to a maximum of 2.25 patients in a group of 5 patients. The 3-D displacement vector ranged from 2.43 mm to 6.65 mm. Table 4.25 Summary of Pt. Axial Rotation and 3D Displacement Vector for the Prone Patient Setup Pt Axial Rotation 3D vector Patient Average (degrees) St Dev Average (mm) St Dev 1P P P P P In the prone position, the standard deviation for the patient axial rotation ranges from 1.70 to 3.62 degrees. The 3-D displacement vector ranges from 3.43 to mm. Table Summary of the Comparison between the Supine and Prone Positions Setup Patient Average Pt Axial Rotation (degrees) Average 3D vector (mm) Prone Supine Prone Supine 1P P P P P T-test P-value 4.01E-07 T-test P-value 2.64E-05 52

69 The values obtained for the prone datasets are larger than the values obtained for the supine datasets. The difference between these two positions can be confirmed with a T-test p-value of <0.001 for both criteria. These findings are in agreement with those of Cheng, et al., who found greater variations for the prone 3D displacement vector in the prone setup (13.9 mm) than in the supine setup (7.2 mm) for patients treated with WBI. However, according to Oh, et al., the thorax setup suffers itself from setup uncertainties, being the 3D displacement shift the second highest among other setups for other parts of the body. Table 4.27 Volume Correlation Results in the Prone Position Criteria Correlation Value P Value Breast Thickness (base) and Volume Breast Thickness (middle) and Volume Breast Length and Volume Breast Angle and Volume The strongest correlation values belong to those between the breast lengths, breast angle with the breast volume. The correlation value of the breast length and volume is 0.48 with a p-value <0.05, making it a significant correlation. As the breast length increases, the breast volume increases as well. The strongest correlation belongs to that of the breast angle and volume, with a value of 0.70 and a p-value of < The breast angle and the breast length in the prone position can be affected by the distribution of skin and fatty tissue on the board. Before treatment, portal imaging and CBCT imaging attempt to create a good registration especially around the chest wall and the breast itself. As skin and fatty tissue are being pulled into the board, there is more breast tissue variation and these changes affect the position of the chest wall with respect to the prone board and original simulation position. 53

70 Chapter 5 Conclusions Prone 3D conformal plans provide PTV coverage comparable to that of their supine counterparts. The difference between supine and prone 3D conformal breast plans is the dose given to the organs at risk. Prone 3D conformal plans received a significant 50% decrease to V20, V10, V5 and V30% for the ipsilateral lung in contrast to the supine plans. The heart also experienced a 10% decrease in maximum dose in the prone position, and V25, V20, V10, V5 and V2 had significantly lower values than their counterparts in the supine position. No differences were observed for the dose given to the contralateral lung. IMRT and VMAT breast plans in the supine position had comparable PTV coverage. Organs at risk had a decrease in volume irradiated with regions of high dose, but this shift lead to increases in the irradiated volume by low dose regions. The heart experienced a 10% decrease in maximum dose with inverse modulated plans when compared to the 3D conformal plan. However, V25, V20, V10, V5 and V2 showed higher values in the inverse modulated plans than in the 3D conformal plan. On the issue of skin flash, the use of virtual bolus in the TPS yielded correct dose delivery to this area regardless of the density of said bolus in inverse modulated plans. The largest dose fluctuations for surface dosimetry occurred near areas where the bolus failed to properly conform to the breast contour. It was found that the use of wet-towel bolus improved dose delivery accuracy compared to standard Superflab bolus as the areas of poor bolus conformity adversely affected dose delivery. The use of wet-towel bolus over Superflab bolus for inverse modulated WBI is recommended. 54

