Abstract Book

S196

ESTRO 37

valves, atria and ventricles) were all contoured as organs at risk (OAR) and included in the optimization process. Maximum and mean dose were compared for all organs; second breast and lung cancer risks were estimated by calculating the lifetime attributable risk (LAR), using the LAR ratio (LAR b-VMAT -to- LAR fa-VMAT ) as a comparative measure. CAD risk was estimated by calculating the relative risk (RR), derived from the mean dose received by the sum of coronary arteries. The figure shows the three different anatomical presentations of the thirty patients included in this study.

at risk according to specific lymphoma dose constraints), significantly decreased the RR for CAD, particularly in bulky presentations, with similar second breast and lung cancer risks. OC-0384 Relationship of mean heart dose and cardiac substructure dose over evolving radiation techniques in Hodgkin lymphoma B. Hoppe 1 , N.P. Mendenhall 1 , J.E. Bates 1 , C.G. Morris 1 , S. Flampouri 1 1 University of Florida College of Medicine, Radiation Oncology, Jacksonville, USA Purpose or Objective The understanding that mean heart dose is an important predictor for late cardiac morbidity >20 years after treatment for Hodgkin lymphoma (HL) is based on data from patients treated decades ago with standard large- field 2D radiotherapy (RT) techniques. Over time, RT treatment in HL has evolved with smaller radiation fields and more conformal techniques, such as 3DCRT, IMRT, and proton therapy. We investigated the relationship and correlation between mean heart dose and cardiac substructures with these more conformal techniques. Material and Methods Under IRB approval, we examined a pre-existing dosimetric database of involved-site radiotherapy (ISRT) plans for mediastinal lymphoma including 3D conformal (3D; n=26), intensity-modulated RT (IMRT; n=30), and proton therapy (PT; n=36). The database included heart and cardiac sub-structure contours, including the left anterior descending artery (LAD), mitral valve (MV), tricuspid valve (TV), aortic valve (AV), left and right ventricle (LV, RV), and left or right atria (LA, RA). For each plan, the mean heart dose was evaluated for correlation(R 2 ) and slope(beta) of the relationship with mean dose to the cardiac substructures. Results Mean heart dose for the cohorts was 18.5 Gy, 13.9 Gy, and 10.2Gy for 3D, IMRT, and PT, respectively. The linear regression plots between mean heart dose and the cardiac substructures are shown in Figure 1. Table 1 reports the correlation coefficient and slope for the relationships between mean heart dose and mean cardiac substructure dose, demonstrating less correlation with increasingly conformal delivery. A strong correlation(R 2 ≥70%) was observed for 3D with the RV, LV, and LAD and for IMRT with RV and LV; there was none for PT. A moderate correlation(50%

Results fa-VMAT resulted in lower dose to whole heart (p = 0.0028), whole coronary arteries (p = 0.0001), left ventricle (p = 0.007), aortic valve (p = 0.0004) and in lower V 20Gy to the lungs (p = 0.008). Conversely, b-VMAT resulted in lower mean dose (p = 0.033) and V 4Gy (0.04) to the female breasts. Dosimetric parameters are reassumed in the table. A significant lower RR for CAD was observed for fa-VMAT, with a mean 20% risk reduction (p = 0.0002); the entity of dose reduction was marginally higher for patients with bulky disease (25% vs 15%, p = 0.06). Despite these dosimetric differences, LAR ratio for breast cancer (p = 0.146) and lung cancer (p = 0.148) were not statistically different between fa-VMAT and b-VMAT, respectively, regardless of gender and anatomic presentation of disease.

Conclusion Among a heterogeneous cohort of mediastinal HL patients, reflecting the most frequent clinical presentations, an innovative fa-VMAT planning solution, compared to b-VMAT (both optimized for multiple organs