Encrypted login | home

Program Information

Computational Tumor Modeling for Dose Painting Candidate Stratification

S Harmon

S Harmon*, B Titz, R Jeraj, Department of Medical Physics, University of Wisconsin-Madison

TH-C-213AB-10 Thursday 10:30:00 AM - 12:30:00 PM Room: 213AB

Purpose: Dose painting plans are generally created by redistributing an arbitrarily chosen portion of the prescription dose (D Rx) according to a map of biological characteristics. This study investigated how different levels of redistributed dose affect patient-specific therapeutic gain.

Methods: Twelve canine patients with sinonasal tumors underwent pre-therapy PET/CT imaging of proliferation using [₁₈F]FLT. Following two fractions of uniform dose and two off-treatment days, early proliferative response was assessed via follow-up [₁₈F]FLT PET/CT. Using an imaging-based tumor simulation model, voxel-based, 'effective' linear radio-sensitivity values (a eff) were extracted from registered PET images. Based on these a eff maps, dose painting plans were generated featuring dose redistribution levels of 5%, 10%, 25%, 50%, and 100% D Rx. Individual end-of-treatment responses to each plan were simulated, analyzed, and compared to a reference uniform plan.

Results: With increasing levels of dose redistribution, average cell kill increased in all but one patient, indicating therapeutic gain. As the redistributed dose increased from 5% to 100%, mean proliferative activity decreased by 3%-5% in three patients, while nine patients showed a more pronounced reduction by 8%-38%. For these nine benefiting patients, most substantial gains were observed when redistributing 10%, 25%, and 50% D Rx (up to 14%, 25%, and 33% increase in average cell kill, respectively). When redistributing 5% D Rx, the decrease in average cell survival was below 4% in ten patients, suggesting minimal therapeutic gain relative to the uniform plan. Similarly, increasing the redistributed dose from 50% to 100% D Rx resulted in only modest gains (less than 5% on average) in cell kill for all patients.

Conclusions: Results indicate that the majority of patients would benefit from dose painting. Computational modeling based on imaged proliferative response to a uniform test dose could allow for candidate patient stratification. For suitable patients, the redistributed dose should be optimized to within 10-50% D Rx.

Contact Email