ESTRO 38 Abstract book

S452 ESTRO 38

dose in percentage was studied. The results were further analysed by using Wilcoxon rank sum test. Analysis was done separately per each toxicity group and its comparison group.

(RT) has been proven effective to prolong biochemical control, but is associated with a higher risk of grade ≥2 toxicity. We aimed to develop and validate a magnetic resonance imaging (MRI)-based radiomics signature with prognostic value in patients with high risk PCa, in order to guide patients management especially regarding the use of adjuvant RT. Material and Methods One hundred-seven patients with histologically proven PCa with high-risk features namely pT3-4, and/or R1, and/or Gleason 8-10 and treated with RP with or without lymphadenectomy from 2010 to 2016 at our institution, and with available preoperative pelvic MRI were included. Adjuvant treatment, postoperative PSA>0.04 ng.mL and c/pN1 were exclusion criteria. Prostatic tumors were delineated on both the ADC and T2 MRI sequences using 3D Slicer® v4.8.0 and IBSI-validated radiomic features were extracted. The cohort was randomly split into training (n=70) and testing (n=37) sets. The selection of a subset of 10 radiomics features was performed in the training set using an aggressive false discovery reduction procedure relying on stability and correlation checks and robustness score. Correlation with BCR was assessed using ROC curves and Cox regression analysis was performed to identify independent predictive variables. Results After a median follow-up of 52.0 months, 17 (16.0%) patients experienced BCR. In the training set, none of the clinical features was significantly correlated with BCR. Amongst the 10 radiomic features pre-selected, an ADC feature outperformed the others with an AUC of 0.85 to predict biochemical relapse-free survival (bPFS) (p = 0.0001). In the testing set, this feature remained significantly predictive of BCR (AUC of 0.76) and prognostic of bPFS (p = 0.0236). Conclusion One ADC radiomic features appeared to be strongly predictive of BCR following RP. This feature could help in redefining the population who would mostly benefit from treatment intensification such as adjuvant RT. PO-0858 Dosimetric correlation analysis of observed toxicities in prostate cancer patients treated with SBRT N. Pienimäki 1 , K. Vuolukka 2 , J. Heikkilä 2 , J. Palmgren 2 , J. Seppälä 2 1 University of Eastern Finland, Department of Applied Physics, Kuopio, Finland ; 2 Kuopio University Hospital, Centre of Oncology, Kuopio, Finland Purpose or Objective The aim of this study was to investigate the correlation between dosimetric parameters and registered toxicities in ultra-hypofractionated radiotherapy of prostate cancer (PCa). Material and Methods Between the years 2012 and 2015 220 patients received ultra-hypofractionated radiotherapy (5x7.0 Gy / 7.25 Gy) delivered with CyberKnife robotic system for the radical treatment of PCa at Kuopio University Hospital, Finland. Toxicities were observed in 36 patients which were divided in four different groups; Intermediate-term Genitourinary toxicities (GU, n=19), intermediate-term rectal toxicity (GI, n=7), Infectious problems toxicity (INF, n=7) and Acute toxicity (ACUTE, n=4). Comparison groups of patients were formed for each toxicity group by selecting similar patients without any toxicity according to age, size of PTV and hormonal treatment. Average DVH curves (Fig.1) for rectum and bladder were produced for each group and were used to choose representative dose- volume parameters for correlation analysis. In correlation analysis the mean values of absolute volume doses, mean values of relative volume doses and relative volumes for

Results For the INF patients significant differences between the groups were discovered for the higher doses of bladder (D5% (p=.007) and D10% (p=.005)). In addition, for bladder the results between INF patients and comparison group differed significantly with V35 (p=.0006) and somewhat notably with V30 (p=.07). When investigating the doses delivered to rectum, significant differences were observed with medium dose value V10 (p=.038). Also, with V15 (p=.07) and D50% (p=.07) some notable differences were observed. For the GI patients in the analysis for rectum, it was observed that the doses of D5cc(p=.07) and D10cc(p=.05) differed remarkably between the groups. When analysing relative volumes for dose in percentage, it was observed that for rectum V15(p=.07) the volumes for doses were remarkably different. Further research is still needed in order to verify the significance of these observed values. Significant correlation between toxicity symptoms and dosimetric values were not found for the GU and for the ACUTE groups, respectively. All the notable results are presented in Table 1.

Conclusion High maximum doses to critical organs was found to correlate strongly with observed INF and GI toxicities for the PCa patients treated with ultra-hypofractionated radiotherapy. For the GU and ACUTE groups no correlation between the groups in toxicity and dose distributions was found. For the medium range doses there were only slight differences between toxicity and control group, respectively. The daily variation of the treated anatomy could have an effect for the non-significance of the lower doses since dose received by the critical organs of may variate by fraction. PO-0859 Validation of genetic variants associated to late severe toxicity after prostate cancer RT A. Cicchetti 1 , T. Rancati 2 , F. Ballarini 3 , M. Dispinzieri 4 , T. Giandini 5 , B. Avuzzi 4 , C. Cozzarini 6 , C. Fiorino 7 , R. Valdagni 8