PRS-344/S095012 in solid tumors: a 4-1BB/PD-L1 bispecific stopped in Phase 1

What was tried

PRS-344/S095012 is a tetravalent bispecific fusion protein that joins 4-1BB-binding Anticalin proteins to a PD-L1-targeting antibody. It was developed by Pieris Pharmaceuticals and partnered to Servier. The design intent was to confine 4-1BB agonism to PD-L1-high tissue, so that costimulation of T cells happens where the tumor already expresses PD-L1 rather than systemically (Pieris and Servier announcement, 2021).

NCT05159388 was a first-in-human, open-label, single-group Phase 1/2 dose-escalation and expansion study in patients with advanced solid tumors that had progressed on standard of care. The registered actual enrollment was 46 participants. Dosing ran from 12 mg through 80 mg, with an additional arm combining 36 mg and obinutuzumab. The primary outcomes were safety measures: dose-limiting toxicities in the first 28 days, all adverse events, and adverse events leading to discontinuation. The study was terminated, with the posted reason "Sponsor decision to terminate during Phase 1," and an actual primary completion date of April 1, 2025 (ClinicalTrials.gov, NCT05159388).

The biological hypothesis

4-1BB, encoded by TNFRSF9 and also called CD137, is an inducible costimulatory receptor on activated T and natural killer cells. Ligation upregulates antiapoptotic molecules, drives cytokine secretion, and restores effector function in exhausted T cells (Chester et al., Blood 2017). Pairing 4-1BB agonism with PD-L1 blockade is mechanistically attractive because the two signals are complementary: one releases a brake while the other presses an accelerator. Preclinical data on PRS-344 reported that simultaneous engagement of PD-L1 and 4-1BB was superior to a combination of separate PD-L1 and 4-1BB antibodies (Pieris and Servier, 2021).

The genetic anchoring for this target in cancer is weak. The Open Targets v4 association score between TNFRSF9 and neoplasm is 0.139, driven by expression and literature evidence rather than human genetic causality. That score places 4-1BB among targets where the rationale is immunological and preclinical rather than genetically validated.

What actually happened

The trial closed during Phase 1 dose escalation by sponsor decision, before any efficacy expansion was read out. Pieris stated it did not intend to pursue further development of the asset, and Servier stopped dosing (pharmaphorum, 2021).

The posted results record a heavily pretreated population. Serious adverse events affected most participants in each cohort, and all-cause deaths during the study were frequent: 14 of 14 in the 60 mg arm, 8 of 13 in the 36 mg plus obinutuzumab arm, and 3 of 4 in the 12 mg arm (ClinicalTrials.gov results posting, NCT05159388). These deaths are all-cause in a metastatic, progression-after-standard-of-care population, so they reflect underlying disease far more than drug effect. The trial was not designed or powered to separate the two, and no efficacy endpoint was reported because the study stopped in escalation.

Failure mechanism, best guess

The stated reason is a sponsor decision, not a safety stop or an efficacy readout. The mechanistic context is what makes the decision legible. Systemic 4-1BB agonism has been hard to dose. Urelumab produced inflammatory liver toxicity at doses above 1 mg/kg, which capped exposure below the level where responses appeared, while utomilumab was safer but a weaker agonist (Chester et al., Blood 2017). Agonist activity depends on Fc-gamma-receptor-mediated hyperclustering, and that same clustering drives the hepatotoxicity that kept first-generation antibodies out of Phase 3 (Claus et al., Science Translational Medicine 2019). Engineering work since then confirmed that epitope and FcgammaRIIB cross-linking, not high affinity, govern both antitumor activity and liver safety (Ho et al., Molecular Cancer Therapeutics 2020).

Tumor-conditional designs like PRS-344 exist precisely to widen this therapeutic window. The class problem is that conditional activation has not yet translated into a clear monotherapy or combination efficacy signal that justifies continued spend against a crowded field. The most recent preclinical entrants still frame their value as escaping the dose-limiting hepatotoxicity that has constrained the class (Liu et al., Oncoimmunology 2026). A sponsor looking at an unread Phase 1, a saturated competitive landscape, and weak genetic validation has a defensible reason to reallocate.

How to prevent this next time

Two quantitative tools would have sharpened the go or no-go before escalation consumed 46 patients.

