Program Description


The theme for this year's conference was New Frontiers in Particle Therapy. Objectives for the 6th annual conference were to: Maintain best practices for proton therapy patient selection by recognizing maximum opportunity for toxicity reduction and/or tumor control probability improvement; appropriately identify patients for clinical trials and employ resources to overcome barriers to improve enrollment; determine where and when to implement advanced treatment planning approaches such as Monte Carlo, relative biological effectiveness (RBE), and linear energy transfer (LET); and consider limitations of proton therapy and recognize when applications of dual modality or non-proton therapy options are indicated to achieve optimal patient outcomes.

Target Audience

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

  • Radiation Oncologists

  • Medical Physicists

  • Dosimetrists

  • Residents

  • Radiation Therapists

Particle Therapy Cooperative Group North America (PTCOG-NA) 2019 Committees

PTCOG-NA Executive Committee

  • Eugen B. Hug, M.D., President

  • Hesham E. Gayar, M.D., Vice-President

  • Anita Mahajan, M.D., Secretary

  • Carl J. Rossi, M.D., Treasurer

PTCOG-NA Host Committee Members

  • Michael D. Chuong, M.D.

    Director, Clinical Research and Education, Miami Cancer Institute, Miami, Florida

  • Marcio Fagundes, M.D.

    Medical Director, Radiation Oncology, Miami Cancer Institute, Miami, Florida

  • Alonso N. Gutierrez, Ph.D., MBA

    Assistant Vice-President, Chief Physicist, Radiation Oncology, Miami Cancer Institute, Miami, Florida

  • Minesh P. Mehta, M.D.

    Deputy Director and Chief, Radiation Oncology, Miami Cancer Institute, Miami, Florida

PTCOG-NA Conference Planning Committee Members

  • Michael D. Chuong, M.D.

    Director, Clinical Research and Education, Miami Cancer Institute, Miami, Florida

  • Marcio Fagundes, M.D.

    Medical Director, Radiation Oncology, Miami Cancer Institute, Miami, Florida

  • Hesham E. Gayar, M.D.

    Medical Directort, McLaren Proton Therapy Center, Karmanos Cancer Institute, Flint, Michigan

  • Alonso N. Gutierrez, Ph.D., MBA

    Assistant Vice-President, Chief Physicist, Radiation Oncology, Miami Cancer Institute, Miami, Florida

  • William F. Hartsell, M.D., FACR, FACRO

    Medical Director, Chicago Proton Center, Radiation Oncologist, Northwestern Medicine, Chicago, Illinois

  • Eugen B. Hug, M.D.

    Professor, Radiation Oncology, Medical Director, MedAustron Ion Therapy Center, Wiener Neustadt, Austria

  • Anita Mahajan, M.D.

    Radiation Oncologist, Brain Tumor Program, Mayo Clinic, Rochester, Minnesota

  • Minesh P. Mehta, M.D.

    Deputy Director and Chief, Radiation Oncology, Miami Cancer Institute, Miami, Florida

  • Carl Rossi, M.D.

    Medical Director, California Protons Cancer Therapy Center, San Diego, California

Abstract 1: Average LET of X-Rays and Gamma Radiation for Radiobiological Modelling

BACKGROUND: Differences in proton and photon physics is an underlying reason for differences in biological effectiveness. Photons set in motion delta-electrons, which transfer energy to biological targets. According to Bethe-Bloch formula, an electron has the same LET as a proton travelling with the same velocity. This suggests that differences in energy spectra of electrons and protons are crucial for understanding differences in biological effectiveness. Electron spectra are complex and the term itself is prone to misinterpretation. We define the spectrum unambiguously as the probability distribution of energy of electrons entering a microscopic biological target.

METHODS AND MATERIALS: We account only for electrons that reach the target volume. In previous studies, average LETs were calculated using the source spectra, i.e. distributions of initial electron energies at the point of origin. These are not the spectra of electrons reaching the target. Electrons are mostly produced outside the target and lose some energy before they reach it. Using source spectra overestimates RBE variation with beam energy, NCRP 181. We calculate spectra of electrons entering microscopic volumes by solving an electron transport equation with Monte Carlo.

RESULTS: We report dose- and frequency averaged LETs for x-ray, brachytherapy, and gamma sources. For a 60Co source, our LETD is 4.6 keV/μm, an order of magnitude higher than reported previously, and higher than LET of protons with energies >10 MeV. The latter implies proton RBE<1 at higher energies. Our model (Vassiliev et al. Phys. Med. Biol. 2018) resolves this problem. It substantiates the use of LETF, which in our calculations is 0.40 keV/μm for 60Co.

CONCLUSIONS: We propose a definition of delta electron spectrum for radiobiological modelling and an algorithm for calculating it. We report average LETs for several sources. Our results are very different from those previously reported. They support using LETF instead of LETD.

Abstract 3: Intensity Modulated Proton Therapy Using Dose-Painting Pencil Beam Scanning for High-Risk Hepatocellular Carcinomas

BACKGROUND: Hypofractionated proton therapy with passive techniques for hepatocellular carcinoma (HCC) may have limitations when tumors are adjacent to organs-at-risk (OARs), which may result in tumor underdosage and lead to inferior local control. We present the first series of HCC patients treated with pencil beam scanning (PBS) intensity-modulated proton therapy (IMPT) using a simultaneous-integrated boost and protection (SIB/SIP) technique to escalate tumor dose while protecting adjacent OARs.

METHODS AND MATERIALS: Sixteen consecutive HCC patients were treated between 2015–2018 with a 15-fraction regimen using IMPT SIB/SIP. SIB/SIP dose levels generally ranged from 37.5 to 67.5 GyRBE to minimize dose to OARs at their respective dose-limiting thresholds (e.g. luminal gastrointestinal organs, chest wall). Hepatotoxicity was defined by a Child-Pugh (CP) score increase of 2 or greater and/or any RTOG grade 3 enzyme elevation. Other toxicities were graded by CTCAEv5.0. Overall survival and local-progression-free survival were calculated using the Kaplan-Meier method.

RESULTS: Patients most commonly had BCLC stage C (50%) and CP-A cirrhosis (71%). Median gross tumor volume (GTV) size was 12.7cm (599cc [228–1617]), and 38% had gross vascular invasion. Median GTV dosimetric parameters included: maximum prescription dose 67.5 GyRBE (60–67.5), mean 62.5 GyRBE (54.0–69.5), D1 68.0 GyRBE (61.6–71.3), and D99 50.4 GyRBE (33.4–67.7). Median liver-GTV parameters included: volume 1403.4cc (805–2130), mean 14.6 GyRBE (11.1–19.6), V30GyRBE of 27% (12%–35%), and V20GyRBE of 32% (21%–47%). At a median follow-up of 447 days (164–894) in alive patients, the median survival and 1-year overall survival was 22 months and 60%, respectively. Local control was 88% with no isolated local failures. Three patients experienced hepatotoxicity with no RILD-related deaths. No acute or late GI grade ≥2 occurred. One patient developed grade 3 chest wall toxicity.

CONCLUSION: In our series of HCC patients with large tumors near OARs, IMPT SIB/SIP results in excellent local control and acceptable toxicities.

Abstract 4: A University of Florida and Proton Collaborative Group Phase I/II Study of Hypofractionated Proton Therapy with Concurrent Chemotherapy for Stage II–III NSCLC

BACKGROUND: While hypofractionated (Hypo) radiotherapy (RT) has widely replaced standard fractionated RT in the treatment of early-stage non-small cell lung cancer (NSCLC), major side effects have limited its use in advanced NSCLC. Based on the improved dosimetry of proton therapy (PT), we investigated the HypoPT approach for stage II–III NSCLC.

METHODS AND MATERIALS: Between March 2013 and November 2018, 28 patients from 4 centers were enrolled on a clinical trial of HypoPT with concurrent chemotherapy followed by adjuvant systemic therapy. Patients could be simultaneously enrolled in a phase 1 study and receive doses of 2.5 (n=14), 3 (n=6), 3.53(n=7), and 4 Gy/fraction (n=1) to a total dose of 60 GyRBE according to the open arm and organ-at-risk (OAR) dosimetric constraints. Patients had stage IIA (n=3), IIB (n=3), IIIA (n=15), and IIIB (n=7) NSCLC. CTCAE, v4.0, was used for toxicity assessment. The primary endpoint of the study was 1-year overall survival (OS). The study closed prematurely due to slow accrual.

RESULTS: The median follow-up for surviving patients was 23 months (range, 1–60). The 1- and 2-year OS rates were 89% and 66%, and the 1- and 2-year progression-free survival rates were 70% and 60%, respectively. Three patients died within 3 months of completing HypoPT: 1 from a bronchial hemorrhage; 1 from congestive heart failure following infectious pneumonia and C. Diff colitis; and 1 from paraneoplastic SIADH after completing just 42 Gy at 3.53 Gy/fraction. Additionally, 6 patients died more than 1 year after HypoPT: 5 from disease progression and 1 from a cardiac event.

CONCLUSIONS: In this phase I/II study, HypoPT at 2.5 to 3.53 Gy per fraction to a total 60 Gy (RBE) with concurrent chemotherapy was well tolerated with favorable PFS and OS. A large randomized clinical trial comparing HypoPT with standard fractionated RT or PT is warranted, especially in the setting of consolidation immunotherapy.

Abstract 5: Early Results of Re-Irradiation for Rectal Cancer Using Pencil-Beam Scanning Proton Therapy Are Promising

BACKGROUND: Re-irradiation (Re-RT) for rectal cancer (RC) in patients with prior pelvic RT has been shown to be safe and effective. However, limited data exists with the use of proton therapy (PT). We hypothesize that PT is a safe and feasible for re-treatment and may allow for decrease in toxicity or treatment escalation.

METHODS AND MATERIALS: We performed a single institutional retrospective IRB-approved analysis of all RC patients with any prior pelvic RT re-irradiated with Pencil-Beam Scanning proton therapy (PBSPT). We collected patient and treatment characteristics including prior diagnosis and treatment; RC diagnosis and re-irradiation records; and toxicities. Outcomes, including overall Survival (OS) and Local Control (LC), were estimated using Kaplan-Meier.

RESULTS: Twenty-six patients (median follow-up 15.3 months) received proton PBSPT Re-RT from 2016–2018: 16 patients w/ recurrent RC [median prior dose 52.2 Gy (43.2–63.0)] and 10 patients w/ de novo RC and variable prior RT (9 for prostate, 1 for ovarian). Median Re-RT dose was 44.4 Gy [(16.0–60.0); 20/26 BID], and 22 received concurrent chemotherapy. Five underwent surgical resection (all R0). Three patients experienced grade 3 acute toxicities, and no acute Grade 4–5 toxicities were observed. Two patients had grade 3+ late toxicities, including a grade 5 toxicity occurring in a patient with history of significant injury from prior RT. One-year LC and OS were 76.5% (95% CI 66.0–86.9%) and 77.7% (95% CI 68.8–86.6%), respectively.

CONCLUSION: In this largest such series, early results of PT for Re-RT for RC are promising, with longer follow-up needed.

Abstract 6: What's Gone Wrong in Head and Neck Rhabdomyosarcoma?

BACKGROUND: Recent trends including proton therapy and reduced-dose cyclophosphamide have been adapted in head and neck rhabdomyosarcoma (HN-RMS) to reduce late morbidity. Our primary goal was to analyze local control and survival outcomes after photon versus proton irradiation in pediatric patients with HN-RMS, with the secondary goal of analyzing the effect of cyclophosphamide dose on disease outcomes.

METHODS AND MATERIALS: This was a cohort study comprising 76 pediatric patients treated with definitive chemoradiation for HN-RMS from 2000 to 2018. Fifty-one patients (67%) were treated with intensity-modulated photon radiation therapy (IMRT) and 25 patients (33%) were treated with proton therapy.

RESULTS: Local failure (LF) at 3 years was 21.8% for parameningeal RMS and 0% for orbital RMS and other head and neck sites (p=0.24). Patients who were treated with protons were more likely to have received reduced dose cyclophosphamide (p<0.0001). The 3-year LF was 10.0% in the IMRT cohort versus 21.6% in the proton cohort (p=0.07). Cyclophosphamide dose was associated with LF: the 3-year LF was 3.9% for patients who received a cumulative dose of >20g/m2 versus 18.4% for ≤20g/m2 (p=0.04). Among patients with parameningeal RMS (n=59), both the cumulative cyclophosphamide dose and dose-intensity were associated with local failure (p=0.01). There were no differences in survival outcomes among the IMRT and proton cohorts. There was a trend toward worse event-free survival in patients with parameningeal RMS who received reduced dose-intensity cyclophosphamide (46.1% versus 67.5%, p=0.11).

CONCLUSIONS: Longer follow-up is needed in the proton cohort, although it appears that the dose of cyclophosphamide, rather than radiation modality, is likely the factor affecting local disease control. Efforts focused on further evaluating the optimal dose of cyclophosphamide or alkylating agents needed to balance disease control with toxicity are needed.

Abstract 10: Image-Guided Hypofractionated Proton Therapy in the Management of Early-Stage Non-Small Cell Lung Cancer

BACKGROUND: Due to the excellent outcomes with image-guided stereotactic body radiotherapy (SBRT) for patients with early-stage non-small cell lung cancer (NSCLC), and the low treatment-related toxicities using proton therapy, we investigated treatment outcomes and toxicities for delivering hypofractionated proton therapy (PT).

METHODS AND MATERIALS: Between 2009 and 2018, 22 patients with T1–T2N0M0 NSCLC (45% T1, 55% T2) were enrolled and treated with image-guided hypofractionated PT on an IRB-approved phase II clinical trial. The median age at diagnosis was 72 years (range, 58 – 90). Patients underwent 4-dimensional computed tomography (CT) simulation following fiducial marker placement, and daily image guidance was performed. Nine patients (41%) were treated with 48 GyRBE in 4 fractions for peripheral lesions, and 13 patients (59%) were treated with 60 GyRBE in 10 fractions for central lesions. Patients were assessed for CTCAEv4 toxicities weekly during treatment, and at regular follow-up intervals with CT imaging for tumor assessment. Overall survival, cause-specific survival, local control, regional control, and metastases-free survival were evaluated using cumulative incidence with competing risks.

RESULTS: The median follow-up for all patients was 3.5 years (range, 0.2–8.8 years). The overall survival rates at 3 and 5 years were 81% and 49%, respectively. The cause-specific survival rates at 3 and 5 years were 100% and 75%, respectively. The 3-year local, regional, and distant control rates were 86%, 85%, and 95%, respectively. Four patients experienced in-field recurrences. The median time to local recurrence was 26.5 months (range, 19–47 months). One patient (5%) developed a late grade 3 bronchial stricture that required hospitalization and stent.

CONCLUSIONS: Image-guided hypofractionated PT for early-stage NSCLC provides promising local control and long-term survival with low toxicities. Regional nodal and distant relapses remain a problem.

Abstract 13: Initial Clinical Outcomes for Prostate Cancer Patients Undergoing Adjuvant or Salvage Post-Prostatectomy Proton Therapy (PT)

BACKGROUND: To report clinical outcomes associated with post-prostatectomy PT. Toxicity outcomes for this cohort were recently published.

METHODS AND MATERIALS: The first 100 consecutive patients from 2010–2016 were retrospectively assessed. Biochemical failure (BF; 2 consecutive rises above the nadir), first site of clinical failure – local, regional, and/or distant metastasis (DM) – and overall survival were recorded from start of radiation. BF- and DM-free survival Kaplan-Meier curves were estimated. Cox proportion hazards model was used to assess uni- (UVA) and multivariable association (MVA) with BF; variables with <0.1 were included in the multivariable model.

RESULTS: Median age and months after surgery were respectively 64 years (range 42–77) and 25 (5–216). PT received was 70.2Gy (RBE) (89%), salvage (93%), prostate bed-only (80%), pencil beam scanning (86%), with intensity-modulated radiation therapy (31%), and with androgen deprivation (34%). Median follow-up was 55mo (16–80). BF was noted in 39 patients (39%). Median time to BF was 23mo (5–69). For patients with BF, local failure was eventually noted in 1 (1%) patient at 30mo. Regional pelvic nodal failure was noted in 4 patients (4%) – all treated to the prostate bed-only – at median 32mo (10–38), 2 of whom also had DM. DM occurred in 6 patients (6%) at median 30mo (10–41), 5 with bony and 1 with lung involvement. There was 1 death at 24mo, unrelated to prostate cancer. In summary, 4.5 yr BF free-, DM free-, and overall-survival were 61%, 94%, and 99%, respectively, in this single institution cohort treated primarily to the prostate bed only without androgen deprivation. On MVA, Gleason >7 (HR 3.55, 95% CI 1.83–6.88, p=0.000) and whole-pelvis with prostate bed PT (HR 0.28, 95% CI 0.10–0.79, p<0.016) were associated with BF.

CONCLUSIONS: Post-prostatectomy PT is feasible with comparable clinical outcomes to historical photon outcomes.

Abstract 15: Clinical Outcomes Following Proton Beam Therapy for Locally Advanced Non-Small Cell Lung Cancer: Analysis of a Multi-Institutional Prospective Registry

BACKGROUND: For most disease sites, level 1 evidence is lacking for proton beam therapy (PBT). By identifying target populations that would benefit most from PBT, prospective registries could overcome the challenges in clinical trials enrollment. Herein, we report clinical outcomes of patients treated with PBT for locally advanced non-small cell lung cancer (LA-NSCLC).

METHODS AND MATERIALS: Data were obtained from the multi-institutional prospective database of the Proton Collaborative Group (PCG). Inclusion criteria of our study were stage III LA-NSCLC, use of PBT, and availability of follow-up data. Survival time was calculated from the start of treatment until death or last follow-up. Kaplan-Meier curves were generated for groups of interest and compared with log-rank tests. Cox regression modeling was used to evaluate the relationship between selected covariates and overall survival (OS).

RESULTS: A total of 195 patients were included in the analysis. PBT alone was given to 93% of patients with a median equivalent dose in 2 Gy fractions (EQD2) of 63.8 Gy(RBE). Pencil beam scanning (PBS) was used in 20% of treatments. Treatment-related grade 3 adverse events (AEs) were rare: one pneumonitis, two dermatitis, and three esophagitis. No grade 4 events were reported. Two grade 5 events occurred, both cardiological, probably unrelated to PBT. The median follow-up time for living patients was 13.6 months and the median OS was 19.0 months. On multivariate analysis, good performance status (HR=0.26, 95% CI 0.15–0.47, p<0.0001), PBS use (HR=0.45, 95% CI 0.20–0.99, p=0.046), and increased EQD2 (HR=0.97, 95% CI 0.96–0.98, p<0.0001) were associated with decreased mortality.

CONCLUSION: PBT appears to yield low rates of AEs with encouraging OS for the treatment of LA-NSCLC. PBS use and increased EQD2 can potentially increase OS. Prospective databases such as the PCG registry could play a key role in the future but need meticulous updates to reflect the clinical reality.

Abstract 18: Outcomes after Proton Therapy for Skull-Base Chordoma: A Prospective Study

PURPOSE: To evaluate treatment outcomes following definitive or adjuvant high-dose, image-guided proton therapy for patients with skull-base chordoma.

METHODS AND MATERIALS: Between February 2007 and February 2018, 91 patients with a median age of 53 years (range, 22–78 years) with a skull-base chordoma were treated with passively scattered 3D-conformal proton therapy to a median dose of 73.8 Gy(RBE) (range, 69.6–75.6 GyRBE) on a prospectively collected, IRB-approved outcomes tracking protocol.

RESULTS: The median age was 53 years (range, 22–78 years). Two patients received a component of intensity-modulated radiotherapy. Seventy percent (n=64) were men and 30% (n=27) were woman. Eighty-two percent (n=75) of patients had macroscopic disease at the time of radiotherapy; 18% (n=16) had undergone a macroscopic gross total resection. Overall survival, cause-specific survival, local control, and RT-related grade 3 toxicity-free survival were calculated. Proton therapy-related toxicities were scored using CTCAE v4.0. With a median follow-up of 3.7 years (range, 0.2–10 years), 26 patients experienced disease recurrence, including 26 local, 0 regional, and 1 distant recurrence. The median time to local progression was 2.2 years (range, 0.4–7.0 years). At the time of last follow-up, 66 patients were alive (56 with no evidence of disease progression) and 25 were deceased (18 with disease progression). There were no acute grade 3 toxicities related to the radiation therapy. The 4-year actuarial rates of overall survival, cause-specific survival, local control, and radiation therapy-related grade 3 toxicity-free survival were 83%, 87%, 76% and 83%, respectively.

CONCLUSION: Definitive or adjuvant high-dose passively scattered 3-dimensional conformal proton therapy for skull-base chordoma provides acceptable local control, comparing favorably to historic photon data, with no acute grade ≥3 radiation-related toxicity and an acceptable rate of grade ≥3 late toxicity. Further follow-up of this cohort is necessary to better characterize long-term disease control and late toxicities.

Abstract 19: Postoperative or Salvage Proton Radiotherapy for Prostate Cancer Following Radical Prostatectomy

BACKGROUND: Post-prostatectomy radiation improves disease control, but limited data exist regarding outcomes, toxicities, and patient-reported quality-of-life with proton therapy.

METHODS AND MATERIALS: The first 101 consecutive patients treated with double-scattered proton therapy (DSPT) between 2006 and 2015 were retrospectively reviewed. Seventy-eight patients received DSPT to the prostate bed only. Twenty-three received a DSPT prostate-bed boost following prostate-bed and pelvic-node treatment (3 with conformal 3-dimensional radiotherapy; 20 with intensity-modulated radiotherapy). Ten adjuvant patients received a median dose of 66.6 GyRBE (range, 66.0–70.2). Ninety-one salvage patients received a median dose of 70.2 GyRBE (range, 66.0–78.0). Three patients had a single positive lymph node. Forty-seven patients received androgen deprivation therapy (ADT) for a median of 9 months (range, 1–30). Toxicities (CTCAE, v4.0) were prospectively graded, and patient-reported quality of life data were reviewed.

RESULTS: The median follow-up was 5.0 years (range, 0.8–11.3). Five-year biochemical relapse-free and distant metastases-free survival rates were 69% and 90% for adjuvant patients, 58% and 96% for salvage patients, and 58% and 95% overall. ADT was associated with improvement in biochemical relapse-free survival on univariate analysis (p=0.0286). Acute and late >grade 3 genitourinary toxicity rates were 1% and 6%. No patients had >grade 3 gastrointestinal toxicity. Acute and late grade 2 gastrointestinal toxicities were 5% and 2%. The median and range International Prostate Symptom Score, International Index of Erectile Function, and Expanded Prostate Cancer Index Composite bowel function and bother were 5 (range, 0–33), 7 (range, 2–25), and 100 (range, 42–100), 92 (range, 57–100), respectively at baseline, and 11 (range, 0–34), 5 (range, 4–24), 92 (range, 17–100) and 89 (range, 42–100) at the 5-year follow-up.

CONCLUSION: High-dose post-prostatectomy PT provides effective long-term biochemical control and freedom from metastasis, with low acute and long-term gastrointestinal and genitourinary toxicity.

Abstract 20: Reirradiation with Proton Therapy for Recurrent Malignancies of the Esophagus and Gastroesophageal Junction: Results of the Proton Collaborative Group Multi-Institutional Prospective Registry Trial

BACKGROUND: Patients with esophageal or gastroesophageal junction (GEJ) cancers with isolated locoregional recurrences have limited definitive treatment options. Reirradiation with modest photon doses is associated with high toxicity rates. Proton therapy improves normal tissue sparing and may more safely allow reirradiation dose escalation. Outcomes and toxicities from the Proton Collaborative Group registry assessing proton reirradiation for esophageal/GEJ cancer locoregional recurrences are reported.

METHODS AND MATERIALS: This IRB-approved prospective, multi-institutional registry was queried for recurrent esophageal/GEJ cancer treated with a second course of radiotherapy using protons. Baseline demographics, treatment details, outcomes, and toxicities (CTCAEv4.0) were evaluated.

RESULTS: Twenty-five consecutive patients retreated from 7/2012–4/2018 were analyzed. Patients were a median of 70 years (52–82) and predominantly male (88%) and non-Hispanic Caucasian (88%) with adenocarcinoma (68%). Initial stage was: T1–4%, T2–8%, T3–76%, unknown–8%; N0–24%, N1–48%, N2–20%, unknown–8%; stage I–4%, IIA–4%, IIB–24%, IIIA–40%, IIIB–16%, IV–4%, unknown–4%. During their initial treatment (median 50.4/1.8Gy), all received concurrent chemotherapy, and five patients (20%) underwent resection. Median time to recurrence was 14.0 months (4.1–134.0). Proton reirradiation was delivered using uniform scanning or passive scattering (n=21) or pencil beam scanning (n=4) to a median 45.9Gy(RBE) (30.1–60.5) in 12–42 fractions. Eleven patients (44%) received concurrent chemotherapy. Median follow-up from reirradiation completion was 23.1 months among living patients. Fifteen patients died; 6/18-month survivals were 48%/24%. Only two patients (8%) developed locoregional recurrences (9.2 and 11.1 months following reirradiation) both salvaged with surgical resection. Grade 3 toxicities occurred in 20% (anemia=1; anorexia=2; dysphagia=2; esophagitis=2) and grade 2 in 44% (most commonly, fatigue=6). Grade ≥2 esophagitis and pneumonitis occurred in 16% and 0%, respectively. No esophageal fistula/stricture/necrosis and no grade 4–5 events occurred.

CONCLUSION: Proton reirradiation for locoregionally recurrent esophageal/GEJ cancers is feasible, achieves durable local control, and has relatively limited toxicity. Additional prospective investigations and analyses of cumulative dose constraints are warranted.

Abstract 21: Clinical Outcomes after Proton Partial Breast Radiotherapy for Early Stage, Hormone Receptor Positive Breast Cancer: 3-Year Outcomes of a Phase II Trial

BACKGROUND: Proton partial breast irradiation (PBI) may decrease morbidity versus photon PBI by allowing superior normal tissue sparing. Single-institutional studies have reported feasibility of proton PBI but with conflicting toxicity results. We report 3-year outcomes of a prospective phase II trial investigating efficacy and toxicity of proton PBI.

METHODS AND MATERIALS: This multi-center Proton Collaborative Group phase II trial (PCG BRE007-12) recruited women age ≥50 years with AJCC stage 0–2, node negative, ER-positive, ≤3cm, non-lobular invasive breast cancer or ductal carcinoma in situ who underwent breast- conserving surgery. Proton PBI was delivered to 40Gy(RBE) over 2 weeks. All received uniform scanning proton therapy except 1 receiving passively scattered protons; ≥2 treatment fields were used. Primary endpoint was progression- free survival. Adverse events were prospectively graded using CTCAEv4.0. BTCOS was used to assess patient-reported quality of life (PRQOL) endpoints and cosmesis (scored 1–4).

RESULTS: Thirty-eight evaluable patients (left-sided=21 [55%]) were enrolled between 2/2013–11/2016. Median tumor size was 0.95cm. Median age was 67 years (50–79). At a median follow-up of 35 months (12–62), all patients were alive, and no patient had local, locoregional or distant disease progression. One patient developed new ER-negative invasive ductal carcinoma of the contralateral breast. Twenty-nine patients (76.3%) received hormonal therapy. Seven grade 2 toxicities occurred (radiation dermatitis=1; lymphedema=1; hot flashes=3; dyspnea=1; fatigue=1). No grade ≥3 toxicities attributable to radiotherapy were observed. Five patients (13%) assigned a BTCOS score of 4 at 1- or 3-year follow-up. Median heart volume receiving 5Gy (V5Gy) was 0%; lung V20Gy was 0% and V10Gy 0.17%.

CONCLUSION: At 3 years, proton PBI provides 100% cancer control for early stage, ER-positive breast cancer with minimal toxicity and acceptable long-term PRQOL. These findings, together with excellent dosimetric sparing of critical organs achieved with proton PBI, provide evidence that protons are safe and effective for PBI.

Abstract 22: Using Breath Hold at the End of Exhale for Lung Treatment in Pencil Beam Scanning Proton Therapy

INTRODUCTION: The treatment of targets with considerable motion is a big challenge in proton therapy, especially for pencil beam scanning (PBS) technique. The interplay effect can be significant particularly for small target treated with hypofractionation or SBRT. Deep inspiration breath hold (DIBH) can't apply to all patients, especially for lung patients with poor lung function. Here we present two SBRT lung cases treated with breath hold at the end of exhale (EoE) using PBS proton therapy.

METHODS AND MATERIALS: Two patients prescribed to receive SBRT lung treatment of 55 Gy in 5 fractions show target motion more than 1.5 cm on their 4DCT images. Both patients could not tolerate DIBH. Repeated CT scans were performed with patient holding their breath at EoE using Anzai respiratory gating system. A 4D movie was created from these BH scans to check the BH reproducibility. An average BH scan was used as the planning CT. Robust optimization was used on all BH scans with 4% range uncertainty and 5 mm setup uncertainty. Layer repainting and volumetric repainting were also used to minimize interplay effects.

RESULTS: Both patients tolerated BH well at EoE. Daily CBCT and verification CT (VFCT) showed good reproducibility of the target location. The average BH duration was around 15 seconds and patient could quickly resume the BH after resting two or three regular breaths. The clinic plan was forward calculated on the VFCT and both target and OAR DVHs remained similar to the nominal plan.

CONCLUSIONS: BH at EoE is a viable option for patient who can't tolerate DIBH, which reduces target motion and mitigate the interplay effect of PBS treatment. Multiple factors should be considered carefully when implementing this technique in clinic, for example, gating system selection, BH reproducibility, image verification during treatment, treatment plan robustness, and beam delivery time etc.

Abstract 23:Comparative Effectiveness of Proton Therapy Versus Photon Therapy as Part of Concurrent Chemo-Radiotherapy for Locally Advanced Cancer: Analysis of Patients with Private Health Insurance

BACKGROUND: Proton therapy may increase the tolerability/effectiveness of concurrent chemo-radiotherapy for locally advanced cancers but is unproven and generally not covered by private insurers. There is very little data on the comparative effectiveness of proton vs. photon chemo-radiotherapy among private insurance patients to guide payers on coverage policies for protons.

