Sarah Verhoeff

123 [89Zr]Zr-DFO-avelumab PET/CT in early stage NSCLC patients DISCUSSION This is the first study on PD-L1 PET-imaging in a neo-adjuvant setting, using [89Zr]Zr-DFOavelumab PET/CT in patients with non-small cell lung cancer (NSCLC). [89Zr]Zr-DFO-avelumab PET/CT was considered feasible and safe with pre-medication. Tumor [89Zr]Zr-DFO-avelumab accumulation before avelumab treatment was correlated to pathological response. Selecting the optimal protein dose for [89Zr]Zr-DFO-avelumab PET-imaging is challenging, since PET quantifications and pharmacokinetic analyses do not favor for a particular dose. The PET/CT imaging quality and quantification of the lowest protein dose of 2 mg was considered suboptimal, despite the high tumor to blood (TTB) ratios found. This discrepancy is most likely the result of trapping of the radiolabeled antibody in the spleen. The spleen acts as a sink organ and subsequently the dose radiolabeled antibody in the circulation and available to bind to the tumor PD-L1 is perhaps suboptimal22. Although the TTB ratio in higher protein doses of 10mg and 50mg was slightly less, the SUVpeak values with these protein doses were higher and comparable. This ‘sink organ phenomenon’ might also affect the choice for optimal protein dose and imaging time points in patients with a high tumor load or systemic immune activation, and thus total potential target binding sites23. We therefore recommend for future studies to ensure an equal distribution of tumor load over different dose cohorts when optimizing the protein dose for imaging. Ultimately, we selected protein dose 10mg based on optimal visual assessment at four days after injection. It provided feasible images, with a protein dose and imaging time-point that is supported by previous studies24. As opposed to preclinical studies10,25, clinical PD-(L)1 PET-imaging studies report varying concordance of tracer-accumulation with IHC-determined PD-L1 expression11-13,26. We were able to correlate the IHC-determined PD-L1 to tracer accumulation, using only fresh tumor biopsies. Interpretation of these results should be taken with caution since each study reported different staining and scoring procedures. More importantly, tumor tracer-accumulation reflects all PD-L1 positive cells, including PD-L1 positive infiltrating immune cells in the whole tumor. This percentage of PD-L1 positive immune cells is not consistently incorporated in the reported IHC-determined PD-L1 score. The first two studies with [89Zr]Zr-atezolizumab and [89Zr]Zr-nivolumab and [18F]F-BMS-986192 (anti PD-L1) showed a correlation between tracer-uptake and treatment response, whereas the most recently published studies with [89Zr]Zr-pembrolizumab and [89Zr]Zr-durvalumab only demonstrated a trend but no significant correlation. In agreement with the first two studies, we also report a correlation between tracer uptake and treatment response. However, in this neo-adjuvant setting, pathological response is reported instead of RECIST 1.1 assessment as a definition for treatment response. More imaging studies in this setting are needed to determine its clinical value. Additionally, we studied the correlation of tumor metabolism to ICI response, as demonstrated in patients with metastatic NSCLC treated with pembrolizumab27. Based on our data, [18F]FDG SUVpeak was not related to treatment response in neo-adjuvant setting. 6