Abstract Book

S868

ESTRO 37

2 The Royal Marsden Hospital NHS Foundation Trust, Radiotherapy, Sutton, United Kingdom

population of patients with variable prognoses. This heterogeneity presents a challenge to both physicians developing treatment recommendations and patients who ultimately choose a specific treatment approach. The purpose of this study was o prospectively evaluate multiparametric MRI (mpMRI) staging accuracy as an independent predictor of outcome as compare to traditional clinical variable in patients treated with high- dose-rate (HDR) brachytherapy (BT) and external beam radiotherapy (EBRT) for intermediate-risk localized prostate cancer. Material and Methods In September 2010 we launched a prospective study of mpMRI guided HDR-BT (15Gy) and supplemental EBRT (37.5Gy) for intermediate and high-risk prostate cancer. The primary endpoints were biochemical (nadir +2ng/mL) and metastatic failure. Clinical variables included baseline PSA value, clinical T- stage, Gleason score, percentage of positive cores on prostate biopsy, use of hormonal therapy, MRI T-stage, risk-group based on clinical T-stage and risk-group based on MRI T-stage. We specifically analyzed the intermediate-risk subgroup (NCCN classification) according to classical clinical variables (PSA, clinical T stage prior to mpMRI and Gleason score) so as to assess the impact of extracapsular extension or seminal vesicle invasion (i.e T3) on outcomes. Cumulative hazard rate function was estimated through the Nelson-Aalen method in order to describe the cumulative risk of the events of interest. Univariate and multivariate hazard ratios were obtained from proportional hazards Cox regression models. Log rank tests were used to compare hazards across categories of MRI-based T stages. Results Eighty-one patients were prospectively treated, median age was 71 years (range 56-82); Thirty-nine percent were Gleason 6, 36 % Gleason 7 (3+4) and 25% Gleason 7 (4+3); median baseline PSA was 8 ng/mL (1.9-19 ng/mL). Median follow-up was 45 months (range 20-81). At the time of the current analysis, 6 (7.4%) patients had a biochemical failure and 2 (3.7%) had developed distant metastases. The 5-year cumulative risk of biochemical failure for patients with MRI-T1/T2 stage and MRI-T3 stage were 0% and 18% respectively (p=0.002). The 5-year cumulative risk of metastatic failure was 0% and 8% respectively (p=0.018). None of the traditional variables (Primary pattern of Gleason score, PSA or clinical T-stage) predicted for clinical outcomes. The only independent predictors of biochemical and metastatic failure ware the presence of extraprostatic extension on mpMRI and the percentage of positive cores on biopsy. Conclusion In clinical intermediate-risk localized prostate cancer, apart from the well-known risk factor of percentage of positive cores on biopsy, the presence of extraprostatic extension in pre-treatment mpMRI is an independent predictor biochemical failure and metastatic failure. A. Pathmanathan 1 , M. Schmidt 1 , D. Brand 1 , L. Delacroix 2 , C. Eccles 2 , A. Gordon 2 , T. Herbert 2 , H. McNair 2 , N. Van As 1 , R. Huddart 1 , A. Tree 1 1 The Institute of Cancer Research and The Royal Marsden NHS Foundation Trust, Radiotherapy and Imaging, Sutton, United Kingdom EP-1613 Comparison of prostate delineation on multi- modality imaging for MR-guided radiotherapy

Purpose or Objective MR-guided radiotherapy (MRgRT) of the prostate requires sequence optimisation for accurate and timely prostate definition for planning and intrafractional imaging. There has been interest in magnetic resonance imaging (MRI) sequences enhancing the signal void of fiducials (1, 2), required for accurate MR and computed tomography (CT) fusion (3). T2*-weighted (T2*W) MRI uses multiple echo times (4) also resulting in a more defined prostate capsule and reliable depiction of fiducials. This study assesses the variability and accuracy of prostate delineation by radiographers using CT, T2-weighted Five radiographers experienced in delineation or registration of the prostate on CT, delineated the prostate on CT, T2W and T2*W MRI using Monaco v.5.19.02 (research version, Elekta AB, Stockholm, Sweden) for ten patients scanned within the PACE trial (NCT01584258), following a training session. Time taken for delineation was recorded and images were scored (0- 10) for ‘image quality’ and ‘confidence in contouring’. There was a minimum of two weeks between contouring (T2W) and T2*W MRI. Material and Methods (1) Interobserver variability (IOV) by comparing individual contours to a simultaneous truth and performance level estimation (STAPLE) contour created from all radiographer contours (2) Accuracy by comparison of radiographer contours to a ‘gold standard’ created from the STAPLE of three physician contours. Contours were assessed using Monaco ADMIRE software v.2.0 (research version, Elekta AB). For each comparison, the overlap measures Dice similarity co-efficient (DSC) and Cohen’s kappa were recorded, (higher values indicate greater agreement). In addition, distance measures of Hausdorff distance (HD) and mean distance between contours were recorded (lower values indicate greater agreement). SPSS statistics v.23 was used to perform separate Friedman’s tests for each comparison between modalities. Where significant, pair-wise group comparison was undertaken using Wilcoxon’s signed rank (Bonferroni corrected). Results to minimise recall bias. Assessment was made of;