72 Chapter 4 mg unlabeled atezolizumab followed by [89Zr]Zr-atezolizumab (37 MBq, ~ 1 mg antibody) intravenously. Imaging was performed at 4- and 7-days post-injection, and subsequent atezolizumab treatment was started until disease progression (Figure 1). The investigators concluded that [89Zr]Zr-atezolizumab injection was safe, apart from one grade 3 infusion-related reaction. In normal tissue, increased [89Zr]Zr-atezolizumab uptake over time was observed in intestines, kidney, liver, and bone marrow. Furthermore, [89Zr]Zr-atezolizumab demonstrated high but variable uptake in nonmalignant lymph nodes and spleen. The standardized uptake value (SUV)max [89Zr]Zr-atezolizumab tumor uptake was 10.4 (95% confidence interval [CI] 8.5–12.7; range 1.6– 46.1), with major intra-tumoral and intertumoral heterogeneity (Figure 2). [89Zr]Zr-atezolizumab tumor uptake was related to clinical response; patients with complete response had a 2.35-fold higher SUVmax than patients with immediate progression (95% CI 98%–476%; p<.001). Niemeijer and colleagues53 published a study using the anti-PD-1 monoclonal antibody nivolumab radiolabeled with 89Zr. A total of 13 patients with advanced NSCLC received [89Zr]Zr-nivolumab (37 MBq, ± 10%, 2 mg antibody) followed by PET/CT at 162 hours (~7 days) post-injection. Subsequently, nivolumab treatment was initiated. The tracer injection was considered safe in the absence of grade ≥3 tracer-related adverse events. Quantitative PET-analyses revealed high tracer accumulation in spleen and liver. [89Zr]Zr-nivolumab tumor uptake was higher in patients with immunohistochemically proven PD-1 positive tumor-infiltrating immune cells as compared with PD-1 negative tumors (median SUVpeak 7.0 vs 2.7, p= .03). [ 89Zr]Zr-nivolumab uptake did not correlate with high tumor PD-L1 expression. Visual [89Zr]Zr-nivolumab tumor uptake revealed significant heterogeneity, both between and within patients. [89Zr]Zr-nivolumab tumor uptake was related to clinical response; of lesions with a diameter of ≥20 mm, the SUVpeak was numerically higher in responding lesions compared with nonresponding lesions (median SUVpeak 6.4 vs 3.0, p= .019). In addition to the first clinical studies with radiolabeled PD-1/PD-L1 monoclonal antibodies, Xing and colleagues54 were the first to demonstrate clinical single-domain antibodies (sdAbs) or nanobody SPECT-imaging. They reported results from 16 patients with advanced NSCLC using NM-01 (anti-PD-L1), which was radiolabeled site specifically with 99mTc. The administered dose of 3.8 to 10.4 MBq/kg, corresponding to 100 or 400 mg of NM-01, was safe. SPECT scans were acquired at 1- and 2-hours post-injection, resulting in visible tumor uptake at 2 hours postinjection. Average tumor to blood pool ratios (T:BP) at 2 hours varied from 1.24 to 3.53 and correlated well with PD-L1 expression measured immunohistochemically. Visual intra-tumoral and inter-tumoral heterogeneity of tracer uptake was noted in some primary tumors and among several patients. Patients with a PD-L1 expression ≤1% tend to show a lower T:BP radio (mean 1.89 vs 2.49, p= .048).