Two Novel PET Tracers Reveal Limits of PD-L1 Imaging in Advanced Lung Cancer

Immunotherapy has transformed the treatment of advanced non-small cell lung cancer, but deciding which patients should receive checkpoint inhibitors remains one of the most stubborn problems in thoracic oncology. A new study published in the European Journal of Nuclear Medicine and Molecular Imaging has put two experimental PET imaging agents to the test in patients with stage IV disease, and the results offer both a technical success and a sobering reality check. Researchers led by Jasper Smit of the Netherlands Cancer Institute and Adrianus J. de Langen, working with teams at Amsterdam University Medical Centers, Yale School of Medicine, Yale Cancer Center, Leiden University Medical Center and Bristol Myers Squibb, evaluated two fluorine-18-labeled tracers that bind to programmed death-ligand 1, or PD-L1, the same molecular target that pathologists measure on tumor biopsies using immunohistochemistry.

The two tracers, known as 18F-BMS-986,229 and 18F-BMS-986,192, represent a different generation of immune imaging compared with the antibody-based agents studied earlier in the field. Because they are labeled with fluorine-18, a radionuclide with a half-life of about 110 minutes, they can be produced in a way that is compatible with routine clinical PET operations, unlike zirconium-89-labeled antibodies that require imaging days after injection. Both agents share the same target, which allowed the investigators to run a direct head-to-head comparison within the same patients, a design that removes much of the inter-patient variability that complicates tracer studies.

The study enrolled patients with stage IV non-small cell lung cancer who were scheduled to undergo baseline PET/CT scans with both tracers before starting treatment. Tracer uptake in tumors was quantified using standard measures, the SUVpeak and SUVmean, which describe the concentration of radioactivity in the hottest and average parts of a lesion respectively. The imaging findings were then correlated with PD-L1 expression measured on tumor biopsies and with clinical outcomes, including the best overall response to subsequent therapy and overall survival. The trial was registered as NCT03001882 and conducted under the Declaration of Helsinki with institutional approval at both the Netherlands Cancer Institute and Yale School of Medicine.

The headline technical finding was one of equivalence. The two tracers performed equally with respect to tumor uptake, with a mean SUVpeak of 2.75 for 18F-BMS-986,192 and 2.68 for 18F-BMS-986,229, a difference that was statistically indistinguishable with a p value of 0.780. For a field that has struggled with tracers that vary widely in their biodistribution, binding affinity and nonspecific liver uptake, this parity matters. It suggests that either agent could serve as a workhorse for PD-L1 imaging programs, and that the choice between them can be driven by practical considerations such as manufacturing logistics rather than by differences in imaging performance.

The more provocative result, however, was what the scans failed to show. When the researchers compared tracer uptake with PD-L1 immunohistochemistry, the standard biomarker used to select patients for pembrolizumab monotherapy and to guide combination regimens, they found no correlation. Patients whose tumors scored below the 50 percent tumor proportion score threshold actually showed a mean SUVpeak of 3.77, numerically higher than the 3.54 seen in patients at or above that threshold, with a p value of 0.8175 confirming the absence of any meaningful relationship. In other words, the living, whole-body picture of PD-L1 availability did not mirror the snapshot provided by a single biopsy stained in the laboratory.

This discordance will not surprise anyone who has followed the literature on PD-L1 heterogeneity. PD-L1 expression is known to vary between the primary tumor and its metastases, between different regions of the same lesion, and over time under the pressure of treatment. A biopsy samples one location at one moment, whereas a PET tracer surveys every lesion in the body simultaneously. Yet the failure of uptake to track with immunohistochemistry in this cohort also raises the harder question of what the tracers are actually measuring, and whether the small size of the study, which the authors themselves acknowledge, limits the strength of any conclusion drawn from the comparison.

The exploratory survival analysis added an intriguing twist. When the team examined the association between baseline tumor tracer uptake and overall survival, they observed that a low SUVpeak effectively excluded a long survival, whereas a higher SUVpeak did not discriminate between patients with short or long overall survival. The signal is asymmetric, and that asymmetry is potentially useful. If confirmed in larger cohorts, a low-uptake scan could act as a rule-out test, identifying patients unlikely to benefit durably from checkpoint blockade and steering them toward alternative strategies earlier than current biomarkers allow. The absence of a corresponding signal at the high end suggests that abundant PD-L1, as imaged by these tracers, is not by itself a guarantee of benefit, which echoes the clinical experience that many PD-L1-positive patients still fail immunotherapy.

The study builds on a decade of work in immunoPET, a field that has attempted to move immune biomarkers from the pathology slide to the whole-body scan. Earlier efforts using zirconium-89-labeled atezolizumab, durvalumab and pembrolizumab demonstrated that antibody-based imaging is feasible in lung cancer and melanoma, and that uptake patterns can be strikingly heterogeneous across metastatic sites in the same patient. The fluorine-18 adnectin and macrocyclic peptide tracers developed at Bristol Myers Squibb, including the two agents tested here, were designed to overcome the pharmacokinetic limitations of full antibodies, which circulate for weeks and produce high blood-pool background. Small scaffolds clear rapidly and can be imaged on the same day they are injected, which is precisely what makes them attractive for baseline staging workflows.

There are important caveats. The cohort was small, the survival analysis was explicitly exploratory, and the study was financially supported by Bristol Myers Squibb, which developed both tracers and employs two of the coauthors. The correlation between tracer uptake and response to checkpoint inhibitors, the question that matters most to patients and oncologists, was not established in this report. What the study does establish is a rigorous quantitative benchmark: two fluorine-18 PD-L1 tracers behave identically in human tumors, and neither reproduces the immunohistochemistry score that currently governs treatment decisions. That negative correlation is itself a piece of scientific information, forcing the field to confront the possibility that static PD-L1 expression, whether measured on slides or by PET, is an incomplete predictor of immunotherapy response.

The path forward will likely involve combining PD-L1 imaging with tracers that visualize other components of the anti-tumor immune response, such as activated T cells, and with dynamic scanning protocols that measure not just how much tracer accumulates but how quickly it binds and clears. The authors conclude that low tumor tracer uptake may be associated with shorter overall survival, a finding they say warrants further investigation. For now, the study stands as a careful, technically disciplined reality check for a technology that many hoped would finally solve the biomarker problem in lung cancer immunotherapy. The scans work; the biology, it turns out, is more complicated than the biomarker we have been using to describe it.

Subject of Research: PD-L1 PET/CT imaging with fluorine-18 tracers in advanced-stage non-small cell lung cancer

Article Title: PD-L1 PET/CT imaging with 18F-BMS-986,229 and 18F-BMS-986,192 in patients with advanced-stage non-small cell lung cancer

Article References: Smit, J., Borm, F. J., Gerards, N. R., Wiegers, S. E., Holden, D., Leung, D., Grootendorst, D. J., Windhorst, B., Nabulsi, N., Cai, Z., Huang, Y., Carson, R. E., Gettinger, S., Smit, E. F., Boellaard, R., & de Langen, A. J. (2026). PD-L1 PET/CT imaging with 18F-BMS-986,229 and 18F-BMS-986,192 in patients with advanced-stage non-small cell lung cancer. European Journal of Nuclear Medicine and Molecular Imaging. https://doi.org/10.1007/s00259-026-08188-4

Image Credits: AI Generated

DOI: 10.1007/s00259-026-08188-4

Keywords: PD-L1, PET/CT, non-small cell lung cancer, immunotherapy, radiotracers, biomarkers, immunohistochemistry, molecular imaging, checkpoint inhibitors, overall survival, 18F-BMS-986229, 18F-BMS-986192