Prospective development of an individualised predictive model for treatment coverage using offline cone beam computed tomography surrogate measures in post-prostatectomy radiotherapy

The aim of this study is to prospectively evaluate and model surrogate explanatory variables (SEVs) of target coverage and rectal dose pertaining to soft tissue anatomy visualised on cone beam computed tomography (CBCT) for incorporation into post-prostatectomy treatment coverage verification protocols. Twenty post-prostatectomy patients treated with conformal prostate bed radiotherapy (64-74 Gy) underwent CBCT daily at fractions 1 to 5, and then weekly. Treatment coverage was defined on each CBCT using 'PTV95', percentage of the CBCT PTV covered by original treatment fields, and 'RECTD50', dose delivered to 50% of CBCT rectal volume by original treatment fields. Three candidate SEVs for treatment coverage were defined for each scan: anterior rectal wall movement, change in bladder length and bladder base movement. Both anterior rectal wall movement and increase in bladder length predicted for the decreased PTV95 (P < 0.001 for each). Anterior movement of the anterior rectal wall predicted for increased RECTD50 (P < 0.001). Predictive models for the PTV95 and RECTD50 that accept the significant SEVs as inputs were developed. We developed simple CBCT-acquired soft tissue anatomic surrogate measures that signal changes in target coverage and rectal dose during post-prostatectomy radiotherapy. Conventional bony anatomy patient position verification protocols were inadequate in accounting for soft tissue target and organ variation seen with CBCT.