UC San Diego

Background

To test the hypothesis that atlas-based active bone marrow (ABM)-sparing intensity modulated radiation therapy (IMRT) yields similar dosimetric results compared to custom ABM-sparing IMRT for cervical cancer patients.

Methods

We sampled 62 cervical cancer patients with pre-treatment FDG-PET/CT in training (n=32) or test (n=30) sets. ABM was defined as the subvolume of the pelvic bone marrow (PBM) with standardized uptake value (SUV) above the mean on the average FDG-PET image (ABM_Atlas) vs. the individual's PET (ABM_Custom). Both were deformed to the planning CT. Overlap between the two subvolumes was measured using the Dice coefficient. Three IMRT plans designed to spare PBM, ABM_Atlas, or ABM_Custom were compared for 30 test patients. Dosimetric parameters were used to evaluate plan quality.

Results

ABM_Atlas and ABM_Custom volumes were not significantly different (p=0.90), with a mean Dice coefficient of 0.75, indicating good agreement. Compared to IMRT plans designed to spare PBM and ABM_Custom, ABM_Atlas-sparing IMRT plans achieved excellent target coverage and normal tissue sparing, without reducing dose to ABM_Custom (mean ABM_Custom dose 29.4Gy vs. 27.1Gyvs. 26.9Gy, respectively; p=0.10); however, PTV coverage and bowel sparing were slightly reduced.

Conclusions

Atlas-based ABM sparing IMRT is clinically feasible and may obviate the need for customized ABM-sparing as a strategy to reduce hematologic toxicity.