METHODS AND MATERIALS: We conducted a comparative effectiveness study of adult non-metastatic cancer patients with non-Medicare private insurance treated with curative-intent proton chemo-radiotherapy vs. photon chemo-radiotherapy from 2011–2016 at Penn. The choice of radiation modality was largely determined by the insurer's proton coverage policy. Data on adverse events (AEs) and survival were gathered prospectively using standardized templates. Primary endpoint was 90-day AEs associated with unplanned hospitalizations (CTCAEv4 grade ≥3 AEs). Secondary endpoints included 90-day grade ≥2 AEs, decline in ECOG performance status during treatment, disease-free survival (DFS), and overall survival (OS). Modified Poisson regression models with inverse propensity score weighting were used for adverse event outcomes and weighted Cox proportional hazards models were used for survival outcomes. Propensity scores were estimated using an ensemble machine-learning approach.

RESULTS: 920 patients were included (178 proton/742 photon). Median age was 57. Disease sites included H&N(25 proton/296 photon); CNS(44/128); lung(41/120), upper GI(34/78), and lower GI/GYN(34/120). Race, Charlson-Deyo comorbidity score, BMI, baseline toxicity, and baseline performance status were similar (p>0.05 for all). In propensity score weighted-analyses, proton chemo-radiotherapy was associated with significantly lower relative risk (RR) of 90-day grade ≥3 AEs (RR 0.51, 95%CI 0.32–0.81, p<0.01) and 90-day grade ≥2 AE's (RR 0.91, 95%CI 0.83–0.99, p=0.03). Decline in performance status (RR 0.85, 95%CI 0.70–1.04, p=0.11) and adjusted DFS and OS all favored proton therapy, but the differences were not statistically significant.

CONCLUSIONS In adults with locally advanced cancer who have private insurance, proton chemo-radiotherapy was associated with significantly reduced acute adverse events causing unplanned hospitalizations, with similar DFS and OS.

Abstract 27: Minimal Severe Acute Toxicity in Anal Cancer Patients Treated with Proton Beam Therapy

BACKGROUND: Definitive chemoradiation (CRT) for anal squamous cell carcinoma (SCC) is curative for most patients. Despite the high conformality of x-ray therapy most patients experience at least grade 2 (G2) acute toxicity as demonstrated in RTOG 0529. Proton beam therapy (PBT) offers significant dosimetric sparing of normal organs in lower dose ranges that may reduce toxicity although there is a lack of published clinical outcomes.

METHODS AND MATERIALS: We retrospectively compared acute toxicity outcomes of patients treated with CRT using protons or x-rays at our institution. Acute toxicity was evaluated per CTCAE version 4.0 criteria and was defined as during or within 90 days of PBT completion. Hematologic toxicity was not evaluable.

RESULTS: Fifty-one non-metastatic anal SCC patients received protons (n=12) or x-rays (n=39). PBT patients were older (median 69 vs. 62 years), more often female (91.7 vs. 76.9%), had more advanced T stage (T3–4: 50 vs. 18%) and less advanced N stage (N1–3: 16.7 vs. 28.2%). Median dose was similar (52.2 vs. 50.4 Gy) as was use of concurrent 5-fluorouracil/mitomycin-C (83.3 vs. 87.2%). PBT was delivered with pencil beam scanning in the supine position for all proton and 48.7% of x-ray patients. Median follow-up was 1.7 months (proton) versus 5 months (x-ray) from start of CRT. The incidence of acute G2 toxicities in proton versus x-ray patients was: GI (16.7% vs. 28.2%); GU (2.6% vs. 12.8%), skin (50% vs. 66.7%), proctitis (50% vs 51.3%). The incidence of acute G3 toxicities in proton versus x-ray cohorts was: GI (0% vs. 2.6%); GU (both 0%), skin (8.3% vs. 7.7%), proctitis (0% vs. 7.7%).

CONCLUSION: Our acute toxicity outcomes with both protons and x-rays are favorable compared to RTOG 0529. Future studies should prospectively evaluate whether PBT leads to clinically significant differences particularly for GI, hematologic, and sexual dysfunction toxicities.

Abstract 28: Organ Sparing with Proton Radiation Therapy for Pediatric Hodgkin Lymphoma

BACKGROUND: Hodgkin lymphoma (HL) is the most commonly diagnosed adolescent cancer; with survival rates exceeding 95%, attention has turned to minimizing long-term morbidity and mortality associated with chemotherapy and radiation treatment (RT). Secondary malignancy (SM) and cardiovascular disease exceed population norms as the leading causes of death in long-term HL survivors, and breast cancer risk increases in a radiation dose-dependent manner. In contrast to photons, protons lack an exit dose, therefore decreasing the volume of heart, lungs, and breasts exposed to RT; this reduction is expected to decrease late toxicities and SM.

METHODS AND MATERIALS: Between 2012–2018, 49 HL patients were treated with protons via passive scatter or pencil beam scanning techniques on an institutional review board-approved outcomes monitoring protocol. Relapse free (RFS) and overall survival (OS) were calculated using Kaplan Meier.

RESULTS: Median age at treatment was 16 years (range 11–21); 61% were female. Fifty-five percent were high risk per AHOD1331; 35%intermediate, 6% favorable, and 4% relapsed/refractory. Most patients were treated according to Children's Oncology Group protocols: AHOD0331 (n=20), AHOD0831 (n=3), or AHOD1331 (n=12). Median dose was 21.6Gy RBE (range 21–36) to the mediastinum (n=46), neck (n=34), axilla (n=5), spleen (n=4), abdomen (n=4), and bone metastases (n=2). Median ipsilateral mean breast dose was 0.51Gy; heart mean dose was 4.51Gy; ipsilateral lung V20 was 8.6%. RFS was 90% and 85% at 2 and 4 years, respectively; OS was 100%. Five patients recurred at a median of 9 months post-treatment (range 3–25) in and out of the field. There were no marginal recurrences. Median clinical follow-up was 24 months. No SMs were reported. Four patients (8%) had transient grades 1–3 toxicities potentially related to RT.

CONCLUSION: Proton treatment resulted in favorable RFS, OS, and low doses to the heart, lungs, and breasts in a pediatric HL population, with low rates of acute and late toxicity.

Abstract 30: Ten-Year Outcomes from Three Prospective Clinical Trials of Image-Guided Proton Therapy in Prostate Cancer

PURPOSE: This study reports 10-year outcomes on 3 image-guided proton therapy (IGPT) trials for prostate cancer.

METHODS AND MATERIALS: Ten-year outcomes were assessed for 3 prospective trials of IGPT including 89 low-risk (LR), 82 intermediate-risk (IR), and 40 high-risk (HR) patients. Median ages for LR, IR, HR were 64, 68, and 72 years (range, 40–88). Treatment was 78–82 cobalt gray equivalent (CGE) in 39–41 fractions; HR patients also received weekly docetaxel (20 mg/m2) during IGPT and 6 months of androgen deprivation therapy. CTCAE, v4.0, was used for toxicity scoring and the International Prostate Symptom Score (IPSS) and Expanded Prostate Index Composite (EPIC) for patient-reported outcomes. Cumulative incidences were calculated for freedom from biochemical progression (FFBP) and toxicity rates with Kaplan-Meier estimates in parentheses for FFBP. EPIC domain score changes considered significant were 4–6 (bowel); 5–7 (urinary irritative/obstructive); 6–9 (urinary incontinence); 10–12 (sexual).

RESULTS: Median follow-up was 11 years. Ten-year FFBP rates for LR, IR, and HR were 96% (96), 93% (91), and 64% (58), respectively. Twenty-five patients had progression at a median time of 9.9. 7.2, and 4.0 years for LR, IR, and HR. With PSA progression, 3 also had local, 4 had nodal, and 6 had distant progression. Eleven of 14 HR patients with progression would now be classified as NCCN “very high risk.” The rate of 10-year grade ≥3 gastrointestinal and urologic toxicity incidences were 0.7% and 4.8%, respectively. Median pretreatment and >8-year IPSS scores were 8 and 6 for LR and IR and 9 and 10 for HR. There were no significant changes in EPIC bowel, urinary irritative/obstructive, and urinary incontinence scores at >8 years compared to baseline.

CONCLUSION: Ten-year clinical outcomes with IGPT remain excellent. Local progression was rare. Most progression occurred in “very high risk” patients. Long-term (>5-year) follow-up is necessary for outcome assessment.

Abstract 31: Computer-Based Assessment of Neurocognitive Decline in Pediatric Patients Treated with Proton Therapy: A Prospective Study

BACKGROUND: Neurocognitive sequelae are well-recognized late toxicities of radiotherapy in children. In this prospective study, we assessed the effects of proton therapy on cognitive function over time in pediatric patients treated for primary central nervous system malignancies with the hypothesis that higher radiation dose to the brain would be associated with impaired cognition.

METHODS AND MATERIALS: Between 2014–19, 26 patients ages 4–21 were enrolled in this prospective trial. Patients completed neurocognitive testing using the National Institutes of Health Toolbox Cognitive Battery (TCB), a novel, validated, computer-based assessment analyzing cognitive function along seven unique domains. Testing was completed at baseline and at annual follow-up visits following completion of radiotherapy. Linear mixed effects models were used to analyze the effects of age, time since treatment, total dose, dose to critical organs at risk, and socioeconomic status on TCB metrics with significant cognitive dysfunction (SCD) defined as a decline in score of >1 standard deviation during follow up.

RESULTS: Of the 26 patients enrolled, 14 received craniospinal irradiation, and 12 were treated with involved-field radiotherapy. Median follow-up was 24 months (range:12–36 months), and at last follow-up median scores declined in overall cognitive function (96.8 v. 92.0, p=0.02) as well as executive functioning (95.8 v. 83.5, p=0.005) and working memory (105.5 v. 96.0, p=0.008) domains. Higher poverty level was associated with greater declines in executive functioning (p=0.03), but no other associations were seen with sociodemographic factors. On dosimetric analysis, brain V50 was most strongly associated with decreased overall cognitive function score (p=0.01), and patients with brain V50 >10% had a higher-odds (OR=6.38, 95% confidence interval: 1.03 – 51.78, p=0.048) of SCD.

CONCLUSIONS: In this study, higher doses of radiotherapy were associated with greater declines in neurocognitive function. By reducing integral dose, proton therapy may help mitigate these sequelae

Abstract 33*: Concurrent Pencil Beam Scanning Proton Therapy and Hyperthermia: Growing Experience with Promising Results

BACKGROUND: Hyperthermia (HT) has been regularly used as an excellent radiosensitizer with conventional radiotherapy (RT). There is a paucity of safety/efficacy data for the concurrent use of proton therapy (PT) and HT due to the lack of institutions with capabilities for both modalities. We report, herein, the largest, and growing, clinical experience with concurrent Pencil Beam Scanning Proton Therapy (PBS-PT) and HT to date.

METHODS AND MATERIALS: At our institution, PBS-PT has been utilized in over 1,400 patients, of which 30 courses/sites (25 curative, 5 palliative) have been delivered with concurrent superficial-HT in 27 patients. Histologies include sarcoma (n=11), breast (n=9), vulvar (n=1), skin (n=1), mesothelioma (n=1), ovarian (n=1), head/neck (n=1), anal (n=1), and ureteral (n=1) cancers. PBS-PT doses ranged from 36 to 70.2 Gy(RBE) (median 57 Gy[RBE]) including altered/hypo-fractionation. The BSD-500 platform was utilized for all superficial-HT administrations (median 8, range 4–28 HT sessions per course).

RESULTS: With a median follow-up of 7.4 months (range 1–31 months), concurrent superficial-HT and PBS-PT has been well tolerated. There were no acute/subacute grade 4–5 toxicities. Grade 3 toxicity arose in only 5 patients: acute desquamation (n=3), chronic lymphedema (n=2). Grade 1–2 toxicities included radiation dermatitis, pain, hyperpigmentation, and GI disturbance. Twenty-two patients (81.4%) remain alive, while 20 (74.1%) are locally controlled and 18 (66.7%) remain free of disease.

CONCLUSION: Concurrent PBS-PT and superficial-HT is well tolerated. While long-term follow-up and prospective data are needed, superficial-HT has represented a safe adjunct to particle therapy in the largest institutional experience to date with this combination.

*Previously submitted in similar form to the 2019 PTCOG58.

Abstract 36: Clinical Sensitivity of a Prototype Proton Radiography System

BACKGROUND: Proton Radiography (P-Rad) images provide information on the patient's position within the beam path in addition to the proton water equivalent pathlengths through the patient. A P-Rad can therefore be used for patient alignment as well as a pre-treatment range consistency check. A quantification of the sensitivity of these parameters was performed for a prototype P-Rad system.

METHODS AND MATERIALS: To quantify the alignment accuracy, Monte Carlo simulated P-Rad images of a brain tumor patient were created from CT image sets offset from a nominal position with known translations and rotations. Single proton positions and the residual energy were scored in the P-Rad detectors and processed through an iterative most likely path back-projection algorithm. To quantify the sensitivity to pathlengths changes, actual P-Rad images were acquired of a pediatric head phantom with a 2cm square cavity in the center. Various inserts of known WET were placed in the cavity and compared to the known thickness of the inserts.

RESULTS: Patient alignment results show that the P-Rad images could be used to obtain a positional correction vector in the Lt/Rt, Ant/Post and Sup/Inf directions by an average of 0.0mm, 0.0mm and −0.1mm with a standard deviation of 0.4mm, 0.4mm and 0.3mm respectively. Average angular accuracy in the Yaw, Pitch and Roll directions were −0.1deg, 0.1deg and 0.1deg with a standard deviation of 0.2deg, 0.2deg and 0.5deg respectively. WET consistency results of actual P-Rad images of a head phantom were able to detect pathlengths differences as small as 0.5mm with statistical certainty.

CONCLUSION: P-Rad images in the head region can be used to correct for positional offsets and quantify pre-treatment pathlengths consistency.

Abstract 39: Patterns of Care and Referral for Proton Therapy in Breast, Lung, And Prostate Cancer

OBJECTIVE: To report demographic and clinical characteristics of patients who were more likely to receive proton beam therapy (PBT) than photon therapy from facilities with access to proton centers.

METHODS: We utilized the national cancer database to identify the facilities with access to PBT between 2004 – 2015 and compared the relative usage photons and PBT for demographic and clinical scenarios in breast, prostate, and lung cancer.

RESULTS: In total, 231 facilities with access to proton centers accounted for 16,8323 breast, 39,975 lung, and 77,297 prostate cancer patients treated definitively. PBT was used in 0.5%, 1.5%, and 8.9% of breast, lung, and prostate cases. PBT correlated with a farther distance traveled and longer start time from diagnosis for each site (p<0.05). For breast, demographic correlates of PBT were treatment in the west coast (OR=4.8), age <60 (OR=1.25), white race (OR=1.94), and metropolitan area (OR=2.8). Left sided cancers (OR=1.28), T1 (OR=1.28), N2 (OR=1.71), non-ER+/PR+/Her2Neu- cancers (OR=1.24), accelerated partial breast irradiation (OR=1.98), and hypofractionation (OR=2.35) were predictors of PBT. For lung, demographic correlates of PBT were treatment in the southwest (OR=2.6), metropolitan area (OR=1.72), and Medicare insurance (OR=1.64). Higher comorbid score (OR=1.36), later year treated (OR=3.16), and hypofractionation (not SBRT) (OR=3.7) were predictors of PBT. For prostate, correlates of PBT were treatment in the west coast (OR=2.48), age <70 (OR=1.19), white race (OR=1.41), metropolitan area (OR=1.25), higher income/education (OR=1.25), and treatment at an academic center (OR=33.94). Lower comorbidity score (OR=1.42), later year treated (OR=1.84), low risk disease (OR=1.45), definitive compared to postoperative (OR=6.10), and conventional fractionation (OR=1.64) were predictors of PBT.

CONCLUSION: Even for facilities with established referrals to proton centers, PBT utilization is low, and socio-economic status may be a factor. PBT was more often used with left-sided breast and low-risk prostate cancers, without a clear clinical pattern in lung cancer.

Abstract 41: Should We Contour to the Skin? The Evaluation of Skin Dose During Pencil Beam Proton Therapy for the Chest Wall

BACKGROUND: Proton beam dosimetry enables far greater sparing of the lungs and heart than possible using photon beams but their use may be limited by skin toxicity if this is not carefully considered. This study investigated skin dose when undergoing chest wall radiotherapy with either photon or pencil beam scanning proton therapy (PBS-PT). Optimization methods in the setting of PBS-PT were used to better match skin doses encountered in photon chest wall irradiations.

METHODS AND MATERIALS: Calibrated film (in-house) and TLD flat packs from the Radiation Dosimetry Services at the MDACC were used on a CIRS phantom in the treatment geometry. Plans were developed using the RayStation TPS for both modalities. The PBS-PT plan used two enface beams with gradient matching to encompass the large field while the photon plan utilized a pair of tangential beams. Skin optimization was accomplished mainly by using a variable minimum radiological depth requirement for the Bragg peaks and evaluated using a skin contour.

RESULTS: Delivery consisting of 28 fractions of a 6 MV beam with 5 mm of bolus every other day resulted in 76% of the prescription dose at the skin as measured by TLDs. PBS-PT plans delivered 91% and 97% of the prescription dose for the skin- and non-optimized plans, respectively when measured by TLDs.

CONCLUSIONS: The dose to the skin during traditional photon chest wall radiotherapy is not 100% of the prescription and therefore such a requirement of proton plans might not be appropriate. Surface dose may be considered in order to achieve acceptable erythema and better cosmesis.

Abstract 44: Clinical Workflow to Evaluate and Incorporate a High Density Implant into Pencil Beam Scanning Proton Therapy Planning

BACKGROUND: In situations, where the high density implant can't be avoided, it is critical to identify its relative stopping power (RSP) before proton beam planning. Herein, we proposed a clinical workflow to estimate the RSP of the high density implants and incorporating it into the clinical practice.

METHODS AND MATERIALS: The RSP of a hip implant system (DePuy CeramaxTM) was investigated as well as several high density materials (Titanium, Tungsten, Platinum, and Al2O3). To estimate the RSP, we use: (1) the software package Stopping and Range of Ions in Matter (SRIM; (2) the National Institute of Standards and Technology (NIST) online program PSTAR ( and (3) If a sample implant was available, the RSP was measured. The data calculated by SRIM must be verified by at least either (2), (3), or using both, before using it in clinical plans. During the entire treatment course, the implant location is closely monitored via daily CBCT. If any change was noted, the physician was informed to review the dosimetric effect and to decide whether a re-plan was needed.

RESULTS: The RSP of the Titanium alloy stem and Ceramic cap of the hip implant system was calculated using SRIM. The result was less than 0.8% from the measured value (3.22 vs. 3.22 for the stem, 3.43 vs. 3.46 for the cap). The RSPs of other materials (Titanium, Tungsten, Platinum, and Al2O3) encountered in our clinic were estimated using SRIM and verified by PSTAR, and their discrepancies were within 1%.

CONCLUSION: The proposed clinical workflow is practically achievable, and clinically useful. This is especially important when the high density implant lie inside the target volume and therefore directing beams through target volume could be favored due to improved target volume coverage, plan complexity and robustness.

Abstract 46: Evaluating Proton Relative Biological Effectiveness for Spinal Cord Toxicity

BACKGROUND: Uncertainty in proton relative biological effectiveness (RBE) may be a cause of radiation-induced normal tissue toxicity. Within late responding tissues such as the spinal cord, risk of radiation necrosis and paralysis resulting from changes in dose delivery and biological effect are of particular concern. Quantification of proton RBE is complex, as this dynamic value is dependent upon many experimental parameters, including linear energy transfer (LET), fractionation scheme and biological endpoint. The purpose of this study was to investigate the dependence of RBE on dose and on position along the Bragg curve in a living mouse spinal cord model.

METHODS AND MATERIALS: Cervical spinal cords of female, 8 week old C57BL/6J mice were irradiated (20–80 Gy; lateral opposed beams) at low-LET (entrance) or high-LET (Bragg peak) positions along the proton curve. Animals were anesthetized and restrained in a custom fabricated set up for treatment. Endpoint was defined as onset of radiation induced myelopathy, while weight and general health changes were recorded weekly. Rotarod tests were used to evaluate motor function on a bi-weekly basis. RBE will be calculated as the ratio of the tolerance dose at 50% effect probability (grade II paresis) upon completion of the study (300 days post irradiation).

RESULTS: Acute toxicities of temporary weight loss and skin abrasions were observed in both cohorts of mice. Forelimb paralysis was manifest in the highest dose groups and accompanied by performance deficits on the rotarod. Preliminary results suggest BP treated mice exhibit a shorter latency time to paralysis.

CONCLUSIONS: Robust characterization of proton RBE in our in vivo spinal cord model is crucial for informing clinical treatment strategies that strive to account for the biological variability of proton therapy in order to reduce short- and long-term side effects to critical organs.

Abstract 47: Real Time Patient Specific Quality Assurance

BACKGROUND: Real time treatment is when the patient is imaged, planned and then treated while the patient is not moved from the treatment couch. One of the obstacles to implementation is quality assurance. While the treatment plan may be compared to an independent dose calculation before treatment to verify that the plan is of high quality, there is a challenge to confirm that the treatment machine will properly deliver the plan before treating the patient without actually treating the plan on the machine. Mayo Clinic Rochester, in cooperation with Hitachi Ltd., has developed a Real Time Patient Specific Quality Assurance process (RT PSQA) to mitigate this challenge.

METHODS AND MATERIALS: The workflow for Real Time PSQA is the same workflow as any new plan, except that the DICOM file of the treatment plan is transferred to a QA computer at the time of the Radiation Oncologist and Medical Physicist approval. The plan then enters the treatment preparation process and is eventually transferred to the treatment machine's Work List Manager (WLM) and Treatment Control Station (TCS) just before treatment. From the TCS, the machine format plan is transferred to the QA computer and the data compared to the DICOM plan for consistency and deliverability.

RESULTS: The product allows verification that the same plan that was reviewed by the Radiation Oncologist is the one that is about to be treated and verification of the integrity of the translation of the treatment field into machine code before treatment. Following these two verifications the patient is then treated.

CONCLUSION: Real time adaptive treatment planning requires new approaches to PSQA. The method presented here ensures fidelity of the real time plan at the time of treatment, eliminating the need for separate QA plan delivery and enabling real time treatment.

Abstract 48: Pilot Study on Critical Structure Dosing for Ocular Melanoma Radiation Techniques: An Analysis Of I-125 Brachytherapy Plaque and Dedicated Proton Eye Beamline Treatment Planning Dosimetry

BACKGROUND: To analyze critical structure dose distributions using I-125 brachytherapy plaques and proton plans generated with a dedicated eye beamline for ocular melanoma treatment.

METHODS: Nine ocular melanoma tumor cases were treated at our institution with I-125 brachytherapy plans (prescribed to 85Gy at tumor apex delivered over 7 days) during proton cyclotron maintenance. Clinically comparable proton plans were generated for the same tumors on the 67.5 MeV eye beamline using EYEPLAN (prescribed to 56GyE in 4 daily fractions). Maximum absolute dose (Dmax) and maximum dose as a percentage of prescription (Dmax/%Rx) were extracted for optic disc (OD), macula, and lens, with the latter metric accounting for differences in prescription values between regimens. Due to the small sample size, distributions were compared using a nonparametric Wilcoxon matched pairs test with statistical significance at p-value<0.05.

RESULTS: Tumors characteristics were as follows [range (median)]: (a) maximum clinical basal diameter [6.7–15.8mm (9.7mm)]; (b) tumor height [1.1–10.75mm (3.4mm)]; and (c) OD-tumor distance [1.2–14mm (4.6mm)]. Dmax to OD was significantly different for protons versus plaques (p=0.01). Median (range) for Dmax OD was 0 GyE(0–52.1GyE) for protons and 40Gy(10–120Gy) for plaques, with 7 proton plans sparing the disc. Dmax/%Rx OD was also significantly different, 0%(0–93) and 47.1%(11.8–141.2) for protons and plaques, respectively (p=0.01). Dmax and Dmax/%Rx to macula was significantly different between protons and plaques, (p=0.01 and p=0.03, respectively) with the latter metric at 56%(0–94) versus 64.7%(11.8–225.9), respectively. Dmax to lens was significantly different at 2.8 GyE(0–56) for protons versus 16Gy(9–90) for plaques, respectively (p=0.02); Dmax/%Rx to lens did not differ at 5%(0–100) versus 18.8%(10.6–105.9), (p=0.37).

CONCLUSION: Dedicated eye proton beam plans demonstrated significantly different dose distributions compared with I-125 plaque plans with increased critical structure sparing and lower median doses for the optic disc and macula for both Dmax and Dmax/%Rx, and Dmax to lens.

Abstract 49: Dual-Energy CT-Based Accurate Relative Stopping Power Mapping Using Deep Learning

BACKGROUND: The accuracy of dual-energy CT (DECT)-derived parametric maps is directly affected by the level of photon noise and image artifacts. Such inconsistency degrades the accuracy of the physics-based mapping technique and affects subsequent processing for clinical applications. In this study, we propose a deep-learning-based method to accurately generate a relative stopping power map (RSPM) as an alternative to physics-based dual-energy approaches.

METHODS AND MATERIALS: We manually segmented head-and-neck DECT images into brain, bone, fat, soft-tissue, lung and air, and then assigned different RSP values into the corresponding tissue types to generate a reference RSPM, which is the training target of our deep-learning model. We proposed to integrate a residual block concept into a cycle-consistent generative adversarial network framework to learn the nonlinear mapping between DECT and reference RSPM. This learning-based RSPM generation method was tested with 18 head-and-neck cancer patients. Mean absolute error (MAE) and mean error (ME) were used to quantify the differences between the generated and reference RSPM.

RESULTS: The average MAE between generated and reference RSPM was 0.031±0.004 and the average ME was 0.015±0.005 for all patients. Comparing to the physics-based method, the proposed method could significantly improve RSPM accuracy and had comparable computational efficiency after training.

CONCLUSIONS: We have developed a novel learning-based method to effectively capture the relationship between DECT data of tissue substitutes and reference RSPM, subsequently used it to generate accurate RSPM, and demonstrated its reliability. The proposed deep-learning-based approach has the potential advantages of producing unbiased and robust RSPM for proton dose calculation.

Abstract 50: Deep-Learning-Based Relative Stopping Power Mapping Generation for CBCT-Guided Adaptive Proton Radiotherapy

BACKGROUND: Compared to helical CTs, cone beam CTs (CBCT) have less accurate Hounsfield Units and degrade image quality, limiting their potential use for proton dose calculation. In this study, we developed a learning-based approach to accurately estimate relative stopping power (RSP) from to daily CBCTs, to enable future CBCT-guided online proton dose evaluation and adaptive proton planning.

METHODS AND MATERIALS: We first built a set of multiple paired training images including dual-energy CT (DECT) acquired for treatment planning and CBCT captured during the first treatment fraction. Then a deformable CBCT-DECT registration was performed to reduce anatomical differences between the images. The RSP mapping was generated using physics-based dual-energy approach to serve as training targets (ground truth). We used a cycle-consistent generative adversarial network framework with integrated residual block minimization to learn the nonlinear mapping between CBCT and DECT-based RSP map. This CBCT-based RSP generation algorithm was tested with 22 head-and-neck cancer patients with a leave-one-out cross-validation method. Mean absolute error (MAE), mean error (ME) and normalized cross-correlation (NCC) were used to quantify the differences between DECT-based and estimated RSP maps.

RESULTS: The average MAE and ME were 0.056±0.012 and −0.005±0.031 between DECT-based and CBCT-based RSP maps, and the mean NCC was 0.965±0.009 for all patients.

CONCLUSION: We have developed a novel learning-based method to generate accurate RSP mapping from daily CBCT imaging and demonstrated its reliability. The absolute value agreement and image similarity between DECT-based and CBCT-based RSP maps warrant further study and development of a CBCT-guided adaptive workflow for proton radiotherapy.

Abstract 52: Carbon Ion Dose Constraints for Organs at Risk at MedAustron

MedAustron is a dual particle therapy facility that started clinical operation with protons in December 2016 and is on schedule for the first patient to be treated with Carbon Ion Radiotherapy (CIRT) in July 2019.

Dose constraints for organs at risk (OARs) in carbon ion radiotherapy have been established in the clinical routine of already treating facilities but are, up to now, not completely harmonized.

In CIRT, as compared to photons and protons RT, it is less straightforward to derive dose constraints for OARs from published data especially because of the different RBE models employed.

Dose constraints based on the analysis of clinical outcome in patients treated with CIRT in Japan are available for: visual pathways, brainstem, brain, skin, rectum, duodenum and maxillary bone. These data are based on the Kanai semi-empirical / modified microdosimetric kinetic model (MKM).

Dose constraints based on German clinical data from patients treated with the local effect model (LEM-I) are available for the temporal lobes. Moreover unpublished constraints routinely used in clinical practice in Germany at the Heidelberg Ion Therapy Center (HIT), In Japan at National Institute of Radiological Sciences (NIRS) and at Gunma University (GHMC) and in Italy at National Center for Hadrontherapy (CNAO) have been generously shared by these Institutions.

The translation of mMKM constraints into LEM-I values is being extensively studied in CNAO. Ongoing activities encompass the analysis of toxicity and pattern of relapse in patients treated with Japanese schedule but using LEM-I and recalculation of LEM optimized plans with mMKM.

All these data were reviewed and were used to estimate dose constraints to be used in MedAustron with CIRT employing both a fractionation similar to HIT (3 Gy RBE per fraction at 5 fractions per week to a total of 20-22 fractions) and a fractionation similar to Japanese centers after conversion for different RBE models as performed in CNAO (4.1–4.8 Gy RBE per fraction at 4 fractions per week to a total of 16 fractions).

Abstract 59: Outcomes of Intracranial High-Grade Glioma Treated With Proton Therapy

BACKGROUND: Data on the use of proton therapy for treatment of pediatric high grade glioma is limited. The purpose of this study is to report 4 year disease control.

METHODS AND MATERIALS: Patients aged ≤21 years with nonmetastatic intracranial high grade glioma treated with proton therapy were enrolled in a prospective outcome study. The Kaplan-Meier method was used to calculate survival and control rates.

