## Program Description

### Overview

The theme for this year's conference was Certainty in an Uncertain World.

The 4th annual conference focused on advances in IMPT, robustness in treatment planning, quantifying and managing the uncertainty in range and tumor motion, incorporating 3-D imaging with daily delivery, and biological effects. The clinical applications centered on the traditionally recognized clinical indications, including breast, lung, head and neck, gastrointestinal tract and pediatric cancers. Hypo-fractionation and combined modality approaches were also highlighted. The event included updates on clinical trials and lively discussions around future protocols, grant opportunities, multi-institutional collaborative efforts, and a recognition of the value of large, multi-center registry data.

### Target Audience

Healthcare professionals who treat cancer patients using radiation therapy/particle therapy and specifically:

• Medical Physicists

• Dosimetrists

• Residents

### Particle Therapy Cooperative Group North America (PTCOG-NA) 2017 Committees

#### PTCOG-NA Conference Planning Committee Members

Eugen B. Hug, MD, PTCOG-NA President

Medical Director, Procure Proton Therapy Center, Somerset, NJ

Hesham E. Gayar, MD, PTCOG-NA Vice President

Department of Radiation Oncology, Karmanos Cancer Institute, Flint, MI

William F. Hartsell, MD (Local host), PTCOG-NA Executive Committee Member

Medical Director, Northwestern Medicine Chicago Proton Center, Warrenville, IL

Anita Mahajan, MD, PTCOG-NA Secretary

Professor, Department of Radiation Oncology, Mayo Clinic, Rochester, MN

Carl Rossi, Jr., MD, PTCOG-NA Treasurer

Medical Director, Scripps Proton Therapy Center, San Diego, CA

## Oral Abstracts Abstract 1: Breast

### Contour-based Lung Dose Prediction for Breast Proton Therapy

#### Chuan Zeng, Kevin Sine, Dennis Mah

##### ProCure Proton Therapy Center – NJ

PURPOSE: This study evaluates the feasibility of lung dose prediction based on target contour and patient anatomy for breast patients.

METHODS: Sixteen randomly selected patients were included in the cohort, who were treated to 50.4 – 66.4 Gy(RBE) to the left (14) or right (two) breast with uniform scanning (15) or pencil beam scanning (one). Anterior-oblique beams were used for each patient. The prescription doses were all scaled to 50.4 Gy(RBE) for the current analysis. Isotropic expansions of the planning target volume of various margins m were retrospectively generated and compared with isodose volumes in the ipsilateral lung. The fractional volume V of each expansion contour within the ipsilateral lung was compared with dose-volume data of clinical plans to establish the relationship between the margin m and dose D for the ipsilateral lung such that VD=V(m). This relationship enables prediction of dose-volume VD from V(m), which could be derived from contours before any plan is generated, providing a goal of plan quality. Lung V20 Gy(RBE) was considered for this pilot study, while the results could be generalized to other dose levels.

RESULTS: The actual V20 Gy(RBE) ranged from 6% to 23%. No statistically significant difference in V20 Gy(RBE) was found between breast irradiation and chest wall irradiation (p=0.8) or between left side and right side treatment (p=0.9). It was found that V(1.1 cm) predicted V20 Gy(RBE) to within 5% root-mean-square deviation.

CONCLUSIONS: A contour-based model was established to predict dose to ipsilateral lung in breast treatment. Clinically relevant accuracy was demonstrated. This model facilitates dose prediction before treatment planning. It could serve as a guide towards realistic clinical goals in the planning stage.

## Abstract 2: Breast

### Incorporating Breast Cancer Biologically Modeled Dose Constraints to Reduce Brachial Plexopathy Risk of Proton Therapy

#### Stephanie Wick, Krishan Jethwa, Thomas Whitaker, Kimberly Corbin, Elizabeth Yan, Sean Park, Nicholas Remmes, Chris Beltran, Robert Foote

##### Mayo Clinic

PURPOSE: Brachial plexopathy is a rare but potentially debilitating late side effect of breast cancer regional nodal irradiation (RNI) that has been correlated with dose to the brachial plexus (BP). The radiobiologic effective dose (RBE) rises with increasing linear energy transfer (LET) towards the end of proton tracks. However, there is no consensus on whether or how RBE heterogeneity should be accounted for in treatment planning. We instituted a biological dose (BD) constraint for the BP in the context of MC1631, a randomized trial of 15 vs 25 fraction intensity modulated proton therapy (IMPT) after mastectomy in patients requiring RNI. Our purpose was to determine the dosimetric consequences of the BD constraint. In addition, we compare target coverage and normal tissue doses using: 1) commercially available Eclipse treatment planning system (TPS, RBE = 1.1), 2) in-house GPU-based Monte Carlo physical dose (MCPD) simulation (RBE=1.1) and 3) in-house Monte Carlo biologic dose (MCBD) simulation which assumes a linear relationship between RBE and LET (RBE = BD).

METHODS: IMPT plans of 13 patients treated prior to the implementation of a BP BD constraint (cohort 1) were compared with IMPT plans of 38 patients treated after the implementation of a BP BD constraint on MC1631 (cohort 2). The BP BD constraint was D0.01cc < 120% of prescription. Eclipse, MCPD, and MCBD dose-volume parameters were evaluated.

RESULTS: Prior to implementation of a BP BD constraint, the mean Eclipse TPS D0.01cc of the BP was 107.3% and the MCBD estimate was 130.8% (i.e. 65.5 Gy [RBE=BD] in 25 fractions for a 50 Gy [RBE=1.1] prescription), compared with 101.7% and 116.0% after the implementation of the BP BD constraint, respectively. Implementation of the BP BD constraint did not detrimentally impact CTV coverage (table). IMN coverage and heart doses were greater in the MCBD plans compared with the Eclipse TPS and MCPD plans.

CONCLUSION: Institution of a BP BD constraint resulted in more conservative brachial plexopathy risk estimates without compromising target coverage. Routine MCBD plan evaluation provides valuable information to physicians making clinical judgments regarding relative priority of target coverage vs normal tissue sparing.

## Abstract 3: Breast

### Dosimetric Comparison of Left-Sided Comprehensive Breast and Nodal Irradiation with Photons at Deep Inspiration Breath Hold vs. Pencil Beam Scanning Protons at Free Breathing

#### Rajal Patel, Michelle Gentile, Fatima Pathiawala, Estelle Batin, Irene Helenoswki, Eric Donnelly, John Hayes, Jonathan Strauss

##### Northwestern University

PURPOSE: Patients with left-sided breast cancer have several effective options to decrease heart and lung dose when receiving radiotherapy. For patients receiving comprehensive breast and nodal irradiation, a comparison was performed of pencil beam scanning proton radiotherapy vs. photon irradiation at deep inspiration breath hold (DIBH) with regard to reducing dose to organs at risk (OAR).

METHODS: Five consecutive patients diagnosed with left-sided, lymph node positive breast cancer with available free-breathing and DIBH CT simulation imaging were included. Contouring was performed on both scans according to the RADCOMP atlas used in proton planning. Additionally, each DIBH scan was contoured according to the NSABP-B51 protocol for a clinically meaningful photon comparison. Radiation planning was performed to maximize coverage (≥95% target coverage) to 45 GyRBE while abiding by institutional OAR constraints for proton planning. Photon plans were all created using a partially wide tangent technique with 3D conformal radiotherapy. NASBP-B51 protocol standards were utilized for photon planning. RCA dose was not considered in planning optimization. Primary endpoints included mean heart, mean left anterior descending (LAD) artery, mean right coronary artery (RCA), and mean ipsilateral lung doses. Statistical differences between groups were calculated using the Wilcoxon Rank Sum Test.

RESULTS: Median mean doses for each organ at risk are listed below. Protons yielded significantly lower mean heart, LAD, and ipsilateral lung doses when compared to photon plans (p≤0.02). Both photon plans had significantly lower RCA dose as compared to protons (p<0.03).

CONCLUSIONS: In comprehensive breast and nodal radiotherapy, protons yielded lower mean heart and LAD doses while DIBH photons yielded lower RCA dose. However, the absolute mean dose for these endpoints was quite low with both treatment modalities, and the clinical significance of these small differences remains unknown. Mean ipsilateral lung dose was more than halved with protons, which may decrease the incidence of radiation-associated secondary malignancies for younger patients. Radiobiologic modeling of the effect of reduction in lung dose and cost-effectiveness comparisons may be helpful in optimal patient selection.

## Abstract 4: Breast

### Excellent Acute Toxicity Outcomes with Proton Therapy for Partial Breast Irradiation in Early Stage Breast Cancer: Initial Results of a Multi-institutional Phase II Trial

#### J. Isabelle Choi, Andrew Chang

##### Scripps Proton Therapy Center, ProCure Proton Therapy Center – Oklahoma City

BACKGROUND AND PURPOSE: Partial breast irradiation (PBI) with proton therapy after lumpectomy for early stage invasive breast cancer is an area of active investigation. To date, multiple single-institutional studies have reported conflicting results on the acute toxicity of PBI. This prospective phase II trial investigates the feasibility, safety, and efficacy of delivering PBI with proton therapy in a multi-institutional setting.

METHODS: Patients over the age of 50 years with ER positive nonlobular invasive breast cancer or ductal carcinoma in situ <=3 cm in size who had undergone lumpectomy with at least 2 mm negative surgical margins were treated with proton therapy to a dose of 40 Gy delivered over 10 daily fractions. In this initial analysis, we assess early toxicity and treatment efficacy of proton PBI. Patients were followed 4 weeks post-treatment and annually thereafter, along with annual mammograms. Patient-reported quality of life and physician-reported cosmesis assessments including photographs were obtained at 1 and 3 years post-treatment.

RESULTS: Of 40 enrolled patients, 38 were evaluable. At a median follow-up of 17.8 months (range 2-36 months), all patients had overall breast cosmesis that was scored “good” or “excellent.” Of 6 grade 2 acute adverse events that occurred, only 1 was radiation dermatitis, with others including lymphedema, hot flashes, and fatigue. One grade 3 acute toxicity occurred 3 weeks after radiation completion in the form of vascular disease requiring stent placement, highly unlikely to be attributable to radiation effects. Patient-reported quality of life outcomes were recorded using the standardized Breast Cancer Treatment Outcome Scale (BCTOS) scored from 1-4 (1: none; 2: mild; 3: moderate; 4: large). Patients assigned a score of 4 for change in nipple appearance (n=2), breast shape (n=2), and scar tissue formation (n=2). To date, local, locoregional, and distant disease control are 100%, although one patient has developed a new hormone receptor negative invasive ductal carcinoma of the contralateral breast.

CONCLUSION: Proton PBI provides excellent early cancer control with acceptable cosmetic outcomes and minimal adverse effects as per patient- and physician-reported assessments. On continued follow-up, late toxicity and cosmesis, as well as long-term disease control outcomes, will be assessed.

## Abstract 5: Breast

### The Evolution of PBS Proton Therapy Whole Breast Treatment Planning

#### Samantha Hedrick, Ben Robison, Mark Blakey, Niek Schreuder, Jackson Renegar, Mark Artz, Allen Meek

##### Provision Center for Proton Therapy

In the era of double scattering proton therapy, whole breast irradiation was not indicated due to high skin dose. However, with the advent of pencil beam scanning (PBS) proton therapy, whole breast irradiation is performed with good results. We have delivered whole breast irradiation with PBS for over 3 years, treating 125 patients. During that time, our treatment planning process has evolved based on in-house data to improve heart, lung, rib, and skin sparing.

Typical lung constraints for whole breast irradiation are based on RTOG 1005. After 1 year, we evaluated a sample of treated patients and created reduced in-house lung constraints. We have continuously evaluated skin cosmesis and evolved our planning technique to improve skin sparing. Based on emerging data of heart and left anterior descending artery (LAD) radiosensitivity, we decreased our heart mean dose and LAD maximum dose constraints. Finally, we evaluated rib dose for our patients and adjusted our planning technique to improve sparing.

According to RTOG constraints, PTV eval coverage of V95%>95% is met for all patients. Our ipsilateral lung constraints for breast only, breast and nodes without internal mammary (IM), and with IM are V20CGE less than 10%, 15%, and 20%, respectively. To improve skin sparing, we increased the minimum depth at which the planning system places a spot, and we have continued to increase this depth, based on clinical results, to >1.5cm from patient surface. Heart mean dose constraint is 1CGE and LAD max dose is D0.3cc<3CGE. The heart mean dose is achievable in all whole breast patients, even those with IM involvement. Finally, the rib constraint is D1cc<50CGE and utilizes a dose gradient at the end of the beam to account for the LET distal end effect.

Continuing evaluation of our whole breast patients has led to improved OAR sparing and consistent plan quality. By reducing OAR constraints, we have ensured that every patient is treated with the best possible plan and that whole breast irradiation utilizing PBS proton therapy is feasible and provides improved heart and lung sparing compared to conventional modalities.

Abstract 6 is a Poster Abstract

## Abstract 7: Breast

### Feasibility of Proton Treatments of Breast Cancer Patients with Expanders under Uniform Scanning

#### Dennis Mah, Gabriely Del Rosario, Chin-Cheng Chen, Oren Cahlon, Jana Fox, Chuan Zeng

##### ProCure Proton Therapy Center – New Jersey, Memorial Sloan Kettering Cancer Center, Montefiore Medical Center

INTRODUCTION: Breast cancer patients treated with protons may have lower lung and heart doses than photon treatments. Proton treatments of breast cancer patients with tissue expanders have only been reported using pencil beam scanning (PBS). Two approaches have been developed. Either the high Z magnet in the expander is avoided by junctioning fields around it or treating directly through. We investigated treating through the magnet using uniform scanning (US).

METHOD: Three breast cancer patients with expanders were evaluated. A template of the expander was generated based upon the manufacturer's specifications. The transmission of the uniform scanning beam through the implant was verified by measurement. Two sets of plans were generated to a dose of 50.4 Gy (RBE). Clinical plans using PBS to junction two fields around the magnet were compared to en face US plans treating through the magnet. For the latter, compensators were generated with a smear of 1 cm to account for possible displacements of the magnet relative to bony anatomy.

RESULTS: Both sets of plans were found to be clinically acceptable. PTV-skin V95 was >95% for both sets of plans. Dosimetric parameters were comparable for a given patient. For instance, dose to ipsilateral lung V20= [19.5; 18.5; 17.2 Gy (RBE)] for PBS and V20 = [20.1; 19.3; 16.3 Gy (RBE)] for US. Mean heart doses (MHD) were very well correlated. MDH was [1.0, 0.43, 2.6 Gy (RBE)] for PBS and [1.03, 0.43, and 2.32 Gy (RBE)]. The skin dose for the uniform scanning plans was higher with US since US has poorer proximal conformality.

CONCLUSIONS: Treatment of breast cancer patients through the magnet of tissue expanders with uniform scanning is feasible. The US distributions and DVHs are comparable to PBS. This study suggests that breast expander patients could be treated in proton facilities where PBS is unavailable.

## Abstract 8: Breast

### Does Deep Inspiration Breath Hold (DIBH) Improve Cardiac Dose for Left-sided Breast Cancer Treated with Double Scattered Proton Therapy?

#### Tamara Smith, Xiaoying Liang, Michael Rutenberg, Nancy Mendenhall, Julie Bradley

##### Ohio State University, University of Florida

PURPOSE: Deep inspiration breath hold (DIBH) can decrease cardiac dose in the treatment of left-sided breast cancer with photons. Proton therapy results in low cardiac dose, even in the setting of comprehensive regional nodal irradiation. The purpose of this study is to investigate the role of DIBH in double scattered proton therapy.

METHODS: Five consecutive patients with left-sided breast cancer underwent DIBH and free-breathing (FB) scans as part of routine CT simulation, with contours delineated on both image sets. Heart was delineated along the pericardium and the ventricle contour consisted of left and right ventricles. All patients were treated with double scattered proton therapy to the left breast (n=3) or chestwall with reconstruction (n=2) and regional lymphatics including internal mammary nodes with FB to a total dose of 50 Gy(RBE) in 25 daily fractions. Each patient had a comparison plan designed on the DIBH CT simulation scan. Dose-volume parameters for target coverage, heart, and lung were assessed for DIBH and FB.

RESULTS: Full target coverage was achieved for all patients on both the DIBH and FB scans (defined as 95% of the prescription dose covering ≥ 95% volume) for the breast/chestwall and the regional nodes. Median mean heart dose was 0.2Gy for DIBH (range, 0-1.6Gy) vs. 0.4Gy for FB (range, 0.3-1.1Gy). Differences in heart V5, heart V25, ventricles V5, and ventricles V25 were minimal between FB and DIBH plans. The primary benefit with DIBH was reduction in maximum point dose to the heart. DIBH decreased dose to the LAD in 2 cases and increased dose in 2 cases. Left lung V5 and V20 were comparable between DIBH and FB.

CONCLUSIONS: In most cases, there was minimal difference in dose to the heart and its substructures and left lung between DIBH and FB double scattered proton plans in treatment of the breast/chestwall and regional lymphatics for women with left-sided breast cancer. Maximum point dose to the heart was decreased in most cases with the use of DIBH, but this may be insufficient to justify the routine use of DIBH, which increases the complexity of treatment delivery and burdens the patient.

## Abstract 9: Genitourinary

### Misalignment Sensitivity of Matched PBS Fields Treating Prostate and Pelvic Lymph Nodes

#### Steve Laub, Mark Pankuch

##### Northwestern Medicine Chicago Proton Center

PURPOSE: Evaluate CTV coverage in the presence of isocentric misalignment of two matched pencil beam scanning (PBS) fields used to treat the prostate and pelvic lymph nodes (PLN).

METHODS: Ten clinical treatment plans that treated the prostate and PLN with two matched, lateral PBS fields were evaluated. Each field was optimized to deliver 50% of the prescription dose to the prostate and iliac nodal region, while delivering full dose to that field's proximal lymph node chain. The isocenter of one of the lateral fields was shifted by 0.3, 0.5, and 1.0 cm in the superior, inferior, anterior, and posterior directions. Lateral misalignments were not considered as these will create minimal dosimetric differences. These shifts simulate beam alignment to the prostate when its position has changed between the first and second field. The Prostate CTV V95% and PLN CTV V95 were evaluated in each scenario. CTV V95% > 95% was used as criteria for clinical acceptability.

RESULTS: The PLN CTV was most sensitive to superior/inferior misalignments: 60% of tested plans failed at 0.3 cm superior, 40% failed at 0.3 cm inferior. There were no failures for 0.3 cm misalignments in the anterior/posterior direction. For 0.5 cm misalignments, 100% failed in the superior and inferior directions, 20% failed in both the anterior and posterior directions. For 1.0 cm misalignments, 100% failed in the superior and inferior directions, 90% failed in the anterior and posterior directions. When evaluating the prostate CTV: 20% failed for the 1.0 cm shift superior, and 30% failed for the 1.0 cm shift inferior. There were no failures for smaller shifts.

CONCLUSION: The PLN CTV is more sensitive to isocenter misalignments than the prostate CTV. PLN CTV coverage may degrade at offsets as small as 0.3 cm, whereas decreased prostate CTV coverage was not observed for any shift less than 1.0 cm. If the prostate position shifts between fields, the dosimetric match line should be preserved, rather than aligning the second field to the prostate, as long as the prostate has not shifted more than 0.5 cm between fields.

