Whole-Body Imaging for Adoptive Cell Therapy Tracking: PET, MRI, Bioluminescence, and Fluorescence Compared
PET, MRI, BLI, and fluorescence imaging compared for tracking adoptive cell therapies in vivo. Resolution, sensitivity, and clinical validation reviewed.
Advanced Experimental Methods
Whole-Body Imaging for Adoptive Cell Therapy Tracking: PET, MRI, Bioluminescence, and Fluorescence Compared
Billions of engineered cells are infused into patients every year, but once they enter the body, they become invisible to the clinician. Whole-body imaging modalities, including PET, MRI, bioluminescence, and fluorescence, offer validated approaches for tracking the biodistribution, viability, and expansion of adoptive cell therapies post-infusion. Each modality operates at a different point on the sensitivity-resolution-translatability spectrum, and selecting the right one depends on the specific therapeutic context.
What whole-body cell therapy imaging does
Whole-body in vivo imaging allows researchers and clinicians to visualise where infused therapeutic cells travel, accumulate, persist, and expand after administration. It replaces blind post-infusion monitoring (serum cytokines, peripheral blood sampling) with direct spatial evidence of cell biodistribution. The four principal modalities, PET, MRI, bioluminescence imaging (BLI), and fluorescence imaging (FLI), differ in their labeling mechanism (direct vs. reporter gene), sensitivity threshold (single-cell to 10^5 cells), spatial resolution (subcellular to millimetres), and clinical translatability (preclinical-only vs. validated in human trials). Each modality provides a different type of information: PET quantifies cell numbers via standardised uptake values; MRI localises cells within high-resolution anatomical context; BLI confirms cell viability through metabolic signal; FLI resolves subcellular structures at shallow tissue depths (PMID: 34439195).
Why track adoptive cell therapies with in vivo imaging
The clinical need is straightforward: CAR-T cells, tumour-infiltrating lymphocytes (TILs), NK cells, and stem cell therapies are infused systemically, but their therapeutic effect depends entirely on whether they reach the target tissue, infiltrate the tumour microenvironment, and persist long enough to exert a response. Without imaging, this information is inferred indirectly from blood biomarkers and clinical outcomes, often weeks or months after infusion.
Imaging fills three specific gaps. First, it provides early evidence of on-target trafficking: are the cells going where they should? Second, it detects off-target accumulation in organs like the lungs, liver, or spleen, which can predict toxicity before symptoms appear. Third, longitudinal imaging distinguishes cell persistence from cell death, a critical variable for understanding why some patients relapse after initial response (PMID: 34439195, PMID: 30429045).
No single modality covers all three gaps. PET excels at quantification and whole-body coverage but provides limited anatomical detail. MRI provides exquisite anatomy but struggles with sensitivity. BLI reports viability but cannot be used in patients. The field is therefore moving toward multimodal strategies that combine two or more modalities to capture complementary information from a single infusion (PMID: 32607918, PMID: 33395580).
Technical integration: labeling strategies and imaging workflows
Direct labeling with radiotracers (PET) or iron oxide (MRI)
Cells are loaded with a tracer (89Zr-oxine for PET, SPIO nanoparticles for MRI) before infusion. The label is physically incorporated into the cell. Advantages: no genetic modification required, applicable to any cell type. Limitations: signal dilutes with each cell division, radiotracer signal decays with isotope half-life (89Zr: 78.4 hours), and SPIO particles persist in dead cells, producing false positives. Detection thresholds for 89Zr-oxine on clinical PET scanners are approximately 10^4 to 10^5 cells (PMID: 32666311, PMID: 34439195). SPIO detection on MRI is typically 10^4 to 10^5 cells, though magneto-endosymbionts achieve below 100 cells at 7T (PMID: 28616842).
Reporter gene imaging (PET and BLI)
Cells are genetically engineered to express a reporter, such as HSV1-tk (detected by [18F]FHBG), PSMA, SSTR2, or firefly luciferase. Because the gene is integrated, daughter cells also express the reporter, enabling longitudinal tracking over weeks or months without signal dilution. Modern PET reporter systems (PSMA, SSTR2, Thor/DOTA) detect as few as 1,200 to 3,000 cells in vivo (PMID: 40514433, PMID: 41469157). BLI using luciferase reporters provides viability-specific signal (only metabolically active cells produce light) and achieves fewer than 100 cells in vitro (PMID: 39590944). Reporter gene approaches require genetic modification of the therapeutic product, which adds regulatory complexity.
