ESTRO 38 Abstract book

S323 ESTRO 38

constraining V64 < 3cc (corresponding to 4x4cm 2 skin surface) should keep the risk of G3 ACT below 10%. Considering therapy DVHs in place of planning DVHs does not seem to improve prediction.

This large registry study showed that for the patients treated with RT for H&N cancer, the strongest risk factor for late xerostomia is the dose to the total parotid volume. There does not seem to be any benefit in predominantly sparing the contralateral parotid. OC-0611 Planned and delivered DVHs of the skin predict acute cutaneous toxicity after IGRT for HN cancer M. Mori 1 , S. Foti 2 , I. Dell' Oca 2 , M. Branchini 1 , S. Broggi 1 , G.M. Cattaneo 1 , N. Di Muzio 2 , C. Fiorino 1 1 San Raffaele Scientific Institute, Medical Physics, Milan, Italy ; 2 San Raffaele Scientific Institute, Radiotherapy, Milan, Italy Purpose or Objective To explore the ability of skin dose-volume histograms (DVH) to predict the risk of acute cutaneous toxicity (ACT) after Radio-chemotherapy for Head and Neck (HN) cancer patients (pts) and to test if DVHs recalculated during therapy may improve prediction. Material and Methods Seventy HN pts were treated with Helical Tomotherapy (HT) with radical intent (SIB technique: 54/66 Gy to PTV1/PTV2 in 30fr) and concurrent chemotherapy (excluding Cetuximab). The skin was defined as a superficial body layer 2mm thick (SL2). Prospectively evaluated CTCAE v4.0 ACT data were available. Average absolute DVH of SL2 for pts who developed severe (G3) and severe/moderate (G3/G2) ACT were assessed. The differences against DVHs of patients without ACT (G0/G1/G2 and G0/G1 respectively) were analyzed by two-tails t-test; univariable logistic analyses were performed selecting DVH values corresponding to the lowest p-values at t-tests. Multivariable logistic analyses were also performed considering CTV volume, age, sex, chemotherapy as potential predictive factors. In addition, the effect of dose changes during therapy was quantified in 32/70 pts, by recalculating DVHs at half and end therapy with a previously validated dose-of-the-day calculation method. The association between therapy DVHs and G3/G2 ACT was tested and compared against the one using planning DVH. Results Sixty-one % of pts experienced G2/G3 ACT (rate of G3= 19%). As reported in Fig 1 (left), differences in skin DVHs (G2/G3 vs G0/G1) were significant in the range 53-68Gy. V56 was the most predictive parameter (OR= 1.12, 95%CI= 1.03-1.21, p= 0.001), with a best cut-off of 7.7cc. The logistic model for V56 was well calibrated (slope/R 2 of calibration plot: 0.97/0.99). When considering G3 ACT (Fig 1, right), V64 was the best predictor (OR= 1.13, 95%CI= 1.01-1.26, p= 0.027) with a best cut-off of 2.7cc. Average V64 were 2.2cc and 6cc for the two groups (G3 vs G0-G2); the logistic model for V64 also showed good calibration (slope/R 2 : 0.99/0.60). The association between DVHs and G2/G3 ACT was significant also in the subgroup of 32 pts with dose-of-the-day calculations available; t-tests on average DVHs at planning, half and end of treatment (Fig 2), showed none improvement of the association with ACT when considering therapy DVHs in place of planning DVHs.

OC-0612 A case-control study of brainstem substructures and morbidity following pediatric proton therapy C. Stokkevåg 1 , D.J. Indelicato 2 , L.F. Fjæra 3 , Y. Lassen- Ramshad 4 , K.S. Ytre-Hauge 3 , Z. Li 2 , L. Toussaint 5 , O. Casares-Magaz 5 , S. Flampouri 2 , R. Mikkelsen 6 , C. Pedro 7 , O. Dahl 1 , L.P. Muren 5 1 Haukeland University Hospital, Department of Oncology and Medical Physics, Bergen, Norway ; 2 University of Florida, Department of Radiation Oncology, Jacksonville, USA ; 3 University of Bergen, Department of Physics and Technology, Bergen, Norway; 4 Aarhus University Hospital, Danish Centre for Particle Therapy, Aarhus, Denmark ; 5 Aarhus University Hospital, Department of Medical Physics, Aarhus, Denmark; 6 Aarhus University Hospital, Department of Neuroradiology, Aarhus, Denmark; 7 Portuguese Institute of Oncology Francisco Gentil de Lisboa, Department of Radiotherapy, Lisbon, Portugal Purpose or Objective Brainstem toxicity is a rare but severe condition that can follow treatment of pediatric brain tumors. Dose received by the brainstem is one of the treatment-specific causes of brainstem morbidity. Current data on brainstem toxicity are limited and the brainstem is often regarded as a single organ with dose constraints determined without considering potential regional sensitivity within the brainstem. The brainstem core has in some studies been found more radiosensitive than the outer surface, and diffusion tensor imaging has recorded non-uniform changes across different substructures of the organ. In order to investigate if outcome could be explained by different radio-sensitivities across the brainstem, dose/volume parameters in different anatomical brainstem regions were evaluated for pediatric brain tumor patients with brainstem toxicity vs. patients with no toxicity. Material and Methods A matched case-control was performed within a cohort of 954 pediatric brain tumor patients treated with proton therapy from 2006 to 2017. Ten patients presenting with symptomatic grade 2+ brainstem toxicity were each matched to three controls based on diagnosis, age within ± 1.5 years and by whole brainstem dose metrics (D0.1cc within ± 2 Gy, and D10% within ± 2 Gy). The brainstem core (brainstem cropped by 3 mm) and the regions medulla oblongata, midbrain, pons were segmented on T1/T2 MRI sequences fused with the CT scan used for treatment planning. The pons was further divided into four transversal zones (posterior, middle posterior, middle anterior and anterior) to approximate fiber tracts (Figure 1). Dose/volume metrics (obtained using an RBE of 1.1) for

Conclusion The relationship between SL2 DVH and ACT after HT of HN pts with a SIB approach was quantified. The results should help in identifying pts at risk and in implementing more effective skin-sparing planning strategies. In particular,