Table 1

Failure rates of pelvic lymph nodes (PLN) CTV and prostate-only CTV for isocenter misalignments induced in 10 clinical treatment plans

## Abstract 11: Physics and Biology

### Clinical Evaluation of a Trolley Patient Transport System for Proton Therapy

#### Ulrich Langner, Charles Simone, Mark Mirsha, Katja Langen, Jamie Lepage, Megan Sank

##### University of Maryland

PURPOSE: The Maryland Proton Treatment Center became the first center to implement clinical use of a novel patient throughput trolley transport system (Qfix, Avondale, PA). We aimed to investigating the effectiveness of the trolley transport system, and we hypothesized that its use would allow for quicker in-room patient times, thus maximizing clinic efficiency.

METHOD: Patients were set up in full immobilization with laser alignment outside of the treatment room on the trolley system. Patients were then transported into the treatment room on the trolley and transferred to the treatment couch with an air bearing system. After treatment, patients were transferred back to the trolley and transported out of the room. Time was recorded from the moment the patients entered the treatment room until they exited. This time was compared to the time for patients treated for the same disease sites prior to the implementation of the trolley system.

RESULTS: For breast cancer patients (n=195 fractions – no trolley and n=94 with trolley), the median fractional time a patient spent inside the room was 26.42 +/- 12.48 minutes for patients walking in and out, and 25.63 +/- 10.70 minutes for patients using the trolley system (p = 0.21). For prostate cancer patients (n=822 fractions – no trolley and n=264 with trolley) the median fractional time a patient spent inside the room was 16.63 +/- 10.08 minutes for patients walking in and out and 15.98 +/- 10.10 minutes for patients using the trolley system (p = 0.18).

CONCLUSION: In the first-ever report of a novel trolley transport system designed to quicken patient throughout, we failed to find a meaningful reduction in total in-room time for patients with breast and prostate cancers. Overall patient set-up and treatment times actually increased when using the trolley system since extra patient and therapist time was needed for setup outside of the treatment room. We will analyze which subsets of patients may benefit most from the trolley system among breast and prostate patients (i.e., patients who are elderly, have mobility difficulties, are difficult to set up anatomically, etc.) and expand the use of this system to other disease sites in order to more definitively determine its clinical potential for proton throughput efficiency.

## Abstract 12: Lung and Mediastinum

### Proton Therapy for Thymic Malignancies: Multi-Institutional Patterns-of-Care and Early Clinical Outcomes from the Proton Collaborative Group Registry

#### Catherine Mercado, Bradford Hoppe, Randal Henderson, Henry Tsai, Carlos Vargas, George Laramore, Shahed Badiyan, William Hartsell

##### University of Florida, ProCure Proton Therapy Center – New Jersey, Mayo Clinic – Scottsdale, University of Washington, University of Maryland, Northwestern Medicine Chicago Proton Center

OBJECTIVE: Thymic malignancies (TM) are rare tumors with long-term survivorship, causing concerns for radiotherapy-related late side effects. Proton therapy (PT) reduces the radiation dose to organs at risk, potentially improving disease control and decreasing long-term toxicity in the management of TM. This study reports on the patterns-of-care and early clinical outcomes of patients treated with PT for thymoma and thymic carcinoma.

METHODS: From January 2008 through March 2017, 30 patients with TM were treated on the institutional review board-approved outcomes tracking protocol at our institution or the PCG registry with post-operative or definitive PT. Clinical outcomes, pathology, treatment dose, acute toxicities, and follow-up information were analyzed.

RESULTS: Twenty-one patients were treated for a thymoma with a median age of 52.1 years (range, 23-72). Median RT dose was 54 CGE (range, 45-70). PT was delivered postoperatively in 86% of patients, while definitive PT +/- chemotherapy was delivered to the remaining 14%. Twenty-four percent of patients received adjuvant chemotherapy. Nine patients were treated for a thymic carcinoma. Median age was 65.5 years (range, 38-88). Median RT dose was 60 CGE (range, 42-70). Postoperative PT was delivered to 67% of patients; 33% underwent definitive PT. Thirty-three percent of patients received chemotherapy. Median follow-up for the entire cohort was 22 months (range, 3-60 months). Two thymoma and one thymic carcinoma patient had a local recurrence. Median time from completion of PT to local recurrence was 13 months (range, 2-26). There have been three deaths, including one with locally recurrent thymic carcinoma, one due to metastatic disease, and one due to intercurrent disease. No patient experienced grade ≥3 toxicities after PT.

CONCLUSION: Adjuvant and definitive PT is being used in the treatment of thymic malignancies. Early results demonstrate an acceptable rate of recurrence with a tolerable toxicity profile. Longer follow-up and a larger patient cohort are needed to confirm these findings.

## Abstract 13: Lung and Mediastinum

### Evaluation of Proton Therapy Analytic Algorithm Distal Dose Calculation in Static Lung Phantom

#### Daniela Branco, Paige Taylor, Dragan Mirkovic, Xiaodong Zhang, Pablo Yepes, Narayan Sahoo, Stephen Kry, David Followill

##### MD Anderson Cancer Center

PURPOSE: The Imaging and Radiation Oncology Core (IROC) anthropomorphic moving proton lung phantom has shown a discrepancy between planning systems' analytic dose calculations and measurements in the target and distal region of the treatment fields. This study investigates the accuracy of proton therapy treatment planning system's (TPS) pencil beam algorithm dose calculations in lung using a static geometrical lung phantom.

METHODS: A geometrical lung phantom was designed to create a simple configuration to mimic a diseased human lung, decoupling the motion uncertainties introduced in the typical proton lung phantom. Solid water and balsa wood, with tissue-equivalent CT and stopping power values, were used to mimic human tissue and lung, respectively. Figure 1 shows the phantom designed for this project, with a 3 cm diameter cylindrical target. Radiochromic film can be inserted into the center of the phantom at a 5° angle with respect to both the left and the anterior wall, in order to try to minimize film quenching and streaming effects. A CT simulation was used to create a pencil beam scanning treatment plan (6 Gy[RBE] prescription) using the Eclipse TPS with an anterior and a left lateral field. The phantom was irradiated 2 separate times. The measured film dose distributions were compared to the clinical TPS pencil beam algorithm dose calculations and Monte Carlo (MC).

RESULTS: Profiles in the center of the target were analyzed, and both left-right and anterior-posterior profiles showed general shape agreement between both sets of calculations and the film measurements. The film showed slight under-dosing in the shoulder regions in the anterior-posterior profiles when compared to TPS and MC, which was a similar pattern observed in the full IROC lung phantom. Additionally, the left-right film dose distribution fell short in comparison to both clinical calculations.

CONCLUSION: The plan delivered showed an under-dosing around the shoulder regions in the anterior-posterior profiles and an overestimation of the TPS and MC range for the left proton field.

## Abstract 14: Lung and Mediastinum

### Machine Learning Analysis for Predicting Survival in Stage III Non-small Cell Lung Cancer Patients Receiving Definitive Chemotherapy and Proton Radiation Therapy

#### Dan Kunaprayoon, Hao Zhang, Gary Larson, Henry Tsai, George Laramore, Pranshu Mohindra, Shahed Badiyan, Carlos Vargas, William Hartsell, Charles Simone

##### University of Maryland School of Medicine, ProCure Proton Therapy Center- Oklahoma City and New Jersey, SCCA Proton Therapy Center, Mayo Clinic – Scottsdale, Northwestern Medicine Chicago Proton Center

OBJECTIVES: To evaluate stage III non-small cell lung cancer (NSCLC) patients undergoing definitive treatment with chemotherapy and proton radiation therapy using a machine-learning analytic approach to identify factors predictive for survival.

METHODS: 179 consecutive stage III NSCLC patients receiving chemoradiation at six proton centers enrolled on PCG registry REG001-09 from 2010 to 2017 were retrospectively studied. Predictive features included patient demographics (age, gender, race, smoking history, ECOG performance status), tumor parameters (histology, primary tumor location, clinical and pathologic stage), and treatment parameters (target site, dose, fractionation, radiation modality, curative intent, treatment interruptions, chemotherapy agents and timing, prior surgery, prior treatment history). Multiple predictive models were constructed based on different machine learning models (Naive Bayes [NB], logistic regression [LR], neural network [NN], support vector machine [SVM], random forest [RF]) to evaluate overall survival (OS) predictability. The number of cases and percentage of cases predicted correctly were collected and the area under the receiver operating characteristic (ROC) curve (AUC), sensitivity, and specificity were determined.

RESULTS: The five models exhibited similar predictive capabilities. For 6-month OS, the maximum AUC, sensitivity, and specificity of the predictive models were 0.65, 0.76, and 0.80. For 1-year OS, the maximum AUC, sensitivity, and specificity were 0.68, 0.67, and 0.69.

CONCLUSION: This is the largest survival predictive analysis for lung cancer patients treated with proton therapy performed to date. We demonstrated that a machine-learning analytic approach using features of patient demographics, tumor parameters, and treatment characteristics shows promise for predicting overall survival in stage III lung cancer patients treated with chemotherapy and proton radiotherapy. Additional prospective data is needed to validate these findings and identify which features are most predictive, which may improve future prediction models.

## Abstract 15: Lung and Mediastinum

### Cardiac Biomarkers among Patients with Hodgkin Lymphoma Treated with Anthracycline-Based Chemotherapy and Proton Therapy

#### Bradford Hoppe, Lidia Gizhva, Nancy Mendenhall, Chris Morris, Stella Flampuri

##### University of Florida Health Proton Therapy Institute

PURPOSE: Late cardiac complications from thoracic radiation of patients with Hodgkin lymphoma are of great concern. We investigated whether cardiac biomarkers could identify patients with early cardiac damage from thoracic radiation.

MATERIALS AND METHODS: Following completion of anthracycline-based chemotherapy, 14 patients with stage I-IIIA mediastinal Hodgkin lymphoma were enrolled on an IRB-approved prospective trial and agreed to serum evaluation of troponin, CK-MB, and BNP before beginning radiation, after completing radiation, and every 3 to 6 months for 5 years, during follow-up or until relapse.

RESULTS: Among the 14 patients, median follow-up for the entire cohort was 5 years. No evidence in abnormal values in troponin or CK-MB was observed among the patient cohort. BNP levels demonstrated statistically significant higher values at various follow-up time points. Higher levels of BNP were observed among patients receiving higher anthracycline doses (>250 mg), but not among patients with higher mean heart radiation doses.

CONCLUSIONS: In the first 5 years after treatment, no rise in CK-MB or troponin was identified. BNP levels significantly increased following treatment, with larger increases among patients who received higher doses of anthracyclines. Further investigation is warranted.

## Abstract 16: Lung and Mediastinum

### Mitigating Caveolin-1 Mediated Radioresistance in A549 Non-small Cell Lung Cancer Through Enhanced Relative Biological Effectiveness With Proton Therapy

#### Santanu Samanta, Javed Mahmood, John Eley, Allen Alexander, Sandrine Soman, Michael Creed, Tami Kinsbury, Soren Bentzen, Charles Simone, Zeljko Vujaskovic

##### University of Maryland

INTRODUCTION: Radioresistance in non-small cell lung cancer (NSCLC) cells is a major contributor to radiation treatment failure and Caveolin1 is highly expressed in many radioresistant tumors. In this in vitro study, we sought to demonstrate the modulation of Caveolin1 expression (overexpression and knockout) in a radioresistant A549 lung cancer cell line and the possible mitigation of radioresistance through the use of proton therapy, using their higher relative biologic effectiveness (RBE) near the end of the spread-out Bragg peak (SOBP).

METHODS: A549 cells were treated with photon therapy on a weekly basis (4Gy per week) to a dose of 80 Gy to develop the acquired resistant A549 cell line that was confirmed with Clonogenic assay, MTS assay and increased Caveolin-1 expression. The radioresistant cell line was then treated with scanned pencil beam proton therapy at different doses, and comparison was made between the cell survival fractions (SF) at the distal end of a 10cm width SOBP versus the middle of the SOBP, for equal absorbed dose exposures, and also compared against that from 250 kVp Xray. The linear quadratic model was fit to each data set and used to characterize differences in radiosensitivity and RBE.

RESULTS: Our acquired radioresistant A549 cell line demonstrated stable radioresistance with increased Caveolin-1 expression. The SF for this resistant cell line at the end of SOBP was significantly lower compared to the middle of the SOBP. We found alpha values of 0.0704 and 0.0903 Gy-1, and beta values of 0.0179 and 0.0234 Gy-2 for midSOBP protons, and distal SOBP protons respectively. These fits revealed a 17-21% increase in RBE seen for the distal-SOBP protons compared with the mid-SOBP protons for the dose range of 2-8 Gy.

CONCLUSIONS: We successfully established an acquired radioresistant A549 cell line and demonstrated that modulating the expression of Caveolin-1 can modulate resistance, thereby supporting a possible role of Caveolin-1 in radioresistance. We demonstrated a decreased survival fraction at the end of SOBP compared to the mid of SOBP, suggesting the RBE in this region might be considerably higher than 1.1. Our findings may inform optimal treatment approaches for NSCLC and give justification for pursuing RBE-based treatment planning.

## Abstract 17: Central Nervous System and Pediatrics

### Proton Radiotherapy for Whole Lung Treatment: Potential for Clinically Relevant Improvements?

#### Limin Song, William Ichter, Shikui Tang, Chang Chang, Victor Mangona

##### Texas Center for Proton Therapy

PURPOSE: To study potential dosimetric benefits with spot scanning intensity-modulated proton radiotherapy (IMPT) compared with volumetric modulated arc radiotherapy (VMAT) for whole lung treatment.

METHODS: Two patients treated with bilateral lungs using IMPT at a single institution are selected in this study. Both 3D slow scan and 4DCT was performed during CT simulation for the patients. The clinical IMPT plans were created using Raystation v5.0.2 on either the slow-CT or the 4D averaged CT image data set, and then evaluated using CT data obtained on both full inspiration and full expiration phase of patient's breath cycle. To mitigate interplay effect between spot delivery and target motion, layer repainting and volumetric repainting was utilized. The plans were optimized with 4D optimization and multi-field optimization (MFO) combined with 4% robustness on range uncertainty. For comparison, VMAT plans with 2-3 full arcs were also created for the two patients within the same planning system using 6MV photon. The target coverage was scaled to the same value between the IMPT and VMAT plans.

RESULTS: Various degrees of dose reduction are observed for nearly all OARs (heart, esophagus, spinal cord etc.) using IMPT. In particular, significant dose reduction is seen for the heart in the proton plan. For both patients, the mean heart dose is reduced by about 2 Gy, and V5 decreased by 31% and 23%, respectively. In IMPT plans, the target remains adequate coverage if CT images at different breath phases were used.

CONCLUSION: Comparing with conventional VMAT plans, the IMPT plans for whole lung treatment show dosimetric benefit based on reduction of total integrated dose and doses to OARs, especially the heart. This can potentially lower the risk of late-toxicity for these patients. The results indicate a potential role of IMPT for bilateral whole lung treatment.

## Abstract 18: Lung and Mediastinum

### 4D Robustness Optimization Strategy for Lung IMPT Treatment Planning

#### Mingyao Zhu, Tejan Diwanji, Charles Simone, Pranshu Mohindra, Katja Langen

##### University of Maryland

Beam delivery and tumor motion interplay is a challenge when treating thoracic tumors with proton PBS. To account for a moving target surrounded by low-density lung tissue, the ITV or PTV is often forced to tissue density on the averaged CT of a 4D-CT (CT-avg), which increases the dose to normal tissues. Recently, 4D robustness optimization (RO) became available in commercial TPS. The most robust plan should be optimized using all phases of 4D-CT images; however, this can be very time-consuming. The purpose of this study is to find the optimal number of CT images for 4D RO. Three right-sided locally advanced non-small-cell lung cancer patients were randomly selected. The CTV was contoured on CT-avg and on all ten 4D-CTs, and the ITV was the union of the 11 CTV contours. RO parameters were 5mm for isocenter offset and ±5% of range uncertainty. Four plans were created using: (1) ITV as target, RO only on CT-avg; (2) CTV as target, RO on CT-avg plus two 4D-CTs at maximum inhalation (0%) and maximum exhalation (50%); (3) CTV as target, RO on CT-avg plus four 4D-CTs at 0%, 20%, 50%, and 70%; and (4) CTV as target, RO on CT-avg plus all ten 4D-CTs. Each plan was then re-calculated on the ten 4D-CTs, and robustness was evaluated on the CT-avg and ten 4D-CTs by recalculating the dose with RO parameters, which yield 132 scenarios per plan (12 scenarios per CT image). Worst case scenario was the value among all 12 scenarios for each CT phase. The average value of the 10 phases is listed in Table 1, along OAR doses. For patient 2 and 3, compared with plan 1, plans 2 and 3 have 3.9%-10.1% lower heart mean doses and 1.1%-3.1% lower right lung mean dose without losing robustness (≤0.7%). For patient 1, CTV D99 robustness continuously improved (up to 4.4%) with an increasing number of phases used. RO is now feasible when planning for patients with thoracic malignancies, and using CT-avg with 4 phased CTs appears to allow for the optimal trade-off between effectiveness and efficiency for RO.

Table 1

Comparison of the dosimetric metrics of different robustness optimization methods using (1) AVG: only average CT image, (2) AVG+2: average CT plus 2 (0% and 50%) phased CTs; (3) AVG+4: average CT plus 4 (0%, 20%, 50%, and 70%) phased CTs, and (4) AVG+10: average CT plus all 10 phased CTs. All plans were normalized to ITV on CT-avg D99 of 6000 cGy in 30 fractions.

Abbreviations: RO: robustness optimization; AVG: average; ITV: internal target margin; CTV: clinical target volume; Dmean: mean dose; D99: dose to 99% of the volume; D1: dose to 1% of the volume; Rt: right; WCS: worst case scenario.

## Abstract 20: Central Nervous System and Pediatrics

### A Prospective Outcomes Study of Proton Therapy for Skull-Base Chondrosarcomas

#### Adam Holtzman, Ronny Rotondo, Michael Rutenberg, Daniel Indelicato, Christopher Morris, William Mendenhall

##### University of Florida

PURPOSE: To evaluate the effectiveness of definitive or adjuvant external-beam proton therapy (PT) on local control and survival in patients with skull-base chondrosarcoma.

MATERIALS/METHODS: Between January 2007 and February 2016, 43 patients (median age, 49 years; range, 23-80) with skull-base chondrosarcoma were treated with double-scattered PT. Median total dose was 73.8 Gy(RBE) (range, 64.5-74.4); 33 patients received 1.8 Gy(RBE) once-daily fractionations and 10 received 1.2 Gy(RBE) twice-daily fractionations. Primary site included skull-base tumors involving the clivus, cavernous sinus, petrous bone, and sella, and tumors involving the ethmoid or sphenoid sinuses. Surgical resection and biopsy alone were performed in 36 (83.7%) and 7 (16.3%) patients, respectively. Tumor grade was as follows: 19 (44.2%) grade 1, 22 (51.1%) grade 2, and 2 (4.7%) grade 3. Forty patients had gross disease; 7 were treated for locally recurrent disease following surgery. Distant metastases, overall survival, cause-specific survival, local control, and disease-free survival were calculated. PT-related toxicities were scored using CTCAE v4.0.

