Caren van Roekel

176 Chapter 6 shunt fraction estimation (4), ii) a safe and improved detection of extrahepatic depositions (20,21), and iii) a more accurate pretreatment prediction of the intrahepatic distribution (5). These predictive properties may be used for an improved patient selection and a more personalized activity prescription. This can be achieved by using the pretreatment biodistribution of the scout dose as input to a multi-compartment model (e.g. the partition model) (22). The prescribed treatment activity can then be maximized such that the absorbed dose in the parenchymal tissue remains below a certain toxicity threshold, whilst maximizing the tumor absorbed dose (23). Subsequent assessment of predicted tumor absorbed doses can guide patient selection by excluding patients for whom no tumor response is to be expected. To that end, 166 Ho absorbed dose thresholds for specific tumor types need to be established. Future studies will need to focus on a single tumor type, increasing statistical power and enabling the identification of this tumoricidal dose threshold. Similarly, a larger study cohort is needed to establish safe absorbed dose thresholds for the parenchyma. The absorbed dose-response relationship demonstrated in this study shows the feasibility of such an effort and is the first step towards a more individualized treatment planning for 166 Ho- radioembolization. CONCLUSION In this study, an association of tumor absorbed dose with (local) response was found. Moreover, a patient-level metabolic response was associated with a significant increase in overall survival. Personalized dosimetry has the potential for improved outcome in radioembolization, as has been well-established for external beam radiotherapy.