Zirconium-89 PET Tracks Immune Cells Safely in First Human Trial

For nearly half a century, physicians who wanted to watch a patient’s own white blood cells travel through the body had to rely on gamma cameras and SPECT imaging, a technique that sacrifices spatial detail and quantitative precision. That era may now be giving way to something sharper. In a first-in-human clinical trial conducted at the University of Alabama at Birmingham, researchers have shown that a patient’s own leukocytes can be labeled with zirconium-89, a positron-emitting isotope, and imaged with PET/CT safely and with a radiation dose profile that is, if anything, better than predicted. The study, published in the European Journal of Nuclear Medicine and Molecular Imaging, marks a critical translational milestone for PET-based immune cell tracking, a technology with potential applications ranging from infection imaging to monitoring CAR T-cell therapies and even spying on immune cells invading the brain.

The logic behind the advance rests on radiochemistry that is deceptively simple. Since the mid-1970s, clinicians have labeled white blood cells with gamma emitters such as technetium-99m and indium-111, then reinjected them to locate abscesses and inflammatory lesions. Indium-111, with a physical half-life of about 2.8 days, became the workhorse for this purpose because it persists long enough to follow cells for days. Zirconium-89, with a half-life of 3.3 days, offers a nearly identical temporal window but emits positrons, enabling PET imaging, which delivers substantially higher sensitivity and better spatial and temporal resolution than SPECT. The UAB team exploited 8-hydroxyquinoline, known as oxine, a lipophilic chelator that shuttles zirconium-89 across cell membranes, allowing the radionuclide to become trapped inside the leukocytes rather than merely attached to their surfaces.

The trial enrolled four healthy adult women between 37 and 53 years of age, a deliberate choice grounded in radiation protection logic. Extrapolations from decades of indium-111 oxine dosimetry predicted a higher effective-dose coefficient for females, 1.14 mSv/MBq, than for males, 0.855 mSv/MBq, so studying women represented the more conservative test of safety. There was also a clinical rationale: the team’s longer-term goal is to image neuroinflammation in myalgic encephalomyelitis/chronic fatigue syndrome, a condition that disproportionately affects women. Each participant donated roughly 120 milliliters of blood, from which leukocytes were isolated, incubated with zirconium-89 oxine, washed, and returned intravenously within strict time limits. The mean administered activity was 10.9 plus or minus 2.9 MBq, a modest amount by nuclear medicine standards.

Quality control was rigorous. Every labeled cell preparation had to meet predefined release criteria before reinjection, including radiochemical purity of at least 95 percent, radionuclidic purity of at least 99.5 percent, verified leukocyte viability before and after labeling, sterility, and endotoxin testing. The radiolabeling itself was performed at a licensed commercial radiopharmacy under USP Chapter 797 standards, while the zirconium-89 oxine was produced in the UAB Cyclotron PET Production Facility under USP Chapter 823 guidelines. Only products passing every specification were administered, a framework designed to mirror the established clinical workflow for FDA-approved indium-111 oxine labeling and thereby ease eventual regulatory translation.

Imaging followed a demanding schedule. Each participant underwent whole-body PET/CT at five time points: immediately after injection, then at 4 to 6, 24, 48, and 72 hours. The scans captured a dynamic biological story. Early images showed the labeled cells concentrated in the liver, spleen, heart, and blood pool, as would be expected for cells still circulating. By 24 hours and increasingly at 48 and 72 hours, uptake shifted to the liver, spleen, and red marrow, particularly within the pelvis and iliac crests, reflecting the physiological homing of leukocytes to the organs where they are naturally cleared and stored. The pattern was qualitatively similar to that long documented for indium-111 oxine-labeled leukocytes and consistent with prior mouse and non-human primate studies of zirconium-89 oxine labeling.

The dosimetry numbers carry the study’s headline significance. The highest mean absorbed-dose coefficients appeared in the spleen at 3.52 plus or minus 1.38 mGy/MBq, followed by the liver at 2.1 plus or minus 0.45, the adrenals at 1.62 plus or minus 0.24, the uterus at 1.48 plus or minus 0.98, and the kidneys at 1.36 plus or minus 0.12 mGy/MBq. The mean effective-dose coefficient came out to 0.92 plus or minus 0.06 mSv/MBq, numerically lower than the predicted 1.14 mSv/MBq, a difference that reached statistical significance in an exploratory one-sample t-test with a P value of 0.006. The measured biodistribution also diverged from the assumptions used for prediction: roughly 32 percent of activity localized to the liver, 8 percent to the spleen, and 15 percent to red marrow, rather than the 30-30-34 split assumed from indium-111 data. The authors attribute much of the discrepancy to methodological differences, since PET/CT-based whole-organ dosimetry is inherently more direct than the SPECT-only methods used historically.

