ESTRO 2020 Abstract Book

S858 ESTRO 2020

Toxicities in arm A and B were compared with Mann Whitney U tests and a log rank test in a time to event analysis of RP≥G2. The relationship of clinical and dosimetric parameters with RP≥Grade2 was investigated with Chi2 (binary variables) and Mann Whitney U tests. For RP≥G2, a multivariate logistic regression with regularization and a 10-fold cross-validation (Least Absolute Shrinkage and Selection Operator (LASSO) function for general linear models) was used to avoid overfitting. Results RP, dyspnea and cough were not significantly different (p- values ranged 0.348 (RP) to 0.982 (Cough pre RT)) between the two arms. Mean lung dose did not reach significance (MLD_A=14.9Gy [5.8,23.1], MLD_B=17.5Gy [8.6,24.8], p=0.075). A time to event analysis of RP≥G2 (Figure 1) indicated a tendency towards higher risk of RP in arm B but was not significant (p=0.180). The patients were henceforth analyzed as one group. Dyspnea at baseline (G1-4) and increase of cough during RT were related to RP≥G2 (p=0.044, p=0.008, Figure 2), while cough at baseline and increase of dyspnea during RT were not significant. Risk factors for RP were significant for infection at baseline or during RT (OR=8.087, p=0.026), GTV size (OR=1.077/10cm3, p=0.002) and PTV size (OR=1.264/100cm3, p=0.004). Smoking (OR=0.160, p=0.063) and MLD (OR=1.077/Gy, p=0.067) did not reach significance. None of the recorded pulmonary function measures were linked to RP. In the multivariate logistic regression analysis, the null model (equaling the general probability of RP≥G2 in the cohort) was within one standard deviation of the minimum deviance.

Conclusion There were no significant differences in pulmonary toxicities between the two arms. For RP, target size and infection were significantly linked to the risk of RP. Patients presenting dyspnea at start and increase of cough during RT showed higher incidence of RP as well. No multivariate RP model could be established for RP, most likely because of low sample size. PO-1582 Dose rate effects in robotic radiosurgery treatments E. Zoros 1 , A. Moutsatsos 2 , L. Lekas 2 , P. Pantelakos 2 , E. Pantelis 1 1 Medical School of National and Kapodistrian University of Athens, Medical Physics Laboratory, Athens, Greece ; 2 Iatropolis Clinic, CyberKnife and TomoTherapy department, Athens, Greece Purpose or Objective Treatment time in robotic radiosurgery depends on the size and shape of the target, the number and size of collimators, as well as the nominal dose-rate of the system. In this study, the influence of dose-rate on the biologically effective dose (BED) was investigated for a patient cohort suffered from vestibular schwannomas. Material and Methods For the purpose of this study, a group of 24 patients treated for vestibular schwannomas with a G4 CybeKnife (Accuray Inc., CA, USA) model was selected. Target volumes varied from 0.27 to 5.66 cc, with a mean value of 2.71 cc. Collimator sizes of 5, 7.5, 10 and 12.5 mm were used. On average 146 beams were used ranging from 90 to 202. All cases were treated with a prescription dose of 14 Gy in a single fraction planned using the MultiPlan v.4.6 (Acurray Inc) treatment planning system. An in-house ray tracing dose calculation algorithm based on the measured beam data was developed to calculate the dose-rate in a voxel-level from each beam of each case using 800 and 1000 MU/min linear accelerator (linac) nominal output values. Second quartile dose rate distributions were created in each case indicating the dose-rate that