Optimization of ⁸⁹Zr PET imaging using both phantom and clinical studies with [⁸⁹Zr]-Df-IAB22M2C, an anti-CD8 minibody.

Introduction: The use of ⁸⁹Zr labeled radiotracers has gained interest recently, in part because its half-life (78.4 hrs) is compatible with the slower clearance and targeted uptake of larger imaging probes such as antibodies/antibody fragments. For example, the anti-CD8 minibody, [⁸⁹Zr]-Df-IAB22M2C, shows good visualization of normal and tumor tissues rich in CD8+ cells at 24 hrs post-injection. Unlike ¹⁸F, however, ⁸⁹Zr has a low positron fraction (22.3%) and a 909-keV gamma in 100% of decays that adds radiation dose without adding signal. Down-scattering of this gamma into the photopeak window also adds random coincidences and may impact scatter correction. These characteristics of ⁸⁹Zr impact both visual image quality and quantitative accuracy. The goal of this study was to improve the image quality of ⁸⁹Zr PET studies through acquisition and reconstruction choices, while reducing radiation dose to the patient.

Methods: The SNMMI Clinical Trials Network (CTN) torso phantom was imaged at the University of Iowa (UI) and the University of Pennsylvania (UP). A GE Discovery MI (4 rings, 20-cm axial field-of-view AFOV) at UI imaged the phantom with 0.8 mCi in the background to simulate phase 1 trial imaging of a 3-mCi injection of [⁸⁹Zr]-Df-IAB22M2C. The list data were sub-sampled to emulate an injected dose range of 0.5-3.0 mCi. At UP a Philips Ingenuity (18-cm AFOV) imaged the phantom with 0.3 mCi in the background, followed by imaging on the PennPET Explorer long AFOV system (112-cm AFOV) to determine the impact of scanner sensitivity on quantitative performance with ⁸⁹Zr. Two patients with melanoma were imaged on the Ingenuity following a 3-mCi injection of [⁸⁹Zr]-Df-IAB22M2C; the data were sub-sampled to lower effective doses. In addition, the data were reconstructed following reduction of the upper energy threshold from 715 keV to 560 keV to mitigate the impact of the 909-keV gamma. Two additional patients were imaged ~24 hrs following a 1-mCi injection of [⁸⁹Zr]-Df-IAB22M2C on both the Ingenuity (18x5-min/bed positions) and PennPET Explorer (2x30-min/bed positions). We studied the noise equivalent counts (NEC) as a function of energy window for the patient data on the Ingenuity and PennPET Explorer scanners. In addition, we studied image noise as a function of emulated injected activity and the impact of reconstruction parameters. Results: CTN phantom results demonstrated increasingly high bias and worse precision in lesion uptake measures as the injected dose and/or scan time was reduced. These effects were mitigated with a smoother reconstruction (fewer iterations, greater post-filtering); the phantom results were used to guide the choice of dose and scan duration for the patient studies. In addition, reduction of the upper energy threshold yielded higher NEC, most notably on the PennPET Explorer. The extended AFOV of the PennPET Explorer is clearly a benefit in terms of increasing image quality and quantitative accuracy, even while reducing dose and scan duration compared to scanners with conventional AFOV.

Conclusions: The injected dose of [⁸⁹Zr]-Df-IAB22M2C can be reduced from 3 mCi to 1 mCi and still yield high quality images, especially if a smoother reconstruction and narrower energy window are employed. Further studies planned include imaging a larger CTN phantom to mimic the whole-body distribution of ⁸⁹Zr tracers and additional patient studies on both Ingenuity and PennPET Explorer scanners. Acknowledgment: IAB22M2C was kindly provided by ImaginAb, Inc., Inglewood, CA.