RESULTS: Median follow-up was 3.7 years (range, 0.7-10.1). The 3-year actuarial overall survival rate was 95%, cause-specific survival was 100%, disease-free survival was 93%, local control was 93%, and freedom from distant metastases was 100%. Three patients experienced local recurrence with no regional or distant recurrence observed. At last follow-up, 38 patients were alive with no evidence of recurrent disease, 2 were alive with recurrent disease, 2 were dead from intercurrent disease, and 1 was dead of recurrent disease. The time to local progression was 2, 2.5, and 3 years (n=3) following PT. We did not observe any ≥grade 3 acute toxicities. 9.3% (n=4) of patients experienced ≥grade 3 late toxicity: grade 3 hearing loss (n=3) and a grade 3 radiation-induced tumor (meningioma 9 years 9 months after treatment). We observed a grade 2 temporal lobe and brainstem radiation necrosis, both in the same patient.

CONCLUSION: High-dose conformal PT alone or following surgical resection for skull-base chondrosarcoma is an effective treatment with a high rate of local control and with a relatively low toxicity profile. Further follow-up of this cohort is necessary to better characterize long-term disease control and late toxicities.

## Abstract 21: Central Nervous System and Pediatrics

### Dosimetric Impact of Brainstem Injury Risk-mitigation Strategies during Proton Therapy for Posterior Fossa Tumors

#### Jeffrey Gross, Brad Kreydick, Evan Chang, William Hartsell, Mark Pankuch, John Chang, Vinai Gondi

##### Northwestern University, Northwestern Medicine Chicago Proton Center

PURPOSE: Enhanced relative biological effectiveness at the distal range of the proton therapy (PT) Bragg peak during treatment of posterior fossa tumors has led to the implementation of various strategies meant to mitigate risk of brainstem injury. Such strategies include uniform scanning (US) with blocking, range feathering, and range pushing, and pencil beam scanning (PBS). This analysis compares target coverage and organ-at-risk (OAR) dosimetry using these risk-mitigation strategies to PT delivered with US without risk mitigation.

METHODS: Treatment plans for children and adults with medulloblastoma and posterior fossa ependymoma who received postoperative PT to 54-55.8 CGyE at a single institution were analyzed retrospectively for planned target volume (PTV) V95 and D95, brainstem mean dose, brainstem surface maximum dose, brainstem core maximum dose, left and right hippocampal mean doses, left and right cochlear mean doses, pituitary mean dose, and optic chiasm maximum dose. Comparisons of categorical variables were made with Fisher's exact test, and comparisons of continuous variables were made using one-way analysis-of-variance and Bonferroni post-hoc tests, with statistical significance set at p<0.05.

RESULTS: 120 patients who met inclusion criteria were treated between 1/1/2011 and 3/31/2017. Patient, treatment and dosimetric details are outlined in Table 1. Of 96 patients receiving mitigation, 13 (13.5%) had blocking, 38 (39.6%) had range feathering, 10 (10.4%) had range pushing, and 35 (36.5%) had PBS. Mitigation was associated with lower brainstem surface maximum dose (p=0.03). Blocking and range pushing were associated with lower median PTV V95 (93.7% and 89.4% respectively) compared to range feathering and PBS (98.0% and 100% respectively, p=0.0001). Mitigation with blocking, range feathering or range pushing did significantly elevate hippocampal mean dose, cochlear mean dose, pituitary mean dose, or optic chiasm maximum dose. Treatment to posterior fossa or involved field was not associated with any differences in these OAR dosimetric outcomes.

CONCLUSIONS: Risk-mitigation strategies improved brainstem dosimetry without increasing OAR exposure, however blocking and range pushing appeared to compromise target coverage compared to range feathering or PBS. Analysis of the clinical impact of these mitigation strategies is ongoing.

Abstract 23 is a Poster Abstract

## Abstract 24: Central Nervous System and Pediatrics

### Insurance Coverage and Related Referral Delays in Children Irradiated With Proton Therapy in the Pediatric Proton Consortium Registry

#### Clayton Hess, Daniel Indelicato, William Hartsell, Arnold Paulino, Christine Hill-Kayser, Stephanie Perkins, Nadia Laack, Ralph Ermoian, Suzanne Wolden, Andrew Chang

##### Massachusetts General Hospital, University of Florida, Northwestern Medicine Chicago Proton Center, MD Anderson Cancer Center, University of Pennsylvania, Washington University, Mayo Clinic – Rochester, University of Washington – Seattle, Memorial Sloan Kettering Cancer Center, ProCure Proton Therapy Center – Oklahoma City

BACKGROUND/OBJECTIVES: Third-party payer denial of proton radiotherapy can delay or prevent children from receiving toxicity-sparing treatment. We report the prevalence and type of insurance coverage and related treatment delays in children irradiated with proton therapy in the Pediatric Proton Consortium Registry.

DESIGN/METHODS: A multi-institutional registry of integrated demographic, clinical, dosimetric, radiographic, and patient-reported data for childhood cancer patients undergoing proton radiation therapy was conceived in May 2010 and opened to enrollment in summer 2012. Data was frozen for analysis on February 3, 2017.

RESULTS: Of 1,578 children enrolled from 49 states and 20 foreign countries, 1,438 (91%) had available insurance records. A total of 1,350 (86%) were covered by private, public, or foreign insurance (79% fully covered) while 88 (6%) were uninsured and were either charity-sponsored or self-paid. Principal insurers included employer-based or private companies (48%), foreign entities (18%), Medicaid (17%), and the military (2%). Ninety (6%) children experienced a delay in proton therapy while awaiting insurance approval, 51 (57%) and 32 (35%) of whom were insured privately and publicly by Medicaid, respectively. The frequency of treatment delay due to insurance approval was 51/1029 (5%) for private/foreign insurance compared to 32/275 (12%) among Medicaid patients (p<0.001).

CONCLUSION: Proton therapy delay from the insurance approval process occurs in six percent of children and the rate is twice as high in Medicaid patients. Proton therapy remains more common and less delayed in privately insured children. High-quality late effects research is needed to justify further expansion of timely access to toxicity-sparing proton therapy for all American children.

Abstract 25 is a Poster Abstract

## Abstract 26: Central Nervous System and Pediatrics

### Multi-Institutional Pediatric Treatment Planning Project

#### Thomas Whitaker, Anita Mahajan, Chris Beltran, Ben Nelms, Nadia Laack

##### Mayo Clinic, ProKnow

PURPOSE: Proton therapy has been used in the treatment of pediatric tumors for many decades. However, limited access to proton therapy and advances in technology has made it difficult to form consensus on treatment-planning techniques. With the goal of starting a community-wide conversation on how best to deliver treatment to the pediatric population, a multi-institutional treatment-planning project was conducted.

METHODS: A clinical case for a 17-year-old male with a recurrent craniopharyngioma was the subject of the planning project. The goal was to create a radiation-therapy plan that treated the target to 54 Gy in 30 fractions. Web-based software for treatment planning education was used to perform the project. Using the software, a set of treatment planning objectives was defined. A scoring system was developed that assigned users points based on how well each planning objective was met and the clinical importance of the objective. Participants recruited from the PTCOG-pediatric subcommittee were asked to create a treatment plan according to their institution's standard of practice using the CT and RT-structure sets provided. Upon uploading their resulting RT-plan and RT-dose files into the software, participants immediately received their plan score but could not see other participants' scores. Participants were allowed multiple submissions to improve their score. Upon closing participation, the aggregate results were analyzed for correlations between technique and plan score.

RESULTS: Thirty-three plans were submitted from 1 Asian, 10 European, and 13 North American institutions. Of the 150 possible points, 15 plans received 145 points or greater, 13 plans received between 130 and 145 points, and 5 received less than 130 points. The numbers of beams used in each plan were: 6 with 1-2 beams, 13 with 3-4 beams, and 4 with 5-7 beams. No correlation between number of beams and plan score was found.

CONCLUSION: It was found that high-scoring plans could be created regardless of the number of beams chosen. Further analysis should be performed for determining if other metrics, such as beam direction, correlated with plan score. This project was an important first step for multi-institutional learning and consensus building in pediatric treatment planning for proton therapy.

## Abstract 28: Head and Neck

### Clinical Experience of Treatment of Head and Neck Cancer Patients in a Seated Position in Case of Proton Therapy

#### Alexander Pryanichnikov, Aleksandr Shemyakov, Nataliia Strelinkova

##### Protom Ltd

BACKGROUND: A special mobile patient positioning and immobilization unit has been developed within the proton therapy complex “Prometheus.” This unit is much cheaper than a gantry and is suitable for a low-cost system designed to be used with a fixed treatment beam and a rotating seated patient. This paper reports the results of the verification (process which is carried out immediately before the proton therapy session) for the first 50 patients going through clinical treatment at this facility. It contains a list of advantages of the presented system of patient positioning and immobilization in contrast to the standard methods and a gantry used in cases of head and neck cancer treatment. This system has been adapted for proton and ion therapy facilities working with the pencil beam scanning (PBS) technique.

PURPOSE: To report the first clinical experience of using new patient setup unit in a seated treatment position.

MATERIALS AND METHODS: A patient positioning and immobilization unit as a part of proton therapy complex “Prometheus,” civco thermoplastic masks, an X-Ray cone-beam computer tomograph, daily verification procedures

RESULTS: The obtained results of patient setup displacement were slight. We got measurements for each of the dimensions: at the X-axis, -0.1±0.8 mm; at the Y-axis, -0.4±2.0 mm; at the Z-axis, 0.1±1.5 mm. Five hundred thirty daily verification procedures were held for 50 patients. The average time of treatment session was 5 min 39 sec.

CONCLUSIONS: This study demonstrates that the designed patient setup unit is a real alternative for gantry systems in cases of head and neck cancer. It meets all clinical requirements. This device can be used as an independent system of patient positioning and immobilization, or can be a part of complex facilities with gantries and other systems.

## Abstract 29: Head and Neck

### Range Shifting Helmet for Spot Size Reduction in Shallow Head Tumors

#### Daniel Mundy, Stephen Corner, Tyler King

##### Mayo Clinic

With minimum proton energies of around 70 MeV, the treatment of shallow targets with scanned beam proton therapy requires a range shifter to bring dose to the surface from the inherent minimum depth of ∼4cm. The range shifter, usually mounted in the nozzle, scatters the proton beam and increases the spot size at the target, thereby reducing dose falloff and decreasing plan quality. Positioning the range shifter as close to the patient as possible keeps spot size small, thereby preserving critical structure sparing. To this end, a range shifting helmet system was developed for treatment of shallow targets in the head. The system includes three helmets (small, medium, and large), a set of risers to increase the height of the helmet from the treatment couch, and a localization jig. The helmet sizes were selected based on measurements of adult and pediatric patients treated in our center. Helmets were produced using a Fortus 400MC 3D printer with Polycarbonate-ISO filament (Stratasys Ltd., Eden Prairie, MN) and have a water-equivalent-thickness (WET) of 42mm. The design explicitly avoided hard edges and steep WET transitions. The localization jig consists of two aluminum plates where the bottom plate indexes to the treatment couch and the top plate localizes the helmet (with an insert for each helmet size) and can slide, via a rack and ball detent design, up to 7cm longitudinally (the distance between indexing pegs on the couch) relative to the fixed lower plate. This allows the helmet to be positioned as far inferior as possible relative to the patient to allow treatment of nasal tumors with inferior extent. The risers are made of the same material and allow treatment of both pediatric and adult patients using the same set of helmets. Clinically, the helmet has improved plan quality in head tumors immediately adjacent to critical structures such as the optic apparatus and has been shown to reduce dose to healthy tissue in patients with brain tumors. Efforts continue in the development of a system that can accommodate head and neck patients with inferior nodal involvement.

## Abstract 30: Head and Neck

### Verification Computed Tomography Scans Impact on Clinical Decision-Making: An Analysis of Pencil Beam Scanning Proton Therapy in Head and Neck Malignancy

#### Jaden Evans, Daniel Mundy, Aman Anand, William Harmsen, Yolanda Garces, Daniel Ma, Michelle Neben-Wittich, Chris Beltran, Samir Patel, Robert Foote

##### Mayo Clinic – Rochester and Scottsdale

PURPOSE/OBJECTIVES: To determine the clinically impactful timing for obtaining verification computed tomography (CT) scans of patients (pts) treated with pencil beam scanning intensity-modulated proton therapy (IMPT) for head and neck cancers. To identify when verification CT scans influenced physician decision-making to re-plan and test for variables that may be associated with a re-plan.

MATERIALS AND METHODS: Retrospective review of 50 consecutive patients treated with IMPT undergoing weekly verification CT scans for primary or recurrent head and neck cancers. Kaplan-Meier estimates were used for the cumulative probability of a re-plan by week. Univariate Cox modeling was used to evaluate absolute change in weekly weight as a time-dependent covariate. Categorical variables were evaluated with various inference methods to test potential predictive variables for a re-plan.

RESULTS: Eighteen of 50 (36%) pts had verification CT scans prompting a re-plan. All re-plans were initiated from CT verification scans ≤week 5 (fxs 1-25). Cumulative probability of a re-plan by week 1 (fxs 1-5): 10.0% (95% CI: 1.3-17.9), week 2 (fxs 6-10): 20.0% (95% CI: 8.1-30.4), week 3 (fxs 11-15): 28.0% (95% CI: 14.4-39.4), week 4 (fxs 16-20): 32.0% (95% CI: 17.8-43.8), and week 5 (fxs 21-25): 36.0% (95% CI: 21.2-48.0). Decreased target coverage (7 pts, 39%), increased dose to organs at risk (4 pts, 22%), or both (4 pts, 22%) were common reasons for re-planning. Weight as a time-dependent covariate was not associated with re-planning. Pts treated with 4 fields trended towards being re-planned more often with 11 (55%) of 20 cases re-planned compared to 6 (25%) of 24 with 3 fields and 1 (17%) of 6 with 2 fields (p=0.073). There were 16 (46%) of 35 pts with primary tumors re-planned compared to 2 (13%) of 15 pts with recurrent tumors (p=0.052).

CONCLUSIONS: Weekly verification CT scans should be utilized for pts undergoing pencil beam scanning IMPT for head and neck cancers, particularly when daily non-volumetric image guidance is used (e.g. orthogonal kV x-rays). Our data indicates that the first 5 weekly CT verification scans were clinically impactful particularly for pts with primary tumors.

## Abstract 31: Head and Neck

### Outcomes in the Primary Treatment of Head and Neck Cancer Treated With Proton Beam Therapy

#### Katarina Petras, Michelle Gentile, Irene Helenowski, Carl Rossi, Lane Rosen, Shaehed Badiyan, George Laramore, Carlos Vargas, William Hartsell, John Chang

##### Northwestern University, Scripps Proton Therapy Center, Willis Knighton, University of Maryland, University of Washington, Mayo Clinic – Scottsdale, Northwestern Medicine Chicago Proton Center

PURPOSE: Proton beam therapy (PBT) may have an improved therapeutic ratio compared to traditional photon/IMRT therapy across several disease sites. There is limited data on the outcomes of large cohorts of head and neck (H&N) cancer patients treated with PBT in the primary setting.

MATERIALS AND METHODS: We performed a retrospective analysis of 238 patients with cancer of major H&N sites treated in the primary setting from 2010-2017. These patients were included in an ongoing prospective multi-institutional proton database from 8 centers. Statistical outcomes were estimated by the Kaplan-Meier method. Treatment toxicity was scored by the CTCAE version 4.0.

RESULTS: The most common disease sites included salivary glands (26.9%), oropharynx (19.3%), paranasal sinuses (13.9%), oral cavity (13.4%), nasal cavity (11.8%), and nasopharynx (5.9%). The most common histologies included squamous cell carcinoma (34.9%), adenoid cystic carcinoma (7.6%), carcinoma NOS (7.1%), and mucoepidermoid carcinoma (5%). In this cohort, 20% of patients were p16 positive, at least 33% were treated post-operatively, 36% had stage III/IV disease, and 36.5% received concurrent chemotherapy. Median follow-up time: 10 months (range 0-71 months). Median age at the start of radiotherapy: 59 years (range 5-91). Median total dose: 54.1 Gy(RBE) [range 10-82.3 Gy(RBE)], including 23% of patients who were treated <50 Gy(RBE). Fifty-percent of patients were treated with uniform scanning therapy and 29% were treated with passive scatter. No acute or late grade 4 toxicities were recorded. Acute and late grade 3 toxicity was experienced in 32% and 6% of patients, respectively. Local recurrence-free survival at 1 and 2 years was 98.9% (95% CI, 95.6-99.7) and 98.9% (95% CI, 95.6-99.7), respectively. Progression-free survival at 1 and 2 years was 88.4% (95% CI, 82.3-92.5) and 74% (95% CI, 64.9-81), respectively. Distant metastasis-free survival at 1 and 2 years was 98.2% (95% CI, 94.3-99.5) and 97.2% (95% CI, 92.3-99.0), respectively. Overall survival at 1 and 2 years was 91% (95% CI, 85.2-94.6) and 76.5% (95% CI, 67.1-83.5), respectively.

CONCLUSIONS: Primary PBT for treatment of H&N cancer provides excellent outcomes with minimal late grade 3 toxicity. Further analysis and longer follow-up is needed to determine outcomes among individual sites representing a variety of histologies and range of doses for this promising treatment modality.

## Abstract 32: Head and Neck

### Dose Sparing Effectiveness of Spot-Scanning Proton Arc (SPArc) Therapy Versus Intensity-Modulated Proton Radiotherapy in Oropharyngeal Carcinoma: Potential for Less Toxicities

#### Thomas Lanni, Xuanfeng Ding, An Qin, Xiaoqiang Li, Peyman Kabolizadeh, Kimberly Marvin, Craig Stevens

##### Beaumont Health

PURPOSE: Recently, our group introduced a novel proton beam therapy concept Spot-Scanning Proton Arc (SPArc) therapy which is able to deliver a robust, efficient and continuous arc therapy in clinical settings. This study is to evaluate the cost effectiveness of intensity modulated proton therapy (IMPT) and SPArc in the management of toxicities for the treatment of head and neck cancer (HNC).

MATERIALS AND METHODS: Ten patients previously treated for oropharyngeal carcinoma were planned with IMPT and SPArc to compare the dose constraints for target volumes, organs at risk (OAR) and normal tissues. A prescription dose of 70 Gy was planned to gross target volume (V70Gy ≥ 95%) while 60 Gy were planned to intermediate risk target volume and to elective nodal regions respectively. A 3mm setup and 3.5% range uncertainties robustness optimization parameter was used in both SPArc and IMPT planning. Actual direct costs for complications from treatment were determined by actual inpatient admissions for head and neck patients undergoing chemo-radiation at Beaumont Hospital. A cost-effectiveness analysis was performed using incremental cost effectiveness ratios (ICER) to compare toxicities by treatment option.

RESULTS: There were statistically significant differences in median doses to all OARs when comparing IMPT vs SPArc (Table 1). Toxicities for dysphagia and xerostomia were based on previously reported data for VMAT and IMPT1,2. The average direct costs for hospital inpatient admissions at Beaumont was ($4,232). Factoring in the complications for these toxicities based on current planning options, IMPT was found to be more effective ($-6,274.50) than VMAT. Given SPArc's dose sparing potential, we hypothesize there could be a potential for fewer side effects and lower costs.

CONCLUSIONS: SPArc demonstrated consistently better normal tissues sparing compared with IMPT. We hypothesize there will be less toxicities when planning with SPArc then IMPT, therefore the potential for even less cost to manage toxicities for patients undergoing treatment. Ultimately, the comparison of treatment planning technique requires a detailed assessment of both local control, survival and overall toxicity burden.

