Functional Mri-Guided Bone Marrow-Sparing Intensity Modulated Radiotherapy For Pelvic Malignancies

Patients undergoing concurrent chemoradiotherapy (CRT) for pelvic cancer are at risk for acute hematologic toxicity (HT). We are testing whether image-guidance with functional MRI can guide delineation of hematopoietically active bone marrow (BM), to assist design of BM-sparing radiation plans, which may reduce HT. We enrolled 9 patients with pelvic cancers undergoing CRT onto a prospective imaging protocol. Seven patients' studies were used in the pilot phase of protocol development. We report results for the 2 latest subjects, 1 male with Stage II anal cancer and 1 female with Stage II cervical cancer. Both patients underwent MRI fat fraction mapping (FFMRI) from L4 to the ischial tuberosities using T2∗ IDEAL (Iterative Decomposition of water and fat with Echo Asymmetry and Least-Squares Estimation). The FFMRI decomposes fat and water proton signals based on their resonant frequency difference, distinguishing active BM (aBM) by its relatively lower fat content. The FFMRI images were rigidly registered to simulation CT images and aBM was automatically segmented using commercially available software. Pelvic aBM was defined as BM with a fat fraction <50% (selected a priori based on literature), expressed as a fraction of the total pelvic BM volume. One patient underwent serial FFMRI to evaluate changes in aBM volume. In a planning comparison, the delivered plan for each patient was compared to 2 experimental plans: (1) BM-sparing IMRT using pelvic BM delineated on CT alone as the principal avoidance structure (BMS-IMRT) and (2) image-guided BMS-IMRT (IG-BMS-IMRT) using aBM as an avoidance structure. The primary aim was to reduce the volume of aBM receiving ≥20 Gy (aBM-V20). Experimental plans were normalized to cover PTV with 95% of the prescription dose (45–52 Gy). For the planning comparison in the anal cancer patient, aBM-V20 for the delivered plan, BMS-IMRT, and IG-BMS-IMRT were 65% vs. 55% vs. 53%, respectively. Corresponding mean aBM dose was 27 Gy vs. 22 Gy vs. 21 Gy, respectively. For the cervical cancer patient, aBM-V20 for the delivered plan, BMS-IMRT, and IG-BMS-IMRT were 74% vs.66% vs. 55%, respectively. Corresponding mean aBM dose was 28 Gy vs. 25 Gy vs. 23 Gy, respectively. The extent of aBM-sparing achieved with image guidance appeared to be related to variations in aBM distributions between patients rather than variations in aBM definition based on fat fraction thresholding. In the anal cancer patient, the aBM volume decreased from 41% pretreatment, to 22% at Week 3 (midtreatment). Here we describe a novel method for delineating aBM in patients undergoing CRT for pelvic cancer. Marked reductions in aBM volume were observed during pelvic RT. The technique is feasible and may be useful in optimizing RT planning and quantifying effects of radiation on BM.