Cone-beam CT-based inter-fraction localization errors for tumors in the pelvic region

PURPOSE: To evaluate inter-fraction tumor localization errors (TE) in the RapidArc® treatment of pelvic cancers based on CBCT. Appropriate CTV-to PTV margins in a non-IGRT scenario have been proposed. METHODS: Data of 928 patients with prostate, gynecological, and rectum/anal canal cancers were retrospectively analyzed to determine systematic and random localization errors. Two protocols were used: daily online IGRT (d-IGRT) and weekly IGRT. The latter consisted in acquiring a CBCT for the first 3 fractions and subsequently once a week. TE for patients who underwent d-IGRT protocol were calculated using either all CBCTs or the first 3. RESULTS: The systematic (and random) TE in the AP, LL, and SI direction were: for prostate bed 2.7(3.2), 2.3(2.8) and 1.9(2.2) mm; for prostate 4.2(3.1), 2.9(2.8) and 2.3(2.2) mm; for gynecological 3.0(3.6), 2.4(2.7) and 2.3(2.5) mm; for rectum 2.8(2.8), 2.4(2.8) and 2.3(2.5) mm; for anal canal 3.1(3.3), 2.1(2.5) and 2.2(2.7) mm. CTV-to-PTV margins determined from all CBCTs were 14 mm in the AP, 10 mm in the LL and 9-9.5 mm in the SI directions for the prostate and the gynecological groups and 9.5-10.5 mm in AP, 9 mm in LL and 8-10 mm in the SI direction for the prostate bed and the rectum/anal canal groups. If assessed on the basis of the first 3 CBCTs, the calculated CTV-to-PTV margins were slightly larger. CONCLUSIONS: without IGRT, large CTV-to-PTV margins up to 15 mm are required to account for inter-fraction tumor localization errors. Daily IGRT should be used for all hypo-fractionated treatments to reduce margins and avoid increased toxicity to critical organs.