Estimation of Duration-Dependent Margins for Prostate Radiotherapy

Abstract

Despite daily pre-treatment localisation and online corrections, intra-fraction motion uncertainties during prostate radiotherapy are expected. The magnitude of motion uncertainty is associated with the duration of the patient lying on the in-room treatment couch. This study reports a minute-by-minute association and calculates the impact of this motion on duration-dependent margins using real-time intra-fraction motion monitored by four-dimensional transperineal ultrasound (4D TPUS). A total of 55 patients were recruited prospectively. Intra-fraction motion of the prostate was monitored real-time using a 4D TPUS system. A total of 1745 monitoring sessions were analysed. Van Herk’s margin recipe (2.5∑+1.64((σ2 + σp2)1/2 - σp)) was used to estimate the duration-dependent margins for every minute, up to the 15th minute. The systematic component (∑) was derived from the standard deviation (SD) of the individual mean intra-fraction motion based on the average position across the entire imaging/treatment phase. Random error (σ) was calculated to include the root sum square of two components: 1) the root mean square (RMS) of the SD calculated from the mean intra-fraction motion for each patient; and 2) RMS of the average of the SD associated to the intra-fraction motion within each fraction for each patient. Linear regression analysis was then performed on the overall margins against time and direction. The mean intra-fraction motion was 0.8mm Inf, 0mm lateral and 0.9mm Post at the 15th minute. A minimum margin expansion of 2.4mm (Sup/Inf), 1mm (Lt/Rt) and 2.7mm (Ant/Post) was required for an 8-minute treatment compared to 4.1mm (Sup/Inf), 1.9mm (Lt/Rt) and 4.5mm (Ant/Post) for a 15-minute treatment. The required margin expansion increased linearly (R2 = 0.99) in all directions (p<0.01). However, while there was no statistical significant difference (p=0.09) in the required margin expansion in the Sup/Inf and Ant/Post directions respective of the time duration, the margins were much bigger compared to those in the Lt/Rt direction (p<0.01). We report our first experience in deriving the minimum duration-dependent margin expansion to generate the required planning target volume for prostate radiotherapy. The required margin expansion increases linearly in all directions within a 15-minute duration.