A Comparison of Three Planning Techniques for Bilateral Reconstructed Chest Wall Patients Undergoing Deep Inspiration Breath Hold (DIBH) Breast Irradiation

E616

International Journal of Radiation Oncology  Biology  Physics

non-uniform plans, demonstrating a standard deviation of dose within target of 2.20.9% (VMAT uniform; P<0.001) compared to 5.11.4% (VMAT non-uniform) and 19.62.1% (GK). The normalizations for uniform, non-uniform and GK plans were 90.52.5%, 79.93.4% and 48.72.2%, respectively. The treatment efficiency was superior for VMAT (3.01.0 min, uniform; 3.41.0 min, non-uniform) compared to GKE (55.415.7min; p<0.001). Conclusion: With the same planning targets and critical structures, LINAC VMAT plans can achieve comparable conformity as GK plans, and better target uniformity and superior treatment efficiency. GK plans have high dose gradient outside target, hence provide better protection for the normal tissue outside the prescription volume. Author Disclosure: J. Phan: None. J.N. Yang: None. A.J. Ghia: None. P.D. Brown: None. A.S. Garden: None. D.I. Rosenthal: None. C. Wang: None. S. Tung: None. D. Luo: None. H. Wang: None.

Author Disclosure: C. Valentini: None. L. Boldrini: None. G. Mattiucci: None. G. Mantini: None. G. Chiloiro: None. D. Pasini: None. S. Bral: None. J.V. Turian: None. J.C. Chu: None. V. Valentini: None.

3544 Multicenter Autocontouring Effectiveness Evaluation in Lung Cancer C. Valentini,1 L. Boldrini,1 G.C. Mattiucci,1 G. Mantini,1 G. Chiloiro,1 D. Pasini,1 S. Bral,2 J.V. Turian,3 J.C.H. Chu,3 and V. Valentini1; 1Sacred Heart Catholic University of Rome, Rome, Italy, 2OLV Ziekenhuis Aalst, Brussel, Belgium, 3Rush University Medical Center, Chicago, IL Purpose/Objective(s): This hypothesis-generating prospective multicentric study has the aim to validate in clinical setting the reliability of a pre-release commercial algorithm for lung target volumes automatic segmentation for Stereotactic Body Radiation Therapy (SBRT) and 4D Radiation Therapy (4DRT). Materials/Methods: Three centers participated in this study in which the same 20 patients with histologically proven lung lesions diagnosed between 2010 and 2014 (primary and secondary) and treated with SBRT or 3D conformal Radiation therapy (3DCRT) were retrospectively collected.13 patients with primary lung cancers or lung metastases (9 patients with stage I lung cancer, 3 patients with lung metastases from rectal adenocarcinoma). -7 patients with locally-advanced-lung-cancer stage II-III were also selected to test the software with larger volumes.-Nodal Volumes were not taken into consideration. Two Radiation Oncologists (RO) experts in lung cancers from each center agreed the delineation of the lung lesions. Two different RO (per center) performed the Automatic Delineation (AD) of the same lesions using the beta version of a commercially available contouring system. The 2 expert RO edited the automatic contour after a month, drawing Edited Delineation (ED).The obtained contours (AD and ED) were compared to MC. Pooled comparisons were made in terms of geometrical overlap, by analyzing: Mean Dice-Similarity-Index (mDSI), Mean-Slicewise-Hausdorff-Distances (MSHD) and Volume difference (cc). Results: The pooled-analysis of the contouring settings showed: MC vs AD: mean mDSIZ0.76 (1SD0.11; 95% CI: 0.73-0.79); a MSHDZ1.16 mm (1SD2.29; 95% CI: 0.57-1.76) and a mean volume differenceZ7.44cc (1SD12.9; 95% CI: 2.83-9.53). MC vs ED mean mDSIZ0.82 (1SD0.08; 95% CI: 0.79-0.84); a MSHDZ0.53mm (1SD0.76; 95%CI: 0.33-0.73) and a mean volume differenceZ6.18cc (1SD12.9; 95%CI: 2.83-9.53). When comparing the two groups a statistically significant greater concordance between MC vs ED than MC vs AD for DSI (pZ0.0016) and for MSHD (p<0.027) was observed, but not for volume difference. Lung basis and hilum were the sites where the majority of discrepancies among contours were observed, probably due to structures similar density. Conclusion: The tested autosegmentation software seems promising in the automatic delineation of lung lesions both for SBRT or 4DRT purposes and for 3DCRT in more advanced stages, offering an acceptable overlap with expert drawn contours, although human expert revision is always warranted. Further investigations in a multicentric setting (taking into account also the contouring time factor) are ongoing on a larger number of patients to define a benchmark for early stage tumor/metastases and for locally advanced stages.