Safety results were unambiguous. No adverse events, clinically significant laboratory abnormalities, or pharmacological effects were observed in any participant through 72 hours, and follow-up contact 24 hours after protocol completion turned up nothing further. Vital signs remained stable throughout. One intriguing observation warrants mention: activity concentrations in plasma exceeded those in whole blood, especially at early time points, which the researchers suggest could reflect leukocyte lysis, efflux of the radionuclide from cells, or incomplete buffy coat formation during processing. This phenomenon, they note, deserves closer investigation in subsequent studies.

A key technical concern with any intracellular label is stability. Free zirconium-89 has a well-documented affinity for hydroxyapatite and accumulates in mineralized bone, so progressive cortical bone uptake would signal that the label was leaking from cells. To probe this, the team performed an exploratory skeletal analysis comparing marrow-rich regions of the spine and pelvis with cortical tibial bone. Uptake remained consistently higher in the marrow-rich compartment, while cortical tibial activity stayed low with no progressive accumulation, supporting the interpretation that skeletal signal reflects genuine leukocyte homing to red marrow rather than liberated radionuclide. The authors caution that low-level efflux beyond 72 hours cannot be fully excluded and that longer imaging windows will help settle the question.

The study also showcased a modern dosimetry workflow. Organ regions of interest were generated automatically on CT using an in-house deep learning segmentation tool, then manually refined and propagated across time points. This automation reduces interobserver variability and could make dosimetry scalable across larger, multi-institution cohorts. Time-integrated activity coefficients were computed with Simpson’s rule, and doses were calculated in OLINDA 2.1 using standard adult female phantoms and ICRP 103 tissue weighting factors. Beyond the final 72-hour scan, activity was assumed to decay only physically, a conservative assumption that likely overestimates doses for organs still clearing activity biologically.

Looking ahead, the implications stretch well beyond infection imaging. The same labeling strategy could track CAR T-cells and other cell-based therapies as they home to tumors, offering oncologists a real-time window into treatment delivery. It may prove especially valuable in neurology: animal models of stroke, traumatic brain injury, multiple sclerosis, and Alzheimer’s and Parkinson’s diseases all show leukocyte invasion of the brain, but these events have never been directly visualized in living humans because available PET tracers decay too quickly, while the critical window of immune cell infiltration spans three to four days. Zirconium-89’s half-life fits that window precisely. The research team plans to apply the technique to patients with suspected central nervous system inflammation, including multiple sclerosis and ME/CFS, while working to reduce the blood volume needed for labeling and to improve labeling efficiency. Limitations remain, including the small all-female cohort and the absence of functional testing of labeled cells beyond viability, but the foundation is now laid: for the first time, human dosimetry data confirm that PET imaging of a patient’s own immune cells is both safe and feasible, opening a new chapter in molecular imaging of the immune system.

Subject of Research: First-in-human radiation dosimetry and biodistribution of zirconium-89 oxine-labeled autologous leukocytes imaged with PET/CT

Article Title: First-in-human dosimetry and biodistribution of [89Zr]Zr-oxine-labeled autologous leukocytes using PET/CT

Article References: Gultekin, K., Bartels, J. L., Cardenas, C. E., Kumar, Y., Gimblet, G. R., Jones, C. L., Bankston, C., Jeffers, C. D., White, S. L., Lapi, S. E., Younger, J. W., & McConathy, J. E. (2026). First-in-human dosimetry and biodistribution of [89Zr]Zr-oxine-labeled autologous leukocytes using PET/CT. European Journal of Nuclear Medicine and Molecular Imaging. https://doi.org/10.1007/s00259-026-08158-w

Image Credits: AI Generated

DOI: 10.1007/s00259-026-08158-w

Keywords: zirconium-89, PET/CT, leukocytes, cell tracking, dosimetry, radiopharmaceuticals, nuclear medicine, oxine, immunimaging, neuroinflammation, CAR T-cells, first-in-human trial