Comparative Investigation of Internal Dosimetry Methodologies

Objectives The need for individual patient dosimetry is increasingly perceived as relevant for patient care and for the development of new therapeutic radiopharmaceuticals (2013/59/EURATOM). Different methodologies for dose calculation exist and their accuracy is an important factor in dosimetry. In this work we aimed (1) to investigate and validate voxel S dose calculation for patient-specific safety and tumor dosimetry and (2) to compare two dose calculation tools, the ICRP-endorsed IDAC 1.0 package, and OLINDA/EXM 1.1, commonly used for phantom-based safety dosimetry. All evaluations were performed using QDOSE, a comprehensive software solution allowing for direct comparison of different methodologies.

Methods Voxel S dose calculations were validated against the spherical model implemented in the OLINDA/EXM software using spherical objects of masses of 10, 50, 100 and 1000 g for the isotopes 177Lu, 90Y, 131I and 188Re using implemented or interpolated voxel S kernels depending on the voxel size of the images. For safety dosimetry, the software tools IDAC 1.0 and OLINDA/EXM 1.1, which are both based on the stylized Cristy-Eckerman phantom series, were compared for patient data acquired with 177Lu-Dotatoc.

Results The voxel S mean doses in spherical objects were in good agreement with the spherical model showing differences of -10.5 % to -12.6% for 177Lu, -4.5% to -10.5% for 90Y, -6.1% to -22% for 131I and -7.3% to -12.4% for 188Re. No differences between implemented and interpolated voxel S kernels were observed. Comparing IDAC 1.0 to OLINDA/EXM 1.1, the absorbed doses to all organs for 177Lu-Dotatoc were within 2%, except for the remainder body with a 4.8% and red marrow with a 31.6% difference when kidneys, spleen and remainder body were included as source organs. Adding red marrow as a source organ, the red marrow dose with IDAC 1.0 was lower than with OLINDA/EXM 1.1 with a maximum difference of -27.5% depending on the residence time ratio of remainder body to red marrow.

Conclusions Voxel S dose calculation using kernels based on EGSnrc allows for accurate tumor efficacy or patient-specific organ safety dosimetry. Differences between voxel S and spherical model are mostly related to improvements in radiation transport simulation in the Monte Carlo code EGSnrc (DOSXYZnrc) used for the voxel S generation compared to EGS4 and MCNP used for the spherical model. The differences between IDAC 1.0 and OLINDA/EXM 1.1 are of the same magnitude or lower than potential errors in dose calculation due to the difference between individual patients and the stylized reference phantom. They are related to a red marrow correction implemented in OLINDA/EXM while IDAC 1.0 follows MIRD and assumes only self-absorption for electrons. The successor IDAC 2.0 uses the ICRP reference phantom and will incorporate measured absorbed fractions for electrons and therefore surpass the previous limitations for more accurate dosimetry including for the red marrow.