RESULTS: From 2008 to 2019, 29 consecutive patients with a median age of 11.3 years (range, 2–20.5 years) received a median dose of 59.4GyRBE (range, 54–59.4GyRBE) using passive scatter proton therapy, most commonly with a 1cm initial CTV margin and no CTV expansion for the reduction. Majority of patients had gross total resection (GTR) (79%), while 21% had biopsy/subtotal resection (STR). 83% received chemotherapy. Median tumor size was 3.6cm (range, 1.5–7.6cm). The most common histologies included anaplastic astrocytoma (n=8) and glioblastoma multiforme (n=9). 52% of patients had WHO grade 3 tumors, 34% WHO grade 4, and 14% high grade not specified. With a median follow-up of 2.3 years for all patients and 3.7 years for living patients (range, 0.3–10.9), 4 year overall survival was 62% (95% CI, 41–79%) and 4 year progression-free survival was 52% (95% CI, 33–71%). 4 year local control was 53% (95% CI, 33–73%). 2/6 patients with STR/biopsy and 14/23 with GTR remained free of disease. Median time to local recurrence was 1.1 years (range, 0.05–6.2years). Only one patient developed distant disease without local progression. Two patients developed symptomatic local progression during radiotherapy.

CONCLUSION: Proton therapy for intracranial high grade glioma allows for conformal radiation delivery without marginal failures. Local control remains a challenge. Improved understanding of tumor biology and molecular composition may assist in identifying children with better prognoses who may benefit from advanced radiotherapy techniques that maximize normal tissue sparing.

Abstract 62: Quantification and Verification of Skin Dose for Patients with Head and Neck Cancer Receiving Intensity Modulated Proton Therapy

PURPOSE: Acute skin reactions in patients with head and neck cancer (HNC) receiving radiotherapy are commonly observed. The grade of radiation dermatitis clinically varies according to dose. Concerns for greater skin reaction with proton therapy in head and neck patients have been raised. The purpose of this study is to quantify and verify skin doses relative to the underlying target volume (TV) prescription for patients with HNC receiving intensity modulated proton therapy (IMPT) to the bilateral neck.

METHODS AND MATERIALS: Skin doses were measured at specified locations using TLDs. IMPT plans used a SIB approach (dose range: 5600-7000 CcGE), had been optimized for TV coverage without a specific skin dose constraint, and used three fields (RAO, LAO, and PA). TLD powder packets were placed by a single radiation oncologist at four neck locations: R/L upper (level II) and R/L lower (level IV). TLDs were read following established dosimetry protocol.

RESULTS: The skin doses in the treatment plans were found to be linearly dependent on the distance of the point of interest to the edge of the TV (d). It was found to be 100% or more of the TV dose for d < 2 mm, reducing to about 90% at d=0.5 cm, and to about 70% at d=1.3 cm. TLD doses and the estimated doses from treatment plans agreed within +/−10%, with mean of deviation being close to 1%.

CONCLUSIONS: The skin doses in treatment plans compare reasonably well with TLD measurements on the skin surface. Additionally, skin dose sparing relative to the underlying TV prescription was observed, but varied according to its distance from the TV and the shape of the TV. Planning strategies to further reduce IMPT skin dose without compromising TV coverage are being explored.

Abstract 63: Treatment Planning Strategies to Mitigate the Relative Biolgical Effectiveness Uncertainty in Proton Therapy for Targets in the Brain

BACKGROUND: Uncertainty in the value of relative biological effectiveness (RBE) of proton beam, especially in the distal dose fall off region, is an important consideration in designing safe treatment plans for proton therapy for many brain cancer patients where organs at risk (OARs) are located in the close proximity of target volumes. The purpose of this study is to explore novel strategies that can be used to mitigate the effect of RBE uncertainty in the proton therapy treatment plans for targets in the brain.

METHODS AND MATERIALS: Some of the treatment planning strategies applied to reduce the effect of RBE uncertainty are the use of: (1) suitable beam angles to minimize the dose to OARs from the distal fall off region of the field dose, (2) beam specific planning target volumes (BSPTVs) to force the beam stop further away from the OARs, (3) reduced biological effective doses (BEDs) to OARs by considering approximate higher RBE values in the longitudinal penumbral (LP) region of proton beam, and (4) multiple plans with different beam angles to spread the dose from LP region to different locations.

RESULTS: It was found that the use of the above strategies in designing treatment plans helped to reduce the BED of OARs so as to meet their required dose volume constraints. In many cases, use of multiple plans with different beam angles led to reduced doses to the OARs compared to one plan with limited beam angles while keeping the treatment delivery time same.

CONCLUSIONS: A combination of beam angle selection, use of BSPTVs, use of multiple plans and use of reduced BED dose constraints on OARs located in the longitudinal penumbra region of the proton beam are found to mitigate the RBE uncertainty in proton therapy plans for targets in the brain.

Abstract 76: Interplay Evaluation of High Frequency Percussive Ventilation (HFPV) for Motion Greater than 10 MM During Proton Beam Scanning

BACKGROUND: Layer and volumetric repainting can mitigate interplay effects, however they become ineffective for respiratory motion beyond 10 mm. In this retrospective study, heterogeneity effects of HFPV for motion greater than 10 mm is investigated.

METHODS AND MATERIALS: Five volunteers' free- and HFPV- respiratory chest wall motion curves were imported into a motion platform holding GAFChromic film which cycled accordingly. Their peak- to-peak free breathing (FB) amplitudes were >10 mm with mean frequency of 6.397 Hz (HFPV) and 0.2318 Hz (FB). Two matched fields were used to deliver a 3GyRBE 12×12cm square pattern. To evaluate uncertainties of hot and cold spot distributions, the plan was optimized as one Bragg peak. One motion curve was used to deliver the plan 5 times under free- and HFPV- motion. The plan was re-optimized to deliver 3GyRBE across a 7.5 cm SOBP. Films were placed at 6, 9, and 12 cm depths for each motion curve. Static films were irradiated for baseline. No repainting was performed.

RESULTS: Compared to free-breathing curves, hot and cold spots reduced from >30 % to <10 % using HFPV. Average Gamma pass rates at 3%3mm for HFPV- and free- breathing were 99.60 % (SD: 0.5) and 76.62 % (SD: 0.65), respectively. Although heterogeneity effects along the SOBP were greater for free-breathing motions, films placed at depths of 12 cm (Gamma: >50 %, SD: >9.0) were statistically different relative to those at 6cm (Gamma: >70 %,SD: >6 %) and 9cm (Gamma: >70 %, SD: >8 %) indicating heterogeneity effects for motion greater than 10 mm can vary along the SOBP while no significant difference was noticed for HFPV.

CONCLUSIONS: HFPV is a novel technique that significantly reduces heterogeneities from motion without repainting. HFPV is an effective technique that can be used to mitigate interplay effects that exist along the SOBP for motions larger than 10 mm.

Abstract 81: Proton Therapy for Pediatric Bladder Rhabdomyosarcoma

BACKGROUND: Pediatric patients with rhabdomyosarcoma (RMS) of the bladder rely on definitive radiotherapy to achieve local control with bladder preservation. Proton therapy has been shown to confer dosimetric advantages over conventional photon radiotherapy. Here we describe the simulation, planning, and treatment techniques used to manage pediatric bladder RMS patients with proton therapy.

METHODS AND MATERIALS: With IRB approval, 5 pediatric patients <18y with bladder RMS, treated with proton therapy, were identified. Clinical and radiation treatment planning data were retrospectively abstracted from the electronic medical record.

RESULTS: Median age at simulation was 2.4y (range 1.3–4.9y). Four of 5 presented with de novo disease, and 1/5 presented with a bladder recurrence of a vaginal RMS previously managed without radiotherapy. All had embryonal or fusion-negative disease. Among the 4/5 with de novo presentation, all had Group 3, 2/4 had Stage 2, and 2/4 had Stage 3 disease. In 4/5 patients, simulation and treatment setup was performed with the bladder filled to 50–70 mL via an 8 French urinary catheter, which was then clamped. Four of 5 patients were planned to a total dose of 50.4 CGE; 1/5 was treated to 41.4 CGE. The CTV was asymmetrically expanded to account for proton distal range uncertainty. The bladder, femoral heads, large bowel, and rectum were contoured. Beam arrangements include left and right posterior obliques in 4/5, and 2 left posterior obliques in 1/5. Among the 4/5 patients treated to 50.4 CGE, mean dose to the bladder was 36.6 CGE (±13.5), femoral heads 1.3 CGE (±1.0), and rectum 44.4 CGE (±10.3); average maximum dose to the large bowel was 4.1 CGE (±3.1).

CONCLUSIONS: With attention to simulation, planning, and daily setup, proton therapy can offer pediatric RMS patients accurate, reproducible target coverage and excellent organ-at-risk sparing.

Abstract 82: Impact of Proton Technology Evolution on Patient Volumes at a Large Academic Medical Center

BACKGROUND: Proton therapy has undergone dramatic technical evolution with regard to treatment delivery in the past decade. These advancements, including the evolution from double scattered proton therapy to pencil beam scanning, as well as improved imaging, have expanded the ability to treat more disease sites. This study evaluates the impact of technology changes on both proton therapy and conventional x-ray radiation patient treatment volumes at a center with both modalities available under one roof.

METHODS AND MATERIALS: Between 1/2010 and 2/2018, 21,482 patients were treated with radiation at the Perelman Center for Advanced Medicine protons and/or conventional radiation. Proton therapy was delivered to 5416 patients. Data for each patient were recorded through ARIA and interfaced with Tableau. Volumes of new patient starts (NPS) were tracked by modality, provider, and disease site across multiple fiscal years. Analysis of changes in NPS was performed based on the trends and addition of new technologies.

RESULTS: There has been a 64.2% increase in NPS volume since the opening of the Roberts Proton Center at the main academic site in 2010. As more treatment rooms were opened and the technology was updated, the number of proton NPS gradually increased, until reaching a plateau of about 820 NPS per year. Conventional volumes have increased 20.2% during the same timeframe at this location. Technology advancements have expanded disease site utilization of proton therapy diversifying and expanding treatment from 7 original disease sites in 2010 to 14 in 2018. For example, head and neck cancer NPS volumes increased from 2% to 12.7% with the addition of PBS on the gantry.

CONCLUSIONS: Our analysis shows that the addition of proton therapy to a conventional radiation department increases the overall NPS volumes, having a positive impact on conventional treatment volumes. This volume increase appears to be related to the technology advancements in PBS and imaging instruments.

Abstract 83: Insurance Coverage for Adjuvant Proton Therapy in the Definitive Treatment of Breast Cancer

BACKGROUND: To determine which factors influence insurance approval for patients seen in consultation following breast cancer surgery for whom adjuvant proton therapy (PT) was recommended.

METHODS: 131 insured patients were seen in consultation April 2014–November 2018 and recommended adjuvant PT as part of definitive treatment. 108 patients (82%) had Tis-T2 cancers; 23 (18%) had T3–T4 malignancies. Laterality included right, 26 patients (20%); left, 105 (80%). Four patients with bilateral cancers were lateralized and staged according to the more advanced side. Forty-three patients (33%) had N0 cancers; the remainder had N1–N3. Adjuvant chemotherapy was administered to 97 patients (74%); 72 (55%) received adjuvant hormonal therapy. Insurance status included commercial, 76 patients (58%); Medicare, 41 (31%); and Medicaid, 14 (11%). Ninety-six (73%) had policies that “covered” PT. Insurance “coverage” for PT did not assure approval for PT treatment nor did lack of “coverage” mean PT would not be approved. Factors for multivariate analysis of predictors for insurance approval included T stage (‘Tis-T2 vs T3–T4); N stage (N0 vs N1–N3); laterality (left or bilateral vs right); insurance type (commercial vs Medicare/Medicaid) combined with potential insurance coverage (covered vs not covered); and time period (2014–2016 vs 2017–2018).

RESULTS: Medical review was required for 73 patients (56%); peer-to-peer review for 20 (15%); comparative dosimetry for 34 (26%). One patient (1%) appealed the insurance denial to a federal administrative law judge. Insurance approval stratified by insurance type and coverage included commercial-covered, 52/52 patients (100%); Medicare/Medicaid-covered, 41/44 patients (93%); commercial-not covered, 16/22 patients (73%); and Medicare/Medicaid-not covered, 7/13 patients (54%). Multivariate analysis revealed the following: T stage, p=0.4268; N stage, p=0.4046; laterality, p=0.1070; insurance type combined with potential coverage, p<0.0001; and time period p=0.1647.

CONCLUSION: Laterality and stage were not associated with approval. The only parameter significantly associated with approval was insurance type combined with potential coverage.

Abstract 86: Improving Proton Plan Robustness Evaluation Standardization and Efficiency Through Scripting

BACKGROUND: Robust perturbations are a key component in pencil beam scanning proton therapy plan evaluation. A physicist can evaluate the isocenter shifts, rotations, and range perturbations for a treatment plan to assess the sensitivity of the proton plan to various uncertainties. The production and evaluation of these perturbations can be tedious and time consuming. We have developed a script to automate the calculation of at least 18 unique perturbations and output the DVH data for targets and OARs to an html file for quick and efficient physics evaluation.

METHODS AND MATERIALS: We developed a Python script, run in RayStation, that allows the user to choose setup shifts, range uncertainty, setup rotations, and independent beam isocenter shift perturbations. The user also chooses the targets and OARs to be exported. The script then calculates each of the perturbations in RayStation and exports the data to a JSON file that can be read in html. The output displays each target and OAR DVH for every perturbation on a single graph. The user can turn target and OAR DVH visualization on and off, as well as turn perturbations on and off. The data is also summarized in a table, visualizing the max dose and max dose deviation from nominal for each OAR. For each target, the max dose, max dose deviation from nominal, D95%, and D95% deviation from nominal are shown.

RESULTS: This script has standardized robust perturbation calculation and evaluation across all of dosimetry and physics. Dosimetry can visualize robustness before a plan is complete, allowing more efficient plan improvements and reducing plan failure by physics. Physics evaluation time has been significantly reduced.

CONCLUSIONS: This robust evaluation script has reduced plan failures following optimization, improved perturbation standardization, and reduced evaluation time for dosimetry and physics.

Abstract 92: Feasibility of Predicting Proton Stopping Power Ratio Using the MRI-Measured Material Hydrogen Density

BACKGROUND: Uncertainties associated with CT-derived proton stopping power ratios (SPR) are a limiting factor in exploiting the full benefits of proton therapy. We investigate the feasibility of a model that predicts SPR using MRI-measured material hydrogen densities.

METHODS AND MATERIALS: A model (SPR-H) that related the medium's hydrogen density with its SPR was constructed by considering twenty-two materials taken from the National Institute of Standards and Technology (NIST) database. Subsequently, the model was used to predict the proton range and water equivalent thickness (WET) for nine tissue-surrogate materials of CIRS phantom, and the predicted values were compared with measurements using a multilayer ionization chamber (MLIC), as well as the treatment planning system (TPS) calculated WET. Additionally, a phantom setup was designed and irradiated and for each plug, ion-chamber measured dose was compared to the TPS dose calculations based on SPR-H model. For further validation, proton-density and T2-weighted MRI scans of twelve salt-water solutions with different salt concentrations (hydrogen densities) were used by the SPR-H model, and the model- predicted SPRs were compared with their corresponding MLIC measurements.

RESULTS: MRI pixel values of both scans correlated well with the measured SPRs. For all salt-water solutions and tissues-surrogates (except lung), model predicted SPR and WET were within 3% and 2% of the measurements and TPS predicted values. Additionally, the SPR-H predicted and measured doses were within 1.5% agreement. Larger deviations of 9% for SPR, and 2.6% for dose were observed for lung-tissue.

CONCLUSIONS: SPR predicted by SPR-H model can facilitate dose calculation while reducing range uncertainty in proton therapy.

Abstract 98: Evaluations of a Flat-Panel Based Compact Daily QA Device for Proton Pencil Beam Scanning (PBS) System

BACKGROUND: To evaluate the clinical usability of a flat-panel based compact PBS daily QA device for constancy measurements of beam spot characteristics, beam energies, flatness of Spread-out-Brag-peak (SOBP), output, X-ray/proton coincidence, and area uniformity.

METHOD AND MATERIALS: The phantom device is composed of a 20×20cm2 flat-panel imager mounted on a portable frame with removable/re-configurable modules (high density plastic blocks dedicated for energy checks of 100, 150, 200 MeV protons and SOBP flatness checks, RW3 block for PPC05 chamber based output checks). X- ray/proton spot coincidence, spot location accuracy, and area uniformity checks can be performed on open area of the imager. For x-ray/proton spot coincidence checks, 3 types of radio-opaque BBs were evaluated. For spot location checks, maps of spots with 1mm, 2mm and 3mm shifts were acquired and compared to the original maps. For energy checks, maps were acquired and analyzed for modified plans with 0.1 MeV increase/decrease from the 3 energies for a continuous energy degrader system and for 1, 2 and 3 energy steps increases/decreases for a stepped energy degrader system. Detector quenching effect was evaluated with different proton energies. For SOBP flatness evaluations, maps were acquired using a full modulation SOBP delivered through different sections of the wedge block.

RESULTS: The daily QA device detected millimeter changes of spot location and detected 0.1 MeV energy changes with its energy blocks; up to 8% quenching effect was observed from 100 to 227 MeV and can be corrected to obtain water equivalent results. Full-modulation SOBP delivered through thicker half of the wedge block mitigated the impact of quenching effect and rendered flat representations of the SOBP. Area uniformity is within 2%.

CONCLUSIONS: The flat-panel based compact daily QA device is capable of efficient daily checks of x-ray/proton coincidence, beam spot location and energy, area uniformity, and flatness of SOBP.

Abstract 99: The Impact of Dose Algorithms on Tumor Control Probability in Intensity-Modulated Proton Therapy for Breast Cancer

PURPOSE: To evaluate the dosimetric and radiobiological impact of PB algorithm versus MC algorithm in intensity-modulated proton therapy (IMPT) plans for breast cancer treatment.

METHODS AND MATERIALS: Twenty (20) breast cancer patients (stage T1–T2, post-mastectomy or post-lumpectomy) who received adjuvant proton IMPT radiotherapy to the breast/chest wall and regional lymphatics were included in this study. For each patient, 2 IMPT plans were generated: a PB-optimized plan and a MC-optimized plan. The radiobiological and dosimetric impact of the dose algorithms was assessed. The Poisson Linear-Quadratic model was used to estimate the tumor control probability (TCP). The influence of the model parameter uncertainties on the TCP was tested against different sets of published model parameters. In addition, we also studied the influence of α/β ratios.

RESULTS: The PB-optimized plans significantly under-dosed the target as compared to the MC-optimized plans. The median (range) differences in CTV D95% and CTV Dmean were 3.8% (2.4% - 6.2%) and 2.4% (1.0% - 3.8%) of the prescription dose. The median (range) difference in CTV V95% was 20.8% (0.9% - 41.8%). The TCP was lower in the PB-optimized plans than the MC-optimized plans. The α/β ratios has minimal influence on the calculated TPS. The median (range) of the TCP differences (ΔTCP) were 4% (2% - 6%), 3% (2% - 5%), and 2% (1% - 3%), respectively, when calculated using 3 different model parameter sets. The ΔTCP correlated with the CTV dose difference, and moderately correlated with the CTV volume.

CONCLUSION: Due to the inaccurate dose modeling, PB-optimized plans under-dose the target and therefore yield a lower TCP compared to MC-optimized plans in breast IMPT. The magnitude of the resulting difference in TCP reached 6% in our study.

Abstract 100: Spot-Scanning Proton Arc Therapy (SPArc) for Improved Optic Pathway Sparing in Paranasal Sinus and Nasal Cavity Cancer Treatment

INTRODUCTION: Paranasal sinus cancers are challenging to irradiate due to proximity to eyes, lenses, optic nerves, chiasm and brainstem. These tumors often have adverse pathologic features, necessitating high radiation doses for cure, which may exceed tolerance of the optic pathways. This is the first known study of paranasal sinus and nasal cavity cancers to evaluate potential dosimetric improvements by using a novel proton therapy modality: Spot-Scanning Proton Arc (SPArc).

METHODS AND MATERIALS: Ten patients with sino-nasal cancer undergoing high dose (60–66 GyE) Intensity Modulated Proton Therapy (IMPT) based on the Single-Field Optimization (SFO) were re-planned with SPArc technique, using Monte Carlo dose calculation. Clinical IMPT plans using multiple static fields, including vertex beams, were compared to SPArc plans using an axial and/or vertex arc with 2.5 degree control point sampling frequency.

RESULTS: With similar target coverage, maximum point doses (in GyE, to 0.03 cm3) for organs at risk were as follows for SFO IMPT versus SPArc plans, respectively: Ipsilateral Lens 12.14 vs 3.71, Contralateral Lens 7.63 vs 2.25, Ipsilateral Eye 46.15 vs 36.55, Contralateral Eye 35.59 vs 27.59, Ipsilateral Optic Nerve 53.69 vs 41.52, Contralateral Optic Nerve 49.44 vs 36.85, Optic Chiasm 39.49 vs 31.32, Brainstem 46.02 vs 31.01 GyE; all with p<0.05.

CONCLUSIONS: SPArc has the potential to significantly reduce maximal dose levels to the eyes by approximately 8 GyE, and to the optic nerves by 12 GyE beyond what was clinically delivered with IMPT plans. SPArc is capable of providing new and clinically meaningful optic pathway sparing, potentially reducing the risk of cataracts and vision loss.

Abstract 101: Effectiveness of Motion Interplay Effect Mitigation Using Spot-Scanning Proton Arc (SPArc) Therapy in Lung Stereotactic Body Radiotherapy (SBRT): A Quantitative Comparison Study Based on a Digital Phantom

PURPOSE: This study is to quantitatively evaluate the effectiveness of motion interplay mitigation via spot-scanning proton arc (SPArc) therapy for lung stereotactic body radiotherapy (SBRT).

METHODS AND MATERIALS: A set of digital lung tumor (diameter:4cm) phantoms 4DCT with different breathing induced motion in superior inferior (SI) direction (with amplitudes: 5, 10, 15 and 20mm) were created to mimic mobile lung SBRT targets. Two-field Intensity Modulated Proton Therapy (IMPT) plans were generated with single field optimization (SFO) technique and SPArc plans were generated using a partial arc from 180 to 30 degree with sampling frequency of 2.5 degree. Both plans used the same robust optimization parameter with ±3.5% range and 5mm setup uncertainties. 6000cGy relative biological effectiveness [RBE] was prescribed to internal target volume (ITV) in 5 fractions. To assess the breathing induced interplay effect, the 4D dynamic dose was calculated by synchronizing the breathing pattern with the simulated proton machine delivery sequence. Volumetric (IMPTvol_rep) and iso-layered (IMPTiso_rep) repainting technique using IMPT were compared with SPArc treatment respectively.

RESULTS: Target coverage degraded as the tumor motion amplitude increased. SPArc significantly mitigated the interplay effects in comparison with the IMPT without volumetric repainting through all the different motion amplitudes. More specially, the average target D99 degradation were 2.51% (IMPT) vs 0.0% (SPArc) (p=0.001), 4.01% (IMPT) vs 0.10% (SPArc) (p<0.001), 6.61% (IMPT) vs1.29% (SPArc)(p<0.001), 8.40% (IMPT) vs 1.70% (SPArc) (p<0.001) respectively. Target coverage can be compensated by increasing the total number of repainting in IMPTvol_rep and reducing the maximum MU per spot in IMPTiso_rep. The study found that the effectiveness of such motion mitigation using SPArc was about 5 to 7 times of IMPTvol_rep, as well as from 1.0 to 0.4 MU per spot at 2cm motion amplitude. With increased number of repainting, the deliver time increased to 698s in IMPTvol_rep compared to 490s in SPArc.

CONCLUSIONS: SPArc is capable to effectively mitigate the interplay effect in lung SBRT depending motion amplitude.

Abstract 109: Status on Carbon Ion Commissioning at MedAustron Ion Therapy Center

BACKGROUND: MedAustron is Austria's synchrotron-based dual particle therapy facility under operation with protons since December 2016. Until June 2019, 400 patients were successfully treated in 2 treatment rooms using horizontal and vertical fixed-beam lines. By early 2019 carbon ion acceptance and medical commissioning started. We report on first commissioning results and specificities of MedAustron's carbon ion therapy system.

METHODS AND MATERIALS: Dosimetry equipment was commissioned for carbon ions and used to measure the following beam properties: 242 energies from 120–402.8MeV/n (ranges in water from 2.9cm–27cm in 1mm steps using 2 crossed 1D Ripple Filters), spot sizes (in FWHM at isocenter in air) from 6mm to 10mm. The patient positioning system was designed for non-isocentric, air-gap reduced setups. This allows decreasing treatment penumbra and reducing uncertainty for pencil beam algorithm-based dose calculation using a range shifter (RaShi) mounted in the nozzle. Hence, the beam delivery system as well as the TPS using the LEM-I RBE-model were commissioned for different air gaps from 7cm to 65cm.

RESULTS: Measured ranges in water at isocenter agreed within 0.3mm with the specified. Circular shaped spots (roundness tolerance; 10%/1mm) could be achieved for the whole air gap range. TPS commissioning showed very good results for setups close to the nozzle with and without RaShi. Only for isocentric treatments the carbon pencil beam algorithm cannot model RaShi beams with sufficient accuracy and hence only non-isocentric setups (reduced air gap as for protons) will be used.

CONCLUSION: The commissioning results for the fourth European carbon facility are very promising and the process is on time. TPS commissioning results in non-isocentric patient setup are in very good agreement with measurements. It is the first time that the carbon functionality within RayStation 8B will be clinically used and hence final integrative checks as well as the specifically developed dosimetric e2e audit procedures are performed. Patient start is expected in the first week of July.

Abstract 110: Enhancing LET in Proton Beam Therapy for Pancreas Cancer

PURPOSE: Proton beam therapy treatment planning can be impacted by linear energy transfer (LET). Higher LET is correlated with higher radiobiological effectiveness. This work seeks to investigate the feasibility of enhancing the average LET in the gross tumor volume (GTV) for pancreas cancer treatment with intensity modulated proton therapy.

METHOD AND MATERIALS: The LET enhanced treatment planning optimization was performed with a gradient based and iterative optimization algorithm. Two aspects contributed to LET enhancement. 1) Extra beams from different directions. Additional beams permitted putting Bragg peaks in GTV while maintaining prescription dose. 2) Proton beam spot was iteratively weighted by LET ratio between target and organs at risk during optimization. The resulting LET enhanced optimization plan was compared with the conventional pancreas proton plan that was used to treat the patient. The conventional plans were optimized with two posterior oblique fields and no LET weighting in optimization. All dose and LET was computed by an in-house Monte Carlo.

RESULTS: Average LET could be enhanced from ~2keV/um to 4–5keV/um. The magnitude of LET enhancement depended on GTV size, GTV location with clinical target volume (CTV), robustness of the plans, etc. When additional anterior beams were utilized, LET enhanced plans had additional dose delivered to bowels at low dose level. The highest LET enhancement was achieved in plans with worst robustness to range uncertainties.

CONCLUSIONS: LET enhancement in the GTV was feasible for pancreas cancer proton beam therapy. Various factors including plan robustness and low dose spillage should be taken into account during planning.

Abstract 111: Evaluation of the Gain in Using Higher Order Components of the Thoraco-Abdominal Surface to Estimate Volumetric Tidal Flow

BACKGROUND: In the absence of motion management, respiratory tumor motion and the presence of cycle-to-cycle variations greatly degrades the accuracy of radiation targeting. By measuring the volumetric tidal flow, spirometry has been shown to be an accurate surrogate for internal displacement. However, spirometers are prone to drift. We investigate the performance in estimating volumetric tidal flow when integrating higher order information associated with monitoring the entire thoraco-abdominal surface rather than the commonly used technique of monitoring a single patch/point.

METHOD AND MATERIALS: Breathing patterns of five healthy volunteers was recorded simultaneously using two common, non-invasive breathing methods: volumetric tidal flow using spirometry (SDX, Dyn'R, France), and abdominal height displacement with Real Time Position Management (RPM, Varian, USA). Simultaneous to the two measurements, a prototype, research version, optical surface imaging system (AlignRT, Vision RT, UK), was used to capture a dense point- cloud of the thoraco-abdominal surface with a rate of 15fps and reconstructed as a watertight surface on a rectangular grid to perform a principal component (PC) analysis. The RPM signal, and the first three components of the surface were normalized. The first ten seconds of the measurement was used to train a principal least square model to estimate volumetric tidal flow for the remaining 100 seconds.

RESULTS: Both surface surrogates (RPM and VisionRT) showed a high degree of correlation with tidal flow. In comparison to the RPM, a three PC surface model leads to a reduction in estimated tidal flow RMSE by as much as one order of magnitude. One volunteer exhibited a one dimensional breathing pattern with the first surface PC correlated with tidal flow correlated (RMSE=0.001).

CONCLUSIONS: Our study suggest that monitoring the entire surface rather than a single point can lead to improvements when estimating tidal flow, and hence represents a more accurate breathing metric.

Abstract 112: Evaluation of Stopping Power Ratio Prediction of Lung Tissue from Dual-Energy CT

BACKGROUND: The accuracy of stopping power ratio (SPR), particularly for low density materials such as lung tissue, is a limiting factor in proton therapy plan robustness. We investigate the potential improvements associated with the use of dual-energy CT (DECT) for estimating SPR of lung tissue samples.

METHOD AND MATERIALS: Single energy CT (SECT) and DECT scans of six sheep and one cow lung tissue samples were acquired on a Siemens SOMATOM Definition Edge CT. SPR images were created for each lung samples from the DECT generated relative electron densities (ρe) and effective atomic numbers (Zeff). To calculate water equivalent thickness (WET), the generated SPR and acquired SECT images were imported into Eclipse (v13.7, Varian Medical Systems, Palo Alto, CA) treatment planning system (TPS). The SPR-to-SPR (straight line with a slope 1) and clinical HU-to-SPR calibration curves were applied to the SPR and SECT image sets, respectively to calculate water equivalent thickness (WET) and SPR of each lung tissue sample. For each sample, the calculated SPR values estimated from SECT and DECT scans were compared with the measurement performed using a multilayer ionization chamber (MLIC).

