Change in Image-Guided Planning Strategies Over Time Impacts Global Outcomes for Intracavitary Cervical Cancer Brachytherapy.

Abstract

PURPOSE/OBJECTIVE(S)\nIntracavitary cervical brachytherapy (BT) has transitioned from a two-dimensional non-volumetric (NV) dosimetry system to three-dimensional (3D) CT and/or MRI-based planning techniques. The purpose of this study is to retrospectively evaluate the relative improvements in image-guided planning strategies over time with regards to dosimetry, survival, and acute toxicity to address the hypothesis that modern image-guidance improves outcomes because of accurate target definition.\n\n\nMATERIALS/METHODS\nNinety-five locally advanced cervical cancer (LACC) patients treated with concurrent chemoradiation and high dose rate (HDR) BT from 2009-2016 at a single academic institution were retrospectively divided into three BT planning groups: point-A based NV dosimetry using CT imaging (2008 - 2011, n\u202f=\u202f37), CT-based volumetric dosimetry (2012 - 2014, n\u202f=\u202f33), and MRI-based volumetric dosimetry (2015 - 2016, n\u202f=\u202f25). To compare the dosimetry data of the 3 cohorts, the NV cohort had organs at risk (OAR) and high-risk clinical target volumes (HR_CTV) contoured. Overall survival (OS), progression free survival (PFS), and local control (LC) at 5 years were plotted using Kaplan-Meier curves for the 3 cohorts. To identify potential predictor variables (age, FIGO stage, HR CTV volume, HR_CTV D90 (dose to 90% of the HR CTV) and image planning technique) on outcomes, we used the univariate (UVA) and multivariate (MVA) cox proportional-hazards model, with significant hazard ratios (HZ) reported. Finally, acute grade 3-4 toxicities for all patients were compared using a 2-sided Pearson χ2 test.\n\n\nRESULTS\nFor all patients, the mean BT HR CTV D90 was significantly lower for MRI (P\u202f=\u202f0.015) and CT (P < 0.001) compared to NV (point A prescriptions were higher for the NV cohort by physician choice), with no significant difference between MRI and CT (P\u202f=\u202f0.231). However, for patients with larger tumors (HR CTV ≥ 30 cc), the mean HR CTV D90 was higher for CT (41.6 Gy, P\u202f=\u202f0.473) and MRI (41.6 Gy, P\u202f=\u202f0.256) based-planning compared to that of NV (38.2 Gy). Both MRI and CT had significantly less D2cc to bowel (P < 0.001) and sigmoid (P < 0.001) compared to NV-based planning. Interestingly, CT had significantly less D2cc to bowel compared to MRI-based planning (P\u202f=\u202f0.002). On MVA, age (< 60 vs ≥ 60 yrs) was significant for worse 5-year OS (HZ: 2.48) and LC (HZ: 5.25); MRI, with NV as the reference, had significantly improved 5-year OS (HZ: 0.26), PFS (HZ: 0.34) and LC (HZ: 0.16), while the CT cohort was only significant for improved LC (HZ: 0.29). There was no significant difference in acute grades 3-4 toxicities, but MRI did trend towards improved acute grade 3-4 gastrointestinal toxicity compared to NV (P\u202f=\u202f0.055).\n\n\nCONCLUSION\nCT and MRI-based 3D planning had significantly less D2cc to bowel and sigmoid. MRI-based planning had significant improvement in 5-year OS, PFS, and LC compared to NV on MVA, while CT-based 3D planning had improved LC. Modern image-guidance in this time-based series improves global patient outcomes in LACC.\n\n\nAUTHOR DISCLOSURE\nP. D Cunha: None. Y. Gonzalez: None. C.R. Nwachukwu: None. B.A. Hrycushko: None. P.M. Medin: None. T. Banks: Develop/vet educational website content; AAPM.A. Owrangi: None. X. Jia: None. K.V. Albuquerque: Research Grant; Astra Zeneca. Honoraria; ACR, ARRT. Travel Expenses; ACR, ARRT, ASCO.