Intra-adaptational motion during online adaptive radiotherapy: analysis of couch shifts and implications for margin design.

Abstract

PurposeOnline adaptive radiotherapy (oART) minimizes inter-fractional variation, yet intra-fractional motion (IFM) remains a source of uncertainty. We quantified intra-adaptational motion (IAM) using routinely recorded couch shifts between pre-adaptation (CBCT1) and pre-treatment (CBCT2) scans to derive motion-related components of the CTV-PTV margins in a large clinical cohort.MethodsWe retrospectively analyzed 1378 fractions from 69 prostate cancer patients treated with 60 Gy in 20 fractions as oART on the Varian Ethos platform (vision 1.1; Varian Medical Systems, Palo Alto, CA, USA). Couch shifts were calculated based on rigid registration targeting the CTV. Systematic and random errors were calculated for right-left (RL), anterior-posterior (AP), and superior-inferior (SI) directions. Margins were estimated using the van Herk formula. Correlations between couch shifts and the CBCT1-CBCT2 time interval were assessed using Spearman's ρ.ResultsCouch shifts showed substantial inter-patient variability. Population mean, systematic, and random errors ranged from 0.05 to 1.4 mm, 0.6 to 1.3 mm, and 1.0 to 1.4 mm, respectively. The mean adaptation time was 17.5 ± 5.1 min (median: 16.6 min, interquartile range [IQR]: 14.0-20.3 min). The resulting IAM-related margins were 2.3 (R), 2.2 (L), 2.2 (S), 3.2 (I), 2.9 (A), and 5.7 mm (P). Positive correlations were observed between the time interval and motion in AP (ρ = 0.26), SI (ρ = 0.30), and vector magnitude (ρ = 0.29; all p ConclusionIntra-adaptational motion varies substantially between patients, supporting the consideration of patient-specific margins. Verification CBCT2 is essential in oART due to potentially large positional shifts occurring during the adaptation process. The calculated margins represent upper bounds for motion-related margins, as they incorporate registration errors and cover longer intervals than actual beam delivery.