Treg therapy needs better trials

Key points

• The next generation of regulatory T cell (Treg) trials must extend beyond safety to prove that cells reach the right tissue, remain stable in inflammatory environments, and drive measurable immune reprogramming.
• Engineered Treg products should be prioritized only when engineering solves a clear translational bottleneck, such as tissue localization, persistence, or scalable manufacturing.
• Informative trials should align product, patient population, timing, endpoints, and biomarkers to assess therapeutic effectiveness and explain success or failure.

Introduction: from persuasive biology to disciplined translation

In their recent Frontiers in Science lead article, Bluestone et al. make a strong case that regulatory T cells (Tregs) should no longer be regarded simply as suppressive cells, but as adaptive coordinators of immune tolerance, tissue repair, and homeostasis (1). From a translational perspective, the field has clearly moved from asking whether Tregs matter to asking how they can be measured, stabilized, manufactured, and used therapeutically.

The attention Tregs are receiving reflects a convergence of biological understanding and technological opportunity. Forkhead box P3 (FOXP3) has given the field a lineage-defining marker, while human genetic disease has revealed the consequences of failed regulatory control. Single-cell, spatial, and epigenetic methods now allow Tregs to be mapped in tissues rather than inferred from blood (2). Genome editing, chimeric antigen receptors, synthetic receptors, cytokine engineering, and in vivo delivery technologies have made it possible to modify Tregs with some of the same sophistication that transformed effector T-cell therapy (3). In transplantation and autoimmunity, where the clinical problem is misdirected immune activation, Tregs are a logical therapeutic platform—but that rationale must now be tested in biologically informative trials.

First-generation clinical studies have established safety and feasibility, and some have provided signals of reduced immunosuppression, improved immune regulation, or preservation of residual function (4, 5). These achievements should not be minimized: the trials were technically difficult, clinically bold, and scientifically necessary. However, safety is not the same as efficacy, and feasibility is not the same as disease modification. The next phase must be more demanding.

The early clinical record provides useful lessons. In type 1 diabetes, expanded autologous polyclonal Tregs were safe and trackable after infusion, but the study was not designed to establish durable disease modification (6). The ONE Study showed that regulatory cell therapies can be delivered across international kidney transplant programs within a harmonized framework, although its non-randomized, multiproduct structure made efficacy and mechanism difficult to disentangle (7). ARTEMIS and TASK were also instructive: ARTEMIS exposed manufacturing challenges linked to post-transplant Treg dysfunction, while TASK highlighted the need to track infused Tregs’ persistence, lineage identity, and trafficking (8, 9).

Manufacturing is biology

A persistent weakness in the field is the tendency to separate the “product” from the biology. For Tregs, this distinction is artificial. Cell source, isolation strategy, expansion conditions, cryopreservation, antigen specificity, metabolic state, and FOXP3 stability all define the therapeutic agent (10). A Treg product that passes release criteria but has lost tissue-homing capacity, suppressive resilience, or epigenetic identity may be technically acceptable but biologically inadequate.

This matters more as the field moves from polyclonal products toward antigen-specific, chimeric antigen receptor (CAR)-Treg, T cell receptor (TCR)-engineered, and possibly in vivo-induced approaches. Engineering can add specificity and potency, but it also adds complexity. The relevant question is whether additional engineering solves a real clinical problem. In solid organ transplantation, polyclonal or enriched alloreactive Tregs may be sufficient if they can engage bystander suppression and infectious tolerance, whereas defined autoantigen-driven disease may require greater specificity. Precision should be earned by the biology of the indication, not assumed as a universal requirement.

The field is now very good at engineering Treg products in preclinical systems, but it is still unclear about which should move forward and why. Too often, development is driven by technical capability rather than a clear model of how the therapy will be used in patients. However, an engineered feature should not be taken forward simply because it improves a readout in vitro or in mice, but because it addresses a barrier that is likely to matter in a clinical trial. A construct may show convincing function in vitro or in a short-term animal model but lack a credible path to scalable manufacturing, regulatory release, repeat dosing, tissue delivery, or combination with existing care. This is not an argument for less ambition, but for more disciplined criteria and product selection.