## Abstract 33: Genitourinary

### An Evaluation of Prostate Volume and Proton Therapy-related Toxicities

#### Elizabeth Krcik, Patrick Lee, Megan Boroczk, Mark Pankuch, William Hartsell

##### Northwestern Medicine Chicago Proton Center

PURPOSE: To analyze trends in the occurrence rate and severity of patient reported adverse proton beam therapy toxicities as a function of prostate size for low and intermediate risk prostate cancer patients.

METHODS: A prospective collection of data was obtained for 688 patients treated at a single institution from October 2010 through October 2016. All patients were enrolled in the Proton Collaborative Group (PCG) registry study and treated with definitive proton therapy to the intact prostate to 70.0 Gy(RBE) at 2.5 Gy(RBE)/fx, or 79.2 Gy(RBE) at 1.8 Gy(RBE)/fx, without pelvic nodal irradiation. Prostate volumes were defined primarily by MRI and separated into four groups: <30cc, 30-60cc, 60-100cc, and >100cc. Genitourinary, gastrointestinal, and dermatological toxicities were collected, and the frequency of each toxicity grade was calculated for each prostate size group.

RESULTS: There were no significant trends in gastrointestinal or dermatological toxicities based on prostate volume. Many urinary symptoms, including IPSS/AUA scores, also showed no correlation. However, there were notable increases in grade 1 and 2 hematuria with increasing prostate size. Smaller prostate volumes (<30 cc and 30-60 cc) experienced 9.7% and 11.5% grade 1 toxicity respectively, and 0.6% and 1.0% grade 2 toxicity. Larger prostate volumes of 60-100 cc and >100 cc reported grade 1 toxicities of 16.7% and 27.3% respectively, and grade 2 toxicities of 5.0% and 4.5%. There were very few high grade toxicities overall, with no significant correlation to prostate volume.

CONCLUSION: While most proton therapy-related toxicities exhibited no significant relationship to prostate volume, there is a correlation between prostate size and incidence of low grade hematuria. The low occurrence rates of all grade 3 toxicities demonstrate a low severity of symptoms in prostate patients treated with proton therapy, irrespective of prostate size.

## Abstract 34: Genitourinary

### Proton Therapy Is Associated With Superior Survival and Decreased Risk of Complications Compared to IMRT for Intermediate Risk Prostate Cancer: A Medicare/SEER Database Study

#### William Hartsell, Hassan Bentefour, David Dooling, Nancy Mendenhall

##### Northwestern Medicine Chicago Proton Center, IBA, Innovative Oncology Business Solutions, University of Florida

RATIONALE: For patients treated with external radiotherapy for prostate cancer, there is a significantly higher integral dose given with IMRT compared to protons. This difference could increase the risk of complications or second malignancies with IMRT.

METHODS: The SEER/Medicare database was used to evaluate patients with prostate cancer, with initial treatment from 2006-2012, treated with complete course of either protons or intensity modulated radiation therapy (IMRT). There were 29,554 IMRT patients and 1,018 proton patients, with complete records available on 28,102 IMRT patients and 851 proton patients. The proton patients were matched to IMRT patients in a 1:5 matching process, using these parameters: comorbidity index, cancer stage, tumor grade, adjuvant chemo/hormone therapy, age, zip code and ethnicity. Medicare billing codes were evaluated for GI, GU, erectile dysfunction, bone/hip fracture, endocrine dysfunction and “other” complications.

RESULTS: For the entire cohort of unmatched prostate cancer patients the 5 year overall survival was 93.4% for PBT vs 85.25% for IMRT (p=0.00001). A total of 3,867 IMRT patients were matched to 788 PBT patients. Using the matched patient cohort, the 5-yr overall survival for PBT was 93.25% vs 88.43% for IMRT (p=0.045). The matched patient group of intermediate risk patients included 736 PBT and 3,591 IMRT patients, with 5 year overall survival of 93.65% vs 88.27% (p=0.022). There was no significant difference in erectile or bone/hip complications, but significantly more patients with GI, GU, endocrine, or “other” complications with IMRT compared to protons. In addition, there were greater numbers of patients with secondary malignancies with IMRT, with the curves beginning to separate after 3 years from treatment (10.5% for IMRT vs 6.1% for PBT at 5 years, p=0.09).

CONCLUSIONS: Proton therapy for prostate cancer is associated with fewer long term complications and a trend toward a lower risk of second malignancies. There is a difference in overall survival at 5 years; further work is necessary to determine if this difference is due to complications, second malignancies or prostate cancer recurrence.

## Abstract 35: Genitourinary

### Rectal Toxicity for Patients Treated With Proton Beam Radiation Diagnosed With Prostate Cancer Using Free Rectal Water, Water Filled Rectal Balloons, or Interstitial Hydrogel for Rectal Immobilization

#### Victoria Casablanca, William Hartsell, Ashley Pruneau, Rachel deLongpre, Megan Boroczk, Lorena Botello, Stephanie Perry, Angie Guderjan, Rebecca Price

##### Northwestern Medicine Chicago Proton Center

BACKGROUND: Multiple techniques are used to decrease the rectal dose and improve internal immobilization for treatment of prostate cancer. We hypothesized that patients with an implanted biodegradable tissue expander would have lower grade rectal toxicities than the alternatives (rectal fill of free water or rectal balloon).

METHODS: Prospective collection of data through the prostate registry from a single institution with Proton Collaborative Group from October of 2010 to September of 2015. Most were treated with daily fractionated proton radiation therapy to 79.2 Cobalt Gray Equivalents (CGE) in 1.8 CGE fractions. Three consecutive cohorts of patients were treated, first with free rectal water (385 patients), then with rectal balloons (96 patients) and most recently with an interstitial hydrogel (167 patients). We evaluated GI toxicities (proctitis, fecal incontinence, and rectal hemorrhage) for each of the three techniques.

RESULTS: Grade 3 GI toxicities were rare, occurring in only 2/648 patients (0.31%). Grade 2 toxicities were more common in patients treated with free water (57/385, 14.81%) or rectal balloon (16/96, 16.67%), than with hydrogel (7/167, 4.19%). A significantly larger portion of hydrogel patients experienced no GI toxicity (Grade 0 – 134/167, 80.24%) compared to free water (205/385, 53.25%) or rectal balloon (52/96, 54.17%).

CONCLUSION: The incidence of >Grade 2 rectal toxicity is significantly lower in patients who are treated with an implanted biodegradable hydrogel expander, compared to prior techniques of rectal immobilization. (See Table 1.0 and Table 1.1)

## Abstract 37: Genitourinary

### Hip Toxicity in Patients Receiving Proton Beam Therapy for Prostate Cancer

#### Howard Lee, Jing Zeng, Meghan Macomber, Matthew Spraker, Adriana Blakaj, Jay Liao, Kenneth Russell, George Laramore, Ramesh Rengan, Smith Apisarnthanarax

##### Duke University School of Medicine, University of Washington

PURPOSE: Proton beam therapy (PBT) often delivers higher mean doses to the femoral heads than photon-based treatments, but the clinical implications of this are unclear. We investigated the rate of hip toxicity and associated clinical and dosimetric correlates in prostate cancer patients treated with PBT.

METHODS: One-hundred forty patients on an IRB approved registry study were treated with PBT from 2013-2015. Hip toxicity was assessed retrospectively based on the CTCAE v4 bone pain criteria. Univariate logistic regression and Kruskal Wallis ANOVA tested correlations of patient characteristics and dosimetric parameters with G1+ hip toxicity.

RESULTS: For the 140 patients treated, 36 had low risk, 72 intermediate risk, and 32 high risk prostate cancer. Ninety-five percent of patients received ≥ 78 CGE and nearly all (99%) were treated with opposed lateral beams. Forty-two patients were treated with uniform scanning (US), and 98 with pencil beam scanning (PBS) techniques. Median follow-up was 2 years and minimum follow-up was 1 year. Hip toxicity was observed in 13% (n=18) of all patients with a maximum duration of 9 months. Grade 1 toxicity occurred in 11 patients (8%), grade 2 in 6 patients (4%), and grade 3 in 1 patient (<1%). Most patients (61%, n=11) experienced hip toxicity within 1 year of treatment completion; 4 patients experienced pain during treatment only and 3 patients after 1 year. Dosimetric parameters did not correlate with GR1+ toxicity (p > 0.49). Mean and max femoral head doses in patients experiencing toxicity were 23.95 Gy and 36.51 Gy respectively, versus 24.91 Gy and 36.84 Gy in those not experiencing toxicity. No clinical parameters were significantly related to G1+ toxicity, including androgen deprivation, whole pelvis radiation, number of fields/day (1 or 2), and age (p > 0.5). However, 5% of US patients and 16% of PBS patients developed GR1+ hip pain (p=0.057).

CONCLUSIONS: Hip pain is a clinically relevant toxicity in PBT treated prostate patients, but this study found no correlation of pain with femoral head dosimetry or clinical comorbidities. Additional femoral head dosimetric parameters as well as other organs at risk, such as the acetabular fossa and pelvic musculature, are under investigation.

## Abstract 38: Genitourinary

### Potential Improved Outcomes with Proton Therapy in Prostate Cancer: A Comparison of IMRT and Proton Cohorts

#### Nancy Mendenhall, William Wong, Curtis Bryant, Sujay Vora, Randal Henderson, Sameer Keole, Bradford Hoppe, Romaine Nichols, William Mendenhall, Steven Schild

##### University of Florida, Mayo Clinic – Scottsdale

BACKGROUND: The ProtecT trial underscores the importance of definitive treatment in prostate cancer patients with life expectancies >10 years and the validity of radiation therapy (RT). To shed light on the controversy of whether Proton Therapy (PT) may offer advantages over photon based intensity modulated RT (IMRT), clinical outcomes of PT and IMRT cohorts from two institutions were directly compared.

METHODS: The first comprised 1,214 men treated with image-guided PT using implanted fiducials from 2006 to 2010 to a dose of 78 Gy(RBE) in 39 fractions. The second comprised 301 men treated with image-guided IMRT using transabdominal ultrasound from 2000 to 2005 to a dose of 75.6 Gy in 42 fractions. Median age and follow-up were 66 yrs and 5.6 yrs for PT and 74 yrs and 7.2 yrs for IMRT. Hormone therapy (ADT) was used with PT and IMRT, respectively, in 7% and 3% of low-risk, 9.9% and 25% of intermediate-risk, and 57.8% and 91% of high-risk patients. Comparative endpoints were GR ≥3 gastrointestinal (GI) and urologic (GU) toxicity, and 5 Y freedom from biochemical progression (FFBP).

RESULTS: The prevalence of GR ≥3 toxicity at last follow-up for the PT and IMRT cohorts were were 0.1% vs 1.3% (p=0.0065) for GI toxicity and 0.1% vs 4.3% (p<0.0001) for GU toxicity. FFBP rates for PT and IMRT for low-risk were 98.9% and 92.2% (p<0.0001), for intermediate-risk patients, 94.5% and 87.3% (p=0.0226), and for high-risk patients, 74.4% and 80.3% (P=0.5154).

CONCLUSIONS: In this retrospective comparison of outcomes of men treated with IMRT and PT for prostate cancer, FFBP rates were better with PT for men with low- and intermediate-risk disease and similar in men with high-risk disease, despite longer and more frequent use of ADT in the IMRT cohort. It is notable that toxicity was less in the PT cohort despite the use of a higher median dose. While this study identified difficulties in comparing retrospective series (differences in age, RT dose and fraction size, and ADT use between cohorts), the magnitude of improvement with PT is intriguing and warrants prospective testing.

## Abstract 39: Genitourinary

### Does Rectal Culture and Sensitivity Reduce the Risk of Bacterial Sepsis After Placement of Fiducial Markers for Prostate Radiotherapy?

#### William Mendenhall, Curtis Bryant, Christopher Morris, Gail Sarto, Christopher Williams, Bradford Hoppe, Randal Henderson, Romaine Nichols, Nancy Mendenhall

##### University of Florida

PURPOSE: Bacterial sepsis is a potentially life-threatening complication that occurs in 3-5% of patients undergoing prostate biopsy or fiducial marker placement prior to radiotherapy (RT) regardless of whether performed by the transrectal or transperineal route. The responsible organism is usually quinolone resistant E. Coli.

METHODS: Between August 2015 and January 2017, rectal cultures were obtained in 250 consecutive patients to identify quinolone resistant E. coli before transrectal (154 patients) or transperineal (96 patients) fiducial marker placement (3 to 4 markers per patient) prior to definitive proton RT for prostate cancer. The prophylactic antibiotic regimen was altered based on the sensitivity profile for patients with resistant E. coli. The median IPSS score was 8 (range, 0 to 33). The interval between fiducial placement and most recent prostate biopsy was less than 6 months for 214 patients (86%), and 65 patients (26%) had 1 or more additional prior prostate biopsies. SpaceOar placement, usually with transperineal fiducial marker placement, was performed on 96 patients (38%).

RESULTS: Quinolone resistant E. coli were identified in 28 patients (11%). Three patients (1%) were hospitalized after fiducial marker placement for syncope (1), urinary obstruction (1), and deep vein thrombosis (1). No patient was hospitalized for bacterial sepsis.

CONCLUSION: Rectal culture to identify quinolone resistant E. coli likely reduces the risk of bacterial sepsis after placement of fiducial marker placement prior to definitive RT for prostate cancer.

## Abstract 40: Genitourinary

### Five-year QOL and 7-year Prostate Cancer Outcomes among Men 60 and Under Treated with Proton Therapy

#### Bradford Hoppe, Clement Ho, Curtis Bryant, Randal Henderson, William Mendenhall, Romaine Nichols, Chris Morris, Zuofeng Li, Nancy Mendenhall

##### University of Florida Health Proton Therapy Institute

PURPOSE: Limited data exist for outcomes among younger men receiving definitive treatment with radiotherapy. The present study reports 5-year quality of life and 7-year disease specific outcomes for men ≤60 years old treated with proton therapy for prostate cancer.

METHODS AND MATERIALS: From August 2006 through January 2010, 254 men ≤60 years old were enrolled on an IRB-approved outcomes tracking protocol and treated with proton therapy alone for prostate cancer without androgen deprivation therapy. Risk group stratification included 56% with low-, 42% with intermediate-, and 2% with high-risk disease. Patients were treated with proton therapy at doses of 76-82 Gy at 2 Gy/fraction or 70-72.5 Gy at 2.5 Gy/fraction. Before treatment and every 6-12 months thereafter, patients were evaluated by a physician, answered Health Related Quality of Life (HRQOL) survey including the EPIC, IIEF, and IPSS, and had PSA evaluated.

RESULTS: Median follow-up for the entire cohort was 7.1 years with a 7-year biochemical-free survival of 97.8%. Eight men (1 high risk; 5 intermediate risk; 2 low risk) experienced biochemical progression, including 1 who died of disease 9 years after treatment. Potency, defined by erections firm enough for sexual intercourse, was 90% at baseline, declined to 72% at the first-year follow-up, but declined to only 67% at 5 years. Only 2% of patients developed urinary incontinence requiring the use of pads. The bowel habits mean score declined initially from a baseline of 96 to 88 at 1 year, which improved over the following years to 93 at 5-year follow-up.

CONCLUSIONS: Young men with prostate cancer continue to have excellent results with respect to 7-year biochemical control and erectile function, without clinically significant urinary incontinence 5 years after proton therapy. Comparative effectiveness studies of proton therapy with surgery and IMRT are needed.

## Abstract 41: Gastrointestinal

### Preliminary Toxicities for a Trial of Dose-escalated Proton Radiotherapy with Elective Nodal Irradiation for Pancreatic Adenocarcinoma (UFHPTI PC04)

#### Romaine Nichols Jr., Bradford Hoppe, Michael Rutenberg, Zuofeng Li, Stella Flampouri, Christopher Morris

##### University of Florida

INTRODUCTION: To review preliminary toxicities for the first patients treated at the University of Florida Health Proton Therapy Institute (UFHPTI) on a trial of dose-escalated proton therapy (PT) with elective nodal irradiation for patients with unresectable, borderline resectable, or medically inoperable pancreatic adenocarcinoma (UFHPTI PC04 Trial).

METHODS: The UFHPTI PC04 trial was activated on March 31, 2016. By April 29, 2017, 8 patients had completed PT. Protocol therapy delivers a dose of 40.50Gy(RBE) in 18 fractions to an initial PTV volume (PTV1) including an internal gross tumor volume (iGTV) and an elective nodal volume consisting of a 2-cm expansion around the most proximal 1cm of the celiac artery and the most proximal 2.5cm of the superior mesenteric artery. A second volume (PTV2) including the iGTV receives an additional 22.50Gy(RBE) in 10 fractions subject to normal tissue constraints. Normal tissue constraints for the duodenum, stomach, and bowel space match the constraints of the NRG 1201 protocol. Patients receive capecitabine chemotherapy 1000mg PO BID on PT treatment days. Patients may receive chemotherapy or other systemic therapy prior to protocol therapy.

RESULTS:

Median age, 77 years (62 to 88)

6 male; 2 female

T4, 4 pts. T3, 3 pts. T2, 1 pt.

2 grade 2 toxicities: 1 pt with abdominal discomfort and weakness related to ascites that developed during the third week of treatment. One pt with grade 2 dermatitis without moist desquamation reported in the final week of PT. Two patients experienced interruptions in treatment. One patient with ascites required a 10-day break for paracentesis and PT re-planning. One patient was hospitalized with urosepsis unrelated to protocol therapy but was able to complete PT after a 27-day break. Median weight change from the first to sixth week of treatment was -2.6 lbs (range +12.3 to -14.2).

CONCLUSION: The lack of significant acute treatment-related toxicity for patients treated on this protocol is encouraging. Updated toxicity data will be presented at PTCOG-NA in October 2017. In May 2017, The Proton Collaborative Group (PCG) made the PC04 protocol available as a multi-institutional study.

Abstract 42 is a Poster Abstract

## Abstract 43: Gastrointestinal

### A Multi-institutional Analysis of Radiation Modality for Elderly Patients with Esophagus Cancer

#### Scott Lester, Michael Chuong, Neha Bhooshan, Zhongxing Liao, Ritsuko Komaki, Michael Haddock, Minesh Mehta, Christopher Hallemeier, Kenneth Merrell

##### Mayo Clinic, MD Anderson Cancer Center, Miami Cancer Institute, University of Maryland

PURPOSE: The therapeutic gains of neoadjuvant chemoradiotherapy followed by esophagectomy may be offset by morbidity and mortality in elderly patients. Advanced radiation modalities may reduce collateral dose to the heart and lungs and possibly improve outcomes. We evaluated the impact of radiation modality on outcomes of trimodality therapy for elderly patients with esophagus cancer.

METHODS AND MATERIALS: We evaluated 571 patients treated with trimodality therapy at 3 high-volume tertiary U.S. cancer centers from 2007 to 2013. Two hundred two of 571 (35%) patients were 65 years or older at diagnosis and were classified as elderly. Eighty-seven (43%) elderly patients received 3-D conformal radiation (3DRT), 73 (36%) received intensity-modulated radiation (IMRT), and 42 (21%) received passive scatter proton beam therapy (PBT). Outcomes were analyzed by radiation treatment modality and compared using univarible (UVA) and multivariable (MVA) logistic analyses.

