A Comparison of PET-Defined, Active Bone Marrow Threshold Methods in Predicting Acute Hematologic Toxicity in Cervical Cancer Patients Undergoing Definitive Chemoradiation

Abstract

Hematologic toxicity (HT) during pelvic chemoradiation may cause treatment modifications or delays. There is emerging interest in using PET-defined regions of active bone marrow (ABM) to spare hematopoetically active bone marrow. However, the optimal methods for defining ABM and its dosimetric parameters are undefined. In this study, a comprehensive analysis of dosimetric predictors for two commonly used ABM metrics is performed and compared to total bone as a surrogate for BM. Thirty-one patients treated for cervical cancer from 5/2011 – 10/2017 were reviewed. All patients received pelvic radiation using IMRT (12 patients, 38.7%) or 3D-CRT (19 patients, 61.3%) with concurrent cisplatin. Pelvic BM (PBM) was defined using a total bone surrogate including L4, L5, proximal femur to lesser trochanter, ilium, sacrum, ischium, and pubis. Whole body mean SUV ABM (WB-ABM) was defined as the ABM greater than the total body mean SUV minus brain. Bone mean SUV ABM (B-ABM) was defined analogous to the NRG-GY006 protocol as ABM with an SUV greater than the mean SUV of bone. Receiver Operating Characteristic (ROC) curve were used to assess optimal cutpoint criteria for grade 3+ HT for any cell line based on the CTCAEv4. Conventional dosimetric parameters (mean dose, V10, V20, V40) to PBM, WB-ABM, and B-ABM were assessed as well as the absolute volume (cc) spared 10, 20, and 40 Gy for each defined volume. The volumes of ABM were compared using a paired t-test. At least one grade 3+ HT occurred in 26 patients (83.9%). The B-ABM volumes (mean 598 cc, 415-875 cc) were significantly smaller than the WB-ABM volumes (mean 1078 cc, 481-1603 cc) (p<0.0001). The mean dose, V10, V20, V40, volume spared 10 Gy and 20 Gy of WB-ABM, B-ABM, and PBM; as well as volume spared 40 Gy of B-ABM and PBM, did not predict HT (table). The absolute volume of WB-ABM spared 40 Gy (<853cc, AUC 0.777, p=0.0016) was the best predictor for grade 3+ HT. In patients with <742 cc WB-ABM spared 40 Gy, 16/16 (100%) had grade 3+ HT compared to 10/15 (67%) patients with >742 cc spared 40 Gy (p=0.04). Using PET to define metabolically active BM is feasible. Thresholding ABM using the mean SUV of bone (B-ABM) yields volumes that are significantly smaller than whole body mean (WB-ABM) volumes and may not predict for grade 3+HT. This analysis suggests that absolute volume of WB-ABM spared 40 Gy (>742 cc) is a strong predictor of grade 3+HT and should be further evaluated for clinical use.Abstract TU_16_3476: Table 1ROC Analysis for grade 3+ HTPelvic BonepWB-ABMpB-ABMpMean dose0.6540.15010.5850.42460.5770.467V100.5380.76220.5270.82260.5540.682V200.6690.1320.6350.2280.560.589V400.510.5380.7430.5120.911cc spared 10 Gy0.5850.54060.6770.120.5770.521cc spared 20 Gy0.5040.9770.5850.480.5080.951cc spared 40 Gy0.7310.03810.7770.00160.70.061 Open table in a new tab