Contouring and Target Variability in Vaginal Tumors Treated with MR-Based Interstitial Brachytherapy: A Multi-institutional Contouring Study

Abstract

The treatment of vaginal tumors can be challenging, often with advanced disease in close proximity to adjacent normal tissue. Interstitial brachytherapy (ISBT) is an effective treatment option for vaginal tumors due to highly conformal dose distributions. With the recent advent of 3D image-based planning in brachytherapy, ISBT is becoming more commonly practiced. As there are currently no consensus guidelines for 3D vaginal brachytherapy planning, we hypothesize that there is a wide variation in target definitions for vaginal tumors and that differences in MR-based contouring practice exist. The goal of this study is to evaluate the variability in practices and target contouring among 15 Canadian interstitial brachytherapy physicians. A contouring study was conducted in 10 academic cancer institutions, with 15 participating radiation oncologists. The study included a short survey and three vaginal ISBT case studies. The cases were comprised of two vaginal squamous cell cancers (post-hysterectomy and intact uterus) and one vaginal recurrence of endometrial cancer. Participants were provided clinical case information, and complete MRI image sets taken prior to external beam radiotherapy, pre-brachytherapy (pre-BT) and at brachytherapy (BT). Participants were to answer brachytherapy planning questions pertaining to each case and to contour on the pre-BT MRI (GTV-Tres) and the BT MRI (GTV-Tres, CTVHR, and CTVIR). The agreement between contours was analyzed with kappa statistics and conformity index (CI). Median ISBT experience among participants was 4 years (1-15). Two institutions plan on MRI images while eight incorporate MRI for contouring. There was agreement among all participants for the general definition of GTV-Tres, while variations were seen in how CTVHR and CTVIR were conceptualized. Recommended doses to the CTVHR ranged from 66 to 90 Gy (EQD2). For all cases, kappa and CI were highest in pre-BT GTV-Tres contours (mean 0.56, 0.44) as compared to BT GTV-Tres contours (mean 0.46, 0.36), indicating a higher level of agreement. Kappa and CI were higher with bulky tumors (mean 0.59, 0.45) as compared to the case featuring small residuum (mean 0.29, 0.21). There was highest level of agreement in CTVIR as compared with GTV-Tres and CTVHR (kappa mean = 0.58 vs 0.46, 0.44). ISBT with 3D planning for vaginal tumors is readily available across Canada. Variation exists in how CTVHR and CTVIR are defined. Contouring seems to be more consistent on pre-BT MRI as compared to BT MRI suggesting an imaging-specific effect from the needles, affecting interpretation at target delineation. Highest variability was seen in the case with small residual disease at the time of BT. Further work is warranted to establish common practices for vaginal tumor brachytherapy.