3545 Is the Liver a Static Organ: Intraorgan Deformation Quantified Through Fiducial Displacement on 4DCT A.K. Paulsson, S.S. Yom, M. Anwar, D. Pinnaduwage, A. Sudhyadhom, A.R. Gottschalk, A.J. Chang, and M. Descovich; University of California, San Francisco, San Francisco, CA Purpose/Objective(s): Stereotactic body radiation treatment (SBRT) is a well-tolerated modality for treatment of primary and metastatic liver lesions. To account for respiration-induced tumor motion, fiducials are used to dynamically track the lesion during treatment. Due to potential morbidity associated with implantation in the tumor, fiducials are often not placed inside the target. Fiducial placement outside the lesion is performed assuming the liver moves as a unit and the relative position of the tumor to the fiducial remains constant. Our purpose was to test this assumption by quantifying intra-organ deformation in the liver. Deformation was categorized as a function of liver disease (healthy vs. cirrhotic) and lesion location (central vs. peripheral). Materials/Methods: Nine patients treated with SBRT between 2011 and 2015 were included in the study. All patients had multiple fiducials (2-4) to guide tracking and a 4DCT to evaluate target motion. Point contours were placed at the center of each fiducial as a reference from which interfiducial distance was measured. Data points were collected at the 0%, 50% inspiration and 100% inspiration points. The distance between each pair of fiducials was measured at each phase and used to calculate deformation in the fiducial configuration during respiration (rigid body error). Extent of liver disease was calculated using Child-Pugh score using laboratory values within 3 months of initiation of treatment. A peripheral lesion was defined as one within 1 cm of the liver edge. Results: Six patients had primary liver disease and 3 had metastatic disease. 2 patients had Child-Pugh Class C liver disease, 6 had Class A liver disease, and one had insufficient data for calculation (metastatic breast cancer). 5 patients had centrally located fiducials, 2 had peripheral fiducials, and 2 had lesions that necessitated peripheral and central fiducials. For 3 patients with only centrally located fiducials, the fiducial configuration moved together (rigid body error < 1.9 mm). Deformations in configuration were observed in the other 6 patients (range 2.2-11.3 mm, median 2.8 mm). The largest excursions were in a patient with both a central and a peripheral fiducial and in a patient with 4 peripheral fiducials. We did not find an association between deformation and degree of liver disease, although only 2 patients had class C Child-Pugh disease. Conclusion: The liver frequently experiences internal deformation and should not be assumed to act as a static organ. While it is often assumed that cirrhotic livers are more rigid, we could not establish there is less deformation. We observed greater deformation at the periphery than for central lesions, which may have important clinical implications. This finding holds particular relevance for lesions at the inferior liver edge, which often abut bowel. Unanticipated deformation may place the bowel in a high dose region. Author Disclosure: A.K. Paulsson: None. S.S. Yom: None. M. Anwar: None. D. Pinnaduwage: None. A. Sudhyadhom: None. A.R. Gottschalk: None. A.J. Chang: None. M. Descovich: None.