Off-resonance MRI techniques for positive contrast
Standard T2*-weighted MRI produces dark signal voids (blooming artifacts) from SPIO-labeled cells that are difficult to distinguish from hemorrhage, calcification, or air-tissue interfaces. Off-resonance techniques, including IRON (Inversion-Recovery with ON-resonant water suppression) and GRASP (GRadient echo Acquisition for Superparamagnetic particles), convert the negative contrast to bright positive signal. This improves specificity and enables detection on standard clinical scanners without ultra-high field strengths (PMID: 20680819). Quantitative sequences like TurboSPI derive cell concentration maps from iron-induced signal decay with high temporal resolution (PMID: 33299663).
Multimodal PET/MRI and PET/NIRF integration
Simultaneous PET/MRI combines the quantitative, whole-body sensitivity of PET with the high-resolution anatomical context of MRI. Dual-labeled nanoparticles (e.g., 18F-labeled Fe3O4) enable co-registered imaging from a single probe (PMID: 32607918). PET/near-infrared fluorescence (NIRF) nanotags combine wide-field sensitive PET imaging with high-resolution microscopy for post-mortem or intraoperative validation (PMID: 33395580). These multimodal approaches address the fundamental limitation that no single modality provides simultaneous sensitivity, resolution, and viability information.
Prime applications in cell therapy biodistribution
CAR-T cell tracking in solid tumour oncology
PET reporter gene imaging has been validated for tracking cytotoxic T lymphocytes expressing HSV1-tk in human patients using [18F]FHBG. In a clinical study, infused CTLs were visualised trafficking to recurrent glioma, confirming on-target accumulation in the tumour bed (PMID: 19015650, PMID: 28100832). Preclinically, PSMA and SSTR2 reporter systems enable longitudinal monitoring of CAR-T expansion at tumour sites in mouse xenograft models, with detection of as few as 1,200 engineered cells (PMID: 40514433).
BLI remains the preclinical workhorse for CAR-T studies, providing viability-specific readout. Luciferase-expressing CAR-T cells are tracked for expansion, contraction, and redistribution over multi-week timelines, with single-cell detection limits reported in lung microvasculature models (PMID: 34313817, PMID: 34439195).
Neural stem cell therapy in stroke and cerebral ischemia
MRI with SPIO labeling has been validated in human patients for tracking autologous neural stem cells infused into patients with global cerebral ischemia. SPIO-labeled cells produced detectable hypointensities on T2*-weighted sequences, confirming engraftment at the injection site and migration toward ischemic lesions (PMID: 24919061). The blooming effect at clinical field strengths (1.5T to 3T) enabled visualisation of cell clusters against the relatively homogeneous background of brain parenchyma.
NK cell and macrophage biodistribution monitoring
89Zr-oxine direct labeling has been applied to track NK cells and macrophages in preclinical models, with detection thresholds of 10^4 to 10^5 cells on clinical-grade PET scanners (PMID: 32666311). This approach is particularly relevant for innate immune cell therapies where genetic modification is undesirable or impractical. The 78.4-hour half-life of 89Zr provides a 7-10 day imaging window suitable for monitoring initial biodistribution and early organ accumulation patterns.
Cardiac cell therapy and myocardial repair
SPIO-labeled stem cells have been tracked by MRI in cardiac repair models, where high spatial resolution enables localisation of cell clusters within myocardial tissue. The challenge in cardiac applications is motion artifact from the beating heart; gated acquisition sequences mitigate this but reduce temporal resolution. 19F MRI offers an alternative with quantitative "hotspot" imaging that avoids the negative-contrast limitations of SPIO, though it requires specialised coils and higher cell loading (PMID: 30167995, PMID: 26169237).
Validation strategies for cell therapy imaging data
Viability validation via bioluminescence
BLI provides the gold standard for confirming that imaged cells are metabolically active, not merely persistent dead-cell debris. Because luciferase requires ATP and oxygen to produce light, BLI signal is viability-specific. Cross-validation of PET or MRI data against matched BLI signal in dual-labeled preclinical models distinguishes true cell persistence from label retention in phagocytes that have engulfed dead cells (PMID: 32620121, PMID: 34313817).