RESULTS: For all lung tissue samples, for SECT the TPS WET was less than the measured WET by 15% to 28% (average=21.0%). For DECT, the TPS calculated WET values were within ±8% (average=3.1%) of the measurement. Similarly, DECT derived SPR values agreed with the measurements to within ±8% (average=5.22%) for sheep and 0.54% for cow lung tissues, whereas SECT derived values (using HU-to-SPR calibration) deviated from measurement by as much as 38% (sheep: <38%, average=28.34% and cow =18.26%).

CONCLUSION: Our study suggests that using DECT can improve the accuracy of SPR and WET predictions for low density tissues such as lung.

Abstract 113: The Importance of Verification Computed Tomography Quality Assurance Scans in Patients Treated with Spot-Scanning Intensity Modulated Proton Therapy for Head and Neck Cancers

PURPOSE: To assess the role of CT-QA scans for IMPT when treating head and neck malignancies and to determine risk factors associated with the need for adaptive replanning.

METHODS AND MATERIALS: A prospectively collected quality improvement study of patients with head and neck cancer treated using spot-scanning IMPT who underwent weekly verification CT-QA scans. Kaplan-Meier estimates were used to determine the cumulative probability of an adaptive re-plan by week. Risk factors associated with adaptive re-planning were determined using univariate and multivariate cox models. Logistic regression was used to determine odds ratios.

RESULTS: Of the 160 patients, 79 (49.4%) had verification CT-QA scans which prompted an adaptive re-plan. The cumulative probability of a re-plan by week 1 was 13.7% (95%CI: 8.82–18.9), week 2, 25.0% (95%CI: 18.0–31.4), week 3, 33.1% (95%CI: 25.4–40.0), week 4, 45.6% (95% CI: 37.3–52.8), and week 5/6, 49.4% (95%CI: 41.0–56.6). Predictors for adaptive re-planning were sinonasal disease site (UVA:HR 1.82, p=0.0443; MVA:HR 3.64, p=0.0303), advanced stage disease (UVA:HR 4.68, p=0.0016; MVA:HR 3.10, p<0.05), dose >60 GyE (RBE 1.1) (UVA:HR 1.99, p=0.0035; MVA:HR 2.20, p=0.0079), primary disease (UVA:HR 2.00 vs. recurrent, p=0.0133; MVA:HR 2.46, p=0.0138), concurrent chemotherapy (UVA:HR 2.05, p=0.0023; MVA not SS), definitive intent treatment (UVA:HR 1.70 vs. adjuvant, p=0.0179; MVA not SS), bilateral neck treatment (UVA:HR 2.07, p=0.0340; MVA not SS), and higher number of beams (5 beam UVA:HR 5.55 vs. 1 or 2 beams, p=0.0157; MVA not SS). Maximal weight change from baseline was associated with higher odds of a re-plan (≥3kg OR 1.97, p=0.0438; ≥5kg OR 2.13, p=0.0246).

CONCLUSIONS: An adaptive replan was required for nearly half of patients undergoing IMPT for head and neck malignancies because of unacceptable deviations in the dose distribution. Weekly CT-QA scanning for head and neck cancer patients treated with spot-scanning IMPT is essential.

Abstract 115: Evaluation of U-Net Based Synthetic CT for MRI-Only Treatment Planning of Prostate Cancer Proton Therapy

BACKGROUND: Magnetic resonance imaging (MRI) has been increasingly utilized for target delineation, treatment-response evaluation and plan adaptation of prostate cancer radiotherapy. The purpose of this study is to evaluate the feasibility and accuracy of a deep-learning based synthetic CT(S-CT) generation method for MR-only proton treatment planning of prostate cancer.

METHODS AND MATERIALS: A 2-D U-net was trained from paired CT and T2-weighted MR images of 36 prostate patients to generate S-CT from MR. Ten additional patients, randomly selected from previous-treated patients in our institution, were included in this study. S-CTs were generated from MR T2 images and evaluated against original CTs. The discrepancy of CT number was analyzed in terms of MAE (Mean absolute error). Intensity modulated proton therapy (IMPT) plans with two opposed lateral beams were optimized on the original CT with a prescription dose of 79.2Gy(1.8Gy×44fx). The plans were then calculated on both original CT and S-CT with Monte Carlo algorithm and 1-mm resolution to get the final doses. The dose discrepancy was evaluated by the DVH parameters of target and critical organs, as well as the 3D-gamma passing rate.

RESULTS: The computational time for generating a new S-CT was 3.84–7.65 s. The MAE(mean±Std) between the original CT and sCTs was 30.23±5.53HU. The gamma passing rate (global 3D gamma, 10% threshold) is 93.17%, 99.56% and 99.90 for 1%/1mm, 2%/2mm, 3%/3mm criteria, respectively. The differences in target DVHs were 0.1±0.2% and 0.1±0.3% for D99, D95, respectively. For critical organs, Rectum D5, Bladder D40 and femoral head D2 have a difference of −0.3%±0.8%, 0.1±1.7%, and −0.1±0.2, between S-CT and original CT.

CONCLUSIONS: The proposed U-net can generate S-CTs that are in good agreement with original CTs for proton planning purpose in a few seconds. An efficient MRI-only planning and adaptation workflow is potentially feasible for proton therapy in prostate cancer.

Abstract 117: Reducing Pencil Beam Proton Lateral Penumbra by Using Layer-By-Layer Adaptive Aperture, Bolus, and Smaller Air Gap

BACKGROUND: Pencil beam proton therapy often has inferior lateral penumbra compared to photon-based therapy at clinically relevant depths. We test the hypothesis that by using layer-by-layer adaptive aperture (AA), it would be possible to achieve sharper penumbra compared to that achieved with conventional photon therapy at typical tumor target depths.

METHOD AND MATERIALS: The Mevion Hyperscan Pencil Beam Scanning system, consisting of an in-line range shifter and layer-by-layer mechanical aperture was used to measure lateral penumbra (distance between the 80%–20% dose levels) of a 10×10cm field at varying depths. Measurements were taken with and without a 10cm-thick bolus and with varying air gaps (AG), with and without AA. This was compared to penumbra from 6MV photons delivered using a Varian Trilogy system.

RESULTS: Most of our Hyperscan treatments require at least a 10cm AG, which can often be reduced to 3cm or less using a 10cm-deep bolus attached to the head. AA reduces measured penumbra by about 1/3 at shallow, clinically relevant depths. Use of 10cm bolus with 3cm AG leads to significant flattening of penumbra by 32% compared to 10cm AG with adaptive aperture, and a reduction of 54% compared to using 10cm AG without adaptive aperture. While 6MV photons have sharper penumbra below 11.5cm for protons without AA at 10cm AG, the sharper lateral penumbra for photons is reduced to depths below 7cm when 3cm AG with AA is used.

CONCLUSIONS: AA permits approximately 30% sharper penumbra at shallower depths clinically applicable for most superficial tumor targets. Use of layer-by-layer AA combined with bolus and smaller AG significantly decreases the depth at which protons start to have tighter lateral penumbra compared to 6MV photons.

Abstract 118: A Sub-Cubic Millimeter Single Particle Sensor for In Vivo Dosimetry

BACKGROUND: Range uncertainty is a major limiting factor in precision targeting of charged particle therapy. Ongoing efforts investigating accurate prediction of energy deposition would benefit from knowledge of the actual dose deposited in tumor and normal structures. Current dosimetric techniques measure the total or average energy deposited, but do not capture individual particle energy which may have distinct biological properties depending on its location within the Bragg curve. To address this, we developed a sub-cubic millimeter single particle dosimeter measuring particle flux and individual particle linear energy transfer (LET).

METHOD AND MATERIALS: A 1mm×0.5mm×0.3mm dosimeter consisting of 2,048 sensing elements was designed and fabricated in a 65nm complementary metal oxide semiconductor process. Each sensing element contains a differential pair of 1μm2 photodiodes; a proton hit produces a transient voltage pulse of 9.5–309mV, proportional to particle LET. Pulse duration, monotonic function of the energy deposited (LET), was measured. The sensor was placed in 67.5MeV proton beam (Crocker Nuclear Laboratory, UC Davis). Current (particle flux) was varied from 0.1–3nA and water column thickness (WCT) was varied from 0–3cm. TOPAS was used to simulate sensor response.

RESULTS: LET of individual protons was measured. Increasing beam current from 0.1 to 3nA resulted in a measured particle flux of 1.08×104–2.6×105 protons/s/mm2. At 3nA, the mean LET increased with water column thickness; the Bragg peak was identified at 2.7cm, with a pulse duration 1.76 times that of the entry dose. Flux decreased slightly with increasing WCT and rapidly fell off after the Bragg peak. The tail consisted of a few high LET protons.

CONCLUSION: We have demonstrated single particle LET detection in a clinical proton beam. Direct implantation of these devices can yield real time single particle in vivo dosimetry, improving accuracy of treatment planning and removing range uncertainty even in the setting of tumor and OAR motion.

Abstract 124: The Impact of Dual Energy CT Scans on TPS Calculated Dose Accuracies for Biological and Non-Biological Samples

BACKGROUND: Range uncertainty is a major limiting factor in precision targeting of charged particle therapy. Ongoing efforts investigating accurate prediction of energy deposition would benefit from knowledge of the actual dose deposited in tumor and normal structures. Current dosimetric techniques measure the total or average energy deposited, but do not capture individual particle energy which may have distinct biological properties depending on its location within the Bragg curve. To address this, we developed a sub-cubic millimeter single particle dosimeter measuring particle flux and individual particle linear energy transfer (LET).

METHOD AND MATERIALS: A 1mm×0.5mm×0.3mm dosimeter consisting of 2,048 sensing elements was designed and fabricated in a 65nm complementary metal oxide semiconductor process. Each sensing element contains a differential pair of 1μm2 photodiodes; a proton hit produces a transient voltage pulse of 9.5–309mV, proportional to particle LET. Pulse duration, monotonic function of the energy deposited (LET), was measured. The sensor was placed in 67.5MeV proton beam (Crocker Nuclear Laboratory, UC Davis). Current (particle flux) was varied from 0.1–3nA and water column thickness (WCT) was varied from 0–3cm. TOPAS was used to simulate sensor response.

RESULTS: LET of individual protons was measured. Increasing beam current from 0.1 to 3nA resulted in a measured particle flux of 1.08×104–2.6×105 protons/s/mm2. At 3nA, the mean LET increased with water column thickness; the Bragg peak was identified at 2.7cm, with a pulse duration 1.76 times that of the entry dose. Flux decreased slightly with increasing WCT and rapidly fell off after the Bragg peak. The tail consisted of a few high LET protons.

CONCLUSION: We have demonstrated single particle LET detection in a clinical proton beam. Direct implantation of these devices can yield real time single particle in vivo dosimetry, improving accuracy of treatment planning and removing range uncertainty even in the setting of tumor and OAR motion.

Abstract 129: Proton Irradiation Using Pencil Beam Scanning (PBS) Correlates with Taste/Odor Changes Identified by the Layer Defined Dose and Bragg Peak Delivery to the Olfactory Structure (OS) for Brain Tumors

BACKGROUND: Patients undergoing brain irradiation indicate unpleasant chemosensory changes during treatment. The ability to predict when unpleasant changes occur will lead to better patient compliance. This study evaluates the correlation between brain proton irradiation and the effects on the olfactory system.

METHOD AND MATERIALS: Fifteen patients were enrolled in a prospective study. All patients received proton PBS craniospinal irradiation with a posterior fossa boost. Patients depressed a buzzer when taste/odor was detected and again when taste/odor dissipated, for three consecutive days. Monitor units and layer were recorded for each buzz and correlated with dosimetry/treatment layer. A Post-treatment survey identified the negative chemosensory description/intensity. Assessment of the number of overlapping days the taste/odor change occurred and coloration between the layer-defined OS dose and Bragg peak location, in proximity to the OS occurred.

RESULTS: All 15 patients experienced changes in odor and 5 noted changes in taste. 9 patients buzzed after every layer indicating no smell detected during the pause between layer switching. On average, patients indicated the odor/taste was moderate/difficult to tolerate (sewer, burning, chlorine). Mean incidence dose to olfactory structures reported on days 0, 1, 2, and 3 were 0.4, 2.7, 4.9, and 16.5 cGyRBE respectively. This best fits a normal-tissue complication probability model yielding a TD50 of 4 cGyRBE. The number of days which layers overlapped the taste/odor period correlates with the OS proximal to the Bragg peak (p=0.036). For layers causing 100% odor change, the OS was proximal to the Bragg peak 95% for all 3 days. Nasal cavity dose was zero in all cases.

CONCLUSION: During brain irradiation, taste/odor changes correlate with the timing of layer-defined dose delivery to the OS and the location of the Bragg peak proximal to the OS. Radiation-induced chemosensory changes that contribute to the taste/odor changes, during irradiation, may be able to be predicted.

Abstract 134: Estimating the Individual Quality of Life Benefit and Cost Effectiveness of Proton Therapy for Patients with Oropharyngeal Cancer

PURPOSE: Using a quantitative decision-support system to estimate the quality of life burden from normal tissue complications (NTCs) after definitive RT for oropharyngeal cancer (OPC), comparing photon and proton RT.

METHOD AND MATERIALS: The normal tissue complication probability (NTCP) for dysphagia, esophagitis, hypothyroidism, xerostomia and oral mucositis was estimated for 33 OPC patients, comparing delivered photon IMRT plans with intensity- modulated proton therapy (IMPT) plans generated using clinical protocols at a collaborating PT center. Plans had equivalent target coverage and robustness optimization was used for IMPT plans. Latencies and durations of NTCs were modeled while accounting for the disease-specific age-, sex-, and smoking status-adjusted conditional survival probability. The quality- adjusted life years (QALYs) lost attributable to each NTC were calculated by assigning quality-adjustment factors based on complication severity. Cost effectiveness was modeled based on the upfront cost of IMPT ($36,659) and IMRT ($20,257), and interventions related to NTCs, with 3%/year discounting of QALYs and long-term costs.

RESULTS: The average QALYs lost from all NTCs were 1.52y and 1.15y for IMRT and IMPT, respectively, with average 0.37y spared with IMPT (95% CI: 0.27y–2.53y). The QALYs spared with proton RT varied considerably between patients, from 0.06 to 0.84 QALYs. Younger patients with p16-positive tumors and ≤10 pack-years smoked had the greatest estimated benefit of average 0.56 QALYs. IMPT cost effectiveness varied greatly between patients with a wide variety of incremental cost effectiveness ratios (ICERs), which are a measure of cost per spared QALY.

CONCLUSION: Using this decision-support tool we identified patients for which IMPT is estimated to have the greatest benefit and would be most cost effective. This can help optimize resource allocation and patient selection for trials aimed at improving the quality of life for OPC patients.

Abstract 135: Routine Neurocognitive Testing of Patients Treated with Proton Therapy on a Prospective Registry Study: Feasibility and First Results

BACKGROUND: Patients with CNS and skull base tumors treated with proton therapy undergo repeated neurocognitive testing as part of our Registry Study. 112 of 114(98%) eligible patients performed the tests. We report on feasibility, patient acceptance, and early results.

METHOD AND MATERIALS: A battery of standardized EORTC tests was employed. The Hopkins Verbal Learning Test (HLTV-R) tests immediate and delayed verbal memory. Trail Making Test Part A and B (TMT-A und B) assesses concentration, processing speed and cognitive flexibility. The Controlled Oral Word Association Test (COWA) evaluates verbal fluency and executive function. The Grooved Pegboard Test evaluates visual motor coordination. Tested time points were at baseline, at treatment completion, at 3, 6,12 months and once yearly thereafter. Testing time for every period were 30 minutes.

RESULTS: The first 4 time points (baseline, treatment end, at 3 and 6 months) were evaluated in 112, 94, 61 and 40 patients. Test completion rates were 99%, 96%, 98%, and 98 % at the time points. No significant differences were found in verbal memory (HVLT-R), cognitive flexibility, executive function (TMT-B) and motor coordination. Between baseline and at 6 months significant worsening was found in speed processing (TMT-A, p=.002), however, significant improvement in verbal fluency (COWA, p=.000).

CONCLUSION: First and early results on neurocognitive function testing after proton therapy demonstrate variable changes of different neurocognitive functions. The tests show high degree of patient acceptance. They are feasible with admissible effort and can be initiated in every center opened in order to collect long term data for neurocognitive toxicity.

Poster Abstract 2: Impact of Practice CT Date on Anesthesia Utilization for Pediatric Patients

HYPOTHESIS: Offering a structured practice CT simulation date to pediatric patients ages 4–9 receiving proton therapy reduces the need for daily anesthesia.

BACKGROUND: For pediatric patients undergoing radiation therapy, it is necessary for the patient to receive a CT scan for treatment planning. The scan requires the child to stay motionless, as does daily treatment. Prior to March 2016, there was no formal process for pediatric patients completing a CT simulation at the Northwestern Medicine Chicago Proton Center. For children aged 8 or less, the staff in the CT department would attempt to get the child through the CT simulation without anesthesia on the day of their consultation. If this was unsuccessful, the patient would undergo CT with anesthesia and proceed with daily anesthesia for treatment.

INTERVENTION: A practice CT was established as of January 16th, 2017. This included a pre-call from the child life specialist (CLS) to evaluate the patient and parent needs and expectations. The day of practice CT simulation, time is scheduled for child life to meet with the patient and the parent. The child life specialist used age appropriate explanations and visual aids to guide the patient through what will happen, then worked with the child throughout CT simulation to help them tolerate the procedure. If patient is unable to tolerate the scan after one or two attempts, the patient will complete the CT simulation, as well as daily proton therapy treatment, under anesthesia.

CONCLUSION: Based on pre-screening, 23 out of 46 children aged 4–9 were determined to be viable candidates for receiving an official practice day. Of these 23 children, 22 have successfully completed their CT simulation, and subsequent treatment, without anesthesia. This translates into a 97% success rate. This processes has reduced the treatment burden on the child and family by eliminating the need for daily anesthesia, thus reducing the total treatment time and treatment cost.

Poster Abstract 7: Clinical Implementation of Pre-Operative Short-Course Pencil Beam Scanning Proton Therapy for Patients with Rectal Cancer

BACKGROUND: For the treatment of rectal cancer, pencil beam scanning proton therapy (PBS-PT) may offer a reduction in radiation exposure to normal tissues compared to 3D-conformal photon radiotherapy (3DCRT). The purpose of this study is to report the clinical implementation and dosimetric analysis of pre-operative short-course PBS-PT (25 Gray in 5 fractions) for rectal cancer.

METHODS AND MATERIALS: This was a retrospective review of ten patients with stage IIA-IVB rectal cancer who received pre- operative short-course PBS-PT between 2018 and 2019. Patients received 25 Gy (radiobiological effectiveness =1.1) in 5 fractions on consecutive weekdays immediately preceding curative-intent total mesorectal excision. Patients were treated supine with a full bladder on a foam pad and knee cushion with vacuum mold immobilization of the legs. Plans were generated using Eclipse treatment planning system (Version 13.7, Varian Medical Systems). The clinical target volume (CTV) encompassed the rectal tumor, mesorectum, and regional lymphatics. PBS-PT plans were generated using single-field optimization with 2 posterior-oblique fields. The goal CTV coverage was V100% ≥ 98%. Robust optimization of CTV coverage was performed with positional uncertainty of +/− 5 mm in x, y, and z directions and range uncertainty of +/− 5%. Verification CT scans were performed on the first 3 days of treatment. Each patient had a backup 3DCRT plan with standard 3–4 pelvic fields generated for dosimetric comparison.

RESULTS: CTV coverage was similar between PBS-PT and 3DCRT plans. PBS-PT had significant reductions in dose to the small bowel, large bowel, bladder, and femoral heads. All 10 patients completed PBS-PT as planned without treatment break or need for replanning. All CT verification scans demonstrated good target coverage with CTV V100 > 95%.

CONCLUSION: Pre-operative short-course PBS-PT has been successfully implemented and offers a significant reduction of radiation dose to normal tissues. Prospective studies are warranted to evaluate if dosimetric advantages translate into clinical benefit.

Poster Abstract 8: Intensity-Modulated Proton Therapy for A Post-Mastectomy Breast Patient with Dual-Port Tissue Expander

BACKGROUND: A bilateral post-mastectomy patient diagnosed with right breast cancer was treated with a dual-port tissue expander using intensity-modulated proton therapy (IMPT).

METHOD AND MATERIALS: The dual-port tissue expander (Allox2®, SIENTRA, INC., CA) features two metallic ports in different dimensions: one for saline injection and a second for serous fluid drainage. The dual-port was carefully contoured and its density was overridden based on the information provided by the manufacturer (Grade 2 Titanium casing and Neodymium 35EH magnet). Three single-field uniform dose IMPT plans were generated: a. 2-field non-coplanar, b. 3-field coplanar, and c. 3-field non-coplanar beam arrangements. All plans were inverse optimized and forward calculated using the Monte Carlo dose algorithm. Dosimetric comparisons were evaluated to select the optimum plan for patient treatment. Patient setup and treatment time were also recorded.

RESULTS: A dose cold spot was noted behind the dual-port within the tissue expander. A clinical target volume for evaluation was defined for the chest wall after subtracting the ipsilateral tissue expander and cropping 5 mm from the body surface for skin. The volume receiving 95% of the prescribed dose (V95%) was 98.58% in the 3-field non-coplanar plan, compared with 97.68% and 96.25% for the 2-field non-coplanar and 3-field coplanar plans. The dose cold spots behind the dual-port were compensated best with the 3-field non-coplanar beam arrangement. All three beam arrangements resulted in comparable normal tissue doses; ipsilateral lung V20% ranged 15.4–16.4% and heart mean dose ranged 0.81–0.87 Gy(RBE). The patient was treated with the 3-field non-coplanar plan and aligned with dual-port contours for daily IGRT.

CONCLUSIONS: The 3-field non-coplanar plan provided excellent target coverage for a patient with a dual-port tissue expander. Treatment time increased to ~ 30 minutes compared with 15 minutes (one field per fraction) in patients with a single-port tissue expander.

Poster Abstract 9: Pathologic Complete Response (pCR) Rates and Outcomes after Neoadjuvant Chemoradiotherapy with Proton or Photon Radiation for Distal Esophageal Adenocarcinoma

BACKGROUND: Pathologic complete response (pCR) after neoadjuvant chemoradiotherapy is associated with improved survival in patients treated for esophageal cancer. While proton beam therapy (PBT) has demonstrated reduced toxicities, limited reports have evaluated pCR rates between modalities.

METHODS: Single-institutional review of patients from 2015–2018 with distal esophageal adenocarcinoma treated with trimodality therapy including PBT PBT patients were compared 2:1 to patients treated in a contemporary timeframe with photons.

RESULTS: Fifteen consecutive PBT patients were identified and compared to 30 consecutive photon patients All patients received concurrent chemotherapy with carboplatin/paclitaxel. Median radiation dose in both cohorts was 50.4Gy; all courses were delivered in 1.8Gy fractions Age, gender and race were well balanced (Table 1) Patients treated with PBT had a more advanced AJCC stage (p=0.038) with increased nodal burden (N2: 33% PBT vs. 3.0% photon, p=0.015). Despite this, nodal clearance rates were equivalent (58% PBT vs. 53% photon, p=0.427). pCR rates did not significantly differ between modalities (20% PBT vs. 23% photon, p=0.800). (Table 2) One grade 4 and 2 grade 5 perioperative complications occurred in the photon group. 1 PBT patient experienced a grade 4 event, no PBT patients experienced a grade 5 event. Acute and chronic RT toxicities were balanced between groups (Tables 3–4) At a median follow-up of 22 months (3–44) 18-month progression-free survival (PFS) was 58% (95% CI 49–67); this was equivalent between modalities (55% v. 58%, p=.634).

CONCLUSIONS: The use of PBT in trimodality therapy for distal esophageal adenocarcinoma is safe and yields pCR rates comparable to photon radiation and historical controls. pCR and nodal clearance rates and 18-month PFS did not significantly differ despite PBT patients having higher AJCC stage and nodal disease burden.

Poster Abstract 11: The Efficacy of Definitive Concurrent Proton Chemoradiotherapy for Esophageal Cancer

BACKGROUND: Current approaches to esophageal cancer focus on multimodality treatment with chemotherapy, radiation therapy (RT), and/or surgery. To date, data on the use of definitive chemoradiation in non-operable esophageal carcinoma are limited, particularly for treatment regimens involving proton RT. Here we report disease-control outcomes in patients who received definitive intent non-operative therapy consisting of concurrent proton chemoradiotherapy for non-operable esophageal cancer.

METHODS: We conducted a retrospective cohort study including patients with unresectable (either medically or by disease characteristics) esophageal cancer who received concurrent proton chemoradiotherapy without surgery at the University of Pennsylvania between 1/2008 and 1/2019. Descriptive statistics were used with Kaplan-Meier analysis for overall survival (OS) and disease-free survival (DFS).

RESULTS: A total of 42 patients were evaluable (Stage I: 7%, Stage II: 48%, Stage III: 31%, Stage IV: 2%, Unknown: 12%) with a median follow-up time of 13.67 months post-RT. Patients received carboplatin/paclitaxel (76%), fluorouracil-based (22%), or single-agent paclitaxel (2%) chemotherapy, with a median of 5 concurrent cycles administered (range, 1–7). Additional induction and/or adjuvant chemotherapy was given to 38% of patients. Median total RT dose was 5047 cGy (range, 4473–6015 cGy) delivered in a median of 28 fractions (range, 24–33), with 93% of patients receiving a median cone down dose of 541 cGy (range, 180–1442 cGy) delivered in a median of 3 fractions (range, 1–9). Loco-regional recurrence and distant metastasis occurred in 29% and 31% of patients at a median time of 16.92 months and 11.28 months post-RT, respectively. OS was 64%, 33%, 21%, and 7% at 1, 2, 3, and 5 years, respectively. Corresponding values for DFS were 55%, 29%, 17%, and 5%, respectively.

CONCLUSIONS: Treatment of non-operable esophageal cancer with definitive concurrent proton chemoradiotherapy has acceptable disease-control outcomes. Continued follow-up to evaluate long-term outcomes and further investigation in large cohort studies is warranted.

Poster Abstract 12: The Toxicity Profile of Definitive Concurrent Proton Chemoradiotherapy for Esophageal Cancer

BACKGROUND: Curative-intent management of esophageal cancer (EC) typically incorporates chemotherapy, radiotherapy (RT), and surgery. Many patients, due to medical comorbidity, advanced disease, and/or patient preference, are not surgical candidates, and receive definitive concurrent chemoradiotherapy (DC-CRT). Proton RT has the potential to reduce toxicity compared to photon RT, though data for DC-CRT for EC are limited. We report toxicity outcomes from our experience treating EC with proton DC-CRT.

METHODS: We performed a retrospective cohort study analyzing patients with inoperable (due to comorbidity, advanced disease, and/or patient preference) EC who underwent proton DC-CRT at our institution between 1/2008–1/2019. Toxicities were prospectively collected. Descriptive statistics were used to report CTCAE v4.0–5.0 acute (<6 months post-RT) and late (≥6 months post-RT) toxicities.

RESULTS: At a median follow-up of 12 months (range 0.5–91) post-RT, 40 patients were evaluable, and typically received a median of 5 cycles (range, 1–7) of concurrent carboplatin/paclitaxel (78%) or fluorouracil-based (20%) chemotherapy with proton RT. 37% received induction and/or adjuvant chemotherapy. Patients received a median total RT dose of 5046 cGy (range, 4473–6015) in 28 fractions (range, 24–33). Acute grade 3 toxicities included dysphagia (20%), anorexia (20%), fatigue (8%), dyspnea (5%), dehydration (5%), and esophagitis (3%), with 2 patients requiring feeding tube placement during or within 4 weeks after RT. The most common acute grade 2 toxicities were dysphagia (45%), fatigue (45%), anorexia (38%), esophagitis (30%), esophageal pain (25%), and nausea (15%). Of the 11 patients evaluable for late toxicity data, at a median follow-up of 22 months (range, 11–91), there was 1 grade 3 toxicity (dysphagia). The most common late grade 2 toxicities were dysphagia (27%), fatigue (27%), and anorexia (27%). There were no acute or late grade≥4 toxicities.

CONCLUSIONS: At a median follow-up of 12 months (range 0.5–91) post-RT, 40 patients were evaluable, and typically received a median of 5 cycles (range, 1–7) of concurrent carboplatin/paclitaxel (78%) or fluorouracil-based (20%) chemotherapy with proton RT. 37% received induction and/or adjuvant chemotherapy. Patients received a median total RT dose of 5046 cGy (range, 4473–6015) in 28 fractions (range, 24–33). Acute grade 3 toxicities included dysphagia (20%), anorexia (20%), fatigue (8%), dyspnea (5%), dehydration (5%), and esophagitis (3%), with 2 patients requiring feeding tube placement during or within 4 weeks after RT. The most common acute grade 2 toxicities were dysphagia (45%), fatigue (45%), anorexia (38%), esophagitis (30%), esophageal pain (25%), and nausea (15%). Of the 11 patients evaluable for late toxicity data, at a median follow-up of 22 months (range, 11–91), there was 1 grade 3 toxicity (dysphagia). The most common late grade 2 toxicities were dysphagia (27%), fatigue (27%), and anorexia (27%). There were no acute or late grade≥4 toxicities.

Poster Abstract 14: Pediatric Bithalamic Glioma with EGFR Exon 20 Insertion Mutation: First Case Report of Concommitant Radiotherapy and Osimertinib

BACKGROUND: Midline gliomas are common in childhood and frequently associated with H3K27 mutations. Pediatric bithalamic gliomas (PBTG), however, are rare tumors that exhibit a diffuse infiltrative growth pattern and generally do not harbor histone mutations. There are no established therapeutic options in this population other than radiotherapy, and outcomes remain poor. The purpose of this report is to describe the first pediatric PBTG patient treated with concomitant osimertinib and radiotherapy.

METHODS AND MATERIALS: We report a 3-year old male who presented with a one-week history of worsening headaches, nausea, vomiting, and gait instability. MRI demonstrated a large, non-enhancing bithalamic mass involving the cerebral peduncles and midbrain with extension into the prepontine cistern. The mass effaced the 3rd ventricle and Sylvian aqueduct, causing obstructive hydrocephalus. Endoscopic third ventriculostomy and biopsy was consistent with PBTG and negative for H3K27M mutation. Next generation sequencing demonstrated an EGFR exon 20 insertion mutation, which has been described in lung adenocarcinoma but rarely in pediatric gliomas. Pathogenic mutations in TP53 and CREBBP were also identified.

