SUMMARY

The physics behind proton therapy support the notion that some cancer treatments may benefit from the therapeutic use of protons. Even though proton therapy typically offers the more favorable dose profile compared to conventional treatments, such as photon therapy, its clinical advantages have not yet been sufficiently demonstrated. In response to this lack of evidence, the National Cancer Institute (NCI) has begun development of the first epidemiological cohort study directly comparing pediatric patient outcomes between modern photon and proton therapy techniques, including 5,000 pediatric patients treated via passive scattering proton therapy. With radiotherapy treatment data currently being collected for the cohort, large-scale retrospective dosimetry methods have the potential to connect the resulting risk estimates to the site-specific radiation doses. The objective of the current work was to support the organ dose reconstruction efforts of this cohort study by leveraging dosimetry infrastructure built at the Massachusetts General Hospital (MGH). Monte Carlo models of the passive scattering beamlines used at MGH were integrated with the Biowulf supercomputing cluster at the NCI, with detailed electronic radiotherapy records used to reconstruct the patient-specific treatment fields. Methods were developed to obtain in-field and out-of-field organ dose estimates and dose-volume metrics for thirteen passive scattering patients treated at MGH. In the process, groundwork was laid for future dosimetry efforts with the remaining MGH patients enrolled in the cohort; special attention was paid to workflow automation and the optimization of histories needed to reach fair statistics on the simulation results. Additionally, the impact of Monte Carlo physics models on out-of-field dose estimates were also investigated. These results showed at least a 25% variation in out-of-field dose due to differences in the estimated generation of secondary particles within the patient.