RESULTS: Elderly patients had a higher risk for postoperative cardiac (UVA: OR 2.2, p<0.001; MVA: OR 2.07, p=0.004) and pulmonary toxicities (UVA: OR 2.0, p<0.001; MVA: OR 2.03, p<0.001), and a higher 90-day postoperative mortality (5.4 vs 2.2%, p=0.049) than younger patients. Elderly patients treated with PBT had a lower rate of 90-day postoperative mortality than patients treated with IMRT or 3DRT (0 vs 6.9%). Use of IMRT or PBT was associated with significantly lower cardiac complications in elderly patients compared to 3-D conformal radiotherapy (Table 1). Use of PBT was associated with significantly lower pulmonary complications than IMRT and 3DRT (Table 1). 8.1% of the elderly treated with 3DRT or IMRT experienced acute respiratory distress syndrome (ARDS) compared to 2.4% of those treated with PBT (p=0.307). Significantly more elderly patients treated with PBT were discharged within 7 days of surgery (23.8 vs 6.3%, p=0.002). Rates of pathologic complete response (28.6 vs 30.0%, p =1.000), relapse (42.9 vs 45%, p =0.863), and 3-year survival (63.2 vs 54.4%, p=0.237) for patients treated with PBT compared to those treated with 3DRT or IMRT.

CONCLUSIONS: Proton beam therapy was associated with lower rates of 90-day postoperative mortality, postoperative pulmonary complications particularly ARDS, postoperative cardiac toxicity, and a shorter time to discharge when compared with 3DRT and IMRT.

## Abstract 44: Lung and Mediastinum

### Comparative Toxicities and Costs of Preoperative Chemoradiotherapy of IMRT and Proton Beam Therapy for Esophageal Cancer

#### Steven Lin, Kaiping Liao, Sherif Shaaban, Ning Zhang, Reza Mehran, Wayne Hofstetter, James Cox, Ritsuko Komaki, Sharon Giordano

##### The University of Texas MD Anderson Cancer Center

BACKGROUND: To better understand the value of proton beam therapy (PBT), we conducted a comparative toxicity and cost analysis of PBT versus Intensity Modulated Radiation Therapy (IMRT) in patients treated with neoadjuvant CRT for esophageal cancer at a single, high volume center.

MATERIALS AND METHODS: Consecutive patients (N=237, IMRT=156 and PBT=81) from 2006 to 2013 who had at least 41.4 Gy, ≥30 days post-CRT follow-up, and Ivor-Lewis esophagectomy were analyzed. Cost analysis was based on aggregate health care claims from initial consultation to time of hospital discharge. Charges were discounted to the level of 1st quarter 2007. Differences in charges were examined by using generalized linear model (GLM), and multivariable generalized linear regression models assessed the independent association of patient characteristics and treatment costs.

RESULTS: The two groups were well balanced in patient, tumor, treatment characteristics, and underlying comorbid illnesses. In the postoperative period, rates of complications were slightly more for IMRT in pulmonary (19% vs 13%, p=NS) and GI (15% vs. 14%, p=NS), but higher in cardiac complications (10% vs 2.5%, p<0.05). As expected, the average cost of neoadjuvant CRT is greater for PBT ($128,427 ± 15,800) than IMRT ($71,108 ± 14,858) (p<0.0001). Conversely, the costs postoperatively were significantly higher for IMRT than PBT ($109,807 vs.$92,512). Age had a significant influence on this cost differential: patients <65 years, IMRT was more costly per patient by $5,935, but for 65-74 and ≥75 years old, IMRT exceeds PBT by$20,292 and \$67,876, respectively. The cost of hospitalization remains unchanged for PBT regardless of age group, but rises significantly for IMRT patients older than 65. This is largely due to greater number of procedures and supportive care used for IMRT patients, as well as the length of hospitalization (+2 days for 65-74 years and +7.11 days for ≥75 years old).

CONCLUSIONS: Despite higher initial costs, the overall cost differential of PBT and IMRT may in fact be smaller than expected due to reduced postoperative complications for PBT. This is particularly significant for elderly patients > 65 years old. These results may help determine the optimal utilization of PBT in esophageal cancer.

## Abstract 45: Gastrointestinal

### Proton Therapy for Pancreatic and Ampullary Carcinoma

#### Romaine Nichols Jr., Christopher Morris, Laura Tottenham, Zuofeng Li, Stella Flampouri, Soon Huh, Meng Wei Ho, Nancy Mendenhall, Bradford Hoppe, Michael Rutenberg

##### University of Florida

PURPOSE: Review outcomes for a series of patients treated with proton therapy for non-metastatic pancreatic and ampullary carcinoma.

METHODS: From 2009 to 2016, 58 patients with non-metastatic pancreatic (n=57) or ampullary (n=1) cancer were treated with proton therapy (PT). Patients were enrolled on either the UFHPTI PC01, PC02, PC03, PC04 protocols or the IRB-approved Outcomes Tracking Protocol (OTP). The median patient age was 68.9 years (range 45.7 to 89.4). Treatment indications included: unresectable (n=34), borderline resectable (n=6), medically inoperable (n=2), postoperative (n=8), and planned preoperative (n=8). The median doses in Gy (RBE) for the 5 groups were 59.4, 63.0, 63.0, 54.9, and 50.4, respectively. Most patients received concomitant chemotherapy of capecitabine 1000mg PO BID on treatment days.

RESULTS: Patients underwent toxicity evaluation using the CTCA Evs4.0 grading system weekly during treatment. No patient experienced grade 3+ gastrointestinal (GI) toxicity; 15 patients (26%) experienced grade 2+ GI toxicity at some point during treatment. The median weight loss from start to end of PT was 3.0 lbs (range -18.2 to +12.3). Treatment interruptions were minimal with only 6 (10%) patients experiencing more than a 3-day delay for any reason. The median follow-up for all patients was 10.2 months. Two-year survival data for the groups are as follows:

All patients: 21% 2YS with 11 still at risk - 14/58 alive

Unresectable + Borderline Resectable + Inoperable: 23% 2YS with 10 still at risk - 9/42 alive

Adjuvant Postop: 44% 2YS with 2 still at risk - 5/8 alive

Planned Preop: 13% with 2 still at risk 0/8 alive (1 died after 2 years)

CONCLUSION: Chemoradiation with protons was well-tolerated even with dose escalation. Updated survival data will be available at PTCOG-NA, 2017.

Abstracts 46 and 47 are Poster Abstracts

## Abstract 48: Physics Range and Uncertainty

### Validation of Relative Stopping Power Calculations Based on Electron Density and Atomic Number from a Commercial Dual Energy Software Package

#### John Roeske, Isabel Alemida, Guillaume Landry, Rakesh Patel, Frank Verhaegen, Katia Parodi, Mark Pankuch, George Coutrakon, Reinhard Schulte

##### Loyola University, MAASTRO Clinic, University of Munich, Ludwig-Maximilians, Northwestern Medicine Chicago Proton Center, Northern Illinois University, UCSF

PURPOSE: To evaluate two independent methods, one of which is based on a commercial software package, for the generation of relative stopping power (RSP) from a dual energy (DE) CT.

METHODS: A customized phantom was designed to provide calibration data for this study. Materials with known atomic number and electron density were individually inserted into the phantom and scanned using a CT having sequential DE imaging capabilities (80 kVp and 140 kVp). Based on the DE images, the scanner software produced relative electron density (rho) and atomic number (Z) images. These images were used as spatially dependent parameters for the Bethe equation to produce RSP images. Separately, the methods of Landry et al (PMB 58[19]:6851, 2013) and of Saito et al (MedPhys 39[4]:2021, 2012), based on the ratio of the attenuation coefficients measured at high and low energies and the weighted difference of the CT numbers, respectively, were used to generate RSP images as described and validated by measurements in Hudobivnik et al (MedPhys 43[1]:495, 2016). To compare both methods, an anthropomorphic head phantom was scanned with DECT using the previously listed energies. Regions of interest (ROI) were obtained within homogeneous regions of brain and bone for evaluation.

RESULTS: Based on 11 material inserts, the average difference between RSP values based on the Bethe equation using known material values and those obtained from the rho/Z images was -0.13% +/- 1.00%. A comparison between RSP values using the rho/Z images and those obtained using the Landry/Saito method was 0.04% +/- 0.74%. This difference excludes the lung inserts as the rho/Z method does not provide Z values for HU<-500. An analysis of the head phantom data showed overall good agreement. In soft tissue regions, the ratio of RSP values for the Landry/Saito method vs. the rho/Z was 0.996, while within bone, the average was 0.998.

CONCLUSIONS: Use of the rho/Z images obtained directly from the commercial DECT scanner software results in good agreement with RSP values in comparison to the Landry/Saito method. Either approach may increase the accuracy of proton dose calculations compared to the stoichiometric technique used with standard X-ray CT.

## Abstract 49: Physics and Biology

### Pencil Beam or Monte Carlo Calculation, Which Should Be Used for IMPT Plan Optimization?

#### Mingyao Zhu, Ulrich Langner, Katja Langen

##### University of Maryland

Historically, proton treatment plans are based on analytical PB calculation. Monte Carlo dose calculation is more accurate; however, it hasn't been available in commercial TPS until recently and is time consuming. Furthermore, the improvement in dose accuracy varies among treatment sites and delivery hardware use (e.g. range shifter). The purpose of this work is to quantify the difference between PB and MC for 4 disease sites: prostate, lung, breast, and head and neck. Four previously treated patients, one for each site, were randomly selected. Three inversely optimized plans were created using: (i) only PB for optimization and final dose; (ii) plan (i) recalculated with MC only for the final dose; and (iii) MC for optimization and the final dose. The MC optimization dose was calculated with 10,000 protons/spot, and the statistical uncertainty of final MC dose calculation was 0.5%. No range shifter was used for the prostate plan, 3cm range shifter was used for the lung plan, and the 5cm range shifter was used for the breast and HN plans. SFO was used for prostate, lung, and breast plans, and MFO for HN. The target coverage, mean dose, hot spot, and H-index (homogeneity index) are listed in Table 1. The difference between plan (i) and (ii) indicates the improvement of MC dose calculation. For the prostate plan, the difference is less than 2% for the dose and the H-index; therefore, MC optimization doesn't provide significant improvement. For the lung, the difference for D95 is 11.72%, Dmean is 6.93%, D1 is 3.92%, and H-index is 9%. For the breast plan, the difference is 6.71%, 2.54%, 3.57%, and -2.03% for CTV D95, IMN D95, CTV Dmean, and D1; and the H-index is 10% different. For HN, the CTV D95, Dmean, D1 dose difference is 7.66%, 4.6%, -0.64% respectively; and the H-index is 9% different. For all sites, using MC for optimization with 10,000 protons/spot yields target coverage and H-index similar to the PB plans. These results indicate that using MC during optimization can substantially improve the PB calculation inaccuracy for the lung, breast, and HN plans, but only marginally for prostate plan.

Table 1

Comparison of the dose metrics between the plans created with (i) PB: pencil beam calculation for optimization and final dose, (ii) MC calc: plan (i) recalculated with MC for the final dose, and (iii) MC: Monte Carlo used for both optimization and final dose calculation.

Abbreviations: HN: head and neck; CTV: clinical target volume; D95: dose to 95% of the volume; Dmean: mean dose; D1: dose the 1% of the volume; H-index: ratio of D1 to D95.

## Abstract 50: Physics and Biology

### Predicting Dose-Volume Histograms for Organs at Risk in PBS Proton Therapy Planning

#### Samantha Hedrick, Jackson Renegar, Ben Robison, Marc Blakey, Mark Artz, Neik Schreuder

##### Provision Center for Proton Therapy

With the move towards pencil beam scanning delivery and inverse planning techniques in proton therapy, it's possible to achieve escalated target dose and improved healthy tissue sparing, so it's more important than ever to ensure plans are developed with the highest quality. One method to ensure quality is knowledge based planning (KBP), which provides a priori knowledge of achievable OAR sparing and reduces planning time. We have developed a scripting tool within our treatment planning system to evaluate the patient-specific feasibility of clinical goals.

Our KBP script is based on geometric overlap of a target and OARs. We evaluated a sample of patients and developed expansion margins to mimic the 95%, 90%, 80%, and 60% prescription isodose lines. A GUI allows the user to choose one or more OARs and the margins for robust optimization. The script expands the prescription target based on robust margins and each isodose expansion, and an overlap volume is calculated for each OAR. It's assumed that the target plus the robust margin will receive V100% = 100%. Thus, the overlap with each isodose expansion is the theoretical achievable volume receiving its respective dose. The results are printed to a notepad file and can be automatically included in the optimization constraints. To validate the script, we tested low- and high-risk prostate, H&N, brain, and anal. For each site, a plan was developed without the script to the best of the planner's ability; then the script was run and a new plan was developed with the DVH values included as planning constraints.

All plans passed for deliverability and robustness. Application of the KBP script led to consistent improvement in OAR sparing across all evaluated treatment sites. The figure highlights an H&N example, with significant improvement in parotid and larynx sparing. After validation was complete, dosimetrists now run the script before a plan is started and integrate the results in to optimization.

We have developed and validated a simple knowledge based planning script to provide a priori knowledge of achievable OAR sparing which has improved our PBS proton therapy treatment plan quality and reduced planning time.

## Abstract 51: Physics Range and Uncertainty

### Thermoacoustic Range Verification of Proton Beams: A Simulation Study Integrating TOPAS Monte Carlo with k-Wave Acoustic Software

#### Sarah Patch, Tianyu Zhao

##### UW – Milwaukee

BACKGROUND, MOTIVATION AND OBJECTIVE: The purpose of this work is to examine robustness of thermoacoustic range verification with respect to beam direction as applied to prostate cancer. Thermoacoustic emissions are bandlimited below 150 kHz, corresponding to wavelengths exceeding 10 mm. Thermoacoustic imaging resolution is limited to one-half wavelength. To improve accuracy beyond 5 mm and robustness to additive white noise, a priori information was utilized.

CONTRIBUTION/METHODS: TOPAS Monte Carlo simulations modeled energy density deposited by oblique and horizontal pencil beams. For each beam angle, proton range was translated in ±5 mm water equivalent increments. The energy density was computed on the same mesh as the planning CT, with 1 mm spatial discretization. A three-speed soundspeed map was generated using the planning CT with 3200 m/s, 1540 m/s and 1480 m/s for bone, muscle and organ, and fat tissues, respectively. Three-dimensional k-wave simulations propagated thermoacoustic pulses throughout the planning CT volume. Pulses were recorded at virtual transducer locations on a 5 cm side-fire TRUS array placed distally and laterally to the oblique and lateral pencil beams respectively. One-way beamforming of thermoacoustic emissions yielded range estimates in the sagittal plane, indicated by a yellow “o” in Fig 1c. Thermoacoustic emissions were compared to those simulated from a synthetic control beam with range nearest the beamformed estimate. Time shifts between the measured and control thermoacoustic emissions were estimated by applying the Fourier shift theorem. Stepping from the known range of the control beam by a distance determined from the time shifts yielded the final range estimate.

RESULTS/DISCUSSION: Thermoacoustic pulses generated during oblique and lateral delivery were highly sensitive and insensitive to range change, respectively (Fig. 1d). Utilizing simulated data improved accuracy and robustness to noise over one-way beamforming. Range errors for an oblique beam angle of 48° were less than 1.5 mm and 3.0 mm for noise-free and SNR=1/2 data.

CONCLUSION: Accuracy and robustness vary with the location and orientation of the ultrasound receivers with respect to the proton beam.

CLINICAL RELEVANCE: Real-time range verification could enable more aggressive and targeted treatment planning. For instance, delivery could be halted when oblique or anterior beams dose the radiosensitive rectum.

## Abstract 52: Physics and Biology

### Predicting Layer Time for Pencil Beam Scanning

#### Chuan Zeng, Haoyang Liu, Chin-Cheng Cehn, Dennis Mah

##### ProCure Proton Therapy Center – New Jersey

PURPOSE: The interplay between spot scanning and breathing motion presents challenge on motion management. The knowledge of layer time becomes essential in selection of patient-specific repainting strategies. The ratio of layer time to patient breathing period directly determines the effectiveness of layer repainting. This study establishes models of multiple levels of complexity and accuracy to predict layer time in order to facilitate efficient clinical treatment planning decision making.

METHODS: Logs from the treatment control system were used to establish layer time prediction models. Eight deliveries of one 47-layer field were used to estimate the intrinsic fluctuation in actual delivery time, which set the practical goal of accuracy. An empirical model of layer time was proposed including: i) a constant offset representing overhead time for each layer, ii) number of spots ns which affects the total spot-settling time, iii) the product of ns and spot spacing corresponding to the time spent to adjust the scanning magnet between spots, and iv) the meterset, which corresponds to the actual irradiation time. The first three terms scale with layer repainting. A simplified one-term model was also determined from the dominant term among these four. Both models were tested across a variety of fields from different patients and disease sites and compared with delivery logs.

RESULTS: The fluctuation of layer time between repetitive deliveries was ∼0.1 s. The four-term model successfully predicted the layer time to subsecond accuracy. The one-term model had slightly higher uncertainty. The four-term model was implemented in our treatment planning system (TPS) as a one-click solution. The one-term model led to a derived geometric model which allows prediction of average layer time before a plan is generated.

CONCLUSIONS: A series of models were established to predict layer time. The spectrum of compromise between complexity and accuracy enables efficient implementation based on the functions readily available with particular TPS's.

Abstracts 53 and 54 are Poster Abstracts

## Abstract 55: Physics and Biology

### Dosimetric Performance of Pencil-Beam Scanning With a Travelling, Single-Spot MLC

#### James Cooley, Daniel Catanzano, Kai Huang, Sylvester Nyamane, Townsend Zwart, Mark Jones

##### Mevion Medical Systems

The Adaptive Aperture is a mechanically actuated, travelling multi-leaf collimator designed to improve the lateral penumbra of pencil-beam-scanned proton therapy treatments. The device shapes only a single spot or a small number of spots at one time, but moves with the scanned beam in order to collimate any or all sides of a spot at any location in a treatment field. As a result, the device has a number of unique capabilities that other collimation technologies lack. It can dosimetrically replicate the effect of machined static apertures and blocks and deliver treatment plans that prescribe patient-specific static apertures without requiring the fabrication of such apertures. It can also trim different energy layers with different aperture shapes, and trim with aperture shapes, such as “island” blocks, not otherwise achievable. We present beam data demonstrating the unique capabilities of a pencil beam scanning system with the Adaptive Aperture. Measured dose distributions comparing the performance of an Adaptive Aperture-trimmed treatment plan with that from the same plan delivered to a machined static aperture will demonstrate the device's ability to replicate the dosimetric effect of a brass aperture. We also present measured dose distributions delivered with unique aperture geometries, such as island blocks and layer-dependent aperture shapes .

Abstracts 55, 56, 57, 58 and 59 are Poster Abstracts

## Abstract 60: Lung and Mediastinum

### Considerations for Further Pulmonary Toxicity Reduction With 4DCT Derived Functional Avoidance Based Intensity Modulated Proton Therapy

#### Michele Dougherty, Yevgeniy Vinogradskiy, Edward Castillo, Austin Faught, Craig Stevens, Inga Grills, Thomas Guerrero

##### Beaumont Health System, University of Colorado

PURPOSE: With the advent of 4DCT based functional image-guided radiotherapy for lung cancer, CT ventilation images have been translated into clinical usage for functional avoidance radiotherapy. The purpose of this study is to evaluate the potential dosimetric gains for functional avoidance with Intensity Modulated Proton Therapy (IMPT).

