Two-years’ experience with a Zr-89 PET/CT scanner validation program for clinical trials

1399 Introduction: 89Zr is a positron-emitting radionuclide with a half-life of 78.4 hours, a positron branching ratio of 22.3%, and a ubiquitous 909 keV gamma. It is seeing increasing use with the development of PET tracers for the imaging of monoclonal antibodies/fragments, where slow pharmacokinetics of tens of hours require a longer half-life radionuclide. As these clinical trials often have quantitative endpoints, a validation schema is critical to assure that participating sites have both dose calibrators and PET/CT scanners calibrated for Zr-89. Below we report details and results of a scanner validation program operational for two years for [89Zr]-Df-IAB22M2C. Methods: The primary goal of the Zr-89 validation program is to provide assurance to the study sponsor that imaging data is quantitatively accurate and that data between sites is comparable. Imaging the CTN oncology phantom annually was the chosen approach. The fill was designed with concentrations 3.9 kBq/mL (0.1 µCi/mL) and contrasts (10:1) that mimicked those measured in [89Zr]-Df-IAB22M2C phase 1 trials. Phantom scan duration mirrored trial imaging protocols. Initial [89Zr]-Df-IAB22M2C clinical trials used a human injected dose of 111 MBq (3 mCi). The phantom fill was tuned to these conditions. When a subsequent trial reduced the injected dose to 37 MBq (1 mCi), phantom fill activities were proportionately reduced. Reconstruction parameters are prescribed per PET/CT scanner model to prospectively harmonize quantitative performance. Reconstructions parameters were selected based upon higher-count F-18 studies, so there was some question of applicability under these low count conditions. All participating imaging sites were required to adjust dose calibrator gains to match Zr-89 supplier measurements, as no NIST standard exists. As additional Zr-89 production sites were added, standardization of measure between producers was required. The Zr-89 oxalate provided for phantom studies requires on-site chelation with EDTA at, and by, the imaging site to remain stable in the pH neutral phantom environment. Instructions and EDTA are provided. Results: Currently 42 PET/CT validations have been completed on 34 scanners. 9 different PET/CT models are represented, ranging from state of the art SiPM systems to older non-ToF systems. Primary acceptance criteria for phantom studies included scanner calibration accuracy for Zr-89 ± 10%, co-registration accuracy with CT within 1 voxel, and recovery-coefficient (RC) curve above minimum predetermined levels. Additional data collected includes noise measures, artifact assessment, and consistency of the scanner’s RC curve with standard RC curve. Lastly, image quality is assessed by the ability to discern a 7mm sphere and discriminate two contiguous 10mm spheres. All scanners were able to pass validation criteria, but several required a repeat phantom fill/scan. In three cases scanners failed due to subtle artifacts that were traced to a specific scanner model scatter correction failure specific to Zr-89 and low count images. Software version upgrades resolved the issue. Notably these artifacts were unmasked by the CTN anthropomorphic phantom, but not with other phantoms. Overall calibration accuracy averaged 0.99 ± 0.05. Image roughness as measured by the standard deviation of background voxels averaged 19% ± 5%, but noise was predictably scanner model/reconstruction dependent. Harmonization was largely successful, but several reconstructions with a large number of iterative updates had substantial noise increases at lower activities resulting in spuriously high SUVmax values. Recommended reconstructions were modified in these cases.Conclusions: A Zr-89 PET/CT scanner validation program using an anthropomorphic phantom filled to simulate [89Zr]-Df-IAB22M2C concentrations and contrast has been successfully implemented over a two-year period. Harmonization strategies were largely successful over a broad range of PET/CT systems.