71 Target volume fluctuations can result from edema, weight gain/loss, side effects, and must be taken into consideration when expanding the dose fluence in the integration of skin flash. The values obtained during this research correspond to an interfractionary volume change rather to an intrafractionary volume analysis. The findings from this research work confirm that the 1-cm-thick bolus for inverse modulated plans is appropriate to implement skin flash throughout the treatment course. However, a 4-D simulation CT might be of use to help quantify changes due to respiratory movement in patients whose chest movement clearly exceeds 1 cm. The correlation between different parameters (thickness and length) measured in breast and the breast volume being irradiated demonstrated the importance of good patient setup techniques. Even though the changes in breast volume are affected by different factors, it is important to keep in mind that correct positioning as well as the use of the correct immobilization devices can make a difference in dose distribution and generation of an optimal treatment plan. For example, when the patient has her arms up during a treatment, the reproducibility of this position affects the tissue distribution at the base of the breast as well as its angle. In the supine position, it is very important that both the vaclok setup and the butterfly board retain the position of the arms at simulation and during the course of treatment. When looking at the anthropomorphic phantom setup and real patients setup, some uncertainty is related to the immobilization device itself, especially in the case of the prone setup. The most important aspect for the supine setup is arm positioning, which controls tissue at the base of the breast. The most important aspect for the prone setup is breast positioning (breast angle) which is affected by the roll of the patient s body. During prone setup, an axial marker for the breast could help with reproducibility as well 55

72 as a device that delimits the area of skin and fatty tissue around the breast to be treated. Good arm support and a shape-retaining immobilization device could decrease the level of rotation in prone setup patients during simulation and course of treatment. 56

73 References Breast cancer Facts and Figures World Cancer Research Fund International Berg, R., Gefter, J., Kimsey, F., Kirby-Harris, L., Phillips, R., & Mckay, J. (2012). VMAT In Vivo Dose Delivery Verification. International Journal of Radiation Oncology Biology Physics, 84(3S). CDC Breast Cancer Statistics. (2015, August 20). Cheng, K., & Wu, V. (2014). Comparison of the effectiveness of different immobilization systems in different body regions using daily megavoltage CT in helical tomotherapy. The British Journal of Radiology BJR, 87, Dumane, V., Hunt, M., Green, S., Lo, Y., & Bakst, R. (2014). Dosimetric Comparison of Volumetric Modulated Arc Therapy, Static Field Intensity Modulated Radiation Therapy, and 3D Conformal Planning for the Treatment of a Right-Sided Reconstructed Chest Wall and Regional Nodal Case. Journal of Radiotherapy, Fisher B, Anderson S, Bryant J, et al. Twenty-year follow-up of a randomized trial comparing total mastectomy, lumpectomy, and lumpectomy plus irradiation for the treatment of invasive breast cancer. N Engl J Med. Oct ;347(16): Formenti, S., Dewyngaert, J., Jozsef, G., & Goldberg, J. (2012). Prone vs Supine Positioning for Breast Cancer Radiotherapy. JAMA, 308(9), Formenti, S., et al. Seminars in Radiation Oncology, Howlader N, Noone AM, Krapcho M, Garshell J, Miller D, Altekruse SF, Kosary CL, Yu M, Ruhl J, Tatalovich Z,Mariotto A, Lewis DR, Chen HS, Feuer EJ, Cronin KA 57

74 (eds). SEER Cancer Statistics Review, , National Cancer Institute. Bethesda, MD, =sect_23_table.10.html, Ito, S., Parker, B., Levine, R., Sanders, M., Fontenot, J., Gibbons, J., & Hogstrom, K. (2011). Verification of Calculated Skin Doses in Postmastectomy Helical Tomotherapy. International Journal of Radiation Oncology Biology Physics, 81(2), Javedan, K., Zhang, G., Mueller, R., Harris, E., Berk, L., & Forster, K. (2009). Skin dose study of chest wall treatment with tomotherapy. Jpn J Radiol Japanese Journal of Radiology, 27, Kataria, T., Sharma, K., Subramani, V., Karrthick, K., & Bisht, S. (2012). Homogeneity Index: An objective tool for assessment of conformal radiation treatments. Journal of Medical Physics, 37(4), Krengli, M., Masini, L., Caltavuturo, T., Pisani, C., Apicella, G., Negri, E.,... Gambaro, G. (2013). Prone versus supine position for adjuvant breast radiotherapy: A prospective study in patients with pendulous breasts. Radiation Oncology, 8, 232. Michalski, A., Atyeo, J., Cox, J., & Rinks, M. (2012). Inter- and intra-fraction motion during radiation therapy to the whole breast in the supine position: A systematic review. Journal of Medical Imaging and Radiation Oncology, 56(5), Mitchell, J., Formenti, S., & Dewyngaert, J. (2010). Interfraction and Intrafraction Setup Variability for Prone Breast Radiation Therapy. International Journal of Radiation Oncology, Biology, Physics, 76(5),

