Caren van Roekel

175 Dose-response relationship in 166 Ho-radioembolization The semi-automatic method of thresholding the FDG scans to define tumor volumes, as used in the study, decreased the variance typically induced with manual delineation. By subsequently applying these masks to the corresponding 166 Ho dose maps using an automatic registration routine, the current method offered a non-subjective measure for both dose and response, maximizing reproducibility. The current study was performed with a limited sample size of patients with hepatic metastases of different origins. Consequently, there was not enough statistical power to model the differences in FDG avidity, tumor biology and radio-sensitivity that might exist between the different tumor types. Furthermore, differences in patient positioning and breath-hold policy between PET and SPECT scans, combined with the relatively low resolution and contrast of the low-dose CT of the SPECT/CT increased the likelihood of (local) misregistrations. These effects increased the error in dose estimates of each response group, contributing to a larger spread in each response category, decreasing separability between response groups. It has been argued that the different radioembolization devices (e.g. resin or glass) result in differences in micro-distribution and consequently the absorbed dose needed for tumor response and toxicity (3). 166 Ho-microspheres are positioned between resin and glass microspheres with respect to the number of injected particles and particle size (19). Based on these data, we expect the ‘apparent’ radio-sensitivity of 166 Ho-microspheres to lie in between as well. However, this will need to be confirmed in a future study in which only patients with the same tumor type are included. The administered activity in the HEPAR I and II studies was based on the MIRD mono-compartment method. In this method, the activity calculation was based on the intended mean absorbed dose to the target liver mass. This method disregards the actual tumor load and the preferential uptake of the microspheres in the tumors, assuming a uniform microsphere distribution in the target volume. This can lead to a wide range in actual absorbed tumor doses. However, the treatment with 166 Ho-radioembolization is usually preceded by the administration of a smaller amount of the same microspheres. This scout dose has been shown, relative to 99m Tc-MAA, to enable: i) a more accurate lung 6