Session Item

Cardiac arrhythmia is a recognised potential complication of thoracic radiotherapy (RT) but the cardiac substructures responsible for arrhythmogenesis in this setting have not been identified. Arrhythmogenic tissue is commonly located in the pulmonary veins (PVs) of cardiology patients with atrial fibrillation (AF), however these structures are not currently considered organs-at-risk during RT planning. Pre-clinical and clinical research recently indicates that the PVs can be isolated with SABR, akin to invasive cardiac ablation, yet a standardised approach for delineation has not been published. Herein we developed and evaluated a PV atlas for RT planning.

The gross and radiological anatomy relevant to AF was derived from the cardiology and radiology literature by a team of radiation oncologists, electrophysiology cardiologists and thoracic radiologists. A region of interest and contouring instructions for 4-dimensional CT RT planning scans were iteratively developed in-house, with input from external collaborators. Given that the myocardial sleeve tissue, which is located within a short central portion of the PV walls, is the source of AF in cardiology, this was identified as the region of interest for RT treatment planning purposes. To evaluate the application of this atlas, radiation oncologists (n=5) and radiation technologists (n=2) contoured the PVs on the 4D planning scans of five patients with locally advanced lung cancer. Interobserver variation in the segmentation was then assessed by comparing the manual contours to reference contours agreed by the lead researchers, using geometric and dosimetric parameters.

Using the atlas designed the mean dose to the PVs in the five datasets for evaluation was 35% of the prescription dose and the mean Dmax was 61%. Geometric and dosimetric similarity of the observer contours with reference contours was good when the atlas was applied, with an overall mean Dice of 0.80 ± 0.02. The right superior PV (mean DSC 0.83 ± 0.02) had better overlap than the left superior PV (mean DSC 0.80 ± 0.03), but the inferior PVs were equivalent (mean DSC of 0.78). The mean magnitude of directional shifts was 1.8mm and shifts in the right-left axis were equally balanced in direction. Shifts in the cranio-caudal and dorso-ventral axes were in the one direction only however. Hausdorff distances were also low generally. The mean difference in mean dose was 0.79Gy (1.46%) and the overall mean Dmax difference was 0.28Gy (0.58%).

A PV atlas with multidisciplinary approval led to reproducible delineation for RT planning, with Dice values comparable to those for other cardiac substructure atlases. Clinical dosimetric studies of consistently contoured PVs could facilitate the derivation of safe dose thresholds below which RT-related arrythmia is unlikely. The atlas outlined may also have applications in the evolving field of SABR for cardiac arrhythmia in providing guidance for consistent identification of the PVs.