ESTRO 36 Abstract Book

S477 ESTRO 36 2017 _______________________________________________________________________________________________

permits computation of clinically meaningful differences in context, but differences in spinal cord dose mandating ART should be a rare event. PO-0880 Using accumulated delivered dose to predict rectal toxicity in prostate radiotherapy L.E.A. Shelley 1,2,3 , J.E. Scaife 1,4 , A.M. Bates 1,4 , J.R. Forman 1,5 , K. Harrison 1,6 , R. Jena 1,4 , D.J. Noble 1,4 , M.A. Parker 1,6 , M.R. Romanchikova 1,3 , M.P.F. Sutcliffe 1,2 , S.J. Thomas 1,3 , N.G. Burnet 1,4 1 Cambridge University Hospitals NHS Foundation Trust, Cancer Research UK VoxTox Research Group, Cambridge, United Kingdom 2 University of Cambridge, Department of Engineering, Cambridge, United Kingdom 3 Cambridge University Hospitals NHS Foundation Trust, Department of Medical Physics and Clinical Engineering, Cambridge, United Kingdom 4 Cambridge University Hospitals NHS Foundation Trust, Department of Oncology, Cambridge, United Kingdo 5 Cambridge University Hospitals NHS Foundation Trust, Cambridge Clinical Trials Unit, Cambridge, United Kingdom 6 University of Cambridge, Department of Physics- Cavendish Laboratory, Cambridge, United Kingdom Purpose or Objective Dose-volume tolerances for organs at risk (OARs) adopted during radiotherapy planning have been historically derived from normal tissue complication probability (NTCP) models linking toxicity with planned dose. On-treatment image guidance facilitates daily tumour localisation ensuring target coverage. However, the positional variation of neighbouring OARs is often disregarded. Anatomical deviations from the pre- treatment CT due to interfraction motion can introduce discrepancies between the planned and delivered dose. One objective of the VoxTox research programme is to test the hypothesis that delivered radiation dose can be a stronger predictor of toxicity than planned dose. Material and Methods For 109 prostate cancer patients treated with TomoTherapy® (74Gy/37#), daily megavoltage CT scans were acquired. An in-house autocontouring algorithm determines the rectal position, incorporating the effect of displacement and deformation, and an independent dose calculation is performed. Processing is fully automated within the VoxTox study. Dose surface maps (DSMs) of the rectal wall were generated following the virtual cutting and unfolding method of Buettner et al [ Phys. Med. Biol , 54, 21 (2009)], allowing conservation of spatial dose information (Figure 1). Daily delivered DSMs were summated to produce an accumulated DSM (Figure 1b) over the whole treatment. Planned and accumulated DSMs were parametrised by calculating 1) the equivalent uniform dose (EUD) and 2) the ‘DSM dose-width’, the lateral extent of an ellipse fitted to the largest isodose cluster, for 7 discrete dose levels between 30 and 75 Gy.

Receiver-operator characteristic curves were plotted, linking the extracted dose parameters with six patient- reported clinical endpoints: rectal bleeding, proctitis, sphincter control, rectal pain, and “How big a problem are bowels?” (≥Grade 1, ≥Grade 2). Statistical correlations between planned and accumulated DSMs were compared using the calculated area under the curve (AUC) presented on High-Low plots. Results For rectal bleeding, the 30, 40, and 60 Gy accumulated DSM dose-widths were significant predictors (AUC 0.629, 0.621 and 0.643 respectively), where planned dose was not (Figure 2a). For DSM dose-widths up to 70 Gy, AUC was greater for accumulated dose than planned dose. EUD was the strongest predictor of rectal bleeding from both accumulated (AUC 0.682) and planned (AUC 0.673) DSMs. The only significant predictor of proctitis was EUD of the accumulated DSM (AUC 0.673) (Figure 2b). Neither planned nor accumulated doses were predictive of the other endpoints