RESULTS: The patient completed intensity-modulated proton therapy (54 Gy RBE/30 fractions) with concurrent temozolomide (70 mg/m2 daily) and the 3rd generation, brain-penetrant EGFR receptor tyrosine kinase inhibitor (RTKI), osimertinib (20 mg daily for 1 week, then increased to 40 mg daily). After proton therapy, the patient continued adjuvant temozolomide and maintenance osimertinib (40 mg daily). He developed facial and truncal rash, managed with clindamycin and desonide creams. The patient remains with clinically and radiographically stable disease eight months since diagnosis with no other Grade 2 or higher toxicities.

CONCLUSIONS: To our knowledge, this is the first report of primary up-front combinatorial radiotherapy with osimertinib for PBTG with an EGFR exon 20 insertion mutation. Future study of RTKI in this rare tumor subtype is advised.

Poster Abstract 16: Proton Partial Breast Irradiation Preferentially Spares the Heart and Lungs over Modern Whole Breast Radiotherapy

PURPOSE: To compare target coverage and normal tissue sparing of state-of-the-art photon whole breast irradiation (WBI) and proton partial breast irradiation (PPBI).

METHODS: Consecutive women with node negative breast cancer treated with lumpectomy and adjuvant WBI, without regional nodal irradiation, or PPBI were included. WBI was delivered with 3-dimensional conformal tangential fields, targeting the breast CTV with a 5 mm expansion to PTV to a median dose of 40 Gy in 15 fractions. Left-sided WBI patients were treated in deep- inspiratory breath hold. PPBI was delivered with a median of 2 multi-field optimized beams, targeting the lumpectomy cavity plus 1 cm, to a median dose of 21.9 Gy in 3 daily fractions. Setup uncertainty analyses of +/− 3 mm isocenter shifts in each translational axis and 3% beam range uncertainty were performed to ensure robust target coverage and normal tissue sparing. Dosimetric parameters, collected prospectively, are primarily presented as % prescription and we excluded the boost component of WBI for uniformity of plan comparisons.

RESULTS: 836 women were treated between 2015–2018; 762 received WBI (274 [36%] with boost to the lumpectomy cavity), and 74 with PPBI. Patients treated with PPBI were older and had more favorable breast cancer. Patients treated with PPBI had comparable target coverage but significantly lower heart and lung doses.

CONCLUSIONS: PPBI reduces exposure to the heart and lungs. Follow-up is needed to determine if these dosimetric advantages translate into improved clinical outcomes.

Poster Abstract 17: Comparison of Prone Versus Supine Position and Different Planning Techniques of Anal Canal Cancer Treated with Pencil-Beam Scanning Proton Beam Therapy: Cross Institutional Dosimetric and IGRT Studies

BACKGROUND: A retrospective study comparing impact of prone versus supine ano-rectal cancer patient setup on scanning proton beam therapy is presented.

METHODS AND MATERIALS: Five supine MCR patients planned with Multi field Optimization (MFO) technique; were compared against 5 prone MCA patients planned with Individual Field Simultaneous Optimization (IFSO). Verification plans were generated from weekly verification CTs. All plans were normalized to 95% Isodose lines for comparisons. Plan robustness was evaluated with 3% range and 5mm of setup errors. Verification plans were also generated to further assess the robustness of these plans. Daily positional shifts and image guided setup times were obtained from Record and Verify systems and analyzed.

RESULTS: Results from 4 patients (2 supine and 2 prone) are reported below. Inter-comparison of target coverage and OAR sparing are summarized. Target coverage (D95) was above 98.8% in all verification plans. The average absolute deviations of positional shifts (vertical, longitudinal, lateral, rotation, pitch, and roll) derived from matching bony anatomy were (2.6mm, 5.6mm, 9.2mm, 0.63°, 0.66°, 0.62°) for supine patients and (2.3mm, 6.5mm, 6.1mm, 0.42°, 0.38°, 0.45°) for prone patients. The average setup times over all fractions (in minutes) for prone patients was 2:30 (max: 7:15, min: 0:00 – no shift was required) and 5:58 (max: 37:15, min: 1:13) for supine patients.

CONCLUSIONS: Preliminary result show plan robustness to be generally independent of both setup and treatment planning techniques. However, our initial findings reveal the MFO technique might result in lower skin dose but a higher bone marrow dose in these patients. The average positional shifts were similar between supine and prone setups, but the average setup time was significantly less for prone setups. This is a work in progress. Data on larger cohort will be analyzed to conclude this study.

Poster Abstract 29: Ethical Allocation of Proton Therapy: A Novel Scoring System

BACKGROUND: At an institution operating a single-room proton facility at full capacity, there exists a need to optimize allocation of treatment slots for maximal patient benefit. Additionally, there remains uncertainty regarding insurance authorization for proton therapy in the rapidly evolving landscape of radiation oncology. The clinical care objective was to delineate a novel scoring system for proton therapy utilization and prioritization and to compare this system against insurance approval rates for proton therapy treatment.

METHOD AND MATERIALS: All 125 patients evaluated for proton therapy while the facility was operating at capacity from September 2018 through May 2019 were considered. A scoring system developed by experts from the fields of radiation oncology, medical oncology, and medical ethics was used to assign proton candidacy scores based the following factors: clinical benefit, strength of evidence, expected survival duration, performance status, and other pertinent information such as research protocol enrollment. Patient characteristics (e.g. age, disease site, insurance, etc.) were tabulated. Logistic regression modeling and box-and-whisker plots comparing patients' scores with receipt of proton therapy were performed.

RESULTS: Of the patients evaluated, 84% were ultimately approved by insurance after accounting for denial appeals and peer-to-peer consultation, including all pediatric patients. Logistic regression modeling showed patients with higher scores were significantly more likely to receive proton therapy (OR=1.29; 95% CI=1.1–1.5; p<0.05). However, there was no significant difference in median total score in adults with regards to receipt of proton therapy (p=0.38). Average time from insurance authorization request to insurance response was two days for approval and eleven days for denial. There was no significant association between median score and proton approval among different insurance providers.

CONCLUSIONS: This scoring system would allow for the ethical prioritization of patients seeking proton therapy at our facility based on characteristics derived through a maximally objective, multi-departmental model.

Poster Abstract 32: Proton Therapy Preserves Acute Left Ventricular Ejection Fraction Relative to Conventional X-Ray-Based Therapy in Breast Cancer

BACKGROUND: Radiotherapy (RT) yields a survival benefit in breast cancer treatment but can also result in radiation exposure to the heart. RT using protons (PT) can reduce the mean heart dose (MHD) relative to conventional X-ray-based RT (XRT), but a clinical benefit in terms of decreased cardiac toxicity has not yet been quantified.


  1. Pre-symptomatic LVEF decline can be quantified using detailed 3D analysis of cardiac magnetic resonance images (CMRI);

  2. Cardiac function is preserved in patients treated with PT than XRT;

  3. The severity of cardiac dysfunction scales with heart dose.

METHODS: Under an IRB-approved study, CMRI datasets were acquired pre-RT but after chemotherapy, and at 6–13 months post-RT in 10 left-sided breast cancer patients undergoing RT to the breast/chest wall and regional nodes. The LV endocardial border from end-diastole to end-systole was segmented in multiple image slices and views by 3 independent readers blinded to the treatment modality. Conformal 3D models of the LV endocardial geometry provided precise LVEF estimates.

RESULTS: Post-RT LVEF improved from baseline for 5/6 PT patients (+8.3±5.2%) but decreased for all (3/3) XRT patients (−9.6±2.1%). The change in LVEF was significant (p=0.006) between the groups. The MHD for the PT cohort (0.28+/−0.10 Gy-RBE) was lower (p<0.001) than for the XRT cohort (4.15+/−0.55 Gy). MHD negatively correlated with LVEF change (r=−0.87) with a slope of −5.3% LVEF for every 1 Gy MHD. Doxorubicin was administered to 4/4 of XRT and 4/6 of PT patients. There was no significant difference in baseline LVEF, age, follow-up interval, body mass index, smoking history or prior cardiac history between the groups.

CONCLUSIONS: Early, pre-symptomatic changes in LVEF are measurable using CMRI with conformal 3D LV surface models. LVEF decline was correlated with MHD. LVEF was preserved or improved with PT but diminished with XRT. Further evaluation is warranted.

Poster Abstract 34: Comparative Toxicity Outcomes of Proton Beam Therapy Versus Intensity-Modulated Radiotherapy for Prostate Cancer in the Post-Operative Setting

BACKGROUND: Despite increasing utilization of proton-beam therapy (PBT) in the post-prostatectomy setting, no data exist regarding toxicity outcomes relative to intensity-modulated radiotherapy (IMRT). We compared acute/late genitourinary (GU)/gastrointestinal (GI) toxicity outcomes in prostate cancer (PC) patients treated with post-prostatectomy IMRT versus PBT.

METHODS AND MATERIALS: With IRB approval, we reviewed PC patients who received adjuvant/salvage IMRT or PBT (70.2 Gy with an endorectal balloon) following prostatectomy from 2009–2017. Factors including combined IMRT/PBT and/or concurrent malignancies prompted exclusion. We performed a case-matched cohort analysis using nearest-neighbor 3-to-1 matching by age and GU/GI disorder history. Logistic and Cox regression identified univariate and multivariate associations between toxicities and cohort/dosimetric characteristics. Toxicity-free survival (TFS) was assessed using the Kaplan-Meier method.

RESULTS: Three-hundred-and-seven men (mean age 59.7±6.3; 237 IMRT, 70 PBT) were identified, generating 70 matched pairs. Median follow-up was 48.6 and 46.1 months for IMRT and PBT, respectively. Although PBT was superior at reducing low-range (V10-V40) bladder and rectal dose (all p≤0.01), treatment modality was not associated with differences in clinician-reported acute/late GU/GI toxicities (all p≥0.05). Five-year grade ≥2GU and grade ≥1 GI TFS was 61.1% and 73.7% for IMRT and 70.7% and 75.3% for PBT, respectively; Five-year grade ≥3GU and GI TFS was >95% for both groups (all p≥0.05).

CONCLUSIONS: Post-prostatectomy PBT minimized low-range bladder and rectal dose relative to IMRT; however, treatment modality was not associated with clinician-reported GU/GI toxicities. Future prospective investigation and ongoing follow-up will determine whether dosimetric differences between IMRT and PBT confer clinically meaningful differences in long-term outcomes.

Poster Abstract 35: Development of a Pencil Beam Scanning Proton Breast Treatment Procedure in a New Proton Center

BACKGROUND: Breast cancer is the most common non-cutaneous malignancy among U.S. females. Pencil beam scanning (PBS) proton therapy acheives more optimal non-target tissue sparing, but a standard breast proton treatment algorithm has not yet been established. We developed an optimized breast treatment procedure in a newly-operational proton center with scanning beam-only technique.

METHOD AND MATERIALS: The treatment team visited numerous operational proton centers in the U.S. and observed end-to-end breast radiotherapy delivery. Based on these observations and our previous experiences, a center-specific treatment model was developed.

RESULTS: We developed an standardized procedure for breast treatment delivery: 1. patients are simulated on a wing board with arms above head; a bra or other device may be utilized for target immobilization; 2. images are transferred to the planning system for physician target delineation; 3. during the planning process, robust optimization is always utilized, and (a) most plans include 2 fields (AP/en face) with single-field optimization (rSFO), (b) for bilateral breast cases, 4 fields with 2 isocenters are applied, and (c) for patients with tissues expanders, a 2-field plan with multi-field optimization (rMFO) is utilized; 4. patient-specific QA is performed prior to the first treatment; 5. surface optical guidance is applied daily and CBCT is taken on days 1–3 and weekly to verify set-up; 6. adaptive planning is applied for significant anatomical changes; 7. plans are reviewed in chart rounds during the first treatment week. This procedure may evolve with time to optimize patient safety and treatment delivery.

CONCLUSIONS: Given the recent increase in proton centers in the U.S. and concomitant increase in women receiving proton therapy for breast cancer, we established a standard treatment procedure for PBS proton therapy delivery that identifies ideal treatment workflows and may contribute to eventual standardization of breast PBS treatment n existing and future proton centers.

Poster Abstract 37: Early Experience with Hydrogel Rectal Spacer with Proton Therapy for Prostate Cancer: Stability Assessment with Weekly MRI

BACKGROUND: Hydrogel is a biodegradable compound inserted between the rectum and prostate as a spacer that has been used to reduce rectal toxicity. Hydrogel is designed to disappear over time, but the effect of proton therapy on degradation rate is not well studied. This study describes early experience with hydrogel rectal spacers (SpaceOAR) with proton therapy for 10 patients with weekly assessment of spacer stability using MRI.

MATERIALS AND METHODS: Each patient had hydrogel inserted concurrently with 3 fiducial markers. All patients were treated at Scripps Proton Therapy Center. All patients underwent CT simulation and MRI with contrast. Rectal balloons were not used. Treatment was delivered with 2 beams (left and right lateral) using IMPT with SFO technique. Dose prescription was either 70 Gy in 28 fractions or 80 Gy in 40 fractions. Weekly adaptive simulation was done with CT and MRI. Since current hydrogel is not visualizable on CT, MRI is necessary to evaluate spacer stability. Measurements obtained by MRI: spacer thickness at base, apex, and mid-gland. The measurements were tabulated for each patient and in aggregate.

RESULTS: Variability between patients was greater than variability over treatment course. One patient had complete disappearance of spacer on imaging and required a rectal balloon for remainder of treatment. The images were sent to the vendor for further evaluation. Three patients had reductions in each of the three measured regions with net reductions of 18%, 33%, and 72% across all regions taken together. The remainder of the patients had reductions less than 10% across all regions taken together.

CONCLUSION: Care must be taken ensure consistently high-quality implantation of hydrogel spacers. Routine monitoring of spacer integrity should be conducted regularly throughout the course treatment to allow for appropriate adjustments to treatment plan if significant degradation has occurred.

Poster Abstract 38: Proton Therapy for Head and Neck Cancer Shows Favorable Outcomes & Low Toxicity in a Large, Multi-Institutional Cohort

BACKGROUND: Proton beam therapy (PBT) may have an improved therapeutic ratio compared to traditional photon/IMRT in the treatment of head and neck (H&N) cancer. There is limited data on the clinical outcomes and toxicity profiles of PBT in this patient population.

METHOD AND MATERIALS: We performed a retrospective analysis of 129 patients with cancer of major H&N sites treated in the primary setting with PBT from 2013–2017. These patients were included in an ongoing prospective multi-institutional proton registry database. 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 oropharynx (38%), salivary gland (38%), oral cavity (14.7%), and nasopharynx (7%). The most common histologies included squamous cell carcinoma (53.5%), carcinoma NOS (14.7%), adenoid cystic carcinoma (9.3%), and mucoepidermoid carcinoma (9.3%). Fifty-percent of patients had confirmed p16+ disease and 18% were treated post-operatively. The median dose was 66 Gy(RBE) [range 63.4–70] and 38% received concurrent chemotherapy. Median follow-up time was 25 months (range 18–37). Nineteen patients (14.7%) experienced acute grade 3+ toxicity − dermatitis (8), dysphagia (4), mucositis (4), nausea (1), odynophagia (1), and rash (1). Nine patients (7%) experienced late grade 3+ toxicity − dysphagia (6), trismus (2), and hearing alteration (1). 2-year OS, LR, and DM rate was 89.8%, 94.2%, and 91.5% respectively. On multivariate analysis, statistically significant predictors of LR were male sex (HR 0.17, 95% CI 0.032–0.86, p=0.033) and N3 stage disease (HR 16.63, 95% CI 2.32–118.95, p=0.0051).

CONCLUSION: While heterogeneous, this group of H&N cancer patients represents the excellent clinical outcomes achievable with PBT over a 2-year median follow-up, especially in regards to grade 3+ toxicities compared to historical control data from photon/IMRT patients. We look forward to further follow-up and improving outcomes in our future prospective treatment trials.

Poster Abstract 40: A Retrospective Analysis of The Effect of Reductions in Energy Switching Times on Treatment Capacity in a Multi-Room Proton Therapy Center

BACKGROUND: To quantify how a software upgrade (R7 to R8) changed beam delivery times and impacted efficiency and capacity.

METHODS AND MATERIALS: A 4-room center treating ~90 patients/day, treating for ~7 years with optimized operations, underwent a software upgrade which reduced energy and room switching times from ~30 and ~4 seconds to ~20 and ~0.5 seconds respectively. The center uses RFID data to track patient treatments and has software which links this to beam delivery data extracted from the treatment log server. Two 4-month periods, July–Nov in 2017 and 2018, with comparable patient volume, representing periods before and after the software change, were retrospectively analyzed.

RESULTS: A total of 16,168 and 17,102 fields were analyzed in 2017 and 2018, respectively. For bilateral head and neck and prostate patients, the beam waiting time was reduced by nearly a factor of 3 and the beam delivery times were reduced by nearly a factor of 2.5. Room switching times were reduced more modestly from 0.51±0.49 to 0.32±−0.37 minutes. Capacity in our gantry has increased from approximately 30–35 patients to 40–45 patients in a 16-hour daily operation. For nearly constant patient volume, the center has been able to close an inclined beam line which has reduced beam competition.

CONCLUSIONS: Many proton centers are striving for increased efficiencies in operations by reduction of QA time, treating daily alternating fields, and simplifying IGRT processes. We demonstrated that reductions in energy and room switching time can significantly increase center capacity. Further reductions are possible but greater gains may come from reducing setup time.

Poster Abstract 42: The Impact of Nasal Cavity Air on Head and Neck Cancer Treatment Planning with Pencil Beam Scanning Proton Therapy

BACKGROUND: The changes in the sinus and nasal cavity volume could potentially affect the range of the anterior-to-posterior (AP) protons beams used for the sinonasal cancer patients, which lead to beam overranging in to organs at risk or underranging in the treatment volumes. A three-field beam approach (20% of the prescribed dose contributed by 2 bi-lateral posterior-oblique fields) could potentially reduce the consequences of the changes in the cavity filling.

METHOD AND MATERIALS: Ten sinonasal cancer patients treated with pencil beam scanning proton therapy were evaluated retrospectively in this study. The verification CTs were performed 3–6 times for each patient during the treatment courses. All patients were planned with a single AP field and 3-field beam arrangement (80% AP, 10% right-posterior oblique and 10% left-posterior oblique fields). The initial plans were re-calculated on the verification CTs and the robustness of the target dose coverages and OAR sparing were analyzed in association with the change of the cavity volumes under different beam arrangement.

RESULTS: In the 39 verification CTs examined, an average percent change of the cavity volume from the planning CT was 14%. The variations of the mean dose in the OARs from planning CT were also recorded. The average brainstem and chiasm were 53.4% and 24.1% respectively in the single AP field plans, compared to 39.4% and 18.9% in the 3-field plans. Changes in the PTV coverage were also noted but percent differences were found to be marginal when compared with the nominal plans on the initial CTs.

CONCLUSIONS: On average, the 3-field beam approach may potentially yield robustness in a proton therapy plan with respect to variations and change in the nasal cavity and sinuses.

Poster Abstract 43: A Customized Image Guidance Tool for Verification CTs: Surface Imaging and Scout Guided Setups

BACKGROUND: In proton therapy, verification CT scans (vCT) are used to assess changes in patient anatomy and as part of the adaptive planning process. Unfortunately, CT scanners lack the IGRT tools for overlaying current patient position and relating that to the planning CT scans (pCT). Consequently, it may be difficult to determine if any observed vCT to pCT variation is due to real changes in anatomy or due to patient setup variation that could be corrected with IGRT. We have developed a system that permits image guided adjustments using commercially available hardware and in-house software.

METHOD AND MATERIALS: Using a pair of orthogonally placed video cameras, we have created a surface image tool that overlays surface images from pCT and vCT using opensource packages, OpenCV and PyQt. A pair of reference images of the patient setup is captured during initial simulation and acts as a reference position for the therapists during the vCT. The 2D surface imaging tool is used for initial guidance of the patient setup. Additionally, for fine tuning the setup, we have also developed a software tool which can fuse pCT to vCT scout images, mimicking treatment alignment of bony landmarks, prior to CT scanning.

RESULTS: In a sample image, where the system is applied to a head and neck patient, the user interface permits the RTTs to blend the overlaid images and adjust positioning. Another figure shows a comparison of the neck region of a patient from the population prior to and after implementation: Setup is faster and overall agreement between vCT and pCT has improved.

CONCLUSION: Reproducing the original patient setup using surface imaging and scout images helps facilitate the adaptive planning process. The more accurate vCTs result in improved workflows and more representative vCTs.

Poster Abstract 45: Effect of Prospective Use of Relative Biological Effectiveness Modeling in Treatment Planning on Magnetic Resonance Imaging Radiation-Related Changes in Pediatric Patients with Brain Tumors Treated with Pencil Beam Scanning Proton Therapy

BACKGROUND: An in-house biologic dose model (BDM) for pencil beam scanning proton beam therapy (PBSPT) has been used for treatment planning guidance and prospective plan evaluation since June 2016 at our institution. This study evaluates the incidence of radiation-related changes (RRC) before and after (pre-BDM and post-BDM) the model was used routinely in clinic.

METHOD AND MATERIALS: All patients younger than 18 years of age treated with PBSPT 4500–6000 cGyE (180–200 cGyE/day) between 1/2015 and 1/2018 for a primary brain tumor (non-glioblastoma) with at least one follow-up MRI within 6 months of PBSPT were identified for retrospective analysis. All available post-PBSPT MRIs were reviewed and compared to the pre-PBSPT MRI. RRC was defined as new areas of contrast-enhanced T1-weighted and/or T2-weighted MRI outside of the gross tumor volume. High-dose volumes including 5670 cGyE based on treatment-planning system (TPS5670), biological dose 5670 cGyE (BD5670), and biological dose 5940 cGyE (BD5940) were evaluated.

RESULTS: Sixty-two patients were identified for the analysis: 37 (59.7%) pre-BDM and 25 (40.3%) post-BDM. No significant differences in CTV coverage by prescribed dose (TPS 100% or TPS 95%) or volume of TPS5670 were noted between the two cohorts. BD5670 and BD5940 were significantly larger in the pre-BDM cohort: pre-BDM 81.9 cc vs. post-BDM 48.4 cc (p= 0.005) and pre-BDM 33.7 cc vs. post-BDM 11.9 cc (p= .002), respectively. Twenty-six (41.9%) patients developed RRC: pre-BDM 46% vs. post-BDM 36% (p=0.436). MRI T1-changes and symptomatic RRC's were both more common pre-BDM vs post-BDM: 29.4% vs.11.1% (p=0.292), respectively, for both outcomes.

CONCLUSIONS: Despite similar TPS dose coverage and hot-spot volume, post-BDM plans have a significant reduction of the high BD volume which may be correlated to a reduction in the risk of T1-MRI and symptomatic RRC.

Poster Abstract 51: Ovary Sparing Proton Therapy in Cervical Cancer

BACKGROUND: Ovarian function is exquisitely sensitive to radiotherapy. In pre- menopausal adult women, ovarian failure occurs with increasing frequency above mean doses of 6 Gy, with near universal toxicity above 15 Gy. In women undergoing ovarian transposition prior to pelvic radiotherapy, only half retain ovarian function. Herein we evaluate the feasibility of proton therapy to preserve ovarian function after pelvic radiotherapy.

METHOD AND MATERIALS: A patient with FIGO IIIC1 cervical squamous cell carcinoma was treated with definitive chemoradiotherapy after ovarian transposition. The patient was treated in a prone position. Dose was 45 Gy in 25 fractions to the pelvis with a simultaneous integrated boost of 55 Gy to involved pelvic lymph nodes. Pencil beam scanning proton therapy and volumetric modulated arc therapy (VMAT) plans were evaluated.

RESULTS: Mean dose (Gy) to the left and right ovaries were 0.10 and 0.04 with proton therapy compared to 11.35 and 10.70 with VMAT. Target coverage was excellent with both modalities (>99% of clinical target volumes receiving 99% of prescription dose), with appropriate robustness criteria and planning target volume coverage for proton therapy and VMAT, respectively. Additional dosimetric goals were improved with proton therapy over VMAT: small bowel maximum dose 46.6 Gy versus 54.5 Gy, small bowel V35 of 2% versus 7%, and bladder V40 of 34% versus 38%. Rectum V40 were both 90%.

CONCLUSIONS: Pencil beam scanning proton therapy resulted in clinically significant ovarian sparing during pelvic radiotherapy after ovarian transposition compared to VMAT. Proton therapy enabled mean ovarian dose constraints to be met that had not been achievable with VMAT. Proton therapy can reduce the rate of acute ovarian failure in pre-menopausal adult women undergoing pelvic radiotherapy. We recommend its use in this population.

Poster Abstract 53: Implementation of Carbon Ion Protocols in Clinical Practice at MedAustron

MedAustron is a synchrotron based dual particle therapy facility that started clinical operation with protons in December 2016. In January 2019, technical commissioning of a horizontal beam line for carbons was completed, and medical commissioning is on schedule for the first patient to be treated with Carbon Ions in July 2019. RayStation 8B will be commissioned for carbon ion (CIRT) treatment planning based on local effect model version I (LEM-I) for relative biological effectiveness (RBE). During the first two years clinical activity will follow established protocols, in the attempt to reproduce clinical results obtained in Germany and in Japan. MedAustron will combine 2 seemingly divergent strategies of dose prescription and incorporation of RBE. Prescribed dose will be adapted from Japanese protocols after correction for RBE models. Conversion of prescribed dose from the Japanese RBE model to LEM-I will be based on simulations performed on geometric targets with the same approach employed in the Italian facility (CNAO). For the first time MedAustron will have the opportunity to incorporate the clinical experience accumulated at CNAO using these converted doses.

All indications for CIRT require gross residual disease. A hypo-fractionated schedule of 16 fractions in 4 weeks will be implemented for H&N non-SCC cancer such as salivary gland cancer and mucosal melanoma. Bone and soft tissue sarcomas in the axial body and local recurrence in the pelvic wall from primary rectal cancer will also be treated with a 16 fraction schedule derived from Japanese experience. Spine and skull base chordomas and chondrosarcomas will be treated with 3 Gy per fraction in 22 fractions, according to German protocols.

With the availability of organ motion mitigation strategies, hypo-fractionated CIRT will be used in pancreatic cancer patients with 12 fractions for locally advanced cancer (LAPC) and 8 fractions for preoperative treatment.

For salivary gland cancer and mucosal melanoma the choice between 4.1 and 4.3 Gy (RBE) per fraction will be based on expected risks of bone, brain or soft tissue necrosis. Tumors with massive involvement of pharyngeal mucosa or pterygopalatine fossa (i.e. more than 180° circumferentially or more than 5 cm cranio-caudally or more than 200 cc PTV ) will be treated with lower prescription dose. Tumors with PTV extension into brain parenchyma or alveolar process for more than 10 cc will be treated with the higher dose but coverage with the lower dose will be accepted in order to limit the dose to the brain and teeth bearing bone. Similar considerations apply for the choice between 4.6 and 4.8 Gy (RBE) in case of sarcoma.

Implementation of CIRT at MedAustron can be considered a benchmark for future carbon facilities. For the first time clinical experience obtained with different fractionation schemes and different RBE model will be comprehensively taken into account and merged into one strategy. In addition, RBE conversion will be supported by preexisting clinical data.

Poster Abstract 54: Knowledge-Based Dose-Volume Histogram Prediction for Treatment Planning in Proton Therapy

BACKGROUND: Prediction of achievable dose volume histogram (DVH) of organ at risks (OARs) is expected to improve efficiency and quality of treatment planning in radiotherapy. Many studies have proposed knowledge-based DVH prediction models for intensity-modulated radiotherapy (IMRT). It is unsuitable, however, to apply these prediction models to proton therapy, because dose distribution in proton therapy is strongly dependent on beam angle and weight unlike in IMRT. The aim of this study is to develop a DVH prediction model for OARs using k-means clustering technique for proton therapy planning.

METHODS AND MATERIALS: Each OAR was divided into sub-volumes depending on the distance from the target boundary. The differential DVH of each sub-volume was assumed to be represented as skew-normal probability distribution. Three parameters of skew-normal function of each sub-volume for the case to be predicted were determined based on the similar cases extracted from the database using k-means clustering technique in terms of geometric features. The DVH of the OAR can be calculated as the summation of differential DVHs through all sub-volumes. The model was applied to the cohort of 34 prostate cancer patients treated with proton therapy at Hokkaido University Hospital. Validity of the model was evaluated by the absolute difference between actual and predicted DVH at V37.5 and V60 for rectum and bladder with leave-one-out cross validation method.

RESULTS: The predicted DVH was in good agreement with actual DVH. The prediction error at V37.5 and V60 were 2.1% and 1.3% on average for rectum, respectively, and 1.1% at V37.5 for bladder.

CONCLUSIONS: The DVH prediction method proposed here shows its potential efficacy for patient specific DVH prediction for proton therapy planning.

Poster Abstract 55: Dose-Volume Statistics Comparison of Intensity-Modulated Spot-Scanning Proton Therapy Sparing the Inner Ear and Parotid and Conventional X-Ray Therapy for the Whole Brain in Pediatric and Young Adult Patients with Central Nervous System Tumors

BACKGROUND: Studies demonstrating dosimetric results comparing organs at risk, such as the parotid gland and inner ear, with intensity-modulated spot-scanning proton therapy (IMPT) and conventional X-ray therapy of the whole brain are currently lacking in the literature. The present study analyzed dose-volume statistics for whole brain irradiation with IMPT sparing the inner ear and parotid and conventional X-ray therapy in pediatric and young adult patients with central nervous system tumors.

METHODS AND MATERIALS: Seven pediatric and young adult patients with central nervous system tumors treated with IMPT were selected. The median age was 9 (5–29) years. The prescribed dose to 95% or 99% of the CTV was 23.4–36.0 Gy (RBE) for the whole craniospinal region, and 50.4–56.0 Gy (RBE) for the primary site. To evaluate dosimetric advantages of whole brain IMPT sparing the inner ear and parotid, simulation of whole brain treatment plans with two laterally opposed fields using X-rays were generated. Dose-volume histograms were used to evaluate the dosimetric parameters, and the normal-tissue complication probability (NTCP) for xerostomia and hearing loss was calculated and compared for the two plans.

