ESTRO 38 Abstract book

S840 ESTRO 38

EP-1556 The effect of an endorectal balloon on GI toxicity after EBRT for localized prostate cancer V. Groen 1 , L. Kerkmeijer 1 , E. Monninkhof 2 , M. Van Schie 3 , M. Kunze-Busch 4 , H. De Boer 1 , U. Van der Heide 3 , F. Pos 3 , K. Haustermans 5 , R.J. Smeenk 4 1 UMC Utrecht, Radiation Oncology, Utrecht, The Netherlands ; 2 UMC Utrecht, Epidemiology, Utrecht, The Netherlands ; 3 The Netherlands Cancer Institute, Radiation Oncology, Amsterdam, The Netherlands ; 4 Radboud University Nijmegen Medical Centre, Radiation Oncology, Nijmegen, The Netherlands ; 5 University Hospitals, Radiation Oncology, Leuven, Belgium Purpose or Objective Gastro-intestinal (GI) toxicity is a common side effect following external beam radiation therapy (EBRT) in patients with localized and locally advanced prostate cancer. The introduction of IMRT and IGRT with subsequent margin reduction has shown to reduce the risk of GI toxicity. However, further reducing toxicity remains a challenge. Endorectal balloons (ERB) are applications initially developed to reduce intrafraction variations of the prostate during EBRT. In addition, studies have suggested the potential ability of an ERB to reduce EBRT related GI toxicity. The aim of this study was to compare GI toxicity in patients with and without an ERB after EBRT for localized prostate cancer. Material and Methods We included 568 patients with intermediate and high risk prostate cancer of the phase 3 multicenter FLAME RCT in this observational analysis. The primary purpose of the FLAME trial is to compare EBRT in a standard treatment arm (n=284) of 77 Gy to the entire prostate + seminal vesicles in 35 fractions and a dose-escalation arm (n=284) consisting of an additional integrated boost up to 95 Gy to the macroscopic visible tumour on mpMRI. One of the participating centers applied an ERB during all radiation sessions (n=49). The ERB was filled with 100cc air. GI toxicity was scored by a physician using the CTCAE version 3.0 in all patients. The effect of the ERB in comparison to no ERB on cumulative GI toxicity ≥ grade 2 was assessed by using a binary logistic regression model using the total study cohort irrespective of randomization. The model was adjusted for age, cardiovascular disease (CVD), diabetes mellitus (DM), T stage of prostate cancer, baseline toxicity, hormonal therapy and radiation dose received in 2cc of the rectum (D2cc). To account for missing data, multiple imputation was used. Furthermore, the prevalence of GI toxicity ≥ grade 2 per follow-up moment was analyzed. Results The incidence of cumulative GI toxicity ≥ grade 2 within 3 years after treatment was lower in the ERB group compared to the group without ERB. After imputation the percentage of cumulative GI toxicity ≥ grade 2 was 29.0% in the no ERB group versus 12.6% in the ERB group, p <0.001. With adjustment for age, CVD, DM, T stage, baseline toxicity, hormonal therapy and rectum D2cc the daily application of an ERB showed a decreased risk of cumulative GI toxicity ≥ grade 2 with an odds ratio of 0.35 (0.27 - 0.46 95% CI p <0.001) (table 1). In figure 1 the prevalence of GI toxicity ≥ grade 2 per FU moment is shown.

to the elective lymph node region (CTV56) in 39 fractions. The patients were CT- and MR-scanned (T2 weighted 3D sequence) for treatment planning and additionally MR scanned at the 10 th , 20 th and 30 th treatment fraction. On each scan, target structures were defined by an experienced oncologist who also validated the OARs defined by an experienced RTT. For evaluation of intra- observer variability, all structures were re-delineated on one MR image set per patient at least 30 days after the initial delineations were performed. Delineations and DIRs were performed in Monaco v 5.40 (Elekta AB, Sweden) treatment planning system (TPS) dedicated to the high field MR-Linac treatment. Structures defined on the planning CT and MR were deformed to the additional MR scans resulting in total 18 CT-MR and 18 MR- MR DIRs. Agreement of deformed structures with the gold standard manual delineations was evaluated by mean surface distance (MSD) and Dice similarity coefficients (DSC). Differences between mean DSC and MSD of CT-MR and MR- MR deformation, respectively, were tested with paired t- tests for each structure set. Results An example of a rectum structure delineation as well as CT- and MR-based deformation is shown in the figure. In total, 285 structures were deformed successfully, while three bladder structures were not created because the TPS’ inability to deform ring structures. All mean DSC between deformed and manually delineated structures were better for MR-MR than CT-MR DIR, as seen in the table. Also, MSD was 0.6 – 1.3 mm lower for all deformed structures based on MR relative to CT. Overall, CT-MR deformed structures showed lesser DSC and higher MSD than the intra-observer variation. MR-MR deformed structures showed DSC and MSD of similar magnitude to those of intra-observer variations.

Conclusion Intra-modally deformed OAR and target structures for daily plan adaptation in prostate cancer patients matched manually delineated structures more precisely than inter- modal deformations. Agreement of intra-modal deformations corresponded in magnitude to the intra- observer variation and is recommended for clinical use. However, some manual corrections of deformed structures should be expected.