Analysis of CBCT-based image guidance for a large cohort of lung cancer patients treated with SABR

Abstract

This study evaluates respiratory excursions and their impact on patient setup and dosimetric coverage with and without the use of cone-beam computed tomography (CBCT) for localization in stereotactic ablative radiotherapy. The datasets of 150 non-small cell lung cancer patients were assessed. Four groups of patients were evaluated based on their tumor location: upper lobe (UL)-peripheral (N = 39), UL-chest-wall seated (CWS, N = 37), lower lobe (LL)-peripheral (N = 48), and LL-CWS (N = 26). Tumor excursion and setup error were quantified and correlated. Treatment planning margins were derived based on the van Herk formalism. A dosimetric study investigated the dose coverage with and without the use of CBCT for localization. The percentage of patients showing >5 mm tumor respiratory excursion for UL-peripheral/LL-peripheral was 10.0/42.9%, and was 4.2/46.7% for UL-CWS/LL-CWS. Planning margin magnitudes averaged over all patients were Mresidual(2.7, 3.2, 3.7) mm and Minterfxn(9.0, 14.0, 10.0) mm, respectively with and without the use of CBCT for image guidance. Comparatively, the planning margins for the patients with LL tumors exhibiting the largest motion and setup errors were MresidualLL(2.8, 3.6, 4.4). Pearson correlation coefficients (for LL-peripheral tumors in the S/I direction) between tumor excursion and, respectively, inter-fraction and residual setup errors were 0.62 and 0.32. Based on skin tattoo setup, the overall average difference in D95 dose to the planning target volume (PTV) between the delivered and planned doses was 14.1 ± 9.2%. The use of CBCT for localization reduced the overall average ΔD95 to less than 2%. This analysis is suggestive that: (a) patients with LL tumors undergo the largest respiratory-induced motion, and experience larger setup errors relative to UL tumors; (b) the use of CBCT-based image guidance significantly reduces residual setup errors; planning margins of the order of 5 mm appear to be adequate for proper PTV dose coverage; (c) CBCT image guidance reduces the correlation between respiratory-induced motion and setup errors, implying that there is much less variation in the setup uncertainty between tumors undergoing respiratory motion of varying magnitudes, relative to the variation without CBCT.