Four-dimensional computed tomography prediction of inter- and intrafractional upper gastrointestinal tumor motion during fractionated stereotactic body radiation therapy.

Purpose: Respiratory-induced tumor motion of upper gastrointestinal (GI) tumors during radiation therapy is often assessed using a single 4-dimensional computed tomography (4D-CT) and presumed to be representative during fractionated treatment regimens. The purpose of this study was to examine the intra- and interfraction correlations of tumor motion between pretreatment 4D-CT and real-time fiducial-based motion tracking in patients treated with fractionated stereotactic body radiation therapy (SBRT) for upper GI malignancies.

Methods And Materials: Fourteen patients with upper GI tumors underwent fractionated SBRT using the CyberKnife radiosurgical system with Synchrony respiratory motion management. Before treatment, each patient underwent a free-breathing 4D-CT scan and fiducial motion was tracked for each phase of the respiratory cycle. Real-time fiducial positions recorded during delivery of each SBRT fraction were extracted from the CyberKnife planning system. Displacements were compared between those predicted by 4D-CT and those recorded by Synchrony during treatment in the left-right (LR), anteroposterior (AP), and superoinferior (SI) directions.

Results: The 4D-CT scans demonstrated little correlation with real-time mean fiducial displacement as determined by Pearson correlation with coefficients of 0.45, 0.52, and 0.63 in the SI, AP, and LR directions, respectively. Cohort-averaged maximum fiducial displacements based on 4D-CT and real-time tracking were measured to be 3.86 ± 1.40 mm versus 10.73 ± 7.03 mm, 2.29 ± 1.02 mm versus 4.44 ± 3.33 mm, and 1.45 ± 0.49 mm versus 2.67 ± 2.49 mm in the SI, AP, and LR directions, respectively. Mean fiducial displacements were greater than that predicted by the maximum displacements on the corresponding 4D-CT scan in 39%, 22%, and 25% of SBRT fractions in the SI, AP, and LR directions, respectively.

Conclusions: Comparison of 4D-CT with real-time fiducial tracking demonstrated significant inter- and intrafractional discrepancies, particularly in the SI direction, which could result in compromise of target coverage when planning with a single free-breathing 4D-CT.