75 Moody, A., Mayles, W., Bliss, J., A'hern, R., Owen, J., Regan, J.,... Yarnold, J. (1994). The influence of breast size on late radiation effects and association with radiotherapy dose inhomogeneity. Radiotherapy and Oncology, 33(2), Mulliez, Thomas, Bruno Speleers, Indira Madani, Werner De Gersem, Liv Veldeman, and Wilfried De Neve. "Whole Breast Radiotherapy in Prone and Supine Position: Is There a Place for Multi-beam IMRT?" Radiation Oncology 8.15 (2013): 151. National Cancer Institute. Surveillance, Epidemiology, and End Results ProgramTurning Cancer Data Into Discovery NanoDot Dosimeter Brochure. (2012). Retrieved November 13, 2015, from Nicolini, G., Vanetti, E., Clivio, A., Fogliata, A., & Cozzi, L. (2010). Pre-clinical evaluation of respiratory-gated delivery of volumetric modulated arc therapy with RapidArc. Physics in Medicine and Biology Phys. Med. Biol., 55(12), Oh, Y., Baek, J., Kim, O., & Kim, J. (2014). Assessment of setup uncertainties for various tumor sites when using daily CBCT for more than 2200 VMAT treatments. Journal of Applied Clinical Medical Physics, 15(2), Pasler, M., Lutterbach, J., Björnsgard, M., Reichmann, U., Bartelt, S., & Georg, D. (2015). VMAT techniques for lymph node-positive left sided breast cancer. Zeitschrift Für Medizinische Physik, 25(2), Popescu, C., Beckham, W., Patenaude, V., Olivotto, I., & Vlachaki, M. (2013). Simultaneous Couch and Gantry Dynamic Arc Rotation (C-G Darc) in the 59

76 Treatment of Breast Cancer With Accelerated Partial Breast Radiation Therapy (APBRT). Journal of Applied Clinical Medical Physics, 14(1). Popescu, C., Wai, E., Ansbacher, W., Salter, L., Olivotto, I., Beckham, W., & Otto, K. (2010). Volumetric Modulated Arc Therapy (VMAT) Improves Dosimetry and Reduces Treatment Time Compared to 9-field Conventional Intensity Modulated Radiotherapy (cimrt) for Locoregional Radiotherapy of Left-sided Breast Cancer. International Journal of Radiation Oncology Biology Physics, 76(1), Remeijer, P., Geerlof, E., Ploeger, L., Gilhuijs, K., Herk, M., & Lebesque, J. (2000). 3-D portal image analysis in clinical practice: An evaluation of 2-D and 3-D analysis techniques as applied to 30 prostate cancer patients. International Journal of Radiation Oncology, Biology, Physics, 46(5), Sankar, A., & Velmurugan, J. (2009). Different intensity extension methods and their impact on entrance dose in breast radiotherapy: A study. J Med Phys Journal of Medical Physics, 34(4), Surgent, R., Bhatnagar, A., Brandner, E., Deutsch, M., Sonnick, D., Gerszten, K., & Heron, D. (2006). Does breast size affect the scatter dose to the ipsilateral lung, heart, or contralateral breast in primary breast irradiation using intensity modulated radiation therapy (IMRT). International Journal of Radiation Oncology, Biology, Physics. Tsai, P., Lin, S., Lee, S., Yeh C., Huang, Y. (2012). The feasibility study of using multiple partial volumetric-modulated arcs therapy in early stage left-sided breast cancer patients. Journal of Applied Clinical Medical Physics, 13(5),

77 Wiant, D., Wentworth, S., Maurer, J., Vanderstraeten, C., Terrell, J., & Sintay, B. (2014). Surface Imaging Based Analysis of Intra-Fraction Motion for Breast Radiotherapy Patients. Journal of Applied Clinical Medical Physics, 15(6),

78 Appendix A Supine and Prone Dose Distribution and DVH Comparison for Individual Patients Patient 3P Figure A1. Axial slices comparing the dose distribution for two 3-D conformal treatment plans. a. Supine 3D conformal dose distribution with a hotspot of 4750 cgy. b. Prone 3D conformal dose distribution with a hotspot of 4681 cgy. Figure A2. DVH comparison of 3D conformal dose distributions in the supine (dashed line) and prone positions (solid line). 62