METHODS: Functional ventilation images were created using a prior published algorithm based on deformable registration and underlying CT density characteristics. 4DCT data from patients enrolled in a prospective functional avoidance clinical trial are re-planned with multi-field robustness optimized IMPT (3.5% and 5mm) and compared with VMAT. Dose degradation due to proton spot arrangement and breathing motion was studied by evaluating dynamic dose accumulation based on 4DCT with an in-house built interplay Python script. Patient tumor excursion was estimated by both weekly 4DCBCT and simulation 4DCT. Dosimetric parameters were compared between IMPT and VMAT. Both pencil beam and MC dose algorithms were evaluated. Dose repainting (layer/volume) method was applied to increase target coverage on the IMPT plans and decrease the motion interplay effect.

RESULTS: The CTV target coverage was adequate for all IMPT plans (D95>60Gy). Compared with VMAT, IMPT plans have a relative dose reduction to most of the OARs (P<0.05), including functional mean lung dose (-4.4Gy), fV20 (-5.1%), and fV5 (-39.7%). Compared with pencil beam method, MC engine has a significant regional dose difference of 28% of the total prescription.

CONCLUSION: We perform one of the first studies to investigate IMPT for functional avoidance thoracic RT. We found that IMPT was able to further spare healthy functional lung tissue with repainting technique to avoid excess dose to normal structures while maintaining satisfying target coverage. Studies have shown that the risk of pulmonary complications can be reduced with functional radiotherapy; our data demonstrate that even further toxicity reductions are possible with IMPT.

## Abstract 61: Physics Range and Uncertainty

### A Novel and Fast Method for Proton Range Verification Using a Step Wedge and 2D Scintillator

#### Jiajian Shen, Bryce Allred, Daniel Robertson, Wei Liu, Terence Sio, Nicholas Remmes, Sameer Keole, Martin Bues

##### Mayo Clinic – Scottsdale

PURPOSE: To implement and evaluate a novel and fast method for proton range verification by using a planar scintillator and step wedge.

METHODS: A homogenous proton pencil beam plan with 35 energies was designed and delivered to a 2D flat scintillator with a step wedge. The measurement was repeated 15 times (three different days, 5 times per day). The scintillator image was smoothed, the Bragg peak and distal fall off regions were fitted by an analytical equation, and the proton range was calculated using simple trigonometry. The accuracy of this method was verified by comparing the measured ranges to those obtained using an ionization chamber and a scanning water tank, the gold standard. The reproducibility was evaluated by comparing the ranges over 15 repeated measurements. The sensitivity was evaluated by delivering to same beam to the system with a film inserted under the wedge.

RESULTS: The range accuracy of all 35 proton energies measured over three days was within 0.2mm. The reproducibility in 15 repeated measurements for all 35 proton ranges was ±0.045mm. The sensitivity to range variation is 0.1mm for the worst case. This efficient procedure permits measurement of 35 proton ranges in less than 3 minutes. The automated data processing produces results immediately. The setup of this system took less than 5 minutes. The time saving by this new method is about two orders of magnitude when compared with the time for water tank range measurements.

CONCLUSIONS: A novel method using a scintillator with a step wedge to measure the proton range was implemented and evaluated. This novel method is fast and sensitive, and the proton range measured by this method was accurate and highly reproducible.

## Abstract 62: Physics and Biology

### A Model for Secondary Monitor Unit Calculations of Pencil Beam Scanning Proton Therapy Treatment Plans

#### Greg Schimke, Terry Wu, Joseph Syh

##### Willis-Knighton Cancer Center

PURPOSE: To develop and implement a model by which secondary monitor unit calculations may be performed for pencil beam scanning, single-field uniform dose proton therapy treatment plans.

METHODS: Treatment planning system (TPS) simulations were performed using the local beam model to define relationships between planning target volume (PTV) characteristics and the monitor units (MU) required to meet a given prescription for defined beam orientations and a set of standard single-field uniform dose (SFUD) optimization goals. Relevant attributes include volume, depth (i.e. beam range), surface area, and the projected area of the target volume in the beam's eye view (BEV). The proposed model approximates the PTV as a simplified box ROI as defined by its volume and BEV projected area. Output factors (cGy/MU) were tabulated for this idealized geometry through TPS simulations using ROIs with a range of dimensions expected to be seen clinically. Correction factors were then applied which account for differences in radiological depth and surface area between the PTV and the idealized conditions, and the calculated monitor units for each beam were scaled according to the assigned beam weighting. This method was applied to a number of treatment sites including prostate, lung, and brain, using clinical datasets and standard SFUD optimization goals.

RESULTS: Good agreement was observed between the proposed model and TPS prescribed monitor units for the investigated treatment sites. The total mean percent difference for the 28 plans was +2.01% ± 3.05%.

CONCLUSIONS: This work demonstrates the potential for acceptably accurate independent verification of monitor units in pencil beam scanning proton therapy for SFUD treatment plans with the proposed method. Required inputs are few and are accessible in the TPS, facilitating future clinical implementation. Further investigation will expand the current model to accommodate a broader range of potential optimization problems, treatment sites, and intensity modulated proton therapy treatment plans.

## Abstract 63: Physics and Biology

### Increased Relative Biological Effectiveness (RBE) of Proton Beam at Shallow Depths

#### Vadim Moskvin, Robert Stewart, Jonathan Farr

##### St. Jude Children's Research Hospital

PURPOSE: The concept of a spatially variant Relative Biological Effectiveness (vRBE) arises from fundamental considerations of the well-known and documented trend towards increasing levels of biological damage with increasing particle stopping power. In proton therapy, vRBE considerations strongly suggest that the bio-effective dose at the tip of and beyond the Bragg peak (BP) may be substantially larger than the accepted, spatially invariant clinical RBE of 1.1. In this study we present new findings on the impact of vRBE in spot scanning proton beam for tumors at shallow depths.

METHODS: The Monte Carlo transport code FLUKA was coupled with the Monte Carlo Damage Simulation (MCDS) to determine the RBE-weighted dose (RWD). Particle- and energy-specific information on an endpoint of DNA double strand break (DSB) induction from MCDS was integrated into the FLUKA code systems for protons and light ions. The tumor depths from depth and lateral spatial RBE for DSB induction of protons beams were computed across a range of therapeutic energies from 69.4 MeV to 221.3 MeV for anoxic conditions. A range shifter was utilized to cover tumors at shallow depths (below 4 cm).

RESULTS: The RBE at the pristine BP and the distal edge increases with the decrease in depth of the tumor location. The RBE for proton beam to cover a tumor at 1.5 cm depth reaches 1.73 at the BP, 1.49 for 4 cm and 1.35 for 7 cm depth while 1.25 for deep sited tumors. The RBE shows independence on the field lateral size. Dose average energy of the protons at the distal edge of the BP is about 3 MeV for shallow depths while for the deep situated tumors is of 8-10 MeV. The RBE difference for shallow and deep sited tumors is associated with increases of protons fraction from sequential (n,p) + (p,n) reactions with increasing initial beam energy. RBE differentiates notably proximal of BP at shallow depths.

CONCLUSION: The proton beam may have higher efficacy in treatment of tumors at shallow depths. Including variable RBE may be necessary to account for the higher efficacy of the proton beams for treatment of tumors at shallow depths.

## Abstract 65: Physics Range and Uncertainty

### Transducer Specifications for Thermoacoustic Range Verification

#### Sarah Patch, Robert Lindert, Tianyu Zhao, Daniel Hoff, Tyler Webb, Lee Sobotka

##### UW-Milwaukee, Washington University Medical Center

BACKGROUND, MOTIVATION AND OBJECTIVE: Thermoacoustic range verification offers the potential of online range verification and correlation with patient anatomy, if ultrasound imaging arrays can be customized with transducer elements sensitive to very low frequency and weak thermoacoustic emissions.

Amplitudes of thermoacoustic emissions depend upon proton beam intensity and pulse duration. Instantaneous deposition of 1 cGy in water induces 1 Pa pressure increase, which decays to mPa as it propagates to remote transducer locations. Superconducting synchrocyclotrons deliver O(10 cGy) over a period of 5-10 microseconds. We develop transducer specifications for pencil beam proton therapy accounting for proton pulse duration.

CONTRIBUTION/METHODS: The temporal envelope of the proton beam delivered by a Mevion S250 proton therapy system was measured at low beam current. The shape was nearly Gaussian, with FWHM=6 us. Thermoacoustic emissions were simulated as follows: TRIM Monte Carlo simulations of 90K protons assumed a water target, Gaussian lateral profile upon entry with σ=3.3 mm, and 230 MeV beam energy (Fig 1a). Custom scripts called 3D k-Wave software to propagate thermoacoustic emission pressures due to a Gaussian pulse duration of 6 us FWHM.

RESULTS/DISCUSSION: Results for a transducer located 5 cm distal to the treatment spot are presented. Thermoacoustic emissions are bandlimited below 150 kHz and have peak amplitude of 43 mPa/cGy assuming instantaneous proton delivery (Fig. 1b-c).

6 us delivery time reduces bandwidth to 100 kHz and amplitude to 28 mPa/cGy. Intense proton beams are expected to increase pulse duration to 10 us, reducing bandwidth to 85 kHz and thermoacoustic efficiency to only 19 mPa/cGy.

Transducer elements sensitive to very low frequencies are required for thermoacoustic range verification.

Additionally, they must be sensitive to pressure amplitudes below one Pascal, which is orders of magnitude weaker than pulse echoes detected by piezoelectric transducer arrays during standard ultrasound imaging. Therefore, we anticipate customizing ultrasound imaging arrays with additional receive elements, such as PVDF transducers or optical microphones.

## Abstract 67: Physics Range and Uncertainty

### Image Reconstruction with a Fast, Monolithic Proton Radiography System

#### Don DeJongh, Victor Rykalin, James Welsh, Mark Pankuch, Nicholas Karonis, Caesar Ordonez, Kirk Duffin, John Winans, George Coutrakon

##### ProtonVDA Inc, Loyola University, Northwestern Medicine Chicago Proton Center, Northern Illinois University

PURPOSE: Proton radiography enables proton range verification in addition to the anatomical alignment verification currently obtained with X-ray radiography. Design specifications require that a clinical system be simple, lightweight, easily scaled to large field sizes, operate at high speed to maximize patient throughput, and expose the patient to the minimum possible radiation dose for a given resolution. We are developing a system to produce images of proton stopping power by tracking individual protons before and after the patient and measuring the proton residual range after traversing the patient. Due to multiple scattering effects, each proton deviates randomly from its projected trajectory. To achieve optimal spatial resolution, an image reconstruction algorithm must fully exploit the individual three-dimensional proton position information.

METHODOLOGY: We have developed an iterative algorithm for radiography fully exploiting proton path information to produce projective radiographs with no blurring from multiple scattering. Simulations of our detector design, with and without multiple scattering effects included, determine the expected accuracy of our proton path reconstruction, and the impact on the spatial resolution of the reconstructed image. Tests of our detector components in proton beam demonstrate the performance needed to validate our simulations.

RESULTS: Protons typically scatter transversely from the projected path by 4 mm after 20 cm of water. Our simulations demonstrate path reconstruction of individual protons to better than 1 mm. Our iterative algorithm successfully produces images with 1 mm sharpness. The iterative process necessarily increases pixel noise compared to estimates neglecting multiple scattering and additional protons will be needed to achieve a given contrast. Operation of our detector components in a test beam demonstrates the required proton tracking resolution and efficiency to optimize resolution and dose to the patient.

CONCLUSIONS: A proton radiography system optimizing image sharpness and dose to the patient will individually track protons before and after the patient. An iterative algorithm produces images with spatial resolution given by the tracking accuracy, at the price of increased pixel noise. We are in the process of integrating the necessary components into a fully functional system.

## Abstract 68: Physics and Biology

### A Method to Deliver Energy Modulated Planar Proton Arc Therapy (EMPPAT)

#### Ulrich Langner, Katja Langen, Mingyao Zhu, Sina Mossahebi, John Eley, Warren D'Souza, Byongyong Yi

##### University of Maryland

PURPOSE: Even though arc therapy is regularly used in photon therapy, it is not commercially available in proton therapy. We developed a concept to deliver proton arcs in an effective and practical way and discuss its application and advantages through a proof-of-principle dosimetric planning study.

METHOD AND MATERIALS: The concept of proton arc therapy which uses a tertiary energy modulator has been suggested so that only a single energy is requested from the cyclotron per arc (one gantry revolution). A dose plane with 1-2 cm thickness perpendicular to the beam is assigned for each gantry angle, where each angle will be used as a control point. This plane sweeps the target volume while the gantry moves. Comparison plans were created between a two field single field optimization (SFO) proton plan, a 6x photon volumetric arc plan, and 24 SFO plans from different gantry angles each day, using a cylindrical phantom with 25 cm diameter. Plan quality were evaluated based on volumetric integral dose, robustness of the technique, and distribution of relative biological effectiveness (RBE) dose. Plans with PTVs of varying size and position were evaluated to investigate the advantages of this method compared to current methods.

RESULTS: Worst case GTV robustness D95% increased from 80.2%, to 90.8%, to 90.2% for the proton arc as the target diameter increased from 1cm, to 3cm, to 11cm. As the target was shifted from the surface to the center of the volume, the robustness did not change significantly. For the target closer to the surface the VMAT integral dose was similar to that of the proton plans, but the difference increased markedly as the depth of the target increased. With VMAT, almost the entire volume receives 10% of the dose, while the V10% is substantially less for all the proton techniques.

The integral dose was about a factor of 2 less for proton arcs compared to VMAT and even less than for the standard two field technique, although the V5% was less for the two-field technique the V20% was less for proton arc method, thus decreasing the integral. The mean RBE's of the 5cm diameter PTV volume were 1.11 and 1.26 for the single field SFO plan and for the arc plan, respectively. The peak of the RBE distribution is at the center of the PTV for the arc plan, while it is at the distal end of the PTV for the single beam plan.

CONCLUSION: This planning study shows that proton arc is a promising delivery method, even if it is perceived to give up the largest advantage of protons, i.e. no dose to a large volume. It gave less integral dose to a simulated volume than VMAT and even current proton treatment techniques. The method proposed here also provides significant advantages in RBE dose distribution.

## Abstract 69: Physics and Biology

### Irradiation Time Prediction during Treatment Planning for Pencil Beam Scanning

#### Markus Bach, Kirankumar Singani, Isabel Huth, Christina Marticke

##### Varian Medical Systems Particle Therapy

Precise irradiation time prediction for pencil beam scanning is a basic principle for clinical indications sensitive to organ motion like lung and liver cancer. On one side this would enable the user of a treatment planning system (so without irradiating the plan at the proton facility) to identify if the patient qualifies e.g. for breath-hold techniques. On the other side plans could be optimized for short irradiation times based on these timing values. Currently no commercial planning system considers temporal behavior of the beam application during calculation or optimization.

This work incorporated machine specific temporal dynamics for a clinical configuration of a proton beam facility. The accelerator control system, which optimizes on maximum beam-currents while maintaining system limits to guarantee accurate dose application, was replicated and communication with Eclipse treatment planning (v 13.7) was established by API scripting. Calling the time prediction script takes less than a second and displays irradiation details for individual energy layers.

Three types of motion sensitive targets (two different lung sites, large volume liver) were investigated, each with different time-sensitive parameters like energy steps, minimum MU value and spot spacing. The initial time prediction leads to deviations below 10% (ranging from 5.4% ... 8.9% for irradiation times of 19s and above). In a second step, ProBeam system log-files related to irradiation time were analyzed and compared to timing values generated by the calculation. The prediction could be refined afterwards so that the deviation between calculated times and irradiation for a selected lung plan was reduced to less than 2%, which is 0.4s. This remaining variation is close to the standard deviation of 0.5s for re-irradiating the same plan.

The integrated workflow achieved by the tool would enable dosimetrists to routinely check on irradiation times. The accuracy (including individual spot duration, spot meandering aspects, and energy layer switching) would allow precise 4D dose distribution calculations.

## Poster Abstracts Poster Abstract 6

### Assessment of Setup Errors on Dose-volume Parameters for Chest Wall and Regional Nodal Irradiation with Tissue Expander Using Double Scattered Proton Therapy

#### Xiaoying Liang, Julie Bradley, Dandan Zheng, Michael Rutenberg, Nancy Mendenhall, Zuofeng Li

##### University of Florida, University of Nebraska

PURPOSE: In treatment of chest wall and regional lymphatics in women with immediate expander reconstructions, the high density metal in the tissue expander port requires override with the corresponding relative stopping power. Understanding of the position of the metal port on a daily basis is needed to ensure adequate chest wall coverage.

METHODS: Misalignment of the metal port on daily treatment could occur due to: 1) setup uncertainty, 2) deformation of the reconstructed breast, which leads to a relative shift from the bony structure on the planning CT, and 3) systematic shift of the port from the day of CT simulation. Three patients with tissue expanders were selected for this study. Each patient was planned with 50 CGE in 25 fractions using proton double scattering technique. The aperture margin was 1cm and 6-7mm smearing margin was used in each beam. For daily setup, each patient was aligned by bony structures such as clavicle, humeral head, sternum, and vertebral bodies using orthogonal X-rays. The daily setup imaging from all 25 treatment sessions for each patient was reviewed and the mean misalignment of the metal was obtained. The plan was then retrospectively calculated by applying shifts to the isocenter. In addition, the dose effect to chest wall and axillary nodes of a systematic 5mm setup error in all three directions was investigated, to simulate up to 5mm setup errors due to the relative shift between the setup bony structure and chest wall anatomy or a shift of the tissue expander port.

RESULTS: The misalignment of the metal has minimal impact (<0.2Gy) on the chest wall coverage, <0.5Gy variation on the mean heart dose and < 3% change to the Ipsilateral lung V20. With a 5mm systematic setup error, the largest chest wall D95 coverage loss is 0.61Gy. The variation in D95 for all axillary nodes is within 1Gy for a 5mm systematic setup error.

CONCLUSION: This treatment planning technique has robustness that preserves target coverage with minimal variation when the tissue expander metal port shifts ≤5mm and when the relative shift between the setup bony structure and chest wall anatomy is ≤5mm.

## Poster Abstract 10:

### Evaluation of a Gold Fiducial Marker for Proton Radiotherapy of the Prostate

#### Kaifang Du, Chang Chang, Andrew Lee

##### Texas Center for Proton Therapy

PURPOSE: The purpose of this study is to investigate the use of a gold fiducial marker in setup radiographs, and its dosimetric impact on prostate cancer proton radiotherapy. The evaluated gold fiducial marker is designed to fold when pushed out and deployed using a needle. Its high atomic number and density result in high attention of kV X-ray and may improve visibility in image guidance.

METHODS: Two gold fiducials of different sizes (0.28x10 mm and 0.4x10 mm) were inserted into two hams (14.5x10x4.5 cm each), which were immersed into a water tank phantom in order to simulate its implantation. The transverse dimension of the water tank is 36.8x21.2 cm, similar in size to a male patient's pelvis. The fiducials were assessed for radiographic visibility using both CT and on-board KV imagers in a proton treatment room. Proton pencil-beam scanning treatment plans with a single lateral field were designed using the CT images. The gold fiducials were contoured and comparison was made between plans with and without the fiducial density overridden to gold. The dose distribution was optimized and computed using Monte Carlo dose engine.