RESULTS: The mean of the average parotid dose was 8.1 (95%CI: 3.3–12.9) Gy (RBE)/14.1 (95%CI: 7.6–20.5) Gy in the IMPT/X-ray plans (P=.0013). The mean of the mean inner ear dose was 34.2 (95%CI: 30.1–38.1) Gy (RBE)/38.6 (95%CI: 35.0–42.1) Gy in the IMPT/X-ray plans (P<.0001). The mean NTCP for xerostomia was 25.9 (95%CI: 21.5–30.3) %/31.6 (95%CI: 24.9–38.4) % in the IMPT/X-ray plans (P=.0024). The mean NTCP for hearing loss was 55.9 (95%CI: 37.8–74.0) %/67.0 (95%CI: 49.6–84.5) % in the IMPT/X-ray plans (P<.0001).

CONCLUSIONS: IMPT enabled significantly lower dose exposure to the parotid gland and inner ear than conventional X-ray therapy of the whole brain. Additional simulation data using IMRT will be presented.

Poster Abstract 56: Proton Beam Therapy Versus Intensity Modulated (IMRT) Radiation Therapy for Periorbital Malignancies

BACKGROUND: Periorbital tumor location presents a challenge with IMRT due to higher tumor control doses needed (>60 Gy) in the setting of proximity to orbital structures with lower tolerance (<54 Gy). Proton beam therapy (PBT) is felt to be effective in such cases due to its sharp dose gradient.

MATERIALS AND METHODS: We reviewed our database and identified 23 patients with tumor volume extending within 2 cm of the optic apparatus who were treated with PBT. This included tumors arising from the nasal cavity, orbit, maxillary and ethmoid sinuses, scalp, cheek, base of skull and nasopharynx. Comparison volumetric arc therapy plans were generated at time of treatment. Maximum and mean doses to organs at risk (OARs) of interest are listed in table 1 (not shown). Differences between plans were evaluated using paired Wilcoxon signed rank test with corrections to control the false discovery rate.

RESULTS: Median age was 65 and 14 patients were male. Histology was squamous cell carcinoma in 8 patients. 20 patients underwent resection and received postoperative radiation. 2 patients had elective neck radiation and 3 patients received re-irradiation. Concurrent chemo was used in 5 patients. There was a significant reduction in mean/max dose to the bilateral optic nerves, optic chiasm, bilateral cochlea, bilateral lacrimal glands, pituitary as well as mean dose to the normal brain with PBT. There was a trend to significant reduction in dose to other OARs as well. Local control to date is 87% and no visual toxicity has been observed to date.

CONCLUSIONS: PBT resulted in significant dose reductions to several periorbital and optic structures as compared to IMRT. PBT appears to be the optimal radiation modality in such cases to minimize risk of toxicity to visual structures.

Poster Abstract 57: Proton-Minibeam Radiation Therapy Planning Using a Monte Carlo Simulation Toolkit and Clinical Treatment Planning System

BACKGROUND: Recent studies have proven the concept of proton minibeam radiation therapy (pMBRT) using sub- and supra-millimeter proton beamlets. The unique dosimetric characteristic, i.e. the Peak-to-Valley dose ratio (PVDR) and induced proximal tissue sparing, were quantified based on Monte Carlo simulations and radiochromic film measurements. One method to produce minibeams is to apply the pinhole collimators to a raster-scanned pencil-beam field. The purpose of this study is to evaluate modelling a pinhole collimator in a Monte-Carlo based clinical treatment planning system for pMBRT and compare the results with Monte Carlo simulations and film measurements.

METHOD AND MATERIALS: A spread out Bragg peak (SOBP) was created with energies from 140 MeV to 120 MeV. Dose delivered by the SOBP was calculated using a Monte Carlo simulation toolkit, TPS using a Monte Carlo calculation model, and measured using films. A 3.25 cm thick brass collimator with 3 mm diameter holes and 6 mm center-to-center spacing with hexagonal arrangement was used for all models. The PVDR, a metric quantifying the quality of minibeam and proximal tissue sparing of pMBRT, was measured between 0–14 cm depths in increments of 2 cm in water equivalent phantom.

RESULTS: The PVDRs at surface and in the SOBP regions were 3.50, 5.64, and 4.9 and 1.07, 1.05, and 1.11 for TPS, MC simulations, and film, respectively.

CONCLUSIONS: Modeling the collimator in the TPS is feasible but results in a bigger difference from MC simulations and film measurements in the entrance region. The discrepancy is suspect caused by the averaging effect in a dose grid of a comparable size of the collimator hole. The smallest dose grid currently allowed in the TPS (1 mm) is used. However, in order to accurately model the potential dose sparing effects of pMBRT, smaller dose grid on the level of 0.1 mm is desirable.

Poster Abstract 58: Whole Pelvis Pencil Beam Scanning (PBS) Proton Therapy for Prostate Cancer: A Unique Treatment Planning Approach

BACKGROUND: Whole-pelvis PBS proton therapy may be utilized in both the intact and post-operative settings in patients with prostate cancer, yet treatment planning techniques are not well described. We sought to characterize the robust three-field treatment planning workflow implemented at our institution to maximize target coverage in the presence of uncertainties in treatment delivery.

METHODS AND MATERIALS: A review of the planning techniques of two intact and two post- operative patients was performed. All plans used one superior posterior and two inferior lateral fields. Two plans used conventional optimization (CO) and two used robust multi- field optimization (RMFO). CO plans required field specific target volumes (FSTVs) to account for range uncertainty, and manual feathering between the superior and inferior fields to create a slow dose gradient for uniformity. The RMFO plans were optimized to cover the CTV with uncertainty scenarios explicitly included (setup ±5mm in 3D, proton range ±3.5%), and the slow dose gradient was generated automatically by the optimization algorithm.

RESULTS: RMFO greatly simplified the planning workflow, because neither the FSTVs nor the complicated match structures used for manual feathering had to be created. Both planning techniques achieved acceptable target coverage, however, rectum V60 Gy and bladder V65 Gy were higher with RMFO (V60 Rectum 21.4% vs. 14.1% for intact and 19.3% vs. 16.8% for post-operative; and V65 Bladder 25.5% vs. 20.2% for intact and 47.3% vs. 27.8% for post-operative). Plan evaluation was more complex with RMFO because multiple target volume DVHs must be considered (one for each uncertainty scenario).

CONCLUSIONS: Whole-pelvis PBS proton therapy delivered via three-field technique can be optimized via CO or RMFO, with the latter greatly simplifying the treatment planning workflow. Both techniques create treatment plans with acceptable target coverage, but with potential differences in OAR dose. This differential may be due to inter-patient variance, however further study on a larger sample size is needed.

Poster Abstract 60: Dosimetric Characteristics of Off-Axis Spots with the Use of a Range Shifter

BACKGROUND: Commissioning data requirements of pencil beam scanning systems has been made standard by the treatment planning system. Spot measurements are performed with open field and for spots at the central axis of the beam. The minimum proton energy at the Miami Cancer Institute is 70.0 MeV [approximately 4.0 cm-H2O]. Shallower target volumes require the use of a range shifter (RS). Range shifters change the characteristics of spots, especially away from central axis (CAX).

METHOD AND MATERIALS: Beam measurements were taken in air with a scintillator detector and a multilayer ionization chamber. Doses were acquired with and without the presence of the range shifter with water equivalent thickness of 7.5 cm-H2O. We calculated dose distributions using two different algorithms: pencil beam scanning and Monte Carlo. Results of the discrepancies were compared between the two algorithms and the measured data.

RESULTS: The spots modeled in the treatment plan system are consistently symmetrical in shape. However, measured data shows that spots off-axis are not necessarily symmetrical and these changes are more significant for spots close to the edge of a 30cm × 40cm field. The location of the range shifter along the central axis (CAX) also affects the spot characteristics. The measured volumetric dose distributions can be different from the calculations.

CONCLUSION: Most dose calculation algorithms have limitations when range shifters are used. The effect of the range shifter is more significant between the calculated and the measured doses for spots farther away from the CAX. We will present results of that effect on the geometry of the spots and dose profiles. We will compare the spots used in the calculations with the measured spots with and without a RS, and show the effect of change in air gap in longitudinal dose distributions.

Poster Abstract 61: IMPT Whole Lung Irradiation Versus IMXT Whole Lung Irradiation

BACKGROUND: Whole lung irradiation (WLI) may be necessary for specific clinical indications for children. In this report the technique, dosimetry and early clinical outcomes of WLI with intensity modulated proton therapy (IMPT) will be described and compared to WLI with intensity modulated Xray therapy (IMXT). The comparison measures the differences in setup, delivery, and DVH reports for normal tissues and reports early clinical experience.

METHODS AND MATERIALS: Three girls (ages 11, 12, and 17) and 1 boy (age 14) were planned to receive 12–15Gy WLI. Simulation included a gated 4-dimensional (4D) computed tomography (CT) scan. Inter-thoracic motion was evaluated during target and normal tissue volume delineation. IMPT plans used 2 robustly optimized posterior oblique fields to cover the WLCTV. IMXT plans covered a PTV with multiple VMAT arcs. Normal tissue doses, target coverage, patient setup and delivery times were compared.

RESULTS: Overall, normal tissue sparing, most notably the breasts and heart, was consistently demonstrated with the IMPT plans. Patients' mean total breast ranged from 789cGy–741cGy, and a mean heart dose of 1006cGy–732cGy with IMXT. These ranges decreased to 290cGy–68cGy, and 996cGy–578cGy, respectively, in the IMPT plans. Dose sparing was noted also for the esophagus, cord, thyroid, and liver. 4 patients have received WLI with IMPT and have tolerated the treatment well. One patient has been followed for a year and has no disease recurrence, normal lung function, and no symptoms.

CONCLUSIONS: The IMPT WLI yielded lower doses to normal tissues, specifically the breasts, esophagus, heart, cord, thyroid, and spleen with efficient delivery. Early clinical results are favorable.

Poster Abstract 64: Using Static Collimation in lieu of Layer-by-Layer Collimation in Intensity Modulated Proton Therapy for Lung Patients

BACKGROUND: A compact gantry-mounted pencil beam scanning system (PBS) features fast dose delivery with ultra-short energy layer switching time and an adaptive aperture with the capability of layer-by-layer collimation. Despite the dosimetric advantage of the adaptive aperture, interplay effect in lung tumor suffers from the slow motion of adaptive aperture that lags behind energy layer switch. The purpose of this study was to evaluate the feasibility of using a static aperture for lung treatments in lieu of layer-by-layer field shaping.

METHODS: Ten lung patients were selected with varying target sizes ranging from 23cc– 523cc. For each patient, two plans were created using Raystation; one using an adaptive and the other using a static aperture. Each plan was optimized to achieve clinically acceptable coverage (V95% at 100%) and to reduce OAR dose as much as possible using a fixed spot and EL spacing of 0.5 cm. The lateral treatment margins for each plan were determined on a case basis. For each patient, both plans were evaluated at the dose points of V20 and V5 for lung, the heart mean dose, and V98% dose coverage on the PTV for comparison.

RESULTS: The average absolute difference in the mean heart dose, lung V5, and lung V20 between the two aperture techniques were 17.5 cGy, 1.8% and 0.9%, respectively. The PTV coverage of 98% prescription showed an average difference of only 0.64% and the global max absolute average difference was 6.4 cGy..

CONCLUSION: The advantage of using layer-by-layer collimation shrinks compared to a static aperture when the same target coverage is achieved in lung tumors, suggesting that the use of a static aperture is acceptable to take full advantage of fast energy switching.

Poster Abstract 65: A Dosimetric Comparison of Prostate Boost Treatment with Intensity Modulated Protons Versus Photons

BACKGROUND: Recent data has demonstrated improved results with brachytherapy boost in combination with external radiation for high risk prostate cancers. We aim to compare the potential of stereotactic body proton therapy (SBPT) versus stereotactic body x-ray therapy (SBXT) to simulate prostate brachytherapy boost dosimetry.

METHOD AND MATERIALS: Twenty-five patients treated for prostate cancer at our institution were selected and stratified by prostate size. All cases were re-planned with SBPT and SBXT. For SBXT, the planning target volume was generated by a 2mm isotropic expansion (except 0mm posteriorly) from the prostate. For SBPT, robust optimization was utilized to account for setup (2mm) and proton range uncertainties (2.5%). A dose of 19 Gy in 2 fractions was planned with target doses maximized to the upper limit of normal tissue constraints. Mann-Whitney U test and Spearman's rank correlation were utilized in data analysis.

RESULTS: The median target volume was 43 cc (range 15–134 cc). The mean conformity index for SBPT and SBXT was 1.8 and 1.3 (P < 0.001), respectively. V100 values did not differ significantly between SBPT and SBXT, and the median V100 differences were not correlated with prostate size. However, V125 and V200 volumes were significantly larger for SBPT than SBXT. Median differences for V125 and V200 also correlated with increasing prostate size with Spearman's correlation coefficient of 0.43 (P = 0.03) and 0.66 (P < 0.001), respectively.

CONCLUSION: Larger high dose regions (V125 and V200) were achievable within target tissues with SBPT than SBXT, and were correlated with increasing prostate size. These results are relevant to studies regarding high risk prostate tumors not amenable to brachytherapy boost.

Poster Abstract 66: Pencil-Beam Scanning Proton Radiography Using A Flat-Panel Imager

BACKGROUND: Proton radiography (PRG) allows for measurement of the water-equivalent path length (WEPL) in tissue, which can be used to determine relative stopping power (RSP), reducing range uncertainties and allowing for more conformal proton treatment planning. In this work, a PRG technique using a flat-panel imager and a pencil-beam scanning (PBS) system is demonstrated on phantom data.

METHODS: Proton PBS plans were delivered to a Perkin-Elmer XRD 1611 detector. Each proton plan consisted of energy layers separated by 4.8 MeV, and square field sizes at 15 cm and 25 cm. A pixel-by-pixel detector response function is built to calculate WEPL from intensity data. Tissue equivalent phantoms are used for quantitative WEPL analysis. A spatial resolution phantom and an anthropomorphic body phantom are also imaged. For all experiments, the detector had an effective pixel size of 0.4mm×0.4mm.

RESULTS: The imaging results are summarized in Fig. 1 (not shown). On average, RSP error of the imaging method was 0.35% with an RMSE of 1.29%. The largest error was in the low-density LN-300 lung insert (3.28%). In the spatial resolution phantom, a 2 mm deep pinhole with 1 mm diameter can be seen.

CONCLUSIONS: We have demonstrated an accurate method for measuring the WEPL in phantoms using a single flat-panel imager on a PBS system. The accuracy and spatial resolution of the method show that it could be implemented for patient position verification. Further development could lead to patient-specific measurements of RSP for treatment planning.

Poster Abstract 67: Dosimetric Evaluation of Proton Therapy for Chest Cancer Patients with MRI Safe Vascular Access Port Implants

PURPOSE: To investigate dosimetric discrepancies in proton therapy treatment planning resulting from inclusion of MRI safe vascular access ports in the beam path.

METHODS AND MATERIALS: Vascular access port (VAP) implants are composed of materials of different properties that generate streak artifacts in CT scans and result in increased proton beam transport simulation inaccuracies if not properly taken into account. A study comparing the dose statistics from two TPS plans using HU-to- density tables and manual density overrides was performed on a CT scan of a POWERPORT CLEARVUE ISP MRI safe implantable port (Bard Access Systems, Salt Lake City, Utah) immersed in water. In both CT data sets, a target was placed at a depth of 0.5cm distal to the VAP. A plan calculated with Monte Carlo algorithm to deliver a prescribed dose to the target was done on the first CT set in which VAP components were contoured and overridden. Assuming a worst-case scenario, Bi2O3 was overridden to 11.32g/cm3, the plastic with silicone overmold and polyurethane were overridden to 0.94, and 1.2g/cm3 respectively. An AP/PA opposed beam technique with different beam weighting was considered. The calculated plan was then painted to the second CT set in which our standard TPS HU-to-density conversion table was used.

RESULTS: Target coverage, assessed via the target volume receiving the prescribed dose (V100) deviated by 0.21% between these two plans. The AP/PA opposed beam technique with 1/3 and 2/3 beam weighting respectively, provided a more robust plan compared to one with equal weighting.

CONCLUSIONS: Contouring the VAP and manually overriding its materials in this case had minimal dosimetric impact to the treatment plan quality. It is therefore feasible to use TPS HU-to-density conversion tables during treatment planning of chest cancer patients with MRI safe VAPs if opposed beam and differential beam weighting are used to achieve more robust plans.

Poster Abstract 68: Acute Toxicity Outcomes in Salivary Gland Carcinoma Patients Treated with Adjuvant or Definitive Proton Radiotherapy

BACKGROUND: Proton therapy (PT) has demonstrated reduced radiation dose to adjacent, non-targeted organs, which may reduce radiotherapy-induced toxicity. This series reports acute toxicity (AT) in patients with salivary gland malignancies (SGM) treated with PT.

METHODS: From 2016–September 2018, 21 patients were treated with PT for SGM at a single institution. Charts were retrospectively reviewed according to an Institution Review Board approved protocol to record tumor and treatment-related characteristics. Toxicity was prospectively assessed according to Common Terminology Criteria for Adverse Events v. 4 and recorded in the patient's chart. Fisher's exact test and Wilcoxon Rank Sum test were performed to evaluate association of tumor and treatment-related characteristics with toxicity.

RESULTS: Twenty-one patients underwent adjuvant (N=16) or definitive (N=5) PT for a primary (N=17) or recurrent (N=4) SGM. Median age was 66 (range; 20–80) years. Most patients had SGM arise in a major salivary gland (N=13) vs. minor salivary gland (N=8). The parotid gland was the most common location (N=11).

Median follow-up was 11.8 months. All patients had locoregional control at last follow-up. Two patients developed pulmonary metastases 5 and 8 months following PT. Three patients experienced grade 3 toxicities; dermatitis (N=1) and mucositis (N=2). Twenty patients experienced at least one grade 2 toxicity; 13 patients experienced >1 grade 2 toxicity. Clinical perineural invasion (cPNI) was the only factor associated with >1 gr 2 toxicity (p=0.0068). Stage, anatomical subsite of the primary, PT dose and intent, and elective inclusion of lymphatics or cranial nerves in PT fields were not significantly associated with toxicity.

CONCLUSION: PT for SGM appears to have acceptable rates of AT. cPNI requires high dose PT tracking cranial nerves to the skull base which increases high dose target volume and likely accounts for increased risk of multiple grade 2 toxicities. Longer follow-up is needed.

Poster Abstract 69: Neutron Dose Equivalent Measurements using WENDI-II in Spot Scanning Beam Proton Therapy

BACKGROUND: Radiation-induced second malignancies are one of the important late side effects of radiation therapy. In proton therapy, the additional exposure to the patient's body other than the treatment site usually comes from secondary neutrons produced by proton nuclear interactions. Compared to the passive scattering beam, the spot scanning beam is believed to have less neutron contamination during the treatment since there are fewer materials in the beam to interact with.

METHOD AND MATERIALS: In this study, the magnitude of neutron dose equivalent depending on the location of the detector was measured using a WENDI-II neutron detector and different solid water phantoms. We also measured the neutron dose in different thickness of range shifters. In addtion, the scattered photon dose was measured to analyze the dose component.

RESULTS: Majority of the secondary exposure of proton therapy is contributed by neutron. The neutron dose equivalent is approximately 1mSv per Gy and varies by beam range, beam modulation, distance from the edge of the beam, thickness of range shifter and beam direction.

CONCLUSION: For spot scanning beam proton therapy, the scattered dose equivalent to the patient's healthy tissue is very low. The results also suggest that by applying ALARA principle, using thinner range shifter is preferred during clinical treatment planning.

Poster Abstract 70: Comparison of Fiducial Versus Liver Alignment for Liver-Directed Proton Therapy

BACKGROUND: Our center utilizes fiducials and breath hold (BH) during pencil beam scanning proton therapy (PBS-PT) for liver malignancies. Fiducial placement is an invasive procedure and we hypothesize that liver alignment provides similar target coverage with BH techniques.

METHODS AND MATERIALS: We performed a single-institution retrospective analysis of patients treated with hypofractionated PBS-PT using a voluntary BH spirometry system (SDX, Dyn'R, France) for liver malignancies. Each patient underwent quality assurance computed tomography (QACT) scans during the treatment course which required rigid registration to the initial treatment scan aligned to fiducials. Clinical target volume (CTV) contours were then copied and verified by a physician. In our study, the QACT scans underwent another registration to the initial plan using an external liver contour match. Alignment changes were assessed by collecting couch shift differences in the medial-lateral (ML), anterior-posterior (AP), and cranial-caudal (CC) directions between the two alignment isocenters. Target coverage was evaluated by calculating the CTV receiving 95% and 100% of the prescription dose (V95% and V100%).

Qualitative assessment of liver-registered alignment was performed by evaluating whether the fiducials remained inside a five-millimeter expansion of the original fiducial contour. Statistical significance was performed using paired t-tests (2-sided).

RESULTS: From 2018–2019, 13 QACT plans from seven patients showed a mean isocenter shift in the ML, AP, and CC directions of 1.1 millimeters (mm) (standard deviation (SD) 1.1mm, p=0.616), 1.2mm (SD 1.7mm, p=0.682), and 2.5mm (SD 2.4mm, p=0.994), respectively. The mean ratios of liver to fiducial-matched V95% and V100% were 100.12% (SD 11.46%, p=0.917) and 100.56% (SD 2.78%, p=0.957), respectively. Liver-registered fiducials were all at least partially within the original fiducial contour plus a 5mm margin.

CONCLUSIONS: In patients undergoing liver-directed PBS-PT with breath hold, liver alignment appears to be a promising alternative to fiducial alignment. There was no statistical difference in couch shifts or target coverage, suggesting the potential to avoid invasive fiducial placement in select patients.

Poster Abstract 71: Single Field Optimization versus Limited-Modulation Multifield Optimization Proton Treatment Planning for High-Risk Neuroblastoma

BACKGROUND: Proton therapy is increasingly used in the treatment of neuroblastoma to limit doses to organs at risk (OAR). Multifield optimization (MFO) plans are more susceptible to physical uncertainties than single field optimization (SFO). Limited-modulation MFO (LM- MFO) may achieve better OAR sparing while mitigating these uncertainties. In this study, we compared OAR sparing in SFO and LM-MFO abdominal neuroblastoma patients treated at our institution.

METHOD AND MATERIALS: Five high-risk neuroblastoma patients received 21.6 CGE/12 Fx (with a minimum 18 CGE to adjacent vertebral bodies). For each case, comparison plans were generated using SFO and LM-MFO techniques using three fields. Each plan was optimized to achieve the lowest possible kidney dose while meeting institutional target volume coverage and robustness standards. For each LM-MFO plan, the absolute dose difference between neighboring voxels were collected for each beam separately. This data was then optimally-binned into a histogram and weighted averages were calculated based on the count per bin and the bin edges; this value represented the beam modulation score. The total modulation score was taken as the average of these values divided by 100.

RESULTS: LM-MFO plans resulted in lower mean doses to the both kidneys and all organs at risk in every case compared to SFO plans. The average reduction in dose to the Left and Right kidney was 4.4 CGE and 4.8 CGE, respectively. Target coverage and robustness metrics were equal in all scenarios. Dose modulation scores ranged from 36.99 to 45.26.

CONCLUSIONS: LM-MFO may reduce the dose to the kidneys and other OARs compared to SFO while maintaining target volume coverage and robustness. Additionally, we present a novel method to quantify dose modulation in pencil beam scanning proton therapy.

Poster Abstract 72: Development of an Innovative Tool to Account for Prior Radiotherapy during Treatment Planning

BACKGROUND: Accounting for previous radiotherapy (RT) is instrumental in designing and evaluating a re-irradiation treatment plan. However, given the frequent difficulty with obtaining previous electronic RT dose files, planners often resort to reconstructing dose distributions from scratch, which can be cumbersome and resource intensive. In this study we explore a novel, programming-based method to automatically produce customized dose distributions for incorporation into the treatment planning system (TPS) as background dose.

METHODS AND MATERIALS: In 2019, a patient with recurrent maxillary sinus cancer received treatment at our institution with pencil beam scanning (PBS) to 62.4Gy. The patient was previously treated to 60Gy to the same area in 2010. Maximum composite doses to serial OARs were given priority over target coverage. No electronic previous RT dose files were available; however, paper records indicated the previous maximum doses received to nearby serial organs at risk (OARs). Using this information as inputs, an in-house program was developed to produce uniform, worst case scenario (WCS) dose distributions for each OAR.

RESULTS: Multiple, uniform dose distributions were generated to correspond with each serial OARs' previous maximum dose. A clinical PBS plan was produced using the artificially-created dose distribution as background dose. This allowed the visualization of a composite dose distribution and dose volume histogram (DVH). Additionally, the utilization of composite dose objectives facilitated more intuitive sparing of serial OARs, such as the chiasm and brainstem, compared to specifying course-specific limits. This method avoided time consuming attempts to generate dose distributions using arbitrarily created intensity modulated radiation therapy (IMRT) plans.

CONCLUSION: As patients undergoing re-irradiation become more common, optimal treatment planning is critically important. The tool presented herein helps clinicians to easily account for previous RT doses to nearby serial organs and visualize composite dose when previous RT dose files are unavailable.

Poster Abstract 73: Proton Beam Radiotherapy for Patients with Pituitary Adenomas is Associated with Low Acute and Late Treatment-Related Toxicities

OBJECTIVE: To report on the treatment-related toxicities associated with proton beam radiotherapy (PBT) for patients with pituitary adenomas.

METHODS: Patients treated with PBT for newly diagnosed or recurrent disease were selected from a prospective multi-institutional registry. Acute and late treatment-related toxicity outcomes were measured according the CTCAE v.4.

RESULTS: Forty-six patients met eligibility criteria for inclusion in this study. The median age at treatment was 53 years (Range: 21–82). Ten patients (22%) had previously received photon radiotherapy before salvage PBT. Forty-three patients (93%) underwent resection prior to PBT at a median of 10 months (Range: 2–238). The median dose was 50.4 Gy in 28 fractions (Range: 25.2 –59.5 Gy in 21–60 fractions). Nineteen (41%) patients were treated with pencil beam scanning technique and 27 (59%) were treated with uniform scanning/passive scatter. The median follow-up from the last date of radiotherapy was 32 months (Range: 4 –131). Of the patients with clinical and imaging follow-up, there was only a single patient who had disease progression, and the remaining patients had stable or controlled disease at time of last follow-up. Nine patients (20%) experienced an acute grade 2 toxicity and two patients (4%) experienced acute grade 3 toxicities: blurred vision and headache. Two patients (4%) experienced a grade 2 late toxicity and no grade 3 late toxicities in any domains were observed.

CONCLUSIONS: In this prospective multi-institutional study of patients treated with PBT for pituitary adenomas, few significant acute and late treatment-related toxicities were observed. Prospective neurocognitive and quality-of-life studies are needed to determine additional benefits of this treatment technique in this patient population.

Poster Abstract 74: Maximizing Efficiencies in Radiation Therapy while Maintaining Perfection

BACKGROUND: Incorporating a Proton facility into a well-established department can lead to many challenges. Staff cohesiveness, increased efficiencies, and improved communication are a must.

METHOD AND MATERIALS: Mayo Clinic – Rochester has implemented several staffing models to improve therapist relations. We have found ways to improve efficiencies from patient, radiation therapist and workload perspectives. One method we used to increase efficiency and communication was implementing Vocera communication devices. Looking at workload treatment times, utilization of beam time, and in the future implementing a gatekeeper strategy is continuing to improve.

RESULTS: Since enhancing scheduling, interactions among therapists have improved and appreciations for other modalities work. Intermixing modalities has also help aid in understanding and development of ideas.

CONCLUSIONS: Treating more patients, who are often difficult, with perfection in a timely manner is challenging. Although, with proper communication, staffing, and developing tools our experience has greatly improved.

Poster Abstract 75: A Supplemental Technique for the Analysis of Pre-Treatment Quality Assurance

BACKGROUND: PSQA measurement analysis depends on generating metrics representative of calculation and measurement agreement. This work introduces a dose-plane comparison algorithm, based on a cylindrical search shape (as opposed to ellipse-based γ-test), with search limits empirically determined from machine QA. Dose plane agreement is reported at each measured pixel as the dose difference minimum (DDM) within an empirically-established search radius: ΔDmin(r).

MATERIALS AND METHODS: Beamlet accuracy was sampled over a 6-month period across the full deliverable field dimensions to determine a weighted radial positional accuracy at a 99.5% confidence level. DDM analysis was performed for 75 scanning-beam proton patients. Image-registration shifts were tracked to monitor set-up error and to isolate the search plane radius to beamlet deviation. Pass rate was assigned as the percentage of measurements where <3% dose agreement was found within the determined search limit.

RESULTS: >99.5% of radial beamlet deviations were less than 1.0mm. The pass rate (mean +/− std. dev) saw no change between a 3%/2mm γ-test (97.7 +/− 3.2%) and a 3%/1mm ΔDmin(r) (97.6 +/− 3.2%).

DISCUSSION: Spot-delivery spatial accuracy was well within 1mm based on extensive QA and delivery logs. However, the elliptical shape of the γ-test is too exclusive at 1-mm. The cylindrical search shape of the new algorithm accepts all pixels with <3% agreement inside the search area, proposed herein as more relevant to plan quality, and also provides additional diagnostic information by reporting dose deviation magnitude per pixel.

Poster Abstract 77: Treatment Duration as a Function of Machine Parameters for a Proton Pencil Beam Scanning Synchrotron with Multi-Energy-Extraction

BACKGROUND: Mayo Clinic Rochester switched to treating with multi-energy extraction (MEE) in January 2018. The significant reduction in treatment duration due to MEE has been previously reported. There are several additional factors in synchrotron pencil-beam scanning (PBS) which contribute to the irradiation time of a treatment field; however, changing these parameters would require modifications to the existing machine. This study looked at the contribution of these factors to irradiation times by simulating various treatment deliveries.

