Treatment with PD-1 and PD-L1 immune checkpoint inhibitors has become standard for multiple tumour types, producing durable responses and improved overall survival in some patients. The anti-PD-L1 monoclonal antibody atezolizumab is approved for several indications and is commonly dosed at 1200 mg IV every 3 weeks. Although serum pharmacokinetics show trough concentrations above a preclinical target of 6 µg/mL, the degree of target engagement and receptor saturation within tumour tissue is incompletely understood.
Preclinical murine immunoPET studies with labelled checkpoint inhibitors suggested complete tumour saturation, but translation of those findings to human tumours is limited. Imaging with zirconium-89 labelled atezolizumab (89Zr-atezolizumab PET/CT) offers a non-invasive approach to visualise whole-body distribution and may provide direct information on available PD-L1 binding sites during treatment. The present study therefore aimed to assess tumour PD-L1 saturation using on-treatment 89Zr-atezolizumab PET/CT scans performed after therapeutic atezolizumab dosing.
This open-label imaging study was conducted at the University Medical Centre Groningen in combination with companion atezolizumab treatment studies (ClinicalTrials.gov: NCT02453984, NCT02478099). Eligible adult patients had advanced solid tumours considered likely to benefit from atezolizumab, ECOG performance status 0–2, and RECIST v1.1 measurable disease; stable brain or leptomeningeal metastases were allowed. The study received ethical approval and all patients provided written informed consent.
Two cohorts were defined. In cohort 1, tracer administration and imaging were performed after the first therapeutic atezolizumab dose. In cohort 2, tracer was administered after the first and second therapeutic doses; imaging and biopsy schedules were the same as cohort 1. Treatment consisted of atezolizumab 1200 mg IV over 120 minutes every 3 weeks, for up to 2 years.
Patients received a therapeutic infusion of 1200 mg atezolizumab followed directly by intravenous administration of 37 MBq (approximately 1 mCi; ~10 mg antibody) of 89Zr-atezolizumab. PET/CT scans were acquired on days 4 and 7 after tracer injection. Low-dose CT preceded PET acquisition on Biograph mCT 40- or 64-slice PET/CT scanners; images were reconstructed using the EARL1 algorithm. The tracer production and scan protocol followed previously published methods for 89Zr-atezolizumab.
PET/CT images were analysed using a VOI-based background and lesion approach, quantifying tumour uptake as tumour standardized uptake value (SUVmax), normal tissue uptake as SUVmean, and tumour-to-background contrast as SUVmax/SUVmean background. Tumour biopsies were obtained around day 7 of imaging cycles for autoradiography and PD-L1 immunohistochemistry to corroborate imaging findings at the tissue level. Tumour response was assessed by (i)RECIST v1.1 on diagnostic CT every 6–9 weeks, and adverse events were recorded per NCI CTCAE v4.03.
On-treatment 89Zr-atezolizumab uptake in tumours was quantified at day 7 after tracer injection. The geometric mean tumour SUVmax at day 7 was 5.5 (95% CI, 4.4–6.9) in cycle 1 and 5.4 (95% CI, 4.3–6.8) in cycle 2 (N = 13; P = 0.88), indicating similar lesion uptake between the two cycles on day 7. The geometric mean tumour-to-background ratio was 1.67 (95% CI, 1.15–2.41) in cycle 1 and 1.53 (N = 11, 95% CI, 1.05–2.21) in cycle 2 (P = 0.32).
A positive trend was observed between higher tumour-to-background ratios and best overall response (p trend = 0.069). Autoradiography of on-treatment tumour biopsies demonstrated presence of radiotracer signal within tumour tissue, and PD-L1 immunohistochemistry was performed on biopsy material according to the study plan.
Compared with previously reported pretreatment 89Zr-atezolizumab PET/CT data, on-treatment imaging in this study showed lower uptake in tumour lesions and in the spleen, while the blood pool SUVmean was higher during treatment. These differences were interpreted as indicative of partial PD-L1 tumour saturation during therapeutic dosing with atezolizumab.
The study explicitly compared on-treatment PET uptake to earlier pre-treatment 89Zr-atezolizumab imaging described in prior work. The observed reduction in tumour and splenic tracer uptake on-treatment, alongside elevated blood-pool signal, suggests that therapeutic atezolizumab distributes systemically and occupies some PD-L1 binding sites, reducing subsequent tracer binding in peripheral tissues and tumours. However, the presence of measurable tumour tracer uptake and autoradiography signal indicates incomplete or partial saturation rather than complete blockade of PD-L1 sites by therapeutic antibody.
Response assessment followed (i)RECIST v1.1 on scheduled CT scans. Progression-free survival was defined from treatment initiation to confirmed progression or death. Adverse events were graded per NCI CTCAE v4.03. The imaging study was embedded within companion clinical trials registered as NCT02453984 and NCT02478099.
On-treatment 89Zr-atezolizumab PET/CT after therapeutic dosing of atezolizumab demonstrated lower tumour and splenic tracer uptake with higher blood-pool activity compared with previously reported pretreatment imaging. These findings are consistent with partial PD-L1 tumour saturation during standard atezolizumab dosing. Autoradiography confirmed tracer presence in tumour biopsies obtained on-treatment. The data provide in vivo human evidence about target engagement dynamics in tumours during PD-L1 blockade, which may inform dosing and development of immunoPET as a pharmacodynamic biomarker. Detailed methods and statistical analyses are provided in the Supplementary Data accompanying the original report.