Intrafractional Motion of Target Volumes and Organs at Risk Due to Bladder Filling: Implications for MR-Only Prostate Radiation Therapy

Abstract

As magnetic resonance simulation (MR-SIM) emerges as a primary treatment planning modality for prostate cancer, it becomes critical to quantify the uncertainties introduced in an MR-only workflow. Given the long image acquisition times encountered in MRI, we sought to characterize the temporal and spatial intra-fractional changes between the prostate, seminal vesicles (SVs), and other organs at risk in response to bladder filling conditions. Serial T2-weighted MR-SIM data (3-6 timepoints/subject at 1.0T, 1.5T, or 3.0T, 38 evaluable timepoints) were acquired in 10 immobilized subjects using a fixed bladder filling protocol (bladder void, 20 oz water consumed, 10 oz consumed mid-session). Delineations were performed by one physician including prostate, SVs, and organs at risk (bladder, rectum, penile bulb) to yield 370 evaluable contours. Bony rigid registration (3 degrees of freedom) was performed for serial T2 data to isolate local organ effects. To quantify the impact of systematic bladder filling, temporal assessments of the volumes, center of mass, and Dice similarity coefficients (DSCs) were performed. To assess serial target displacements, centroid differences between the prostate and proximal SVs were compared across timepoints. Subjects had an average bladder volume difference of 62 ± 15% between initial and final timepoints. When compared to the initial prostate volume, mean DSCs of the prostate went from 0.81 ± 0.07 at the middle timepoint to 0.75 ± 0.07 at the final. A large reduction in DSC was observed for the proximal SVs (0.44 ± 0.22 at the middle time point vs. 0.32 ± 0.17 for the final). Overall, the median vector displacement of the penile bulb was 2.2 mm. After excluding a subject with an extreme case of rectal gas, the average rectum volume difference from the initial timepoint was 5.9 ± 10.3%. The prostate showed <1 mm lateral displacement and > 3mm in centroid shift in 6/10 cases (4 posteriorly and 2 anteriorly due to changes in rectal status). The SVs shifted posteriorly with bladder filling (7/10 cases were > 3mm). The centroid differences between the prostate and proximal SVs were >2mm in 4/10, 6/10, and 7/10 of subjects in the left-right, anterior-posterior, and superior-inferior directions, respectively, suggesting that they move independently with bladder status. As bladder filling increased, DSC decreased for the prostate, SVs, and OARs. In addition, large, systematic shifts were observed in the prostate and SVs (particularly anterior/posterior), although these were subject dependent due to confounders such as rectal status. This suggests that care must be taken during patient preparation to minimize these effects. Future work will incorporate dosimetric consequences using MR-only treatment planning in response to non-uniform organ displacement.