3546 A Comparison of Three Planning Techniques for Bilateral Reconstructed Chest Wall Patients Undergoing Deep Inspiration Breath Hold (DIBH) Breast Irradiation T. Volpe,1 R. Margiasso,2 Z. Saleh,3 L. Kuo,4 L. Hong,5 A. Ballangrud,3 D. Gelblum,5 M. Remis,3 J.O. Deasy,5 and X. Tang6; 1Memorial Sloan Kettering West Harrison, West Harrison, NY, 2Memorial Sloan Kettering Cancer Center, West Harrison, NY, 3Memorial Sloan Kettering Cancer Center, New York, NY, 4MSKCC, New York, NY, 5Memorial Sloan Kettering Cancer Center, New York, NY, 6Memorial Sloan Kettering, West Harrison, NY

Volume 93  Number 3S  Supplement 2015

Poster Viewing Session E617

Poster Viewing Abstracts 3546; Table 1 Patient 1

Patient 2

Patient 3

Compared parameters

Tangent

VMAT

IMRT

Tangent

VMAT

IMRT

Tangent

VMAT

IMRT

Heart mean (cGy) LT Lung V20 Gy (%) RT Lung V20 Gy (%) RT breast mean (cGy) Cord max (cGy) MU

293.5 38.9 0.05 120.6 112.2 251

767.5 26 2.2 583.3 642.3 604

1063.8 18.1 1.6 650.9 912.6 1790

2488.2 78 0.58 238.5 4301.5 265

645 27.4 5.7 752.6 3941.4 1042

812 27.5 7.5 401.2 4061.2 1819

1902.3 61.7 0.5 372.8 2869.5 244

664.2 29.8 4.3 725.3 2541 913

1079.4 31.7 4.3 798.6 2780.5 2536

Purpose/Objective(s): Deep Inspiration Breath Hold (DIBH) is utilized to reduce cardiac exposure for left-sided breast cancer treatment. As we continuously see more bilateral chest wall cases, new challenges are being presented to administer DIBH. We herein compare three different planning techniques (tangents, VMAT, and IMRT) for such cases. Materials/Methods: Three left-sided chest wall DIBH patients with bilateral implants were studied. Two of the three patients had nodal involvement. For each patient, tangents, VMAT, and IMRT plans were created in a Treatment Planning System. All three patients had the same prescription (200 cGy x 25 fractions). For the purpose of comparison, all plans were normalized so that 95% of the prescription dose is delivered to 95% of the planning target volume (PTV). The maximum point dose was constrained to less than 120% of the prescription dose. We do not systematically employ strict dose/volume constraints for the normal tissue. Rather, we try to balance the competing goals of target coverage and normal tissue sparing. Since the success of DIBH delivery largely depends on patient’s ability to perform consistent breath hold during beam on time, smaller number of Monitor Units (MU) is in general desired. For each patient, the following information was collected to compare the three planning techniques: heart mean dose, left and right lung V20 Gy, contralateral (right) breast mean dose, cord max dose, and MU. Results: The comparison is presented in the following table. Conclusion: In the setting of bilateral chest wall reconstruction, opposed tangent beams cannot consistently achieve desired heart and left lung sparing. This is contrary to the goal of applying DIBH. VMAT appears to be preferential to IMRT for planning and delivering radiation to patients with bilaterally reconstructed chest walls being treated with DIBH because of 1) consistently lower MU; 2) consistently lower heart mean dose and max cord dose; and 3) comparable lungs and contra-lateral breast dose. More patient data will be collected to provide more statistically meaningful clinical guidelines. Author Disclosure: T. Volpe: None. R. Margiasso: None. Z. Saleh: None. L. Kuo: None. L. Hong: None. A. Ballangrud: None. D. Gelblum: None. M. Remis: None. J.O. Deasy: None. X. Tang: None.

3547 Acute Radiation Dermatitis in Head and Neck Patients Treated With VMAT Versus IMRT E. Sapir, J. Bredfeldt, M. Schipper, K. Masi, M.M. Matuszak, and A. Eisbruch; University of Michigan, Ann Arbor, MI Purpose/Objective(s): VMAT has notable advantages in treatment delivery efficiency, however it may spread low doses to larger volumes compared with IMRT. We had qualitatively observed a trend towards increased skin toxicity for head and neck squamous cell carcinoma (HNSCC) treated with VMAT vs. IMRT. We present a quantitative clinical analysis of skin toxicity in HNSCC patients (pts) treated with VMAT vs. IMRT. In addition, we present the results of surface dose measurements on a phantom with VMAT and IMRT plans. Materials/Methods: Skin toxicity data of consecutive 102 HNSCC pts treated with IMRT and 88 pts irradiated with VMAT between 2013 and 2014 were reviewed. An assessment of the skin toxicity during treatment was conducted by the treating physician using the Common Terminology Criteria for Adverse Events (CTCAE v4.0). Skin toxicity was defined as 2. The Cochran-Armitage trend test was used to evaluate the relationship between treatment type, chemotherapy and skin toxicity. To evaluate any potential surface dose measurement differences between VMAT and