Commercial development is essential if Treg therapies are to reach patients at scale, but the logic of product development inevitably favors products with clear protectability, manufacturability, and portfolio fit. Investigator-led trials therefore remain particularly important because they can address mechanistic questions that may not be commercially decisive in the short term but are essential for the field: What dose is biologically active? What tissue signal changes? What patient group is most informative? What constitutes a meaningful early response?

Clinical trials must become more biological

Many early trials are small, open-label, and safety focused. That was appropriate for first-in-human work, but is less so now. Too many studies risk using endpoints that are convenient rather than decisive, including short-term safety, transient changes in circulating Treg frequency, non-specific inflammatory markers, or composite clinical outcomes poorly matched to the proposed mechanism. At the same time, trials enrolling patients with advanced disease and established tissue damage may set an unrealistic bar if Treg therapy is expected to restore normal physiology.

The therapeutic window is likely to be disease specific and dynamic. In type 1 diabetes, preservation of residual β-cell function is more plausible than restoration after substantial loss (6). The relevant outcome in transplantation may be reduced immunosuppression with stable tissue histology and no molecular injury—not rejection-free survival under standard immunosuppression alone (4). In graft-versus-host disease or fibrotic inflammatory disease, timing may determine whether Tregs can reshape an active immune process or must contend with established tissue damage.

Future studies should therefore be designed around explicit mechanistic hypotheses. If the claim is antigen-specific tolerance, the trial should include assays of antigen-specific effector and regulatory responses. If the claim is tissue repair, tissue-level readouts are required. If the claim is durable immune re-equilibration, longitudinal sampling must extend beyond the short life of the infused product. Integrated biomarker programs should combine product characterization, immune monitoring, tissue or graft profiling where feasible, cell-free DNA or other injury biomarkers, and spatial or single-cell approaches in selected patients. These tools should not be “ornamental”; they should help determine whether the therapy did what it was designed to do.

What would constitute a step change?

A genuine step change would require five linked advances. First, the field needs better alignment between indication, product, and endpoint. Not every disease requires an engineered cell, and not every trial should use the same definition of success. Second, release criteria must move closer to mechanism. Identity, purity, and viability are necessary but insufficient. Potency assays, epigenetic stability, transcriptional fingerprinting, and stress-resilience measures under inflammatory pressure need to be developed in parallel with clinical programs, not retrospectively after failure.

Third, manufacturing must become more reproducible and scalable without losing biological intent. Closed systems, automation, and platform processes are important, but beyond cheaper production, the goal must be consistent delivery of a cell product with a well-understood function. Fourth, trials must ask sufficiently ambitious clinical questions, for example, safe immunosuppression withdrawal, preserved graft integrity, clinical remission, or durable biomarker-defined immune regulation. Finally, the field needs a culture of informative negative results. A failed Treg trial should tell us whether the wrong product, dose, timing, tissue, patient group, or endpoint was selected. At present, failures can be difficult to interpret or the data may not be publicly released.

Bluestone et al. (1) are right to be optimistic. Tregs now sit at the intersection of immunology, cell therapy, tissue biology, and precision medicine. The opportunity is real, but optimism will only be justified if the next generation of studies is more disciplined than the first. The field must now show that Tregs can be deployed at the right time, in the right patients, with the right product, and with endpoints capable of detecting true immune reprogramming. That is the route by which Treg therapy will move from a persuasive idea to a therapeutic step change.

Generative AI statement

The authors declared that generative AI was not used in the creation of this manuscript.

Publisher’s note

All claims expressed in this article are solely those of the authors and do not necessarily represent those of their affiliated organizations, or those of the publisher, the editors and the reviewers. Any product that may be evaluated in this article, or claim that may be made by its manufacturer, is not guaranteed or endorsed by the publisher.

Footnotes

Edited and reviewed by: Shohei Hori, The University of Tokyo, Japan