METHODS AND MATERIALS: A synchrotron model was previously created for a 4D dose calculator. Several parameters of this model were varied and the irradiation times for two patient plans were calculated. The plans were a typical head and neck patient and a liver patient both with and without repainting. The varied parameters were: time to calculate delivered spot characteristics before delivering the subsequent spot (SCT); scanning magnet (SCM) speed; extracted beam current; charge in synchrotron; recapture efficiency between MEE energies; time to switch MEE energies; and max MEE layer switches per synchrotron recharge.

RESULTS: Reducing SCT had the greatest overall impact of any individual parameter as it reduced irradiation times by up to 40%. Improving SCM and the beam current decreased the irradiation times by 7–10% and 9–13%, respectively. The MEE recapture efficiency decreased the non- repainted plans' irradiation times by 16–20% while only decreasing the repainted delivery by 6%. Improving all parameters decreased the irradiation times by 90% for all three deliveries.

CONCLUSIONS: Irradiation times were reduced by a factor of 10–12 for the three deliveries when improving all parameters and would allow a compact synchrotron to treat a liver plan via breath hold without repainting (11–12 s per field). The combination of improving SCT and SCM would allow PBS to approach continuous scanning by reducing inter-spot beam-off time.

Poster Abstract 78: Multi-Energy Isocenter Analysis in Proton Pencil Beam Scanning System

BACKGROUND: Proton pencil beam scanning (PBS) gantries are substantially heavy and usually present mechanical sag, which is generally corrected by means of the beam optics. Because beam optic parameters are energy dependent, fine beam steering and isocenter location tend to be energy dependent. We present a method to quickly evaluate the isocenter location across the energy spectrum as well as the mechanical stability of the gantry known as corkscrew effect in an IBA Proteus Plus gantry.

METHOD AND MATERIALS: The isocenter position is evaluated using Logos XRV-124 (Logos Systems), a device composed of a conical scintillator detector and a camera. A plan with ten energies ranging from 70 MeV to 220 MeV per gantry angle is produced based on a CT scan of the device. Ten energies are shot per gantry angle. Eight to twelve equally spaced gantry angles are selected. Device is aligned as usual based on the planning images. The data is acquired with gantry rotating clockwise (CW) and counter-clockwise (CCW).

RESULTS: Splitting the energies and analyzing them per gantry angle allowed us to visualize the spread of the isocenter location as function of energy and gantry angle. We observed the spread of the isocenter per energy to be around 0.5 mm, while its spread per gantry angle is around 1 mm. We also observed that the longitudinal position of the isocenter is sensitive to the direction of the rotation when corkscrew effects are in place.

CONCLUSION: We observed that fine beam steering is usually tight around 0.5 mm and that the mechanical sag is corrected to about 1 mm. Testing both directions of rotation of the gantry helped to diagnose the corkscrew effect caused by a mechanical problem.

Poster Abstract 79: Fast Tomography for Real-Time Precision Beam Imaging

Pixelated readouts of detectors such as those used for quality assurance of clinical particle therapy beams and patient treatment plans have the advantage of being able to image very general dose distributions. However, the number of signal channels required for pixelated planar readout is linearly dependent on the required sensitive area, and quadratically increases as the required spatial resolution decreases. Alternatively, one may reduce the number of readout channels by judiciously grouping together regions of the sensitive area, and applying a reconstruction technique to infer the pixelated image: a process which we refer to as tomography. A simple example of such an approach is the combination of two orthogonal 1-dimensional strip readouts found in many gas and silicon detectors.

We take a generalization of the strip readout approach using the method of Gaussian Process Tomography (GPT), and examine the imaging capabilities and limitations of various readout configurations in the context of a parallel-plate ionization chamber. For example, we study clinical use cases for a three-dimensional strip-like planar readout, as well as more academic scenarios such as random association of readout pads. Additionally, we study the feasibility of using this generalized approach for real-time imaging of clinical proton beams at a rate of 10 kHz.

We conclude that an approximation of GPT combined with dynamic pixelation of the reconstructed image is both practical and effective for precise real-time imaging of proton therapy beams.

Poster Abstract 80: Development of a Collimated Gamma-ray Detector for In-vivo Range Verification

BACKGROUND: In theory, proton therapy allows doctors to target tumors which sit near critical structures. In practice, doctors must take a more cautious approach due to uncertainty in where the protons stop, known as range uncertainty. Range uncertainty is accounted for in treatment planning primarily using two techniques: choosing low risk, suboptimal treatment plans which do not place the end of range against critical structures, and prescribing a margin (beam overshoot to ensure tumor coverage) of typically 3.5% + 1 mm. These techniques counteract the intrinsic advantages offered by proton therapy. Range uncertainty prevents proton therapy from reaching its full potential.

METHODS AND MATERIALS: One method to reduce range uncertainty is to directly measure the range during treatment. In order to prevent any harmful dose from being delivered, the detector should be able to measure the range to within 1 mm in less than 250 ms. In this work, a compact detector consisting of a collimated array of Thallium-doped Cesium Iodide scintillators was developed to measure the range through detection of prompt gamma-rays emitted during treatment.

RESULTS: Testing of the device was performed at Mayo Clinic in Phoenix, AZ. Using an acrylic phantom, the device is able to measure the proton range in-vitro to within 1.1 mm at depths up to 11 cm in less than 500 ms and up to 7.5 cm in less than 200 ms. Future upgrades are planned with the goal of improving on these results.

CONCLUSIONS: A new device has been tested at a proton therapy center. Initial results suggest that it is feasible to measure the range in-vivo in less than 250 ms using this approach.

Poster Abstract 84: A Planning Comparison of IMRT versus Pencil Beam Scanning For Deep Inspiration Breath Hold Lung Cancers

BACKGROUND: Deep inspiration breath hold (DIBH) has dosimetric advantages for lung cancer patients treated with external beam therapy, but is difficult for many lung cancer patients to perform. Proton therapy for lung cancer permits sparing of the contralateral lung and other organs at risk (OAR) as the finite range spares downstream OARs. We compared conventionally fractionated proton (p) and photon(x) plans on both free breathing (FB) and DIBH planning CTs to determine how much DIBH would affect proton therapy. We also compared DIBH IMRT to FB proton plans.

METHODS AND MATERIALS: We evaluated 6 patients treated with photon DIBH on a prospective protocol from 5/2014–10/2018. Six patients had locally advanced disease and were treated with conventional fractionation. One patient had early stage disease and was treated with moderate hypofractionation (60 Gy in 15 fractions). All patients were re-planned using pencil beam scanning proton (PBS) therapy. Additionally, new plans were generated for FB datasets with both modalities for a total of 4 plans/patient. Prescription dose ranged from 50–70 Gy or Gy(RBE) with the goal of having the 95% isodose line cover 95% of the PTV. To compare the effect of DIBH vs FB, we calculated the difference between dosimetric parameters for heart and lung for x-ray and proton plans. We also compared FBp to DIBHx parameters to quantify how FB proton plans compare to DIBH IMRT plans. A 5 mm PTV margin was used for all plans, which was sufficient to ensure ITV coverage within a ±3.5 % range uncertainty.

RESULTS: Regarding the population differences between the FB and DIBH plans for the two different modalities: For V20, the differences were comparable, suggesting that DIBH has comparable advantages for both PBS and IMRT at the V20 level. For all other metrics, the proton gains of DIBH over FB were smaller than IMRT. Since PBS produces a smaller low-isodose volume than IMRT, the relative advantage that DIBH offers is reduced. The population mean lung for DIBHx V20 (29.6%) and MLD (16.6 Gy) were comparable to FBp V20 (28.4%) and 16.7 Gy (RBE) while other metrics favored FB PBS. For instance, for DIBHx vs FBp the population mean lung V5 and MHD were 57.7% vs. 40.7%, and 15.8 Gy vs. 6.8 Gy (RBE), respectively.

CONCLUSION: DIBH may offer similar relative advantages for PBS as it does for IMRT as measured by V20. The gains from DIBH for V5, MHD and MLD are smaller for PBS than IMRT. On average, FB proton plans offer similar or better lung and heart sparing compared to DIBH IMRT plans. For IMRT patients who have difficulty performing DIBH, FB protons may offer an alternative.

Poster Abstract 85: Commissioning of McLaren Proton Therapy System

INTRODUCTION: McLaren Proton Therapy Center (MPTC) is equipped with a Radiance 330TM synchrotron which is capable of pencil beam delivery (70–330 MeV). The treatment rooms are equipped with a robotic couch for patient positioning and half gantry for beam delivery. The in-room imaging system is capable of acquiring planar and CBCT images while mounted from an independent x-ray gantry.

METHODS AND MATERIAL: The ionization depth doses for 70 to 250 MeV were measured using PTW water tank and Bragg peak chambers. An IBA Lynx was used to measure the beam sigma at isocenter as well as at four different positions relative to isocenter. The IAEA TRS 398 protocol was used to calibrate the delivered dose. The dose distribution was verified using gamma index analysis. The dose calibration and dose distribution were verified by IROC Houston for a prostate phantom. The gantry mechanical isocentricity was measured using an in-house fabricated device. Isocentricity shifts are accommodated by correcting the treatment couch positions for gantry sag for various gantry angles.

RESULTS: The ranges of proton beams are verified to be within 0.5 mm of the tabulated CSDA values. The spots circularity is verified to be within 10% in X and Y axis. The positional accuracy of the spots is within 1.5% of the planned map. The gantry isocentricity is within 0.5 mm radius after couch correction.

CONCLUSION: The gamma analysis of dose distributions had a passing rate of >95% for 2%/2mm. Independent verification by IROC has verified our beam delivery calibration and accuracy.

Poster Abstract 87: Four-Field Head and Neck Pencil Beam Scanning Proton Therapy Optimization

BACKGROUND: Pencil beam scanning (PBS) proton therapy offers dosimetric advantages for several treatment sites, including head and neck (HN). The potential dose reduction in the oral cavity, parotids, esophagus, and larynx can often improve quality of life. To best utilize PBS for these patients, we found that four fields offer better conformality and robustness than three or fewer fields. However, four-field robust optimization is potentially prohibitively time consuming, especially with Monte Carlo (MC) dose calculation. We developed an optimization strategy for HN PBS proton plans to balance robustness and planning time.

METHODS AND MATERIALS: Bilateral HN patients were planned in RayStation v6 with MC optimization and final dose calculation and a four-field modified X configuration. Anterior beams were 30° from 0, and posterior beams were 45° from 180°. Avoidance structures were created to prevent spot placement based on uncertainties. Posterior beams did not place spots inferior to the top of the shoulders. For the anterior beams, an avoidance structure was created using the oral cavity and any dental hardware plus an 8mm expansion. Targets were planned robustly to D97%>97%. Robustness settings included 0.3cm right, left, superior, and inferior and 2% range uncertainty, for a total of 15 perturbations per iteration.

RESULTS: Plans typically require several rounds of 100 iterations, requiring approximately 90 minutes of optimization time per round. After each round, OAR objectives are adjusted to continually improve the plan. Final MC dose is calculated with 0.5% uncertainty. These plans, without independent beam robust optimization, maintain good target coverage and OAR sparing across 18 unique perturbations, including plan shifts and rotations, range uncertainty, and independent beam isocenter shifts.

CONCLUSIONS: Other HN planning techniques require independent beam robust shifts, which can significantly increase planning time. With this technique, we standardized our HN plans and reduced planning and evaluation time.

Poster Abstract 88: Performance of 2D and CBCT Imaging for Patient Positioning at McLaren Proton Therapy Center

BACKGROUND: The characteristics of proton beams allow for steep dose fall off between normal tissue and target. Evaluating the accuracy of patient positioning is important to ensure correct dose delivery. The McLaren Proton Therapy System (MPTS) is equipped with a kV-imaging system capable of orthogonal and CBCT imaging from an independent gantry.

METHODS AND MATERIALS: The system was tested according to requirements outlined in TG-142 and TG-179. The CIRS Isocube device and anthropomorphic phantom were simulated with a slice thickness of 1mm. Orthogonal and CBCT imaging fields were planned for image registration evaluation. The accuracy was assessed by placing the phantoms at a known offset from isocenter and imaging at three couch angles 0, 180 and 270 degrees with both modalities. The imaging hardware is controlled by EhmetDx XIS (X-ray Imaging Software) and the image registration is done with MIM Software.

RESULTS: Deviations from expected values were below 1 mm and 1 degree. Average deviations for orthogonal imaging in all three couch positions were 0.25, 0.29 and 0.16 mm for x, y, z and 0.07, 0.00 and 0.00 degrees for yaw, pitch, and roll. Average deviations for CBCT in all three couch positions were −0.05, −0.11 and 0.02 mm for x, y, z and −0.06, 0.11 and −0.22 degrees for yaw, pitch, and roll.

CONCLUSION: The use of both modalities with a static solid phantom yielded comparable results with repositioning after image registration and all tested couch angles within 1 mm. The imaging system has had an excellent and stable performance during clinical use and routine quality assurance testing.

Poster Abstract 89: Daily Offline Range Verification of Proton Beams at The McLaren Proton Therapy Center Using a Commercial Multi-Layer Faraday Cup

OBJECTIVE: To study the use of a Multi-Layer Faraday Cup (MLFC) for a quick and precise daily range verification of proton beams at McLaren Proton Therapy Center.

METHODS: Depth dose measurements were performed at room iso-center using a water tank and Bragg Peak ion chamber. The IBA Giraffe was used to measure the water equivalent thickness (WET) of the sample copper plates used in the MLFC. The WET measurements provided the range calibration factors for the MLFC. To establish a baseline for in-room measurements, ranges were measured using the MLFC for energies from 70 to 250 MeV in steps of 10 MeV. Daily range verification measurements are performed for five representative beam energies (70, 100, 150, 200 and 250 MeV) with the MLFC, which is permanently placed at the end of the beam line inside the accelerator vault. Data collected over a period of more than 100 days are analyzed and presented.

RESULTS: The centroid channel number in the MLFC where the protons stop was calculated and converted to depth in water and compared to the depth of distal 80% measured in the water tank. The depths agreed to within 2 mm. The daily variation in ranges measured by the MLFC was within ±0.5 mm. The total measurement time is less than 5 minutes.

CONCLUSION: Based on the measurement results, the MLFC can be used for a daily range constancy check with submillimeter accuracy. It is a quick and simple method to perform range constancy verification on a daily basis.

Poster Abstract 90: Commissioning and Implementation of the Automatic Gated Breath Hold Technique for Pencil Beam Scanning Proton System

BACKGROUND: To report the commissioning and implementation of the automatic gated breath- hold system using pencil beam scanning (PBS; ProBeam, Varian, USA).

METHOD AND MATERIALS: Commissioning: The voluntarily breath-hold spirometry system (SDX, Dyn'R, France) with automatic gating module was commissioned at the Maryland Proton Treatment Center by performing point and 2D-planar dose measurements of 5 gated plans (3–4 fields per plan). For each field, four measurements were performed: three with 2, 3 and 5 breath- hold which were evaluated against the one without breath-hold (reference). Implementation: After instructing the patient how to breathe and hold the breath through the spirometer, based on the deepest inspiration breath-hold (DIBH), 70% to 80% of DIBH will be set as the breath-hold level. Next, the CT images will be acquired using SDX breath-hold system as well as normal 4D-CT (as a backup treatment) following by treatment planning on both image sets. After treatment planning, the physician will evaluate both plans. If the breath-hold plan is chosen, both plans get ready for patient specific QA and the non-breath-hold plan is used as a backup plan. At the time of treatment, kV and CBCT images will be acquired at the pre- defined breath-hold level for image guidance. After the patient setup, the treatment will proceed with the automatic gating module active and connected to the ProBeam system.

RESULTS: The maximum percent difference of point dose measurements and the lowest gamma passing rate between non-breath-hold and breath-hold plans were 0.2% and 97.2%, respectively. Of 48 patient that were consulted for SDX treatment, 37 underwent SDX simulation and 35 were treated with SDX plan.

CONCLUSIONS: The automatic gated breath hold system was successfully commissioned and implemented for PBS proton system.

Poster Abstract 91: Dosimetric Comparisons between Breath-Hold and Non-Breath-Hold Plans in Proton Therapy

BACKGROUND: The breath-hold technique can be used to mitigate motion of the target and to minimize the target margin and reduce the dose to normal tissues during proton radiotherapy. The purpose of this study was to perform dosimetric comparison between breath-hold and non-breath- hold plans.

METHOD AND MATERIALS: Twenty-three consecutive patients treated with a voluntarily breath-hold spirometry system (SDX, Dyn'R, France) were used for dosimetric analysis. For all patients, the breath-hold level was set to 75% of deepest inspiration breath-hold. Clinically acceptable breath- hold and non-breath-hold plans were created on breath-hold CT and 4D-CT (with or without compression belt) simulation scans, respectively, and the dose-volume histograms (DVH) of the two plans were compared for each patient.

RESULTS: Patients had mediastinal/lung (n=10), liver (n=11), or upper abdominal (n=6) malignancies. The average target volume of breath-hold scans, CTV1 (initial: 27 patients) and CTV2 (small field boost: 11 patients) were reduced to 70±24% and 69±22% of ITV1 and ITV2 of 4D-CT scans, respectively. Figure 1 (not shown) shows the ratio of breath-hold (SDX) plan parameters normalized to non-breath-hold plan parameters for liver group (11 patients) and mediastinum/lung group (10 patients). Compared to non-breath-hold plans, the mean dose to liver, stomach, kidney, esophagus, heart, and V20 of total lung of breath-hold plans were reduced to 72%, 71%, 67%, 90%, 75%, and 81%, respectively. The small bowel, large bowel, heart, and spinal cord max doses were lowered to 81%, 58%, 94%, and 84%, respectively.

CONCLUSIONS: This study showed that breath-hold plans can significantly reduce the treated target volume and consistently lower the mean and maximum doses to the organs at risks by about 10– 40%.

Poster Abstract 93: Breath-Hold Plan Reproducibility in Proton Therapy Treatment

BACKGROUND: The breath-hold technique minimizes target margins and improves normal-tissue sparing in proton radiotherapy. We investigated the reproducibility of breath-hold plans using frequent quality assurance CT scans (QACTs).

METHOD AND MATERIALS: Eighteen consecutive patients with thoracic or upper gastrointestinal malignancies (7 mediastinum/lung, 7 liver, 2 abdomen, 1 pancreas, 1 esophagus) treated with the voluntarily breath-hold spirometry system (SDX, Dyn'R, France) were analyzed. Reproducibility of the breath-hold plans were assessed using 1–3 QACTs for each patient and re-calculating the initial treatment plan on the QACTs. When the target V95% decreased by more than 5% of the initial plan or the dose to critical organs at risk (OARs) increased significantly, replanning or repeat of the QACT was performed. To evaluate the reproducibility of each plan, variations in dose- volume histograms (DVH) of QACT plans with respect to the initial CT plan were studied and the error window (EW) required to cover the 95th percentile variations was reported.

RESULTS: 4% and 7% EWs were observed with respect to CTV1 (initial, 18 patients) and CTV2 (conedown, 7 patients) volumes. EWs for D95%, maximum, and mean dose of CTV1 were 1.4%, 6.6%, and 1.5%, respectively. For maximum OAR doses, the required EWs were: small bowel (1.2Gy), large bowel (1.5Gy), heart (7.6Gy), and spinal cord (5.5Gy), whereas a 4.4% EW was observed for lung V20. Similarly, EW for OAR specific mean doses were: liver (1.2Gy), stomach (0.7Gy), kidney (0.1Gy), heart (1.5Gy) and esophagus (3Gy).

CONCLUSIONS: When patient anatomical changes are absent, coverage and OAR doses of breath- hold scans are reproducible to within the clinically acceptable margins reported in this study.

Poster Abstract 94: Target Margins versus Field Specific Margins for Proton Brain Plans

BACKGROUND: Eclipse defines the area in which proton spots can be place in two different ways. A target margin (TM) is a geometric expansion about a specified structure, whereas a field specific margin (FS) is a water equivalent expansion. FS is typically larger than TM, allowing potentially higher doses to critical structures.

METHOD AND MATERIALS: Five brain tumor plans were re-optimized using TM's and FS's. The distance specified for the margins was varied: 1, 1.5, and 2 times the spot size (sigma in air). The mean and maximum doses were evaluated for the clinical target volume (CTV) and a total of 74 critical structures. All percentage differences are compared to the TM×1 plan. Averages were calculated only using critical structures that were more than 1% different than the TM×1 critical structures. Robustness was calculated for all plans.

RESULTS: The mean and maximum CTV doses were all within 1%, except one plan (max=1.3%). All plans per patient have comparable robustness. The TM×1 plan best spared critical structures. For all critical structures, 19% had a lower maximum dose (average(maximum) difference=2.8%(7.4%)). Forty-three percent had a higher maximum dose (average(maximum) difference=5.6%(15.1%)). Fifteen percent had a lower mean dose (average(maximum) difference= 1.9%(3.4%)). Eighty-five percent had a higher mean dose (average(maximum) difference=3.4%(11.7%)). The TM plans outperformed the FS plans where the same factor was used for margin creation. Doses to critical structures increased with margin size with the exception of the FSx2, which did better than the FSx1.5 plans in 7 out of 11 parameters investigated.

CONCLUSIONS: TM plans result in lower doses to critical structures than FS plans. The TM×1 plan yielded the most critical structure sparing while maintaining CTV coverage and robustness.

Poster Abstract 95: Dose Verification of the Delivered Proton Pencil Beams at the McLaren Proton Therapy Center Using Treatment Log Files

BACKGROUND: The purpose of this work is to evaluate the discrepancy in dose delivery as monitored by primary ionization chamber (IC) and position accuracy of spot scanning proton beams.

METHODS AND MATERIALS: Proton plans developed by treatment planning system assume all spots will land at the correct positions and MUs delivered accurately. In reality, magnetic hysteresis and the predicated charge calculation model for IC introduces position and MU errors to a delivered plan. We utilized the treatment record logs generated by Treatment Delivery Control System (TDCS), which has been developed by Pyramid Technical Consultants. The treatment record logs the spot positions and MUs per spot for delivered plans at Isocenter. We developed an in-house Matlab code (MathWorks) which converts the log files to DICOM files. The DICOM files are imported into the Varian Eclipse planning system for plan comparison.

RESULTS: We evaluated the log files generated by TDCS for a daily QA plan as well as patients' treatment plans. The results showed that the delivered spot positions are within 1.0 mm from the planned position. The spots in distal layers with larger dose weights are accurately delivered to within 1% while the spots in proximal layers with smaller dose weight showed discrepancy up to 5%. Total MUs delivered as recorded by primary IC are within 0.5% of the planned total MU value. Dose variation within the treatment volume agreed to within 1.5%. It was noticed a large discrepancy on the field edges up to 5% due to field misalignment, as we have not, yet, incorporated setup uncertainties.

CONCLUSION: The methodology for evaluating the log file provided an excellent tool to evaluate the delivery system at McLaren Proton Therapy Center. We observed that scanned spot positions were reproducible to within 1mm and the delivered dose in agreement to within 1.5%.

Poster Abstract 102: Multi-Field Optimization Pencil Beam Scanning Proton Treatment Planning for High Risk Prostate Cancer

BACKGROUND: Multi-Field Optimization (MFO) for Pencil Beam Scanning (PBS) proton therapy provides dosimetric advantages over conventional radiation therapy for high risk prostate cancer, improving bladder and bowel sparing. However, MFO increases potential risks of setup uncertainties and therefore must be planned more robustly. We have developed a method to create more robust MFO PBS plans for high risk prostate cancer.

METHODS AND MATERIALS: Plans were developed in RayStation v6 with Monte Carlo (MC) optimization and dose calculation. Each plan used 2 lateral beams with a midline avoidance structure created by subtracting the CTVs from a union of the bladder and rectum. This avoidance prevents lateral beams from crossing midline and contributing dose to contralateral lymph nodes, creating a dose gradient in the nodes. To maintain a robust, homogenous dose in the prostate GTV, a beam-specific max dose constraint is placed on each beam to limit the gradients in the prostate. The plan is optimized robustly with independent beams, for a total of 75 perturbations, to smooth out the gradients in the nodal region. The plans were evaluated for target coverage and OAR sparing across 18 unique perturbations, including plan shifts and rotations, range uncertainty, and independent beam isocenter shifts.

RESULTS: Plans typically require several rounds of 100 iterations, each round requiring approximately 45 minutes. A final MC dose calculation is performed with 0.5% uncertainty. Each plan maintained appropriate clinical target coverage across all perturbations, with improved sparing compared to a single-field optimization proton plan and a VMAT plan.

CONCLUSIONS: To take full advantage of PBS proton therapy for high risk prostate cancer, MFO must be used, but this introduces uncertainty in plan robustness. With our method of optimization, we can mitigate this uncertainty and develop a more robust plan, while maintaining good bladder and bowel sparing.

Poster Abstract 104: Novel Robust Multicriteria Optmization Method for Intensity Modulated Proton Therapy

BACKGROUND: Intensity modulated proton therapy (IMPT) has the ability to provide treatment plans with highly comformal dose distributions, but it is sensitive to setup and range uncertainties. Incorporating those uncertainties in the treatment plan optimization and selection improves the plans robustness against such uncertainties. We provide a study of a novel method for incorporating robust objectives in multicriteria optimization (MCO) of IMPT.

METHODS: An adaptive MCO method developed previously by our group was extended to include a set of robust objectives in addition to the nominal objectives. The robust objectives represent the worst case from a set of predefined setup and range error scenarios. The method uses an adaptive weighting scheme that generates a well-distributed representation of the Pareto surface for a large number of objectives. The Pareto database was generated for each nominal objective in addition to the worst-case objective from the error scenarios. The method was tested on a brain tumor case with 8 organs at risk.

RESULTS: The robust MCO method was as able to generate Pareto database for 11 nominal and 11 robust objectives. Only 5 objectives showed large deviations (>5%) between the nominal and the robust objectives. The deviations ranged from 7.8% for CTV D98% up to 56.9% for left hippocampus D1%. The CTV D98% robustness showed an average improvement of 2.5% if CTVD98% coverage was reduced by 3.5%, or if the pituitary gland D1% increase in average by 12%. The CTV D98% robustness showed improvements by 2.5% in plans where the dose to the hippocampus, left optic nerve, and optic chiasm D1% decreased in average by 34, 5, and 7.9 % respectively.

CONCLUSION: We have developed a robust MCO method that was able to explore the trade-offs between IMPT plan objectives and the trade-offs between nominal plan quality and the plan robustness.

Poster Abstract 105: Gantry Angle Specific PSQA for a Commercial Proton Gantry – First Year of Clinical Experience

BACKGROUND: Patient-specific QA (PSQA) measurements for proton therapy are commonly performed using a single standardized angle, such as 270°, for all fields. Based on the long-standing experience at our centre to perform PSQA for our in-house developed gantries using the clinical gantry angle, we now developed for the first time a PSQA setup to perform gantry angle specific PSQA measurements for a commercial gantry (Varian ProBeam). We report implementation and first year of clinical experience with the new PSQA set-up.

METHOD AND MATERIALS: A rotatable and extendible water column (∅︀=27cm) was built in- house. Its base-plate was tailored to be mountable with a 90°-angle on the standard ProBeam robotic couch. The water column hosts a PTW Octavius 1500 XDR 2D array and a pT sensor. Water column, pT sensor and 2D array are remotely controlled via a dedicated LabView-based GUI, which integrates PTW BeamAdjust. PTW VeriSoft is used for analysis.

RESULTS: The set-up was extensively used during gantry commissioning and is routinely in use for PSQA. It allows field-specific QA with the clinically used gantry angle. The constancy field is stable within 0.7% (1σ). Rotational alignment shows a slight systematic angular dependency (max 1.5mm at 180°). More than 300 clinical fields were analysed resulting in a mean Gamma(3%,3mm) pass rate of 99.7% (all fields had pass rates>90%) and a mean ‘best match scaling factor’ of 0.44±1.42% (TPS Eclipse, PCS v13.7). Data analysis revealed only minor angular dependencies for our gantry (<1%).

CONCLUSIONS: PSQA set-up showed a solid performance during commissioning and first year of clinical operation. It provides gantry angle specific field verification for each patient. In the initial clinical phase, this helped to monitor and gain confidence in the gantry performance.

Poster Abstract 106: Efficiency and Safety Improvements for Proton Treatment Plan QA by Automation

BACKGROUND: Meticulous treatment planning QA is essential to deliver safe and high-quality proton treatments. Repetitive and extensive plan QA binds significant amounts of resources in clinical routine and is prone to human error. This work presents and evaluates tools developed at our clinic to extend, automatize and standardize the physicist's and clinician's plan review.

METHOD AND MATERIALS: Using the scripting capabilities of the commercial TPS Eclipse (Varian), two plugins for plan QA were developed. A Plan Checker plugin verifies parameters and properties of plans. The comprehensive checks cover the planning CT, the structure set, and the plan itself and generate more than 150 different colour-coded messages, which are classified according to their severity (INFO, PASSED, TO CHECK, WARNING, ERROR). Checks encompass amongst others: correct labelling (e.g. switching of left and right of organs), plan approval by authorized personnel, correct patient orientation, correct field tolerance tables, reasonable air gaps, HU overrides, plan and beam model parameters, the degree of field modulation, normalization, plan and field dose maxima and their positions, etc. A second tool evaluates clinical objectives for a given plan or plan sum and indicates by colour coding if they are met. It allows for easy template generation and can obtain data from the clinicians' prescription. Plan uncertainty can be additionally evaluated more efficiently.

RESULTS: Both tools extend and standardize the clinically performed plan checks while leading to a considerable time-reduction when performing the physicist's and clinician's plan review. At the same time, they lead to a reduction of repetitive, human error-prone work in clinical routine.

CONCLUSIONS: TPS plugins that automatize parts of the plan QA allow to create flexible plan checks and can be tailored to institutional needs and practices. They can contribute to increase safety, standardization and efficiency in proton therapy treatment planning.

Poster Abstract 107: Optimal Electrometer Selection for PBS Dosimetry

BACKGROUND: Most commercially available electrometers used in radiotherapy have been optimized for use in photon and electron beams. It is recommended the uncertainty associated to the electrometer in the overall dosimetry be less than 0.1%. Hence, when selecting the electrometers for use in absolute dosimetry of proton pencil beam scanning (PBS) systems, the exact specifications are not fully realized yet.

