ESTRO 38 Abstract book

S453 ESTRO 38

1 Fondazione IRCCS Istituto Nazionale dei Tumori and University of Pavia, Prostate cancer program/Department of Physics, Milan, Italy ; 2 Fondazione IRCCS Istituto Nazionale dei Tumori, Prostate cancer program, Milan, Italy ; 3 University of Pavia, Physics Department INFN-Section of Pavia, Pavia, Italy ; 4 Fondazione IRCCS Istituto Nazionale dei Tumori, Radiation Oncology 1, Milan, Italy ; 5 Fondazione IRCCS Istituto Nazionale dei Tumori, Medical Physics, Milan, Italy ; 6 San Raffaele Scientific Institute, Radiation Oncology, MIlan, Italy ; 7 San Raffaele Scientific Institute, Medical Physics, MIlan, Italy ; 8 University of Milan and Fondazione IRCCS Istituto Nazionale dei Tumori, Department of Oncology and Hemato-oncology / Prostate cancer program & Radiation Oncology 1, Milan, Italy Purpose or Objective To validate the impact of single nucleotide polymorphisms (SNPs) on radio-induced toxicity (tox) after curative RT for Comprehensive literature search for SNPs associated to increased risk for tox after PC RT was carried out. A panel of 8 SNPs was established. A cohort of PC patients (pts) from a prospective trial on RT-induced toxicity was identified. They were treated with radical IMRT to a total dose of 78Gy at 2Gy/fr or to 65-74.2Gy with 2.55-2.65Gy/fr. Blood samples were collected during the follow up (fup) for DNA extraction (Oct2016-May2017). Selected endpoint was overall moderate-severe late tox assessed at 3 year-fup, i.e. scoring of at least one symptom among the following:grade≥2 rectal bleeding (CTCAE), chronic faecal incontinence (FI) (CTCAE average grade>1 during whole fup), severe urinary dysfunction (15- point worsening in IPSS questions #1to#7), hematuria, stricture and moderate-severe urinary incontinence (by 2-point worsening in ICIQ question #3). Association of selected SNPs with overall late tox was investigated through logistic regression (LR). Confirmed SNPs (similar Odds Ratios) were later included into a LR model together with Equivalent Uniform Dose (EUD) for bladder and rectum, as computed from DVHs corrected with a/b=3Gy). For both bladder and rectum, volume parameters n for overall late tox were determined by literature search. Results 108 pts were included in the study. Overall late tox was scored in 37 pts (34%). 3/8 investigated SNPs were confirmed: rs3931914CG, rs2293054GA and rs845552AG (De Langhe Rad&Onc 2014, see Fig. 1 for details). From literature search, we identified a volume parameter n=0.1 for bladder (p=0.003, i.e. bladder acting as serial organ for late urinary tox) and n=0.25 for rectum (p=0.007, i.e. relevance of medium to high doses is kept, correctly matching the presence of 2 distinct patterns in late rectal tox, which includes both bleeding, affected by small volumes receiving high doses, and FI, affected by mean rectal dose). The final LR model included: bladder EUD (fitted OR=1.10), rectal EUD (fitted OR=1.09) and SNPs (literature OR=2.59, 0.23 and 5.3 for rs845552AG, rs2293054GA and rs3931914CG ). Total p=0.015, calibration slope=0.83 and AUC=0.70, details on calibration in fig 2b. Fig 2a presents the nomogram derived from LR model, where the genetic risk score is reported as the number of SNPs harboured by the single patient. The nomogram highlights that given the same dose to organs at risk (i.e. same bladder EUD+rectal EUD) presence of 1 SNP is increasing tox risk by 5%, while presence of 2 or 3 SNPs increases tox risk by more than a factor 2. prostate cancer (PC). Material and Methods

Conclusion 3 SNPs previously identified as associated with increased risk of RT-induced tox were confirmed in this validation study. Late overall tox incidence was well described by a data+literature driven model (effect size of SNPs and volume parameters for EUD calculations from literature, effect size for bladder and rectal EUD fitted on data). PO-0860 Improving consistency of proximal seminal vesicle delineation for prostate SBRT K. Morrison 1 , N. Van As 1 1 Royal Mardsen NHS Trust, Radiotherapy, London, United Kingdom Purpose or Objective Accurate and consistent clinical target volume (CTV) delineation is important in stereotactic body radiotherapy (SBRT). The proximal 1 - 2 cm seminal vesicles (pSV) are included in the CTV for low/ intermediate risk prostate cancer, however there is no clear consensus on how this should be defined. Within the PACE SBRT trial, inconsistent pSV delineation has been demonstrated between centres. The aim of this study is to investigate a delineation method which will improve consistency in further trials. Material and Methods 21 experienced prostate radiotherapy clinicians (consultants and trainees), were identified at a national uro-oncology conference, and invited to participate in a contouring exercise. Using fused CT and MRI imaging from one selected intermediate-risk prostate cancer case, participants were instructed to contour: superior prostate (partial contour provided); full seminal vesicles (SV); and proximal 1cm SV as defined by the clinician (pSV method A). A further pSV contour was defined by creating a 1cm circumferential prostate margin, and including the section of SV within this region (pSV method B). For each method, investigational contours were compared to a set of reference contours, using measured volume (cc) and calculated conformity indices: DICE similarity coefficient; Geographical Miss Index (GMI); and Disconcordance Index (DI). Results for methods A and B were compared to analyse differences in consistency, using paired t-test to measure statistical significance. Results 19 sets of contours were complete and analysable. For pSV method A, the mean investigational contour volume was 11.07 cc (95% CI 9.04 – 13.11) compared to the reference contour 8.65 cc. For method B, the mean investigational