Differential Response of Pelvic Bone Marrow Fluorodeoxyglucose Uptake in Patients Receiving Concurrent...

Irradiation of pelvic bone marrow (PBM) at the level of the typical low dose bath of intensity-modulated radiotherapy delivery (10-20 Gy) is associated with an increased risk of haematological toxicity, particularly when combined with concurrent chemotherapy. Although sparing of the whole of the PBM at a 10-20 Gy dose level is unachievable, it is known that PBM is divided into haematopoietically active and inactive regions that are identifiable based on the threshold uptake of [¹⁸F]-fluorodeoxyglucose (FDG) seen on positron emission tomography-computed tomography (PET-CT). In published studies to date, the definition of active PBM widely used is that of a standardised uptake value (SUV) greater than the mean SUV of the whole PBM prior to the start of chemoradiation. These studies include those looking at developing an atlas-based approach to contouring active PBM. Using baseline and mid-treatment FDG PET scans acquired as part of a prospective clinical trial we sought to determine the suitability of the current definition of active bone marrow as representative of differential underlying cell physiology.

Active and inactive PBM were contoured on baseline PET-CT and using deformable registration mapped onto mid-treatment PET-CT. Volumes were cropped to exclude definitive bone, voxel SUV extracted and the change between scans calculated. Change was compared using Mann-Whitney U testing.

Active and inactive PBM were shown to respond differentially to concurrent chemoradiotherapy. The median absolute response of active PBM for all patients was -0.25 g/ml, whereas the median inactive PBM response was -0.02 g/ml. Significantly, the inactive PBM median absolute response was shown to be near zero with a relatively unskewed distribution (0.12).

These results would support the definition of active PBM as FDG uptake greater than the mean of the whole structure as being representative of underlying cell physiology. This work would support the development of atlas-based approaches published in the literature to contour active PBM based on the current definition as being suitable.