RESULTS: The visibility of the fiducials was acceptable for clinical use. The presence of fiducials perturbed the dose distribution downstream along the beam direction. When the fiducials were overridden to gold, for the 0.28x10 mm fiducial, a dosimetric cold spot down to 82.2% of Rx-dose was observed 3.1 cm downstream, and max dose was 116.4% at 2.9 cm downstream; for the 0.4x10 mm fiducial, a dosimetric cold spot down to 78.9% of Rx-dose was observed 1.7 cm downstream, and max dose was 121.5% at 0.9 cm downstream. When the fiducials were not overridden, no obvious dose perturbations were observed immediately downstream.

CONCLUSION: These gold fiducials provide good radiograph visibility, but cause dose heterogeneity to the absorbed dose in proton therapy. Multi-beam proton PBS plans are required in order to ensure dose uniformity within the target. The observed dosimetric perturbations suggest it may be better to use these markers in an uncoiled fashion for protons.

## Poster Abstract 19:

### Prospective Phase II Study of Scanned Proton Beam Radiotherapy for Spine Chordomas and Chondrosarcomas

#### Thomas DeLaney, Yen-Lin Chen, Karen De Amorim Bernstein, Beow Yeap, Matthew Dimaria, Nannette Thomas, Susan McGovern, John Mullen, Joseph Schwab, Francis Hornicek

##### Francis H. Burr Proton Therapy Center, Massachusetts General Hospital/Harvard Medical School, MD Anderson Cancer Center

BACKGROUND: Negative surgical margins are uncommon for spine sarcomas; hence, adjuvant or neoadjuvant radiotherapy (RT) may be important for tumor control. Because the required radiation target dose is often substantially higher than the radiation tolerance of adjacent non-target spinal cord, nerve, and viscera, we employed passively scattered protons in a completed prospective phase II study which accrued between December 1997 to March 2005 to effectively deliver the desired radiation dose to the target while sparing normal tissues. We hypothesized that scanned proton beam might allow higher target conformality with improved normal tissue sparing. After conducting a dosimetry planning study that confirmed this hypothesis, we designed and completed a prospective clinical trial to test this hypothesis.

METHODS: Prospective Phase II clinical trial incorporating high dose scanned proton beams. Eligibility: primary or locally recurrent cervical, thoracic, lumbar, and/or sacral spine chordomas or chondrosarcomas. Treatment: pre- and/or post-operative photon/proton RT+/- resection. En bloc resections were preferred; some resections had a planned microscopic positive margin and most primary tumors a non-contaminated or negative but not wide surgical margin.

RESULTS: Sixty patients (51 chordoma, 9 chondrosarcoma) were accrued from January 2013 to March 2017. Patients underwent gross total (43) or subtotal (6) resection or biopsy only (11). Target RT dose was initially 70.2 GyRBE/39 fx for R0/R1 patients (reduced in the later part of the study to 68.4 and later 64.8 GyRBE in patients who underwent core biopsy, preop RT, and en bloc resection with negative margins). 77.4 GyRBE/43 fx used for R2 resection or biopsy only. 19.8 Gy IMRT allowed to avoid scheduling delays when scanned proton beam slots were limited.

All patients received scanned proton beam as planned, although some patients with extensive spine stabilization hardware received a photon component to ameliorate hardware proton scatter. No myelopathies were seen. No unusual radiation related toxicities were noted. Scanned proton plans provided improved skin and subcutaneous sparing. Disease control information with median follow-up of 2 years will be presented.

CONCLUSIONS: High dose scanned proton beam radiation can be given for spine chordomas and chondrosarcomas. It improves target conformality and appears to reduce skin and subcutaneous toxicity.

## Poster Abstract 22:

### Minimization of Dose to Critical Structures with Dynamically Collimated Scanned Proton Beams - A Treatment Planning Study

#### Peter Ahn, Hao Chen, Keith Unger, Jonathan Lischalk, Dalong Pang

##### MedStar Georgetown University Hospital

Cancers of the brain, skull base, liver and prostate have been treated with proton therapy for over 30 years due to the proximity of numerous critical structures. For the brain and skull base, the brainstem, optic structures, cochlea, and normal brain create challenges in delivering a curative dose. Liver and prostate treatments create challenges due to the need to avoid treating the normal liver and rectum, respectively. These sites present good cases to test advances in radiation therapy delivery techniques. While passively scattered protons with apertures allow higher conformality of dose beyond those obtainable with conformal photons, proton pencil beam scanning (PBS) further reduces integral dose. The higher conformity of PBS compared to passively scattered protons comes at the cost of larger penumbra and thus lower conformality at the lateral edge compared to passive scatter or even conventional radiation. HYPERSCAN is the latest advance in PBS and it includes a dynamic multi-leaf collimator (dMLC), called Adaptive Aperture (AA),that can collimate any spot to optimize gradients and spare adjacent tissue. With the installation of the first HYPERSCAN PBS system incorporating the AA dMLC, we have performed comparative treatment planning studies using our commissioning beam data comparing our system with another modern conventional PBS system. Dose comparisons to target and critical structures are presented. Results indicate that significant reduction of dose to adjacent structures can be obtained with the use of carefully designed spot collimation (field-edge or block) over scanned small spots techniques alone. The doses to the target structures are essentially equivalent for both techniques. It is concluded that the use of dynamic spot collimation equals or surpasses the results of small spot treatment delivery. Therefore, combining the best of both passive scattering with its sharper lateral penumbra and the increased conformality of PBS by using the dMLC system, affords plans offering the best balance of target coverage and critical organ avoidance.

## Poster Abstract 23:

### Outcomes of Atypical Teratoid Rhabdoid Tumor (ATRT) Treated with Proton Therapy

#### Sonal Mayekar, William Hartsell, Carlos Vargas, Gary Larson, Henry Tsai, George Laramore, John Chang

##### Rush University Medical Center, Northwestern Medicine Chicago Proton Center, Mayo Clinic – Scottsdale, ProCure Proton Therapy Center – Oklahoma City and New Jersey, University of Washington

Introduction: ATRT are highly malignant tumors of the central nervous system diagnosed predominately in young children. Treatment of these tumors involves multimodality treatment with surgery, chemotherapy and radiation. Patients with this tumor have historically had a poor survival rate (< 30%) when approached similar to PNETs. The Dana Farber Cancer Institute (DFCI) ATRT intensive chemotherapy regimen has started to shift the pendulum in a more positive direction. We evaluated the use of the DFCI regimen with proton radiation therapy (PRT).

Methods: We identified and retrospectively analyzed 14 ATRT patients treated with the DFCI regimen and prospectively enrolled in a nationwide registry (PCG RGE 0119). The patients were treated with proton therapy between March 2010 and November 2016. We calculated the median survival. Disease extent (solitary vs multiple/diffuse), location and surgical techniques were reviewed. Radiation dose, fraction and toxicities were also analyzed.

Results: Median age at time of treatment was 3.7 years (9 months - 15.5 years). Ten patients had solitary lesions and four patients had multiple lesions. Twelve of the 14 patients underwent surgical resection prior to radiation. Nine patients underwent a gross total resection and three patients underwent subtotal resection/debulking. Five tumors were located in the cerebrum, three in the cerebellum, three in the ventricular system, two in the spinal cord and one in the brainstem. The median radiation dose of 47.8 Gy(RBE) (range 14.43 – 54.13) at 1.8 Gy(RBE) per fraction. Eight patients were treated with concurrent chemotherapy. With a median follow-up of 9.5 months (range 1-70 months), seven of the 14 patients were alive. Median survival following radiation was 24.5 months (range 1-70 months). The most common adverse effects were radiation dermatitis (10 of 14 patients), alopecia (10 of 14 patients) and nausea (7 of 14 patients). There were eight grade 3 toxicities (anorexia, nausea and thrombocytopenia). There were no grade 4 or 5 toxicities.

Conclusions: Trimodality with the incorporation of proton therapy is a safe and effective treatment for ATRT. Further follow-up is needed to evaluate for long-term survival and quality of life in this patient cohort.

## Poster Abstract 25:

### Early Outcomes in Pelvic and Lumbosacral Spine Ewing Sarcoma Treated With Proton Therapy

#### Julie Bradley, Daniel Indelicato, Christopher Morris, Ronica Nanda, Natalie Logie, Eric Sandler, Ronny Rotondo

##### University of Florida

PURPOSE: The use of proton therapy for treatment of Ewing sarcoma has been increasing over time, but clinical data remain limited. Pelvic tumors present at a large size and have inferior local control compared to Ewing sarcomas of other sites. This study reports 2-year disease control and toxicity after treatment with proton therapy.

METHODS: Patients aged ≤ 30 years with Ewing sarcoma of the pelvis or lumbosacral spine treated with proton therapy were identified from a prospective database. From 2010 to 2016, 36 patients (median age, 13.7 years; range, 3.4-26.9 years) received a median dose of 55.8 Gy (range, 45-64.8) for definitive (78%), preoperative (5%), or postoperative (17%) radiotherapy. All patients received chemotherapy. Fourteen percent had limited metastatic disease at diagnosis and underwent definitive local therapy to the primary pelvic site and the metastatic disease (lung metastases treated with whole lung irradiation, n=4; solitary bone metastasis, n=1). Median primary tumor size was 10 cm (range, 4-20 cm). The Kaplan-Meier method was used to calculate survival and control rates. A log-rank test statistic assessed significance between strata of prognostic factors.

RESULTS: With a median follow-up of 1.5 years (range, 0.1 – 6.9 years), 2-year overall survival (OS) was 81% and 2-year progression-free survival (PFS) was 65%. 2-year local control was 85%. Distant metastases-free survival at 2 years was 75%. There was no association between OS or PFS and tumor size, patient age, radiation dose, or definitive vs. pre-/postoperative radiotherapy. Patients with M1 disease at diagnosis had worse 2-year PFS compared to those with M0 (25% vs. 75%, p = 0.03). Median time to recurrence was 0.6 years (range, 0.5-1.9 years). Serious toxicity included sacral insufficiency fracture (n=2), limb length discrepancy (n=1), and stage 3 chronic kidney disease (n=1). No grade 4 or 5 toxicities developed.

CONCLUSION: Proton therapy does not compromise early disease control in this cohort of patients with pelvic or lumbosacral Ewing sarcoma. Compared to previous institutional data on patients treated with photons, serious toxicity is lower with protons but longer follow-up is necessary to fully characterize the therapeutic ratio in this setting.

## Poster Abstract 27:

### Clinical Benefits of Spot-scanning Irradiation for Large Field: A Report of Three Pediatric Cases Treated by Whole Craniospinal IMPT

#### Takayuki Hashimoto, Hiroyuki Kobayashi, Akihiro Iguchi, Takashi Mori, Seishin Takao, Taeko Matsuura, Rikiya Onimaru, Shunsuke Terasaka, Shinichi Shimizu, Hiroki Shirato

##### Hokkaido University

Proton therapy provides dosimetric benefits in sparing normal tissue for pediatric patients. We report three pediatric cases treated by craniospinal intensity-modulated proton therapy (IMPT) and describe the clinical benefits of spot-scanning irradiation for large field. CASE 1: A 9-year-old girl was diagnosed as high-risk medulloblastoma. She received whole craniospinal IMPT to 36.0 GyE and tumor bed boost to 55.8 GyE concurrently using carboplatin and vincristine. Treatment time could be shortened by using only one posterior field, which was large enough to cover whole spine. Although Gr 4 neutropenia was observed during and after IMPT, scheduled chemotherapy continued using G-CSF without recurrence at the time of 5 months after IMPT. CASE 2: A 16-year-old boy was diagnosed yolk sac tumor of the pineal region with intraventricular dissemination. He received craniospinal IMPT to 30.6 GyE by sparing the anterior vertebral body to prevent severe myelosuppression and whole ventricular boost to 61.2 GyE concurrently using ifosfamide, cisplatin, and etoposide without interruption. Two posterior fields were large enough to cover whole spine. Although Gr 3 febrile neutropenia was observed during remaining cycles of chemotherapy, neutrophil counts rapidly recovered by G-CSF without delay in treatment schedule. No other Gr 3 or more late toxicity has been observed 6 months after IMPT with local control. CASE 3: A 17-year-old girl was diagnosed as GCT of poor prognosis group. She received whole craniospinal IMPT to 30.6 GyE by sparing the anterior vertebral body to preserve myelopoietic function and whole ventricular boost to 61.2 GyE after one course of chemotherapy with ifosfamide, carboplatin, and etoposide. Two posterior fields covered whole spine sufficiently. No Gr 2 or more myelosuppression was observed during IMPT. Although Gr 4 neutropenia was observed during remaining cycles of chemotherapy, scheduled therapy continued using G-CSF without prolonged severe myelosuppression. No other Gr 3 or more late toxicity has been observed 4 months after IMPT without recurrence.

CONCLUSION: Using large posterior fields for whole spine was helpful to shorten the treatment time for craniospinal IMPT. Especially in adolescent cases, anterior vertebral body sparing in craniospinal IMPT was useful to prevent severe bone marrow toxicity.

## Poster Abstract 36:

### Dosimetric Comparison of Intensity Modulated Proton Therapy (IMPT) Versus Intensity Modulated Photon Radiation Therapy (IMRT) Using Volumetric Arc Therapy (VMAT) and Helical Radiation Therapy (HRT) for Salvage Radiation Post Prostatectomy

#### John Chang, Patrick Lee, Tracy Yates, Heather Parks, Shae Gans, Susan Niles, Trinh Mai, Amanda Duggal, Darren Kaplan, Linnae Campbell, Stacey Schmidt, William Hartsell

##### Northwestern Medicine Chicago Proton Center, Northwestern Medicine Central DuPage Hospital Cancer Center, Northwestern Medicine Delnor Cancer Center

PURPOSE: Evaluation of the dosimetric differences of an IMPT plan relative to 2 types of IMRT for target and normal tissue organs at risk (OAR).

MATERIALS AND METHODS: We compiled the treatment planning for IMPT and then performed VMAT and HRT planning on 15 post radical prostatectomy patients treated at our institution with IMPT to 70.2 Gy(RBE) to the prostate and seminal vesicle surgical bed. Daily prostate immobilization was achieved by utilizing a rectal balloon inflated to a customized amount based upon individual anatomy. The RTOG contouring atlas was utilized for target and OAR delineation. Contours were performed by multiple physicians but reviewed and adjusted by one physician reviewer. Planning target volume (PTV) expansions on the clinical target volume (CTV) were per our institutional guidelines for IMPT and IMRT, respectively. Goals of treatment planned on listed on Table 1. Dose volume histograms (DVH) were compared between IMPT and IMRT plans with a one-tailed T-Test. The differences were considered statistically significant with a p-value of 0.001 given the repetitive analysis of OARs.

RESULTS: The 98% PTV coverage (table 1) was not statistically significant in any of the three modalities, although the mean dose to 95% of the PTV was a bit higher in the IMRT plans [70.6-70.8 Gy(RBE)] over the IMPT plan [70.2 Gy(RBE)]. There was also no difference in the penile bulb, femoral heads, or bowel cavity DVHs. Nearly all the bladder and rectal along with sigmoid colon DVHs are improved with IMPT over IMRT (table 1).

CONCLUSION: IMRT seems to have a dosimetric advantage over IMRT in several categories for the bowel and bladder. There is also the added advantage of lower integral radiation dose. Data are forthcoming regarding whether this will translate into a true clinical advantage.

## Poster Abstract 42:

### Dosimetric Comparison of Pencil Beam Scanning (PBS) Intensity Modulated Proton Therapy (IMPT) and Volumetric Modulated Arc Therapy (VMAT) for Locally Advanced Pancreas Cancer (LAPC)

#### Krishan Jethwa, Broc Giffey, Bret Kazemba, Erik Tryggestad, Michael Haddock, Michelle Neben-Wittich, Kenneth Merrell, Thomas Whitaker, Christopher Hallemeier

##### Mayo Clinic – Rochester

PURPOSE: For LAPC, PBS-IMPT may allow for reductions in dose to organs at risk (OARs) compared to VMAT, while maintaining clinical target volume (CTV) coverage. The purpose of this study was to perform a dosimetric comparison between PBS-IMPT and VMAT in the treatment of LAPC.

METHODS: Between July 2016 and March 2017, 13 patients received PBS-IMPT for LAPC. All patients received a dose of 45 Gy (RBE 1.1) in 25 fractions to the gross pancreas tumor with an elective expansion encompassing the regional lymphatics (CTV45), and a simultaneous integrated boost to 50 Gy to the gross pancreas tumor plus ∼ 1 cm margin (CTV50). Plans were generated on the CT-average of the 4DCT. The typical beam arrangement included a posterior and posterior oblique field. Motion mitigation strategy was repainting for tumor motion < 1 cm and gating for ≥ 1 cm. A VMAT plan was generated for each patient for dosimetric comparison. Mean dosimetric parameters for IMPT and VMAT plans were compared using matched pairs t-test.

RESULTS: There were no differences in mean CTV coverage between IMPT and VMAT (p=ns). IMPT resulted in statistically significant reductions in dose to OARs, including small bowel, duodenum, stomach, large bowel, liver, and kidney (Table 1).

CONCLUSION: Relative to VMAT, IMPT offers a statistically significant reduction in dose to OARs, particularly the mean dose and organ volume receiving low to intermediate doses. Further clinical studies are needed to evaluate if these dosimetric advantages translate into clinically meaningful benefits.

## Poster Abstract 46:

### Development of a Geant4 Monte Carlo Platform for a Pencil Beam Scanning Proton Therapy System

#### Yingzi Liu, Kuan Ling Chen, Hsin Shun Wu

##### Willis-Knighton Cancer Center

Comparing to the analytical pencil beam algorithm utilized in current treatment planning system (TPS), Monte Carlo (MC) methods are more accurate, especially with the present of high heterogeneous interface or artificial materials. This study aims to implement a Geant4 MC dose retrospective check system for the TPS of our pencil beam scanning system. The proton beams modeled in the MC code was first validated by comparing the simulated with the experimental integrated depth-dose collected using a Bragg peak chamber and a multilayer ionization chamber. The validations were done for 17 therapeutic energies ranging from 70 to 226.7 MeV. Several different ranges Spread-Out Bragg Peak with 10 cm modulation were also compared between MC simulation and measurements. The validated energy-dependent beam characteristics, including energy spread, spot size, and beam angular divergence, were then coded into a Python script to facilitate data conversion from the TPS to Geant4 source defining file. To compute delivered dose in patient with Geant4 MC code, the beam sets created by the TPS were imported into Geant4 along with patient's CT images. A realistic voxel model of the patient was created by converting the CT Hounsfield Units to mass density values, and the corresponding biological tissues were then assigned and used in simulations. A high performance supercomputer was used for computation to speed up the calculation process. This developed MC system can be used to calculate dose distribution of tumor presented in highly heterogeneous situation, or to estimate the dose disturbance caused by fiducials placed in the patient body near the tumor, or in any condition that TPS tends to perform poorly.

## Poster Abstract 47:

### Evaluation of Proton Dose Distribution between Analytical Pencil Beam Algorithm and Monte Carlo Algorithm

#### Limin Song, Lea Gentry Swan, Shikui Tang, Chang Chang

##### Texas Center for Proton Therapy

PURPOSE: To compare dose distributions for spot scanning intensity modulated proton therapy (IMPT) calculated using analytical pencil beam algorithm and Monte Carlo algorithm for treatment sites with different heterogeneity complexity.

