Caren van Roekel

174 Chapter 6 DISCUSSION This prospective exploratory study is the first to show clinical evidence of an absorbed dose-response relationship in patients treated with 166 Ho- radioembolization. Specifically, a high tumor absorbed dose was associated with individual tumor and per-patient response and the occurrence of patient- level objective response was associated with a significantly increased overall survival. The efficacy of radioembolization with 166 Ho-microspheres for inducing anatomical response according to RECIST 1.1 has previously been demonstrated by Prince et al. (7). For this study, metabolic metrics were used to measure response. These metrics are more sensitive, often have an earlier onset and can be more predictive of overall survival (14). This was indeed reflected in the higher fraction of patients who were classified as responders at three months follow-up in the present study (12/36; 33%) versus in the study by Prince and colleagues (5/37; 14%). Furthermore, grouping according to metabolic response resulted in significant differences in overall survival between these groups. This metabolic response was associated with a higher tumor absorbed dose. Van der Hoven, et al. conducted a study similar to this one, but with 90 Y resin microspheres in mCRC patients (8). Van der Hoven and colleagues conservatively estimated that a dose of 40-60 Gy would be needed to achieve a significant tumor response. Willowson et al. found ~50 Gy to be sufficient for a metabolic response (15) and Levillain and colleagues found that an average absorbed dose on all tumors higher than 39 Gy was a good predictor of both metabolic response as well as overall survival (16). Flamen et al. found a median of 46 Gy for the metabolic response group (17,18). All these studies used resin microspheres in mCRC patients. In the current study, the estimated dose needed for a local response was higher (geometric absorbed tumor dose was 232 Gy for CR and 168 Gy for PR). This likely reflects differences between the used microspheres and potentially also between the methods used for the actual dose estimation. Furthermore, a direct quantitative comparison with the present study is hampered by the heterogeneity in primary tumor types of included patient cohort. For a valid pair wise comparison, a more homogenous patient group is needed.