Abstract Code: IUC25511-79

Prostate volume changes during longer fractionated and ultra-hypofractionated MR-guided adaptive radiotherapy leveraging the MOMENTUM registry

1. Cooper ¹, H. Akhiat ², S. Alexander ¹, F. Alongi ³, E. Blezer ⁴, A. Choudhury ⁵, J. Christodouleas ², J. Diaz ¹, C. Gani ⁶, S. Hafeez ¹, W. Hall ⁷, P. Jeene ⁸, M. Luzzara ¹, M. Nowee ⁹, T. Schytte ¹⁰, H. Verkooijen ⁹, D. Vesprini ¹¹, J. Van Der Voort Van Zyp ⁴, P. Westhoff ¹², A.C. Tree ¹

(1) Royal Marsden NHS Foundation Trust / Institute of Cancer Research, London – United Kingdom, (2) Elekta, Stockholm – Sweden, (3) IRCCS Ospedale Sacro Cuore-Don Calabria, Cancer Care Center, Negrar Verona – Italy, (4) University Medical Center Utrecht, Utrecht – Netherlands, (5) Division of Cancer Sciences, University of Manchester / The Christie NHS Foundation Trust, Manchester – United Kingdom, (6) Universitätsklinikum Tübingen, Tübingen – Germany, (7) Medical College of Wisconsin / Froedtert Hospital, Milwaukee – United States, (8) Radiotherapiegroep, Deventer/Arnhem – Netherlands, (9) Netherlands Cancer Institute – Antoni van Leeuwenhoek, Amsterdam – Netherlands, (10) Odense University Hospital, Odense – Denmark, (11) Sunnybrook Health Sciences Centre, Toronto – Canada, (12) Radboud Universitair Medisch Centrum, Nijmegen – Netherlands

Purpose/Objectives:

Prostate volume changes during radiotherapy may compromise target coverage [1,2]. We quantified longitudinal swelling across longer fractionation (≥15 fractions)(LFRT) [3] and stereotactic body radiotherapy (SBRT) regimes during MR-guided adaptive radiotherapy (MRgART) to identify predictors of clinically significant swelling (CSS, defined as ≥15% volume increase) [4,5].

Materials/Methods:

This multicentre prospective cohort study analysed patients treated on an MR-Linac between January 2020-April 2025 enrolled in MOMENTUM [6]. Inclusion mandated daily recontouring/adaptation with ≥3 fractions available. Structure volumes were extracted using PyRadiomics (v3.0.1). A hierarchical fuzzy-matching algorithm (Levenshtein distance) harmonised prostate volume nomenclature across centres. Clinical and demographic variables were collected, and fraction 1 counted as baseline volume [7]. Proportion of treatment spent with CSS was normalised to number of fractions. Baseline prostate volume was dichotomised. Linear mixed-effects models (LMM) quantified continuous volume trajectories by cohort. Multivariable logistic regression tested predictors of CSS as a binary outcome.

Results:

From 1907 patients, 222 LFRT and 1390 SBRT patients met eligibility. A total of 10297 fractions were analysed (median 20 for LFRT, 5 for SBRT patients), with median baseline prostate volume of 40.0 cm³ (IQR 28.9-53.6) for LFRT and 47.5 cm³ (IQR 35.8-66.0) for SBRT. Incomplete age data excluded 160 patients from LMM. CSS occurred in 15.2% LFRT and 13.1% SBRT total fractions (figure 1), but SBRT patients spend a higher proportion of treatment in CSS (figure 2). In LMM, larger baseline prostate volume was associated with reduced mean swelling in SBRT (β=-0.030% per cm³, p<0.001) but not LFRT (p=0.30). However, logistic regression showed larger baseline volume was protective in both cohorts (LFRT: OR 0.38, 95% CI 0.32-0.46; SBRT: OR 0.80, 95% CI 0.69-0.92). In secondary complete-case analysis including ADT and T-stage data (n=116 LFRT, n=256 SBRT), no significant association with ADT was observed in either cohort (LFRT p=0.65; SBRT p=0.37), whilst T2-stage disease was associated with increased swelling in SBRT (β=+3.0%, p=0.018) compared to T1-stage.

Conclusion:

SBRT patients experience a faster onset of prostate swelling compared to LFRT. Smaller baseline volumes and T2-stage was associated with increased swelling risk. Incomplete clinical covariate data limits definitive risk stratification, supporting continued fraction-by-fraction adaptive workflows.

Figures