Interim Toxicity Analysis From the Randomized HERMES Trial of 2- and 5-Fraction Magnetic Resonance Imaging–Guided Adaptive Prostate Radiation Therapy

Background Ultra-hypofractionated radiotherapy (UHRT) is an effective treatment for localised prostate cancer with an acceptable toxicity profile; boosting the visible intraprostatic tumour has been shown to improve biochemical disease-free survival with no significant impact on genitourinary and gastrointestinal toxicity1,2. Methods HERMES is a single centre non-comparative randomised phase II trial in men with intermediate or lower high risk prostate cancer. Patients were allocated (1:1) to 36.25 Gray (Gy) in 5 fractions over 2 weeks or 24Gy in 2 fractions over 8 days with an integrated boost to the magnetic resonance imaging (MRI) visible tumour of 27 Gy in two fractions. A minimisation algorithm with a random element with risk group as a balancing factor was used for participant randomisation. Treatment was delivered on the Unity MR-Linac (Elekta AB, Stockholm) with daily online adaption. The primary end point was acute genitourinary (GU) Common Terminology Criteria for Adverse Events version 5.0 (CTCAE v5.0) toxicity with the aim of excluding a doubling of the rate of acute GU 2+ toxicity seen in PACE. Analysis was by treatment received and included all participants who received at least one fraction of study treatment. This interim analysis was pre-specified (stage 1 of a 2-stage Simon design) for when 10 participants in each treatment group had completed the acute toxicity monitoring period (12 weeks after radiotherapy). Findings Acute grade 2 genitourinary toxicity was reported in 1 (10%) patient in the 5-fraction group and 2 (20%) patients in the 2-fraction group. No grade 3 or higher genitourinary toxicities were reported. Interpretation At this interim analysis, the rate of genitourinary toxicity in the 2-fraction and 5-fraction treatment groups was found to be below the pre-specified threshold (5/10 grade 2 or higher) and continuation of the study to complete recruitment of 23 participants per group was recommended.