Method to assess the need for re-planning HDR brachytherapy tandem and ring treatments

High dose rate (HDR) brachytherapy procedures for cervical cancer require multiple applicator insertions for multiple (typically 5) fractions of a single plan, which carries a risk for variability in applicator position between fractions. Due to applicator displacement relative to patient anatomy, the dose to nearby organs-at-risk (OARs) may vary significantly from one fraction to the next. The purpose of this study was to evaluate the effect of changes in HDR tandem and ring (T&R) applicator position on doses to nearby OARs and to present a quick and simple method to estimate doses to OARs inter-fractionally without having to perform a re-plan. Ninety CT image sets for 20 patients, ages 44 to 86, undergoing T&R-based HDR for cervical cancer were used retrospectively for this study. Measures of applicator positional and angular changes relative to the bony anatomy were obtained using image fusion in MIM software, between the planning CT (plan CT) and the CT on the treatment day (CT-TX). Dosimetric data were determined, also using MIM software, using the original (first fraction) dose distribution applied to organs at risk (rectum and bladder), transferred via rigid registration from the plan CT to each CT-TX. Bladder and rectum contours were also transferred from each plan CT to each CT-TX and were tweaked manually to match anatomy on each CT-TX and examined visually for appropriateness. Differences in translation and rotation of the T&R applicator between the planning CT and subsequent individual fractions were recorded and plotted against dose differences between each fraction of treatment and the original (first) fraction. Absolute dose (D2cc) and volume (V50) differences vs positional shifts were calculated and plotted, and the Pearson Product-Moment correlation coefficient between dose parameters and measured positional shifts was determined. Average dosimetric differences between planned dose and subsequent fractional doses obtained through rigid registration were 1.48 ± 1.92 Gy, 14.91 ± 11.92 cm3, 0.56 ± 0.93 Gy, and 1.77 ± 2.18 cm3 for Bladder D2cc, Bladder V50, Rectum D2cc, and Rectum V50, respectively. Correlation between Bladder V50 and sagittal plane rotation gave an r2 of 0.4, showing the most correlation of all parameters studied. Bladder dose and volume increased by a maximum of about 2.7 Gy and 50 cm3 overall for Bladder D2cc and Bladder V50, respectively. Bladder V50 was most sensitive to T&R applicator displacements. We have quantified th