Fractional Variability of High-Risk Clinical Target Volume and Organs-at-Risk during HDR Brachytherapy for Cervical Cancer
Abstract
Purpose
High-dose-rate (HDR) tandem and ring cervical brachytherapy requires significant time and resources from implant, to imaging, planning, and treatment. The high-risk clinical target volume (HR-CTV) and organs at risk (OARs) are contoured on MRI. To save time and resources without compromising plan and treatment quality, this study aimed to retrospectively identify patient subgroups that would benefit from adaptive treatment planning (MR each fraction) versus those who can be safely treated with initial HR-CTV contour without compromising coverage or OAR sparing.
Methods
HR-CTV and OAR (bladder, rectum, and sigmoid) contours from forty patient cases were retrospectively extracted from anonymized DICOM-RT structure sets using 3D Slicer (IRB STUDY00022849). Patients were classified by initial HR-CTV volume into small, medium, and large subgroups. For fractions 1-5, the contour volume changes were quantified using dice similarity coefficient (DSC) and percent volume change from fraction one. Dose parameters evaluated include HR-CTV V100, V90 and D90. Treatment parameters including number of dwell positions and dose to point A were recorded, and any change in parameter from fraction one categorized the fraction as “Replan”. To determine statistical differences a two-sided unpaired t-test was performed for Replan versus No-Replan across target and OAR volume groups.
Results
Inter-fraction HR-CTV volume variability decreased with increasing target size. The small HR-CTV group demonstrated the highest variability, with a mean absolute percentage change of 16.3 ± 12.9%. The HR-CTV dice coefficient was significantly different (p<0.05) between Replan and No-Replan groups for all patients. For small-target patients, there was significant difference in bladder volume between replan and no-replan groups.
Conclusion
Patients with small HR-CTV volumes showed the greatest inter-fraction variability and are most likely to benefit from adaptive treatment planning, whereas large HR-CTV volumes exhibited minimal variability, suggesting that reuse of the initial HR-CTV contour may be feasible.