Voxel-Based Computational Tools Help Liver Dosimetry Calculations of Multiple (External and Internal) Radiation Therapies

Abstract

Voxel-wise absorbed dose distributions, γ-index agreement test and dose-volume histograms (DVH) for the evaluation of dose distributions have been used in stereotactic body radiotherapy (SBRT) for several years. Our research group recently adapted these concepts to liver radioembolization (RE) using Yttrium-90 (90Y) glass microspheres (MS). In this work we demonstrate the application of our dosimetry tools to study the dose distributions within the tumor (PTV) and normal liver (NLV) tissues of a patient treated with multiple radiation therapies (RE after SBRT). Voxel-wise biological effective dose (BED) distributions from SBRT and RE were computed and added up. Moreover, we investigated the role of pre-treatment single-photon emission computed tomography (SPECT) in the planning dosimetry of RE. This individual case describes a 73 years old man with pancreatic carcinoma and a single liver metastasis previously treated with 48 Gy of 10 MV SBRT photon beams in three sessions of 16 Gy. Ten months later he underwent RE due to disease progression. RE was performed according to the MS manufacturer’s recommendations. Hepatic arterial angiography and pre-treatment Technetium-99 m (99mTc) macroaggregated albumin (MAA) scans (planar and SPECT) were performed in preparation of RE. Then, a 90Y-MS activity of 3.239 GBq was calculated and later administered in the liver. 90Y-MS positron emission tomography (PET) images were then used to verify the distribution of the MS in the liver. SPECT-MAA and PET-MS images were used, retrospectively, to calculate the voxel-wise dose distributions. Voxel-wise BED distributions of SBRT and RE were computed and added up to obtain the total BED. Cumulative dose volume histograms (DVHs) were generated to evaluate the safety of multiple radiation therapies. Finally, the voxel-wise intraclass correlation coefficient (ICC) and γ-index between pre and post-treatment dose distributions were computed. The predictive power role of the pre-treatment SPECT-MAA in the planning dosimetry of RE after SBRT was further investigated. In conclusion, the previous SBRT treatment contributed significantly to the BED in the RE PTV with no detectable contribution to the BED in the RE NLV. Therefore, higher activity might have been administered during RE keeping the NLV dose at safe values. Moreover, the verification PET-MS dose distribution, in this individual case, showed good agreement with the pre-treatment SPECT-MAA dose distribution (ICC = 0.73, γ-index = 98%). Consequently, SPECT-MAA could have been used in the planning dosimetry of RE.