86 Chapter 5 for partial volume effect, subgroup analyses were performed for lesions ≥20m (reported in the Supplementary data) PD-L1 immunohistochemistry Fresh or archival cytological or histological samples suitable for PD-L1 staining was available for 27 patients. This involved tumor tissue from recurrent disease (n=12), or metastases in lung (n=7), lymph node (n=7) or bone (n=1). PD-L1 staining was performed using VENTANA PD-L1 (SP263) assay and evaluated by a certified pathologist in head and neck cancer (I.E.), blinded for clinical information. As an internal control, staining for PD-L1 was performed with the clinical validated 22C3 antibody using the DAKO stainer in histological samples of 8 patients8. In all samples, PD-L1 expression was assessed according to the combined positive score (CPS), which describes the number of PD-L1 positive tumor cells plus immune cells per 100 tumor cells, showing positive cell membrane staining and/or a score of <1, 1-20 or >20. Statistical analyses Clinical outcome was evaluated according to an intention-to-treat analyses on a patient and lesion level. This was visualized in a waterfall plot. Furthermore, we assessed the disease control rate (DCR), overall response rate (ORR), progression free survival (PFS) and overall survival (OS). A log-rank test was performed to correlate PD-L1 CPS to PFS. A Cox regression model was used to report hazard ratios (HR) for progressive disease. In the dose finding study, we compared differences in tracer-uptake and TTB ratios between the three dose groups, testing for significance using a two-sided Kruskal-Wallis test. On a patient level, the relation between [89Zr]Zr-DFO-durvalumab patient gm SUVpeak and SUVpeak of the hottest lesion with durvalumab response was explored by Kaplan–Meier survival plots. Similar analyses were performed for gm [18F]FDG SUVpeak, TLG and MTV. Patients were grouped in a below-median and above-median group to evaluate a difference in survival using the log-rank test. The relation between these groups was tested by additional Cox regression models, reporting HRs for progressive disease and/or survival. We correlated [89Zr]Zr-DFO-durvalumab uptake with [18F]FDG SUVpeak and [ 18F]FDG TLG on a lesion level. Additional descriptive analyses were performed to evaluate the per lesion PD-L1 expression to tracer accumulation. For these correlations we report the Spearman correlation coefficient (ρ). Statistical analyses were performed using SPSS Statistics for windows version 22.0. Differences with a p-value of 0.05 or less were considered statistically significant.