METHODS: Raystation v6.0 has implemented both pencil beam (PB) and Monte Carlo (MC) algorithms for IMPT dose calculation. Total 50 patients from 2015 to 2017 treated at a single institution were selected for comparison, 10 for each of the five treatment sites [prostate, lung, brain, head and neck, and breast]. All clinical plans were created using PB algorithm with robust optimization. The same plans were then forward calculated with MC algorithm with 1% uncertainty in dose accuracy. Differences in target coverage were assessed using percentage of prescription dose covering 98% of the CTV. Dmean and Dmax values were compared for organs at risk (OARs).

RESULTS: PB algorithm overestimates target coverage for all treatment sites when compared with MC algorithm. The mean dose overestimate of CTV D98% for prostate, brain, lung, breast, and HN are 2%, 4%, 7%, 9%, and 11%, respectively. Despite the relative homogeneous mass distribution for breast cases, the mean coverage difference for breast plans is greater than that of lung plans. The CTV D98% difference between the two algorithms has the least variation for prostate cases (<1% around the mean), and much larger variation for lung cases (1%-14%). In the HN cases, the corresponding variation spans from 8% to 18%, and the max cord dose is underestimated using PB algorithm in more than 60% cases up to 7 Gy.

CONCLUSIONS: Compared with MC-based dose calculation engine, PB algorithm likely overestimates target coverage and underestimates doses to OARs. The prostate and brain plans are least affected by the algorithms and the lung plans show large differences in CTV coverage as generally expected. The most significant differences in target coverage are seen in the HN and breast cases, which may be contributed from the large air gaps and shallow depths with the use of range shifters. These cases need to be planned with caution when using PB algorithm.

## Poster Abstract 53:

### The Role of Heavy Ions in Neutron Therapy

#### Kozak Oksana, Anatolij Shvedov, Olexandr Morus

##### Kiev Region Oncology Hospital, Institute of Nuclear Research, Support of Nuclear Research Organization

The use of fast neutrons in oncology has a number of features that differ very much from gamma-therapy. They were checked experimentally and in clinic, and sometimes the outcome after neutron irradiation was unpredictable. Early and late reactions from the side of tumor and normal tissue were sudden.

The sequence of processes occurring during the passage of neutrons through the organic tissue is not completely understood and poorly studied. And the main question was not solved yet: What causes such a strong destruction of cells in passing of fast neutrons that DNA breaks cannot be repaired and the destructions cannot be modified?

The aim of this work was to formulate the basic concept of the model for correct assessing of radiobiological effectiveness in irradiation of organic tissue by neutrons.

The analysis of data obtained on cyclotron U-120 was carried out. The cross sections of possible reactions were systemized. It was shown that heavy ions and not recoil protons make the main contribution to the radiobiological effect in irradiating of organic tissue with neutrons.

Nitrogen nucleus takes a special place in process of cell destruction. The levels of excitation of 14N start about 2.3 MeV and the density of excitation levels is much higher than that for other elements, for example, 16O or 12C. Moreover probability of neutron interaction with nitrogen nucleus is much higher than with hydrogen nucleus for the energy of about 6 MeV.

When 14N absorbs a neutron, it is converted to 15N with energy of about 15 MeV, which leaves behind the ruined pieces of cells while moving. Recoil protons are absorbed in the very place of origin. They cannot seriously hurt the cell creating something like a “swarm of flies.”

Just heavy ions play a major role in the destruction of cells after neutron irradiation.

It is necessary to differentiate the processes occurring in the tissue during irradiation.

It seems that effect of irradiation is not uniquely determined by the absorbed dose (as integral value).

## Poster Abstract 54:

### Reference Field Characterization of an Intensity Modulated Proton Therapy System

#### Townsend Zwart, James Cooley, Dan Catazano, Kai Huang, Sylvester Nyamane, Mark Jones, Miles Wagner, Stanley Rosenthal, Michael Tajima, Armin Langenegger

##### Mevion Medical Systems

A radiotherapy system delivers intensity modulated proton therapy fields at water equivalent depths between 1 and 32 cm. Useful measures of the quality of this system include on axis and lateral dosimetric profiles. We present measured field profiles of shallow, moderate and deep one liter reference fields. These fields are characterized with and without the use of an Adaptive Aperture which provides enhancement of the lateral penumbra and flexible blocking capability within a field. Also presented are benchmark data from comparable IMPT systems which rely exclusively on spot size to achieve lateral dose conformality and dose gradients.

## Poster Abstract 56:

### Acceptance and Commissioning of the First Compact PBS Proton Therapy System Using Superconducting Synchrocyclotron in North America

#### Xuanfeng Ding, Xiaoqiang Li, Jingjing Dougherty, Di Yan, John Potts, Thomas Lanni, Lisa Benedetti, Craig Stevens, Peyman Kabolizdeh

##### Beaumont Health, IBA

PURPOSE: To introduce a 7-week acceptance and commissioning procedure along with defining the machine characteristics a superconducting synchrocyclotron accelerator (S2C2) at the Beaumont Proton Therapy Center (PTC).

METHODS: A 7-week acceptance and commissioning schedule was designed and carried out to efficiently use the proton beam irradiation time in parallel with TPS and OIS software integration and configuration and clinical staff training as well as undergoing an IROC TLD independent check. The timeline of such extensive procedure is as follows. Week 1: Beam acceptance testing and data collection including evaluation of (i) integral depth dose; (ii) absolute dose calibration; and (iii) spot profile. Week 2: including imaging, mechanical, isocentric and safety quality acceptance testing while the beam modeling was in progress. Week 3: RayStation beam model validation which includes evaluation of (i) range; (ii) absolute dose; (iii) field size effect; (iv) spot profile/position and (v) IROC TLD irradiation; Week 4: (i) Integration testing and configuration of the proton beam system in RayStation and Mosiaq (ii) couch and immobilization device modeling and range validation. Week 5-7: (i) Therapists' training (ii) end-to-end testing with different treatment sites and techniques using film and a RANDO phantom; (iii) clinical workflow integration with current photon site; (iv) machine QA protocol design e.g. daily, weekly, monthly.

RESULT: Beaumont ProteusONE (S2C2) has spot sigma at isocenter from 3.4mm (227.7MeV) to 8.1mm (70MeV). Spot symmetry of 1% was measured across all gantry angles and X-Y direction. Spot position accuracy is within 1mm across the max treatment field. Proton range pull-pack accuracy is within 0.5mm from the expected value. IROC TLD analysis showed dose accuracy within 1% of the planned dose.

CONCLUSIONS: We report the safe, accurate implementation of a PBS proton delivery system over a 7-week period of integrated acceptance testing and commissioning. This required an extensive preplanning and close cooperation and support from all vendors with some double shifts and weekends work from clinical physics and engineer team. Week 5-7 could be used as buffer time for clinical training and integration test.

## Poster Abstract 57:

### Incorporating a Compact Proton Therapy System Into the Heart of an Existing NCI Designated Comprehensive Cancer Center: Update on the First Year of Operations

#### David Mansur, Brenda Myers, Frederick Jesseph, Mitchell Machtay, Chee-Wai Cheng

##### University Hospitals, Seidman Cancer Center, Case Western Reserve University

INTRODUCTION: A compact system was chosen in order to optimally integrate proton therapy into our crowded urban medical center with 1,000-bed capacity including a children's hospital and adult cancer hospital.

METHODS: We have reviewed our operations as we approach our one-year anniversary. Data from first patient through June 13, 2017, was reviewed.

RESULTS: Groundbreaking was September 20, 2013, and first patient was treated July 26, 2016. Over this time period, 96 patients have completed proton therapy. All but 5 patients were from in-state. The treatment ramp-up began with 2 patients the first week. Average number of patients on treatment by month varied from 4.9 in August 2016 to 14.9 in June 2017 with a maximum of 16.4 in December 2016. This included 3 craniospinal treatments, the first of which was treated within two months of opening. At the time of abstract submission (week 49 of operations) 23 patients were on treatment (Figure 1). The most common disease sites treated were pediatric/AYA 27.5% and adult CNS 27.5%. Prostate cancer comprised 17.5% of patients. Payer mix included 39% commercial, 36% Medicare, 24% Medicaid, and 2% other. Of all patients for whom their attending physician considered proton therapy the treatment of choice, 15% were denied by insurance after appeals were exhausted. Denial rate by payer included Medicare/Medicaid 0%, Health Design Plus 20%, Anthem 31%, Aetna 31%, United Healthcare 40%, and one local state carrier 44%. Denials by disease site varied from a low of 0% for pediatric to a high of 50% for prostate.

CONCLUSIONS: Time from groundbreaking to first patient was relatively rapid at 2 years and 10 months. Patient ramp for our center was steady, reaching near capacity in approximately 1 year. Preliminary results indicate favorable approval rates by Medicare/Medicaid and most commercial payers in our region.

## Poster Abstract 58:

### Optical Patient Setup Tracking: Using Dynamic Photographic Images to Reduce kV Imaging and Improve Soft-tissue Alignment

#### Brad Kreydick, Leah Chung, Katelyn Zastrow, Mark Artz, Blake Macnair, Andries Schreuder, William Hartsell, Mark Pankuch

##### Northwestern Medicine Chicago Proton Center, Provision Cares Proton Therapy Center

PURPOSE: To reduce patient imaging dose, setup time, and possible soft-tissue misalignment using a beta version of the Patient Tracker PCPT, a MacOS application that compares live photographic images to reference images captured during simulation.

METHODS: A rigid phantom was used to measure the “return-to-reference” capability of the Patient Tracker PCPT system using a robotic couch with six degrees of freedom (d.o.f). After using the system to capture reference images, the phantom and the treatment table are displaced by a random six d.o.f vector to a new position, then returned to the reference position using only the visual aids provided by Patient Tracker PCPT. The discrepancy between the returned and reference positions was measured with an NDI Polaris Spectra infrared camera.

RESULTS: An average of 1.05 mm return-to-reference translation and 0.25 degrees rotation was achieved in initial trials. Geometric optimization of the Patient Tracker PCPT system enhanced clarity in visual distinction of misalignments between the live patient position images and the desired reference position.

CONCLUSION: Patient Tracker PCPT may increase the efficiency of current Image Guided Proton Therapy systems by assisting radiation therapists in accurately positioning patients prior to X-ray imaging. The system will allow therapists to align patients in the simulation and treatment rooms more quickly, limit patient exposure to imaging radiation, and improve throughput. Continued optimization is needed to increase positioning accuracy to the level of clinical acceptability.

## Poster Abstract 59:

### A Range Compensation Method Using a Range Shifter to Reduce the Inter-fractional Dose Variation in SSPT

#### Takaaki Fujii, Taeko Matsuura, Seishin Takao, Shusuke Hirayama, Naoki Miyamoto, Kikuo Umegaki, Shinichi Shimizu, Hiroki Shirato

##### Graduate School of Medicine, Hokkaido University

PURPOSE: In Spot Scanning Proton-beam Therapy (SSPT), range variation due to the inter-fractional anatomical change is a crucial issue because it causes significant changes in the planned dose distribution. We proposed a method to correct the range on a field- by-field basis by adjusting the thickness of the gantry-mounted range shifter (RS). As a preliminary study, we applied this method to prostate cases and evaluated the impact to the dose distribution.

MATERIALS AND METHODS: For one prostate cancer patient, two CTs taken at the same day were used, which will be denoted as plan CT (pCT) and secondary CT (sCT) respectively. SFUD plans was made on pCT using four fields, and the spot information was exported in the DICOM-RT format. sCT was rigidly transformed so that fiducial-marker positions on the sCT match those of pCT, and the ROIs contoured in pCT were deformed into the sCT by commercial deform-software. By using the in-house software, the values of Water-Equivalent Length (WEL) along the proton-beam-path were calculated over all spots in pCT and sCT. The values of the WEL deviation between the two CTs (ΔWEL) were averaged over all spots and then used as the required size of the range correction. RS thickness was corrected accordingly. The comparison was made for the dose evaluation indices (D99 for prostate as CTV) between (1) original treatment-plan on pCT, (2) treatment-plan that is made by applying the original planned beam to sCT and (3) treatment-plan (2) + range correction using RS.

RESULTS AND CONCLUSIONS: The size of ΔWEL values were greater than 1 mm in one field (Field No.4: -1.0 mm). In this particular case, differences of the D99 among (1)(2)(3) were negligibly small due to the short-time-interval of taking the two CTs. However, we consider that the proposed method has the potential to reduce the ITV margin. We will further explore the effectiveness of the proposed method for other patients in the future.

## Poster Abstract 66:

### Operational Metrics of a Compact Proton Therapy System: First Year Experience

#### Omar Zeidan, Ethan Pepmiller, Twyla Willoughby, Sanford Meeks, Naren Ramakrishna

##### Orlando Health

We report on various metrics for first-year operation of our compact double scattering proton system. A total of 130 patients were treated between April 2016 and April 2017 with the following site breakdown: 45% prostates, 18% adult CNS, 16% pediatrics, 9% H&N, 7% thorax/lung, and 5% breast/chest wall. Nearly 86% of patients lived within 100 miles radius from the center and nearly 70% of them within a 50-mile radius. A total of 3,833 fractions were delivered, which is nearly 28% more than pro-forma assumptions. The maximum fractions delivered per day was 27 nearly a year from go-live and corresponding to 12 hours of treatment. The highest number of fractions treated in a given month was 494. Total number of fractions canceled due to downtime was 196 with 4 months with no treatment cancellations corresponding to an overall uptime of nearly 95%. The maximum number of cancellations in a given month was 33 (9.6% of scheduled fractions for the month). Of the 421 beams delivered, the min, max, and mean of beam ranges were 5.0, 30.6, and 16.3 cm, respectively. The mode and mean modulation were 9.0 cm and 8.7 cm, respectively. The relatively compact circular geometry of the applicator housing allows for smaller air gaps compared to larger PT systems. The min, max and mean air gaps for all treatments was 4.0, 23.0, and 8.2 cm, respectively. The most widely used gantry angles for treatment were 90, 180, and 0 degrees. Combined they were nearly 40% of all available angles. Robotic couch angles of 0, 180, and 270 degrees comprised nearly 72% of all couch angles. The system snout accommodates only two-size applicators with maximum field sizes of 14 cm and 25 cm. The min, max, and mean field diameters (maximum aperture opening) for all beams were 3.4, 28.0, and 13.8 cm, respectively. We were the first center to use the Pinnacle treatment planning system for proton planning. In addition, we were the first to use the Airo 32-slice mobile CT for in-room patient anatomy verification and localization in a compact room setting.

## Poster Abstract 70:

### Multi-institutional Results of Breast Proton Radiation Therapy: An Analysis of the PCG Registry

#### Daniel Brunnhoelzl, Gary Larson, Henry Tsai, Carl Rossi, Lane Rosen, William Hartsell, Christine Fang, Elizabeth Nichols, Carlos Vargas

##### Creighton University School of Medicine – PRC, ProCure Proton Therapy Center – Oklahoma City and New Jersey, Scripps Proton Therapy Center, Willis-Knighton Cancer Center, Northwestern Medicine Chicago Proton Center, University of Washington Medical Center, University of Maryland Medical Center, Mayo Clinic – Scottsdale

PURPOSE: This study aimed to determine disease-specific outcomes and toxicities associated with proton therapy for breast cancer treatment.

METHODS: Records of 335 breast cancer patients with localized disease treated with proton radiotherapy (RT) in a multi-institutional prospective registry between 2011 and 2016 were analyzed. Late toxicity is (CTCAE V4.0) any event beginning or persisting for 6 months or longer from the start of radiotherapy and acute Adverse Events (AE) before that. Cancer events were measured from the completion of RT.

RESULTS: Median follow-up was 1.1 years (0.2-3.7 years) and 73% for longer than 6 months. Nodal stage was pN1-3 in 68%. Re-irradiation was delivered to 52 (15.5%), chest wall RT 135 (40.3%), whole breast 108 (32.3%), and partial breast RT 40 (11.9%). Loco-regional failures were seen in 2.4%, distant failures in 6.9%, breast cancer deaths in 10 (3.0%), and deaths in 15 (4.5%). AEs ≥G3 or ≥G2 were different between retreatment (9, 17.3%; 43, 82.7%) and de novo (20, 7.1%; 191, 67.5%), p=0.028 and p=0.032. Only one chronic G3 AE was seen. Retreatment was associated with similar loco-regional failure rates 3 (5.8%) vs 5 (1.8%), p=0.11. However, distant failures (7 and 13.5% vs 16 and 5.7%; p=0.06), breast cancer deaths (4 and 7.7% vs 6 and 2.1%; p=0.05), and overall deaths (6 and 11.5% vs 9 and 3.2%; p=0.017) were higher. Volume treated was a predictor of G2 or G3 AEs; AEs ≥G3 or ≥G2 were different between patients treated to the regional lymph nodes (20, 10.7%; 165, 88.2%) vs local RT (0, 0.0%; 25, 26%), p=0.0003 and p= 0.0001. A relationship between dose and AEs was found; for ≥G3 the dose threshold was 61.3 Gy (6 and 17.5% vs 14 and 5.6%; p=0.02) and for ≥G2 it was 45.5 Gy, (17 and 50.0% vs 173 69.4%; p=0.0319). Use of PBS (Pencil Beam Scanning) revealed decrease in ≥G2 (36, 63.8%) vs. 155 (90.0%) for double scatter, p=0.0008.

CONCLUSION: Proton beam radiation was associated with low chronic G3 AEs including re-treatment cases. Volume and dose thresholds were found, as well as a benefit for the use of PBS.

## Poster Abstract 71:

### MultiCT-based Optimization of Intensity Modulated Proton Therapy

#### Xiaodong Zhang, Xianliang Wang, Heng Li, Bo Jiang, Xiaorong Ronald Zhu

##### MD Anderson Cancer Center

Background and purpose: Intensity modulated proton therapy (IMPT) is sensitive to patient anatomical changes. We hypothesized that a plan's robustness to anatomical changes can be improved by optimizing with multiple CTs of a patient. The major purpose of this study was to validate whether there is a plan meets dose criteria of multiple CTs.

Material and methods: Eight lung cancer patients who underwent IMPT at our institution were selected for this retrospective study. For each patient, two CTs, primary planning CT (PCT) and adaptive planning CT (ACT), were combined to optimize one plan. The selective robust plan of PCT was introduced into ACT for dose calculation (P-ACT plan). The dose difference between selective robust plan, MultiCT-based plan and P-ACT plan was compared.

Results: For each patient, the MultiCT-based plan was able to ensure the dose coverage (D95%) of target as compared with the selective robust plan. MultiCT-based plans led to an average increase of lung V20 (PCT 2.49%, ACT 2.89%), spinal cord Dmax (PCT 3.45%, ACT 3.34%) and heart V30 (PCT 2.61%, ACT 9.34%), but all within acceptable limits. There was no statistical significant difference of OARs dose between P-ACT plans and MultiCT-based plans. However, the CTV dose coverage of ACT was significantly improved by the MultiCT-based plan as compared with the P-ACT plan.

Conclusions: The results of this study indicate that a plan which meets dose criteria of both PCT and ACT is exist. The MultiCT-based optimization can increase the plan robustness to anatomical change. How to complete a MultiCT-based plan according to a PCT combined with a forecasting model or AI is the next research direction.