Diffusion-weighted MRI in image-guided adaptive brachytherapy: Tumor delineation feasibility study and comparison with GEC-ESTRO guidelines.

PURPOSE: To examine the feasibility of using diffusion-weighted images (DWIs) coregistered with T2-weighted (T2w) sequence in treatment planning system to improve target delineation for image-guided adaptive brachytherapy (IGABT) in locally advanced cervical cancer patients.

METHODS AND MATERIALS: We retrospectively examined the records of consecutive patients who were referred to our institution for pulse-dose-rate IGABT between May and December 2015 after concurrent chemoradiation. An MRI with T2w and DWI sequences with a maximum diffusion factor of 1000 s/mm2 on a 1.5 or 3.0-T MR scanner was performed after placement of the vaginal mold applicator. T2w defined gross tumor volume (GTVT2) and diffusion volume on DWI sequences (CTVDWI) were retrospectively delineated, if achievable, by radiologist and radiation oncologist. Concordances between CTVDWI and GTVT2, high-risk and intermediate-risk clinical target volumes (CTVHR and CTVIR) were evaluated.

RESULTS: Forty-four patients were identified: 23 patients (52%) had a GTVT2 delineated, CTVDWI was delineated in 42 patients (95%). Intraobserver and interobserver conformity indexes were <0.75 in 11 patients (26%) and 23 patients (54%), respectively. There was a positive correlation between GTVT2 and CTVDWI volumes (p = 0.038, r = 0.58). Median CTVHR and CTVDWID90 were 37.3 Gy (17.1-48.9 Gy) and 33 Gy (22-97 Gy), respectively (p = 0.659). CTVDWI could have pointed CTVHR delineation modifications in a total of 32 of 44 (73%) patients with CTVDWI/CTVHR conformity < 1. CTVDWI volume was totally encompassed by CTVIR in all patients.

CONCLUSIONS: These results suggest that DWI images as anatomical sequence without apparent diffusion coefficient mapping might have led to CTVHR modifications. Still, interobserver and intraobserver variations in delineation are substantial, and artifacts make it difficult to implement.