Irvine. Travel Expenses; AACR, Academic Institutions, American Board ... D.P. Lindsay,1 J.M. Caster,1 K. Wang,1 J.H. Myung,2 R.C. Chen,1. B. Chera,1 G.P. ...
S50
International Journal of Radiation Oncology Biology Physics
multi-institutional datasets is required. Clinically, this is a promising approach to identify a subset of head and neck radiotherapy patients who may benefit from hypoxia-based dose modification. Author Disclosure: M. Crispin-Ortuzar: None. A. Apte: None. M. Grkovski: None. J. Oh: None. N. Lee: Consultant; Lily. Advisory Board; Merck, Pfizer, Vertex. Partnership; AGS. J. Humm: None. J.O. Deasy: AAPM.
release in regions that are likely associated with swallowing function, and the good ability of the mean dose on S.05 to predict the dysphagia outcome is confirmed by a relatively high AUC. We believe that the VBA may provide new insights into spatial signature of radiation sensitivity in a highly composite region such as the H&N district. Author Disclosure: S. Monti: None. G. Palma: None. V. D’Avino: None. M. Gerardi: None. D. Ciardo: None. D. Alterio: None. B.A. JereczekFossa: None. M. Quarantelli: None. R. Pacelli: None. L. Cella: None.
106 Regional Dose Differences Associated with Radiation-Induced Acute Severe Dysphagia S. Monti,1 G. Palma,2 V. D’Avino,2 M. Gerardi,3 D. Ciardo,3 D. Alterio,3 B.A. Jereczek-Fossa,3,4 M. Quarantelli,2 R. Pacelli,5 and L. Cella2; 1IRCCS SDN, Napoli, Italy, 2National Research Council, Institute of Biostructures and Bioimaging, Napoli, Italy, 3Istituto Europeo di Oncologia, Milano, Italy, 4Department of Oncology and Hemato-oncology, University of Milan, Milano, Italy, 5Federico II University School of Medicine, Department of Advanced Biomedical Sciences, Napoli, Italy Purpose/Objective(s): To apply a voxel-based analysis (VBA) to explore possible correlation between radiation-induced acute dysphagia (RIAD) and local dose release in a cohort of patients treated for head and neck (H&N) tumor with RT. Materials/Methods: In a cohort 42 H&N patients treated with definitive RT (35 fractions of 2 Gy) and no previous surgery, 9 subjects developed severe RIAD (CTCAE v 4.0 grade 3). Each planning CT was registered to a common coordinate system (CCS) via a log-diffeomorphic approach. The obtained deformation fields were used to map the dose of each patient into the CCS. The dose warping accuracy was evaluated by Dice, modified Hausdorff distance (MHD) and dose-organ overlap (DOO) scores. A preliminary significance analysis of the dose differences was performed according to the normalized maximum dose difference (Tmax). Then, a statistical mapping scheme for nonparametric multiple permutation inference on dose maps with threshold-free cluster enhancement (TFCE) was applied, and the corresponding P < 0.05 subregions (S.05) were generated. A ROC-based test was performed on the mean dose extracted from S.05. Results: The registration process resulted in a geometrically robust and accurate dose warping. In Table, pre- and post-registration scores are reported along with the significance of accuracy improvement. A significantly higher dose was delivered to RIAD patients (Tmax statistic: P Z .038) in two voxel clusters (overall volume of 5.1 cm3 according to TFCE) located in correspondence of Cricopharyngeus Muscle (CM) and Cervical Esophagus, as shown in Table 1. The ROC-AUC from the mean dose on S.05 was .81. Conclusion: In this study, a method for the dosimetric VBA was applied to a cohort of H&N patients. To the best of our knowledge this represents the first attempt to go beyond the organ-based philosophy in normal-tissue complication probability modeling in the H&N region and to explore a correlation between radiation-induced toxicity and local dose differences. Compared to other districts, H&N poses serious difficulties in terms of registration accuracy, due to the high motility of the structures and considerable inter-patient anatomical variability. Notably, the adopted registration strategy succeeded in a sharp normalization of patients’ doses to a CCS and the following statistical analysis was able to highlight dose differences between RIAD and non-RIAD subjects in very elusive structures, such as the CM. The TFCE approach pointed out a higher dose
107 An Automated Multiparametric MRI Quantitative Imaging Prostate Habitat Risk Scoring System for Defining External Beam Radiation Therapy Boost Volumes A. Pollack,1 M.C. Abramowitz,1 D. Kwon,2 K.R. Padgett,3 I.M. Reis,2 Y. Tschudi,1 F.M. Chinea,1 O.N. Kryvenko,1 S. Punnen,1 and R. Stoyanova1; 1University of Miami, Miami, FL, 2University of Miami Sylvester Comprehensive Cancer Center, Miami, FL, 3Department of Radiation Oncology, University of Miami / Sylvester Comprehensive Cancer Center, Miami, FL Purpose/Objective(s): Dose escalation above 80 Gy improves the control of intermediate to high risk prostate cancer. We hypothesize that GTV dose escalation, as opposed to whole prostate dose escalation, will improve tumor control without increasing side effects. A major obstacle is defining the GTV in a systematic reproducible way. A habitat risk scoring system based on prostate multiparametric MRI (mpMRI) has been developed and applied using an automated method to define radiotherapy (RT) boost volumes in the background of a randomized Phase II clinical trial (BLaStM) comparing two methods of increasing dose to the mpMRIdefined habitat tumor region(s). Materials/Methods: An automated pixel by pixel method was optimized, using Dynamic Contrast Enhanced (DCE)-MRI and Apparent Diffusion Coefficient (ADC) sequences, to be associated Gleason score (GS). A “Habitat Risk Score” (HRS) was devised in ten subcategories with increasing levels associated with a greater risk of harboring higher GS’s and depicted as a heat map. The defined mpMRI habitats were first related to radical prostatectomy (RP) GS tumor volumes in 3-dimensions as contoured by the study pathologist. A workflow for RT planning was created in MIM where the HRS contours were migrated to the planning CT to define the GTV using rigid fusion. Results: There were 39 regions of interest in 12 patients who underwent RP (RP-ROIs) after mpMRI; these were also assigned a PI-RADS score 4/ 5. There were 18 GS 7 and 21 GS 6. The HRS algorithm had a higher sensitivity than PI-RADS at most levels (Table 1), ranging from 98.9% for HRS 4 to 36.6% for HRS 9, compared to 44.0% for PI-RADS. There were no HRS 10 in this cohort. However, the specificities for HRS ranged from 27.6% to 96.1% and were lower than the specificity of 98.3% for PIRADS. In particular, HRS 6 had 83.4% sensitivity and 73.5% specificity. HRS 6 had higher accuracy (78.4%) than PIRADS (71.6%). The volumes of HRS 6 and RP ROIs were strongly correlated (r Z 0.97; P < 0.0001) using Spearman’s correlation coefficient. HRS maps were created for the first 41 patients on the BLaStM trial to direct MRI-guided prostate biopsies at the time of fiducial marker placement and to guide boost volumes. In total, there were 59 prostate mpMRI lesions with the mean number of lesions being 1.4 per patient. The mean lesion volume (+/-SD) was 2.28 2.83 cm3. The median lesion volume was 1.48 (range 0.19-15.34) cm3.
Abstract 106; Table 1 Pre Median Dice MHD (mm) DOO
.67 2.31 .59
S.05 Mean Dose (Gy)
33
Abstract 107; Table 1
Post Range
.03-.83 .53-24.13 .32-.78
Median .80 1.22 .87
Non-RIAD 10-57
Range .59-.82 .78-4.46 .74-.92
P-value