TH-A-9A-07: Investigation of Localization Errors Using CBCT for SABR-Based Treatment of Lung Cancer and Impact On Planning Margins and Dose Coverage

Abstract

Purpose: In this study, we quantify setup errors and their impact on planning margin and dose coverage in the context of CBCT-based IGRT for SABR of lung tumors. The effect of respiratory excursion on setup up error was also investigated. Methods: The datasets of 150 patients, divided into four groups according to tumor location: upper-chest well seated (CWS), upper-island, lower-CWS, and lower-island, were evaluated. Setup errors and planning margins were calculated and compared. Degradation of dose coverage due to setup errors was investigated for 8 patients. For each patient, 4 CBCT localizing images were used to create internal target volume (ITVCBCT), which was considered a surrogate of respiratory-induced tumor excursion during treatment. Results: The percentage of patients showing >5 mm tumor respiratory excursion for upper-island/lower-island was 10.0/42.9% and for upper-CWS/lower-CWS was 4.2/46.7%. Without image-guidance for patient localization, ∑interfxn and σinterfxn of the inter-fraction setup errors were 2.2–3.9/2.9–5.0 mm and 2.3–4.2/3.0–5.1 mm for upper-island/lower-island and upper-CWS/lower-CWS, respectively. Errors were correlated to tumor excursion. Pearson coefficient for upper-island/lower-island was 0.36/0.60, and for upper-CWS/lower-CWS was 0.20/0.44. With the use of CBCT for daily image guidance, ∑residual and σresidual of the residual setup errors were 0.7–1.2/0.8–1.4 mm and 0.7–1.2/0.8–1.3 mm for upper-island/lower-island and upper-CWS/lower-CWS, respectively. The planning margin vectors (Ant./Post., Sup./Inf., Left/Right) for upper-island/lower-island tumor were (2.7, 2.8, 3.6)/(2.8, 3.6, 4.4) mm, and for upper-CWS/lower-CWS tumors were (2.4, 3.3, 3.3)/(2.8, 2.9, 3.5) mm. Average dose difference (D95) between delivered doses to ITVCBCT and planned doses was 4.5±0.9% and 7.4±1.9% with and without using CBCT for localization, respectively. Conclusion: This large cohort analysis suggests that tumor location may be predictive of tumor excursion and corresponding inter-fraction setup errors. The use of CBCT reduces setup uncertainty substantially. 3–5 mm ITV-to-PTV planning margin would be adequate to compensate for residual setup errors and provide sufficient dosimetric coverage.