KEY POINTS
- The Japanese DeSC claims-database study included 7,860 patients starting primary definitive prostate radiotherapy between July 2018 and December 2024. Hydrogel spacers were used in 1,754 patients (22.3%).
- Treatments included three-dimensional conformal radiotherapy, intensity-modulated radiotherapy, stereotactic body radiotherapy, particle therapy and brachytherapy. The median follow-up was 19.1 months.
- Propensity-score weighting estimated the average treatment effect among spacer-treated patients. After weighting, the effective non-spacer population was 1,878.2, although residual imbalance remained for facility type.
- The primary endpoint was gastrointestinal endoscopic hemostasis, used as a claims-based surrogate for clinically significant bleeding proctitis. Spacer placement reduced its risk by 74%: subdistribution hazard ratio 0.26 (95% confidence interval, 0.10–0.69; p = 0.007).
- At 36 months, cumulative incidence of endoscopic hemostasis was 0.8% with a spacer versus 2.3% without one, an absolute reduction of 1.5 percentage points.
- Diagnosed radiation proctitis was also less frequent with a spacer: subdistribution hazard ratio 0.29 (95% confidence interval, 0.16–0.53; p < 0.001). Thirty-six-month incidence was 1.4% versus 4.8%.
- The association remained directionally consistent across evaluated clinical subgroups, including patients receiving anticoagulants and those treated with particle therapy, although some subgroup analyses contained few events.
- The database contained no tumour stage, prostate-specific antigen, radiotherapy dose, detailed fractionation, physician-graded toxicity or spacer-placement complications. Ultra-hypofractionated treatment was uncommon, and unmeasured confounding remained possible.
CLINICAL TAKEAWAY
The analysis supports a clinically meaningful reduction in rectal bleeding and diagnosed proctitis with hydrogel spacing even in the modern image-guided radiotherapy era. It does not establish that every patient needs a spacer: absolute event rates were low, allocation was non-randomized and the database could not measure procedure-related harm, patient-reported bowel function or cost-effectiveness.