IMRT, a model HN patient case was planned on an anthropomorphic phantom for both IMRT and VMAT to achieve clinically similar dose distributions. Surface dose was measured using OSLDs in 4 locations on the phantom neck surface. Each plan was delivered 8 times between which the phantom position was reproduced. After correcting for average dose differences, a two sample t-test was performed to determine the difference in mean surface dose between IMRT and VMAT. Results: Pts treated with IMRT or VMAT were: 49 and 47 pts oropharyngeal, 34 and 20 oral cavity, 2 and 4 nasopharyngeal, 3 and 5 hypopharyngeal, 9 and 7 laryngeal and 5 and 5 unknown primary cancers, respectively. The majority of pts had stage III/IV disease (91% and 93% respectively), and the rest had stage I/II. 66 (65%) pts in the IMRT group and 62 (70%) pts in VMAT arm were treated definitively, the rest received adjuvant radiation. The median target dose for definitive radiation was 70 Gy in each group whereas median dose for adjuvant treatment was 60 Gy in IMRT irradiated pts as opposed to 66 Gy in VMAT. 70 (69%) pts in the IMRT group had chemoradiation compared with 72 (88%) pts in the VMAT group. In pts treated with RT alone, VMAT radiation delivery was marginally associated with higher skin toxicity compared with IMRT (pZ 0.06). This trend disappears with addition of chemotherapy to radiation, which by itself was a strong predicting factor for acute dermatitis (p< .0001). In our phantom study, mean fractional surface dose for VMAT (1.33 Gy) was measured to be 6% higher (p<.0001) than for IMRT (1.24 Gy). Conclusion: When treated with RT alone, HNSCC pts irradiated with VMAT had higher rate of radiation dermatitis compared with pts treated with IMRT, supported by findings in the irradiated phantom. Skin dose reduction treatment planning techniques should be considered for HN cancer patients treated with VMAT. Author Disclosure: E. Sapir: None. J. Bredfeldt: None. M. Schipper: None. K. Masi: None. M.M. Matuszak: None. A. Eisbruch: None.

3548 CTV-based Robustness Optimization Versus PTV-based Conventional Optimization for Intensity Modulated Proton Therapy Planning N. Cao,1 J. Saini,1 S.R. Bowen,2 S. Apisarnthanarax,2 R. Rengan,2 and T.P. Wong1; 1Seattle Cancer Care Alliance Proton Therapy, A ProCure Center, Seattle, WA, 2University of Washington, Seattle, WA Purpose/Objective(s): To investigate whether CTV-based robustness optimization is less sensitive to uncertainties in beam range and patient setup than PTV-based conventional optimization for intensity modulated proton therapy (IMPT) planning for liver and lung cases. Materials/Methods: 3 liver cases and 3 lung cases were planned with both CTV-based robustness optimization method and PTV-based conventional optimization method. The CTV was generated from all phases of the 4DCT data. PTV was expanded from CTV by adding 5-7 mm margin. For CTV-based robustness optimization, 3.5% uncertainty in beam range, as well as 3 mm setup uncertainties in superior-inferior, left-right lateral and anterior-posterior directions were incorporated in the plan optimization process. The planning optimization was aimed to ensure CTV coverage without using PTV. For PTV-based optimization, no uncertainties were included for planning and optimization was based on achieving sufficient CTV and PTV coverage. For both approaches, robustness analysis with 3% changes in beam range, and 3 (and 5) mm change in superiorinferior, left-right lateral, anterior-posterior directions were generated. The