A 49-year-old man presented in February 2017 with chest wall pain and weight loss, leading to the diagnosis of a ccRCC with metastases to the lungs and a rib (Fig. 1). Based on his presentation with anemia, hypercalcemia, and the need for prompt initiation of systemic therapy, his International Metastatic Renal Cell Carcinoma Database Consortium (IMDC) scoring predicted a poor prognosis with a median survival of 7.8 months [8]. Initial management included a right radical cytoreductive nephrectomy, which required a partial hepatectomy. Pathological analyses showed a 9 cm ccRCC invading into the perirenal and renal sinus adipose tissue as well as the ipsilateral adrenal gland of ISUP grade 4 with extensive sarcomatoid differentiation. Eight out of eight lymph nodes were positive for metastatic disease. Staging was consistent with a pT4N1 tumor. IHC studies showed positivity for CK AE1/AE3, and CA-IX. CK7 was negative. PBRM1 and BAP1 were present suggestive of a wild-type state. PD-L1 was expressed in more than 30% of tumor cells. Given the age of the patient, germline testing was pursued using a CancerNext-Expanded genetic panel including genes such as VHL, BAP1, FLCN and PTEN, but did not reveal any mutations.Fig1Fig. 1

Clinical case. a Coronal contrast-enhanced CT images of a lytic metastasis in the left 10th rib (red arrow) before and after SABR and HD-IL2. b Axial contrast-enhanced CT image of new lytic metastasis in the right distal anterolateral femur (red arrow), which developed after SABR/HD-IL2 therapy. c Coronal proton density fat saturated MR imaging of an osseous metastasis in right glenoid (red arrow) that developed while on pazopanib therapy. d Clinical images illustrating radiation recall dermatitis 11 days after first nivolumab infusion at two prior sites of radiation, the left rib (A, radiated six months prior) and the right knee (B, radiated one month prior). Outlined is an area of subcutaneous edema and discoloration (C) attributed to drainage from lesion A. e Axial contrast-enhanced CT scan of the chest of representative lingular nodule (red arrow) improving with nivolumab therapy. f Hematoxylin and eosin stains of left colon biopsy with increased intraepithelial lymphocytes and cryptitis representative of autoimmune colitis

We radiolabeled atezolizumab, a monoclonal anti-PD-L1 antibody with a mutant Fc, with zirconium-89 [⁸⁹Zr]. Zirconium-89 is a well-studied positron emitting radioisotope used to label antibodies with a half-life of 78 h, which is compatible with the slower pharmacokinetics of antibodies [14]. This allows imaging to be performed for several days after injection to improve tumor-to-background signal. Accumulation of the isotope in tumor sites over time and clearance from other sites improves contrast.

Methodologically, the antibody was conjugated to the chelator deferoxamine (DFO) at a molar ratio of 1:1.9 and radiolabeled with ⁸⁹Zr (5 mCi per mg of DFO-atezolizumab conjugate) using previously published protocols [15, 16]. Briefly, the DFO-atezolizumab conjugate was incubated with neutralized ⁸⁹Zr for 1 h, and the reaction was quenched with 50 mM diethylenetriamine pentaacetic acid. The radiolabeled antibody fraction was purified using the Zeba™ centrifugal spin columns (40 K MWCO) and eluted in 0.2 M sodium acetate buffer containing 5 mg/mL gentisic acid (pH 5.5–5.6). The conjugate had a specific radioactivity of 2–4 mCi/mg protein, with high radiochemical purity (≥ 99%). The immunoreactivity of the radiolabeled immunoconjugate was confirmed using an in vitro cell-based Lindmo assay [17] and was 86.2 ± 4% (n = 6). In addition, the conjugate was tested for stability in plasma and was found to be quite stable (> 80% of ⁸⁹Zr activity retained with atezolizumab in rat serum at 37 °C after 7 days).

Mice bearing the patient-derived tumorgrafts were injected intravenously (by tail-vein) with ~ 100 μCi of ⁸⁹Zr-DFO-atezolizumab. A second tumorgraft line from a tumor expressing low levels of PD-L1 (< 1%) by IHC was chosen as a negative control (Fig. 2c). PD-L1 IHC procedures and interpretations were standardized (Biocare Medical, Clone ACI3171A,C; 1:300) and results were scored by a pathologist blinded to other results.

Mice were serially imaged on a Siemens Inveon PET/CT system. PET quantification was performed blinded. PET imaging at day 6 post-injection (d.p.i.) showed a statistically significant difference in ⁸⁹Zr signal between the index patient tumorgrafts (4.2 ± 0.6% injected dose/g [%ID/g]; n = 3) and the controls (3.1 ± 0.5% ID/g; n = 3) (p = 0.028) (see Fig. 2). Similar results were observed with a second independent cohort of tumorgrafts (5.2 ± 0.4% ID/g; n = 3) compared to the same control group (p = 0.002).

Differences in tumor uptake could not be accounted for by differences in tumorgraft volumes, which were not significantly different between the index and control groups (831.9 ± 473 mm³ versus 1010.3 ± 492.6 mm³; p = 0.62, respectively). Further, the tumor/muscle contrast in the index tumorgrafts was 4.4 ± 0.4, which is also significantly higher than controls (2.7 ± 0.6) (p < 0.05). (All statistical analyses were performed using GraphPad Prism 7 by un-paired t-tests without correction for multiple comparisons and an alpha value of 0.05.) Following the last PET scan, mice were sacrificed, and tumorgrafts and other vital organs were collected for IHC assays. IHC analyses of the tumorgrafts confirmed expected levels of PD-L1 expression.