Comparing Maximum Dose to Central Airways in Mid-Ventilation Versus Average Intensity Projection-Based 4DCT Planning for Lung Stereotactic Body Radiation Therapy (SBRT)

Abstract

Four-Dimensional Computed Tomography (4DCT) is typically used in radiation treatment planning for improved targeting of intrathoracic tumors. While tumor definition uses all in- and expiration image sets to account for the full range of respiratory motion, the definition of organs at risk (OARs) is less well defined. We investigated OAR dose differences for two commonly used image sets for SBRT dose calculation: average intensity projection (AIP) and mid-point in ventilation (midV). Determining accurate OAR doses is particularly important for proximal airways, as RTOG 0813 results have shown that increased dose to these structures can result in unacceptable, at times fatal toxicity. Fifteen 4DCT-based lung SBRT plans originally generated on 30% MidV image sets were transferred to associated AIP images. Fractionation schedules were 50 Gy in 5 fractions in 5, 48 Gy in 4 fractions in 9, and 40 Gy in 5 fractions in 1 plan. Internal gross tumor volumes (iGTV) were delineated on maximum intensity projection (MIP) images. OARs were contoured per RTOG 0813 protocol on both MidV and AIP datasets. Maximum doses were reported for each OAR (trachea, proximal bronchial tree (PBT), heart, and great vessels (GV)) on original MidV plans and after plan transfer to AIP image sets with field arrangements and monitor unit (MU) values unchanged. AIP volumes were, on average, greater by 4.26 ± 5.44 cc for trachea and 13.11 ± 12.21 for PBT, and reduced by 21.6 ± 70.42 for heart and 6.06 ± 41.14 cc for GV. Average superior-inferior movement of tumor and carina were 7.1 ± 4.1 and 7.6 ± 4.7 mm, respectively. The average absolute difference between AIP and MidV in maximum doses to trachea, PBT, heart, and GV were 1.01 ± 1.37, 1.68 ± 3.10, 1.45 ± 2.05, and 1.07 ± 1.21 Gy, respectively. Maximum dose to PBT was increased in 3 of 15 (20%) plans after transfer to AIP image sets, with an average increase of 3.03 ± 2.97 Gy (0.09 to 7.10). Of the remaining 12 plans (80%), this was decreased by an average of 1.35 ± 3.04 Gy (0.01 to 11.31). Fourteen (93%) plans exhibited an absolute difference in maximum dose of > 1 Gy to at least one OAR. Absolute difference in maximum dose to heart showed a positive correlation with carina motion (R2 = 0.61), and GV a weak positive correlation with tumor motion (R2 = 0.40). Our results show a difference in maximum dose to OARs between AIP and MidV image sets. These differences are confounded by tumor and carina motion and associated differences in contoured volumes. Although this study does not allow to identify the better image set, AIP images are more likely to account for a true positional average, and AIP-based plans are expected to be more representative of the true maximum OAR dose. To calculate correct OAR doses, both tumor and OARs require contouring and planning on all respiratory phases which might be necessary for critical central structures such as PBT, if they are in the high dose area.