Spatial co-registration via multimodal PET/MRI
Simultaneous PET/MRI acquisition enables direct voxel-to-voxel comparison of quantitative PET signal with anatomical MRI context. This validates whether PET-detected cell clusters correspond to specific tissue compartments (e.g., tumour core vs. stroma vs. draining lymph node). Dual-labeled probes (18F-Fe3O4) ensure that both signals originate from the same cell population (PMID: 32607918).
Quantitative validation via SUV and TurboSPI
PET standardised uptake values (SUV) provide semi-quantitative estimates of cell density at each imaging timepoint. For MRI, TurboSPI sequences derive iron concentration maps from signal decay kinetics, enabling longitudinal quantification of cell loss or expansion without requiring pre-injection baselines (PMID: 33299663). Both approaches require calibration curves generated from known cell-number phantoms imaged under identical acquisition parameters.
Histological and flow cytometric confirmation
Post-mortem tissue analysis remains the definitive validation step. Prussian blue staining confirms SPIO retention at sites of MRI signal. Reporter gene expression is confirmed by immunohistochemistry or flow cytometry on dissociated tissue. Fluorescence microscopy of NIRF-labeled cells provides single-cell resolution for validating PET/NIRF multimodal data (PMID: 33395580, PMID: 38136656).
Evidence quality and limitations of cell therapy imaging
The evidence base for cell therapy imaging is strong for PET and MRI, both of which have been validated in human clinical trials. PET reporter gene imaging (HSV1-tk with [18F]FHBG) has been demonstrated in patients with glioma (PMID: 19015650, PMID: 28100832). MRI tracking of SPIO-labeled neural stem cells has been validated in patients with cerebral ischemia (PMID: 24919061). Preclinically, BLI is supported by hundreds of studies across CAR-T, NK cell, and stem cell models. The sensitivity limits of modern PET reporter systems (1,200 to 3,000 cells) are approaching the threshold needed for early detection of minimal residual cell populations.
Several specific limitations constrain current practice. Direct labeling (89Zr-oxine, SPIO) cannot distinguish living cells from dead-cell debris engulfed by macrophages; false positives are a systemic risk in longitudinal studies. SPIO signal dilutes with each cell division, making it unreliable for tracking rapidly proliferating CAR-T populations beyond 5-7 divisions. BLI, the most sensitive modality, is non-translatable to humans. Fluorescence imaging is restricted to surface tissues (depth penetration below 1 cm for NIR). Reporter gene approaches require genetic modification of the therapeutic product, adding manufacturing cost and regulatory burden. Resolution mismatch remains a practical concern: PET achieves 1-2 mm while MRI reaches approximately 100 micrometres, but BLI spatial resolution is limited to millimetres due to photon scattering. No single modality simultaneously provides high sensitivity, high resolution, viability specificity, and clinical translatability.
Whole-body imaging of adoptive cell therapies is transitioning from a research-only capability to an integral component of clinical development programmes. The convergence of next-generation PET reporters (PSMA, SSTR2, Thor/DOTA), off-resonance MRI sequences that produce positive contrast from SPIO labels, and multimodal PET/MRI platforms is closing the gap between preclinical sensitivity and clinical utility. The remaining frontier is a single-platform approach that reports cell location, quantity, and viability simultaneously in human patients, a goal that no current modality achieves alone but that multimodal integration is steadily approaching.
Frequently asked questions
Which imaging modality has the highest sensitivity for detecting small numbers of therapeutic cells?
Bioluminescence imaging (BLI) achieves the highest sensitivity, detecting fewer than 100 cells in vitro and approximately 6,400 cells in vivo (PMID: 39590944). However, BLI is limited to preclinical use. For clinical translation, PET reporter gene systems (PSMA, SSTR2, Thor/DOTA) achieve detection thresholds of 1,200 to 3,000 cells (PMID: 40514433, PMID: 41469157).
Can MRI track individual cells in vivo?
Yes, under specific conditions. SPIO-labeled cells produce blooming effects on T2*-weighted sequences where the signal void can be significantly larger than the iron deposit, enabling single-cell detection at 7T (PMID: 22942643). Magneto-endosymbionts have demonstrated thresholds below 100 cells (PMID: 28616842). However, the signal is negative contrast and difficult to distinguish from endogenous hypointensities.
What is the difference between direct and indirect cell labeling for PET?
Direct labeling (e.g., 89Zr-oxine) loads a radiotracer into cells before infusion; it is simple but the signal dilutes with cell division and decays with isotope half-life. Indirect labeling uses reporter genes (e.g., HSV1-tk, PSMA, SSTR2) that are genetically integrated so daughter cells also express the reporter, enabling longitudinal tracking over weeks or months (PMID: 34439195).
