2359 - Evaluation of Cumulative Doses to Organs at Risk during Lung Reirradiation

Purpose/Objective(s): Limited data exists to guide Radiation Oncologists on optimal planning strategies during lung reirradiation. Rigid image registrations based on bony anatomy are often used to calculate cumulative radiation doses but may not provide the most accurate results due to changes in lung anatomy between treatments. In this study we investigate multiple image registration techniques and analyze differences found in cumulative doses to organs at risk (OARs).

Materials/Methods: Twenty-five patients treated with at least two courses of ipsilateral lung reirradiation were evaluated. At least one course of SBRT was used in 22/25 patients, conventional fractionation was used in 10/25. Typical fractionations were 48 Gy in 4 fx and 66 Gy in 33 fx. Rigid image registrations (RIR) were created for spine and carina (MIM software Inc.) followed by a deformable image registration (DIR), and a hybrid DIR using lung lobe contours (DIR w/ lobes) based on each RIR. Geometric differences between registrations were analyzed based on manually defined landmarks (29-30 per patient). Registrations resulting in the smallest overall landmark error were considered the most accurate. Cumulative BED doses were calculated for Esophagus, Heart, Proximal Bronchial Tree (PBT), Spinal Cord (Cord), and Great Vessels (GV). For each OAR, the most accurate registration (A) was compared to the corresponding DIR (C), e.g., Carina DIR w/ lobes to Carina DIR w/o lobes, and then to Spine RIR (S). A Wilcoxon Rank Sum test compared A vs C to A vs S.

Results: DIR (both w/ and w/o lobes) was the most accurate in 24/25 (p<0.001) patients. DIR w/ lobes was the most accurate in 17/25 (p=0.0046). For D0.035cc, Heart showed the largest mean BED difference (>7 Gy) and Cord the smallest (see table). For volumetric metrics, Heart showed the largest dose differences (> 6 Gy), Esophagus the smallest. RIR to the spine resulted in larger dose differences, except for Cord D0.035cc and Esophagus D5cc. Differences were significant between A vs C and A vs S for Esophagus D5cc (p=0.029), Heart D0.035cc (p=0.033), PBT D5cc (p=0.0004), and GV D10cc (p=0.015).

Conclusion: DIR resulted in reduced cumulative dose uncertainty compared to RIR in this cohort of lung reirradiation. Large dose differences were observed particularly for Heart and PBT. Using DIR and including lung lobes in the registration process may reduce dose uncertainty during reirradiation planning. A better understanding of dose uncertainty will support establishing dose-effect relationships for reirradiation.