Toxicity Correlations of Per-Fraction Delivered Organ Dose in An MRI-Guided Prostate Radiotherapy Clinical Trial
J Lamb*, G Ibrahim, Y Abdulkadir, M Cao, T Ma, M Casado, Y Gao, V Basehart, H Wilhalme, K Sheng, Y Yang, D Low, M Steinberg, A Kishan, UCLA School of Medicine, Los Angeles, CA
Presentations
PO-GePV-M-244 (Sunday, 7/10/2022) [Eastern Time (GMT-4)]
ePoster Forums
Purpose: Magnetic image resonance (MRI) guided radiotherapy greatly improves the accuracy of per-fraction delivered dose due to the use of simulation-quality images for daily image guidance. In this work we analyze daily delivered dose in a prostate stereotactic body radiotherapy clinical trial and correlate per-fraction dosimetry with physician-reported toxicity.
Methods: 48 consecutive patients treated on the MRI-guided arm of a single-institutional clinical trial were analyzed. Bladder and rectum were contoured for each daily fraction. A deep-learned based auto-contouring routine was used for initial contours, which were then corrected by an experienced treatment planner. Per-fraction delivered dose was calculated on the daily image. In case of interrupted treatment fractions, deformable registration was performed between the corresponding setup images and dose was deformably mapped to the partial fraction representing the majority of partial fraction dose. Clinical dose-volume histogram (DVH) metrics used for treatment planning and plan evaluation were calculated on the per-fraction DVH. Treatment planning scans were also re-contoured by the same observer. DVH metrics were compared between the treatment plan and the mean per-fraction dosimetry, and both were correlated with physician reported toxicity scores (CTCAE version 4.03).
Results: Correlations between treatment plan and per-fraction delivered DVH metrics ranged from 0.23-0.80 (mean: 0.52). Correlations between treatment plan and mean delivered DVH metrics ranged from 0.42-0.91 (mean: 0.70). A higher degree of correlation was observed for large-volume metrics. Correlations between delivered dose metrics and toxicity ranged from -0.18 to 0.19 (mean: 0.04). Correlations between planned dose metrics were on average slightly higher and ranged from -0.15 to 0.23 (mean: 0.05).
Conclusion: Considerable variation was observed in dose-volume metrics between daily delivered dose and planned dose. More work is needed to identify multi-variate and potentially non-linear predictors of toxicity in this data.
Funding Support, Disclosures, and Conflict of Interest: Authors Cao, Kishan, Lamb, and Low report personal fees from ViewRay, Inc, outside the presented work. Steinberg reports membership on the ViewRay Strategic Advisory Board.
Keywords
Not Applicable / None Entered.
Taxonomy
Not Applicable / None Entered.
Contact Email