METHOD AND MATERIALS: The electrometers evaluated were IBA Dose 2, Keithley 35614, and Standard Imaging (SI) Max4000. The ion chamber used was a parallel plate ion chamber IBA PPC05 with bias set to 300V. We evaluated the setup under spread-out- Bragg-peak (SOBP) at shallow and deep depths for 10cm × 10cm field size. We verified the charge drift and repeatability in continuous mode for all electrometers under background condition after the required stabilization period and zeroing the electrometers.

RESULTS: The total charge collected was around 3.4nC for 2 Gy. Due to high current generated in the IC by the pencil beam, both Dose 2 and Max4000 were set to high current range and, for the Max 4000, the resolution is limited to two decimal figures. We found under the current configuration, the Dose 2 presented with the highest charge drift under background conditions of up to 0.022nC over 100 seconds, while Max 4000 had 0.01 nC, and Keithley of 0.001 nC. This translates to 0.64% of the total collected charge for the Dose 2, 0.29% for Max4000, and 0.03% for Keithley.

CONCLUSIONS: Accurate absolute pencil beam scanning dosimetry requires some electrometers to work under high current mode with minimal drift. Some electrometers may require modifications to allow accurate dosimetry in PBS systems.

Poster Abstract 108: The Effects of Scanning Protocol, Reconstruction Filter, Window and Level Selection on Identifying Surgical Implants Geometry in Proton Therapy Using A Dual Energy CT

BACKGROUND: Surgical implants (SI) in CT images are typically manually contoured and their materials and densities are overridden with known material in proton therapy. However, due to metal artifacts and beam hardening effects in CT images, there are significant uncertainties in deterring the geometry of the implants, leading to inaccurate dose calculation. The purpose of this study is to identify an appropriate scanning protocol on a dual-energy CT scanner for accurate contours of the implants.

METHOD AND MATERIALS: A titanium pelvic bolt (Φ8.5×90mm, SI1), pedicle screw (Φ5.5×35mm, SI2), lateral mass screw (Φ3.5×14mm, SI3), and a cobalt-chrome cortical screw (Φ5.0×30mm, SI4) were used. The screws were placed in a water phantom, an acrylic phantom, and a PVC pipe, respectively, to simulate different beam hardening scenarios. The images were reconstructed with various metal artifact reduction settings and 190keV synthetic monoenergetic reconstructions. Different window and level settings were investigated to find a robust setting accounting for beam hardening effect. Dose calculation of the phantom was conducted with both actual implants volume and contours drew using the selected protocol. Gamma analysis was conducted to test the dosimetric effect.

RESULTS: Compared to all different metal artifact reduction algorithms, a 190keV monoenergetic reconstruction using a threshold value of 50% the maximum HU for each implant provided contours closest to the true size of the metal in all scenarios. Mean diameter differences between the contoured and the actual implants are 0.00 ±0.16mm, −0.06 ±0.05mm, −0.26 ±0.27mm, and 0.00 ±0.07mm, for SI1, SI2, SI3, and SI4, respectively. The maximum differences are 0.20mm, −0.10mm, 0.60mm, and 0.10mm, respectively. The corresponding Gamma passing rates are 93.4%, 97.2%, 92.5%, and 96.9%, using a 2%/2mm threshold.

CONCLUSIONS: We have developed a universal scanning protocol that can detect and contour pelvic surgical implants with high geometric accuracy. Gamma analysis shows high agreement.

Poster Abstract 114: Efficiency and Accuracy Improvements for Patient Plan Quality Assurance with a Passive Scattering Proton Therapy System

BACKGROUND: All proton therapy plans for treatment at our institution using the Mevion S250 passive scattering therapy system are verified for correct monitor units by patient-specific dosimetric measurements. We describe software methods that are designed to streamline the workload, improve accuracy, and reduce the possibility of errors in the overall process of plan quality assurance and monitor unit determination.

METHOD AND MATERIALS: The RayStation treatment planning system (TPS) incorporates a complete implementation of the Python scripting language. The scripting interface allows for directly writing custom spreadsheets that are used in each of the patient QA measurements. The measurements are performed using the IBA MatrixX detector and “myQA” software. To aid in the accuracy of the measurements, a predictive starting monitor unit is extracted from the TPS. A large set (over 500) of measurements were made, covering all of the 24 beam options, to determine the correlation of the plan “meterset” with the actual, measured, clinically valid monitor unit.

RESULTS: The relationship between the treatment plan meterset and the clinically validated treatment monitor unit has been shown to be linear with R-factor's in excess of 0.999. The measured linear function is incorporated into the software that generates the patient QA spreadsheet, to give a starting point for the QA measurement that will be close to the final value, improving the accuracy of the measurement. A limitation of the commercial software is that regions-of-interest (ROI's) used in analysis can only be rectangular in shape. To overcome this limitation, and provide additional analysis information, a software tool was written that can compare the measured and TPS generated dose distributions with an arbitrary, user defined, ROI.

CONCLUSIONS: Using custom produced software can improve both the efficiency and accuracy of patient-specific quality assurance of double-scattering (aperture and compensator) treatment plans.

Poster Abstract 116: International Dosimetry Inter-Comparison Between Auditing Institutions in Proton Beam Therapy: MedAustron and IROC

BACKGROUND: MedAustron (MA, Austria) in collaboration with the National Physical Laboratory (NPL, UK) has developed a novel dosimetry audit based on end-to-end (E2E) testing with alanine dosimeters and a Farmer chamber for proton beam therapy. The aim of this work is to compare this novel dosimetry audit procedure with the well-establish audit procedure developed by IROC (Houston, USA).

METHOD AND MATERIALS: The measurements at first were performed in water at the fixed horizontal proton beam line at MA. A single-energetic layer square field and a SOBP were delivered to a water phantom. A Farmer chamber and 9 alanine pellets (MA- NPL) were placed at the same reference depths in water as the TLD detectors (IROC). A physical dose of 6.0 Gy was delivered to all the detectors.

As the second step, comparison E2E testing measurements were performed with the anthropomorphic MA head phantom loaded with alanine detectors and with the IROC brain proton phantom loaded with TLDs. A physical dose of 5.5 Gy was delivered with plan composed by two opposing oblique beams.

RESULTS: The average dose of the 9 alanine pellets at each depth and the Farmer results agreed within 1.5% in water for all the measured depths. An agreement within 0.5% was observed between TLDs and alanine in the proximal part and in the center of SOBP, but in the plateau and in the distal part of the SOBP slightly larger deviations were observed. The E2E tests for anthropomorphic phantoms performed at the same proton beam line with both audit techniques showed average deviations from the TPS planned dose within 2%. The differences in results were within the uncertainties of the used dosimetric techniques.

CONCLUSIONS: The MA-NPL dosimetry audit system was successfully compared to IROC confirming a potential application for future proton clinical trials within Europe.

Poster Abstract 120: A Novel Design of Proton Computed Tomography Detected by Multiple-Layer Ionization Chamber with Strip Chambers: A Feasibility Study with Monte Carlo Simulation

BACKGROUND: Uncertainty in proton range can be reduced by proton computed tomography (CT). A novel design of proton CT using a multiple-layer ionization chamber with two strip ionization chambers on the surface is proposed to simplify the imaging acquisition and reconstruction.

METHOD AND MATERIALS: Two strip ionization chambers facing the proton source were coupled into a multi-layer ionization chamber (MLIC). The strip chambers measured the location and lateral profile of each incident proton beamlet after exiting the imaging object, while the integral of depth dose was recorded in the MLIC. Simulation was performed at 5 levels of imaging dose to demonstrate the feasibility and performance expectation of our design. The energy of the proton beamlet was set to 150 ± 0.6 MeV. A collimator with a round slit of 1 mm in diameter was placed in the central beam axis upstream from steering magnets. Proton stopping power ratio (SPR) was reconstructed through inverse radon transform. The imaging phantom was a water-based cylinder with 13 tissue-equivalent inserts according to ICRP 1975. The phantom was 10 cm in diameter and all inserts were 1 cm in diameter. Percentage discrepancies were reported by comparing to the ground truths. The imaging dose and quality were also evaluated.

RESULTS: Deviation of reconstructed SPR of all inserts were within 1% from the ground truths. Imaging dose was 0.54 cGy if 5×102 protons were used in each beamlet. Imaging quality was acceptable for planning purpose and held consistently through all levels of imaging dose. Spatial resolution was consistently measured as 5 line pairs per cm in all simulations varying in imaging dose.

CONCLUSIONS: The proposed design of proton CT simplifies data acquisition while achieving excellent accuracy in proton SPR and acceptable spatial resolution, making it potentially a great tool for localization and plan adaption in proton therapy.

Poster Abstract 125: Improving IMPT Delivery Efficiency by Increasing Minimum Spot MU in Plan Optimization

BACKGROUND: Intensity modulated proton therapy (IMPT) beam delivery time is proportional to target size and can take several minutes for breast treatments. This work studies the feasibility of improving beam delivery efficiency for breast IMPT treatments by increasing the minimum spot MU during plan optimization without compromising plan quality.

METHODS AND MATERIALS: Thirty IMPT plans for 5 bi-lateral breast with comprehensive lymph node patients were generated using the Monte-Carlo dose calculation engine in RayStation TPS (RaySearch Laboratories, Sweden). For each patient, six plans with minimum spot MU set to 1–2 and 8 were created and sensitivity of plan quality to minimum spot MU was studied. Plans used 2–3 field configurations, used the same organ-at-risk (OAR) constraints, and were optimized to 5040 cGy (180 cGy × 28) covering 95% of the clinical target volume (CTV). Each plan was subsequently delivered using a clinical IMPTC system, and the beam-on time was recorded and compared.

RESULTS: The CTVs were 1323 milliliter for patient 1 and 2184 milliliter for patient 2. CTV D1 were 105.10±0.05% for patient 1 and 104.37±0.03% for patient 2; and there was no difference in the OAR doses among the 5 plans for each patient. The delivery time ranges from 1.48 minutes to 5.38 minutes per beam. Plans with higher minimum spot MU delivered within less time; compared with plan 1 (minimum MU was 1MU), the beam-on time reduced 31.9–34.7%, 43.8–47.4%, 46.6–49.5%, and 48.9–51.9% when the minimum MU were increased to 2MU, 3MU, 4MU, and 5 MU, respectively.

CONCLUSION: Increasing the minimum spot MU reduces the beam delivery time without compromising plan quality. This method can improve IMPT treatment delivery efficiency, decrease the waiting time for multi-room proton treatment facility, and increase patient throughput.

Poster Abstract 126: Proton Radiation Beam Matching and Patient Transfer Workflow Among Rooms in a Multi-Vendor Software Environment: Conveniences, Challenges, and Potential Solutions

BACKGROUND: Proton radiation beam matching option is available for a multi-room proton therapy center. The purpose of this work is twofold: First, as the first proton center to become clinical with a unique combination of RayStation (v6.1.1.2), ARIA (v13.7), and adaPT-IBA-system (v11), we highlight the challenges and potential solutions for patient transfer workflow among different beam matched rooms in a multi-vendor software environment. Second, we present the comprehensive dosimetric results of proton beam matching for an IBA ProteusPLUS PBS proton system.

METHOD AND MATERIALS: The measured proton beam matching parameters for each treatment room include: in-air spot size, absolute dose output, range, and beam coincidence (X-ray vs. proton beam). PBS proton and imaging (kV-planar and CBCT) machines are configured in RayStation, ARIA, and adaPT to facilitate patient transfer from one room to another.

RESULTS: Although three gantries are beam matched dosimetrically, several challenges and potential solutions related to patient transfer among rooms were identified within ARIA and IBA system. Our current beam matching measurements for spot size, range, and beam coincidence include energies from 70 MeV to 225 MeV at an increment of 2.5 MeV. The spot size and range measurements among rooms were found to be within ±5%/±0.25mm and ±1 mm, respectively, of each other. The beam coincidence was within ±1 mm. Absolute dose output was measured for energies from 70 MeV to 225 MeV at an increment of 5 MeV, and the results were within ±1% with exception at lower energies.

CONCLUSION: Beam matching provides the convenience of treating the same patient in any given room. However, patient transfer among rooms using ARIA and adaPT is non-trivial with several in-house solutions including changes in machine configurations.

Poster Abstract 127: Secondary Neutrons in Spatially Fractionated Radiotherapy with Protons Using Collimators

BACKGROUND: New technique is trending in spatially fractionated radiotherapy, SFRT, with protons: to utilize the spot scanning ability together with a physical collimator to obtain minibeams. The aim of this project is to address the concern of secondary neutron dose formed when a dense collimator is put directly in the beamline.

METHOD AND MATERIALS: A 3.25 cm thick brass collimator with 3 mm diameter holes and 6 mm center-to-center spacing with hexagonal arrangement was used to obtain minibeams via the open holes of the collimator. The operational quantity ambient dose equivalent, H*(10), per treatment proton dose, D, was assessed using Monte Carlo code TOPAS and measured using WENDI-II detector at different angles (135, 180, 225 and 270 degrees) and distances (11 cm, 58 and 105 cm) from the phantom for two cases: with and without physical collimation. Assessments were performed for a uniform iso-energy (120 MeV) layer of a field size of 9×9 cm2 at isocenter.

RESULTS: Without collimation H*(10)/D varied from 0.0013 mSv/Gy (at θ=135 degrees and 105 cm distance) to 0.242 mSv/Gy (at θ=0 degrees and 11 cm distance). With collimation H*(10)/D varied from 0.017 (at θ=0 degrees and 105 cm distance) to 3.23 mSv/Gy (at θ=135 degrees and 11 cm distance).

CONCLUSIONS: Without the collimation, most of the neutrons are created at the distal end of the phantom owing to the forward directedness of the secondary neutrons. With the collimation, most of the secondary neutrons are created in the periphery of the collimator. Results suggest that the secondary neutron dose will increase tenfold when the physical collimator is used. Regardless, even in the worst-case scenario, negligible neutron dose is created as compared to uniform scanning.

Poster Abstract 130: Fast Proton Beam Fluence and Position Detector Array with Multi-Coordinate Readout

Using Pencil Beam Scanning (PBS), treatment plans are fully described by a set of beam spot parameters such as energy, time duration, position, angle, etc. Even as PBS has become the new standard in proton therapy, most quality assurance instrumentation has not been designed to complement PBS. A detector capable of measuring beam parameters spot by spot in real time would enable richer diagnostics and further restriction on proximal margins.

A novel planar detector array, which can measure position and fluence at a frame rate of up to 25 kHz has been conceptualized, built, and is being characterized. Capable of sub-millimeter spatial resolution in beam positioning and excellent radiation hardness, the new device overcomes factors that often limit the overall performance of planar detector arrays with a conventional pixelated arrangement of sensors.

Gas ionization is collected by a planar arrangement of strips projected along three angles, from which the beam size and position are reconstructed for each frame. Such proposed multi-directional readout provides a large, isotropic and continuous active area, while using fewer data channels (vs. pixel-based arrays). Combined with a new approach to tomographic reconstruction, the presented preliminary experimental results support the feasibility of precise registration of the proton beam size and position, while providing additional avenues for enhancing machine and patient level quality assurance through its superior timing capabilities.

Poster Abstract 131A: A Golden Beam Data Commissioning Framework of Monte Carlo Dose Calculation Algorithms of Two Pencil Beam Scanning Treatment Planning Systems

PURPOSE: Golden beam data (GBD) can potentially shorten commissioning time and minimize errors. However, there is no guidance on how to commission Monte Carlo Dose Calculation (MCDC) algorithms for Pencil Beam Scanning (PBS) using GBD.

METHODS AND MATERIALS: A GBD framework was developed to commission MCDC algorithms in two treatment planning systems using water and heterogeneous phantoms. Measurements included Bragg peaks and profiles of PBS single-spots and PBS field outputs for Varian ProBeam. The phase parameters, numbers of protons per MU were obtained from in-air measurement and PBS outputs of 100 cm2 square fields at 2 cm depth, while spot profiles and more PBS fields at more depths were used to validate TPS.

RESULTS: The maximum differences of phase parameters spot sigma and divergence between MCDC algorithms are below 4 μm and 0.26 mrad in air, respectively. Comparing TPS to measurement at depths, both MC algorithms predict the spot sigma within 0.5 mm, the resolution of measurement device. One MCDC algorithm is found to underestimate numbers of protons per MU by ~2% and requires user adjustment to match measurement, while another algorithm is within 1% of measurement using Auto model. Site-specific gamma criteria vary from 2 mm/2% to 5 mm/5% with comparing TPS to exit dose measurement.

CONCLUSION: The proposed GBD framework can detect potential issues during PBS beam data collection or TPS commissioning processes. Therefore, it can shorten commissioning time and improve dosimetric accuracies. Secondary MCDC can be used to identify the root sources of disagreement between primary MCDC and measurement.

Poster Abstract 131B: Evaluation of Accumulated Dose Distribution Utilizing Log Data on Spot Scanning Proton Therapy for Liver Tumors

BACKGROUND: In spot scanning proton therapy (SSPT), spot positions relative to a target can fluctuate by tumor motion even in the case of gating irradiation utilizing a fiducial marker. We have developed an evaluation system of delivered dose distribution on the basis of log data about tumor motion and spot information. The purpose of this study is to validate reliability of dose distributions for liver tumors treated with real-time image guided SSPT by evaluation of accumulated dose distributions.

METHODS AND MATERIALS: The evaluation system synchronizes the delivered spot information and the marker position on the basis of log data about the timing of spot irradiation and fluoroscopic X-ray irradiation. Then a treatment planning system reconstructs the actual delivered dose distribution utilizing the synchronized record data. Dose distributions accumulated for all fractions were reconstructed for 8 liver cases with treatment plans made by the single-field uniform optimization. The log data were acquired in all 168 fractions for the 8 cases. The evaluation was performed for the values of maximum dose, minimum dose, D99 and D5–D95 for the clinical target volumes (CTVs) and mean liver dose (MLD) scaled by prescribed dose. Statistical t-test was performed to compare dosimetric parameters between planed dose distribution and reconstructed actual dose distribution.

RESULTS: The reconstructed dose distribution showed the dose distortion was suppressed by gated irradiation. As for minimum dose and D99 of the CTVs, there were no significant differences between planed and actual dose distributions. Maximum dose and D5–D95 of the CTVs significantly increased. The variation of average values were +1.3% and +0.5% for maximum dose and D5–D95, respectively. Regarding MLD, the actual dose distribution well reproduced the planned dose distribution.

CONCLUSIONS: Reliability of dose distributions for liver tumors treated with real-time image guided SSPT was validated by evaluation of accumulated dose distributions.

Poster Abstract 132: Disagreement of Measured Small-Field Output with Treatment Planning System for a Probeam System

PURPOSE: For pencil-beam scanning protons systems, in-air non-Gaussian halo can significantly impact output at small field sizes and low energies. Since the halo is typically not modelled in treatment planning systems (TPS), this can potentially lead to significant differences in planned and delivered treatment. Here, we report the magnitude of such disagreements.

MATERIALS AND METHODS: A small-volume ion chamber was used to measure absolute output from a ProBeam nozzle in water, and the results were validated with a diamond detector. Field sizes from 2–20 cm were employed with energies ranging from 70–240 MeV. Measurements were taken at the water surface and at half-range for each proton energy. A clinical Monte Carlo algorithm was used for output modeling.

RESULTS: The extent of the halo is featured in Fig. 1 (not shown - plots at Z300 show the spot size just after the protons leave the snout, and Z0 shows the spot size 300 mm downstream). The output measurements exhibited a 4–5% disagreement with the TPS for the 2 cm field with 100 MeV, and an 8% disagreement in output for the 2 cm field with 70 MeV.

CONCLUSIONS: We found that the clinical TPS overestimated output by as much as 8% for small field sizes of 2 cm at extremely low energy of 70 MeV. The in-air halo of low energy extension to 2–3 cm diameter may potentially lead to underdosage of patients treated with small fields.

Poster Abstract 133: Pseudo Proton Radiography Beam Validation of Monte Carlo Dose Calculation Algorithms

PURPOSE: To validate Monte Carlo Dose Calculation (MCDC) in heterogeneous media, one must ensure accurate calculation of proton range, absolute dose, and modeling of scattering properties. We investigated the feasibility using pseudo proton radiography beams through an anthropomorphic phantom to validate the newly released MCDC in two treatment planning systems (TPSs).

MATERIALS AND METHODS: Stoichiometric calibration of CT Hounsfield Unit versus mass density was used in TPSs. Anterior pencil beam scanning beams of 197, 200, and 210 MeV were used to penetrate the pelvis of a male anthropomorphic phantom. Two-dimensional ionization chamber array was positioned directly under the treatment couch to detect exit dose. Measurements were compared to three-dimensional dose cubes calculated with MCDC from both TPSs using site-specific gamma criteria.

RESULTS: Based on the best three-dimensional match of dose distributions, range disagreements for the anterior-posterior travel through the pelvis torso of ~20 cm water equivalent thickness were within 4 mm between TPS and measurements. However, there was a potential underestimate of the proton range up to 4 mm when sacral bones were dominant, i.e., ~2% level for this anthropomorphic phantom. For the 210 MeV beam in uniform regions, one TPS calculated dose was ~2% lower than that of another TPS. To achieve 90% pass rate, gamma criteria range from 2 mm/2% in brain to 4 mm/4% in lung and 5 mm/5% in pelvis.

CONCLUSION: Pseudo proton radiography beam measurements are a feasible technique to determine site-specific MCDC accuracy in heterogeneous media.

Poster Abstract 136: Use of Optical Coherence Tomography (OCT) As Routine Base Line Examination in Meningioma Patients before Proton Beam Radiation

INTRODUCTION: The assessment of visual field (VF) measured via automatic perimetry (AP) is a standard examination. Optical coherence tomography (OCT) is a non-invasive, non-contact and painless imaging technique that provides high-resolution measurements and cross-sectional imaging of the retina and retinal nerve fibre layer (RNFL). The RNFL thickness is of particular interest in clinically manifest as well as subclinical optic neuropathies.

MATERIAL AND METHODS: Visual parameters including VF and RNFL thickness were measured before start of radiation. VF was measured by AP, RNFL via OCT. The examination was performed prior to treatment planning for proton therapy. Additionally, the involvement of the anterior visual pathway (optic nerve, chiasma) was defined on the planning MRI.

RESULTS: Twenty-four patients with no ophthalmologic comorbidities were included. The mean age at time of radiation was 55.4 a (+/−12.8 a). At baseline a restriction of the VF was detected via AP in 12 patients on the left and in 7 patients on the right. Via OCT in 13 patients a deficiency was detected on the right side and in 9 patients on the left. On MRI the right optic nerve was in direct contact to the meningioma in 13 patients, the left optic nerve in 16 and the chiasma in 11 patients, respectively.

CONCLUSION: In this cohort the detection of the anterior visual pathway disorders was higher with OCT compared to AP. OCT provides additional base line information which is beneficial for treatment planning, follow-up and as endpoint in future clinical trials.

Poster Abstract 137: IMPT with Simultaneous-Integrated Boost Substantially Reduces Integral Dose in the Treatment of Low-Grade Glioma in Comparison to VMAT

BACKGROUND: Intensity-modulated radiation therapy in combination with simultaneous integrated boost (SIB) reveals advantages in the sparing of normal tissues. Aim of the present study was to identify relevant differences with a focus on integral dose between intensity modulated proton therapy (IMPT) and volumetric modulated photon arc therapy (VMAT).

METHOD AND MATERIALS: Three grade-II and four grade-III glioma patients (WHO-classification) were treated at MedAustron with protons (IMPT). IMPT plans were optimized using single-field integrated boosts (SFIB). All IMPT-plans were clinically applied and subsequently compared to VMAT-plans that were generated using identical clinical goals for targets and OARs. A median number of 4 fields (range: 2 to 4 fields) arranged in the coronal plane was used for IMPT plans. VMAT plans were generated with a median number of 2 coplanar 360°-arcs (range: 1 to 2 arcs) arranged in the axial plane. CTV1 included GTV with isotropic margin of 1.0 - 1.5 cm adapted to anatomic barriers. CTV2 (=GTV) was defined as boost volume. Respective isotropic PTV- margins were 0.3 cm. 50.4/54.0 Gy in 28 fractions to PTV1/2 were prescribed to WHO grade-II glioma, 54.0/60.0 Gy in 30 fractions to PTV1/2 were prescribed to WHO grade-III glioma. For protons a constant RBE of 1.1 was considered. To assess integral dose a volume called total normal-tissue volume brain (TNTVb = brain - PTVs) was generated.

RESULTS: Median target volumes were 313 cm3 (173–479 cm3) and 154 cm3 (42–192 cm3) for PTV1 and PTV2, respectively. Dosimetric parameters for PTVs were evaluated in terms of D98%, D50% (= Dprescr.), D2%, HI and CI. Following the prescription, similar values were obtained for IMPT and VMAT plans.

CONCLUSION: VMAT as well as IMPT resulted in clinically acceptable plans by meeting the prescribed dosimetric goals. Protons bear substantial advantages for CNS in the low-dose region. Neurocognitive testing is part of prospective follow-up and a correlation of proton dose to long-term clinical outcome is planned.

Poster Abstract 138: IROC Audits of Proton Therapy Centers Around the Globe

BACKGROUND: IROC performs quality assurance audits for proton therapy centers. The audits range from simple output checks to complex end-to-end tests to comprehensive on-site audits.

METHODS AND MATERIALS: IROC uses TLD for remote output checks. Anthropomorphic phantoms have been designed for six disease-site-specific audits (brain, H&N, lung, liver, prostate, and spine). On-site audits cover machine performance, treatment planning practices, and quality assurance procedures.

RESULTS: IROC Houston has audited 44 proton facilities in eleven countries. There have been over 650 TLD beam output checks performed. The mean TLD/institution ratio is 0.997+/−0.020. There have been 253 phantom irradiations analyzed and the overall pass rate is 74%. Phantom failure rate correlates with number of targets and motion. Another phantom trend observed is that the treatment planning systems overestimate dose to the target. For the lung phantom, Monte Carlo shows an improvement over pencil beam algorithms in the absolute and relative dose agreement, particularly in the proton beam range direction. IROC Houston has performed 36 on-site audits of 28 proton therapy centers. Several facilities with multiple delivery modalities have received two on-site audits. The mean number of recommendations that an institution receives is 4. Most of the recommendations from the on-site audits are for QA practices or for the HU-RLSP conversion curve. 22 of the 28 proton centers that received on-site audits have gone on to enroll patients in NCI-funded clinical trials.

CONCLUSIONS: IROC Houston has developed a robust audit program for proton therapy. While beam calibration is very tight across institutions, there are still areas for improvement in clinical practice, such as the dose modelling for beam delivery in the lung and accuracy of HU-RLSP conversion. Proton institutions around the world can benefit from the in-depth peer review that IROC Houston's QA program provides.

Poster Abstract 139: Automated Stoichiometric Calibration Using Scripting

BACKGROUND: The stoichiometric CT calibration process is sensitive to many parameter values and human inputs which could be prone to error. The purpose of this work to provide baseline script solution for new institutions to use as a second check for calibrations.

MATERIALS AND METHODS: A generalized stoichiometric calibration script was developed using the .NET framework and the libraries openCV and Evil DICOM. A GAMMEX 467 phantom was scanned with various tissue equivalent plugs on a Phillips Brilliance CT scanner and the DICOM files were exported. The DICOM files were read and a 3D matrix for computer vision processing was created from the voxels. A plug finding algorithm was created to find all of the inserts, locate their center coordinates and sample an average HU on the center slice. The materials were determined using a descriptor file containing the polar coordinates of the inserts for the scan. The relative stopping power and relative electron densities as a function of HU for real tissues were reported.

RESULTS: Given only a directory containing DICOM files and the plug coordinate descriptor file, the software was able to correctly detect and identify all 16 plugs. A PDF report was generated containing mean HU of each plug, mean error of the measurement, graphs and tables of the RSP and RED for the materials in 3.01 seconds after user interaction.

CONCLUSION: We have demonstrated a scripting solution for stoichiometric calibrations. It is hoped the speed and ease provided can improve new centers in validating their datasets. The framework is generalizable to any phantom by use of a phantom descriptor file. Future work includes deploying the code as a freely available web service.

Poster Abstract 140: The Impact of Dual Energy CT Scans on TPS Calculated Dose Accuracies for Biological and Non-Biological Samples

BACKGROUND: Current clinical practice accounts for proton range uncertainties by additional clinical target volume margins. A main part of these uncertainties stem from the use of clinical HU-to-SPR calibration curves that are derived from single energy CT scans (SECT). We investigate the clinical impact of using dual-energy CT scans (DECT) on TPS calculated dose accuracies for biological and non-biological samples.

METHOD AND MATERIALS: SECT and sequential DECT scans for four tissues consisting of cow-liver, cow-brain, pork-liver, and pork-foot were acquired using a SIEMENS SOMATOM Definition Edge CT Scanner (Siemens, Germany). Similarly, DECT and SECT scans for six non-biological tissue surrogate plugs from the CIRS phantom (Norfolk, VA) were acquired. SPR images were constructed from the DECT scans using the Syngo.via software generated ρe and Zeff, values. Calculated SPR values were verified by water equivalent thickness (WET) measurements with a multi-layer ionization chamber. For each sample, the same plan was optimized and recalculated on SECT and DECT scans using the clinical Monte-Carlo dose engine in the RayStation TPS. Each plan was delivered and planar dose 2D gamma analysis was performed for TPS calculated dose distributions on SECT and DECT scans.

RESULTS: For the pork-foot, SPR measurement was within 1% agreement with DECT calculated values in comparison to 7% difference for SECT. However, no significant difference was observed for other biological tissue. For non-biological tissue samples, DECT calculated SPRs were within 3% agreement of the measurement, while SECT computed values deviated by as much as 15%. Similarly, higher gamma passing rates were consistently obtained when using DECT SPR images.

CONCLUSIONS: These early results indicate that utilization of DECT in proton therapy can lead to significant reduction of range uncertainties when encountering non-biological tissues such as breast implants, chemo-ports, while maintaining the current standard of dose calculation accuracy for biological tissues.