Calculation of tumour and normal tissue biological effective dose in 90Y liver radioembolization with different dosimetric methods

Abstract

PurposeRadioembolization with 90Y microspheres is an effective treatment for unresectable liver tumours. Two types of microspheres are available: resin (SIR-Spheres®) and glass (Theraspheres®). The aim of this study is to compare biological effective dose (BED) values obtained with three different dosimetric methods. Methods29 HCC patients were included in this study: 15 were treated with resin(mean injected activity 1.5GBq, range 0.8–2.7GBq) and 14 with glass microspheres (2.6GBq, range 1.3–4.1GBq). Average doses to tumours and normal liver tissues were calculated with AAPM, multi-compartmental MIRD and Voxel-based methods and consequently the BED values were obtained. Planar images were used for the AAPM method: 99mTc-MAA SPECT-CT attenuation and scatter corrected images (resin) and 99m Tc-MAA SPECT attenuation corrected (glass) were employed for the other two methods. ResultsRegardless of type of microspheres, both for tumours and normal liver tissues, no significant statistical differences were found between MIRD and Voxel for both doses and BED values. Conversely AAPM gave discordant results with respect to the other two methods (Mann-Whitney p-values⩽0.01). For resin spheres the calculated tumour-to-normal tissue ratios on planar images were on average 14 times greater than those obtained on SPECT-CT images, while they were 4 times greater on glass. A linear correlation was observed between MIRD and Voxel BEDs. ConclusionsThe AAPM method appears to be less precise for absorbed dose and BED estimation, while MIRD and voxel based dosimetry are more confident each other.