Cross-modality applicability of rectal normal tissue complication probability models from photon- to proton-based...

Proton therapy (PT) is currently being studied to improve normal tissue (NT) sparing beyond what can be achieved with conventional photon-based therapy. Compared to photons, PT dose distributions have a reduced NT low-to-intermediate 'dose bath' and a different biological effectiveness, questioning the applicability of photon-based NT complication probability (NTCP) models to PT. The aim of this study was to assess the applicability of photon-based NTCP models to rectum morbidity outcomes following PT.

Treatment planning and morbidity data from 1151 prostate cancer patients treated with passive scattering PT and from 159 patients treated with conventional 3D conformal four-field photon therapy were analysed. Prospectively scored gastrointestinal morbidities (grade >=2) were analysed, with a total of 184 events (protons; medical and procedural) and 12 events (photons; procedural only), respectively. Rectal dose volume histograms were extracted for all patients in both cohorts and used as input to two different NTCP models, with up to six different published photon-based parameter sets.

Photon-based rectal NTCP models either over- or underestimated the clinically observed gastrointestinal morbidity when used on the proton cohort, depending on the choice of endpoint (p < 0.05 for all parameter sets, for both morbidity classifications). Four of the six photon-based NTCP models showed a good fit to the photon outcome data (p > 0.05).

There were large differences in morbidity predictions between cohorts and modalities, indicating that the validity of NTCP models and parameters across institutions and treatment modalities should be carefully investigated prior to clinical application.