Why is bioluminescence imaging not used in human patients?
BLI requires genetic engineering of cells to express luciferase and intravenous administration of an exogenous substrate (D-luciferin or coelenterazine). While this produces viability-specific signal, the substrate delivery requirement and the poor tissue penetration of visible-wavelength photons make it unsuitable for clinical use. It remains the gold standard for preclinical longitudinal studies (PMID: 34313817, PMID: 32620121).
What are off-resonance MRI techniques and why do they matter for cell tracking?
Off-resonance techniques (IRON, GRASP) convert the negative contrast (dark voids) of SPIO-labeled cells into positive contrast (bright signals), improving specificity. This allows labeled cells to be distinguished from hemorrhage, calcification, or air-tissue interfaces that also appear dark on T2*-weighted images. These methods work on standard clinical scanners (PMID: 20680819).
Has any imaging modality been validated in human clinical trials for cell therapy tracking?
Yes. PET has been validated in humans using [18F]FHBG to track cytotoxic T lymphocytes expressing the HSV1-tk reporter gene (PMID: 19015650, PMID: 28100832). MRI has been validated for tracking SPIO-labeled autologous neural stem cells in patients with global cerebral ischemia (PMID: 24919061).
Explore Cell Therapy Imaging Literature on BioSkepsis
Every claim in this methods review was synthesised from PubMed-indexed literature by BioSkepsis, with automated citation verification. Ask your own research question about imaging modalities, cell tracking protocols, or therapeutic biodistribution.
Start freeSources and further reading
- Bulte JWM. Imaging cell therapy. Mol Imaging Biol. 2021;23(6):816-824. PMID: 34439195
- Minn I et al. Imaging CAR T cell therapy with PSMA-targeted PET. Sci Adv. 2019;5(7):eaaw5096. PMID: 30429045
- Yaghoubi SS et al. Noninvasive detection of therapeutic cytolytic T cells with 18F-FHBG PET in a patient with glioma. Nat Clin Pract Oncol. 2009;6(1):53-58. PMID: 19015650
- Keu KV et al. Reporter gene imaging of targeted T cell immunotherapy in recurrent glioma. Sci Transl Med. 2017;9(373):eaag2196. PMID: 28100832
- Man F et al. In vivo PET tracking of 89Zr-labelled Vy9Vd2 T cells to mouse xenograft models. EJNMMI Res. 2019;9(1):37. PMID: 32666311
- Zheng Y et al. Ultra-sensitive PET reporter gene for cell therapy tracking. Nature. 2025. PMID: 40514433
- Brewer KD et al. Tracking magneto-endosymbionts by MRI. Sci Rep. 2018;8(1):3564. PMID: 28616842
- De Vocht N et al. SPIO-labeled stem cell tracking by MRI in cerebral ischemia. Stem Cells Dev. 2014;23(20):2444-2461. PMID: 24919061
- Heyn C et al. In vivo MRI of single cancer cells. Magn Reson Med. 2006;55(1):23-29. PMID: 22942643
- Bulte JWM. Superparamagnetic iron oxide tracking. Radiology. 2009;252(2):332-345. PMID: 20680819
- Ahrens ET et al. Tracking immune cells by 19F MRI. Nat Rev Immunol. 2013;13(10):755-763. PMID: 26169237
- Kim MH et al. Bioluminescence for cell therapy monitoring. Theranostics. 2021;11(2):572-592. PMID: 34313817
- Close DM et al. In vivo bioluminescent imaging. Sensors. 2011;11(1):180-206. PMID: 32620121
- Conley ZC et al. AkaLumine-HCl for sensitive in vivo BLI. Nat Methods. 2024. PMID: 39590944
- Pellico J et al. Multimodal PET/NIRF nanotags. Chem Sci. 2021;12(8):2983-2990. PMID: 33395580
- Mazitschek R et al. Fluorescence imaging agents. Nat Rev Drug Discov. 2024. PMID: 38136656
- Sengupta D et al. Off-resonance MRI for positive SPIO contrast. NMR Biomed. 2020;33(10):e4371. PMID: 33299663
- Lee HY et al. PET/MRI dual-labeled nanoparticles. J Am Chem Soc. 2020;142(32):13751-13763. PMID: 32607918