An improved method for analyzing and reporting patterns of in-field recurrence after stereotactic ablative radiotherapy in early-stage non-small cell lung cancer

Introduction: Patterns of local, regional, and distant failure after stereotactic ablative radiotherapy (SABR) for early-stage non-small cell lung cancer (NSCLC) have been widely reported. However, reliable methods for analyzing causes of local failure are lacking. We describe a method for analyzing and reporting patterns of in-field recurrence after SABR, incorporating dosimetric parameters from initial treatment plan as well as geometric information from diagnostic images at recurrence. Material and methods: Diagnostic CT images at recurrence were registered with initial treatment planning images and radiation dose by deformable image registration. Recurrent gross tumor volume (rGTV) and centroid (geometric center of rGTV) were delineated. In-field failure was classified as centroids originating within the original planning target volume. Dose-volume histograms for each rGTV were used to further classify in-field recurrences as central high-dose (dose to 95% of rGTV [rGTVD95%] ≥95% of dose prescribed to PTV) or peripheral high-dose (rGTVD95% <95% of dose prescribed to PTV). Results: 634 patients received SABR from 2004 to 2014 with 48 local recurrences. 35 of these had evaluable images with 16 in-field recurrences: 9 central high-dose, 6 peripheral high-dose, and 1 had both. Time to and volume of recurrence were not statistically different between central versus peripheral high-dose recurrences. However mean rGTV dose, mean centroid dose, and rGTVD95% were higher for central versus peripheral high-dose recurrences. Conclusion: We report a standardized method for analysis and classification of in-field recurrence after SABR. There were more central as opposed to peripheral high-dose recurrences, suggesting biological rather than technical issues underlying majority of in-field failures.