Partition Model Based 99mTc-MAA SPECT/CT Predictive Dosimetry Compared with 90Y TOF PET/CT Posttreatment Dosimetry in Radioembolization of Hepatocellular Carcinoma: A Quantitative Agreement Comparison

Y-90-microsphere selective internal radiation therapy (SIRT) is a valuable treatment in unresectable hepatocellular carcinoma (HCC). Partition-model predictive dosimetry relies on differential tumor-to-nontumor perfusion evaluated on pretreatment Tc-99m-macroaggregated albumin (MAA) SPECT/CT. The aim of this study was to evaluate agreement between the predictive dosimetry of Tc-99m-MAA SPECT/CT and posttreatment dosimetry based on Y-90 time-of-flight (TOF) PET/CT. Methods: We compared the Tc-99m-MAA SPECT/CT results for 27 treatment sessions (25 HCC patients, 41 tumors) with Y-90 SIRT (7 glass spheres, 20 resin spheres) and the posttreatment Y-90 TOF PET/CT results. Three-dimensional voxelized dose maps were computed from the Tc-99m-MAA SPECT/CT and Y-90 TOF PET/CT data. Mean absorbed dose (D-mean) 1 was evaluated to compute the predicted-to-actual dose ratio (DRmean = D-Mean(MAA)/D-mean(90Y)) in tumor volumes (TVs) and nontumor volumes (NTVs) for glass and resin spheres. The Lin concordance (rho(c)) was used to measure accuracy (C-b) and precision (rho). Results: Administered activity ranged from 0.8 to 1.9 GBq for glass spheres and from 0.6 to 3.4 GBq for resin spheres, and the respective TVs ranged from 2 to 125 mL and from 6 to 1,828 mL. The mean dose D-mean(90Y) was 240 Gy for glass and 122 Gy for resin in TVs and 72 Gy for glass and 47 Gy for resin in NTVs. DRmeanTV was 1.46 +/- 0.58 (0.65-2.53) for glass and 1.16 +/- 0.41 (0.54-2.54) for resin, and the respective values for DRmeanNTV were 0.88 +/- 0.15 (0.56-1.00) and 0.86 0.2 (0.58-1.35). DR variability was substantially lower in NTVs than in TVs. The Lin concordance between D-Mean(MAA) and D-mean(90Y) (resin) was significantly better for tumors larger than 150 mL than for tumors 150 mL or smaller (rho(c) = 0.93 and C-b = 0.95 vs. rho(c) = 0.57 and C-b = 0.93; P < 0.05). Conclusion: In Y-90 radioembolization of HCC, predictive dosimetry based on Tc-99m-MAA SPECT/CT provided good estimates of absorbed doses calculated from posttreatment Y-90 TOF PET/CT for tumor and nontumor tissues. The low variability of DRNTV demonstrates that pretreatment dosimetry is particularly suitable for minimizing radiation-induced hepatotoxicity.