Glioblastoma remains one of medicine’s most formidable opponents. Median survival after diagnosis is still under a year despite surgery, radiation and chemotherapy, and magnetic resonance imaging, the clinical workhorse for tumor detection, frequently underestimates how far malignant cells have infiltrated surrounding brain tissue. Because cutting away more of the tumor margin measurably delays progression, clinicians have a strong incentive to see the tumor and its microenvironment more clearly. That is where PET comes in. Rather than showing anatomy, PET can reveal molecular activity, and one target has drawn particular attention: the 18-kDa translocator protein, or TSPO, which sits on the outer membranes of mitochondria in activated immune cells.

In glioblastoma, TSPO lights up on two cell types that define the tumor microenvironment: tumor-associated microglia and macrophages, the inflammatory cells the tumor recruits and manipulates, and, crucially, the glioma cells themselves, which previous work has shown are actually the dominant cellular source of TSPO signal in this disease. The radiotracer [18F]GE-180 was developed to bind TSPO with high affinity, and preclinical and clinical studies have demonstrated robust tracer accumulation in TSPO-expressing gliomas. Yet [18F]GE-180 penetrates the intact healthy brain poorly, and critics have argued that in tumors the signal might largely reflect passive flooding of tissue through a leaky blood-brain barrier rather than genuine binding to the target protein. The debate has split the imaging community and complicated the tracer’s interpretation in the clinic.

The Munich-based research team, led by Leonie Hoermann, Laura M. Bartos and Matthias Brendel of LMU University Hospital, designed an unusually rigorous set of experiments to settle the question. They worked with mice implanted with SB28, a weakly immunogenic mouse glioblastoma line that grows in an intact immune system, imaging the animals at an early stage one week after implantation and a late stage three weeks in. Each mouse received [18F]GE-180 and underwent a dynamic 60-minute microPET scan. Critically, some animals first received a massive dose of non-radioactive GE-180, 25,000 times in excess of the tracer dose. This cold ligand saturates every available TSPO binding site, so any radioactivity that still accumulates after blocking cannot be specific binding and must instead come from non-specific processes such as passive extravasation. Comparing blocked and unblocked scans therefore quantifies the true specific fraction of the signal.

The results were striking. In early-stage tumors, blocking wiped out 72 percent of the PET signal, indicating that nearly three quarters of what the scanner saw was genuine, displaceable binding. In late-stage tumors, the blockable fraction dropped to 59 percent. Contralateral hemispheres and sham-injected animals showed reductions in the 61 to 68 percent range, but the absolute drop in distribution volume was substantially larger inside tumors, confirming a bigger pool of specifically bound tracer there. To rule out tracer-specific quirks, the team repeated key experiments with [18F]DPA-714, an independent TSPO radioligand, and observed comparable blocking effects. Ex vivo gamma counting of dissected tissue corroborated the in vivo findings, showing a 55 percent radioactivity reduction in tumor after blocking versus 28 percent in non-tumor tissue.

The next question was whether the specific signal could be trusted at the level of individual cells. Here the team deployed a technique the group calls scRadiotracing: injecting the radioactive tracer into living animals, then dissociating the tumors and sorting the cells with immunomagnetic separation and flow cytometry before measuring radioactivity in each cell fraction with an ultra-sensitive gamma counter. When cells came from blocked animals, their radioactive burden collapsed. Tumor cells showed 95 percent less uptake after blocking, tumor-associated microglia and macrophages 98 percent less, and remaining cells 99 percent less, with the pattern holding at both tumor stages and in sham animals. In other words, virtually every radioactive molecule inside a cell was bound specifically to TSPO. The specificity problem, wherever it existed, was not a cellular one.

That left the extracellular compartment. To measure blood-brain barrier disruption directly, the researchers injected a fluorescent dextran with a molecular weight of 3,000 daltons into the bloodstream. The molecule is small enough to slip through a disrupted barrier but far too large to cross an intact one, making it a clean readout of pathological permeability. Confocal microscopy of brain sections revealed that dextran extravasation surged as tumors progressed, with the fluorescence intensity sum increasing roughly sevenfold from early to late stage. Most tellingly, the amount of dextran leakage correlated strongly with the residual PET signal that survived blocking (a correlation coefficient of 0.809) and inversely with the percentage reduction achieved by blocking (R = -0.824). Bigger, leakier tumors carried proportionally more non-specific signal.

The team also exploited the natural heterogeneity of glioblastoma to answer a subtle follow-up question: does barrier disruption merely add background noise, or does it also boost the specific signal by delivering more tracer to cells? Dividing tumors into twenty regional clusters and comparing blocked with unblocked scans in the same animals, they found no correlation between regional residual signal, a surrogate of local barrier leakiness, and the regionally specific signal. Cellular tracer uptake, in other words, was not being driven open-barrier delivery. Combined with the near-complete blocking of every cell fraction, this allowed the researchers to allocate the entire non-specific TSPO-PET signal to the extracellular compartment, where free tracer pools in tissue flooded through a broken barrier without ever entering a cell.

The findings reconcile what had seemed like contradictory positions in the literature. Skeptics who called [18F]GE-180 a failed radioligand were right that a substantial and variable slice of its signal tracks barrier disruption rather than the target protein. Advocates who pointed to robust tracer accumulation in gliomas were also right that the majority of the signal, particularly early in disease, represents true molecular binding inside tumor cells and immune cells. Both phenomena coexist, and their balance shifts as the tumor grows, its vasculature remodels and increasingly leaks. The practical implication for imaging specialists is that TSPO-PET remains a genuinely specific window on the glioblastoma microenvironment, but interpretation must account for tumor size and stage, because the non-specific extracellular component grows with disease progression.

The implications stretch well beyond one tracer. Blood-brain barrier disruption complicates the interpretation of molecular imaging and drug delivery across the entire spectrum of brain disease, and vascular permeability varies dramatically between tumor regions and across disease time courses. The methodological framework established here, pairing blocking studies with single-cell radiotracer allocation and direct permeability measurement in the same animal, can now be applied to other PET tracers used in glioma imaging and to other neurological conditions in which the barrier is compromised. The authors note limitations, including the use of a single tumor cell line and exclusively female mice, and previous studies have reported sex-related differences in TSPO expression and tracer uptake that future work will need to address. Still, for a field that has argued for years over what its scans really see, this study provides something rarer than a new tracer: a quantitative map of exactly where the light comes from.

Subject of Research: Specificity of TSPO-PET imaging signals and blood-brain barrier disruption in experimental glioblastoma

Article Title: Allocation of TSPO-PET signals to cellular and extracellular compartments in dependence of blood-brain barrier disruption in experimental glioblastoma

Article References: Hoermann, L., Seker, F. B., Joseph, E., Kirchleitner, S. V., Blobner, J., Kunze, L. H., Thevis, J. F., Hummel, S., Englert, A. L., Wind-Mark, K., Holzgreve, A., Lindner, S., Werner, R. A., von Baumgarten, L., Albert, N. L., Plesnila, N., Brendel, M., & Bartos, L. M. (2026). Allocation of TSPO-PET signals to cellular and extracellular compartments in dependence of blood-brain barrier disruption in experimental glioblastoma. European Journal of Nuclear Medicine and Molecular Imaging. https://doi.org/10.1007/s00259-026-08187-5

Image Credits: AI Generated

DOI: 10.1007/s00259-026-08187-5

Keywords: glioblastoma, TSPO-PET, blood-brain barrier, radiotracer specificity, GE-180, tumor microenvironment, microglia, macrophages, molecular imaging, scRadiotracing, DPA-714, preclinical imaging