79 Patient 4P Figure A3. Axial slices comparing the dose distribution for two 3-D conformal treatment plans. a. Supine 3D conformal dose distribution with a hotspot of 5400 cgy. b. Prone 3D conformal dose distribution with a hotspot of 5400 cgy. Figure A4. DVH comparison of 3D conformal dose distributions in the supine (dashed line) and prone positions (solid line). 63

80 Patient 1S Figure A5. Axial slices comparing the dose distribution for two 3-D conformal treatment plans. a. Supine 3D conformal dose distribution with a hotspot of 5450 cgy. b. Prone 3D conformal dose distribution with a hotspot of 5500 cgy. Figure A6. DVH comparison of 3D conformal dose distributions in the supine (dashed line) and prone positions (solid line). 64

81 Patient 5P Figure A7. Axial slices comparing the dose distribution for two 3-D conformal treatment plans. a. Supine 3D conformal dose distribution with a hotspot of 5350 cgy. b. Prone 3D conformal dose distribution with a hotspot of 5300 cgy. Figure A8. DVH comparison of 3D conformal dose distributions in the supine (dashed line) and prone positions (solid line). 65

82 Patient 6P Figure A9. Axial slices comparing the dose distribution for two 3-D conformal treatment plans. a. Supine 3D conformal dose distribution with a hotspot of 5600 cgy. b. Prone 3D conformal dose distribution with a hotspot of 5200 cgy. Figure A10. DVH comparison of 3D conformal dose distributions in the supine (dashed line) and prone positions (solid line). 66

83 Appendix B 3D CRT/VMAT/IMRT Supine Dose Distribution and DVH Comparison for Individual Patients Patient 3P Figure B1. Axial slices comparing the dose distribution for two inverse-planned treatments with bolus and without bolus in the supine position. The plans were optimized with bolus and then this density override was later removed. a. 3D conformal plan with a hotspot of cgy. b. IMRT with bolus plan with a hotspot of 4650 cgy. c. VMAT with bolus plan with a hotspot of 4535 cgy. d. IMRT without bolus plan with a hotspot of 4600 cgy. e. VMAT without bolus plan with a hotspot of 4800 cgy. 67

84 Figure B2. DVH comparison of dose distributions: IMRT with bolus (thick dashed line), VMAT with bolus (thick solid line), and 3D conformal (thin dashed line). Figure B3. DVH comparison of dose distributions: IMRT without bolus (thick dashed line), VMAT without bolus (thick solid line), and 3D conformal (thin dashed line). 68

85 Patient 4P Figure B4. Axial slices comparing the dose distribution for two inverse-planned treatments with bolus and without bolus in the supine position. The plans were optimized with bolus and then this density override was later removed. a. 3D conformal plan with a hotspot of cgy. b. IMRT with bolus plan with a hotspot of 5600 cgy. c. VMAT with bolus plan with a hotspot of 5200 cgy. d. IMRT without bolus plan with a hotspot of 5630 cgy. e. VMAT without bolus plan with a hotspot of 5700 cgy. 69

86 Figure B5. DVH comparison of dose distributions: IMRT with bolus (thick dashed line), VMAT with bolus (thick solid line), and 3D conformal (thin dashed line). Figure B6. DVH comparison of dose distributions: IMRT without bolus (thick dashed line), VMAT without bolus (thick solid line), and 3D conformal (thin dashed line). 70

87 Patient 1S Figure B7. Axial slices comparing the dose distribution for two inverse-planned treatments with bolus and without bolus in the supine position. The plans were optimized with bolus and then this density override was later removed. a. 3D conformal plan with a hotspot of cgy. b. IMRT with bolus plan with a hotspot of 5570 cgy. c. VMAT with bolus plan with a hotspot of 5330 cgy. d. IMRT without bolus plan with a hotspot of 5580 cgy. e. VMAT without bolus plan with a hotspot of 5480 cgy. 71

88 Figure B8. DVH comparison of dose distributions: IMRT with bolus (thick dashed line), VMAT with bolus (thick solid line), and 3D conformal (thin dashed line). Figure B9. DVH comparison of dose distributions: IMRT without bolus (thick dashed line), VMAT without bolus (thick solid line), and 3D conformal (thin dashed line). 72

