Caren van Roekel

235 Use of an anti-reflux catheter in 166 Ho-radioembolization lobe delineation was done on the accompanying low-dose CTs of the baseline [ 18 F]-FDG PET/CT using the cone-beam CT images on the side as a reference. The tumor contours were obtained using a threshold-based approach, based on the PERCIST guidelines. The resulting volumes of interest were transferred from the [ 18 F]-FDG PET/CT to the 166 Ho-SPECT/CT using a rigid coregistration of the accompanying low-dose CTs, as described before (20) (Figure 2). Statistical analyses The sample size calculation, based on a difference of 0.4 in mean tumor to non-tumor (T/N) activity ratio between the catheters, showed that at least 23 patients needed to be treated (15). The intent was to treat 25 patients (15). The differences in mean post-treatment T/N activity ratio and mean tumor- and healthy liver-absorbed dose on SPECT/CT between administration with an anti- reflux catheter and a standard microcatheter were assessed using a paired t-test. The infusion efficiency was calculated as percentage residual activity and compared using a McNemar’s test for paired data. The predictive value of the 166 Ho-scout was assessed using Bland-Altman analysis. The relation between tumor-absorbed dose and response was best explained using a linear mixed- effects regression model, using a random intercept per patient, to account for correlation of tumors within patients. The influence of the anti-reflux catheter on tumor response was analyzed with logistic regression. Analyses were primarily performed according to the intention-to-treat (ITT) principle. Per- protocol analyses were also performed. A subgroup analysis was performed in patients in whom the anti-reflux catheter was deployed in the right hepatic artery, under the hypothesis that its effect on hemodynamics and dose distribution would be most notable in wide vessels. Furthermore, a subgroup analysis was performed in liver lobes treated with the anti-reflux catheter only, to evaluate the influence of spasm (as evident during angiography) on T/N activity concentration ratio. Overall survival was defined as the interval between treatment and death from any cause. Cox regression models were made using Firth’s correction for small sample bias (21). Analyses were performed using R statistical software for Windows, version 3.6.2. We report effect estimates with associated 95% confidence intervals and corresponding two-sided p-values. 9