First, historical base-rate adjustment. The likelihood of approval from Phase 1 in oncology is approximately 5.3 percent, against roughly 9.6 percent across all indications (Wong, Siah, Lo, Biostatistics 2019). A 4-1BB costimulation program with no human genetic anchor (Open Targets 0.139) should enter with a prior below the oncology base rate, not above it, and the development plan should state in advance what interim signal would move that prior.

Second, a Bayesian posterior probability of success at a prespecified interim, integrating the response data against that adjusted prior:

PP(success)=∫θ​P(final success∣θ,interim data)⋅π(θ∣interim data)dθ

If the posterior at the end of escalation falls below a preset futility threshold, the program stops on a rule rather than on sentiment. A competitive landscape red-team belongs alongside this. By 2021 the 4-1BB field already held urelumab, utomilumab, multiple tumor-targeted bispecifics, and 4-1BBL fusions, so a clear-eyed read of differentiation should have set a high bar for the expansion decision (Hashimoto, Cancers 2021).

The single highest leverage change would have been to define, before first dose, the exact PD-L1-conditional pharmacodynamic and response signal that would justify expansion, and to bind continuation to a Bayesian futility rule keyed to the oncology Phase 1 base rate.

What this means for similar programs

The Claidex Mechanism Risk Score for TNFRSF9 in this run is 29 out of 100, in the yellow band. The score is dominated by the genetic deficit component (12.9 of 15), which reflects the Open Targets association of 0.139. This is the second TNFRSF9 program in the Claidex graph, after a PD-L1 by 4-1BB bispecific shelved in melanoma. The pattern across 4-1BB assets is consistent: programs are stopped by sponsor or strategic decisions before efficacy is established, not by a single decisive trial. Teams pursuing 4-1BB costimulation should treat tumor-conditional activation as necessary but not sufficient, and should carry an explicit, prespecified efficacy bar that survives contact with a crowded field.

Open questions

Did PRS-344 reach pharmacodynamically active exposure, and was there any conditional 4-1BB engagement signal in PD-L1-high tumors before termination? Would the obinutuzumab combination arm have changed the risk-benefit if escalation had continued? More broadly, can any tumor-localized 4-1BB agonist convert a wider therapeutic window into a durable response rate that clears the oncology base rate, or is the costimulation thesis in solid tumors structurally limited by the biology of clustering-dependent agonism?

Sources

1. • ClinicalTrials.gov, NCT05159388 (official title, single-group Phase 1/2 design, actual enrollment 46, terminated, "Sponsor decision to terminate during Phase 1," primary completion April 1, 2025, results posting cohort serious adverse event and all-cause death counts). https://clinicaltrials.gov/study/NCT05159388 - Pieris Pharmaceuticals and Servier, first-patient-dosed announcement describing the 4-1BB/PD-L1 Anticalin bispecific design (2021). https://www.pieris.com/investors/news-events/press-releases/detail/673/pieris-pharmaceuticals-and-servier-announce-dosing-of-first - pharmaphorum, report that Servier stopped dosing and Pieris ended development (2021). https://pharmaphorum.com/news/pieris-loses-servier-partner-cancer-immunotherapies - Chester C, et al. Immunotherapy targeting 4-1BB: mechanistic rationale, clinical results, and future strategies. Blood.

2. [DOI]( - Claus C, et al. Tumor-targeted 4-1BB agonists for combination with T cell bispecific antibodies as off-the-shelf therapy. Sci Transl Med.

3. [DOI]( - Ho SK, et al. Epitope and Fc-mediated cross-linking, but not high affinity, are critical for antitumor activity of CD137 agonist antibody with reduced liver toxicity. Mol Cancer Ther.

4. [DOI]( - Hashimoto K. CD137 as an attractive T cell co-stimulatory target in the TNFRSF for immuno-oncology drug development. Cancers.

5. [DOI]( - Liu S, et al. FTL008.16, a 5T4-conditional 4-1BB bispecific antibody. Oncoimmunology.

6. [DOI]( - Wong CH, Siah KW, Lo AW. Estimation of clinical trial success rates and related parameters. Biostatistics.

7. [DOI]( - Open Targets Platform v4, TNFRSF9 (ENSG00000049249) to neoplasm (EFO_0000616) association score 0.139. https://platform.opentargets.org/target/ENSG00000049249.