89 Patient 5P Figure B10. Axial slices comparing the dose distribution for two inverse-planned treatments with bolus and without bolus in the supine position. The plans were optimized with bolus and then this density override was later removed. a. 3D conformal plan with a hotspot of cgy. b. IMRT with bolus plan with a hotspot of 5400 cgy. c. VMAT with bolus plan with a hotspot of 5450 cgy. d. IMRT without bolus plan with a hotspot of 5450 cgy. d. VMAT without bolus plan with a hotspot of 5500 cgy. 73

90 Figure B11. DVH comparison of dose distributions: IMRT with bolus (thick dashed line), VMAT with bolus (thick solid line), and 3D conformal (thin dashed line). Figure B12. DVH comparison of dose distributions: IMRT without bolus (thick dashed line), VMAT without bolus (thick solid line), and 3D conformal (thin dashed line). 74

91 Patient 6S Figure B13. Axial slices comparing the dose distribution for two inverse-planned treatments with bolus and without bolus in the supine position. The plans were optimized with bolus and then this density override was later removed. a. 3D conformal plan with a hotspot of 5400 cgy. b. IMRT without bolus plan with a hotspot of 5650 cgy. c. VMAT with bolus plan without a hotspot of 5650 cgy. d. IMRT with bolus plan with a hotspot of 5450 cgy. e. VMAT with bolus plan with a hotspot of 5550 cgy. 75

92 Figure B14. DVH comparison of dose distributions: IMRT with bolus (thick dashed line), VMAT with bolus (thick solid line), and 3D conformal (thin dashed line). Figure B15. DVH comparison of dose distributions: IMRT without bolus (thick dashed line), VMAT without bolus (thick solid line), and 3D conformal (thin dashed line). 76

93 Patient 7S Figure B17. DVH for 3D conformal plan Figure B18. DVH comparison of dose distributions: IMRT without bolus (thick dashed line), and VMAT without bolus (thick solid line). 77

94 Figure B19. DVH comparison of dose distributions: IMRT with bolus (thick dashed line), and VMAT with bolus (thick solid line) 78

95 Figure B20. Axial slices comparing the dose distribution for two inverse-planned treatments with bolus and without bolus in the supine position. The plans were optimized with bolus and then this density override was later removed. a. 3D conformal plan with a hotspot of 5650 cgy. b. VMAT without bolus plan with a hotspot of 5300 cgy. c. IMRT with bolus plan without a hotspot of 5750 cgy. d. VMAT with bolus plan with a hotspot of 5350 cgy. e. IMRT with bolus plan with a hotspot of 5600 cgy. 79

96 Patient 8S Figure B21. Axial slices comparing the dose distribution for two inverse-planned treatments with bolus and without bolus in the supine position. The plans were optimized with bolus and then this density override was later removed. a. 3D conformal plan with a hotspot of 5400 cgy. b. VMAT without bolus plan with a hotspot of 5500 cgy. c. IMRT with bolus plan without a hotspot of 5500 cgy. d. VMAT with bolus plan with a hotspot of 5450 cgy. e. IMRT with bolus plan with a hotspot of 5450 cgy. 80

97 Figure B22. DVH comparison of dose distributions: IMRT with bolus (thick dashed line), VMAT with bolus (thick solid line), and 3D conformal (thin dashed line). Figure B23. DVH comparison of dose distributions: IMRT without bolus (thick dashed line), VMAT without bolus (thick solid line), and 3D conformal (thin dashed line). 81

98 Patient 9S Figure B24. Axial slices comparing the dose distribution for two inverse-planned treatments with bolus and without bolus in the supine position. The plans were optimized with bolus and then this density override was later removed. a. 3D conformal plan with a hotspot of 4650 cgy. b. VMAT without bolus plan with a hotspot of 4900 cgy. c. VMAT with bolus plan without a hotspot of 4750 cgy. d. IMRT with bolus plan with a hotspot of 4600 cgy. e. IMRT with bolus plan with a hotspot of 4680 cgy. 82

99 Figure B25. DVH comparison of dose distributions: IMRT with bolus (thick dashed line), VMAT with bolus (thick solid line), and 3D conformal (thin dashed line). Figure B26. DVH comparison of dose distributions: IMRT without bolus (thick dashed line), VMAT without bolus (thick solid line), and 3D conformal (thin dashed line). 83