Geometric and Clinical Factors Affecting Active Fiducial Tracking in Robotic Stereotactic Body Radiation Therapy in the Lung

Abstract

The movement of lung tumors during the breathing cycle presents a challenge in active tumor and fiducial tracking in the robotic stereotactic body radiation therapy (SABR). Robotic SABR manufacturer guidelines for fiducial placement rarely allows 6D tracking of lung targets, which requires a minimum of 3 tracked fiducials. Advances in navigational bronchoscopy are allowing more controlled fiducial placement. This review of 4D geometric and clinical factors affecting the number of fiducials tracked can potentially optimize fiducial placement. Patients treated at our institution with robotic SABR for lung tumors where error data was available were identified, and their clinical characteristics collected. When available, the fiducials were contoured in each of 10 phases on 4D CT using HU thresholds, and the centroids of the objects calculated in Matlab. The number of fiducials tracked, and the errors in the breathing model that prevented at least 3 fiducials from being tracked were recorded. Clinical correlations were determined using Chi-square tests, comparison of maximum and minimum inter-fiducial distances and errors by Mann-Whitney U, and local control/survival by Kaplan-Meier analysis. 76 patients with 83 treated lesions were identified, 48 patients with available PFTs, 28 patients with fiducial centroid data. Lung primaries and oligometastatic disease were included. Overall survival and local control were not affected by the number of tracked fiducials (p = 0.20, p = 0.79) with 2.5 year follow up. COPD was associated with higher likelihood of having > = 3 fiducials tracked (p = 0.034). Elements of the PFTs, smoking history, and prior lung surgery were not correlated with > = 3 fiducials tracked. The errors preventing at least 3 fiducials to be tracked included rigid body errors (60.2%), spacing (37.3%), uncertainty (15.7%), fiducial shadowing (6.0%), and high collinearity (1.2%). Tumors in the lower lobes had higher rates of uncertainty errors (p = 0.015), especially the left lower lobe (p = 0.007). High maximum inter-fiducial spacing was correlated with rigid body errors (p = 0.022), with incidence rising with fiducial spacing above 3 cm. Low minimum inter-fiducial distance was correlated with spacing errors (p = 0.017). Tumor control and survival were not affected by having <3 fiducials tracked. The increased uncertainty errors in the lower lobes could be due to increased motion, and on the left lower lobe due from superimposed cardiac motion. Fiducials may be more easily tracked in patients with COPD due to the loss of lung elasticity. The expansile nature of lung tissue could increase rigid body errors with increased inter-fiducial distance. Further study is needed to determine how fiducial placement can be optimized to improved tracking and influence how clinicians determine the expansions used to create PTVs.