Voxel-based dose heterogeneity and dose-volume effects in ⁹⁰Y-microsphere therapy dosimetry

Full text of publication follows. Objective: standard ⁹⁰Y-microsphere therapy planning (STD) conservatively assumes uniform uptake to total liver (normal-liver and tumor) to minimize radiation induced toxicities but fails to provide tumor doses. The three-compartment model (3CM) that accounts for preferential uptake in tumor has been used to estimate tumor doses (Ho 1996). However, ambiguity in estimation of tumor uptake and differences between planning ^99mTc-MAA and post-therapy ⁹⁰Y-microsphere images lead to significant dose uncertainties (MacLellan SNMMI 2012). We propose a new method for calculating 3-dimensional dose distributions (3D-dose) based on post-therapy quantitative ⁹⁰Y SPECT/CT to investigate dose heterogeneity and dose-volume effects. Materials and methods: 3D-doses were generated from DOSXYZnrc Monte Carlo (MC) simulations and ATTILA a grid-based Boltzmann solver (GBBS). We retrospectively computed normal-liver and tumor doses for 5 patients after ⁹⁰Y-microsphere therapy using MC and GBBS. Cumulative dose-volume histograms (DVH) were calculated for total liver using 3D-dose, 3CM, and STD. For evaluation of heterogeneity, we computed ratios of maximum to minimum dose in normal-liver and tumor; the integral uniformity of the 3D-dose in normal liver (IUNL) and tumor (IUT) was also calculated. We computed percent differences of minimum doses covering 100% of normal-liver (DL100) and tumor (DT100) and median dose to normal-liver (DL50) and tumor (DT50) for 3D-dose and 3CM. Results: the average total liver dose calculated by 3Ddose (using both MC and ATTILA) agreed to within 5% of the STD. 3D-dose calculations showed large dose heterogeneity with over 400% variation in both normal-liver and tumor compartments relative to 3CM. The ranges of max/min doses in normal-liver and tumor were 1.8-to-5.0 and 2.3-to-6.4, respectively. IUNL ranged from 29% to 67%. IUT ranged from 40% to 73%. Compared to 3CM, 3D-dose estimates of DL100 were higher (mean 400%, range -50% to 1600%), while those of DT100 were lower (mean -40%, range -67% to +3%). Compared to 3CM, 3D-dose estimates of DL50 were higher (mean 900%, range 38% to 2600%) while those of DT50 were lower (mean -13%, range -56% to +69%). Conclusion: Doses quantified from 3D-dose exhibited substantial heterogeneity in normal-liver and tumor. While previous works have shown lower normal-liver and higher tumor doses with the 3CM compared to STD, 3D-dose suggests that 3CM may be under/over estimating normal-liver/tumor doses. Dose heterogeneity and dose-volume effects likely play a key role in understanding tumor response and normal tissue toxicity in ⁹⁰Y-microsphere therapy. 3D-dose evaluations for N=20 patients are ongoing. (authors)