SIM0348: a TIGIT and PVRIG bispecific closed by business decision before either arm was tested

What was tried

Jiangsu Simcere Pharmaceutical opened NCT05718219 on 29 March 2023 as a multicentre, open-label, non-randomised first-in-human Phase 1 study of SIM0348 in adults with advanced solid tumours who had failed at least one standard therapy. SIM0348 is a bispecific antibody against TIGIT and PVRIG, also called CD112R, infused on days 1, 8, 15 and 22 of each cycle.

Part 1A escalated SIM0348 as monotherapy, Part 1B added sintilimab at 200 mg every three weeks, and Parts 2A and 2B expanded the selected doses. Primary endpoints were dose-limiting toxicity through cycle 1 and response rate by RECIST v1.1. Eligibility required ECOG status 0 or 1, with no biomarker requirement.

Enrolment reached 49 participants and primary completion was recorded on 30 April 2025. The record was updated on 25 August 2026 to Terminated, reason "Business decision", with no results posted. A drug-profile database records the asset moving to Shanghai Xianxiang Medical Technology in October 2023, and a December 2025 readout we could not retrieve.

The biological hypothesis

TIGIT and PVRIG are inhibitory receptors in the poliovirus receptor cosignalling network. TIGIT binds CD155, PVRIG binds CD112, and both compete with the activating receptor DNAM-1 for the same nectin ligands. They are co-expressed on activated T and NK cells, which is the argument for blocking them together: relieving one arm leaves the other free to suppress the same cell.

SIM0348 tested that argument directly, with anti-PVRIG nanobodies fused to the N terminus of an anti-TIGIT antibody. In the published preclinical work, TIGIT and PVRIG expression was measured on tumour-infiltrating lymphocytes from patients with non-small cell lung cancer (63) and colorectal cancer (26). The bispecific blocked both receptors from their ligands and produced a 2.8-fold increase in T-cell activation and a 1.8-fold increase in NK-cell cytotoxicity, each at P below 0.05, with antitumour activity alone and with anti-PD-1 or anti-PD-L1 in mouse models.

Genetic anchoring is thin on both arms. Open Targets scores PVRIG against neoplasm at 0.097, carried entirely by literature evidence at 0.798 with no clinical and no genetic contribution. TIGIT scores 0.139, from literature at 0.989 and clinical at 0.125, again with no genetic component. Open Targets lists no clinical-stage PVRIG-directed molecule, against seven for TIGIT.

What actually happened

The trial enrolled 49 participants, reached its primary completion date, and was terminated for a stated business decision. No safety reason is given or implied, and nothing was posted to ClinicalTrials.gov: no dose-limiting toxicity count, no recommended Phase 2 dose, no response rate, no adverse event table.

The surrounding context is the informative part. Of the seven TIGIT-directed molecules Open Targets records at clinical stage, four reached Phase 3: ociperlimab, tiragolumab, domvanalimab and vibostolimab. The Claidex graph holds two TIGIT post-mortems, both efficacy failures in non-small cell lung cancer. SIM0348 was terminated into that landscape, not out of a safety finding of its own.

Failure mechanism, best guess

We classify this as strategic_reprioritization because no trial-level evidence points anywhere else. The asset changed hands in October 2023, six months after dosing began, and the study closed with a two-word explanation.

The mechanistic question is what a bispecific of this design was positioned to prove. It inherits the TIGIT development risk in full while adding a target with no clinical precedent. If the dominant problem with TIGIT blockade is the one its Phase 3 programme surfaced, that adding TIGIT to PD-1 blockade does not reliably convert into survival, then the PVRIG arm must carry the differentiating hypothesis alone. Open Targets records no clinical evidence for PVRIG, so nothing outside the preclinical package supports it doing so.

An unresolved design question sits inside the TIGIT arm too. Antitumour activity of anti-TIGIT antibodies depends partly on Fc gamma receptor engagement, and the field is divided on whether that is a feature. Fc-silent antibodies potentiate antitumour immunity without depleting regulatory T cells, while Fc-competent ones act through effector-cell engagement and Treg depletion. Fusing a nanobody to an IgG scaffold makes an implicit choice there, and no clinical data exist to say whether it was right. The study closed before it could answer its own question, in a class where the base rate of answering it favourably had already fallen.

How to prevent this next time

No endpoint-level data were released, so a Bayesian update or a power calculation would be fabrication. The available levers are qualitative.

Base-rate adjustment is the largest. Any programme entering a checkpoint target after four molecules have reached Phase 3 against it should carry that base rate into its go and no-go criteria, rather than treating a novel combination partner as a reset. Two documented TIGIT efficacy failures on file are a prior, not a footnote.

Biomarker enrichment is the second. The preclinical work measured TIGIT and PVRIG co-expression on tumour-infiltrating lymphocytes, exactly the material a trial could enrol on. A study stratified by that co-expression, or by CD112 and CD155 tumour expression, could have produced an interpretable signal from 49 patients. An all-comers population will not.

A red-team review is the third: which arm would the programme keep if the other failed, and what would establish that? Neither arm was independently controlled, so no outcome could have separated them.

The single highest leverage change would have been to enrol on measured TIGIT and PVRIG co-expression rather than unselected advanced solid tumours, so that 49 patients could speak to the PVRIG hypothesis instead of to neither target.

What this means for similar programs

PVRIG enters the Claidex graph with a mechanistic risk score of 26 of 100, yellow band. The genetic component dominates at 13.5 of 15, reflecting an Open Targets score of 0.097 with no genetic evidence behind it. Saturation contributes 2.7 of 15, this being the first PVRIG programme on file. The score describes evidentiary thinness, not a demonstrated failure of PVRIG biology.

The read-across runs two ways. For PVRIG itself, nothing here is disconfirming: the hypothesis was never tested in a way that could produce a negative, and a biomarker-selected study would still be informative. For the broader pattern of pairing a novel checkpoint with a heavily failed one, the structure repeats. The crowded arm supplies credibility, the novel arm differentiation, and a business decision arrives before either resolves.

Open questions

What did the 49 patients show? A first-in-human readout was reported at a meeting in late 2025 that we could not retrieve, so the human profile of a TIGIT-by-PVRIG bispecific remains unavailable.

Was the Fc region silenced or intact? That choice determines whether the molecule acts through checkpoint release, Treg depletion, or both. Enrolment measured no TIGIT, PVRIG, CD155 or CD112 expression, leaving any efficacy signal uninterpretable.

Sources

1. ClinicalTrials.gov NCT05718219, 29 August 2026. https://clinicaltrials.gov/study/NCT05718219 Lin Y, et al. Mol Cancer Ther 2025.https://Stamm H, et al. Mamm Genome 2018.https://Wu B, et al. J Exp Clin Cancer Res 2021.https://Rousseau A, et al. ESMO Open 2023.https://Preillon J, et al. Mol Cancer Ther 2021.https://Piovesan D, et al. Cancer Res 2024.https://Open Targets API v4, ENSG00000213413, 29 August 2026. https://platform.opentargets.org/target/ENSG00000213413 openFDA drug event API, sintilimab, 29 August 2026. https://api.fda.gov/drug/event.json.